I N V I T E D R E V I EW S AND S YN TH E S E S

Extreme environments and the origins of biodiversityAdaptation and speciation in sulphide spring fishes

Michael Tobler1 | Joanna L Kelley2 | Martin Plath3 | Reuroudiger Riesch4

1Division of Biology Kansas State

University Manhattan KS USA

2School of Biological Sciences Washington

State University Pullman WA USA

3Shaanxi Key Laboratory of Molecular

Biology for Agriculture College of Animal

Science and Technology Northwest AampF

University Yangling Shaanxi China

4School of Biological Sciences Centre for

Ecology Evolution and Behaviour Royal

Holloway University of London Egham

Surrey UK

Correspondence

Michael Tobler Division of Biology Kansas

State University Manhattan KS USA

Email toblerksuedu

Funding information

Army Research Office GrantAward

Number W911NF-15-1-0175 National

Science Foundation Division of Integrative

Organismal Systems GrantAward Numbers

IOS-1557795 IOS-1557860

Abstract

Organisms adapted to physiochemical stressors provide ideal systems to study evo-

lutionary mechanisms that drive adaptation and speciation This review study

focuses on livebearing fishes of the Poecilia mexicana species complex (Poeciliidae)

members of which have repeatedly colonized hydrogen sulphide (H2S)-rich springs

H2S is a potent respiratory toxicant that creates extreme environmental conditions

in aquatic ecosystems There is also a rich history of research on H2S in toxicology

and biomedicine which has facilitated the generation of a priori hypotheses about

the proximate mechanisms of adaptation Testing these hypotheses through the

application of high-throughput genomic and transcriptomic analyses has led to the

identification of the physiological underpinnings mediating adaptation to H2S-rich

environments In addition systematic natural history studies have provided a

nuanced understanding of how the presence of a physiochemical stressor interacts

with other sources of selection to drive evolutionary change in a variety of organis-

mal traits including physiology morphology behaviour and life history Adaptation

to extreme environments in P mexicana also coincides with ecological speciation

and evolutionarily independent lineages span almost the full range of the speciation

continuum from panmixia to complete reproductive isolation Multiple mechanisms

of reproductive isolation are involved in reducing gene flow between adjacent popu-

lations that are adapted to contrasting environmental conditions Comparative stud-

ies among evolutionarily independent lineages within the P mexicana species

complex and more recently other members of the family Poeciliidae that have colo-

nized H2S-rich environments will provide insights into the factors facilitating or

impeding convergent evolution providing tangible links between micro-evolutionary

processes and macro-evolutionary patterns

K E YWORD S

ecological genomics extremophile fishes hydrogen sulphide local adaptation Poeciliidae

speciation

1 | INTRODUCTION

Organisms adapted to extreme environmental conditions provide

excellent systems to investigate basic questions about the origins of

biodiversity Extreme environments are characterized by the pres-

ence of physicochemical stressors that are lethal to most organisms

(Bell 2012 Riesch Tobler amp Plath 2015 Waterman 1999) Most

extreme environments harbour relatively simple biological communi-

ties with low levels of diversity both because there are constraints

for adaptation to physiochemical stressors and because resources

are often scarce (Gibert amp Deharveng 2002 Wall amp Virginia 1999)

Comparing highly specialized extremophiles to ancestral lineages in

Received 25 July 2017 | Revised 8 January 2018 | Accepted 10 January 2018

DOI 101111mec14497

Molecular Ecology 201827843ndash859 wileyonlinelibrarycomjournalmec copy 2018 John Wiley amp Sons Ltd | 843

ldquonormalrdquo habitats provides unique opportunities to illuminate evolu-

tionary processes that give rise to diversity both in terms of pheno-

typic novelty that mediates organismal function under adverse

environmental conditions and the origin of species The clearly

defined selective regimes in extreme environments enable hypothe-

sis-driven tests of responses at all levels of biological organization

and facilitate integrative approaches to study the causes and conse-

quences of adaptive evolution (Nevo 2011 Tobler Riesch amp Plath

2015) In addition the availability of replicated lineages exposed to

the same physicochemical stressors allows for evolutionary analyses

in a comparative context spanning vast geographic and phylogenetic

distances (Riesch Plath Schlupp Tobler amp Langerhans 2014 Riesch

et al 2016) Here we review recent research on fishes inhabiting

hydrogen sulphide-rich springs that sheds light onto mechanisms of

adaptation speciation and convergent evolution

Hydrogen sulphide (H2S) is a noxious gas that profoundly shaped

the diversification of life on the planet It served as a substrate for

anaerobic energy production for early life forms that lived in the

oxygen-poor oceans of the Proterozoic (Olson amp Straub 2015) and

caused mass extinctions after aerobic organisms diversified during

the Cambrian explosion (Kump Pavlov amp Arthur 2005) In todayrsquos

well-oxygenated atmosphere H2S is relatively rare but it is pro-

duced through geochemical and biological processes in a variety of

aquatic ecosystems (Tobler Passow Greenway Kelley amp Shaw

2016) The presence of H2S creates extreme environments for meta-

zoans because it interferes with fundamental biochemical and physi-

ological processes that mediate the function of aerobic organisms

(Bagarinao 1992) Most importantly H2S binds to cytochrome c oxi-

dase (COX) thereby interrupting the oxidative phosphorylation path-

way (OXPHOS) in mitochondria and inhibiting the aerobic

production of ATP (Cooper amp Brown 2008) Consequently H2S is

highly toxic for most metazoans even in micromolar concentrations

(Beauchamp Bus Popp Boreiko amp Andjelkovich 1984 Reiffenstein

Hulbert amp Roth 1992)

In southern Mexico H2S occurs naturally in multiple freshwater

spring complexes that are fed by groundwater sources (Rosales

Lagarde 2012) These sulphide springs are distributed across multi-

ple river drainages (Rıos Tacotalpa Puyacatengo Ixtapangajoya and

Pichucalco) that are part of the Rıo Grijalva basin (Figure 1) and

they exhibit H2S concentrations orders of magnitude higher than

what is considered lethal for most metazoans (Tobler et al 2011)

All springs directly drain into adjacent streams or rivers where H2S

is immediately diluted Thus there are contrasting sulphidic and non-

sulphidic habitat types that lack physical separation Despite the high

levels of toxicity multiple lineages of small livebearing fish of the

Poecilia mexicana species complex (subgenus Mollienesia family Poe-

ciliidae) have independently colonized the H2S-rich springs in each

of the river drainages Members of this species complex inhabit a

variety of freshwater and brackish habitats from northern Mexico

south to Panama (Alda Reina Doadrio amp Bermingham 2013 Pala-

cios Voelker Arias-Rodriguez Mateos amp Tobler 2016) and the

nominal species P mexicana is common in nonsulphidic habitats sur-

rounding the sulphide springs In the eastern two drainages with

sulphide springs (Rıos Tacotalpa and Puyacatengo) colonization of

the H2S-rich habitats has occurred relatively recently

(lt100000 years ago Pfenninger et al 2014) and H2S-adapted pop-

ulations are closely related tomdashand taxonomically considered the

same species asmdashP mexicana in adjacent nonsulphidic habitats (Pala-

cios et al 2013) In the western two drainages (Rıos Ixtapangajoya

and Pichucalco) sulphide springs have been colonized earlier

(gt200000 years ago Pfenninger et al 2014) and sulphide spring

populations in these drainages are more closely related to Poecilia

limantouri which primarily occurs north of the Trans-Mexican Vol-

canic Belt in northeastern Mexico (Brown et al 2017 Palacios et al

2013 2016) These older extremophile lineages have been described

as distinct species that are endemic to sulphide springs Poecilia ther-

malis (Steindachner 1863) in the Ixtapangajoya and Poecilia sulphu-

raria (Alvarez del Villar 1948) in the Pichucalco river drainage All

sulphide spring populationsmdashirrespective of their taxonomic statusmdash

represent locally adapted ecotypes that diverged in a variety of phe-

notypic traits (Riesch et al 2016 Tobler et al 2011) Sulphide

spring ecotypes can tolerate high H2S concentrations whereas eco-

types from adjacent nonsulphidic sites succumb to the toxic effects

of H2S over short periods of time (Plath et al 2013 Tobler et al

2011) indicating strong survival selection that impacts fitness across

small spatial scales Consequently high tolerance to environmental

H2S in sulphide spring fish can be viewed as an evolutionary innova-

tion that facilitated the colonization of unoccupied ecological niches

The sulphide spring fish of the P mexicana species complex pro-

vide an ideal system for an integrative approach to studying adapta-

tion and speciation for several reasons H2S provides a strong source

of selection with predictable physiological effects and known bio-

chemical consequences (Bagarinao 1992 Li Rose amp Moore 2011

Tobler et al 2016) which facilitates hypothesis-driven approaches

to studying mechanisms underlying physiological adaptation In addi-

tion the environmental gradients are clearly defined and replicated

which allows for an assessment of convergent patterns of evolution

across independent lineages that have colonized the same habitat

type The ease of cultivation of livebearing fish in the laboratory also

permits experimental approaches to corroborate patterns docu-

mented in the field and establish causal relationships between speci-

fic sources of selection and the evolution of specific traits Finally

growing genomic resources for fish of the family Poeciliidae facilitate

genetic analyses (Kelley et al 2012 Keuroustner et al 2016 Schartl

et al 2013)

2 | UNCOVERING PROXIMATEMECHANISMS OF ADAPTATION

Recent technological and analytical breakthroughs have revolution-

ized the genomewide quantification of genetic variation and gene

expression (Metzker 2010 Wang Gerstein amp Snyder 2009) facili-

tating the identification of genetic biochemical and physiological

underpinnings of adaptation to physiochemical stressors in natural

systems The rapid generation of large data sets allows for inductive

844 | TOBLER ET AL

Pichucalco Ixtapangajoya Puyacatengo Tacotalpa

Pichucalco

Teapa

Tacotalpa

Tapijulapa

4 km

(a)

(b) (c)

(d) (e)

F IGURE 1 Overview of the distribution of sulphide springs and their inhabitants in southern Mexico (a) Map depicting the four river drainages(Pichucalco Ixtapangajoya Puyacatengo and Tacotalpa) that exhibit sulphide springs inhabited by members of the P mexicana species complexThe locations of sulphide springs are indicated with yellow triangles black triangles represent commonly used field sites for the collection ofecotypes from nonsulphidic waters Note that roads (black lines) and major towns (grey shaded areas) were added for orientation The insethighlights the location of the study region within Mexico (bndashd) Examples of H2S-rich habitats in the Pichucalco (b) Puyacatengo (c) and Tacotalpa(d) river drainages (e) Representative individuals from the nonsulphidic (top) and the sulphidic (bottom) ecotypes from the Pichucalco river drainage

TOBLER ET AL | 845

inference of putative mechanisms underlying adaptation even in sys-

tems without a priori predictions (Kell amp Oliver 2004) However

genetic and gene expression data alone are insufficient to infer

adaptation at specific loci because neutral evolutionary processes

demographic history and genetic hitchhiking can generate false posi-

tives in genomewide analyses of selection and gene expression (Bar-

ton 1998 Ramirez-Soriano Ramos-Onsins Rozas Calafell amp

Navarro 2008 Whitehead amp Crawford 2006) This underscores the

importance of generating a priori hypotheses about the molecular

underpinnings of adaptation and testing whether genetic and gene

expression variation at specific loci actually affects organismal per-

formance under different environmental conditions (Dalziel Rogers

amp Schulte 2009 Storz amp Wheat 2010)

21 | Developing and testing a priori hypotheses

There is a rich history of research in toxicology and biomedicine that

has investigated the molecular targets of H2S illuminating the mech-

anisms of H2S toxicity and detoxification in model organisms and

human cell cultures (see Li et al 2011 Olson Donald Dombkowski

amp Perry 2012 Wang 2012 Wallace amp Wang 2015 for recent

reviews) In a recent attempt of ldquoseparating hype and hoperdquo in

potential biomedical applications of H2S Olson (2011) lamented that

H2S ldquoseemingly affects all organ systems and biological processes in

which it has been investigatedrdquo While it often remains unclear

whether the documented effects are a direct consequence of H2S

exposure or a result of the disruption of cellular homeostasis this

body of work provides fruitful grounds for the development of

hypotheses about the molecular mechanisms underlying adaptation

to H2S-rich environments that can then be tested using genome

scans for selection and analyses of gene expression variation

between sulphide-adapted and nonadapted populations (see Table 1

for an overview) Most importantly candidates include direct toxicity

targets involved in OXPHOS (Cooper amp Brown 2008) and oxygen

transport (Pietri Roman-Morales amp Lopez-Garriga 2011) anaerobic

metabolism (Dubilier Giere amp Grieshaber 2005) and oxidative stress

response systems (Eghbal Pennefather amp OrsquoBrien 2004) that may

help mitigate adverse consequences of H2S exposure as well as

pathways involved in enzymatic H2S detoxification sulphur process-

ing and sulphur excretion (Hildebrandt amp Grieshaber 2008) It is

important to note that H2S is also produced endogenously by meta-

zoans in the cytosol and in mitochondria through enzymatic reac-

tions associated with the reverse trans-sulphuration pathway

(Stipanuk 2004 Stipanuk amp Ueki 2011) At concentrations well

below the toxicity threshold endogenously produced H2S serves as

a cellular signalling molecule that modulates physiological processes

including the function of ion channels that underlie smooth muscle

control and the function of neural cells (Peers Bauer Boyle Scragg

amp Dallas 2012) and it interacts with transcription factors that con-

trol inflammation cell survival and proliferation the expression of

cellular protective enzymes and hypoxia responses (Li et al 2011

Wallace amp Wang 2015 Wang 2012) Organisms may thus not strive

to merely eliminate H2S from their system and avoid effects of

toxicity Instead they are likely selected for maintaining optimal

endogenous concentrations that facilitate H2S-mediated signalling

and assure proper physiological functioning As a consequence

endogenous H2S production pathways and targets of H2S signalling

may also be modified during adaptation to environmental H2S

A series of studies in members of the P mexicana species com-

plex have tested hypotheses about the molecular underpinnings of

H2S adaptation by analysing patterns of molecular evolution in can-

didate genes (Barts et al 2018 Pfenninger et al 2014) deploying

genomewide scans for selection based on FST outliers (Pfenninger

et al 2015) and quantifying genomewide patterns of gene expres-

sion (Kelley et al 2016 Passow Henpita et al 2017) Collectively

the results of these studies have indicated a striking congruence

between candidate genes inferred from toxicological and biomedical

research and genes with empirical evidence for selection andor dif-

ferential expression between at least some of the adjacent H2S-

adapted and nonadapted populations (Table 1) (i) There was evi-

dence for selection on genes encoding the primary toxicity target of

H2S COX (complex IV in OXPHOS Pfenninger et al 2014) In addi-

tion genes encoding other components and assembly factors of

OXPHOS are either under selection or differentially expressed in sul-

phide spring populations (Kelley et al 2016 Pfenninger et al

2015) Evidence for selection and differential expression was also

uncovered for subunits of haemoglobin whose function is directly

impaired by H2S (Barts et al 2018) (ii) Genes involved in anaerobic

ATP production and in coping with oxidative stress exhibited evi-

dence for selection and differential expression suggesting a shift

towards alternative sources of ATP production and an increased abil-

ity to withstand reactive oxygen species that are produced as a by-

product of OXPHOS inhibition by H2S (Kelley et al 2016 Pfen-

ninger et al 2015) (iii) There is overwhelming evidence for selection

on and differential regulation of genes encoding components of the

sulphidequinone oxidoreductase pathway which mediates enzymatic

H2S detoxification in metazoans (Kelley et al 2016 Pfenninger

et al 2015) In addition there is evidence for upregulation of genes

encoding transporters of oxidized sulphur species which are puta-

tively involved in sulphur excretion (Kelley et al 2016)

These results suggest that adaptation to H2S-rich environments

in members of the P mexicana species complex involves both an

increased ability to cope with elevated endogenous H2S and an

increased ability to keep endogenous H2S concentrations low

through detoxification and excretion However there is a noticeable

absence of evidence for genes and physiological pathways associ-

ated with endogenous H2S production and H2S-mediated cell sig-

nalling being involved in adaptation (see Kelley et al 2016

Pfenninger et al 2015) There are three nonmutually exclusive

hypotheses that could explain this finding First the ability of popu-

lations from sulphidic habitats to regulate endogenous H2S through

detoxification may be sufficient to maintain normal H2S-mediated

signalling Second compared to the ability for detoxification and for

mitigation of direct toxic effects in terms of ATP production and

oxidative stress H2Srsquos influence on cell signalling may have relatively

low fitness consequences with concomitantly smaller evolutionary

846 | TOBLER ET AL

responses Last inferences from biomedical studies about the causal

relationship between endogenously produced H2S and various physi-

ological functions may be weak and some documented physiological

effects may instead be an indirect consequence of the disruption of

homeostasis that cascades through interconnected pathways operat-

ing in a cell In this case evolutionary studies on natural systems

exposed to H2S may provide important insights in terms of prioritiz-

ing research on biomedically relevant aspects of H2S biology Study-

ing ancestral populations susceptible to H2S will allow for

quantifying consequences of disruptions in H2S homeostasis at all

levels of biological organization (like in presently available animal

models) Adding derived H2S tolerant populations will further allow

for the detection of critical modifications in physiological processes

affected by H2S Consequently such a comparative approach may

provide a powerful tool to disentangle direct and indirect physiologi-

cal consequences of deviations in H2S homeostasis which is cur-

rently difficult in traditional biomedical models

Despite some of these discrepancies the studies exploring mech-

anisms underlying H2S tolerance in sulphide spring fishes highlight

the utility of research in toxicology and biomedicine in generating

testable hypotheses about evolutionary processes Biomedical stud-

ies on metabolic disorders (Aspiras Rohner Martineau Borowsky amp

Clifford 2015) stress responses (Rohner et al 2013) responses to

low oxygen (Eales Hollinshead amp Tennant 2016) and effects of

xenobiotics (Stockinger Di Meglio Gialitakis amp Duarte 2014) have

also facilitated our understanding of adaptive evolution in other nat-

ural systems Considering the increasingly specialized and frag-

mented research areas in the life sciences interdisciplinary

approaches and reciprocal exchange of ideas and methodologies

continue to be critical for facilitating breakthroughs in basic and

applied research on biomedical models and natural systems (Carroll

et al 2014 Nesse amp Stearns 2008 Schartl 2014)

22 | Convergent and nonconvergent evolution inindependent lineages

Our discussion of putative candidate genes that are under selection

andor differentially expressed in sulphide spring populations so far

focused on general patterns documented across evolutionarily inde-

pendent lineages in different river drainages Selection scans (Pfen-

ninger et al 2015) and analyses of differential gene expression

(Kelley et al 2016) have indicated high congruence of biological

processes associated with candidate genes across lineages in differ-

ent river drainages indicating that evolution has repeatedly and pre-

dictably modified the same physiological pathways during adaptation

to environmental H2S However the signal of evolutionary conver-

gence among sulphide spring lineages decreases with the level of

organismal organization (from physiological pathways down to geno-

mic regions genes and individual SNPs in selection scans Pfenninger

et al 2015 from physiological pathways down to genes and tran-

scripts in expression studies Passow Brown et al 2017) indicating

that different lineages have often modified the same physiological

pathways in different ways While knowledge about the biochemical

TABLE 1 Adaptation to H2S can occur through three principlemechanisms (1) Resistance involves direct toxicity targets whosefunctions are impaired by H2S Adaptive modifications of thesetoxicity targets are predicted to assure proper function despitethe presence of H2S (2) Mitigation involves alternativephysiological pathways that compensate for the inhibition ofdirect toxicity targets Adaptive modifications of these alternativephysiological pathways are predicted to allow for the maintenanceof organismal function by other means (3) Regulation involvesphysiological pathways associated with the maintenance ofendogenous H2S homeostasis Adaptive modifications ofregulatory pathways are predicted to maintain low endogenousconcentrations despite continuous flux of H2S from theenvironment into the body

Candidate pathways Evidence

(1) Resistance

OXPHOS bull Positive selection on COX in some lineages of

P mexicana (Puyacatengo and Pichucalco)a as

well as some other genera of sulphide spring

poeciliidsb COX amino acid substitutions in P

mexicana allow for the maintenance of function

in the presence of H2Sa

bull Upregulation of cytochrome c and some COX

subunits as well as downregulation of compo-

nents of complex I in all sulphide spring lineages

of P mexicana indicating complex changes in

aerobic ATP productionc

Oxygen transport

genesbull Positive selection on and upregulation of

several oxygen transport genes (including myo-

globin and multiple haemoglobin subunits) in

sulphide spring lineages of the genera

Gambusia Limia Poecilia and Pseudoxiphophorus

indicating modifications to putatively maintain

oxygen transport in the presence of H2Sd

(2) Mitigation

Anaerobic

metabolismbull Upregulation of genes associated with glycolysis

and gluconeogenesis in all sulphide spring lin-

eages of P mexicana indicating increased

capacity of anaerobic ATP productionc

Oxidative stress

responsesbull Upregulation of genes associated with

oxidative stress responses in all sulphide

spring lineages of P mexicana indicating

increased capacity to cope with H2S-induced

oxidative stressc

(3) Regulation

H2S detoxification bull Positive selection on and upregulation of sev-

eral genes associated with enzymatic H2S

detoxification in all sulphide spring lineages of

P mexicanace Upregulation of H2S detoxifica-

tion enzymes in other sulphide spring lineages

of the genera Gambusia Limia Poecilia Pseudox-

iphophorus and Xiphophorusb This indicates an

increased ability to maintain low endogenous

H2S concentrations during exposure

H2S production bull No evidence for modifications of genes associ-

ated with enzymatic production of H2Sc

aPfenninger et al (2014) bR Greenway J L Kelley amp M Tobler unpub-

lished data cKelley et al (2016) dBarts et al (2018) ePfenninger et al

(2015)

TOBLER ET AL | 847

and physiological consequences of H2S may thus enable us to pre-

dict the physiological pathways in which adaptive modifications

occur it may be impossible to predict specific modifications within

those pathways This contrasts with a common expectation in the

field that convergent evolution of molecular mechanisms underlying

phenotypic traits (ie parallel evolution) necessarily hinges on modi-

fication of the same loci across lineages as found in some of the

best-known examples of parallelism (Colosimo et al 2005 Feldman

Brodie Brodie amp Pfrender 2012 Grant Grant Markert Keller amp

Petren 2004) However modification of the same loci may only be

common under particular circumstances (Rosenblum Parent amp

Brandt 2014) for example when selection acts on standing genetic

variation (Colosimo et al 2005) when adaptive alleles are intro-

gressed across population boundaries through gene flow (Grant

et al 2004) or when there are strong genetic constraints (Feldman

et al 2012) Exploration of molecular mechanisms underlying con-

vergent evolution in many systems may therefore be better

approached from the perspective of interacting genes that are orga-

nized in pathways in which alternative genetic changes can have

equivalent functional consequences (ie functional redundancy)

Developing and testing hypotheses about parallel evolution at the

pathway level will hinge on improving genome annotations (Pavey

Bernatchez Aubin-Horth amp Landry 2012) and provide a more

nuanced understanding of the factors determining the predictability

of evolution at different levels of organismal organization (Rosen-

blum et al 2014)

Despite evidence for convergent evolution in some candidate

physiological pathways it also needs to be emphasized that this

does not apply to all P mexicana lineages or to all candidate physio-

logical pathways associated with H2S adaptation investigated to

date Nonconvergent evolutionmdashevidenced by unique lineage-speci-

fic signals of selection and differential expressionmdashmay be indicative

of unique molecular mechanisms to cope with H2S toxicity For

example signatures of positive selection on COXmdashH2Srsquos primary

toxicity targetmdashcan be found in sulphide spring lineages of the Puy-

acatengo and Pichucalco river drainages (a lineage with recent diver-

gence time and a lineage with older divergence time respectively)

but not in those of the Tacotalpa drainage (Pfenninger et al 2014)

The presence of nonconvergent genetic and gene expression differ-

ences between populations from sulphidic and non-sulphidic habitats

in different river drainages highlights the limitations of inferences

that can be drawn based on data sets generated with high-through-

put DNA and RNA sequencing The potential importance of modifi-

cations at candidate locimdashor the lack thereofmdashcan only be assessed

through functional assays that document how alternative alleles and

changes in gene expression actually affect organismal performance

The functional consequences of genetic and gene expression varia-

tion at most candidate loci remain to be investigated However

enzyme activity assays have indicated that COX maintains its activity

with increasing H2S concentrations in the two sulphide spring popu-

lations with evidence for positive selection on mitochondrially

encoded subunits (Puyacatengo and Pichucalco Pfenninger et al

2014) Key amino acid substitutions occur in the D-pathway channel

of COX1 which putatively narrow the channel width and reduce

H2Srsquos access to the binuclear centre of the protein Consequently

some lineages of P mexicana exhibit an H2S-resistant COX that can

maintain its activity and aerobic ATP production even if endogenous

H2S concentrations are elevated In contrast COX activity in sul-

phide spring fish of the Tacotalpa river drainage and in fish from

nonsulphidic habitats decreases when H2S is present which is con-

sistent with mechanisms of H2S toxicity (Pfenninger et al 2014) In

vitro assays on isolated mitochondria have nonetheless indicated

that an H2S-resistant COX is not necessary to maintain aerobic ATP

production during H2S exposure Unlike in populations from nonsul-

phidic habitats mitochondrial spare respiratory capacity is main-

tained in the presence of H2S even in the Tacotalpa sulphide spring

populations that lack an H2S-resistant COX (C Henpita M Tobler amp

JH Shaw unpublished data) This may indicate that modification of

detoxification pathways in sulphide spring fishes is sufficient to

maintain mitochondrial function and aerobic ATP production

23 | Ensuing challenges linking genotypes tofitness

Understanding the mechanistic basis of adaptation requires linking

genotypic variation to phenotypic traits that ultimately shape fitness

along selective gradients The evolution of H2S tolerance in sulphide

spring fishes appears to be a consequence of modifications in many

genes and many physiological pathways whose functions interact to

eliminate H2S from the system minimize effects of elevated endoge-

nous concentrations on direct toxicity targets and mitigate adverse

effects that cascade through cellular metabolism during disruption of

homeostasis (Table 1) Such complex genetic changes are likely com-

mon during adaptation and have been documented in other systems

that involve major evolutionary shifts including the emergence of

parasitic lifestyles (Werren et al 2010) eusociality (Simola et al

2013) asexual reproduction (Fradin et al 2017) and internal gesta-

tion with livebearing (Schartl et al 2013) Consequently comple-

mentary approaches will be required to uncover the molecular

mechanisms of adaptation to H2S in the future

1 The involvement of many genes and physiological pathways in

shaping tolerance emphasizes the importance of quantitative

genetic approaches that leverage experimental crosses between

H2S tolerant and nontolerant populations and test whether toler-

ancemdashand adaptive trait modifications that contribute to itmdashis

explained by a few major effects loci or whether it is a conse-

quence of a number of genes with small effects (Klerks Xie amp

Levinton 2011 Saltz Hessel amp Kelly 2017) Combining analyses

of phenotype-fitness associations and their underlying genetic

mechanisms via QTL mapping with the results of selection scans

from population genomic analyses (Pfenninger et al 2015) will

allow for an integrative understanding of how selection acting on

the genome results in phenotypic variation that mediates adapta-

tion to different environments (Stinchcombe amp Hoekstra 2008)

From a functional perspective breaking up correlations between

848 | TOBLER ET AL

different physiological traits (associated with resistance mitiga-

tion and regulation Table 1) in F2 crosses will further allow test-

ing what molecular mechanisms are necessary for tolerance

whether there are alternative strategies with equivalent fitness

benefits and whether there are complementary modifications

with additive effects on tolerance

2 Considering the number of genes and physiological pathways

likely involved in H2S tolerance assessing the functional conse-

quences of genetic and gene expression variation is a daunting

task At the same time such functional tests can significantly

contribute to closing the genotype-to-phenotype gap because

physiochemical stress responses can be assessed along the con-

tinuum of organismal organization (Gjuvsland Vik Beard Hunter

amp Omholt 2013) Standardized assays can be used to quantify

the biochemical makeup of organelles cells and tissues (eg pro-

tein abundance and enzyme activity) physiological performance

in vitro and in vivo (eg rates of H2S detoxification ATP produc-

tion and reactive oxygen production) and fitness-relevant proxies

of whole organism performance (eg H2S tolerance Figure 2)

This highlights the importance of methods development for the

quantification of complex phenotypic traits or even whole phe-

nomes that can keep pace with rapid data generation through

high-throughput sequencing (Houle Govindaraju amp Omholt

2010 Ritchie Holzinger Li Pendergrass amp Kim 2015) In addi-

tion genotypendashphenotype links established through such correla-

tive approaches will ideally be verified through targeted gene

editing While gene editing with CRISPRndashCas9 is now routinely

employed in model organisms (Doudna amp Charpentier 2014 Sha-

lem Sanjana amp Zhang 2015) applying this technology to non-

models still remains challenging especially in livebearing species

like poeciliid fishes where in vitro culture of embryos is not rou-

tinely employed (Martyn Weigel amp Dreyer 2006) Nonetheless

the effects of specific mutations can potentially be verified

through the establishment and manipulation of cells in culture

(Liu et al 2014) or by generating transgenics (Lin et al 2016)

3 Organismal responses to physiochemical stressors occur at multi-

ple hierarchical levels that can be plastically modulated in

response to endogenous and exogenous cues (Whitehead 2012)

Understanding adaptation to physiochemical stressors conse-

quently requires disentangling the effects of genetic variation

developmental plasticity and short-term acclimation in gene

expression patterns on the emerging physiological processes

Common garden experiments on sulphide spring fishes have

shown that many genes differentially expressed between popula-

tions from sulphidic and nonsulphidic habitats in nature remain

differentially expressed in laboratory-reared fish that have never

been exposed to H2S indicating evolved differences in constitu-

tive gene expression (Passow Henpita et al 2017) At the same

time expression patterns in the laboratory also exhibit evidence

for plasticity in response to short-term H2S exposure (Passow

Henpita et al 2017) Some of these plastic responses are shared

between populations from sulphidic and nonsulphidic habitats

(likely representing shared ancestral plasticity) while others have

been lost or gained during sulphide spring colonization A core

challenge for future research is to overcome logistical constraints

that have so far prevented long-term H2S exposures of fish in

the laboratory Exposing individuals throughout development will

be required to test how developmental plasticity and short-term

acclimation shape responses to H2S in individuals from adapted

and nonadapted populations Finally the next big scientific

advance in this context will also require the identification of reg-

ulatory mechanisms that shape expression differences during

adaptation including sequence changes in transcription factors

and their binding sites patterns of DNA methylation and chro-

matin accessibility as well as the production of microRNAs and

competing endogenous RNAs that can modulate transcript abun-

dance (Flores Wolschin amp Amdam 2013 Lopez-Maury Mar-

guerat amp Beuroahler 2008 Silva Bye el Azzouzi amp Wisloslashff 2017)

In addition the role of alternative splicing and other post-tran-

scriptional and post-translational modifications in mediating adap-

tation remains understudied by evolutionary biologists (Ast

2004 Bush Chen Tovar-Corona amp Urritia 2017) This is particu-

larly relevant because H2S has been hypothesized to play a role

in the post-translational modification of proteins with sulphur-

containing amino acids through sulfhydration (Mustafa et al

2009)

Bridging the genotypendashphenotype gap remains one of the major

challenges in biology (Gjuvsland et al 2013 Ritchie et al 2015) and

will provide opportunities to understand how organisms function in

the face of environmental stress and how evolution shapes these

functions during adaptation to different environmental conditions

Establishing the molecular mechanisms of adaptation across repli-

cated lineages exposed to the same sources of selection will also

provide insights into how evolutionary history demography and

genetics shape convergent evolutionary outcomes providing a better

understanding at what hierarchical levels of organismal organization

evolutionary change may actually be predictable (Rosenblum et al

2014)

3 | COMPLEX ORGANISMS IN A COMPLEXWORLD DISENTANGLING THE TANGLEDBANK

Mechanisms of adaptation can be studied through bottom-up

(screening for genetic modifications that allow for inferences about

relevant phenotypic traits) and top-down approaches (assessing fit-

ness-relevant phenotypic differences among populations without

knowledge about the genetic basis see Schluter 2009) Over the

years genomic and transcriptomic studies on sulphide spring popula-

tions of the P mexicana species complex have been complemented

with analyses of phenotypic variation Just as genetic analyses have

indicated that adaptation to H2S is mediated by the modification

and modulation of many genes associated with multiple physiological

pathways phenotypic studies uncovered a variety of differences

TOBLER ET AL | 849

between populations from sulphidic and nonsulphidic environments

Many of these phenotypic differences have evolved in convergence

across replicated sulphide spring lineages in different river drainages

Examples include modifications of morphological (body shape and

organ size Tobler et al 2011 Tobler Scharnweber et al 2015

Schulz-Mirbach et al 2016) physiological (tolerance and metabolic

rates Tobler et al 2011 Passow Arias-Rodriguez amp Tobler 2017)

behavioural (aggression and personality Riesch et al 2009 Bierbach

et al 2012 Sommer-Trembo et al 2016 Bierbach Arias Rodriguez

amp Plath 2018) and life history traits (fecundity and offspring size

Riesch Plath Garcia de Leon amp Schlupp 2010 Riesch et al 2014)

Hence proximate populations in different habitat types have

diverged in complex phenotypes that span many different suites of

traits and a key question is how divergence in multivariate pheno-

types arises In theory H2S could be a key source of selection driv-

ing multivariate phenotypic divergence either because it directly

exerts selection on a multitude of traits that are independently modi-

fied to reach new fitness peaks or because selection on a few fit-

ness-relevant traits causes simultaneous changes in correlated

characteristics Neither explanation seems sufficient because many

Abiotic environmentOxygen concentrations

Ionic compositionTemperature

Structure and flow

Biotic environmentTrophic resources

DemographyPredationParasitism

H2S

Genetic variation

Gene expression

Biochemistry

Physiologicalfunction

Organismal performance

Primary fitness components

Survivalgrowth

reproduction

Organismal function Environment

F IGURE 2 Understanding how organisms function in the context of their environment is a core question in evolutionary biology yet thecomplexities associated with organismal function and environmental variation provide significant challenges High-throughput sequencing hasallowed for the identification of candidate genes subject to selection and differential expression during adaptation to H2S The next step is tobridge the genotype-to-phenotype gap by quantifying the integrated responses to H2S across the continuum of organismal organization includingbiochemical phenotypes physiological functions whole organism performance and ultimately fitness In addition understanding adaptation toH2S-rich environments will also require acknowledging H2Srsquos relation to other abiotic and biotic environmental variables which can exert selectionand modulate aspects of organismal function This will allow for the establishment of cause-and-effect relationships between specific sources ofselection and evolutionary change in specific traits and contribute to our understanding of organismal function in natural systems

850 | TOBLER ET AL

of the divergent traits between populations from sulphidic and non-

sulphidic environments have no clearmdashor at best tenuousmdashlinks to

H2S toxicity and detoxification and there is only limited evidence

for within-population correlations between different traits groups

(Riesch et al 2016)

Evolutionary analyses frequently focus on seemingly simple

sources of selection juxtaposing populations that are assigned to dis-

crete habitat types to understand the evolution of specific traits

(Kaeuffer Peichel Bolnick amp Hendry 2012) But as Darwin (1859)

already observed in his ldquotangled bankrdquo allegory selection in natural

systems is rarely univariate but multifarious selective regimes con-

tribute to trait divergence even among populations that occur along

apparently simple environmental gradients Unquantified environmen-

tal variation that is subsumed in the categorization of habitat types

can profoundly affect trait variation in natural systems (Stuart et al

2017) highlighting the importance of considering and testing alterna-

tive hypotheses about the cause-and-effect relationship between

specific sources of selection and evolution of specific traits Thus

accurate natural history observations that consider organisms as holis-

tic entities consisting of interacting traits in the context of the abiotic

and biotic factors comprising their environment are critical to inform

the design and interpretation of empirical research in natural systems

(Greene 2005) Indeed adjacent sulphidic and nonsulphidic habitat

types do not merely differ in the presence or absence of H2S but also

in a variety of other abiotic and biotic factors (Figure 2) Unlike nonsul-

phidic habitats sulphide springs in southern Mexico are characterized

by extreme hypoxia elevated salinity and lower pH (Rosales Lagarde

2012 Tobler et al 2011) In addition the presence of physiochemical

stressors affects the biotic communities in these sulphide springs Spe-

cies richness of aquatic invertebrates and vertebrates is drastically

reduced as a consequence of H2S toxicity (Greenway Arias-Rodriguez

Diaz amp Tobler 2014) and primary production appears to be driven in

part by chemoautotrophic bacteria that use H2S as an energy source

rather than just by autochthonous and allochthonous photosynthetic

organisms (Roach Tobler amp Winemiller 2011) These changes in the

biotic environment affect the relative importance of intra-vs inter-

specific competition trophic resource use and exposure to predators

and parasites (Riesch Oranth et al 2010 Tobler et al 2014 Tobler

Scharnweber et al 2015)

31 | Alternative drivers of trait evolution

Acknowledging the complexity of selective regimes in sulphide

springs allows for a more nuanced understanding of trait evolution

and for the development of alternative hypotheses about the ulti-

mate mechanisms that cause divergence in specific genetic and phe-

notypic traits In some instances sources of selection other than H2S

can plausibly explain trait divergence between populations from sul-

phidic and nonsulphidic habitats For example hypoxia likely drives

increases in head size and gill surface area (Tobler et al 2011)

changes in the ionic composition of the water can explain differential

expression of genes associated with osmoregulation (Kelley et al

2016) and shifts in trophic resource use are associated with

modifications of the gastrointestinal system (Tobler Scharnweber

et al 2015) In other cases disentangling alternative hypotheses is

not trivial because selection mediated by H2S and other environmen-

tal factors are predicted to affect trait modifications in similar ways

Shifts towards anaerobic metabolism in sulphide spring fish could be

the consequence of inhibition of aerobic ATP production by H2S or

by constraints in oxygen availability Similarly reproductive life his-

tory shifts towards the production of larger but fewer offspring could

be driven by H2S (decreasing offspringrsquos surfacevolume ratio to mini-

mize the diffusion of H2S into the body Powell 1989) by habitat dif-

ferences in the quality and quantity of predation that can affect size-

specific rates of extrinsic mortality (Jorgensen Auer amp Reznick

2011) or by energetic constraints (see below) While comparisons

across different selective regimes can shed light into the ultimate dri-

vers of trait evolution (Moore Riesch amp Martin 2016) isolating the

effects of different sources of selection may not be possible because

they could act in concert and facilitate increases of organismal perfor-

mance in multiple contexts (Ghalambor Walker amp Reznick 2003)

When multifarious selection shapes multivariate phenotypic evo-

lution it is important to also consider the role of trade-offs both in

terms of trait expression (increased investment in one trait may

impose constraints on the expression of other traits) and in terms of

trait function (optimizing the trait for one function may impair the

traitrsquos utility for other functions) For example selection on large off-

spring size in sulphide spring fishes has inadvertent consequences

for fecundity (Riesch et al 2014) and selection for increased gill

surface areas due to hypoxia simultaneously increases exposure to

environmental H2S (Tobler et al 2011) Thus the outcome of phe-

notypic evolution is ultimately a result of balancing competing needs

in a manner that optimizes trait expression and organismal function

in multiple contexts to maximize fitness (Arnold 1983) and the pres-

ence of trade-offs can have cascading indirect effects with selection

on one or a few traits leading to the modification of a multitude of

apparently unrelated characteristics

In this context bioenergetics may represent a nexus in the evo-

lution of complex traits observed in sulphide spring fish and other

extreme environments Living in and adapting to sulphide springs

imposes constraints on organismal energy availability (Plath et al

2007) with according trade-offs for energy allocation Energy limita-

tion in sulphide springs does not arise from a scarcity of food

resources (chemoautotrophic bacteria eaten by the fish are abundant

in sulphide springs Tobler Scharnweber et al 2015) but from

behavioural and physiological adaptation A primary survival strategy

for fish in the toxic and hypoxic springs is to engage in aquatic sur-

face respiration where individuals breathe directly at the water sur-

face to maximize oxygen uptake and minimize H2S exposure This

behaviour mediates short-term survival (Plath et al 2007) but it

also constrains energy acquisition for benthic foragers like P mexi-

cana (Tobler Riesch Tobler amp Plath 2009) and increases susceptibil-

ity to bird predation (Riesch Oranth et al 2010) Costly investments

into H2S detoxification and tissue repair as well as a shift from aero-

bic to less efficient anaerobic metabolism further constrain energy

allocation to other functions (Passow Greenway Arias-Rodriguez

TOBLER ET AL | 851

Jeyasingh amp Tobler 2015) Theory predicts that energetic con-

straints ndash whether mediated by limitations in the rate of resource

acquisition or in the rate of energy allocation to reproduction (as

opposed to maintenance or growth) ndash should exert selection for a

reduction in organismal energy budgets which allows for the maxi-

mization of residual energy allocation for the production of offspring

(Brown Marquet amp Taper 1993) Several lines of evidence support

the hypothesis that energy limitation plays a critical role in pheno-

typic evolution of sulphide spring fish Compared to individuals in

nonsulphidic environments fish in sulphide springs have significantly

lower routine energy demands which is mediated both by reduc-

tions in body size and mass-specific metabolism (Passow Arias-

Rodriguez et al 2017) Reductions in other traits such as energeti-

cally costly behaviours (Bierbach et al 2012 Plath 2008) and the

size of energy-demanding organs (Schulz-Mirbach et al 2016) may

also contribute to the reduced energy demands and shifts in repro-

ductive life history traits (Riesch et al 2014) are consistent with

selection from energy limitation (Reznick amp Yang 1993)

32 | Ensuing challenges Disentanglingenvironmental and organismal complexities

Future research needs to empirically address how multifarious selec-

tive regimes shape the evolution of organisms with complex pheno-

types We have made great advances in our understanding of how

specific traits evolve and mediate adaptation (eg morphology Chan

et al 2010 coloration Dasmahapatra et al 2012 sensory organs

Cartwright Schwartz Merry amp Howell 2017) and we have a thor-

ough understanding of how single environmental factors can shape

organismal responses (eg climatic conditions Johnston amp Bennett

1996 predation Langerhans 2007 competition Pfennig amp Pfennig

2012) In contrast our understanding of phenotypic evolution in

response to multifarious selection is comparatively limited especially

because interactive effects of different environmental factors can lead

to organismal and evolutionary responses that are unpredictable based

on experiments that manipulate single environmental variables in iso-

lation (Gunderson Armstrong amp Stillman 2016 Todgham amp Stillman

2013) Future research thus needs to acknowledge the complexities of

environments and organismal design and address them explicitly in

empirical investigations to gain a robust understanding of biological

diversification (Culumber amp Tobler 2017 Stuart et al 2017)

4 | LINKING ADAPTATION TOSPECIATION

So far we primarily explored the causes of adaptive evolution but

what are its consequences Theory predicts that adaptation to local

environmental conditions can lead to speciation because reproduc-

tive isolation emerges incidentally as a by-product (ecological specia-

tion Nosil 2012) Population genetic analyses of Poecilia spp in

sulphide springs and adjacent populations in nonsulphidic environ-

ments have indicated strong genetic differentiation across small

spatial scales (in some instances lt100 m) and in the absence of

physical barriers that would prevent fish movement (Plath et al

2013) In addition population genetic structuring has been shown to

be stable even in the face of major environmental disturbance

BOX 1 Speciation in sulphide springs and conservation

biology

Adaptation to H2S-rich environments in poeciliids coincides

with ecological speciation giving rise to phenotypically dis-

tinct and reproductively isolated lineages In some

instances this has led to the evolution of highly distinct

micro-endemic species including Poecilia sulphuraria and

P thermalis Limia sulphurophila and Gambusia eurystoma

(Figure 3) Except for P sulphuraria which is found in two

spring complexes of the Rıo Pichucalco drainage of Mexico

(Tobler amp Plath 2009b) the distributions of these sulphide

spring endemics are restricted to single localities that

include only small areas of suitable habitat For example

the largest sulphide spring complex (the Ba~nos del Azufre

in southern Mexico which harbours P sulphuraria and G

eurystoma) includes ~18000 m2 of suitable habitat (equiva-

lent to lt35 football fields) while one of the smallest (La

Esperanza springs harbouring P thermalis) merely covers

~600 m2 (just over a tenth of a football field Brown et al

2017) Accordingly some sulphide spring endemics are

considered critically endangered (IUCN 2012) and pro-

tected by law (SEDESOL 1994)

Many threatened species with restricted ranges exhibit

low genetic diversity and high incidences of inbreeding

(Bouzat 2010 Willi Van Buskirk amp Hoffmann 2006) In

contrast at least P sulphuraria and P thermalismdashthe only

species investigated to datemdashactually exhibit relatively high

effective population sizes (Ne = 1200ndash3800 depending on

the population) and no evidence for inbreeding (FIS lt 0)

(Brown et al 2017) This is likely a consequence of the

long-term stability of sulphide springs and the high pro-

ductivity and low levels of interspecific competition also

support unusually high fish densities and thus population

census sizes (Brown et al 2017 Culumber et al 2016)

Nonetheless sulphide spring endemics are threatened by

habitat destruction which is mediated primarily by the

modification of springs for recreational purposes defor-

estation and conversion of land for agriculture and most

recently high-intensity palm oil plantations (Tobler amp Plath

2009ab) Despite the official protection of sulphide spring

endemics there have been no tangible conservation mea-

sures that foster the sustainable development of land

around sulphide springs with the well-being of the endemic

fishes in mind

852 | TOBLER ET AL

(catastrophic flood event Plath et al 2010) Population genetic dif-

ferentiation between adjacent sulphide-adapted and nonadapted

populations is indicative of ongoing ecological speciation (Box 1) but

different population pairs from sulphidic and nonsulphidic waters

have made variable progress along the speciation continuum towards

complete reproductive isolation (Plath et al 2013) This prompts

questions about the mechanisms involved in reducing gene flow

along the environmental gradient connecting the two habitat types

41 | Prezygotic mechanisms of reproductiveisolation

The simplest form of reproductive isolation between locally adapted

populations arises from selection against migrants between habitat

types (Nosil Vines amp Funk 2005) Translocation experiments in nat-

ural habitats have indicated strong survival selection against

migrants especially those translocated from nonsulphidic to sulphidic

habitats (Plath et al 2013) Fish translocated from nonsulphidic

environments face high mortality within short periods of time

(lt24 hr) which is likely caused by H2S toxicity and hypoxia in sul-

phide springs Hence low gene flow from populations inhabiting

nonsulphidic waters into populations inhabiting sulphidic habitats is

simply explained by a lack of adaptation to the extreme environmen-

tal conditions At least in some drainages there is also appreciable

mortality for sulphide spring fish translocated to nonsulphidic habi-

tats although mechanisms causing inviability are unclear and the

strength of selection appears to be lower (Plath et al 2013) If sur-

vival selection against immigrants was the sole source of reproduc-

tive isolation this should result in unidirectional gene flow from

H2S-adapted to nonadapted populations but population genetic

analyses provided no evidence for this implying that additional

mechanisms are at work (Plath et al 2013)

One such mechanism could be intersexual selection and mate

choice experiments have indicated that migrants from sulphidic to

nonsulphidic habitats are at a significant disadvantage with fish from

nonsulphidic habitats avoiding potential mating partners from sul-

phidic ones (Plath et al 2013 Sommer-Trembo et al 2016) Such

assortative mating has been found to be critical during ecological spe-

ciation because it counteracts the homogenizing effects of gene flow

(van Doorn Edelaar amp Weissing 2009) yet how assortative mating

patterns arise often remains unclear In theory assortative mating can

arise through a variety of processes (Nosil 2012) ldquoMagic traitrdquo mecha-

nisms where mating preferences evolve as a by-product of divergent

selection on ecological traits that serve as cues have recently garnered

much attention (Servedio van Doorn Kopp Frame amp Nosil 2011)

Consistent with magic trait mechanisms assortative mating in P mexi-

cana appears to be linked to adaptive population differences in body

shape which serve as cues to distinguish potential mating partners

from different habitats (Greenway Drexler Arias-Rodriguez amp Tobler

2016) However neither sulphide spring fish (which are unlikely to

encounter migrants from nonsulphidic habitats due to H2S toxicity)

nor allopatric P mexicana from nonsulphidic waters (which have never

experienced migrants from sulphide spring populations) exhibit a

preference for mating partners from their own ecotype suggesting

that assortative mating in this system is a product of reinforcement

(ie direct selection for assortative mating Greenway et al 2016)

This is different from classic magic traits mechanisms because assorta-

tive mating is not a mere by-product of adaptation but adaptive traits

have instead been co-opted as mating signals in response to selection

on mating preferences (Maan amp Seehausen 2012) Moreover males of

some H2S-adapted populations have reduced aggressive behaviour

(Bierbach et al 2012 2018) In staged contests of size-matched males

from different ecotypes intruders had a significantly lower likelihood

to establish dominance than resident males independent of whether

or not they showed reduced aggressiveness overall Hence intrasexual

selection may represent another mechanism mediating reproductive

isolation between locally adapted ecotypes Overall H2S-adapted poe-

ciliids allow disentangling the relative contributions of different forms

of ecological and sexual selectionmdashand their interdependencymdashduring

ecological speciation

42 | Ensuing challenges quantifying postzygoticmechanisms of reproductive isolation

In contrast to prezygotic mechanisms little is known about postzy-

gotic mechanisms of reproductive isolation that potentially act to

reduce gene flow between adjacent populations in sulphidic and

nonsulphidic habitats Putative hybrids between sulphidic and non-

sulphidic ecotypes have occasionally been observed in nonsulphidic

habitats immediately adjacent to confluences with sulphide springs

(R Riesch M Plath and M Tobler personal observations) but it

remains unclear whether and through what mechanisms such hybrids

exhibit reduced fitness compared to parentals

Direct selection of mitochondrially encoded genes in the pres-

ence of H2S (Pfenninger et al 2014) should be accompanied by co-

evolution at nuclear loci that directly interact with mitochondrial

gene products in the formation of protein complexes involved in the

OXPHOS pathway (Sunnucks Morales Lamb Pavlova amp Greening

2017 Willett amp Burton 2004) Disruption of co-evolved gene com-

plexes in hybrids should therefore lead to intrinsic fitness reductions

through mito-nuclear incompatibilities (Burton Pereira amp Barreto

2013 Wolff Ladoukakis Enrıquez amp Dowling 2014) which have

been hypothesized to be a key driver in speciation (Chou amp Leu

2010 Gershoni Templeton amp Mishmar 2009) Empirical tests of

these ideas remain to be conducted in sulphide spring fishes just

like studies on potential ecological and sexual selection against

hybrids that arise from interpopulation mating events

5 | PERSPECTIVES LIFE IN SULPHIDICENVIRONMENTS BEYOND P MEXICANA

Research on sulphide spring fishes of the P mexicana species com-

plex in Mexico has provided insights into mechanisms underlying

adaptation and the links between adaptation and speciation Sul-

phide springs do not only occur in Mexico however but are found

TOBLER ET AL | 853

worldwide and have been colonized by 24 evolutionarily indepen-

dent fish lineages in North America the Neotropics and the Middle

East (Greenway et al 2014) The vast majority of lineages belong to

the order Cyprinodontiformes (95) and livebearing members of the

family Poeciliidae in particular (80) This potentially indicates that

internal gestation of young has facilitated the colonization of

extreme environments because it shields developing embryos from

physiochemical stressors (Blackburn 1999) Among the Poeciliidae

sulphide spring fishes include members of the genera Brachyrhaphis

Gambusia Poeciliopsis Priapichthys Pseudoxiphophorus Xiphophorus

and multiple subgenera of Poecilia (Figure 3 Greenway et al 2014)

spanning over 40 million years of evolution (based on Reznick Fur-

ness Meredith amp Springer 2017)

A handful of studies have started to address molecular and phe-

notypic evolution across a broader taxon sampling of sulphide spring

fishes finding evidence for convergent evolution in morphology

(Riesch et al 2016 Tobler amp Hastings 2011) life history traits

(Riesch Plath et al 2010 Riesch et al 2014) and oxygen transport

genes (Barts et al 2018) In addition ecological speciation in

response to adaptation to H2S-rich environments has also been doc-

umented in sulphide spring poeciliids other than P mexicana (Riesch

et al 2016) These studies have laid a fruitful foundation for the

investigation of mechanisms of adaptation to H2S-rich springs from

genes to phenotypic traits linking micro-evolutionary processes to

macro-evolutionary patterns

The availability of broad-scale evolutionary replication of popula-

tions that have invaded habitats with the same physiochemical stressors

allows for addressing some of the ensuing challenges discussed above

and investigating the factors that give rise to predictable evolutionary

outcomes Convergent evolution is a commonmdashbut not omnipresentmdash

theme in adaptive evolution and multiple factors can determine the

degree of convergence among replicated lineages exposed to the same

sources of selection (Kaeuffer et al 2012 Landry amp Bernatchez 2010

Oke Rolshausen LeBlond amp Hendry 2017 Rosenblum amp Harmon

2011) Thus leveraging replicated lineages that have colonized sulphide

springs to assess molecular mechanisms underlying convergent pheno-

typic changes analyse how environmental heterogeneity among sul-

phide spring habitats causes lineage-specific patterns of trait evolution

and test whether and by what mechanisms adaptation to extreme envi-

ronments is linked to speciation will provide insights into how ecologi-

cal genetic and functional factors contribute to evolutionary

convergence at different levels of organismal organization (Rosenblum

et al 2014) An expanded taxon sampling with population pairs that

span almost the full range of the speciation continuum from panmixia to

complete reproductive isolation (Riesch et al 2016) will also help

addressing questions about temporal aspects of H2S adaptation In par-

ticular comparing younger to older lineages will provide insights about

the order at which genetic and phenotypic modifications accrue during

the adaptation process

Animals have also invaded H2S-rich habitats in other aquatic

ecosystems including cold seeps and deep-sea hydrothermal vents

(Tobler et al 2016) Adaptation to H2S in the P mexicana species com-

plex has involved modification of physiological pathways associated

with H2S toxicity (OXPHOS pathway) and detoxification (sulphide

F IGURE 3 Phylogeny of the family Poeciliidae (adopted from Culumber amp Tobler 2017) Various lineages in the genera Poecilia (subgeneraAcanthophacelus Limia and Mollienesia) Priapichthys Brachyrhaphis Poeciliopsis Xiphophorus Pseudoxiphophorus and Gambusia have colonizedH2S-rich springs in different parts of the Americas (see Greenway et al 2014) Lineages highlighted in yellow represent endemic speciesrestricted to H2S-rich springs while red lineages represent species in which some populations have invaded sulphidic habitats Pictures depict arepresentative of each focal genussubgenus

854 | TOBLER ET AL

quinone oxidoreductase pathway) which have been highly conserved

during the diversification of life on Earth (Saraste 1999 Shahak amp

Hauska 2008) The high replicability of H2S as an environmental factor

that shaped animal evolution across vastly different ecological contexts

consequently offers the opportunity to investigate whether selection

mediated by the presence of H2S has repeatedly modified these con-

served physiological pathways across animal phyla that have diverged

during the Cambrian explosion spanning over 500 million years of evo-

lution Such comparisons would provide unprecedented insights into

the effects of a single physiochemical stressor on evolutionary change

irrespective of ecological or phylogenetic contexts

6 | CONCLUSIONS

H2S-rich environments and their inhabitants represent useful models

for research in organismal biology ecology and evolution illustrating

both opportunities (eg strong and explicit source of selection evolu-

tionary replication) and challenges (eg complexity of environments

and organismal responses) in asking basic question about the origins

of biodiversity Understanding how organisms thrive in apparently

inhospitable environments will provide us with insights into lifersquos

capacities and limitations to adapt to novel environments (Waterman

1999) This is particularly relevant considering that human activities

substantially contribute to the exacerbation of physiochemical stres-

sors at local regional and global scales impacting ecological and evo-

lutionary dynamics (Alberti 2015 Norberg Urban Vellend

Klausmeier amp Loeuille 2012) Thus a focus on ecological and evolu-

tionary consequences of natural stressors that push organisms to the

brink of their physiological capacities may be instrumental to predict

consequences of anthropogenic environmental change

ACKNOWLEDGEMENTS

We would like to thank Lenin Arias-Rodriguez the Universidad

Juarez Autonoma de Tabasco and the Centro de Investigacion E

Innovacion Para la Ense~nanza y El Aprendizaje for their continued

collaboration enabling our work on the P mexicana system Funding

for this review article has been provided by the National Science

Foundation (IOS-1557860 IOS-1557795) and the US Army

Research Office (W911NF-15-1-0175)

AUTHOR CONTRIBUTIONS

MT wrote the first draft of the paper all authors contributed text

and comments

ORCID

Michael Tobler httporcidorg0000-0002-0326-0890

Joanna L Kelley httporcidorg0000-0002-7731-605X

Reuroudiger Riesch httporcidorg0000-0002-0223-1254

REFERENCES

Alberti M (2015) Eco-evolutionary dynamics in an urbanizing planet

Trends in Ecology amp Evolution 30 114ndash126 httpsdoiorg101016j

tree201411007

Alda F Reina R G Doadrio I amp Bermingham E (2013) Phylogeny

and biogeography of the Poecilia sphenops species complex

(Actinopterygii Poeciliidae) in Central America Molecular Phylogenet-

ics and Evolution 66 1011ndash1026 httpsdoiorg101016jympev

201212012

Alvarez del Villar J (1948) Descripcion de una nueva especie de Mol-

lienisia capturada en Ba~nos del Azufre Tabasco (Pisces Poeciliidae)

Anales de la Escuela Nacional de Ciencias Biologicas 5 275ndash281

Arnold S J (1983) Morphology performance and fitness American Zool-

ogist 23 347ndash361 httpsdoiorg101093icb232347

Aspiras A C Rohner N Martineau B Borowsky R L amp Clifford C J

(2015) Melanocortin 4 receptor mutations contribute to the adapta-

tion of cavefish to nutrient-poor conditions Proceedings of the

National Academy of Sciences USA 112 9968ndash9973

Ast G (2004) How did alternative splicing evolve Nature Reviews

Genetics 5 773ndash782 httpsdoiorg101038nrg1451

Bagarinao T (1992) Sulfide as an environmental factor and toxicant

Tolerance and adaptations in aquatic organisms Aquatic Toxicology

24 21ndash62 httpsdoiorg1010160166-445X(92)90015-F

Barton N H (1998) The effect of hitch-hiking on neutral genealogies

Genetic Research 72 123ndash133 httpsdoiorg101017

S0016672398003462

Barts N Greenway R Passow C N Arias Rodriguez L Kelley J L amp

Tobler M (2018) Expression and molecular evolution of oxygen

transport genes in livebearing fishes (Poeciliidae) from hydrogen sul-

fide rich springs Genome httpsdoiorg101139gen-2017-0051

[Epub ahead of print]

Beauchamp R O Bus J S Popp C S Boreiko C J amp Andjelkovich

D A (1984) A critical review of the literature on hydrogen sulfide

CRC Critical Reviews in Toxicology 13 25ndash97 httpsdoiorg10

310910408448409029321

Bell E M (2012) Life at extremes Environments organisms and strategies

for survival Oxfordshire CABI httpsdoiorg101079

97818459381470000

Bierbach D Arias Rodriguez L amp Plath M (2018) Intrasexual competi-

tion enhances reproductive isolation between locally adapted popula-

tions Current Zoology httpsdoiorg101093czzox071 [Epub

ahead of print]

Bierbach D Klein M Sassmannshausen V Schlupp I Riesch R

Parzefall J amp Plath M (2012) Divergent evolution of male aggres-

sive behavior Another reproductive isolation barrier in extremophile

poeciliid fishes International Journal of Evolutionary Biology 2012

148745

Blackburn D G (1999) Viviparity and oviparity Evolution and reproduc-

tive strategies In T E Knobil amp J D Neill (Eds) Encyclopedia of

reproduction Vol 4 (pp 994ndash1003) New York NY Academic Press

Bouzat J L (2010) Conservation genetics of population bottlenecks

The role of chance selection and history Conservation Genetics 11

463ndash478

Brown A P Greenway R Morgan S Quackenbush C R Giordani

L Arias-Rodriguez L Kelley J L (2017) Genome-scale data

reveals that endemic Poecilia populations from small sulfide spring

display no evidence of inbreeding Molecular Ecology 26 4920ndash

4934

Brown J H Marquet P A amp Taper M L (1993) Evolution of body

size Consequences of an energetic definition of fitness American

Naturalist 142 573ndash584 httpsdoiorg101086285558

Burton R S Pereira R J amp Barreto F S (2013) Cytonuclear genomic

interactions and hybrid breakdown Annual Review of Ecology

TOBLER ET AL | 855

Pichucalco Ixtapangajoya Puyacatengo Tacotalpa

Pichucalco

Teapa

Tacotalpa

Tapijulapa

4 km

(a)

(b) (c)

(d) (e)

F IGURE 1 Overview of the distribution of sulphide springs and their inhabitants in southern Mexico (a) Map depicting the four river drainages(Pichucalco Ixtapangajoya Puyacatengo and Tacotalpa) that exhibit sulphide springs inhabited by members of the P mexicana species complexThe locations of sulphide springs are indicated with yellow triangles black triangles represent commonly used field sites for the collection ofecotypes from nonsulphidic waters Note that roads (black lines) and major towns (grey shaded areas) were added for orientation The insethighlights the location of the study region within Mexico (bndashd) Examples of H2S-rich habitats in the Pichucalco (b) Puyacatengo (c) and Tacotalpa(d) river drainages (e) Representative individuals from the nonsulphidic (top) and the sulphidic (bottom) ecotypes from the Pichucalco river drainage

TOBLER ET AL | 845

inference of putative mechanisms underlying adaptation even in sys-

tems without a priori predictions (Kell amp Oliver 2004) However

genetic and gene expression data alone are insufficient to infer

adaptation at specific loci because neutral evolutionary processes

demographic history and genetic hitchhiking can generate false posi-

tives in genomewide analyses of selection and gene expression (Bar-

ton 1998 Ramirez-Soriano Ramos-Onsins Rozas Calafell amp

Navarro 2008 Whitehead amp Crawford 2006) This underscores the

importance of generating a priori hypotheses about the molecular

underpinnings of adaptation and testing whether genetic and gene

expression variation at specific loci actually affects organismal per-

formance under different environmental conditions (Dalziel Rogers

amp Schulte 2009 Storz amp Wheat 2010)

21 | Developing and testing a priori hypotheses

There is a rich history of research in toxicology and biomedicine that

has investigated the molecular targets of H2S illuminating the mech-

anisms of H2S toxicity and detoxification in model organisms and

human cell cultures (see Li et al 2011 Olson Donald Dombkowski

amp Perry 2012 Wang 2012 Wallace amp Wang 2015 for recent

reviews) In a recent attempt of ldquoseparating hype and hoperdquo in

potential biomedical applications of H2S Olson (2011) lamented that

H2S ldquoseemingly affects all organ systems and biological processes in

which it has been investigatedrdquo While it often remains unclear

whether the documented effects are a direct consequence of H2S

exposure or a result of the disruption of cellular homeostasis this

body of work provides fruitful grounds for the development of

hypotheses about the molecular mechanisms underlying adaptation

to H2S-rich environments that can then be tested using genome

scans for selection and analyses of gene expression variation

between sulphide-adapted and nonadapted populations (see Table 1

for an overview) Most importantly candidates include direct toxicity

targets involved in OXPHOS (Cooper amp Brown 2008) and oxygen

transport (Pietri Roman-Morales amp Lopez-Garriga 2011) anaerobic

metabolism (Dubilier Giere amp Grieshaber 2005) and oxidative stress

response systems (Eghbal Pennefather amp OrsquoBrien 2004) that may

help mitigate adverse consequences of H2S exposure as well as

pathways involved in enzymatic H2S detoxification sulphur process-

ing and sulphur excretion (Hildebrandt amp Grieshaber 2008) It is

important to note that H2S is also produced endogenously by meta-

zoans in the cytosol and in mitochondria through enzymatic reac-

tions associated with the reverse trans-sulphuration pathway

(Stipanuk 2004 Stipanuk amp Ueki 2011) At concentrations well

below the toxicity threshold endogenously produced H2S serves as

a cellular signalling molecule that modulates physiological processes

including the function of ion channels that underlie smooth muscle

control and the function of neural cells (Peers Bauer Boyle Scragg

amp Dallas 2012) and it interacts with transcription factors that con-

trol inflammation cell survival and proliferation the expression of

cellular protective enzymes and hypoxia responses (Li et al 2011

Wallace amp Wang 2015 Wang 2012) Organisms may thus not strive

to merely eliminate H2S from their system and avoid effects of

toxicity Instead they are likely selected for maintaining optimal

endogenous concentrations that facilitate H2S-mediated signalling

and assure proper physiological functioning As a consequence

endogenous H2S production pathways and targets of H2S signalling

may also be modified during adaptation to environmental H2S

A series of studies in members of the P mexicana species com-

plex have tested hypotheses about the molecular underpinnings of

H2S adaptation by analysing patterns of molecular evolution in can-

didate genes (Barts et al 2018 Pfenninger et al 2014) deploying

genomewide scans for selection based on FST outliers (Pfenninger

et al 2015) and quantifying genomewide patterns of gene expres-

sion (Kelley et al 2016 Passow Henpita et al 2017) Collectively

the results of these studies have indicated a striking congruence

between candidate genes inferred from toxicological and biomedical

research and genes with empirical evidence for selection andor dif-

ferential expression between at least some of the adjacent H2S-

adapted and nonadapted populations (Table 1) (i) There was evi-

dence for selection on genes encoding the primary toxicity target of

H2S COX (complex IV in OXPHOS Pfenninger et al 2014) In addi-

tion genes encoding other components and assembly factors of

OXPHOS are either under selection or differentially expressed in sul-

phide spring populations (Kelley et al 2016 Pfenninger et al

2015) Evidence for selection and differential expression was also

uncovered for subunits of haemoglobin whose function is directly

impaired by H2S (Barts et al 2018) (ii) Genes involved in anaerobic

ATP production and in coping with oxidative stress exhibited evi-

dence for selection and differential expression suggesting a shift

towards alternative sources of ATP production and an increased abil-

ity to withstand reactive oxygen species that are produced as a by-

product of OXPHOS inhibition by H2S (Kelley et al 2016 Pfen-

ninger et al 2015) (iii) There is overwhelming evidence for selection

on and differential regulation of genes encoding components of the

sulphidequinone oxidoreductase pathway which mediates enzymatic

H2S detoxification in metazoans (Kelley et al 2016 Pfenninger

et al 2015) In addition there is evidence for upregulation of genes

encoding transporters of oxidized sulphur species which are puta-

tively involved in sulphur excretion (Kelley et al 2016)

These results suggest that adaptation to H2S-rich environments

in members of the P mexicana species complex involves both an

increased ability to cope with elevated endogenous H2S and an

increased ability to keep endogenous H2S concentrations low

through detoxification and excretion However there is a noticeable

absence of evidence for genes and physiological pathways associ-

ated with endogenous H2S production and H2S-mediated cell sig-

nalling being involved in adaptation (see Kelley et al 2016

Pfenninger et al 2015) There are three nonmutually exclusive

hypotheses that could explain this finding First the ability of popu-

lations from sulphidic habitats to regulate endogenous H2S through

detoxification may be sufficient to maintain normal H2S-mediated

signalling Second compared to the ability for detoxification and for

mitigation of direct toxic effects in terms of ATP production and

oxidative stress H2Srsquos influence on cell signalling may have relatively

low fitness consequences with concomitantly smaller evolutionary

846 | TOBLER ET AL

responses Last inferences from biomedical studies about the causal

relationship between endogenously produced H2S and various physi-

ological functions may be weak and some documented physiological

effects may instead be an indirect consequence of the disruption of

homeostasis that cascades through interconnected pathways operat-

ing in a cell In this case evolutionary studies on natural systems

exposed to H2S may provide important insights in terms of prioritiz-

ing research on biomedically relevant aspects of H2S biology Study-

ing ancestral populations susceptible to H2S will allow for

quantifying consequences of disruptions in H2S homeostasis at all

levels of biological organization (like in presently available animal

models) Adding derived H2S tolerant populations will further allow

for the detection of critical modifications in physiological processes

affected by H2S Consequently such a comparative approach may

provide a powerful tool to disentangle direct and indirect physiologi-

cal consequences of deviations in H2S homeostasis which is cur-

rently difficult in traditional biomedical models

Despite some of these discrepancies the studies exploring mech-

anisms underlying H2S tolerance in sulphide spring fishes highlight

the utility of research in toxicology and biomedicine in generating

testable hypotheses about evolutionary processes Biomedical stud-

ies on metabolic disorders (Aspiras Rohner Martineau Borowsky amp

Clifford 2015) stress responses (Rohner et al 2013) responses to

low oxygen (Eales Hollinshead amp Tennant 2016) and effects of

xenobiotics (Stockinger Di Meglio Gialitakis amp Duarte 2014) have

also facilitated our understanding of adaptive evolution in other nat-

ural systems Considering the increasingly specialized and frag-

mented research areas in the life sciences interdisciplinary

approaches and reciprocal exchange of ideas and methodologies

continue to be critical for facilitating breakthroughs in basic and

applied research on biomedical models and natural systems (Carroll

et al 2014 Nesse amp Stearns 2008 Schartl 2014)

22 | Convergent and nonconvergent evolution inindependent lineages

Our discussion of putative candidate genes that are under selection

andor differentially expressed in sulphide spring populations so far

focused on general patterns documented across evolutionarily inde-

pendent lineages in different river drainages Selection scans (Pfen-

ninger et al 2015) and analyses of differential gene expression

(Kelley et al 2016) have indicated high congruence of biological

processes associated with candidate genes across lineages in differ-

ent river drainages indicating that evolution has repeatedly and pre-

dictably modified the same physiological pathways during adaptation

to environmental H2S However the signal of evolutionary conver-

gence among sulphide spring lineages decreases with the level of

organismal organization (from physiological pathways down to geno-

mic regions genes and individual SNPs in selection scans Pfenninger

et al 2015 from physiological pathways down to genes and tran-

scripts in expression studies Passow Brown et al 2017) indicating

that different lineages have often modified the same physiological

pathways in different ways While knowledge about the biochemical

TABLE 1 Adaptation to H2S can occur through three principlemechanisms (1) Resistance involves direct toxicity targets whosefunctions are impaired by H2S Adaptive modifications of thesetoxicity targets are predicted to assure proper function despitethe presence of H2S (2) Mitigation involves alternativephysiological pathways that compensate for the inhibition ofdirect toxicity targets Adaptive modifications of these alternativephysiological pathways are predicted to allow for the maintenanceof organismal function by other means (3) Regulation involvesphysiological pathways associated with the maintenance ofendogenous H2S homeostasis Adaptive modifications ofregulatory pathways are predicted to maintain low endogenousconcentrations despite continuous flux of H2S from theenvironment into the body

Candidate pathways Evidence

(1) Resistance

OXPHOS bull Positive selection on COX in some lineages of

P mexicana (Puyacatengo and Pichucalco)a as

well as some other genera of sulphide spring

poeciliidsb COX amino acid substitutions in P

mexicana allow for the maintenance of function

in the presence of H2Sa

bull Upregulation of cytochrome c and some COX

subunits as well as downregulation of compo-

nents of complex I in all sulphide spring lineages

of P mexicana indicating complex changes in

aerobic ATP productionc

Oxygen transport

genesbull Positive selection on and upregulation of

several oxygen transport genes (including myo-

globin and multiple haemoglobin subunits) in

sulphide spring lineages of the genera

Gambusia Limia Poecilia and Pseudoxiphophorus

indicating modifications to putatively maintain

oxygen transport in the presence of H2Sd

(2) Mitigation

Anaerobic

metabolismbull Upregulation of genes associated with glycolysis

and gluconeogenesis in all sulphide spring lin-

eages of P mexicana indicating increased

capacity of anaerobic ATP productionc

Oxidative stress

responsesbull Upregulation of genes associated with

oxidative stress responses in all sulphide

spring lineages of P mexicana indicating

increased capacity to cope with H2S-induced

oxidative stressc

(3) Regulation

H2S detoxification bull Positive selection on and upregulation of sev-

eral genes associated with enzymatic H2S

detoxification in all sulphide spring lineages of

P mexicanace Upregulation of H2S detoxifica-

tion enzymes in other sulphide spring lineages

of the genera Gambusia Limia Poecilia Pseudox-

iphophorus and Xiphophorusb This indicates an

increased ability to maintain low endogenous

H2S concentrations during exposure

H2S production bull No evidence for modifications of genes associ-

ated with enzymatic production of H2Sc

aPfenninger et al (2014) bR Greenway J L Kelley amp M Tobler unpub-

lished data cKelley et al (2016) dBarts et al (2018) ePfenninger et al

(2015)

TOBLER ET AL | 847

and physiological consequences of H2S may thus enable us to pre-

dict the physiological pathways in which adaptive modifications

occur it may be impossible to predict specific modifications within

those pathways This contrasts with a common expectation in the

field that convergent evolution of molecular mechanisms underlying

phenotypic traits (ie parallel evolution) necessarily hinges on modi-

fication of the same loci across lineages as found in some of the

best-known examples of parallelism (Colosimo et al 2005 Feldman

Brodie Brodie amp Pfrender 2012 Grant Grant Markert Keller amp

Petren 2004) However modification of the same loci may only be

common under particular circumstances (Rosenblum Parent amp

Brandt 2014) for example when selection acts on standing genetic

variation (Colosimo et al 2005) when adaptive alleles are intro-

gressed across population boundaries through gene flow (Grant

et al 2004) or when there are strong genetic constraints (Feldman

et al 2012) Exploration of molecular mechanisms underlying con-

vergent evolution in many systems may therefore be better

approached from the perspective of interacting genes that are orga-

nized in pathways in which alternative genetic changes can have

equivalent functional consequences (ie functional redundancy)

Developing and testing hypotheses about parallel evolution at the

pathway level will hinge on improving genome annotations (Pavey

Bernatchez Aubin-Horth amp Landry 2012) and provide a more

nuanced understanding of the factors determining the predictability

of evolution at different levels of organismal organization (Rosen-

blum et al 2014)

Despite evidence for convergent evolution in some candidate

physiological pathways it also needs to be emphasized that this

does not apply to all P mexicana lineages or to all candidate physio-

logical pathways associated with H2S adaptation investigated to

date Nonconvergent evolutionmdashevidenced by unique lineage-speci-

fic signals of selection and differential expressionmdashmay be indicative

of unique molecular mechanisms to cope with H2S toxicity For

example signatures of positive selection on COXmdashH2Srsquos primary

toxicity targetmdashcan be found in sulphide spring lineages of the Puy-

acatengo and Pichucalco river drainages (a lineage with recent diver-

gence time and a lineage with older divergence time respectively)

but not in those of the Tacotalpa drainage (Pfenninger et al 2014)

The presence of nonconvergent genetic and gene expression differ-

ences between populations from sulphidic and non-sulphidic habitats

in different river drainages highlights the limitations of inferences

that can be drawn based on data sets generated with high-through-

put DNA and RNA sequencing The potential importance of modifi-

cations at candidate locimdashor the lack thereofmdashcan only be assessed

through functional assays that document how alternative alleles and

changes in gene expression actually affect organismal performance

The functional consequences of genetic and gene expression varia-

tion at most candidate loci remain to be investigated However

enzyme activity assays have indicated that COX maintains its activity

with increasing H2S concentrations in the two sulphide spring popu-

lations with evidence for positive selection on mitochondrially

encoded subunits (Puyacatengo and Pichucalco Pfenninger et al

2014) Key amino acid substitutions occur in the D-pathway channel

of COX1 which putatively narrow the channel width and reduce

H2Srsquos access to the binuclear centre of the protein Consequently

some lineages of P mexicana exhibit an H2S-resistant COX that can

maintain its activity and aerobic ATP production even if endogenous

H2S concentrations are elevated In contrast COX activity in sul-

phide spring fish of the Tacotalpa river drainage and in fish from

nonsulphidic habitats decreases when H2S is present which is con-

sistent with mechanisms of H2S toxicity (Pfenninger et al 2014) In

vitro assays on isolated mitochondria have nonetheless indicated

that an H2S-resistant COX is not necessary to maintain aerobic ATP

production during H2S exposure Unlike in populations from nonsul-

phidic habitats mitochondrial spare respiratory capacity is main-

tained in the presence of H2S even in the Tacotalpa sulphide spring

populations that lack an H2S-resistant COX (C Henpita M Tobler amp

JH Shaw unpublished data) This may indicate that modification of

detoxification pathways in sulphide spring fishes is sufficient to

maintain mitochondrial function and aerobic ATP production

23 | Ensuing challenges linking genotypes tofitness

Understanding the mechanistic basis of adaptation requires linking

genotypic variation to phenotypic traits that ultimately shape fitness

along selective gradients The evolution of H2S tolerance in sulphide

spring fishes appears to be a consequence of modifications in many

genes and many physiological pathways whose functions interact to

eliminate H2S from the system minimize effects of elevated endoge-

nous concentrations on direct toxicity targets and mitigate adverse

effects that cascade through cellular metabolism during disruption of

homeostasis (Table 1) Such complex genetic changes are likely com-

mon during adaptation and have been documented in other systems

that involve major evolutionary shifts including the emergence of

parasitic lifestyles (Werren et al 2010) eusociality (Simola et al

2013) asexual reproduction (Fradin et al 2017) and internal gesta-

tion with livebearing (Schartl et al 2013) Consequently comple-

mentary approaches will be required to uncover the molecular

mechanisms of adaptation to H2S in the future

1 The involvement of many genes and physiological pathways in

shaping tolerance emphasizes the importance of quantitative

genetic approaches that leverage experimental crosses between

H2S tolerant and nontolerant populations and test whether toler-

ancemdashand adaptive trait modifications that contribute to itmdashis

explained by a few major effects loci or whether it is a conse-

quence of a number of genes with small effects (Klerks Xie amp

Levinton 2011 Saltz Hessel amp Kelly 2017) Combining analyses

of phenotype-fitness associations and their underlying genetic

mechanisms via QTL mapping with the results of selection scans

from population genomic analyses (Pfenninger et al 2015) will

allow for an integrative understanding of how selection acting on

the genome results in phenotypic variation that mediates adapta-

tion to different environments (Stinchcombe amp Hoekstra 2008)

From a functional perspective breaking up correlations between

848 | TOBLER ET AL

different physiological traits (associated with resistance mitiga-

tion and regulation Table 1) in F2 crosses will further allow test-

ing what molecular mechanisms are necessary for tolerance

whether there are alternative strategies with equivalent fitness

benefits and whether there are complementary modifications

with additive effects on tolerance

2 Considering the number of genes and physiological pathways

likely involved in H2S tolerance assessing the functional conse-

quences of genetic and gene expression variation is a daunting

task At the same time such functional tests can significantly

contribute to closing the genotype-to-phenotype gap because

physiochemical stress responses can be assessed along the con-

tinuum of organismal organization (Gjuvsland Vik Beard Hunter

amp Omholt 2013) Standardized assays can be used to quantify

the biochemical makeup of organelles cells and tissues (eg pro-

tein abundance and enzyme activity) physiological performance

in vitro and in vivo (eg rates of H2S detoxification ATP produc-

tion and reactive oxygen production) and fitness-relevant proxies

of whole organism performance (eg H2S tolerance Figure 2)

This highlights the importance of methods development for the

quantification of complex phenotypic traits or even whole phe-

nomes that can keep pace with rapid data generation through

high-throughput sequencing (Houle Govindaraju amp Omholt

2010 Ritchie Holzinger Li Pendergrass amp Kim 2015) In addi-

tion genotypendashphenotype links established through such correla-

tive approaches will ideally be verified through targeted gene

editing While gene editing with CRISPRndashCas9 is now routinely

employed in model organisms (Doudna amp Charpentier 2014 Sha-

lem Sanjana amp Zhang 2015) applying this technology to non-

models still remains challenging especially in livebearing species

like poeciliid fishes where in vitro culture of embryos is not rou-

tinely employed (Martyn Weigel amp Dreyer 2006) Nonetheless

the effects of specific mutations can potentially be verified

through the establishment and manipulation of cells in culture

(Liu et al 2014) or by generating transgenics (Lin et al 2016)

3 Organismal responses to physiochemical stressors occur at multi-

ple hierarchical levels that can be plastically modulated in

response to endogenous and exogenous cues (Whitehead 2012)

Understanding adaptation to physiochemical stressors conse-

quently requires disentangling the effects of genetic variation

developmental plasticity and short-term acclimation in gene

expression patterns on the emerging physiological processes

Common garden experiments on sulphide spring fishes have

shown that many genes differentially expressed between popula-

tions from sulphidic and nonsulphidic habitats in nature remain

differentially expressed in laboratory-reared fish that have never

been exposed to H2S indicating evolved differences in constitu-

tive gene expression (Passow Henpita et al 2017) At the same

time expression patterns in the laboratory also exhibit evidence

for plasticity in response to short-term H2S exposure (Passow

Henpita et al 2017) Some of these plastic responses are shared

between populations from sulphidic and nonsulphidic habitats

(likely representing shared ancestral plasticity) while others have

been lost or gained during sulphide spring colonization A core

challenge for future research is to overcome logistical constraints

that have so far prevented long-term H2S exposures of fish in

the laboratory Exposing individuals throughout development will

be required to test how developmental plasticity and short-term

acclimation shape responses to H2S in individuals from adapted

and nonadapted populations Finally the next big scientific

advance in this context will also require the identification of reg-

ulatory mechanisms that shape expression differences during

adaptation including sequence changes in transcription factors

and their binding sites patterns of DNA methylation and chro-

matin accessibility as well as the production of microRNAs and

competing endogenous RNAs that can modulate transcript abun-

dance (Flores Wolschin amp Amdam 2013 Lopez-Maury Mar-

guerat amp Beuroahler 2008 Silva Bye el Azzouzi amp Wisloslashff 2017)

In addition the role of alternative splicing and other post-tran-

scriptional and post-translational modifications in mediating adap-

tation remains understudied by evolutionary biologists (Ast

2004 Bush Chen Tovar-Corona amp Urritia 2017) This is particu-

larly relevant because H2S has been hypothesized to play a role

in the post-translational modification of proteins with sulphur-

containing amino acids through sulfhydration (Mustafa et al

2009)

Bridging the genotypendashphenotype gap remains one of the major

challenges in biology (Gjuvsland et al 2013 Ritchie et al 2015) and

will provide opportunities to understand how organisms function in

the face of environmental stress and how evolution shapes these

functions during adaptation to different environmental conditions

Establishing the molecular mechanisms of adaptation across repli-

cated lineages exposed to the same sources of selection will also

provide insights into how evolutionary history demography and

genetics shape convergent evolutionary outcomes providing a better

understanding at what hierarchical levels of organismal organization

evolutionary change may actually be predictable (Rosenblum et al

2014)

3 | COMPLEX ORGANISMS IN A COMPLEXWORLD DISENTANGLING THE TANGLEDBANK

Mechanisms of adaptation can be studied through bottom-up

(screening for genetic modifications that allow for inferences about

relevant phenotypic traits) and top-down approaches (assessing fit-

ness-relevant phenotypic differences among populations without

knowledge about the genetic basis see Schluter 2009) Over the

years genomic and transcriptomic studies on sulphide spring popula-

tions of the P mexicana species complex have been complemented

with analyses of phenotypic variation Just as genetic analyses have

indicated that adaptation to H2S is mediated by the modification

and modulation of many genes associated with multiple physiological

pathways phenotypic studies uncovered a variety of differences

TOBLER ET AL | 849

between populations from sulphidic and nonsulphidic environments

Many of these phenotypic differences have evolved in convergence

across replicated sulphide spring lineages in different river drainages

Examples include modifications of morphological (body shape and

organ size Tobler et al 2011 Tobler Scharnweber et al 2015

Schulz-Mirbach et al 2016) physiological (tolerance and metabolic

rates Tobler et al 2011 Passow Arias-Rodriguez amp Tobler 2017)

behavioural (aggression and personality Riesch et al 2009 Bierbach

et al 2012 Sommer-Trembo et al 2016 Bierbach Arias Rodriguez

amp Plath 2018) and life history traits (fecundity and offspring size

Riesch Plath Garcia de Leon amp Schlupp 2010 Riesch et al 2014)

Hence proximate populations in different habitat types have

diverged in complex phenotypes that span many different suites of

traits and a key question is how divergence in multivariate pheno-

types arises In theory H2S could be a key source of selection driv-

ing multivariate phenotypic divergence either because it directly

exerts selection on a multitude of traits that are independently modi-

fied to reach new fitness peaks or because selection on a few fit-

ness-relevant traits causes simultaneous changes in correlated

characteristics Neither explanation seems sufficient because many

Abiotic environmentOxygen concentrations

Ionic compositionTemperature

Structure and flow

Biotic environmentTrophic resources

DemographyPredationParasitism

H2S

Genetic variation

Gene expression

Biochemistry

Physiologicalfunction

Organismal performance

Primary fitness components

Survivalgrowth

reproduction

Organismal function Environment

F IGURE 2 Understanding how organisms function in the context of their environment is a core question in evolutionary biology yet thecomplexities associated with organismal function and environmental variation provide significant challenges High-throughput sequencing hasallowed for the identification of candidate genes subject to selection and differential expression during adaptation to H2S The next step is tobridge the genotype-to-phenotype gap by quantifying the integrated responses to H2S across the continuum of organismal organization includingbiochemical phenotypes physiological functions whole organism performance and ultimately fitness In addition understanding adaptation toH2S-rich environments will also require acknowledging H2Srsquos relation to other abiotic and biotic environmental variables which can exert selectionand modulate aspects of organismal function This will allow for the establishment of cause-and-effect relationships between specific sources ofselection and evolutionary change in specific traits and contribute to our understanding of organismal function in natural systems

850 | TOBLER ET AL

of the divergent traits between populations from sulphidic and non-

sulphidic environments have no clearmdashor at best tenuousmdashlinks to

H2S toxicity and detoxification and there is only limited evidence

for within-population correlations between different traits groups

(Riesch et al 2016)

Evolutionary analyses frequently focus on seemingly simple

sources of selection juxtaposing populations that are assigned to dis-

crete habitat types to understand the evolution of specific traits

(Kaeuffer Peichel Bolnick amp Hendry 2012) But as Darwin (1859)

already observed in his ldquotangled bankrdquo allegory selection in natural

systems is rarely univariate but multifarious selective regimes con-

tribute to trait divergence even among populations that occur along

apparently simple environmental gradients Unquantified environmen-

tal variation that is subsumed in the categorization of habitat types

can profoundly affect trait variation in natural systems (Stuart et al

2017) highlighting the importance of considering and testing alterna-

tive hypotheses about the cause-and-effect relationship between

specific sources of selection and evolution of specific traits Thus

accurate natural history observations that consider organisms as holis-

tic entities consisting of interacting traits in the context of the abiotic

and biotic factors comprising their environment are critical to inform

the design and interpretation of empirical research in natural systems

(Greene 2005) Indeed adjacent sulphidic and nonsulphidic habitat

types do not merely differ in the presence or absence of H2S but also

in a variety of other abiotic and biotic factors (Figure 2) Unlike nonsul-

phidic habitats sulphide springs in southern Mexico are characterized

by extreme hypoxia elevated salinity and lower pH (Rosales Lagarde

2012 Tobler et al 2011) In addition the presence of physiochemical

stressors affects the biotic communities in these sulphide springs Spe-

cies richness of aquatic invertebrates and vertebrates is drastically

reduced as a consequence of H2S toxicity (Greenway Arias-Rodriguez

Diaz amp Tobler 2014) and primary production appears to be driven in

part by chemoautotrophic bacteria that use H2S as an energy source

rather than just by autochthonous and allochthonous photosynthetic

organisms (Roach Tobler amp Winemiller 2011) These changes in the

biotic environment affect the relative importance of intra-vs inter-

specific competition trophic resource use and exposure to predators

and parasites (Riesch Oranth et al 2010 Tobler et al 2014 Tobler

Scharnweber et al 2015)

31 | Alternative drivers of trait evolution

Acknowledging the complexity of selective regimes in sulphide

springs allows for a more nuanced understanding of trait evolution

and for the development of alternative hypotheses about the ulti-

mate mechanisms that cause divergence in specific genetic and phe-

notypic traits In some instances sources of selection other than H2S

can plausibly explain trait divergence between populations from sul-

phidic and nonsulphidic habitats For example hypoxia likely drives

increases in head size and gill surface area (Tobler et al 2011)

changes in the ionic composition of the water can explain differential

expression of genes associated with osmoregulation (Kelley et al

2016) and shifts in trophic resource use are associated with

modifications of the gastrointestinal system (Tobler Scharnweber

et al 2015) In other cases disentangling alternative hypotheses is

not trivial because selection mediated by H2S and other environmen-

tal factors are predicted to affect trait modifications in similar ways

Shifts towards anaerobic metabolism in sulphide spring fish could be

the consequence of inhibition of aerobic ATP production by H2S or

by constraints in oxygen availability Similarly reproductive life his-

tory shifts towards the production of larger but fewer offspring could

be driven by H2S (decreasing offspringrsquos surfacevolume ratio to mini-

mize the diffusion of H2S into the body Powell 1989) by habitat dif-

ferences in the quality and quantity of predation that can affect size-

specific rates of extrinsic mortality (Jorgensen Auer amp Reznick

2011) or by energetic constraints (see below) While comparisons

across different selective regimes can shed light into the ultimate dri-

vers of trait evolution (Moore Riesch amp Martin 2016) isolating the

effects of different sources of selection may not be possible because

they could act in concert and facilitate increases of organismal perfor-

mance in multiple contexts (Ghalambor Walker amp Reznick 2003)

When multifarious selection shapes multivariate phenotypic evo-

lution it is important to also consider the role of trade-offs both in

terms of trait expression (increased investment in one trait may

impose constraints on the expression of other traits) and in terms of

trait function (optimizing the trait for one function may impair the

traitrsquos utility for other functions) For example selection on large off-

spring size in sulphide spring fishes has inadvertent consequences

for fecundity (Riesch et al 2014) and selection for increased gill

surface areas due to hypoxia simultaneously increases exposure to

environmental H2S (Tobler et al 2011) Thus the outcome of phe-

notypic evolution is ultimately a result of balancing competing needs

in a manner that optimizes trait expression and organismal function

in multiple contexts to maximize fitness (Arnold 1983) and the pres-

ence of trade-offs can have cascading indirect effects with selection

on one or a few traits leading to the modification of a multitude of

apparently unrelated characteristics

In this context bioenergetics may represent a nexus in the evo-

lution of complex traits observed in sulphide spring fish and other

extreme environments Living in and adapting to sulphide springs

imposes constraints on organismal energy availability (Plath et al

2007) with according trade-offs for energy allocation Energy limita-

tion in sulphide springs does not arise from a scarcity of food

resources (chemoautotrophic bacteria eaten by the fish are abundant

in sulphide springs Tobler Scharnweber et al 2015) but from

behavioural and physiological adaptation A primary survival strategy

for fish in the toxic and hypoxic springs is to engage in aquatic sur-

face respiration where individuals breathe directly at the water sur-

face to maximize oxygen uptake and minimize H2S exposure This

behaviour mediates short-term survival (Plath et al 2007) but it

also constrains energy acquisition for benthic foragers like P mexi-

cana (Tobler Riesch Tobler amp Plath 2009) and increases susceptibil-

ity to bird predation (Riesch Oranth et al 2010) Costly investments

into H2S detoxification and tissue repair as well as a shift from aero-

bic to less efficient anaerobic metabolism further constrain energy

allocation to other functions (Passow Greenway Arias-Rodriguez

TOBLER ET AL | 851

Jeyasingh amp Tobler 2015) Theory predicts that energetic con-

straints ndash whether mediated by limitations in the rate of resource

acquisition or in the rate of energy allocation to reproduction (as

opposed to maintenance or growth) ndash should exert selection for a

reduction in organismal energy budgets which allows for the maxi-

mization of residual energy allocation for the production of offspring

(Brown Marquet amp Taper 1993) Several lines of evidence support

the hypothesis that energy limitation plays a critical role in pheno-

typic evolution of sulphide spring fish Compared to individuals in

nonsulphidic environments fish in sulphide springs have significantly

lower routine energy demands which is mediated both by reduc-

tions in body size and mass-specific metabolism (Passow Arias-

Rodriguez et al 2017) Reductions in other traits such as energeti-

cally costly behaviours (Bierbach et al 2012 Plath 2008) and the

size of energy-demanding organs (Schulz-Mirbach et al 2016) may

also contribute to the reduced energy demands and shifts in repro-

ductive life history traits (Riesch et al 2014) are consistent with

selection from energy limitation (Reznick amp Yang 1993)

32 | Ensuing challenges Disentanglingenvironmental and organismal complexities

Future research needs to empirically address how multifarious selec-

tive regimes shape the evolution of organisms with complex pheno-

types We have made great advances in our understanding of how

specific traits evolve and mediate adaptation (eg morphology Chan

et al 2010 coloration Dasmahapatra et al 2012 sensory organs

Cartwright Schwartz Merry amp Howell 2017) and we have a thor-

ough understanding of how single environmental factors can shape

organismal responses (eg climatic conditions Johnston amp Bennett

1996 predation Langerhans 2007 competition Pfennig amp Pfennig

2012) In contrast our understanding of phenotypic evolution in

response to multifarious selection is comparatively limited especially

because interactive effects of different environmental factors can lead

to organismal and evolutionary responses that are unpredictable based

on experiments that manipulate single environmental variables in iso-

lation (Gunderson Armstrong amp Stillman 2016 Todgham amp Stillman

2013) Future research thus needs to acknowledge the complexities of

environments and organismal design and address them explicitly in

empirical investigations to gain a robust understanding of biological

diversification (Culumber amp Tobler 2017 Stuart et al 2017)

4 | LINKING ADAPTATION TOSPECIATION

So far we primarily explored the causes of adaptive evolution but

what are its consequences Theory predicts that adaptation to local

environmental conditions can lead to speciation because reproduc-

tive isolation emerges incidentally as a by-product (ecological specia-

tion Nosil 2012) Population genetic analyses of Poecilia spp in

sulphide springs and adjacent populations in nonsulphidic environ-

ments have indicated strong genetic differentiation across small

spatial scales (in some instances lt100 m) and in the absence of

physical barriers that would prevent fish movement (Plath et al

2013) In addition population genetic structuring has been shown to

be stable even in the face of major environmental disturbance

BOX 1 Speciation in sulphide springs and conservation

biology

Adaptation to H2S-rich environments in poeciliids coincides

with ecological speciation giving rise to phenotypically dis-

tinct and reproductively isolated lineages In some

instances this has led to the evolution of highly distinct

micro-endemic species including Poecilia sulphuraria and

P thermalis Limia sulphurophila and Gambusia eurystoma

(Figure 3) Except for P sulphuraria which is found in two

spring complexes of the Rıo Pichucalco drainage of Mexico

(Tobler amp Plath 2009b) the distributions of these sulphide

spring endemics are restricted to single localities that

include only small areas of suitable habitat For example

the largest sulphide spring complex (the Ba~nos del Azufre

in southern Mexico which harbours P sulphuraria and G

eurystoma) includes ~18000 m2 of suitable habitat (equiva-

lent to lt35 football fields) while one of the smallest (La

Esperanza springs harbouring P thermalis) merely covers

~600 m2 (just over a tenth of a football field Brown et al

2017) Accordingly some sulphide spring endemics are

considered critically endangered (IUCN 2012) and pro-

tected by law (SEDESOL 1994)

Many threatened species with restricted ranges exhibit

low genetic diversity and high incidences of inbreeding

(Bouzat 2010 Willi Van Buskirk amp Hoffmann 2006) In

contrast at least P sulphuraria and P thermalismdashthe only

species investigated to datemdashactually exhibit relatively high

effective population sizes (Ne = 1200ndash3800 depending on

the population) and no evidence for inbreeding (FIS lt 0)

(Brown et al 2017) This is likely a consequence of the

long-term stability of sulphide springs and the high pro-

ductivity and low levels of interspecific competition also

support unusually high fish densities and thus population

census sizes (Brown et al 2017 Culumber et al 2016)

Nonetheless sulphide spring endemics are threatened by

habitat destruction which is mediated primarily by the

modification of springs for recreational purposes defor-

estation and conversion of land for agriculture and most

recently high-intensity palm oil plantations (Tobler amp Plath

2009ab) Despite the official protection of sulphide spring

endemics there have been no tangible conservation mea-

sures that foster the sustainable development of land

around sulphide springs with the well-being of the endemic

fishes in mind

852 | TOBLER ET AL

(catastrophic flood event Plath et al 2010) Population genetic dif-

ferentiation between adjacent sulphide-adapted and nonadapted

populations is indicative of ongoing ecological speciation (Box 1) but

different population pairs from sulphidic and nonsulphidic waters

have made variable progress along the speciation continuum towards

complete reproductive isolation (Plath et al 2013) This prompts

questions about the mechanisms involved in reducing gene flow

along the environmental gradient connecting the two habitat types

41 | Prezygotic mechanisms of reproductiveisolation

The simplest form of reproductive isolation between locally adapted

populations arises from selection against migrants between habitat

types (Nosil Vines amp Funk 2005) Translocation experiments in nat-

ural habitats have indicated strong survival selection against

migrants especially those translocated from nonsulphidic to sulphidic

habitats (Plath et al 2013) Fish translocated from nonsulphidic

environments face high mortality within short periods of time

(lt24 hr) which is likely caused by H2S toxicity and hypoxia in sul-

phide springs Hence low gene flow from populations inhabiting

nonsulphidic waters into populations inhabiting sulphidic habitats is

simply explained by a lack of adaptation to the extreme environmen-

tal conditions At least in some drainages there is also appreciable

mortality for sulphide spring fish translocated to nonsulphidic habi-

tats although mechanisms causing inviability are unclear and the

strength of selection appears to be lower (Plath et al 2013) If sur-

vival selection against immigrants was the sole source of reproduc-

tive isolation this should result in unidirectional gene flow from

H2S-adapted to nonadapted populations but population genetic

analyses provided no evidence for this implying that additional

mechanisms are at work (Plath et al 2013)

One such mechanism could be intersexual selection and mate

choice experiments have indicated that migrants from sulphidic to

nonsulphidic habitats are at a significant disadvantage with fish from

nonsulphidic habitats avoiding potential mating partners from sul-

phidic ones (Plath et al 2013 Sommer-Trembo et al 2016) Such

assortative mating has been found to be critical during ecological spe-

ciation because it counteracts the homogenizing effects of gene flow

(van Doorn Edelaar amp Weissing 2009) yet how assortative mating

patterns arise often remains unclear In theory assortative mating can

arise through a variety of processes (Nosil 2012) ldquoMagic traitrdquo mecha-

nisms where mating preferences evolve as a by-product of divergent

selection on ecological traits that serve as cues have recently garnered

much attention (Servedio van Doorn Kopp Frame amp Nosil 2011)

Consistent with magic trait mechanisms assortative mating in P mexi-

cana appears to be linked to adaptive population differences in body

shape which serve as cues to distinguish potential mating partners

from different habitats (Greenway Drexler Arias-Rodriguez amp Tobler

2016) However neither sulphide spring fish (which are unlikely to

encounter migrants from nonsulphidic habitats due to H2S toxicity)

nor allopatric P mexicana from nonsulphidic waters (which have never

experienced migrants from sulphide spring populations) exhibit a

preference for mating partners from their own ecotype suggesting

that assortative mating in this system is a product of reinforcement

(ie direct selection for assortative mating Greenway et al 2016)

This is different from classic magic traits mechanisms because assorta-

tive mating is not a mere by-product of adaptation but adaptive traits

have instead been co-opted as mating signals in response to selection

on mating preferences (Maan amp Seehausen 2012) Moreover males of

some H2S-adapted populations have reduced aggressive behaviour

(Bierbach et al 2012 2018) In staged contests of size-matched males

from different ecotypes intruders had a significantly lower likelihood

to establish dominance than resident males independent of whether

or not they showed reduced aggressiveness overall Hence intrasexual

selection may represent another mechanism mediating reproductive

isolation between locally adapted ecotypes Overall H2S-adapted poe-

ciliids allow disentangling the relative contributions of different forms

of ecological and sexual selectionmdashand their interdependencymdashduring

ecological speciation

42 | Ensuing challenges quantifying postzygoticmechanisms of reproductive isolation

In contrast to prezygotic mechanisms little is known about postzy-

gotic mechanisms of reproductive isolation that potentially act to

reduce gene flow between adjacent populations in sulphidic and

nonsulphidic habitats Putative hybrids between sulphidic and non-

sulphidic ecotypes have occasionally been observed in nonsulphidic

habitats immediately adjacent to confluences with sulphide springs

(R Riesch M Plath and M Tobler personal observations) but it

remains unclear whether and through what mechanisms such hybrids

exhibit reduced fitness compared to parentals

Direct selection of mitochondrially encoded genes in the pres-

ence of H2S (Pfenninger et al 2014) should be accompanied by co-

evolution at nuclear loci that directly interact with mitochondrial

gene products in the formation of protein complexes involved in the

OXPHOS pathway (Sunnucks Morales Lamb Pavlova amp Greening

2017 Willett amp Burton 2004) Disruption of co-evolved gene com-

plexes in hybrids should therefore lead to intrinsic fitness reductions

through mito-nuclear incompatibilities (Burton Pereira amp Barreto

2013 Wolff Ladoukakis Enrıquez amp Dowling 2014) which have

been hypothesized to be a key driver in speciation (Chou amp Leu

2010 Gershoni Templeton amp Mishmar 2009) Empirical tests of

these ideas remain to be conducted in sulphide spring fishes just

like studies on potential ecological and sexual selection against

hybrids that arise from interpopulation mating events

5 | PERSPECTIVES LIFE IN SULPHIDICENVIRONMENTS BEYOND P MEXICANA

Research on sulphide spring fishes of the P mexicana species com-

plex in Mexico has provided insights into mechanisms underlying

adaptation and the links between adaptation and speciation Sul-

phide springs do not only occur in Mexico however but are found

TOBLER ET AL | 853

worldwide and have been colonized by 24 evolutionarily indepen-

dent fish lineages in North America the Neotropics and the Middle

East (Greenway et al 2014) The vast majority of lineages belong to

the order Cyprinodontiformes (95) and livebearing members of the

family Poeciliidae in particular (80) This potentially indicates that

internal gestation of young has facilitated the colonization of

extreme environments because it shields developing embryos from

physiochemical stressors (Blackburn 1999) Among the Poeciliidae

sulphide spring fishes include members of the genera Brachyrhaphis

Gambusia Poeciliopsis Priapichthys Pseudoxiphophorus Xiphophorus

and multiple subgenera of Poecilia (Figure 3 Greenway et al 2014)

spanning over 40 million years of evolution (based on Reznick Fur-

ness Meredith amp Springer 2017)

A handful of studies have started to address molecular and phe-

notypic evolution across a broader taxon sampling of sulphide spring

fishes finding evidence for convergent evolution in morphology

(Riesch et al 2016 Tobler amp Hastings 2011) life history traits

(Riesch Plath et al 2010 Riesch et al 2014) and oxygen transport

genes (Barts et al 2018) In addition ecological speciation in

response to adaptation to H2S-rich environments has also been doc-

umented in sulphide spring poeciliids other than P mexicana (Riesch

et al 2016) These studies have laid a fruitful foundation for the

investigation of mechanisms of adaptation to H2S-rich springs from

genes to phenotypic traits linking micro-evolutionary processes to

macro-evolutionary patterns

The availability of broad-scale evolutionary replication of popula-

tions that have invaded habitats with the same physiochemical stressors

allows for addressing some of the ensuing challenges discussed above

and investigating the factors that give rise to predictable evolutionary

outcomes Convergent evolution is a commonmdashbut not omnipresentmdash

theme in adaptive evolution and multiple factors can determine the

degree of convergence among replicated lineages exposed to the same

sources of selection (Kaeuffer et al 2012 Landry amp Bernatchez 2010

Oke Rolshausen LeBlond amp Hendry 2017 Rosenblum amp Harmon

2011) Thus leveraging replicated lineages that have colonized sulphide

springs to assess molecular mechanisms underlying convergent pheno-

typic changes analyse how environmental heterogeneity among sul-

phide spring habitats causes lineage-specific patterns of trait evolution

and test whether and by what mechanisms adaptation to extreme envi-

ronments is linked to speciation will provide insights into how ecologi-

cal genetic and functional factors contribute to evolutionary

convergence at different levels of organismal organization (Rosenblum

et al 2014) An expanded taxon sampling with population pairs that

span almost the full range of the speciation continuum from panmixia to

complete reproductive isolation (Riesch et al 2016) will also help

addressing questions about temporal aspects of H2S adaptation In par-

ticular comparing younger to older lineages will provide insights about

the order at which genetic and phenotypic modifications accrue during

the adaptation process

Animals have also invaded H2S-rich habitats in other aquatic

ecosystems including cold seeps and deep-sea hydrothermal vents

(Tobler et al 2016) Adaptation to H2S in the P mexicana species com-

plex has involved modification of physiological pathways associated

with H2S toxicity (OXPHOS pathway) and detoxification (sulphide

F IGURE 3 Phylogeny of the family Poeciliidae (adopted from Culumber amp Tobler 2017) Various lineages in the genera Poecilia (subgeneraAcanthophacelus Limia and Mollienesia) Priapichthys Brachyrhaphis Poeciliopsis Xiphophorus Pseudoxiphophorus and Gambusia have colonizedH2S-rich springs in different parts of the Americas (see Greenway et al 2014) Lineages highlighted in yellow represent endemic speciesrestricted to H2S-rich springs while red lineages represent species in which some populations have invaded sulphidic habitats Pictures depict arepresentative of each focal genussubgenus

854 | TOBLER ET AL

quinone oxidoreductase pathway) which have been highly conserved

during the diversification of life on Earth (Saraste 1999 Shahak amp

Hauska 2008) The high replicability of H2S as an environmental factor

that shaped animal evolution across vastly different ecological contexts

consequently offers the opportunity to investigate whether selection

mediated by the presence of H2S has repeatedly modified these con-

served physiological pathways across animal phyla that have diverged

during the Cambrian explosion spanning over 500 million years of evo-

lution Such comparisons would provide unprecedented insights into

the effects of a single physiochemical stressor on evolutionary change

irrespective of ecological or phylogenetic contexts

6 | CONCLUSIONS

H2S-rich environments and their inhabitants represent useful models

for research in organismal biology ecology and evolution illustrating

both opportunities (eg strong and explicit source of selection evolu-

tionary replication) and challenges (eg complexity of environments

and organismal responses) in asking basic question about the origins

of biodiversity Understanding how organisms thrive in apparently

inhospitable environments will provide us with insights into lifersquos

capacities and limitations to adapt to novel environments (Waterman

1999) This is particularly relevant considering that human activities

substantially contribute to the exacerbation of physiochemical stres-

sors at local regional and global scales impacting ecological and evo-

lutionary dynamics (Alberti 2015 Norberg Urban Vellend

Klausmeier amp Loeuille 2012) Thus a focus on ecological and evolu-

tionary consequences of natural stressors that push organisms to the

brink of their physiological capacities may be instrumental to predict

consequences of anthropogenic environmental change

ACKNOWLEDGEMENTS

We would like to thank Lenin Arias-Rodriguez the Universidad

Juarez Autonoma de Tabasco and the Centro de Investigacion E

Innovacion Para la Ense~nanza y El Aprendizaje for their continued

collaboration enabling our work on the P mexicana system Funding

for this review article has been provided by the National Science

Foundation (IOS-1557860 IOS-1557795) and the US Army

Research Office (W911NF-15-1-0175)

AUTHOR CONTRIBUTIONS

MT wrote the first draft of the paper all authors contributed text

and comments

ORCID

Michael Tobler httporcidorg0000-0002-0326-0890

Joanna L Kelley httporcidorg0000-0002-7731-605X

Reuroudiger Riesch httporcidorg0000-0002-0223-1254

REFERENCES

Alberti M (2015) Eco-evolutionary dynamics in an urbanizing planet

Trends in Ecology amp Evolution 30 114ndash126 httpsdoiorg101016j

tree201411007

Alda F Reina R G Doadrio I amp Bermingham E (2013) Phylogeny

and biogeography of the Poecilia sphenops species complex

(Actinopterygii Poeciliidae) in Central America Molecular Phylogenet-

ics and Evolution 66 1011ndash1026 httpsdoiorg101016jympev

201212012

Alvarez del Villar J (1948) Descripcion de una nueva especie de Mol-

lienisia capturada en Ba~nos del Azufre Tabasco (Pisces Poeciliidae)

Anales de la Escuela Nacional de Ciencias Biologicas 5 275ndash281

Arnold S J (1983) Morphology performance and fitness American Zool-

ogist 23 347ndash361 httpsdoiorg101093icb232347

Aspiras A C Rohner N Martineau B Borowsky R L amp Clifford C J

(2015) Melanocortin 4 receptor mutations contribute to the adapta-

tion of cavefish to nutrient-poor conditions Proceedings of the

National Academy of Sciences USA 112 9968ndash9973

Ast G (2004) How did alternative splicing evolve Nature Reviews

Genetics 5 773ndash782 httpsdoiorg101038nrg1451

Bagarinao T (1992) Sulfide as an environmental factor and toxicant

Tolerance and adaptations in aquatic organisms Aquatic Toxicology

24 21ndash62 httpsdoiorg1010160166-445X(92)90015-F

Barton N H (1998) The effect of hitch-hiking on neutral genealogies

Genetic Research 72 123ndash133 httpsdoiorg101017

S0016672398003462

Barts N Greenway R Passow C N Arias Rodriguez L Kelley J L amp

Tobler M (2018) Expression and molecular evolution of oxygen

transport genes in livebearing fishes (Poeciliidae) from hydrogen sul-

fide rich springs Genome httpsdoiorg101139gen-2017-0051

[Epub ahead of print]

Beauchamp R O Bus J S Popp C S Boreiko C J amp Andjelkovich

D A (1984) A critical review of the literature on hydrogen sulfide

CRC Critical Reviews in Toxicology 13 25ndash97 httpsdoiorg10

310910408448409029321

Bell E M (2012) Life at extremes Environments organisms and strategies

for survival Oxfordshire CABI httpsdoiorg101079

97818459381470000

Bierbach D Arias Rodriguez L amp Plath M (2018) Intrasexual competi-

tion enhances reproductive isolation between locally adapted popula-

tions Current Zoology httpsdoiorg101093czzox071 [Epub

ahead of print]

Bierbach D Klein M Sassmannshausen V Schlupp I Riesch R

Parzefall J amp Plath M (2012) Divergent evolution of male aggres-

sive behavior Another reproductive isolation barrier in extremophile

poeciliid fishes International Journal of Evolutionary Biology 2012

148745

Blackburn D G (1999) Viviparity and oviparity Evolution and reproduc-

tive strategies In T E Knobil amp J D Neill (Eds) Encyclopedia of

reproduction Vol 4 (pp 994ndash1003) New York NY Academic Press

Bouzat J L (2010) Conservation genetics of population bottlenecks

The role of chance selection and history Conservation Genetics 11

463ndash478

Brown A P Greenway R Morgan S Quackenbush C R Giordani

L Arias-Rodriguez L Kelley J L (2017) Genome-scale data

reveals that endemic Poecilia populations from small sulfide spring

display no evidence of inbreeding Molecular Ecology 26 4920ndash

4934

Brown J H Marquet P A amp Taper M L (1993) Evolution of body

size Consequences of an energetic definition of fitness American

Naturalist 142 573ndash584 httpsdoiorg101086285558

Burton R S Pereira R J amp Barreto F S (2013) Cytonuclear genomic

interactions and hybrid breakdown Annual Review of Ecology

TOBLER ET AL | 855

inference of putative mechanisms underlying adaptation even in sys-

tems without a priori predictions (Kell amp Oliver 2004) However

genetic and gene expression data alone are insufficient to infer

adaptation at specific loci because neutral evolutionary processes

demographic history and genetic hitchhiking can generate false posi-

tives in genomewide analyses of selection and gene expression (Bar-

ton 1998 Ramirez-Soriano Ramos-Onsins Rozas Calafell amp

Navarro 2008 Whitehead amp Crawford 2006) This underscores the

importance of generating a priori hypotheses about the molecular

underpinnings of adaptation and testing whether genetic and gene

expression variation at specific loci actually affects organismal per-

formance under different environmental conditions (Dalziel Rogers

amp Schulte 2009 Storz amp Wheat 2010)

21 | Developing and testing a priori hypotheses

There is a rich history of research in toxicology and biomedicine that

has investigated the molecular targets of H2S illuminating the mech-

anisms of H2S toxicity and detoxification in model organisms and

human cell cultures (see Li et al 2011 Olson Donald Dombkowski

amp Perry 2012 Wang 2012 Wallace amp Wang 2015 for recent

reviews) In a recent attempt of ldquoseparating hype and hoperdquo in

potential biomedical applications of H2S Olson (2011) lamented that

H2S ldquoseemingly affects all organ systems and biological processes in

which it has been investigatedrdquo While it often remains unclear

whether the documented effects are a direct consequence of H2S

exposure or a result of the disruption of cellular homeostasis this

body of work provides fruitful grounds for the development of

hypotheses about the molecular mechanisms underlying adaptation

to H2S-rich environments that can then be tested using genome

scans for selection and analyses of gene expression variation

between sulphide-adapted and nonadapted populations (see Table 1

for an overview) Most importantly candidates include direct toxicity

targets involved in OXPHOS (Cooper amp Brown 2008) and oxygen

transport (Pietri Roman-Morales amp Lopez-Garriga 2011) anaerobic

metabolism (Dubilier Giere amp Grieshaber 2005) and oxidative stress

response systems (Eghbal Pennefather amp OrsquoBrien 2004) that may

help mitigate adverse consequences of H2S exposure as well as

pathways involved in enzymatic H2S detoxification sulphur process-

ing and sulphur excretion (Hildebrandt amp Grieshaber 2008) It is

important to note that H2S is also produced endogenously by meta-

zoans in the cytosol and in mitochondria through enzymatic reac-

tions associated with the reverse trans-sulphuration pathway

(Stipanuk 2004 Stipanuk amp Ueki 2011) At concentrations well

below the toxicity threshold endogenously produced H2S serves as

a cellular signalling molecule that modulates physiological processes

including the function of ion channels that underlie smooth muscle

control and the function of neural cells (Peers Bauer Boyle Scragg

amp Dallas 2012) and it interacts with transcription factors that con-

trol inflammation cell survival and proliferation the expression of

cellular protective enzymes and hypoxia responses (Li et al 2011

Wallace amp Wang 2015 Wang 2012) Organisms may thus not strive

to merely eliminate H2S from their system and avoid effects of

toxicity Instead they are likely selected for maintaining optimal

endogenous concentrations that facilitate H2S-mediated signalling

and assure proper physiological functioning As a consequence

endogenous H2S production pathways and targets of H2S signalling

may also be modified during adaptation to environmental H2S

A series of studies in members of the P mexicana species com-

plex have tested hypotheses about the molecular underpinnings of

H2S adaptation by analysing patterns of molecular evolution in can-

didate genes (Barts et al 2018 Pfenninger et al 2014) deploying

genomewide scans for selection based on FST outliers (Pfenninger

et al 2015) and quantifying genomewide patterns of gene expres-

sion (Kelley et al 2016 Passow Henpita et al 2017) Collectively

the results of these studies have indicated a striking congruence

between candidate genes inferred from toxicological and biomedical

research and genes with empirical evidence for selection andor dif-

ferential expression between at least some of the adjacent H2S-

adapted and nonadapted populations (Table 1) (i) There was evi-

dence for selection on genes encoding the primary toxicity target of

H2S COX (complex IV in OXPHOS Pfenninger et al 2014) In addi-

tion genes encoding other components and assembly factors of

OXPHOS are either under selection or differentially expressed in sul-

phide spring populations (Kelley et al 2016 Pfenninger et al

2015) Evidence for selection and differential expression was also

uncovered for subunits of haemoglobin whose function is directly

impaired by H2S (Barts et al 2018) (ii) Genes involved in anaerobic

ATP production and in coping with oxidative stress exhibited evi-

dence for selection and differential expression suggesting a shift

towards alternative sources of ATP production and an increased abil-

ity to withstand reactive oxygen species that are produced as a by-

product of OXPHOS inhibition by H2S (Kelley et al 2016 Pfen-

ninger et al 2015) (iii) There is overwhelming evidence for selection

on and differential regulation of genes encoding components of the

sulphidequinone oxidoreductase pathway which mediates enzymatic

H2S detoxification in metazoans (Kelley et al 2016 Pfenninger

et al 2015) In addition there is evidence for upregulation of genes

encoding transporters of oxidized sulphur species which are puta-

tively involved in sulphur excretion (Kelley et al 2016)

These results suggest that adaptation to H2S-rich environments

in members of the P mexicana species complex involves both an

increased ability to cope with elevated endogenous H2S and an

increased ability to keep endogenous H2S concentrations low

through detoxification and excretion However there is a noticeable

absence of evidence for genes and physiological pathways associ-

ated with endogenous H2S production and H2S-mediated cell sig-

nalling being involved in adaptation (see Kelley et al 2016

Pfenninger et al 2015) There are three nonmutually exclusive

hypotheses that could explain this finding First the ability of popu-

lations from sulphidic habitats to regulate endogenous H2S through

detoxification may be sufficient to maintain normal H2S-mediated

signalling Second compared to the ability for detoxification and for

mitigation of direct toxic effects in terms of ATP production and

oxidative stress H2Srsquos influence on cell signalling may have relatively

low fitness consequences with concomitantly smaller evolutionary

846 | TOBLER ET AL

responses Last inferences from biomedical studies about the causal

relationship between endogenously produced H2S and various physi-

ological functions may be weak and some documented physiological

effects may instead be an indirect consequence of the disruption of

homeostasis that cascades through interconnected pathways operat-

ing in a cell In this case evolutionary studies on natural systems

exposed to H2S may provide important insights in terms of prioritiz-

ing research on biomedically relevant aspects of H2S biology Study-

ing ancestral populations susceptible to H2S will allow for

quantifying consequences of disruptions in H2S homeostasis at all

levels of biological organization (like in presently available animal

models) Adding derived H2S tolerant populations will further allow

for the detection of critical modifications in physiological processes

affected by H2S Consequently such a comparative approach may

provide a powerful tool to disentangle direct and indirect physiologi-

cal consequences of deviations in H2S homeostasis which is cur-

rently difficult in traditional biomedical models

Despite some of these discrepancies the studies exploring mech-

anisms underlying H2S tolerance in sulphide spring fishes highlight

the utility of research in toxicology and biomedicine in generating

testable hypotheses about evolutionary processes Biomedical stud-

ies on metabolic disorders (Aspiras Rohner Martineau Borowsky amp

Clifford 2015) stress responses (Rohner et al 2013) responses to

low oxygen (Eales Hollinshead amp Tennant 2016) and effects of

xenobiotics (Stockinger Di Meglio Gialitakis amp Duarte 2014) have

also facilitated our understanding of adaptive evolution in other nat-

ural systems Considering the increasingly specialized and frag-

mented research areas in the life sciences interdisciplinary

approaches and reciprocal exchange of ideas and methodologies

continue to be critical for facilitating breakthroughs in basic and

applied research on biomedical models and natural systems (Carroll

et al 2014 Nesse amp Stearns 2008 Schartl 2014)

22 | Convergent and nonconvergent evolution inindependent lineages

Our discussion of putative candidate genes that are under selection

andor differentially expressed in sulphide spring populations so far

focused on general patterns documented across evolutionarily inde-

pendent lineages in different river drainages Selection scans (Pfen-

ninger et al 2015) and analyses of differential gene expression

(Kelley et al 2016) have indicated high congruence of biological

processes associated with candidate genes across lineages in differ-

ent river drainages indicating that evolution has repeatedly and pre-

dictably modified the same physiological pathways during adaptation

to environmental H2S However the signal of evolutionary conver-

gence among sulphide spring lineages decreases with the level of

organismal organization (from physiological pathways down to geno-

mic regions genes and individual SNPs in selection scans Pfenninger

et al 2015 from physiological pathways down to genes and tran-

scripts in expression studies Passow Brown et al 2017) indicating

that different lineages have often modified the same physiological

pathways in different ways While knowledge about the biochemical

TABLE 1 Adaptation to H2S can occur through three principlemechanisms (1) Resistance involves direct toxicity targets whosefunctions are impaired by H2S Adaptive modifications of thesetoxicity targets are predicted to assure proper function despitethe presence of H2S (2) Mitigation involves alternativephysiological pathways that compensate for the inhibition ofdirect toxicity targets Adaptive modifications of these alternativephysiological pathways are predicted to allow for the maintenanceof organismal function by other means (3) Regulation involvesphysiological pathways associated with the maintenance ofendogenous H2S homeostasis Adaptive modifications ofregulatory pathways are predicted to maintain low endogenousconcentrations despite continuous flux of H2S from theenvironment into the body

Candidate pathways Evidence

(1) Resistance

OXPHOS bull Positive selection on COX in some lineages of

P mexicana (Puyacatengo and Pichucalco)a as

well as some other genera of sulphide spring

poeciliidsb COX amino acid substitutions in P

mexicana allow for the maintenance of function

in the presence of H2Sa

bull Upregulation of cytochrome c and some COX

subunits as well as downregulation of compo-

nents of complex I in all sulphide spring lineages

of P mexicana indicating complex changes in

aerobic ATP productionc

Oxygen transport

genesbull Positive selection on and upregulation of

several oxygen transport genes (including myo-

globin and multiple haemoglobin subunits) in

sulphide spring lineages of the genera

Gambusia Limia Poecilia and Pseudoxiphophorus

indicating modifications to putatively maintain

oxygen transport in the presence of H2Sd

(2) Mitigation

Anaerobic

metabolismbull Upregulation of genes associated with glycolysis

and gluconeogenesis in all sulphide spring lin-

eages of P mexicana indicating increased

capacity of anaerobic ATP productionc

Oxidative stress

responsesbull Upregulation of genes associated with

oxidative stress responses in all sulphide

spring lineages of P mexicana indicating

increased capacity to cope with H2S-induced

oxidative stressc

(3) Regulation

H2S detoxification bull Positive selection on and upregulation of sev-

eral genes associated with enzymatic H2S

detoxification in all sulphide spring lineages of

P mexicanace Upregulation of H2S detoxifica-

tion enzymes in other sulphide spring lineages

of the genera Gambusia Limia Poecilia Pseudox-

iphophorus and Xiphophorusb This indicates an

increased ability to maintain low endogenous

H2S concentrations during exposure

H2S production bull No evidence for modifications of genes associ-

ated with enzymatic production of H2Sc

aPfenninger et al (2014) bR Greenway J L Kelley amp M Tobler unpub-

lished data cKelley et al (2016) dBarts et al (2018) ePfenninger et al

(2015)

TOBLER ET AL | 847

and physiological consequences of H2S may thus enable us to pre-

dict the physiological pathways in which adaptive modifications

occur it may be impossible to predict specific modifications within

those pathways This contrasts with a common expectation in the

field that convergent evolution of molecular mechanisms underlying

phenotypic traits (ie parallel evolution) necessarily hinges on modi-

fication of the same loci across lineages as found in some of the

best-known examples of parallelism (Colosimo et al 2005 Feldman

Brodie Brodie amp Pfrender 2012 Grant Grant Markert Keller amp

Petren 2004) However modification of the same loci may only be

common under particular circumstances (Rosenblum Parent amp

Brandt 2014) for example when selection acts on standing genetic

variation (Colosimo et al 2005) when adaptive alleles are intro-

gressed across population boundaries through gene flow (Grant

et al 2004) or when there are strong genetic constraints (Feldman

et al 2012) Exploration of molecular mechanisms underlying con-

vergent evolution in many systems may therefore be better

approached from the perspective of interacting genes that are orga-

nized in pathways in which alternative genetic changes can have

equivalent functional consequences (ie functional redundancy)

Developing and testing hypotheses about parallel evolution at the

pathway level will hinge on improving genome annotations (Pavey

Bernatchez Aubin-Horth amp Landry 2012) and provide a more

nuanced understanding of the factors determining the predictability

of evolution at different levels of organismal organization (Rosen-

blum et al 2014)

Despite evidence for convergent evolution in some candidate

physiological pathways it also needs to be emphasized that this

does not apply to all P mexicana lineages or to all candidate physio-

logical pathways associated with H2S adaptation investigated to

date Nonconvergent evolutionmdashevidenced by unique lineage-speci-

fic signals of selection and differential expressionmdashmay be indicative

of unique molecular mechanisms to cope with H2S toxicity For

example signatures of positive selection on COXmdashH2Srsquos primary

toxicity targetmdashcan be found in sulphide spring lineages of the Puy-

acatengo and Pichucalco river drainages (a lineage with recent diver-

gence time and a lineage with older divergence time respectively)

but not in those of the Tacotalpa drainage (Pfenninger et al 2014)

The presence of nonconvergent genetic and gene expression differ-

ences between populations from sulphidic and non-sulphidic habitats

in different river drainages highlights the limitations of inferences

that can be drawn based on data sets generated with high-through-

put DNA and RNA sequencing The potential importance of modifi-

cations at candidate locimdashor the lack thereofmdashcan only be assessed

through functional assays that document how alternative alleles and

changes in gene expression actually affect organismal performance

The functional consequences of genetic and gene expression varia-

tion at most candidate loci remain to be investigated However

enzyme activity assays have indicated that COX maintains its activity

with increasing H2S concentrations in the two sulphide spring popu-

lations with evidence for positive selection on mitochondrially

encoded subunits (Puyacatengo and Pichucalco Pfenninger et al

2014) Key amino acid substitutions occur in the D-pathway channel

of COX1 which putatively narrow the channel width and reduce

H2Srsquos access to the binuclear centre of the protein Consequently

some lineages of P mexicana exhibit an H2S-resistant COX that can

maintain its activity and aerobic ATP production even if endogenous

H2S concentrations are elevated In contrast COX activity in sul-

phide spring fish of the Tacotalpa river drainage and in fish from

nonsulphidic habitats decreases when H2S is present which is con-

sistent with mechanisms of H2S toxicity (Pfenninger et al 2014) In

vitro assays on isolated mitochondria have nonetheless indicated

that an H2S-resistant COX is not necessary to maintain aerobic ATP

production during H2S exposure Unlike in populations from nonsul-

phidic habitats mitochondrial spare respiratory capacity is main-

tained in the presence of H2S even in the Tacotalpa sulphide spring

populations that lack an H2S-resistant COX (C Henpita M Tobler amp

JH Shaw unpublished data) This may indicate that modification of

detoxification pathways in sulphide spring fishes is sufficient to

maintain mitochondrial function and aerobic ATP production

23 | Ensuing challenges linking genotypes tofitness

Understanding the mechanistic basis of adaptation requires linking

genotypic variation to phenotypic traits that ultimately shape fitness

along selective gradients The evolution of H2S tolerance in sulphide

spring fishes appears to be a consequence of modifications in many

genes and many physiological pathways whose functions interact to

eliminate H2S from the system minimize effects of elevated endoge-

nous concentrations on direct toxicity targets and mitigate adverse

effects that cascade through cellular metabolism during disruption of

homeostasis (Table 1) Such complex genetic changes are likely com-

mon during adaptation and have been documented in other systems

that involve major evolutionary shifts including the emergence of

parasitic lifestyles (Werren et al 2010) eusociality (Simola et al

2013) asexual reproduction (Fradin et al 2017) and internal gesta-

tion with livebearing (Schartl et al 2013) Consequently comple-

mentary approaches will be required to uncover the molecular

mechanisms of adaptation to H2S in the future

1 The involvement of many genes and physiological pathways in

shaping tolerance emphasizes the importance of quantitative

genetic approaches that leverage experimental crosses between

H2S tolerant and nontolerant populations and test whether toler-

ancemdashand adaptive trait modifications that contribute to itmdashis

explained by a few major effects loci or whether it is a conse-

quence of a number of genes with small effects (Klerks Xie amp

Levinton 2011 Saltz Hessel amp Kelly 2017) Combining analyses

of phenotype-fitness associations and their underlying genetic

mechanisms via QTL mapping with the results of selection scans

from population genomic analyses (Pfenninger et al 2015) will

allow for an integrative understanding of how selection acting on

the genome results in phenotypic variation that mediates adapta-

tion to different environments (Stinchcombe amp Hoekstra 2008)

From a functional perspective breaking up correlations between

848 | TOBLER ET AL

different physiological traits (associated with resistance mitiga-

tion and regulation Table 1) in F2 crosses will further allow test-

ing what molecular mechanisms are necessary for tolerance

whether there are alternative strategies with equivalent fitness

benefits and whether there are complementary modifications

with additive effects on tolerance

2 Considering the number of genes and physiological pathways

likely involved in H2S tolerance assessing the functional conse-

quences of genetic and gene expression variation is a daunting

task At the same time such functional tests can significantly

contribute to closing the genotype-to-phenotype gap because

physiochemical stress responses can be assessed along the con-

tinuum of organismal organization (Gjuvsland Vik Beard Hunter

amp Omholt 2013) Standardized assays can be used to quantify

the biochemical makeup of organelles cells and tissues (eg pro-

tein abundance and enzyme activity) physiological performance

in vitro and in vivo (eg rates of H2S detoxification ATP produc-

tion and reactive oxygen production) and fitness-relevant proxies

of whole organism performance (eg H2S tolerance Figure 2)

This highlights the importance of methods development for the

quantification of complex phenotypic traits or even whole phe-

nomes that can keep pace with rapid data generation through

high-throughput sequencing (Houle Govindaraju amp Omholt

2010 Ritchie Holzinger Li Pendergrass amp Kim 2015) In addi-

tion genotypendashphenotype links established through such correla-

tive approaches will ideally be verified through targeted gene

editing While gene editing with CRISPRndashCas9 is now routinely

employed in model organisms (Doudna amp Charpentier 2014 Sha-

lem Sanjana amp Zhang 2015) applying this technology to non-

models still remains challenging especially in livebearing species

like poeciliid fishes where in vitro culture of embryos is not rou-

tinely employed (Martyn Weigel amp Dreyer 2006) Nonetheless

the effects of specific mutations can potentially be verified

through the establishment and manipulation of cells in culture

(Liu et al 2014) or by generating transgenics (Lin et al 2016)

3 Organismal responses to physiochemical stressors occur at multi-

ple hierarchical levels that can be plastically modulated in

response to endogenous and exogenous cues (Whitehead 2012)

Understanding adaptation to physiochemical stressors conse-

quently requires disentangling the effects of genetic variation

developmental plasticity and short-term acclimation in gene

expression patterns on the emerging physiological processes

Common garden experiments on sulphide spring fishes have

shown that many genes differentially expressed between popula-

tions from sulphidic and nonsulphidic habitats in nature remain

differentially expressed in laboratory-reared fish that have never

been exposed to H2S indicating evolved differences in constitu-

tive gene expression (Passow Henpita et al 2017) At the same

time expression patterns in the laboratory also exhibit evidence

for plasticity in response to short-term H2S exposure (Passow

Henpita et al 2017) Some of these plastic responses are shared

between populations from sulphidic and nonsulphidic habitats

(likely representing shared ancestral plasticity) while others have

been lost or gained during sulphide spring colonization A core

challenge for future research is to overcome logistical constraints

that have so far prevented long-term H2S exposures of fish in

the laboratory Exposing individuals throughout development will

be required to test how developmental plasticity and short-term

acclimation shape responses to H2S in individuals from adapted

and nonadapted populations Finally the next big scientific

advance in this context will also require the identification of reg-

ulatory mechanisms that shape expression differences during

adaptation including sequence changes in transcription factors

and their binding sites patterns of DNA methylation and chro-

matin accessibility as well as the production of microRNAs and

competing endogenous RNAs that can modulate transcript abun-

dance (Flores Wolschin amp Amdam 2013 Lopez-Maury Mar-

guerat amp Beuroahler 2008 Silva Bye el Azzouzi amp Wisloslashff 2017)

In addition the role of alternative splicing and other post-tran-

scriptional and post-translational modifications in mediating adap-

tation remains understudied by evolutionary biologists (Ast

2004 Bush Chen Tovar-Corona amp Urritia 2017) This is particu-

larly relevant because H2S has been hypothesized to play a role

in the post-translational modification of proteins with sulphur-

containing amino acids through sulfhydration (Mustafa et al

2009)

Bridging the genotypendashphenotype gap remains one of the major

challenges in biology (Gjuvsland et al 2013 Ritchie et al 2015) and

will provide opportunities to understand how organisms function in

the face of environmental stress and how evolution shapes these

functions during adaptation to different environmental conditions

Establishing the molecular mechanisms of adaptation across repli-

cated lineages exposed to the same sources of selection will also

provide insights into how evolutionary history demography and

genetics shape convergent evolutionary outcomes providing a better

understanding at what hierarchical levels of organismal organization

evolutionary change may actually be predictable (Rosenblum et al

2014)

3 | COMPLEX ORGANISMS IN A COMPLEXWORLD DISENTANGLING THE TANGLEDBANK

Mechanisms of adaptation can be studied through bottom-up

(screening for genetic modifications that allow for inferences about

relevant phenotypic traits) and top-down approaches (assessing fit-

ness-relevant phenotypic differences among populations without

knowledge about the genetic basis see Schluter 2009) Over the

years genomic and transcriptomic studies on sulphide spring popula-

tions of the P mexicana species complex have been complemented

with analyses of phenotypic variation Just as genetic analyses have

indicated that adaptation to H2S is mediated by the modification

and modulation of many genes associated with multiple physiological

pathways phenotypic studies uncovered a variety of differences

TOBLER ET AL | 849

between populations from sulphidic and nonsulphidic environments

Many of these phenotypic differences have evolved in convergence

across replicated sulphide spring lineages in different river drainages

Examples include modifications of morphological (body shape and

organ size Tobler et al 2011 Tobler Scharnweber et al 2015

Schulz-Mirbach et al 2016) physiological (tolerance and metabolic

rates Tobler et al 2011 Passow Arias-Rodriguez amp Tobler 2017)

behavioural (aggression and personality Riesch et al 2009 Bierbach

et al 2012 Sommer-Trembo et al 2016 Bierbach Arias Rodriguez

amp Plath 2018) and life history traits (fecundity and offspring size

Riesch Plath Garcia de Leon amp Schlupp 2010 Riesch et al 2014)

Hence proximate populations in different habitat types have

diverged in complex phenotypes that span many different suites of

traits and a key question is how divergence in multivariate pheno-

types arises In theory H2S could be a key source of selection driv-

ing multivariate phenotypic divergence either because it directly

exerts selection on a multitude of traits that are independently modi-

fied to reach new fitness peaks or because selection on a few fit-

ness-relevant traits causes simultaneous changes in correlated

characteristics Neither explanation seems sufficient because many

Abiotic environmentOxygen concentrations

Ionic compositionTemperature

Structure and flow

Biotic environmentTrophic resources

DemographyPredationParasitism

H2S

Genetic variation

Gene expression

Biochemistry

Physiologicalfunction

Organismal performance

Primary fitness components

Survivalgrowth

reproduction

Organismal function Environment

F IGURE 2 Understanding how organisms function in the context of their environment is a core question in evolutionary biology yet thecomplexities associated with organismal function and environmental variation provide significant challenges High-throughput sequencing hasallowed for the identification of candidate genes subject to selection and differential expression during adaptation to H2S The next step is tobridge the genotype-to-phenotype gap by quantifying the integrated responses to H2S across the continuum of organismal organization includingbiochemical phenotypes physiological functions whole organism performance and ultimately fitness In addition understanding adaptation toH2S-rich environments will also require acknowledging H2Srsquos relation to other abiotic and biotic environmental variables which can exert selectionand modulate aspects of organismal function This will allow for the establishment of cause-and-effect relationships between specific sources ofselection and evolutionary change in specific traits and contribute to our understanding of organismal function in natural systems

850 | TOBLER ET AL

of the divergent traits between populations from sulphidic and non-

sulphidic environments have no clearmdashor at best tenuousmdashlinks to

H2S toxicity and detoxification and there is only limited evidence

for within-population correlations between different traits groups

(Riesch et al 2016)

Evolutionary analyses frequently focus on seemingly simple

sources of selection juxtaposing populations that are assigned to dis-

crete habitat types to understand the evolution of specific traits

(Kaeuffer Peichel Bolnick amp Hendry 2012) But as Darwin (1859)

already observed in his ldquotangled bankrdquo allegory selection in natural

systems is rarely univariate but multifarious selective regimes con-

tribute to trait divergence even among populations that occur along

apparently simple environmental gradients Unquantified environmen-

tal variation that is subsumed in the categorization of habitat types

can profoundly affect trait variation in natural systems (Stuart et al

2017) highlighting the importance of considering and testing alterna-

tive hypotheses about the cause-and-effect relationship between

specific sources of selection and evolution of specific traits Thus

accurate natural history observations that consider organisms as holis-

tic entities consisting of interacting traits in the context of the abiotic

and biotic factors comprising their environment are critical to inform

the design and interpretation of empirical research in natural systems

(Greene 2005) Indeed adjacent sulphidic and nonsulphidic habitat

types do not merely differ in the presence or absence of H2S but also

in a variety of other abiotic and biotic factors (Figure 2) Unlike nonsul-

phidic habitats sulphide springs in southern Mexico are characterized

by extreme hypoxia elevated salinity and lower pH (Rosales Lagarde

2012 Tobler et al 2011) In addition the presence of physiochemical

stressors affects the biotic communities in these sulphide springs Spe-

cies richness of aquatic invertebrates and vertebrates is drastically

reduced as a consequence of H2S toxicity (Greenway Arias-Rodriguez

Diaz amp Tobler 2014) and primary production appears to be driven in

part by chemoautotrophic bacteria that use H2S as an energy source

rather than just by autochthonous and allochthonous photosynthetic

organisms (Roach Tobler amp Winemiller 2011) These changes in the

biotic environment affect the relative importance of intra-vs inter-

specific competition trophic resource use and exposure to predators

and parasites (Riesch Oranth et al 2010 Tobler et al 2014 Tobler

Scharnweber et al 2015)

31 | Alternative drivers of trait evolution

Acknowledging the complexity of selective regimes in sulphide

springs allows for a more nuanced understanding of trait evolution

and for the development of alternative hypotheses about the ulti-

mate mechanisms that cause divergence in specific genetic and phe-

notypic traits In some instances sources of selection other than H2S

can plausibly explain trait divergence between populations from sul-

phidic and nonsulphidic habitats For example hypoxia likely drives

increases in head size and gill surface area (Tobler et al 2011)

changes in the ionic composition of the water can explain differential

expression of genes associated with osmoregulation (Kelley et al

2016) and shifts in trophic resource use are associated with

modifications of the gastrointestinal system (Tobler Scharnweber

et al 2015) In other cases disentangling alternative hypotheses is

not trivial because selection mediated by H2S and other environmen-

tal factors are predicted to affect trait modifications in similar ways

Shifts towards anaerobic metabolism in sulphide spring fish could be

the consequence of inhibition of aerobic ATP production by H2S or

by constraints in oxygen availability Similarly reproductive life his-

tory shifts towards the production of larger but fewer offspring could

be driven by H2S (decreasing offspringrsquos surfacevolume ratio to mini-

mize the diffusion of H2S into the body Powell 1989) by habitat dif-

ferences in the quality and quantity of predation that can affect size-

specific rates of extrinsic mortality (Jorgensen Auer amp Reznick

2011) or by energetic constraints (see below) While comparisons

across different selective regimes can shed light into the ultimate dri-

vers of trait evolution (Moore Riesch amp Martin 2016) isolating the

effects of different sources of selection may not be possible because

they could act in concert and facilitate increases of organismal perfor-

mance in multiple contexts (Ghalambor Walker amp Reznick 2003)

When multifarious selection shapes multivariate phenotypic evo-

lution it is important to also consider the role of trade-offs both in

terms of trait expression (increased investment in one trait may

impose constraints on the expression of other traits) and in terms of

trait function (optimizing the trait for one function may impair the

traitrsquos utility for other functions) For example selection on large off-

spring size in sulphide spring fishes has inadvertent consequences

for fecundity (Riesch et al 2014) and selection for increased gill

surface areas due to hypoxia simultaneously increases exposure to

environmental H2S (Tobler et al 2011) Thus the outcome of phe-

notypic evolution is ultimately a result of balancing competing needs

in a manner that optimizes trait expression and organismal function

in multiple contexts to maximize fitness (Arnold 1983) and the pres-

ence of trade-offs can have cascading indirect effects with selection

on one or a few traits leading to the modification of a multitude of

apparently unrelated characteristics

In this context bioenergetics may represent a nexus in the evo-

lution of complex traits observed in sulphide spring fish and other

extreme environments Living in and adapting to sulphide springs

imposes constraints on organismal energy availability (Plath et al

2007) with according trade-offs for energy allocation Energy limita-

tion in sulphide springs does not arise from a scarcity of food

resources (chemoautotrophic bacteria eaten by the fish are abundant

in sulphide springs Tobler Scharnweber et al 2015) but from

behavioural and physiological adaptation A primary survival strategy

for fish in the toxic and hypoxic springs is to engage in aquatic sur-

face respiration where individuals breathe directly at the water sur-

face to maximize oxygen uptake and minimize H2S exposure This

behaviour mediates short-term survival (Plath et al 2007) but it

also constrains energy acquisition for benthic foragers like P mexi-

cana (Tobler Riesch Tobler amp Plath 2009) and increases susceptibil-

ity to bird predation (Riesch Oranth et al 2010) Costly investments

into H2S detoxification and tissue repair as well as a shift from aero-

bic to less efficient anaerobic metabolism further constrain energy

allocation to other functions (Passow Greenway Arias-Rodriguez

TOBLER ET AL | 851

Jeyasingh amp Tobler 2015) Theory predicts that energetic con-

straints ndash whether mediated by limitations in the rate of resource

acquisition or in the rate of energy allocation to reproduction (as

opposed to maintenance or growth) ndash should exert selection for a

reduction in organismal energy budgets which allows for the maxi-

mization of residual energy allocation for the production of offspring

(Brown Marquet amp Taper 1993) Several lines of evidence support

the hypothesis that energy limitation plays a critical role in pheno-

typic evolution of sulphide spring fish Compared to individuals in

nonsulphidic environments fish in sulphide springs have significantly

lower routine energy demands which is mediated both by reduc-

tions in body size and mass-specific metabolism (Passow Arias-

Rodriguez et al 2017) Reductions in other traits such as energeti-

cally costly behaviours (Bierbach et al 2012 Plath 2008) and the

size of energy-demanding organs (Schulz-Mirbach et al 2016) may

also contribute to the reduced energy demands and shifts in repro-

ductive life history traits (Riesch et al 2014) are consistent with

selection from energy limitation (Reznick amp Yang 1993)

32 | Ensuing challenges Disentanglingenvironmental and organismal complexities

Future research needs to empirically address how multifarious selec-

tive regimes shape the evolution of organisms with complex pheno-

types We have made great advances in our understanding of how

specific traits evolve and mediate adaptation (eg morphology Chan

et al 2010 coloration Dasmahapatra et al 2012 sensory organs

Cartwright Schwartz Merry amp Howell 2017) and we have a thor-

ough understanding of how single environmental factors can shape

organismal responses (eg climatic conditions Johnston amp Bennett

1996 predation Langerhans 2007 competition Pfennig amp Pfennig

2012) In contrast our understanding of phenotypic evolution in

response to multifarious selection is comparatively limited especially

because interactive effects of different environmental factors can lead

to organismal and evolutionary responses that are unpredictable based

on experiments that manipulate single environmental variables in iso-

lation (Gunderson Armstrong amp Stillman 2016 Todgham amp Stillman

2013) Future research thus needs to acknowledge the complexities of

environments and organismal design and address them explicitly in

empirical investigations to gain a robust understanding of biological

diversification (Culumber amp Tobler 2017 Stuart et al 2017)

4 | LINKING ADAPTATION TOSPECIATION

So far we primarily explored the causes of adaptive evolution but

what are its consequences Theory predicts that adaptation to local

environmental conditions can lead to speciation because reproduc-

tive isolation emerges incidentally as a by-product (ecological specia-

tion Nosil 2012) Population genetic analyses of Poecilia spp in

sulphide springs and adjacent populations in nonsulphidic environ-

ments have indicated strong genetic differentiation across small

spatial scales (in some instances lt100 m) and in the absence of

physical barriers that would prevent fish movement (Plath et al

2013) In addition population genetic structuring has been shown to

be stable even in the face of major environmental disturbance

BOX 1 Speciation in sulphide springs and conservation

biology

Adaptation to H2S-rich environments in poeciliids coincides

with ecological speciation giving rise to phenotypically dis-

tinct and reproductively isolated lineages In some

instances this has led to the evolution of highly distinct

micro-endemic species including Poecilia sulphuraria and

P thermalis Limia sulphurophila and Gambusia eurystoma

(Figure 3) Except for P sulphuraria which is found in two

spring complexes of the Rıo Pichucalco drainage of Mexico

(Tobler amp Plath 2009b) the distributions of these sulphide

spring endemics are restricted to single localities that

include only small areas of suitable habitat For example

the largest sulphide spring complex (the Ba~nos del Azufre

in southern Mexico which harbours P sulphuraria and G

eurystoma) includes ~18000 m2 of suitable habitat (equiva-

lent to lt35 football fields) while one of the smallest (La

Esperanza springs harbouring P thermalis) merely covers

~600 m2 (just over a tenth of a football field Brown et al

2017) Accordingly some sulphide spring endemics are

considered critically endangered (IUCN 2012) and pro-

tected by law (SEDESOL 1994)

Many threatened species with restricted ranges exhibit

low genetic diversity and high incidences of inbreeding

(Bouzat 2010 Willi Van Buskirk amp Hoffmann 2006) In

contrast at least P sulphuraria and P thermalismdashthe only

species investigated to datemdashactually exhibit relatively high

effective population sizes (Ne = 1200ndash3800 depending on

the population) and no evidence for inbreeding (FIS lt 0)

(Brown et al 2017) This is likely a consequence of the

long-term stability of sulphide springs and the high pro-

ductivity and low levels of interspecific competition also

support unusually high fish densities and thus population

census sizes (Brown et al 2017 Culumber et al 2016)

Nonetheless sulphide spring endemics are threatened by

habitat destruction which is mediated primarily by the

modification of springs for recreational purposes defor-

estation and conversion of land for agriculture and most

recently high-intensity palm oil plantations (Tobler amp Plath

2009ab) Despite the official protection of sulphide spring

endemics there have been no tangible conservation mea-

sures that foster the sustainable development of land

around sulphide springs with the well-being of the endemic

fishes in mind

852 | TOBLER ET AL

(catastrophic flood event Plath et al 2010) Population genetic dif-

ferentiation between adjacent sulphide-adapted and nonadapted

populations is indicative of ongoing ecological speciation (Box 1) but

different population pairs from sulphidic and nonsulphidic waters

have made variable progress along the speciation continuum towards

complete reproductive isolation (Plath et al 2013) This prompts

questions about the mechanisms involved in reducing gene flow

along the environmental gradient connecting the two habitat types

41 | Prezygotic mechanisms of reproductiveisolation

The simplest form of reproductive isolation between locally adapted

populations arises from selection against migrants between habitat

types (Nosil Vines amp Funk 2005) Translocation experiments in nat-

ural habitats have indicated strong survival selection against

migrants especially those translocated from nonsulphidic to sulphidic

habitats (Plath et al 2013) Fish translocated from nonsulphidic

environments face high mortality within short periods of time

(lt24 hr) which is likely caused by H2S toxicity and hypoxia in sul-

phide springs Hence low gene flow from populations inhabiting

nonsulphidic waters into populations inhabiting sulphidic habitats is

simply explained by a lack of adaptation to the extreme environmen-

tal conditions At least in some drainages there is also appreciable

mortality for sulphide spring fish translocated to nonsulphidic habi-

tats although mechanisms causing inviability are unclear and the

strength of selection appears to be lower (Plath et al 2013) If sur-

vival selection against immigrants was the sole source of reproduc-

tive isolation this should result in unidirectional gene flow from

H2S-adapted to nonadapted populations but population genetic

analyses provided no evidence for this implying that additional

mechanisms are at work (Plath et al 2013)

One such mechanism could be intersexual selection and mate

choice experiments have indicated that migrants from sulphidic to

nonsulphidic habitats are at a significant disadvantage with fish from

nonsulphidic habitats avoiding potential mating partners from sul-

phidic ones (Plath et al 2013 Sommer-Trembo et al 2016) Such

assortative mating has been found to be critical during ecological spe-

ciation because it counteracts the homogenizing effects of gene flow

(van Doorn Edelaar amp Weissing 2009) yet how assortative mating

patterns arise often remains unclear In theory assortative mating can

arise through a variety of processes (Nosil 2012) ldquoMagic traitrdquo mecha-

nisms where mating preferences evolve as a by-product of divergent

selection on ecological traits that serve as cues have recently garnered

much attention (Servedio van Doorn Kopp Frame amp Nosil 2011)

Consistent with magic trait mechanisms assortative mating in P mexi-

cana appears to be linked to adaptive population differences in body

shape which serve as cues to distinguish potential mating partners

from different habitats (Greenway Drexler Arias-Rodriguez amp Tobler

2016) However neither sulphide spring fish (which are unlikely to

encounter migrants from nonsulphidic habitats due to H2S toxicity)

nor allopatric P mexicana from nonsulphidic waters (which have never

experienced migrants from sulphide spring populations) exhibit a

preference for mating partners from their own ecotype suggesting

that assortative mating in this system is a product of reinforcement

(ie direct selection for assortative mating Greenway et al 2016)

This is different from classic magic traits mechanisms because assorta-

tive mating is not a mere by-product of adaptation but adaptive traits

have instead been co-opted as mating signals in response to selection

on mating preferences (Maan amp Seehausen 2012) Moreover males of

some H2S-adapted populations have reduced aggressive behaviour

(Bierbach et al 2012 2018) In staged contests of size-matched males

from different ecotypes intruders had a significantly lower likelihood

to establish dominance than resident males independent of whether

or not they showed reduced aggressiveness overall Hence intrasexual

selection may represent another mechanism mediating reproductive

isolation between locally adapted ecotypes Overall H2S-adapted poe-

ciliids allow disentangling the relative contributions of different forms

of ecological and sexual selectionmdashand their interdependencymdashduring

ecological speciation

42 | Ensuing challenges quantifying postzygoticmechanisms of reproductive isolation

In contrast to prezygotic mechanisms little is known about postzy-

gotic mechanisms of reproductive isolation that potentially act to

reduce gene flow between adjacent populations in sulphidic and

nonsulphidic habitats Putative hybrids between sulphidic and non-

sulphidic ecotypes have occasionally been observed in nonsulphidic

habitats immediately adjacent to confluences with sulphide springs

(R Riesch M Plath and M Tobler personal observations) but it

remains unclear whether and through what mechanisms such hybrids

exhibit reduced fitness compared to parentals

Direct selection of mitochondrially encoded genes in the pres-

ence of H2S (Pfenninger et al 2014) should be accompanied by co-

evolution at nuclear loci that directly interact with mitochondrial

gene products in the formation of protein complexes involved in the

OXPHOS pathway (Sunnucks Morales Lamb Pavlova amp Greening

2017 Willett amp Burton 2004) Disruption of co-evolved gene com-

plexes in hybrids should therefore lead to intrinsic fitness reductions

through mito-nuclear incompatibilities (Burton Pereira amp Barreto

2013 Wolff Ladoukakis Enrıquez amp Dowling 2014) which have

been hypothesized to be a key driver in speciation (Chou amp Leu

2010 Gershoni Templeton amp Mishmar 2009) Empirical tests of

these ideas remain to be conducted in sulphide spring fishes just

like studies on potential ecological and sexual selection against

hybrids that arise from interpopulation mating events

5 | PERSPECTIVES LIFE IN SULPHIDICENVIRONMENTS BEYOND P MEXICANA

Research on sulphide spring fishes of the P mexicana species com-

plex in Mexico has provided insights into mechanisms underlying

adaptation and the links between adaptation and speciation Sul-

phide springs do not only occur in Mexico however but are found

TOBLER ET AL | 853

worldwide and have been colonized by 24 evolutionarily indepen-

dent fish lineages in North America the Neotropics and the Middle

East (Greenway et al 2014) The vast majority of lineages belong to

the order Cyprinodontiformes (95) and livebearing members of the

family Poeciliidae in particular (80) This potentially indicates that

internal gestation of young has facilitated the colonization of

extreme environments because it shields developing embryos from

physiochemical stressors (Blackburn 1999) Among the Poeciliidae

sulphide spring fishes include members of the genera Brachyrhaphis

Gambusia Poeciliopsis Priapichthys Pseudoxiphophorus Xiphophorus

and multiple subgenera of Poecilia (Figure 3 Greenway et al 2014)

spanning over 40 million years of evolution (based on Reznick Fur-

ness Meredith amp Springer 2017)

A handful of studies have started to address molecular and phe-

notypic evolution across a broader taxon sampling of sulphide spring

fishes finding evidence for convergent evolution in morphology

(Riesch et al 2016 Tobler amp Hastings 2011) life history traits

(Riesch Plath et al 2010 Riesch et al 2014) and oxygen transport

genes (Barts et al 2018) In addition ecological speciation in

response to adaptation to H2S-rich environments has also been doc-

umented in sulphide spring poeciliids other than P mexicana (Riesch

et al 2016) These studies have laid a fruitful foundation for the

investigation of mechanisms of adaptation to H2S-rich springs from

genes to phenotypic traits linking micro-evolutionary processes to

macro-evolutionary patterns

The availability of broad-scale evolutionary replication of popula-

tions that have invaded habitats with the same physiochemical stressors

allows for addressing some of the ensuing challenges discussed above

and investigating the factors that give rise to predictable evolutionary

outcomes Convergent evolution is a commonmdashbut not omnipresentmdash

theme in adaptive evolution and multiple factors can determine the

degree of convergence among replicated lineages exposed to the same

sources of selection (Kaeuffer et al 2012 Landry amp Bernatchez 2010

Oke Rolshausen LeBlond amp Hendry 2017 Rosenblum amp Harmon

2011) Thus leveraging replicated lineages that have colonized sulphide

springs to assess molecular mechanisms underlying convergent pheno-

typic changes analyse how environmental heterogeneity among sul-

phide spring habitats causes lineage-specific patterns of trait evolution

and test whether and by what mechanisms adaptation to extreme envi-

ronments is linked to speciation will provide insights into how ecologi-

cal genetic and functional factors contribute to evolutionary

convergence at different levels of organismal organization (Rosenblum

et al 2014) An expanded taxon sampling with population pairs that

span almost the full range of the speciation continuum from panmixia to

complete reproductive isolation (Riesch et al 2016) will also help

addressing questions about temporal aspects of H2S adaptation In par-

ticular comparing younger to older lineages will provide insights about

the order at which genetic and phenotypic modifications accrue during

the adaptation process

Animals have also invaded H2S-rich habitats in other aquatic

ecosystems including cold seeps and deep-sea hydrothermal vents

(Tobler et al 2016) Adaptation to H2S in the P mexicana species com-

plex has involved modification of physiological pathways associated

with H2S toxicity (OXPHOS pathway) and detoxification (sulphide

F IGURE 3 Phylogeny of the family Poeciliidae (adopted from Culumber amp Tobler 2017) Various lineages in the genera Poecilia (subgeneraAcanthophacelus Limia and Mollienesia) Priapichthys Brachyrhaphis Poeciliopsis Xiphophorus Pseudoxiphophorus and Gambusia have colonizedH2S-rich springs in different parts of the Americas (see Greenway et al 2014) Lineages highlighted in yellow represent endemic speciesrestricted to H2S-rich springs while red lineages represent species in which some populations have invaded sulphidic habitats Pictures depict arepresentative of each focal genussubgenus

854 | TOBLER ET AL

quinone oxidoreductase pathway) which have been highly conserved

during the diversification of life on Earth (Saraste 1999 Shahak amp

Hauska 2008) The high replicability of H2S as an environmental factor

that shaped animal evolution across vastly different ecological contexts

consequently offers the opportunity to investigate whether selection

mediated by the presence of H2S has repeatedly modified these con-

served physiological pathways across animal phyla that have diverged

during the Cambrian explosion spanning over 500 million years of evo-

lution Such comparisons would provide unprecedented insights into

the effects of a single physiochemical stressor on evolutionary change

irrespective of ecological or phylogenetic contexts

6 | CONCLUSIONS

H2S-rich environments and their inhabitants represent useful models

for research in organismal biology ecology and evolution illustrating

both opportunities (eg strong and explicit source of selection evolu-

tionary replication) and challenges (eg complexity of environments

and organismal responses) in asking basic question about the origins

of biodiversity Understanding how organisms thrive in apparently

inhospitable environments will provide us with insights into lifersquos

capacities and limitations to adapt to novel environments (Waterman

1999) This is particularly relevant considering that human activities

substantially contribute to the exacerbation of physiochemical stres-

sors at local regional and global scales impacting ecological and evo-

lutionary dynamics (Alberti 2015 Norberg Urban Vellend

Klausmeier amp Loeuille 2012) Thus a focus on ecological and evolu-

tionary consequences of natural stressors that push organisms to the

brink of their physiological capacities may be instrumental to predict

consequences of anthropogenic environmental change

ACKNOWLEDGEMENTS

We would like to thank Lenin Arias-Rodriguez the Universidad

Juarez Autonoma de Tabasco and the Centro de Investigacion E

Innovacion Para la Ense~nanza y El Aprendizaje for their continued

collaboration enabling our work on the P mexicana system Funding

for this review article has been provided by the National Science

Foundation (IOS-1557860 IOS-1557795) and the US Army

Research Office (W911NF-15-1-0175)

AUTHOR CONTRIBUTIONS

MT wrote the first draft of the paper all authors contributed text

and comments

ORCID

Michael Tobler httporcidorg0000-0002-0326-0890

Joanna L Kelley httporcidorg0000-0002-7731-605X

Reuroudiger Riesch httporcidorg0000-0002-0223-1254

REFERENCES

Alberti M (2015) Eco-evolutionary dynamics in an urbanizing planet

Trends in Ecology amp Evolution 30 114ndash126 httpsdoiorg101016j

tree201411007

Alda F Reina R G Doadrio I amp Bermingham E (2013) Phylogeny

and biogeography of the Poecilia sphenops species complex

(Actinopterygii Poeciliidae) in Central America Molecular Phylogenet-

ics and Evolution 66 1011ndash1026 httpsdoiorg101016jympev

201212012

Alvarez del Villar J (1948) Descripcion de una nueva especie de Mol-

lienisia capturada en Ba~nos del Azufre Tabasco (Pisces Poeciliidae)

Anales de la Escuela Nacional de Ciencias Biologicas 5 275ndash281

Arnold S J (1983) Morphology performance and fitness American Zool-

ogist 23 347ndash361 httpsdoiorg101093icb232347

Aspiras A C Rohner N Martineau B Borowsky R L amp Clifford C J

(2015) Melanocortin 4 receptor mutations contribute to the adapta-

tion of cavefish to nutrient-poor conditions Proceedings of the

National Academy of Sciences USA 112 9968ndash9973

Ast G (2004) How did alternative splicing evolve Nature Reviews

Genetics 5 773ndash782 httpsdoiorg101038nrg1451

Bagarinao T (1992) Sulfide as an environmental factor and toxicant

Tolerance and adaptations in aquatic organisms Aquatic Toxicology

24 21ndash62 httpsdoiorg1010160166-445X(92)90015-F

Barton N H (1998) The effect of hitch-hiking on neutral genealogies

Genetic Research 72 123ndash133 httpsdoiorg101017

S0016672398003462

Barts N Greenway R Passow C N Arias Rodriguez L Kelley J L amp

Tobler M (2018) Expression and molecular evolution of oxygen

transport genes in livebearing fishes (Poeciliidae) from hydrogen sul-

fide rich springs Genome httpsdoiorg101139gen-2017-0051

[Epub ahead of print]

Beauchamp R O Bus J S Popp C S Boreiko C J amp Andjelkovich

D A (1984) A critical review of the literature on hydrogen sulfide

CRC Critical Reviews in Toxicology 13 25ndash97 httpsdoiorg10

310910408448409029321

Bell E M (2012) Life at extremes Environments organisms and strategies

for survival Oxfordshire CABI httpsdoiorg101079

97818459381470000

Bierbach D Arias Rodriguez L amp Plath M (2018) Intrasexual competi-

tion enhances reproductive isolation between locally adapted popula-

tions Current Zoology httpsdoiorg101093czzox071 [Epub

ahead of print]

Bierbach D Klein M Sassmannshausen V Schlupp I Riesch R

Parzefall J amp Plath M (2012) Divergent evolution of male aggres-

sive behavior Another reproductive isolation barrier in extremophile

poeciliid fishes International Journal of Evolutionary Biology 2012

148745

Blackburn D G (1999) Viviparity and oviparity Evolution and reproduc-

tive strategies In T E Knobil amp J D Neill (Eds) Encyclopedia of

reproduction Vol 4 (pp 994ndash1003) New York NY Academic Press

Bouzat J L (2010) Conservation genetics of population bottlenecks

The role of chance selection and history Conservation Genetics 11

463ndash478

Brown A P Greenway R Morgan S Quackenbush C R Giordani

L Arias-Rodriguez L Kelley J L (2017) Genome-scale data

reveals that endemic Poecilia populations from small sulfide spring

display no evidence of inbreeding Molecular Ecology 26 4920ndash

4934

Brown J H Marquet P A amp Taper M L (1993) Evolution of body

size Consequences of an energetic definition of fitness American

Naturalist 142 573ndash584 httpsdoiorg101086285558

Burton R S Pereira R J amp Barreto F S (2013) Cytonuclear genomic

interactions and hybrid breakdown Annual Review of Ecology

TOBLER ET AL | 855

responses Last inferences from biomedical studies about the causal

relationship between endogenously produced H2S and various physi-

ological functions may be weak and some documented physiological

effects may instead be an indirect consequence of the disruption of

homeostasis that cascades through interconnected pathways operat-

ing in a cell In this case evolutionary studies on natural systems

exposed to H2S may provide important insights in terms of prioritiz-

ing research on biomedically relevant aspects of H2S biology Study-

ing ancestral populations susceptible to H2S will allow for

quantifying consequences of disruptions in H2S homeostasis at all

levels of biological organization (like in presently available animal

models) Adding derived H2S tolerant populations will further allow

for the detection of critical modifications in physiological processes

affected by H2S Consequently such a comparative approach may

provide a powerful tool to disentangle direct and indirect physiologi-

cal consequences of deviations in H2S homeostasis which is cur-

rently difficult in traditional biomedical models

Despite some of these discrepancies the studies exploring mech-

anisms underlying H2S tolerance in sulphide spring fishes highlight

the utility of research in toxicology and biomedicine in generating

testable hypotheses about evolutionary processes Biomedical stud-

ies on metabolic disorders (Aspiras Rohner Martineau Borowsky amp

Clifford 2015) stress responses (Rohner et al 2013) responses to

low oxygen (Eales Hollinshead amp Tennant 2016) and effects of

xenobiotics (Stockinger Di Meglio Gialitakis amp Duarte 2014) have

also facilitated our understanding of adaptive evolution in other nat-

ural systems Considering the increasingly specialized and frag-

mented research areas in the life sciences interdisciplinary

approaches and reciprocal exchange of ideas and methodologies

continue to be critical for facilitating breakthroughs in basic and

applied research on biomedical models and natural systems (Carroll

et al 2014 Nesse amp Stearns 2008 Schartl 2014)

22 | Convergent and nonconvergent evolution inindependent lineages

Our discussion of putative candidate genes that are under selection

andor differentially expressed in sulphide spring populations so far

focused on general patterns documented across evolutionarily inde-

pendent lineages in different river drainages Selection scans (Pfen-

ninger et al 2015) and analyses of differential gene expression

(Kelley et al 2016) have indicated high congruence of biological

processes associated with candidate genes across lineages in differ-

ent river drainages indicating that evolution has repeatedly and pre-

dictably modified the same physiological pathways during adaptation

to environmental H2S However the signal of evolutionary conver-

gence among sulphide spring lineages decreases with the level of

organismal organization (from physiological pathways down to geno-

mic regions genes and individual SNPs in selection scans Pfenninger

et al 2015 from physiological pathways down to genes and tran-

scripts in expression studies Passow Brown et al 2017) indicating

that different lineages have often modified the same physiological

pathways in different ways While knowledge about the biochemical

TABLE 1 Adaptation to H2S can occur through three principlemechanisms (1) Resistance involves direct toxicity targets whosefunctions are impaired by H2S Adaptive modifications of thesetoxicity targets are predicted to assure proper function despitethe presence of H2S (2) Mitigation involves alternativephysiological pathways that compensate for the inhibition ofdirect toxicity targets Adaptive modifications of these alternativephysiological pathways are predicted to allow for the maintenanceof organismal function by other means (3) Regulation involvesphysiological pathways associated with the maintenance ofendogenous H2S homeostasis Adaptive modifications ofregulatory pathways are predicted to maintain low endogenousconcentrations despite continuous flux of H2S from theenvironment into the body

Candidate pathways Evidence

(1) Resistance

OXPHOS bull Positive selection on COX in some lineages of

P mexicana (Puyacatengo and Pichucalco)a as

well as some other genera of sulphide spring

poeciliidsb COX amino acid substitutions in P

mexicana allow for the maintenance of function

in the presence of H2Sa

bull Upregulation of cytochrome c and some COX

subunits as well as downregulation of compo-

nents of complex I in all sulphide spring lineages

of P mexicana indicating complex changes in

aerobic ATP productionc

Oxygen transport

genesbull Positive selection on and upregulation of

several oxygen transport genes (including myo-

globin and multiple haemoglobin subunits) in

sulphide spring lineages of the genera

Gambusia Limia Poecilia and Pseudoxiphophorus

indicating modifications to putatively maintain

oxygen transport in the presence of H2Sd

(2) Mitigation

Anaerobic

metabolismbull Upregulation of genes associated with glycolysis

and gluconeogenesis in all sulphide spring lin-

eages of P mexicana indicating increased

capacity of anaerobic ATP productionc

Oxidative stress

responsesbull Upregulation of genes associated with

oxidative stress responses in all sulphide

spring lineages of P mexicana indicating

increased capacity to cope with H2S-induced

oxidative stressc

(3) Regulation

H2S detoxification bull Positive selection on and upregulation of sev-

eral genes associated with enzymatic H2S

detoxification in all sulphide spring lineages of

P mexicanace Upregulation of H2S detoxifica-

tion enzymes in other sulphide spring lineages

of the genera Gambusia Limia Poecilia Pseudox-

iphophorus and Xiphophorusb This indicates an

increased ability to maintain low endogenous

H2S concentrations during exposure

H2S production bull No evidence for modifications of genes associ-

ated with enzymatic production of H2Sc

aPfenninger et al (2014) bR Greenway J L Kelley amp M Tobler unpub-

lished data cKelley et al (2016) dBarts et al (2018) ePfenninger et al

(2015)

TOBLER ET AL | 847

and physiological consequences of H2S may thus enable us to pre-

dict the physiological pathways in which adaptive modifications

occur it may be impossible to predict specific modifications within

those pathways This contrasts with a common expectation in the

field that convergent evolution of molecular mechanisms underlying

phenotypic traits (ie parallel evolution) necessarily hinges on modi-

fication of the same loci across lineages as found in some of the

best-known examples of parallelism (Colosimo et al 2005 Feldman

Brodie Brodie amp Pfrender 2012 Grant Grant Markert Keller amp

Petren 2004) However modification of the same loci may only be

common under particular circumstances (Rosenblum Parent amp

Brandt 2014) for example when selection acts on standing genetic

variation (Colosimo et al 2005) when adaptive alleles are intro-

gressed across population boundaries through gene flow (Grant

et al 2004) or when there are strong genetic constraints (Feldman

et al 2012) Exploration of molecular mechanisms underlying con-

vergent evolution in many systems may therefore be better

approached from the perspective of interacting genes that are orga-

nized in pathways in which alternative genetic changes can have

equivalent functional consequences (ie functional redundancy)

Developing and testing hypotheses about parallel evolution at the

pathway level will hinge on improving genome annotations (Pavey

Bernatchez Aubin-Horth amp Landry 2012) and provide a more

nuanced understanding of the factors determining the predictability

of evolution at different levels of organismal organization (Rosen-

blum et al 2014)

Despite evidence for convergent evolution in some candidate

physiological pathways it also needs to be emphasized that this

does not apply to all P mexicana lineages or to all candidate physio-

logical pathways associated with H2S adaptation investigated to

date Nonconvergent evolutionmdashevidenced by unique lineage-speci-

fic signals of selection and differential expressionmdashmay be indicative

of unique molecular mechanisms to cope with H2S toxicity For

example signatures of positive selection on COXmdashH2Srsquos primary

toxicity targetmdashcan be found in sulphide spring lineages of the Puy-

acatengo and Pichucalco river drainages (a lineage with recent diver-

gence time and a lineage with older divergence time respectively)

but not in those of the Tacotalpa drainage (Pfenninger et al 2014)

The presence of nonconvergent genetic and gene expression differ-

ences between populations from sulphidic and non-sulphidic habitats

in different river drainages highlights the limitations of inferences

that can be drawn based on data sets generated with high-through-

put DNA and RNA sequencing The potential importance of modifi-

cations at candidate locimdashor the lack thereofmdashcan only be assessed

through functional assays that document how alternative alleles and

changes in gene expression actually affect organismal performance

The functional consequences of genetic and gene expression varia-

tion at most candidate loci remain to be investigated However

enzyme activity assays have indicated that COX maintains its activity

with increasing H2S concentrations in the two sulphide spring popu-

lations with evidence for positive selection on mitochondrially

encoded subunits (Puyacatengo and Pichucalco Pfenninger et al

2014) Key amino acid substitutions occur in the D-pathway channel

of COX1 which putatively narrow the channel width and reduce

H2Srsquos access to the binuclear centre of the protein Consequently

some lineages of P mexicana exhibit an H2S-resistant COX that can

maintain its activity and aerobic ATP production even if endogenous

H2S concentrations are elevated In contrast COX activity in sul-

phide spring fish of the Tacotalpa river drainage and in fish from

nonsulphidic habitats decreases when H2S is present which is con-

sistent with mechanisms of H2S toxicity (Pfenninger et al 2014) In

vitro assays on isolated mitochondria have nonetheless indicated

that an H2S-resistant COX is not necessary to maintain aerobic ATP

production during H2S exposure Unlike in populations from nonsul-

phidic habitats mitochondrial spare respiratory capacity is main-

tained in the presence of H2S even in the Tacotalpa sulphide spring

populations that lack an H2S-resistant COX (C Henpita M Tobler amp

JH Shaw unpublished data) This may indicate that modification of

detoxification pathways in sulphide spring fishes is sufficient to

maintain mitochondrial function and aerobic ATP production

23 | Ensuing challenges linking genotypes tofitness

Understanding the mechanistic basis of adaptation requires linking

genotypic variation to phenotypic traits that ultimately shape fitness

along selective gradients The evolution of H2S tolerance in sulphide

spring fishes appears to be a consequence of modifications in many

genes and many physiological pathways whose functions interact to

eliminate H2S from the system minimize effects of elevated endoge-

nous concentrations on direct toxicity targets and mitigate adverse

effects that cascade through cellular metabolism during disruption of

homeostasis (Table 1) Such complex genetic changes are likely com-

mon during adaptation and have been documented in other systems

that involve major evolutionary shifts including the emergence of

parasitic lifestyles (Werren et al 2010) eusociality (Simola et al

2013) asexual reproduction (Fradin et al 2017) and internal gesta-

tion with livebearing (Schartl et al 2013) Consequently comple-

mentary approaches will be required to uncover the molecular

mechanisms of adaptation to H2S in the future

1 The involvement of many genes and physiological pathways in

shaping tolerance emphasizes the importance of quantitative

genetic approaches that leverage experimental crosses between

H2S tolerant and nontolerant populations and test whether toler-

ancemdashand adaptive trait modifications that contribute to itmdashis

explained by a few major effects loci or whether it is a conse-

quence of a number of genes with small effects (Klerks Xie amp

Levinton 2011 Saltz Hessel amp Kelly 2017) Combining analyses

of phenotype-fitness associations and their underlying genetic

mechanisms via QTL mapping with the results of selection scans

from population genomic analyses (Pfenninger et al 2015) will

allow for an integrative understanding of how selection acting on

the genome results in phenotypic variation that mediates adapta-

tion to different environments (Stinchcombe amp Hoekstra 2008)

From a functional perspective breaking up correlations between

848 | TOBLER ET AL

different physiological traits (associated with resistance mitiga-

tion and regulation Table 1) in F2 crosses will further allow test-

ing what molecular mechanisms are necessary for tolerance

whether there are alternative strategies with equivalent fitness

benefits and whether there are complementary modifications

with additive effects on tolerance

2 Considering the number of genes and physiological pathways

likely involved in H2S tolerance assessing the functional conse-

quences of genetic and gene expression variation is a daunting

task At the same time such functional tests can significantly

contribute to closing the genotype-to-phenotype gap because

physiochemical stress responses can be assessed along the con-

tinuum of organismal organization (Gjuvsland Vik Beard Hunter

amp Omholt 2013) Standardized assays can be used to quantify

the biochemical makeup of organelles cells and tissues (eg pro-

tein abundance and enzyme activity) physiological performance

in vitro and in vivo (eg rates of H2S detoxification ATP produc-

tion and reactive oxygen production) and fitness-relevant proxies

of whole organism performance (eg H2S tolerance Figure 2)

This highlights the importance of methods development for the

quantification of complex phenotypic traits or even whole phe-

nomes that can keep pace with rapid data generation through

high-throughput sequencing (Houle Govindaraju amp Omholt

2010 Ritchie Holzinger Li Pendergrass amp Kim 2015) In addi-

tion genotypendashphenotype links established through such correla-

tive approaches will ideally be verified through targeted gene

editing While gene editing with CRISPRndashCas9 is now routinely

employed in model organisms (Doudna amp Charpentier 2014 Sha-

lem Sanjana amp Zhang 2015) applying this technology to non-

models still remains challenging especially in livebearing species

like poeciliid fishes where in vitro culture of embryos is not rou-

tinely employed (Martyn Weigel amp Dreyer 2006) Nonetheless

the effects of specific mutations can potentially be verified

through the establishment and manipulation of cells in culture

(Liu et al 2014) or by generating transgenics (Lin et al 2016)

3 Organismal responses to physiochemical stressors occur at multi-

ple hierarchical levels that can be plastically modulated in

response to endogenous and exogenous cues (Whitehead 2012)

Understanding adaptation to physiochemical stressors conse-

quently requires disentangling the effects of genetic variation

developmental plasticity and short-term acclimation in gene

expression patterns on the emerging physiological processes

Common garden experiments on sulphide spring fishes have

shown that many genes differentially expressed between popula-

tions from sulphidic and nonsulphidic habitats in nature remain

differentially expressed in laboratory-reared fish that have never

been exposed to H2S indicating evolved differences in constitu-

tive gene expression (Passow Henpita et al 2017) At the same

time expression patterns in the laboratory also exhibit evidence

for plasticity in response to short-term H2S exposure (Passow

Henpita et al 2017) Some of these plastic responses are shared

between populations from sulphidic and nonsulphidic habitats

(likely representing shared ancestral plasticity) while others have

been lost or gained during sulphide spring colonization A core

challenge for future research is to overcome logistical constraints

that have so far prevented long-term H2S exposures of fish in

the laboratory Exposing individuals throughout development will

be required to test how developmental plasticity and short-term

acclimation shape responses to H2S in individuals from adapted

and nonadapted populations Finally the next big scientific

advance in this context will also require the identification of reg-

ulatory mechanisms that shape expression differences during

adaptation including sequence changes in transcription factors

and their binding sites patterns of DNA methylation and chro-

matin accessibility as well as the production of microRNAs and

competing endogenous RNAs that can modulate transcript abun-

dance (Flores Wolschin amp Amdam 2013 Lopez-Maury Mar-

guerat amp Beuroahler 2008 Silva Bye el Azzouzi amp Wisloslashff 2017)

In addition the role of alternative splicing and other post-tran-

scriptional and post-translational modifications in mediating adap-

tation remains understudied by evolutionary biologists (Ast

2004 Bush Chen Tovar-Corona amp Urritia 2017) This is particu-

larly relevant because H2S has been hypothesized to play a role

in the post-translational modification of proteins with sulphur-

containing amino acids through sulfhydration (Mustafa et al

2009)

Bridging the genotypendashphenotype gap remains one of the major

challenges in biology (Gjuvsland et al 2013 Ritchie et al 2015) and

will provide opportunities to understand how organisms function in

the face of environmental stress and how evolution shapes these

functions during adaptation to different environmental conditions

Establishing the molecular mechanisms of adaptation across repli-

cated lineages exposed to the same sources of selection will also

provide insights into how evolutionary history demography and

genetics shape convergent evolutionary outcomes providing a better

understanding at what hierarchical levels of organismal organization

evolutionary change may actually be predictable (Rosenblum et al

2014)

3 | COMPLEX ORGANISMS IN A COMPLEXWORLD DISENTANGLING THE TANGLEDBANK

Mechanisms of adaptation can be studied through bottom-up

(screening for genetic modifications that allow for inferences about

relevant phenotypic traits) and top-down approaches (assessing fit-

ness-relevant phenotypic differences among populations without

knowledge about the genetic basis see Schluter 2009) Over the

years genomic and transcriptomic studies on sulphide spring popula-

tions of the P mexicana species complex have been complemented

with analyses of phenotypic variation Just as genetic analyses have

indicated that adaptation to H2S is mediated by the modification

and modulation of many genes associated with multiple physiological

pathways phenotypic studies uncovered a variety of differences

TOBLER ET AL | 849

between populations from sulphidic and nonsulphidic environments

Many of these phenotypic differences have evolved in convergence

across replicated sulphide spring lineages in different river drainages

Examples include modifications of morphological (body shape and

organ size Tobler et al 2011 Tobler Scharnweber et al 2015

Schulz-Mirbach et al 2016) physiological (tolerance and metabolic

rates Tobler et al 2011 Passow Arias-Rodriguez amp Tobler 2017)

behavioural (aggression and personality Riesch et al 2009 Bierbach

et al 2012 Sommer-Trembo et al 2016 Bierbach Arias Rodriguez

amp Plath 2018) and life history traits (fecundity and offspring size

Riesch Plath Garcia de Leon amp Schlupp 2010 Riesch et al 2014)

Hence proximate populations in different habitat types have

diverged in complex phenotypes that span many different suites of

traits and a key question is how divergence in multivariate pheno-

types arises In theory H2S could be a key source of selection driv-

ing multivariate phenotypic divergence either because it directly

exerts selection on a multitude of traits that are independently modi-

fied to reach new fitness peaks or because selection on a few fit-

ness-relevant traits causes simultaneous changes in correlated

characteristics Neither explanation seems sufficient because many

Abiotic environmentOxygen concentrations

Ionic compositionTemperature

Structure and flow

Biotic environmentTrophic resources

DemographyPredationParasitism

H2S

Genetic variation

Gene expression

Biochemistry

Physiologicalfunction

Organismal performance

Primary fitness components

Survivalgrowth

reproduction

Organismal function Environment

F IGURE 2 Understanding how organisms function in the context of their environment is a core question in evolutionary biology yet thecomplexities associated with organismal function and environmental variation provide significant challenges High-throughput sequencing hasallowed for the identification of candidate genes subject to selection and differential expression during adaptation to H2S The next step is tobridge the genotype-to-phenotype gap by quantifying the integrated responses to H2S across the continuum of organismal organization includingbiochemical phenotypes physiological functions whole organism performance and ultimately fitness In addition understanding adaptation toH2S-rich environments will also require acknowledging H2Srsquos relation to other abiotic and biotic environmental variables which can exert selectionand modulate aspects of organismal function This will allow for the establishment of cause-and-effect relationships between specific sources ofselection and evolutionary change in specific traits and contribute to our understanding of organismal function in natural systems

850 | TOBLER ET AL

of the divergent traits between populations from sulphidic and non-

sulphidic environments have no clearmdashor at best tenuousmdashlinks to

H2S toxicity and detoxification and there is only limited evidence

for within-population correlations between different traits groups

(Riesch et al 2016)

Evolutionary analyses frequently focus on seemingly simple

sources of selection juxtaposing populations that are assigned to dis-

crete habitat types to understand the evolution of specific traits

(Kaeuffer Peichel Bolnick amp Hendry 2012) But as Darwin (1859)

already observed in his ldquotangled bankrdquo allegory selection in natural

systems is rarely univariate but multifarious selective regimes con-

tribute to trait divergence even among populations that occur along

apparently simple environmental gradients Unquantified environmen-

tal variation that is subsumed in the categorization of habitat types

can profoundly affect trait variation in natural systems (Stuart et al

2017) highlighting the importance of considering and testing alterna-

tive hypotheses about the cause-and-effect relationship between

specific sources of selection and evolution of specific traits Thus

accurate natural history observations that consider organisms as holis-

tic entities consisting of interacting traits in the context of the abiotic

and biotic factors comprising their environment are critical to inform

the design and interpretation of empirical research in natural systems

(Greene 2005) Indeed adjacent sulphidic and nonsulphidic habitat

types do not merely differ in the presence or absence of H2S but also

in a variety of other abiotic and biotic factors (Figure 2) Unlike nonsul-

phidic habitats sulphide springs in southern Mexico are characterized

by extreme hypoxia elevated salinity and lower pH (Rosales Lagarde

2012 Tobler et al 2011) In addition the presence of physiochemical

stressors affects the biotic communities in these sulphide springs Spe-

cies richness of aquatic invertebrates and vertebrates is drastically

reduced as a consequence of H2S toxicity (Greenway Arias-Rodriguez

Diaz amp Tobler 2014) and primary production appears to be driven in

part by chemoautotrophic bacteria that use H2S as an energy source

rather than just by autochthonous and allochthonous photosynthetic

organisms (Roach Tobler amp Winemiller 2011) These changes in the

biotic environment affect the relative importance of intra-vs inter-

specific competition trophic resource use and exposure to predators

and parasites (Riesch Oranth et al 2010 Tobler et al 2014 Tobler

Scharnweber et al 2015)

31 | Alternative drivers of trait evolution

Acknowledging the complexity of selective regimes in sulphide

springs allows for a more nuanced understanding of trait evolution

and for the development of alternative hypotheses about the ulti-

mate mechanisms that cause divergence in specific genetic and phe-

notypic traits In some instances sources of selection other than H2S

can plausibly explain trait divergence between populations from sul-

phidic and nonsulphidic habitats For example hypoxia likely drives

increases in head size and gill surface area (Tobler et al 2011)

changes in the ionic composition of the water can explain differential

expression of genes associated with osmoregulation (Kelley et al

2016) and shifts in trophic resource use are associated with

modifications of the gastrointestinal system (Tobler Scharnweber

et al 2015) In other cases disentangling alternative hypotheses is

not trivial because selection mediated by H2S and other environmen-

tal factors are predicted to affect trait modifications in similar ways

Shifts towards anaerobic metabolism in sulphide spring fish could be

the consequence of inhibition of aerobic ATP production by H2S or

by constraints in oxygen availability Similarly reproductive life his-

tory shifts towards the production of larger but fewer offspring could

be driven by H2S (decreasing offspringrsquos surfacevolume ratio to mini-

mize the diffusion of H2S into the body Powell 1989) by habitat dif-

ferences in the quality and quantity of predation that can affect size-

specific rates of extrinsic mortality (Jorgensen Auer amp Reznick

2011) or by energetic constraints (see below) While comparisons

across different selective regimes can shed light into the ultimate dri-

vers of trait evolution (Moore Riesch amp Martin 2016) isolating the

effects of different sources of selection may not be possible because

they could act in concert and facilitate increases of organismal perfor-

mance in multiple contexts (Ghalambor Walker amp Reznick 2003)

When multifarious selection shapes multivariate phenotypic evo-

lution it is important to also consider the role of trade-offs both in

terms of trait expression (increased investment in one trait may

impose constraints on the expression of other traits) and in terms of

trait function (optimizing the trait for one function may impair the

traitrsquos utility for other functions) For example selection on large off-

spring size in sulphide spring fishes has inadvertent consequences

for fecundity (Riesch et al 2014) and selection for increased gill

surface areas due to hypoxia simultaneously increases exposure to

environmental H2S (Tobler et al 2011) Thus the outcome of phe-

notypic evolution is ultimately a result of balancing competing needs

in a manner that optimizes trait expression and organismal function

in multiple contexts to maximize fitness (Arnold 1983) and the pres-

ence of trade-offs can have cascading indirect effects with selection

on one or a few traits leading to the modification of a multitude of

apparently unrelated characteristics

In this context bioenergetics may represent a nexus in the evo-

lution of complex traits observed in sulphide spring fish and other

extreme environments Living in and adapting to sulphide springs

imposes constraints on organismal energy availability (Plath et al

2007) with according trade-offs for energy allocation Energy limita-

tion in sulphide springs does not arise from a scarcity of food

resources (chemoautotrophic bacteria eaten by the fish are abundant

in sulphide springs Tobler Scharnweber et al 2015) but from

behavioural and physiological adaptation A primary survival strategy

for fish in the toxic and hypoxic springs is to engage in aquatic sur-

face respiration where individuals breathe directly at the water sur-

face to maximize oxygen uptake and minimize H2S exposure This

behaviour mediates short-term survival (Plath et al 2007) but it

also constrains energy acquisition for benthic foragers like P mexi-

cana (Tobler Riesch Tobler amp Plath 2009) and increases susceptibil-

ity to bird predation (Riesch Oranth et al 2010) Costly investments

into H2S detoxification and tissue repair as well as a shift from aero-

bic to less efficient anaerobic metabolism further constrain energy

allocation to other functions (Passow Greenway Arias-Rodriguez

TOBLER ET AL | 851

Jeyasingh amp Tobler 2015) Theory predicts that energetic con-

straints ndash whether mediated by limitations in the rate of resource

acquisition or in the rate of energy allocation to reproduction (as

opposed to maintenance or growth) ndash should exert selection for a

reduction in organismal energy budgets which allows for the maxi-

mization of residual energy allocation for the production of offspring

(Brown Marquet amp Taper 1993) Several lines of evidence support

the hypothesis that energy limitation plays a critical role in pheno-

typic evolution of sulphide spring fish Compared to individuals in

nonsulphidic environments fish in sulphide springs have significantly

lower routine energy demands which is mediated both by reduc-

tions in body size and mass-specific metabolism (Passow Arias-

Rodriguez et al 2017) Reductions in other traits such as energeti-

cally costly behaviours (Bierbach et al 2012 Plath 2008) and the

size of energy-demanding organs (Schulz-Mirbach et al 2016) may

also contribute to the reduced energy demands and shifts in repro-

ductive life history traits (Riesch et al 2014) are consistent with

selection from energy limitation (Reznick amp Yang 1993)

32 | Ensuing challenges Disentanglingenvironmental and organismal complexities

Future research needs to empirically address how multifarious selec-

tive regimes shape the evolution of organisms with complex pheno-

types We have made great advances in our understanding of how

specific traits evolve and mediate adaptation (eg morphology Chan

et al 2010 coloration Dasmahapatra et al 2012 sensory organs

Cartwright Schwartz Merry amp Howell 2017) and we have a thor-

ough understanding of how single environmental factors can shape

organismal responses (eg climatic conditions Johnston amp Bennett

1996 predation Langerhans 2007 competition Pfennig amp Pfennig

2012) In contrast our understanding of phenotypic evolution in

response to multifarious selection is comparatively limited especially

because interactive effects of different environmental factors can lead

to organismal and evolutionary responses that are unpredictable based

on experiments that manipulate single environmental variables in iso-

lation (Gunderson Armstrong amp Stillman 2016 Todgham amp Stillman

2013) Future research thus needs to acknowledge the complexities of

environments and organismal design and address them explicitly in

empirical investigations to gain a robust understanding of biological

diversification (Culumber amp Tobler 2017 Stuart et al 2017)

4 | LINKING ADAPTATION TOSPECIATION

So far we primarily explored the causes of adaptive evolution but

what are its consequences Theory predicts that adaptation to local

environmental conditions can lead to speciation because reproduc-

tive isolation emerges incidentally as a by-product (ecological specia-

tion Nosil 2012) Population genetic analyses of Poecilia spp in

sulphide springs and adjacent populations in nonsulphidic environ-

ments have indicated strong genetic differentiation across small

spatial scales (in some instances lt100 m) and in the absence of

physical barriers that would prevent fish movement (Plath et al

2013) In addition population genetic structuring has been shown to

be stable even in the face of major environmental disturbance

BOX 1 Speciation in sulphide springs and conservation

biology

Adaptation to H2S-rich environments in poeciliids coincides

with ecological speciation giving rise to phenotypically dis-

tinct and reproductively isolated lineages In some

instances this has led to the evolution of highly distinct

micro-endemic species including Poecilia sulphuraria and

P thermalis Limia sulphurophila and Gambusia eurystoma

(Figure 3) Except for P sulphuraria which is found in two

spring complexes of the Rıo Pichucalco drainage of Mexico

(Tobler amp Plath 2009b) the distributions of these sulphide

spring endemics are restricted to single localities that

include only small areas of suitable habitat For example

the largest sulphide spring complex (the Ba~nos del Azufre

in southern Mexico which harbours P sulphuraria and G

eurystoma) includes ~18000 m2 of suitable habitat (equiva-

lent to lt35 football fields) while one of the smallest (La

Esperanza springs harbouring P thermalis) merely covers

~600 m2 (just over a tenth of a football field Brown et al

2017) Accordingly some sulphide spring endemics are

considered critically endangered (IUCN 2012) and pro-

tected by law (SEDESOL 1994)

Many threatened species with restricted ranges exhibit

low genetic diversity and high incidences of inbreeding

(Bouzat 2010 Willi Van Buskirk amp Hoffmann 2006) In

contrast at least P sulphuraria and P thermalismdashthe only

species investigated to datemdashactually exhibit relatively high

effective population sizes (Ne = 1200ndash3800 depending on

the population) and no evidence for inbreeding (FIS lt 0)

(Brown et al 2017) This is likely a consequence of the

long-term stability of sulphide springs and the high pro-

ductivity and low levels of interspecific competition also

support unusually high fish densities and thus population

census sizes (Brown et al 2017 Culumber et al 2016)

Nonetheless sulphide spring endemics are threatened by

habitat destruction which is mediated primarily by the

modification of springs for recreational purposes defor-

estation and conversion of land for agriculture and most

recently high-intensity palm oil plantations (Tobler amp Plath

2009ab) Despite the official protection of sulphide spring

endemics there have been no tangible conservation mea-

sures that foster the sustainable development of land

around sulphide springs with the well-being of the endemic

fishes in mind

852 | TOBLER ET AL

(catastrophic flood event Plath et al 2010) Population genetic dif-

ferentiation between adjacent sulphide-adapted and nonadapted

populations is indicative of ongoing ecological speciation (Box 1) but

different population pairs from sulphidic and nonsulphidic waters

have made variable progress along the speciation continuum towards

complete reproductive isolation (Plath et al 2013) This prompts

questions about the mechanisms involved in reducing gene flow

along the environmental gradient connecting the two habitat types

41 | Prezygotic mechanisms of reproductiveisolation

The simplest form of reproductive isolation between locally adapted

populations arises from selection against migrants between habitat

types (Nosil Vines amp Funk 2005) Translocation experiments in nat-

ural habitats have indicated strong survival selection against

migrants especially those translocated from nonsulphidic to sulphidic

habitats (Plath et al 2013) Fish translocated from nonsulphidic

environments face high mortality within short periods of time

(lt24 hr) which is likely caused by H2S toxicity and hypoxia in sul-

phide springs Hence low gene flow from populations inhabiting

nonsulphidic waters into populations inhabiting sulphidic habitats is

simply explained by a lack of adaptation to the extreme environmen-

tal conditions At least in some drainages there is also appreciable

mortality for sulphide spring fish translocated to nonsulphidic habi-

tats although mechanisms causing inviability are unclear and the

strength of selection appears to be lower (Plath et al 2013) If sur-

vival selection against immigrants was the sole source of reproduc-

tive isolation this should result in unidirectional gene flow from

H2S-adapted to nonadapted populations but population genetic

analyses provided no evidence for this implying that additional

mechanisms are at work (Plath et al 2013)

One such mechanism could be intersexual selection and mate

choice experiments have indicated that migrants from sulphidic to

nonsulphidic habitats are at a significant disadvantage with fish from

nonsulphidic habitats avoiding potential mating partners from sul-

phidic ones (Plath et al 2013 Sommer-Trembo et al 2016) Such

assortative mating has been found to be critical during ecological spe-

ciation because it counteracts the homogenizing effects of gene flow

(van Doorn Edelaar amp Weissing 2009) yet how assortative mating

patterns arise often remains unclear In theory assortative mating can

arise through a variety of processes (Nosil 2012) ldquoMagic traitrdquo mecha-

nisms where mating preferences evolve as a by-product of divergent

selection on ecological traits that serve as cues have recently garnered

much attention (Servedio van Doorn Kopp Frame amp Nosil 2011)

Consistent with magic trait mechanisms assortative mating in P mexi-

cana appears to be linked to adaptive population differences in body

shape which serve as cues to distinguish potential mating partners

from different habitats (Greenway Drexler Arias-Rodriguez amp Tobler

2016) However neither sulphide spring fish (which are unlikely to

encounter migrants from nonsulphidic habitats due to H2S toxicity)

nor allopatric P mexicana from nonsulphidic waters (which have never

experienced migrants from sulphide spring populations) exhibit a

preference for mating partners from their own ecotype suggesting

that assortative mating in this system is a product of reinforcement

(ie direct selection for assortative mating Greenway et al 2016)

This is different from classic magic traits mechanisms because assorta-

tive mating is not a mere by-product of adaptation but adaptive traits

have instead been co-opted as mating signals in response to selection

on mating preferences (Maan amp Seehausen 2012) Moreover males of

some H2S-adapted populations have reduced aggressive behaviour

(Bierbach et al 2012 2018) In staged contests of size-matched males

from different ecotypes intruders had a significantly lower likelihood

to establish dominance than resident males independent of whether

or not they showed reduced aggressiveness overall Hence intrasexual

selection may represent another mechanism mediating reproductive

isolation between locally adapted ecotypes Overall H2S-adapted poe-

ciliids allow disentangling the relative contributions of different forms

of ecological and sexual selectionmdashand their interdependencymdashduring

ecological speciation

42 | Ensuing challenges quantifying postzygoticmechanisms of reproductive isolation

In contrast to prezygotic mechanisms little is known about postzy-

gotic mechanisms of reproductive isolation that potentially act to

reduce gene flow between adjacent populations in sulphidic and

nonsulphidic habitats Putative hybrids between sulphidic and non-

sulphidic ecotypes have occasionally been observed in nonsulphidic

habitats immediately adjacent to confluences with sulphide springs

(R Riesch M Plath and M Tobler personal observations) but it

remains unclear whether and through what mechanisms such hybrids

exhibit reduced fitness compared to parentals

Direct selection of mitochondrially encoded genes in the pres-

ence of H2S (Pfenninger et al 2014) should be accompanied by co-

evolution at nuclear loci that directly interact with mitochondrial

gene products in the formation of protein complexes involved in the

OXPHOS pathway (Sunnucks Morales Lamb Pavlova amp Greening

2017 Willett amp Burton 2004) Disruption of co-evolved gene com-

plexes in hybrids should therefore lead to intrinsic fitness reductions

through mito-nuclear incompatibilities (Burton Pereira amp Barreto

2013 Wolff Ladoukakis Enrıquez amp Dowling 2014) which have

been hypothesized to be a key driver in speciation (Chou amp Leu

2010 Gershoni Templeton amp Mishmar 2009) Empirical tests of

these ideas remain to be conducted in sulphide spring fishes just

like studies on potential ecological and sexual selection against

hybrids that arise from interpopulation mating events

5 | PERSPECTIVES LIFE IN SULPHIDICENVIRONMENTS BEYOND P MEXICANA

Research on sulphide spring fishes of the P mexicana species com-

plex in Mexico has provided insights into mechanisms underlying

adaptation and the links between adaptation and speciation Sul-

phide springs do not only occur in Mexico however but are found

TOBLER ET AL | 853

worldwide and have been colonized by 24 evolutionarily indepen-

dent fish lineages in North America the Neotropics and the Middle

East (Greenway et al 2014) The vast majority of lineages belong to

the order Cyprinodontiformes (95) and livebearing members of the

family Poeciliidae in particular (80) This potentially indicates that

internal gestation of young has facilitated the colonization of

extreme environments because it shields developing embryos from

physiochemical stressors (Blackburn 1999) Among the Poeciliidae

sulphide spring fishes include members of the genera Brachyrhaphis

Gambusia Poeciliopsis Priapichthys Pseudoxiphophorus Xiphophorus

and multiple subgenera of Poecilia (Figure 3 Greenway et al 2014)

spanning over 40 million years of evolution (based on Reznick Fur-

ness Meredith amp Springer 2017)

A handful of studies have started to address molecular and phe-

notypic evolution across a broader taxon sampling of sulphide spring

fishes finding evidence for convergent evolution in morphology

(Riesch et al 2016 Tobler amp Hastings 2011) life history traits

(Riesch Plath et al 2010 Riesch et al 2014) and oxygen transport

genes (Barts et al 2018) In addition ecological speciation in

response to adaptation to H2S-rich environments has also been doc-

umented in sulphide spring poeciliids other than P mexicana (Riesch

et al 2016) These studies have laid a fruitful foundation for the

investigation of mechanisms of adaptation to H2S-rich springs from

genes to phenotypic traits linking micro-evolutionary processes to

macro-evolutionary patterns

The availability of broad-scale evolutionary replication of popula-

tions that have invaded habitats with the same physiochemical stressors

allows for addressing some of the ensuing challenges discussed above

and investigating the factors that give rise to predictable evolutionary

outcomes Convergent evolution is a commonmdashbut not omnipresentmdash

theme in adaptive evolution and multiple factors can determine the

degree of convergence among replicated lineages exposed to the same

sources of selection (Kaeuffer et al 2012 Landry amp Bernatchez 2010

Oke Rolshausen LeBlond amp Hendry 2017 Rosenblum amp Harmon

2011) Thus leveraging replicated lineages that have colonized sulphide

springs to assess molecular mechanisms underlying convergent pheno-

typic changes analyse how environmental heterogeneity among sul-

phide spring habitats causes lineage-specific patterns of trait evolution

and test whether and by what mechanisms adaptation to extreme envi-

ronments is linked to speciation will provide insights into how ecologi-

cal genetic and functional factors contribute to evolutionary

convergence at different levels of organismal organization (Rosenblum

et al 2014) An expanded taxon sampling with population pairs that

span almost the full range of the speciation continuum from panmixia to

complete reproductive isolation (Riesch et al 2016) will also help

addressing questions about temporal aspects of H2S adaptation In par-

ticular comparing younger to older lineages will provide insights about

the order at which genetic and phenotypic modifications accrue during

the adaptation process

Animals have also invaded H2S-rich habitats in other aquatic

ecosystems including cold seeps and deep-sea hydrothermal vents

(Tobler et al 2016) Adaptation to H2S in the P mexicana species com-

plex has involved modification of physiological pathways associated

with H2S toxicity (OXPHOS pathway) and detoxification (sulphide

F IGURE 3 Phylogeny of the family Poeciliidae (adopted from Culumber amp Tobler 2017) Various lineages in the genera Poecilia (subgeneraAcanthophacelus Limia and Mollienesia) Priapichthys Brachyrhaphis Poeciliopsis Xiphophorus Pseudoxiphophorus and Gambusia have colonizedH2S-rich springs in different parts of the Americas (see Greenway et al 2014) Lineages highlighted in yellow represent endemic speciesrestricted to H2S-rich springs while red lineages represent species in which some populations have invaded sulphidic habitats Pictures depict arepresentative of each focal genussubgenus

854 | TOBLER ET AL

quinone oxidoreductase pathway) which have been highly conserved

during the diversification of life on Earth (Saraste 1999 Shahak amp

Hauska 2008) The high replicability of H2S as an environmental factor

that shaped animal evolution across vastly different ecological contexts

consequently offers the opportunity to investigate whether selection

mediated by the presence of H2S has repeatedly modified these con-

served physiological pathways across animal phyla that have diverged

during the Cambrian explosion spanning over 500 million years of evo-

lution Such comparisons would provide unprecedented insights into

the effects of a single physiochemical stressor on evolutionary change

irrespective of ecological or phylogenetic contexts

6 | CONCLUSIONS

H2S-rich environments and their inhabitants represent useful models

for research in organismal biology ecology and evolution illustrating

both opportunities (eg strong and explicit source of selection evolu-

tionary replication) and challenges (eg complexity of environments

and organismal responses) in asking basic question about the origins

of biodiversity Understanding how organisms thrive in apparently

inhospitable environments will provide us with insights into lifersquos

capacities and limitations to adapt to novel environments (Waterman

1999) This is particularly relevant considering that human activities

substantially contribute to the exacerbation of physiochemical stres-

sors at local regional and global scales impacting ecological and evo-

lutionary dynamics (Alberti 2015 Norberg Urban Vellend

Klausmeier amp Loeuille 2012) Thus a focus on ecological and evolu-

tionary consequences of natural stressors that push organisms to the

brink of their physiological capacities may be instrumental to predict

consequences of anthropogenic environmental change

ACKNOWLEDGEMENTS

We would like to thank Lenin Arias-Rodriguez the Universidad

Juarez Autonoma de Tabasco and the Centro de Investigacion E

Innovacion Para la Ense~nanza y El Aprendizaje for their continued

collaboration enabling our work on the P mexicana system Funding

for this review article has been provided by the National Science

Foundation (IOS-1557860 IOS-1557795) and the US Army

Research Office (W911NF-15-1-0175)

AUTHOR CONTRIBUTIONS

MT wrote the first draft of the paper all authors contributed text

and comments

ORCID

Michael Tobler httporcidorg0000-0002-0326-0890

Joanna L Kelley httporcidorg0000-0002-7731-605X

Reuroudiger Riesch httporcidorg0000-0002-0223-1254

REFERENCES

Alberti M (2015) Eco-evolutionary dynamics in an urbanizing planet

Trends in Ecology amp Evolution 30 114ndash126 httpsdoiorg101016j

tree201411007

Alda F Reina R G Doadrio I amp Bermingham E (2013) Phylogeny

and biogeography of the Poecilia sphenops species complex

(Actinopterygii Poeciliidae) in Central America Molecular Phylogenet-

ics and Evolution 66 1011ndash1026 httpsdoiorg101016jympev

201212012

Alvarez del Villar J (1948) Descripcion de una nueva especie de Mol-

lienisia capturada en Ba~nos del Azufre Tabasco (Pisces Poeciliidae)

Anales de la Escuela Nacional de Ciencias Biologicas 5 275ndash281

Arnold S J (1983) Morphology performance and fitness American Zool-

ogist 23 347ndash361 httpsdoiorg101093icb232347

Aspiras A C Rohner N Martineau B Borowsky R L amp Clifford C J

(2015) Melanocortin 4 receptor mutations contribute to the adapta-

tion of cavefish to nutrient-poor conditions Proceedings of the

National Academy of Sciences USA 112 9968ndash9973

Ast G (2004) How did alternative splicing evolve Nature Reviews

Genetics 5 773ndash782 httpsdoiorg101038nrg1451

Bagarinao T (1992) Sulfide as an environmental factor and toxicant

Tolerance and adaptations in aquatic organisms Aquatic Toxicology

24 21ndash62 httpsdoiorg1010160166-445X(92)90015-F

Barton N H (1998) The effect of hitch-hiking on neutral genealogies

Genetic Research 72 123ndash133 httpsdoiorg101017

S0016672398003462

Barts N Greenway R Passow C N Arias Rodriguez L Kelley J L amp

Tobler M (2018) Expression and molecular evolution of oxygen

transport genes in livebearing fishes (Poeciliidae) from hydrogen sul-

fide rich springs Genome httpsdoiorg101139gen-2017-0051

[Epub ahead of print]

Beauchamp R O Bus J S Popp C S Boreiko C J amp Andjelkovich

D A (1984) A critical review of the literature on hydrogen sulfide

CRC Critical Reviews in Toxicology 13 25ndash97 httpsdoiorg10

310910408448409029321

Bell E M (2012) Life at extremes Environments organisms and strategies

for survival Oxfordshire CABI httpsdoiorg101079

97818459381470000

Bierbach D Arias Rodriguez L amp Plath M (2018) Intrasexual competi-

tion enhances reproductive isolation between locally adapted popula-

tions Current Zoology httpsdoiorg101093czzox071 [Epub

ahead of print]

Bierbach D Klein M Sassmannshausen V Schlupp I Riesch R

Parzefall J amp Plath M (2012) Divergent evolution of male aggres-

sive behavior Another reproductive isolation barrier in extremophile

poeciliid fishes International Journal of Evolutionary Biology 2012

148745

Blackburn D G (1999) Viviparity and oviparity Evolution and reproduc-

tive strategies In T E Knobil amp J D Neill (Eds) Encyclopedia of

reproduction Vol 4 (pp 994ndash1003) New York NY Academic Press

Bouzat J L (2010) Conservation genetics of population bottlenecks

The role of chance selection and history Conservation Genetics 11

463ndash478

Brown A P Greenway R Morgan S Quackenbush C R Giordani

L Arias-Rodriguez L Kelley J L (2017) Genome-scale data

reveals that endemic Poecilia populations from small sulfide spring

display no evidence of inbreeding Molecular Ecology 26 4920ndash

4934

Brown J H Marquet P A amp Taper M L (1993) Evolution of body

size Consequences of an energetic definition of fitness American

Naturalist 142 573ndash584 httpsdoiorg101086285558

Burton R S Pereira R J amp Barreto F S (2013) Cytonuclear genomic

interactions and hybrid breakdown Annual Review of Ecology

TOBLER ET AL | 855

and physiological consequences of H2S may thus enable us to pre-

dict the physiological pathways in which adaptive modifications

occur it may be impossible to predict specific modifications within

those pathways This contrasts with a common expectation in the

field that convergent evolution of molecular mechanisms underlying

phenotypic traits (ie parallel evolution) necessarily hinges on modi-

fication of the same loci across lineages as found in some of the

best-known examples of parallelism (Colosimo et al 2005 Feldman

Brodie Brodie amp Pfrender 2012 Grant Grant Markert Keller amp

Petren 2004) However modification of the same loci may only be

common under particular circumstances (Rosenblum Parent amp

Brandt 2014) for example when selection acts on standing genetic

variation (Colosimo et al 2005) when adaptive alleles are intro-

gressed across population boundaries through gene flow (Grant

et al 2004) or when there are strong genetic constraints (Feldman

et al 2012) Exploration of molecular mechanisms underlying con-

vergent evolution in many systems may therefore be better

approached from the perspective of interacting genes that are orga-

nized in pathways in which alternative genetic changes can have

equivalent functional consequences (ie functional redundancy)

Developing and testing hypotheses about parallel evolution at the

pathway level will hinge on improving genome annotations (Pavey

Bernatchez Aubin-Horth amp Landry 2012) and provide a more

nuanced understanding of the factors determining the predictability

of evolution at different levels of organismal organization (Rosen-

blum et al 2014)

Despite evidence for convergent evolution in some candidate

physiological pathways it also needs to be emphasized that this

does not apply to all P mexicana lineages or to all candidate physio-

logical pathways associated with H2S adaptation investigated to

date Nonconvergent evolutionmdashevidenced by unique lineage-speci-

fic signals of selection and differential expressionmdashmay be indicative

of unique molecular mechanisms to cope with H2S toxicity For

example signatures of positive selection on COXmdashH2Srsquos primary

toxicity targetmdashcan be found in sulphide spring lineages of the Puy-

acatengo and Pichucalco river drainages (a lineage with recent diver-

gence time and a lineage with older divergence time respectively)

but not in those of the Tacotalpa drainage (Pfenninger et al 2014)

The presence of nonconvergent genetic and gene expression differ-

ences between populations from sulphidic and non-sulphidic habitats

in different river drainages highlights the limitations of inferences

that can be drawn based on data sets generated with high-through-

put DNA and RNA sequencing The potential importance of modifi-

cations at candidate locimdashor the lack thereofmdashcan only be assessed

through functional assays that document how alternative alleles and

changes in gene expression actually affect organismal performance

The functional consequences of genetic and gene expression varia-

tion at most candidate loci remain to be investigated However

enzyme activity assays have indicated that COX maintains its activity

with increasing H2S concentrations in the two sulphide spring popu-

lations with evidence for positive selection on mitochondrially

encoded subunits (Puyacatengo and Pichucalco Pfenninger et al

2014) Key amino acid substitutions occur in the D-pathway channel

of COX1 which putatively narrow the channel width and reduce

H2Srsquos access to the binuclear centre of the protein Consequently

some lineages of P mexicana exhibit an H2S-resistant COX that can

maintain its activity and aerobic ATP production even if endogenous

H2S concentrations are elevated In contrast COX activity in sul-

phide spring fish of the Tacotalpa river drainage and in fish from

nonsulphidic habitats decreases when H2S is present which is con-

sistent with mechanisms of H2S toxicity (Pfenninger et al 2014) In

vitro assays on isolated mitochondria have nonetheless indicated

that an H2S-resistant COX is not necessary to maintain aerobic ATP

production during H2S exposure Unlike in populations from nonsul-

phidic habitats mitochondrial spare respiratory capacity is main-

tained in the presence of H2S even in the Tacotalpa sulphide spring

populations that lack an H2S-resistant COX (C Henpita M Tobler amp

JH Shaw unpublished data) This may indicate that modification of

detoxification pathways in sulphide spring fishes is sufficient to

maintain mitochondrial function and aerobic ATP production

23 | Ensuing challenges linking genotypes tofitness

Understanding the mechanistic basis of adaptation requires linking

genotypic variation to phenotypic traits that ultimately shape fitness

along selective gradients The evolution of H2S tolerance in sulphide

spring fishes appears to be a consequence of modifications in many

genes and many physiological pathways whose functions interact to

eliminate H2S from the system minimize effects of elevated endoge-

nous concentrations on direct toxicity targets and mitigate adverse

effects that cascade through cellular metabolism during disruption of

homeostasis (Table 1) Such complex genetic changes are likely com-

mon during adaptation and have been documented in other systems

that involve major evolutionary shifts including the emergence of

parasitic lifestyles (Werren et al 2010) eusociality (Simola et al

2013) asexual reproduction (Fradin et al 2017) and internal gesta-

tion with livebearing (Schartl et al 2013) Consequently comple-

mentary approaches will be required to uncover the molecular

mechanisms of adaptation to H2S in the future

1 The involvement of many genes and physiological pathways in

shaping tolerance emphasizes the importance of quantitative

genetic approaches that leverage experimental crosses between

H2S tolerant and nontolerant populations and test whether toler-

ancemdashand adaptive trait modifications that contribute to itmdashis

explained by a few major effects loci or whether it is a conse-

quence of a number of genes with small effects (Klerks Xie amp

Levinton 2011 Saltz Hessel amp Kelly 2017) Combining analyses

of phenotype-fitness associations and their underlying genetic

mechanisms via QTL mapping with the results of selection scans

from population genomic analyses (Pfenninger et al 2015) will

allow for an integrative understanding of how selection acting on

the genome results in phenotypic variation that mediates adapta-

tion to different environments (Stinchcombe amp Hoekstra 2008)

From a functional perspective breaking up correlations between

848 | TOBLER ET AL

different physiological traits (associated with resistance mitiga-

tion and regulation Table 1) in F2 crosses will further allow test-

ing what molecular mechanisms are necessary for tolerance

whether there are alternative strategies with equivalent fitness

benefits and whether there are complementary modifications

with additive effects on tolerance

2 Considering the number of genes and physiological pathways

likely involved in H2S tolerance assessing the functional conse-

quences of genetic and gene expression variation is a daunting

task At the same time such functional tests can significantly

contribute to closing the genotype-to-phenotype gap because

physiochemical stress responses can be assessed along the con-

tinuum of organismal organization (Gjuvsland Vik Beard Hunter

amp Omholt 2013) Standardized assays can be used to quantify

the biochemical makeup of organelles cells and tissues (eg pro-

tein abundance and enzyme activity) physiological performance

in vitro and in vivo (eg rates of H2S detoxification ATP produc-

tion and reactive oxygen production) and fitness-relevant proxies

of whole organism performance (eg H2S tolerance Figure 2)

This highlights the importance of methods development for the

quantification of complex phenotypic traits or even whole phe-

nomes that can keep pace with rapid data generation through

high-throughput sequencing (Houle Govindaraju amp Omholt

2010 Ritchie Holzinger Li Pendergrass amp Kim 2015) In addi-

tion genotypendashphenotype links established through such correla-

tive approaches will ideally be verified through targeted gene

editing While gene editing with CRISPRndashCas9 is now routinely

employed in model organisms (Doudna amp Charpentier 2014 Sha-

lem Sanjana amp Zhang 2015) applying this technology to non-

models still remains challenging especially in livebearing species

like poeciliid fishes where in vitro culture of embryos is not rou-

tinely employed (Martyn Weigel amp Dreyer 2006) Nonetheless

the effects of specific mutations can potentially be verified

through the establishment and manipulation of cells in culture

(Liu et al 2014) or by generating transgenics (Lin et al 2016)

3 Organismal responses to physiochemical stressors occur at multi-

ple hierarchical levels that can be plastically modulated in

response to endogenous and exogenous cues (Whitehead 2012)

Understanding adaptation to physiochemical stressors conse-

quently requires disentangling the effects of genetic variation

developmental plasticity and short-term acclimation in gene

expression patterns on the emerging physiological processes

Common garden experiments on sulphide spring fishes have

shown that many genes differentially expressed between popula-

tions from sulphidic and nonsulphidic habitats in nature remain

differentially expressed in laboratory-reared fish that have never

been exposed to H2S indicating evolved differences in constitu-

tive gene expression (Passow Henpita et al 2017) At the same

time expression patterns in the laboratory also exhibit evidence

for plasticity in response to short-term H2S exposure (Passow

Henpita et al 2017) Some of these plastic responses are shared

between populations from sulphidic and nonsulphidic habitats

(likely representing shared ancestral plasticity) while others have

been lost or gained during sulphide spring colonization A core

challenge for future research is to overcome logistical constraints

that have so far prevented long-term H2S exposures of fish in

the laboratory Exposing individuals throughout development will

be required to test how developmental plasticity and short-term

acclimation shape responses to H2S in individuals from adapted

and nonadapted populations Finally the next big scientific

advance in this context will also require the identification of reg-

ulatory mechanisms that shape expression differences during

adaptation including sequence changes in transcription factors

and their binding sites patterns of DNA methylation and chro-

matin accessibility as well as the production of microRNAs and

competing endogenous RNAs that can modulate transcript abun-

dance (Flores Wolschin amp Amdam 2013 Lopez-Maury Mar-

guerat amp Beuroahler 2008 Silva Bye el Azzouzi amp Wisloslashff 2017)

In addition the role of alternative splicing and other post-tran-

scriptional and post-translational modifications in mediating adap-

tation remains understudied by evolutionary biologists (Ast

2004 Bush Chen Tovar-Corona amp Urritia 2017) This is particu-

larly relevant because H2S has been hypothesized to play a role

in the post-translational modification of proteins with sulphur-

containing amino acids through sulfhydration (Mustafa et al

2009)

Bridging the genotypendashphenotype gap remains one of the major

challenges in biology (Gjuvsland et al 2013 Ritchie et al 2015) and

will provide opportunities to understand how organisms function in

the face of environmental stress and how evolution shapes these

functions during adaptation to different environmental conditions

Establishing the molecular mechanisms of adaptation across repli-

cated lineages exposed to the same sources of selection will also

provide insights into how evolutionary history demography and

genetics shape convergent evolutionary outcomes providing a better

understanding at what hierarchical levels of organismal organization

evolutionary change may actually be predictable (Rosenblum et al

2014)

3 | COMPLEX ORGANISMS IN A COMPLEXWORLD DISENTANGLING THE TANGLEDBANK

Mechanisms of adaptation can be studied through bottom-up

(screening for genetic modifications that allow for inferences about

relevant phenotypic traits) and top-down approaches (assessing fit-

ness-relevant phenotypic differences among populations without

knowledge about the genetic basis see Schluter 2009) Over the

years genomic and transcriptomic studies on sulphide spring popula-

tions of the P mexicana species complex have been complemented

with analyses of phenotypic variation Just as genetic analyses have

indicated that adaptation to H2S is mediated by the modification

and modulation of many genes associated with multiple physiological

pathways phenotypic studies uncovered a variety of differences

TOBLER ET AL | 849

between populations from sulphidic and nonsulphidic environments

Many of these phenotypic differences have evolved in convergence

across replicated sulphide spring lineages in different river drainages

Examples include modifications of morphological (body shape and

organ size Tobler et al 2011 Tobler Scharnweber et al 2015

Schulz-Mirbach et al 2016) physiological (tolerance and metabolic

rates Tobler et al 2011 Passow Arias-Rodriguez amp Tobler 2017)

behavioural (aggression and personality Riesch et al 2009 Bierbach

et al 2012 Sommer-Trembo et al 2016 Bierbach Arias Rodriguez

amp Plath 2018) and life history traits (fecundity and offspring size

Riesch Plath Garcia de Leon amp Schlupp 2010 Riesch et al 2014)

Hence proximate populations in different habitat types have

diverged in complex phenotypes that span many different suites of

traits and a key question is how divergence in multivariate pheno-

types arises In theory H2S could be a key source of selection driv-

ing multivariate phenotypic divergence either because it directly

exerts selection on a multitude of traits that are independently modi-

fied to reach new fitness peaks or because selection on a few fit-

ness-relevant traits causes simultaneous changes in correlated

characteristics Neither explanation seems sufficient because many

Abiotic environmentOxygen concentrations

Ionic compositionTemperature

Structure and flow

Biotic environmentTrophic resources

DemographyPredationParasitism

H2S

Genetic variation

Gene expression

Biochemistry

Physiologicalfunction

Organismal performance

Primary fitness components

Survivalgrowth

reproduction

Organismal function Environment

F IGURE 2 Understanding how organisms function in the context of their environment is a core question in evolutionary biology yet thecomplexities associated with organismal function and environmental variation provide significant challenges High-throughput sequencing hasallowed for the identification of candidate genes subject to selection and differential expression during adaptation to H2S The next step is tobridge the genotype-to-phenotype gap by quantifying the integrated responses to H2S across the continuum of organismal organization includingbiochemical phenotypes physiological functions whole organism performance and ultimately fitness In addition understanding adaptation toH2S-rich environments will also require acknowledging H2Srsquos relation to other abiotic and biotic environmental variables which can exert selectionand modulate aspects of organismal function This will allow for the establishment of cause-and-effect relationships between specific sources ofselection and evolutionary change in specific traits and contribute to our understanding of organismal function in natural systems

850 | TOBLER ET AL

of the divergent traits between populations from sulphidic and non-

sulphidic environments have no clearmdashor at best tenuousmdashlinks to

H2S toxicity and detoxification and there is only limited evidence

for within-population correlations between different traits groups

(Riesch et al 2016)

Evolutionary analyses frequently focus on seemingly simple

sources of selection juxtaposing populations that are assigned to dis-

crete habitat types to understand the evolution of specific traits

(Kaeuffer Peichel Bolnick amp Hendry 2012) But as Darwin (1859)

already observed in his ldquotangled bankrdquo allegory selection in natural

systems is rarely univariate but multifarious selective regimes con-

tribute to trait divergence even among populations that occur along

apparently simple environmental gradients Unquantified environmen-

tal variation that is subsumed in the categorization of habitat types

can profoundly affect trait variation in natural systems (Stuart et al

2017) highlighting the importance of considering and testing alterna-

tive hypotheses about the cause-and-effect relationship between

specific sources of selection and evolution of specific traits Thus

accurate natural history observations that consider organisms as holis-

tic entities consisting of interacting traits in the context of the abiotic

and biotic factors comprising their environment are critical to inform

the design and interpretation of empirical research in natural systems

(Greene 2005) Indeed adjacent sulphidic and nonsulphidic habitat

types do not merely differ in the presence or absence of H2S but also

in a variety of other abiotic and biotic factors (Figure 2) Unlike nonsul-

phidic habitats sulphide springs in southern Mexico are characterized

by extreme hypoxia elevated salinity and lower pH (Rosales Lagarde

2012 Tobler et al 2011) In addition the presence of physiochemical

stressors affects the biotic communities in these sulphide springs Spe-

cies richness of aquatic invertebrates and vertebrates is drastically

reduced as a consequence of H2S toxicity (Greenway Arias-Rodriguez

Diaz amp Tobler 2014) and primary production appears to be driven in

part by chemoautotrophic bacteria that use H2S as an energy source

rather than just by autochthonous and allochthonous photosynthetic

organisms (Roach Tobler amp Winemiller 2011) These changes in the

biotic environment affect the relative importance of intra-vs inter-

specific competition trophic resource use and exposure to predators

and parasites (Riesch Oranth et al 2010 Tobler et al 2014 Tobler

Scharnweber et al 2015)

31 | Alternative drivers of trait evolution

Acknowledging the complexity of selective regimes in sulphide

springs allows for a more nuanced understanding of trait evolution

and for the development of alternative hypotheses about the ulti-

mate mechanisms that cause divergence in specific genetic and phe-

notypic traits In some instances sources of selection other than H2S

can plausibly explain trait divergence between populations from sul-

phidic and nonsulphidic habitats For example hypoxia likely drives

increases in head size and gill surface area (Tobler et al 2011)

changes in the ionic composition of the water can explain differential

expression of genes associated with osmoregulation (Kelley et al

2016) and shifts in trophic resource use are associated with

modifications of the gastrointestinal system (Tobler Scharnweber

et al 2015) In other cases disentangling alternative hypotheses is

not trivial because selection mediated by H2S and other environmen-

tal factors are predicted to affect trait modifications in similar ways

Shifts towards anaerobic metabolism in sulphide spring fish could be

the consequence of inhibition of aerobic ATP production by H2S or

by constraints in oxygen availability Similarly reproductive life his-

tory shifts towards the production of larger but fewer offspring could

be driven by H2S (decreasing offspringrsquos surfacevolume ratio to mini-

mize the diffusion of H2S into the body Powell 1989) by habitat dif-

ferences in the quality and quantity of predation that can affect size-

specific rates of extrinsic mortality (Jorgensen Auer amp Reznick

2011) or by energetic constraints (see below) While comparisons

across different selective regimes can shed light into the ultimate dri-

vers of trait evolution (Moore Riesch amp Martin 2016) isolating the

effects of different sources of selection may not be possible because

they could act in concert and facilitate increases of organismal perfor-

mance in multiple contexts (Ghalambor Walker amp Reznick 2003)

When multifarious selection shapes multivariate phenotypic evo-

lution it is important to also consider the role of trade-offs both in

terms of trait expression (increased investment in one trait may

impose constraints on the expression of other traits) and in terms of

trait function (optimizing the trait for one function may impair the

traitrsquos utility for other functions) For example selection on large off-

spring size in sulphide spring fishes has inadvertent consequences

for fecundity (Riesch et al 2014) and selection for increased gill

surface areas due to hypoxia simultaneously increases exposure to

environmental H2S (Tobler et al 2011) Thus the outcome of phe-

notypic evolution is ultimately a result of balancing competing needs

in a manner that optimizes trait expression and organismal function

in multiple contexts to maximize fitness (Arnold 1983) and the pres-

ence of trade-offs can have cascading indirect effects with selection

on one or a few traits leading to the modification of a multitude of

apparently unrelated characteristics

In this context bioenergetics may represent a nexus in the evo-

lution of complex traits observed in sulphide spring fish and other

extreme environments Living in and adapting to sulphide springs

imposes constraints on organismal energy availability (Plath et al

2007) with according trade-offs for energy allocation Energy limita-

tion in sulphide springs does not arise from a scarcity of food

resources (chemoautotrophic bacteria eaten by the fish are abundant

in sulphide springs Tobler Scharnweber et al 2015) but from

behavioural and physiological adaptation A primary survival strategy

for fish in the toxic and hypoxic springs is to engage in aquatic sur-

face respiration where individuals breathe directly at the water sur-

face to maximize oxygen uptake and minimize H2S exposure This

behaviour mediates short-term survival (Plath et al 2007) but it

also constrains energy acquisition for benthic foragers like P mexi-

cana (Tobler Riesch Tobler amp Plath 2009) and increases susceptibil-

ity to bird predation (Riesch Oranth et al 2010) Costly investments

into H2S detoxification and tissue repair as well as a shift from aero-

bic to less efficient anaerobic metabolism further constrain energy

allocation to other functions (Passow Greenway Arias-Rodriguez

TOBLER ET AL | 851

Jeyasingh amp Tobler 2015) Theory predicts that energetic con-

straints ndash whether mediated by limitations in the rate of resource

acquisition or in the rate of energy allocation to reproduction (as

opposed to maintenance or growth) ndash should exert selection for a

reduction in organismal energy budgets which allows for the maxi-

mization of residual energy allocation for the production of offspring

(Brown Marquet amp Taper 1993) Several lines of evidence support

the hypothesis that energy limitation plays a critical role in pheno-

typic evolution of sulphide spring fish Compared to individuals in

nonsulphidic environments fish in sulphide springs have significantly

lower routine energy demands which is mediated both by reduc-

tions in body size and mass-specific metabolism (Passow Arias-

Rodriguez et al 2017) Reductions in other traits such as energeti-

cally costly behaviours (Bierbach et al 2012 Plath 2008) and the

size of energy-demanding organs (Schulz-Mirbach et al 2016) may

also contribute to the reduced energy demands and shifts in repro-

ductive life history traits (Riesch et al 2014) are consistent with

selection from energy limitation (Reznick amp Yang 1993)

32 | Ensuing challenges Disentanglingenvironmental and organismal complexities

Future research needs to empirically address how multifarious selec-

tive regimes shape the evolution of organisms with complex pheno-

types We have made great advances in our understanding of how

specific traits evolve and mediate adaptation (eg morphology Chan

et al 2010 coloration Dasmahapatra et al 2012 sensory organs

Cartwright Schwartz Merry amp Howell 2017) and we have a thor-

ough understanding of how single environmental factors can shape

organismal responses (eg climatic conditions Johnston amp Bennett

1996 predation Langerhans 2007 competition Pfennig amp Pfennig

2012) In contrast our understanding of phenotypic evolution in

response to multifarious selection is comparatively limited especially

because interactive effects of different environmental factors can lead

to organismal and evolutionary responses that are unpredictable based

on experiments that manipulate single environmental variables in iso-

lation (Gunderson Armstrong amp Stillman 2016 Todgham amp Stillman

2013) Future research thus needs to acknowledge the complexities of

environments and organismal design and address them explicitly in

empirical investigations to gain a robust understanding of biological

diversification (Culumber amp Tobler 2017 Stuart et al 2017)

4 | LINKING ADAPTATION TOSPECIATION

So far we primarily explored the causes of adaptive evolution but

what are its consequences Theory predicts that adaptation to local

environmental conditions can lead to speciation because reproduc-

tive isolation emerges incidentally as a by-product (ecological specia-

tion Nosil 2012) Population genetic analyses of Poecilia spp in

sulphide springs and adjacent populations in nonsulphidic environ-

ments have indicated strong genetic differentiation across small

spatial scales (in some instances lt100 m) and in the absence of

physical barriers that would prevent fish movement (Plath et al

2013) In addition population genetic structuring has been shown to

be stable even in the face of major environmental disturbance

BOX 1 Speciation in sulphide springs and conservation

biology

Adaptation to H2S-rich environments in poeciliids coincides

with ecological speciation giving rise to phenotypically dis-

tinct and reproductively isolated lineages In some

instances this has led to the evolution of highly distinct

micro-endemic species including Poecilia sulphuraria and

P thermalis Limia sulphurophila and Gambusia eurystoma

(Figure 3) Except for P sulphuraria which is found in two

spring complexes of the Rıo Pichucalco drainage of Mexico

(Tobler amp Plath 2009b) the distributions of these sulphide

spring endemics are restricted to single localities that

include only small areas of suitable habitat For example

the largest sulphide spring complex (the Ba~nos del Azufre

in southern Mexico which harbours P sulphuraria and G

eurystoma) includes ~18000 m2 of suitable habitat (equiva-

lent to lt35 football fields) while one of the smallest (La

Esperanza springs harbouring P thermalis) merely covers

~600 m2 (just over a tenth of a football field Brown et al

2017) Accordingly some sulphide spring endemics are

considered critically endangered (IUCN 2012) and pro-

tected by law (SEDESOL 1994)

Many threatened species with restricted ranges exhibit

low genetic diversity and high incidences of inbreeding

(Bouzat 2010 Willi Van Buskirk amp Hoffmann 2006) In

contrast at least P sulphuraria and P thermalismdashthe only

species investigated to datemdashactually exhibit relatively high

effective population sizes (Ne = 1200ndash3800 depending on

the population) and no evidence for inbreeding (FIS lt 0)

(Brown et al 2017) This is likely a consequence of the

long-term stability of sulphide springs and the high pro-

ductivity and low levels of interspecific competition also

support unusually high fish densities and thus population

census sizes (Brown et al 2017 Culumber et al 2016)

Nonetheless sulphide spring endemics are threatened by

habitat destruction which is mediated primarily by the

modification of springs for recreational purposes defor-

estation and conversion of land for agriculture and most

recently high-intensity palm oil plantations (Tobler amp Plath

2009ab) Despite the official protection of sulphide spring

endemics there have been no tangible conservation mea-

sures that foster the sustainable development of land

around sulphide springs with the well-being of the endemic

fishes in mind

852 | TOBLER ET AL

(catastrophic flood event Plath et al 2010) Population genetic dif-

ferentiation between adjacent sulphide-adapted and nonadapted

populations is indicative of ongoing ecological speciation (Box 1) but

different population pairs from sulphidic and nonsulphidic waters

have made variable progress along the speciation continuum towards

complete reproductive isolation (Plath et al 2013) This prompts

questions about the mechanisms involved in reducing gene flow

along the environmental gradient connecting the two habitat types

41 | Prezygotic mechanisms of reproductiveisolation

The simplest form of reproductive isolation between locally adapted

populations arises from selection against migrants between habitat

types (Nosil Vines amp Funk 2005) Translocation experiments in nat-

ural habitats have indicated strong survival selection against

migrants especially those translocated from nonsulphidic to sulphidic

habitats (Plath et al 2013) Fish translocated from nonsulphidic

environments face high mortality within short periods of time

(lt24 hr) which is likely caused by H2S toxicity and hypoxia in sul-

phide springs Hence low gene flow from populations inhabiting

nonsulphidic waters into populations inhabiting sulphidic habitats is

simply explained by a lack of adaptation to the extreme environmen-

tal conditions At least in some drainages there is also appreciable

mortality for sulphide spring fish translocated to nonsulphidic habi-

tats although mechanisms causing inviability are unclear and the

strength of selection appears to be lower (Plath et al 2013) If sur-

vival selection against immigrants was the sole source of reproduc-

tive isolation this should result in unidirectional gene flow from

H2S-adapted to nonadapted populations but population genetic

analyses provided no evidence for this implying that additional

mechanisms are at work (Plath et al 2013)

One such mechanism could be intersexual selection and mate

choice experiments have indicated that migrants from sulphidic to

nonsulphidic habitats are at a significant disadvantage with fish from

nonsulphidic habitats avoiding potential mating partners from sul-

phidic ones (Plath et al 2013 Sommer-Trembo et al 2016) Such

assortative mating has been found to be critical during ecological spe-

ciation because it counteracts the homogenizing effects of gene flow

(van Doorn Edelaar amp Weissing 2009) yet how assortative mating

patterns arise often remains unclear In theory assortative mating can

arise through a variety of processes (Nosil 2012) ldquoMagic traitrdquo mecha-

nisms where mating preferences evolve as a by-product of divergent

selection on ecological traits that serve as cues have recently garnered

much attention (Servedio van Doorn Kopp Frame amp Nosil 2011)

Consistent with magic trait mechanisms assortative mating in P mexi-

cana appears to be linked to adaptive population differences in body

shape which serve as cues to distinguish potential mating partners

from different habitats (Greenway Drexler Arias-Rodriguez amp Tobler

2016) However neither sulphide spring fish (which are unlikely to

encounter migrants from nonsulphidic habitats due to H2S toxicity)

nor allopatric P mexicana from nonsulphidic waters (which have never

experienced migrants from sulphide spring populations) exhibit a

preference for mating partners from their own ecotype suggesting

that assortative mating in this system is a product of reinforcement

(ie direct selection for assortative mating Greenway et al 2016)

This is different from classic magic traits mechanisms because assorta-

tive mating is not a mere by-product of adaptation but adaptive traits

have instead been co-opted as mating signals in response to selection

on mating preferences (Maan amp Seehausen 2012) Moreover males of

some H2S-adapted populations have reduced aggressive behaviour

(Bierbach et al 2012 2018) In staged contests of size-matched males

from different ecotypes intruders had a significantly lower likelihood

to establish dominance than resident males independent of whether

or not they showed reduced aggressiveness overall Hence intrasexual

selection may represent another mechanism mediating reproductive

isolation between locally adapted ecotypes Overall H2S-adapted poe-

ciliids allow disentangling the relative contributions of different forms

of ecological and sexual selectionmdashand their interdependencymdashduring

ecological speciation

42 | Ensuing challenges quantifying postzygoticmechanisms of reproductive isolation

In contrast to prezygotic mechanisms little is known about postzy-

gotic mechanisms of reproductive isolation that potentially act to

reduce gene flow between adjacent populations in sulphidic and

nonsulphidic habitats Putative hybrids between sulphidic and non-

sulphidic ecotypes have occasionally been observed in nonsulphidic

habitats immediately adjacent to confluences with sulphide springs

(R Riesch M Plath and M Tobler personal observations) but it

remains unclear whether and through what mechanisms such hybrids

exhibit reduced fitness compared to parentals

Direct selection of mitochondrially encoded genes in the pres-

ence of H2S (Pfenninger et al 2014) should be accompanied by co-

evolution at nuclear loci that directly interact with mitochondrial

gene products in the formation of protein complexes involved in the

OXPHOS pathway (Sunnucks Morales Lamb Pavlova amp Greening

2017 Willett amp Burton 2004) Disruption of co-evolved gene com-

plexes in hybrids should therefore lead to intrinsic fitness reductions

through mito-nuclear incompatibilities (Burton Pereira amp Barreto

2013 Wolff Ladoukakis Enrıquez amp Dowling 2014) which have

been hypothesized to be a key driver in speciation (Chou amp Leu

2010 Gershoni Templeton amp Mishmar 2009) Empirical tests of

these ideas remain to be conducted in sulphide spring fishes just

like studies on potential ecological and sexual selection against

hybrids that arise from interpopulation mating events

5 | PERSPECTIVES LIFE IN SULPHIDICENVIRONMENTS BEYOND P MEXICANA

Research on sulphide spring fishes of the P mexicana species com-

plex in Mexico has provided insights into mechanisms underlying

adaptation and the links between adaptation and speciation Sul-

phide springs do not only occur in Mexico however but are found

TOBLER ET AL | 853

worldwide and have been colonized by 24 evolutionarily indepen-

dent fish lineages in North America the Neotropics and the Middle

East (Greenway et al 2014) The vast majority of lineages belong to

the order Cyprinodontiformes (95) and livebearing members of the

family Poeciliidae in particular (80) This potentially indicates that

internal gestation of young has facilitated the colonization of

extreme environments because it shields developing embryos from

physiochemical stressors (Blackburn 1999) Among the Poeciliidae

sulphide spring fishes include members of the genera Brachyrhaphis

Gambusia Poeciliopsis Priapichthys Pseudoxiphophorus Xiphophorus

and multiple subgenera of Poecilia (Figure 3 Greenway et al 2014)

spanning over 40 million years of evolution (based on Reznick Fur-

ness Meredith amp Springer 2017)

A handful of studies have started to address molecular and phe-

notypic evolution across a broader taxon sampling of sulphide spring

fishes finding evidence for convergent evolution in morphology

(Riesch et al 2016 Tobler amp Hastings 2011) life history traits

(Riesch Plath et al 2010 Riesch et al 2014) and oxygen transport

genes (Barts et al 2018) In addition ecological speciation in

response to adaptation to H2S-rich environments has also been doc-

umented in sulphide spring poeciliids other than P mexicana (Riesch

et al 2016) These studies have laid a fruitful foundation for the

investigation of mechanisms of adaptation to H2S-rich springs from

genes to phenotypic traits linking micro-evolutionary processes to

macro-evolutionary patterns

The availability of broad-scale evolutionary replication of popula-

tions that have invaded habitats with the same physiochemical stressors

allows for addressing some of the ensuing challenges discussed above

and investigating the factors that give rise to predictable evolutionary

outcomes Convergent evolution is a commonmdashbut not omnipresentmdash

theme in adaptive evolution and multiple factors can determine the

degree of convergence among replicated lineages exposed to the same

sources of selection (Kaeuffer et al 2012 Landry amp Bernatchez 2010

Oke Rolshausen LeBlond amp Hendry 2017 Rosenblum amp Harmon

2011) Thus leveraging replicated lineages that have colonized sulphide

springs to assess molecular mechanisms underlying convergent pheno-

typic changes analyse how environmental heterogeneity among sul-

phide spring habitats causes lineage-specific patterns of trait evolution

and test whether and by what mechanisms adaptation to extreme envi-

ronments is linked to speciation will provide insights into how ecologi-

cal genetic and functional factors contribute to evolutionary

convergence at different levels of organismal organization (Rosenblum

et al 2014) An expanded taxon sampling with population pairs that

span almost the full range of the speciation continuum from panmixia to

complete reproductive isolation (Riesch et al 2016) will also help

addressing questions about temporal aspects of H2S adaptation In par-

ticular comparing younger to older lineages will provide insights about

the order at which genetic and phenotypic modifications accrue during

the adaptation process

Animals have also invaded H2S-rich habitats in other aquatic

ecosystems including cold seeps and deep-sea hydrothermal vents

(Tobler et al 2016) Adaptation to H2S in the P mexicana species com-

plex has involved modification of physiological pathways associated

with H2S toxicity (OXPHOS pathway) and detoxification (sulphide

F IGURE 3 Phylogeny of the family Poeciliidae (adopted from Culumber amp Tobler 2017) Various lineages in the genera Poecilia (subgeneraAcanthophacelus Limia and Mollienesia) Priapichthys Brachyrhaphis Poeciliopsis Xiphophorus Pseudoxiphophorus and Gambusia have colonizedH2S-rich springs in different parts of the Americas (see Greenway et al 2014) Lineages highlighted in yellow represent endemic speciesrestricted to H2S-rich springs while red lineages represent species in which some populations have invaded sulphidic habitats Pictures depict arepresentative of each focal genussubgenus

854 | TOBLER ET AL

quinone oxidoreductase pathway) which have been highly conserved

during the diversification of life on Earth (Saraste 1999 Shahak amp

Hauska 2008) The high replicability of H2S as an environmental factor

that shaped animal evolution across vastly different ecological contexts

consequently offers the opportunity to investigate whether selection

mediated by the presence of H2S has repeatedly modified these con-

served physiological pathways across animal phyla that have diverged

during the Cambrian explosion spanning over 500 million years of evo-

lution Such comparisons would provide unprecedented insights into

the effects of a single physiochemical stressor on evolutionary change

irrespective of ecological or phylogenetic contexts

6 | CONCLUSIONS

H2S-rich environments and their inhabitants represent useful models

for research in organismal biology ecology and evolution illustrating

both opportunities (eg strong and explicit source of selection evolu-

tionary replication) and challenges (eg complexity of environments

and organismal responses) in asking basic question about the origins

of biodiversity Understanding how organisms thrive in apparently

inhospitable environments will provide us with insights into lifersquos

capacities and limitations to adapt to novel environments (Waterman

1999) This is particularly relevant considering that human activities

substantially contribute to the exacerbation of physiochemical stres-

sors at local regional and global scales impacting ecological and evo-

lutionary dynamics (Alberti 2015 Norberg Urban Vellend

Klausmeier amp Loeuille 2012) Thus a focus on ecological and evolu-

tionary consequences of natural stressors that push organisms to the

brink of their physiological capacities may be instrumental to predict

consequences of anthropogenic environmental change

ACKNOWLEDGEMENTS

We would like to thank Lenin Arias-Rodriguez the Universidad

Juarez Autonoma de Tabasco and the Centro de Investigacion E

Innovacion Para la Ense~nanza y El Aprendizaje for their continued

collaboration enabling our work on the P mexicana system Funding

for this review article has been provided by the National Science

Foundation (IOS-1557860 IOS-1557795) and the US Army

Research Office (W911NF-15-1-0175)

AUTHOR CONTRIBUTIONS

MT wrote the first draft of the paper all authors contributed text

and comments

ORCID

Michael Tobler httporcidorg0000-0002-0326-0890

Joanna L Kelley httporcidorg0000-0002-7731-605X

Reuroudiger Riesch httporcidorg0000-0002-0223-1254

REFERENCES

Alberti M (2015) Eco-evolutionary dynamics in an urbanizing planet

Trends in Ecology amp Evolution 30 114ndash126 httpsdoiorg101016j

tree201411007

Alda F Reina R G Doadrio I amp Bermingham E (2013) Phylogeny

and biogeography of the Poecilia sphenops species complex

(Actinopterygii Poeciliidae) in Central America Molecular Phylogenet-

ics and Evolution 66 1011ndash1026 httpsdoiorg101016jympev

201212012

Alvarez del Villar J (1948) Descripcion de una nueva especie de Mol-

lienisia capturada en Ba~nos del Azufre Tabasco (Pisces Poeciliidae)

Anales de la Escuela Nacional de Ciencias Biologicas 5 275ndash281

Arnold S J (1983) Morphology performance and fitness American Zool-

ogist 23 347ndash361 httpsdoiorg101093icb232347

Aspiras A C Rohner N Martineau B Borowsky R L amp Clifford C J

(2015) Melanocortin 4 receptor mutations contribute to the adapta-

tion of cavefish to nutrient-poor conditions Proceedings of the

National Academy of Sciences USA 112 9968ndash9973

Ast G (2004) How did alternative splicing evolve Nature Reviews

Genetics 5 773ndash782 httpsdoiorg101038nrg1451

Bagarinao T (1992) Sulfide as an environmental factor and toxicant

Tolerance and adaptations in aquatic organisms Aquatic Toxicology

24 21ndash62 httpsdoiorg1010160166-445X(92)90015-F

Barton N H (1998) The effect of hitch-hiking on neutral genealogies

Genetic Research 72 123ndash133 httpsdoiorg101017

S0016672398003462

Barts N Greenway R Passow C N Arias Rodriguez L Kelley J L amp

Tobler M (2018) Expression and molecular evolution of oxygen

transport genes in livebearing fishes (Poeciliidae) from hydrogen sul-

fide rich springs Genome httpsdoiorg101139gen-2017-0051

[Epub ahead of print]

Beauchamp R O Bus J S Popp C S Boreiko C J amp Andjelkovich

D A (1984) A critical review of the literature on hydrogen sulfide

CRC Critical Reviews in Toxicology 13 25ndash97 httpsdoiorg10

310910408448409029321

Bell E M (2012) Life at extremes Environments organisms and strategies

for survival Oxfordshire CABI httpsdoiorg101079

97818459381470000

Bierbach D Arias Rodriguez L amp Plath M (2018) Intrasexual competi-

tion enhances reproductive isolation between locally adapted popula-

tions Current Zoology httpsdoiorg101093czzox071 [Epub

ahead of print]

Bierbach D Klein M Sassmannshausen V Schlupp I Riesch R

Parzefall J amp Plath M (2012) Divergent evolution of male aggres-

sive behavior Another reproductive isolation barrier in extremophile

poeciliid fishes International Journal of Evolutionary Biology 2012

148745

Blackburn D G (1999) Viviparity and oviparity Evolution and reproduc-

tive strategies In T E Knobil amp J D Neill (Eds) Encyclopedia of

reproduction Vol 4 (pp 994ndash1003) New York NY Academic Press

Bouzat J L (2010) Conservation genetics of population bottlenecks

The role of chance selection and history Conservation Genetics 11

463ndash478

Brown A P Greenway R Morgan S Quackenbush C R Giordani

L Arias-Rodriguez L Kelley J L (2017) Genome-scale data

reveals that endemic Poecilia populations from small sulfide spring

display no evidence of inbreeding Molecular Ecology 26 4920ndash

4934

Brown J H Marquet P A amp Taper M L (1993) Evolution of body

size Consequences of an energetic definition of fitness American

Naturalist 142 573ndash584 httpsdoiorg101086285558

Burton R S Pereira R J amp Barreto F S (2013) Cytonuclear genomic

interactions and hybrid breakdown Annual Review of Ecology

TOBLER ET AL | 855

different physiological traits (associated with resistance mitiga-

tion and regulation Table 1) in F2 crosses will further allow test-

ing what molecular mechanisms are necessary for tolerance

whether there are alternative strategies with equivalent fitness

benefits and whether there are complementary modifications

with additive effects on tolerance

2 Considering the number of genes and physiological pathways

likely involved in H2S tolerance assessing the functional conse-

quences of genetic and gene expression variation is a daunting

task At the same time such functional tests can significantly

contribute to closing the genotype-to-phenotype gap because

physiochemical stress responses can be assessed along the con-

tinuum of organismal organization (Gjuvsland Vik Beard Hunter

amp Omholt 2013) Standardized assays can be used to quantify

the biochemical makeup of organelles cells and tissues (eg pro-

tein abundance and enzyme activity) physiological performance

in vitro and in vivo (eg rates of H2S detoxification ATP produc-

tion and reactive oxygen production) and fitness-relevant proxies

of whole organism performance (eg H2S tolerance Figure 2)

This highlights the importance of methods development for the

quantification of complex phenotypic traits or even whole phe-

nomes that can keep pace with rapid data generation through

high-throughput sequencing (Houle Govindaraju amp Omholt

2010 Ritchie Holzinger Li Pendergrass amp Kim 2015) In addi-

tion genotypendashphenotype links established through such correla-

tive approaches will ideally be verified through targeted gene

editing While gene editing with CRISPRndashCas9 is now routinely

employed in model organisms (Doudna amp Charpentier 2014 Sha-

lem Sanjana amp Zhang 2015) applying this technology to non-

models still remains challenging especially in livebearing species

like poeciliid fishes where in vitro culture of embryos is not rou-

tinely employed (Martyn Weigel amp Dreyer 2006) Nonetheless

the effects of specific mutations can potentially be verified

through the establishment and manipulation of cells in culture

(Liu et al 2014) or by generating transgenics (Lin et al 2016)

3 Organismal responses to physiochemical stressors occur at multi-

ple hierarchical levels that can be plastically modulated in

response to endogenous and exogenous cues (Whitehead 2012)

Understanding adaptation to physiochemical stressors conse-

quently requires disentangling the effects of genetic variation

developmental plasticity and short-term acclimation in gene

expression patterns on the emerging physiological processes

Common garden experiments on sulphide spring fishes have

shown that many genes differentially expressed between popula-

tions from sulphidic and nonsulphidic habitats in nature remain

differentially expressed in laboratory-reared fish that have never

been exposed to H2S indicating evolved differences in constitu-

tive gene expression (Passow Henpita et al 2017) At the same

time expression patterns in the laboratory also exhibit evidence

for plasticity in response to short-term H2S exposure (Passow

Henpita et al 2017) Some of these plastic responses are shared

between populations from sulphidic and nonsulphidic habitats

(likely representing shared ancestral plasticity) while others have

been lost or gained during sulphide spring colonization A core

challenge for future research is to overcome logistical constraints

that have so far prevented long-term H2S exposures of fish in

the laboratory Exposing individuals throughout development will

be required to test how developmental plasticity and short-term

acclimation shape responses to H2S in individuals from adapted

and nonadapted populations Finally the next big scientific

advance in this context will also require the identification of reg-

ulatory mechanisms that shape expression differences during

adaptation including sequence changes in transcription factors

and their binding sites patterns of DNA methylation and chro-

matin accessibility as well as the production of microRNAs and

competing endogenous RNAs that can modulate transcript abun-

dance (Flores Wolschin amp Amdam 2013 Lopez-Maury Mar-

guerat amp Beuroahler 2008 Silva Bye el Azzouzi amp Wisloslashff 2017)

In addition the role of alternative splicing and other post-tran-

scriptional and post-translational modifications in mediating adap-

tation remains understudied by evolutionary biologists (Ast

2004 Bush Chen Tovar-Corona amp Urritia 2017) This is particu-

larly relevant because H2S has been hypothesized to play a role

in the post-translational modification of proteins with sulphur-

containing amino acids through sulfhydration (Mustafa et al

2009)

Bridging the genotypendashphenotype gap remains one of the major

challenges in biology (Gjuvsland et al 2013 Ritchie et al 2015) and

will provide opportunities to understand how organisms function in

the face of environmental stress and how evolution shapes these

functions during adaptation to different environmental conditions

Establishing the molecular mechanisms of adaptation across repli-

cated lineages exposed to the same sources of selection will also

provide insights into how evolutionary history demography and

genetics shape convergent evolutionary outcomes providing a better

understanding at what hierarchical levels of organismal organization

evolutionary change may actually be predictable (Rosenblum et al

2014)

3 | COMPLEX ORGANISMS IN A COMPLEXWORLD DISENTANGLING THE TANGLEDBANK

Mechanisms of adaptation can be studied through bottom-up

(screening for genetic modifications that allow for inferences about

relevant phenotypic traits) and top-down approaches (assessing fit-

ness-relevant phenotypic differences among populations without

knowledge about the genetic basis see Schluter 2009) Over the

years genomic and transcriptomic studies on sulphide spring popula-

tions of the P mexicana species complex have been complemented

with analyses of phenotypic variation Just as genetic analyses have

indicated that adaptation to H2S is mediated by the modification

and modulation of many genes associated with multiple physiological

pathways phenotypic studies uncovered a variety of differences

TOBLER ET AL | 849

between populations from sulphidic and nonsulphidic environments

Many of these phenotypic differences have evolved in convergence

across replicated sulphide spring lineages in different river drainages

Examples include modifications of morphological (body shape and

organ size Tobler et al 2011 Tobler Scharnweber et al 2015

Schulz-Mirbach et al 2016) physiological (tolerance and metabolic

rates Tobler et al 2011 Passow Arias-Rodriguez amp Tobler 2017)

behavioural (aggression and personality Riesch et al 2009 Bierbach

et al 2012 Sommer-Trembo et al 2016 Bierbach Arias Rodriguez

amp Plath 2018) and life history traits (fecundity and offspring size

Riesch Plath Garcia de Leon amp Schlupp 2010 Riesch et al 2014)

Hence proximate populations in different habitat types have

diverged in complex phenotypes that span many different suites of

traits and a key question is how divergence in multivariate pheno-

types arises In theory H2S could be a key source of selection driv-

ing multivariate phenotypic divergence either because it directly

exerts selection on a multitude of traits that are independently modi-

fied to reach new fitness peaks or because selection on a few fit-

ness-relevant traits causes simultaneous changes in correlated

characteristics Neither explanation seems sufficient because many

Abiotic environmentOxygen concentrations

Ionic compositionTemperature

Structure and flow

Biotic environmentTrophic resources

DemographyPredationParasitism

H2S

Genetic variation

Gene expression

Biochemistry

Physiologicalfunction

Organismal performance

Primary fitness components

Survivalgrowth

reproduction

Organismal function Environment

F IGURE 2 Understanding how organisms function in the context of their environment is a core question in evolutionary biology yet thecomplexities associated with organismal function and environmental variation provide significant challenges High-throughput sequencing hasallowed for the identification of candidate genes subject to selection and differential expression during adaptation to H2S The next step is tobridge the genotype-to-phenotype gap by quantifying the integrated responses to H2S across the continuum of organismal organization includingbiochemical phenotypes physiological functions whole organism performance and ultimately fitness In addition understanding adaptation toH2S-rich environments will also require acknowledging H2Srsquos relation to other abiotic and biotic environmental variables which can exert selectionand modulate aspects of organismal function This will allow for the establishment of cause-and-effect relationships between specific sources ofselection and evolutionary change in specific traits and contribute to our understanding of organismal function in natural systems

850 | TOBLER ET AL

of the divergent traits between populations from sulphidic and non-

sulphidic environments have no clearmdashor at best tenuousmdashlinks to

H2S toxicity and detoxification and there is only limited evidence

for within-population correlations between different traits groups

(Riesch et al 2016)

Evolutionary analyses frequently focus on seemingly simple

sources of selection juxtaposing populations that are assigned to dis-

crete habitat types to understand the evolution of specific traits

(Kaeuffer Peichel Bolnick amp Hendry 2012) But as Darwin (1859)

already observed in his ldquotangled bankrdquo allegory selection in natural

systems is rarely univariate but multifarious selective regimes con-

tribute to trait divergence even among populations that occur along

apparently simple environmental gradients Unquantified environmen-

tal variation that is subsumed in the categorization of habitat types

can profoundly affect trait variation in natural systems (Stuart et al

2017) highlighting the importance of considering and testing alterna-

tive hypotheses about the cause-and-effect relationship between

specific sources of selection and evolution of specific traits Thus

accurate natural history observations that consider organisms as holis-

tic entities consisting of interacting traits in the context of the abiotic

and biotic factors comprising their environment are critical to inform

the design and interpretation of empirical research in natural systems

(Greene 2005) Indeed adjacent sulphidic and nonsulphidic habitat

types do not merely differ in the presence or absence of H2S but also

in a variety of other abiotic and biotic factors (Figure 2) Unlike nonsul-

phidic habitats sulphide springs in southern Mexico are characterized

by extreme hypoxia elevated salinity and lower pH (Rosales Lagarde

2012 Tobler et al 2011) In addition the presence of physiochemical

stressors affects the biotic communities in these sulphide springs Spe-

cies richness of aquatic invertebrates and vertebrates is drastically

reduced as a consequence of H2S toxicity (Greenway Arias-Rodriguez

Diaz amp Tobler 2014) and primary production appears to be driven in

part by chemoautotrophic bacteria that use H2S as an energy source

rather than just by autochthonous and allochthonous photosynthetic

organisms (Roach Tobler amp Winemiller 2011) These changes in the

biotic environment affect the relative importance of intra-vs inter-

specific competition trophic resource use and exposure to predators

and parasites (Riesch Oranth et al 2010 Tobler et al 2014 Tobler

Scharnweber et al 2015)

31 | Alternative drivers of trait evolution

Acknowledging the complexity of selective regimes in sulphide

springs allows for a more nuanced understanding of trait evolution

and for the development of alternative hypotheses about the ulti-

mate mechanisms that cause divergence in specific genetic and phe-

notypic traits In some instances sources of selection other than H2S

can plausibly explain trait divergence between populations from sul-

phidic and nonsulphidic habitats For example hypoxia likely drives

increases in head size and gill surface area (Tobler et al 2011)

changes in the ionic composition of the water can explain differential

expression of genes associated with osmoregulation (Kelley et al

2016) and shifts in trophic resource use are associated with

modifications of the gastrointestinal system (Tobler Scharnweber

et al 2015) In other cases disentangling alternative hypotheses is

not trivial because selection mediated by H2S and other environmen-

tal factors are predicted to affect trait modifications in similar ways

Shifts towards anaerobic metabolism in sulphide spring fish could be

the consequence of inhibition of aerobic ATP production by H2S or

by constraints in oxygen availability Similarly reproductive life his-

tory shifts towards the production of larger but fewer offspring could

be driven by H2S (decreasing offspringrsquos surfacevolume ratio to mini-

mize the diffusion of H2S into the body Powell 1989) by habitat dif-

ferences in the quality and quantity of predation that can affect size-

specific rates of extrinsic mortality (Jorgensen Auer amp Reznick

2011) or by energetic constraints (see below) While comparisons

across different selective regimes can shed light into the ultimate dri-

vers of trait evolution (Moore Riesch amp Martin 2016) isolating the

effects of different sources of selection may not be possible because

they could act in concert and facilitate increases of organismal perfor-

mance in multiple contexts (Ghalambor Walker amp Reznick 2003)

When multifarious selection shapes multivariate phenotypic evo-

lution it is important to also consider the role of trade-offs both in

terms of trait expression (increased investment in one trait may

impose constraints on the expression of other traits) and in terms of

trait function (optimizing the trait for one function may impair the

traitrsquos utility for other functions) For example selection on large off-

spring size in sulphide spring fishes has inadvertent consequences

for fecundity (Riesch et al 2014) and selection for increased gill

surface areas due to hypoxia simultaneously increases exposure to

environmental H2S (Tobler et al 2011) Thus the outcome of phe-

notypic evolution is ultimately a result of balancing competing needs

in a manner that optimizes trait expression and organismal function

in multiple contexts to maximize fitness (Arnold 1983) and the pres-

ence of trade-offs can have cascading indirect effects with selection

on one or a few traits leading to the modification of a multitude of

apparently unrelated characteristics

In this context bioenergetics may represent a nexus in the evo-

lution of complex traits observed in sulphide spring fish and other

extreme environments Living in and adapting to sulphide springs

imposes constraints on organismal energy availability (Plath et al

2007) with according trade-offs for energy allocation Energy limita-

tion in sulphide springs does not arise from a scarcity of food

resources (chemoautotrophic bacteria eaten by the fish are abundant

in sulphide springs Tobler Scharnweber et al 2015) but from

behavioural and physiological adaptation A primary survival strategy

for fish in the toxic and hypoxic springs is to engage in aquatic sur-

face respiration where individuals breathe directly at the water sur-

face to maximize oxygen uptake and minimize H2S exposure This

behaviour mediates short-term survival (Plath et al 2007) but it

also constrains energy acquisition for benthic foragers like P mexi-

cana (Tobler Riesch Tobler amp Plath 2009) and increases susceptibil-

ity to bird predation (Riesch Oranth et al 2010) Costly investments

into H2S detoxification and tissue repair as well as a shift from aero-

bic to less efficient anaerobic metabolism further constrain energy

allocation to other functions (Passow Greenway Arias-Rodriguez

TOBLER ET AL | 851

Jeyasingh amp Tobler 2015) Theory predicts that energetic con-

straints ndash whether mediated by limitations in the rate of resource

acquisition or in the rate of energy allocation to reproduction (as

opposed to maintenance or growth) ndash should exert selection for a

reduction in organismal energy budgets which allows for the maxi-

mization of residual energy allocation for the production of offspring

(Brown Marquet amp Taper 1993) Several lines of evidence support

the hypothesis that energy limitation plays a critical role in pheno-

typic evolution of sulphide spring fish Compared to individuals in

nonsulphidic environments fish in sulphide springs have significantly

lower routine energy demands which is mediated both by reduc-

tions in body size and mass-specific metabolism (Passow Arias-

Rodriguez et al 2017) Reductions in other traits such as energeti-

cally costly behaviours (Bierbach et al 2012 Plath 2008) and the

size of energy-demanding organs (Schulz-Mirbach et al 2016) may

also contribute to the reduced energy demands and shifts in repro-

ductive life history traits (Riesch et al 2014) are consistent with

selection from energy limitation (Reznick amp Yang 1993)

32 | Ensuing challenges Disentanglingenvironmental and organismal complexities

Future research needs to empirically address how multifarious selec-

tive regimes shape the evolution of organisms with complex pheno-

types We have made great advances in our understanding of how

specific traits evolve and mediate adaptation (eg morphology Chan

et al 2010 coloration Dasmahapatra et al 2012 sensory organs

Cartwright Schwartz Merry amp Howell 2017) and we have a thor-

ough understanding of how single environmental factors can shape

organismal responses (eg climatic conditions Johnston amp Bennett

1996 predation Langerhans 2007 competition Pfennig amp Pfennig

2012) In contrast our understanding of phenotypic evolution in

response to multifarious selection is comparatively limited especially

because interactive effects of different environmental factors can lead

to organismal and evolutionary responses that are unpredictable based

on experiments that manipulate single environmental variables in iso-

lation (Gunderson Armstrong amp Stillman 2016 Todgham amp Stillman

2013) Future research thus needs to acknowledge the complexities of

environments and organismal design and address them explicitly in

empirical investigations to gain a robust understanding of biological

diversification (Culumber amp Tobler 2017 Stuart et al 2017)

4 | LINKING ADAPTATION TOSPECIATION

So far we primarily explored the causes of adaptive evolution but

what are its consequences Theory predicts that adaptation to local

environmental conditions can lead to speciation because reproduc-

tive isolation emerges incidentally as a by-product (ecological specia-

tion Nosil 2012) Population genetic analyses of Poecilia spp in

sulphide springs and adjacent populations in nonsulphidic environ-

ments have indicated strong genetic differentiation across small

spatial scales (in some instances lt100 m) and in the absence of

physical barriers that would prevent fish movement (Plath et al

2013) In addition population genetic structuring has been shown to

be stable even in the face of major environmental disturbance

BOX 1 Speciation in sulphide springs and conservation

biology

Adaptation to H2S-rich environments in poeciliids coincides

with ecological speciation giving rise to phenotypically dis-

tinct and reproductively isolated lineages In some

instances this has led to the evolution of highly distinct

micro-endemic species including Poecilia sulphuraria and

P thermalis Limia sulphurophila and Gambusia eurystoma

(Figure 3) Except for P sulphuraria which is found in two

spring complexes of the Rıo Pichucalco drainage of Mexico

(Tobler amp Plath 2009b) the distributions of these sulphide

spring endemics are restricted to single localities that

include only small areas of suitable habitat For example

the largest sulphide spring complex (the Ba~nos del Azufre

in southern Mexico which harbours P sulphuraria and G

eurystoma) includes ~18000 m2 of suitable habitat (equiva-

lent to lt35 football fields) while one of the smallest (La

Esperanza springs harbouring P thermalis) merely covers

~600 m2 (just over a tenth of a football field Brown et al

2017) Accordingly some sulphide spring endemics are

considered critically endangered (IUCN 2012) and pro-

tected by law (SEDESOL 1994)

Many threatened species with restricted ranges exhibit

low genetic diversity and high incidences of inbreeding

(Bouzat 2010 Willi Van Buskirk amp Hoffmann 2006) In

contrast at least P sulphuraria and P thermalismdashthe only

species investigated to datemdashactually exhibit relatively high

effective population sizes (Ne = 1200ndash3800 depending on

the population) and no evidence for inbreeding (FIS lt 0)

(Brown et al 2017) This is likely a consequence of the

long-term stability of sulphide springs and the high pro-

ductivity and low levels of interspecific competition also

support unusually high fish densities and thus population

census sizes (Brown et al 2017 Culumber et al 2016)

Nonetheless sulphide spring endemics are threatened by

habitat destruction which is mediated primarily by the

modification of springs for recreational purposes defor-

estation and conversion of land for agriculture and most

recently high-intensity palm oil plantations (Tobler amp Plath

2009ab) Despite the official protection of sulphide spring

endemics there have been no tangible conservation mea-

sures that foster the sustainable development of land

around sulphide springs with the well-being of the endemic

fishes in mind

852 | TOBLER ET AL

(catastrophic flood event Plath et al 2010) Population genetic dif-

ferentiation between adjacent sulphide-adapted and nonadapted

populations is indicative of ongoing ecological speciation (Box 1) but

different population pairs from sulphidic and nonsulphidic waters

have made variable progress along the speciation continuum towards

complete reproductive isolation (Plath et al 2013) This prompts

questions about the mechanisms involved in reducing gene flow

along the environmental gradient connecting the two habitat types

41 | Prezygotic mechanisms of reproductiveisolation

The simplest form of reproductive isolation between locally adapted

populations arises from selection against migrants between habitat

types (Nosil Vines amp Funk 2005) Translocation experiments in nat-

ural habitats have indicated strong survival selection against

migrants especially those translocated from nonsulphidic to sulphidic

habitats (Plath et al 2013) Fish translocated from nonsulphidic

environments face high mortality within short periods of time

(lt24 hr) which is likely caused by H2S toxicity and hypoxia in sul-

phide springs Hence low gene flow from populations inhabiting

nonsulphidic waters into populations inhabiting sulphidic habitats is

simply explained by a lack of adaptation to the extreme environmen-

tal conditions At least in some drainages there is also appreciable

mortality for sulphide spring fish translocated to nonsulphidic habi-

tats although mechanisms causing inviability are unclear and the

strength of selection appears to be lower (Plath et al 2013) If sur-

vival selection against immigrants was the sole source of reproduc-

tive isolation this should result in unidirectional gene flow from

H2S-adapted to nonadapted populations but population genetic

analyses provided no evidence for this implying that additional

mechanisms are at work (Plath et al 2013)

One such mechanism could be intersexual selection and mate

choice experiments have indicated that migrants from sulphidic to

nonsulphidic habitats are at a significant disadvantage with fish from

nonsulphidic habitats avoiding potential mating partners from sul-

phidic ones (Plath et al 2013 Sommer-Trembo et al 2016) Such

assortative mating has been found to be critical during ecological spe-

ciation because it counteracts the homogenizing effects of gene flow

(van Doorn Edelaar amp Weissing 2009) yet how assortative mating

patterns arise often remains unclear In theory assortative mating can

arise through a variety of processes (Nosil 2012) ldquoMagic traitrdquo mecha-

nisms where mating preferences evolve as a by-product of divergent

selection on ecological traits that serve as cues have recently garnered

much attention (Servedio van Doorn Kopp Frame amp Nosil 2011)

Consistent with magic trait mechanisms assortative mating in P mexi-

cana appears to be linked to adaptive population differences in body

shape which serve as cues to distinguish potential mating partners

from different habitats (Greenway Drexler Arias-Rodriguez amp Tobler

2016) However neither sulphide spring fish (which are unlikely to

encounter migrants from nonsulphidic habitats due to H2S toxicity)

nor allopatric P mexicana from nonsulphidic waters (which have never

experienced migrants from sulphide spring populations) exhibit a

preference for mating partners from their own ecotype suggesting

that assortative mating in this system is a product of reinforcement

(ie direct selection for assortative mating Greenway et al 2016)

This is different from classic magic traits mechanisms because assorta-

tive mating is not a mere by-product of adaptation but adaptive traits

have instead been co-opted as mating signals in response to selection

on mating preferences (Maan amp Seehausen 2012) Moreover males of

some H2S-adapted populations have reduced aggressive behaviour

(Bierbach et al 2012 2018) In staged contests of size-matched males

from different ecotypes intruders had a significantly lower likelihood

to establish dominance than resident males independent of whether

or not they showed reduced aggressiveness overall Hence intrasexual

selection may represent another mechanism mediating reproductive

isolation between locally adapted ecotypes Overall H2S-adapted poe-

ciliids allow disentangling the relative contributions of different forms

of ecological and sexual selectionmdashand their interdependencymdashduring

ecological speciation

42 | Ensuing challenges quantifying postzygoticmechanisms of reproductive isolation

In contrast to prezygotic mechanisms little is known about postzy-

gotic mechanisms of reproductive isolation that potentially act to

reduce gene flow between adjacent populations in sulphidic and

nonsulphidic habitats Putative hybrids between sulphidic and non-

sulphidic ecotypes have occasionally been observed in nonsulphidic

habitats immediately adjacent to confluences with sulphide springs

(R Riesch M Plath and M Tobler personal observations) but it

remains unclear whether and through what mechanisms such hybrids

exhibit reduced fitness compared to parentals

Direct selection of mitochondrially encoded genes in the pres-

ence of H2S (Pfenninger et al 2014) should be accompanied by co-

evolution at nuclear loci that directly interact with mitochondrial

gene products in the formation of protein complexes involved in the

OXPHOS pathway (Sunnucks Morales Lamb Pavlova amp Greening

2017 Willett amp Burton 2004) Disruption of co-evolved gene com-

plexes in hybrids should therefore lead to intrinsic fitness reductions

through mito-nuclear incompatibilities (Burton Pereira amp Barreto

2013 Wolff Ladoukakis Enrıquez amp Dowling 2014) which have

been hypothesized to be a key driver in speciation (Chou amp Leu

2010 Gershoni Templeton amp Mishmar 2009) Empirical tests of

these ideas remain to be conducted in sulphide spring fishes just

like studies on potential ecological and sexual selection against

hybrids that arise from interpopulation mating events

5 | PERSPECTIVES LIFE IN SULPHIDICENVIRONMENTS BEYOND P MEXICANA

Research on sulphide spring fishes of the P mexicana species com-

plex in Mexico has provided insights into mechanisms underlying

adaptation and the links between adaptation and speciation Sul-

phide springs do not only occur in Mexico however but are found

TOBLER ET AL | 853

worldwide and have been colonized by 24 evolutionarily indepen-

dent fish lineages in North America the Neotropics and the Middle

East (Greenway et al 2014) The vast majority of lineages belong to

the order Cyprinodontiformes (95) and livebearing members of the

family Poeciliidae in particular (80) This potentially indicates that

internal gestation of young has facilitated the colonization of

extreme environments because it shields developing embryos from

physiochemical stressors (Blackburn 1999) Among the Poeciliidae

sulphide spring fishes include members of the genera Brachyrhaphis

Gambusia Poeciliopsis Priapichthys Pseudoxiphophorus Xiphophorus

and multiple subgenera of Poecilia (Figure 3 Greenway et al 2014)

spanning over 40 million years of evolution (based on Reznick Fur-

ness Meredith amp Springer 2017)

A handful of studies have started to address molecular and phe-

notypic evolution across a broader taxon sampling of sulphide spring

fishes finding evidence for convergent evolution in morphology

(Riesch et al 2016 Tobler amp Hastings 2011) life history traits

(Riesch Plath et al 2010 Riesch et al 2014) and oxygen transport

genes (Barts et al 2018) In addition ecological speciation in

response to adaptation to H2S-rich environments has also been doc-

umented in sulphide spring poeciliids other than P mexicana (Riesch

et al 2016) These studies have laid a fruitful foundation for the

investigation of mechanisms of adaptation to H2S-rich springs from

genes to phenotypic traits linking micro-evolutionary processes to

macro-evolutionary patterns

The availability of broad-scale evolutionary replication of popula-

tions that have invaded habitats with the same physiochemical stressors

allows for addressing some of the ensuing challenges discussed above

and investigating the factors that give rise to predictable evolutionary

outcomes Convergent evolution is a commonmdashbut not omnipresentmdash

theme in adaptive evolution and multiple factors can determine the

degree of convergence among replicated lineages exposed to the same

sources of selection (Kaeuffer et al 2012 Landry amp Bernatchez 2010

Oke Rolshausen LeBlond amp Hendry 2017 Rosenblum amp Harmon

2011) Thus leveraging replicated lineages that have colonized sulphide

springs to assess molecular mechanisms underlying convergent pheno-

typic changes analyse how environmental heterogeneity among sul-

phide spring habitats causes lineage-specific patterns of trait evolution

and test whether and by what mechanisms adaptation to extreme envi-

ronments is linked to speciation will provide insights into how ecologi-

cal genetic and functional factors contribute to evolutionary

convergence at different levels of organismal organization (Rosenblum

et al 2014) An expanded taxon sampling with population pairs that

span almost the full range of the speciation continuum from panmixia to

complete reproductive isolation (Riesch et al 2016) will also help

addressing questions about temporal aspects of H2S adaptation In par-

ticular comparing younger to older lineages will provide insights about

the order at which genetic and phenotypic modifications accrue during

the adaptation process

Animals have also invaded H2S-rich habitats in other aquatic

ecosystems including cold seeps and deep-sea hydrothermal vents

(Tobler et al 2016) Adaptation to H2S in the P mexicana species com-

plex has involved modification of physiological pathways associated

with H2S toxicity (OXPHOS pathway) and detoxification (sulphide

F IGURE 3 Phylogeny of the family Poeciliidae (adopted from Culumber amp Tobler 2017) Various lineages in the genera Poecilia (subgeneraAcanthophacelus Limia and Mollienesia) Priapichthys Brachyrhaphis Poeciliopsis Xiphophorus Pseudoxiphophorus and Gambusia have colonizedH2S-rich springs in different parts of the Americas (see Greenway et al 2014) Lineages highlighted in yellow represent endemic speciesrestricted to H2S-rich springs while red lineages represent species in which some populations have invaded sulphidic habitats Pictures depict arepresentative of each focal genussubgenus

854 | TOBLER ET AL

quinone oxidoreductase pathway) which have been highly conserved

during the diversification of life on Earth (Saraste 1999 Shahak amp

Hauska 2008) The high replicability of H2S as an environmental factor

that shaped animal evolution across vastly different ecological contexts

consequently offers the opportunity to investigate whether selection

mediated by the presence of H2S has repeatedly modified these con-

served physiological pathways across animal phyla that have diverged

during the Cambrian explosion spanning over 500 million years of evo-

lution Such comparisons would provide unprecedented insights into

the effects of a single physiochemical stressor on evolutionary change

irrespective of ecological or phylogenetic contexts

6 | CONCLUSIONS

H2S-rich environments and their inhabitants represent useful models

for research in organismal biology ecology and evolution illustrating

both opportunities (eg strong and explicit source of selection evolu-

tionary replication) and challenges (eg complexity of environments

and organismal responses) in asking basic question about the origins

of biodiversity Understanding how organisms thrive in apparently

inhospitable environments will provide us with insights into lifersquos

capacities and limitations to adapt to novel environments (Waterman

1999) This is particularly relevant considering that human activities

substantially contribute to the exacerbation of physiochemical stres-

sors at local regional and global scales impacting ecological and evo-

lutionary dynamics (Alberti 2015 Norberg Urban Vellend

Klausmeier amp Loeuille 2012) Thus a focus on ecological and evolu-

tionary consequences of natural stressors that push organisms to the

brink of their physiological capacities may be instrumental to predict

consequences of anthropogenic environmental change

ACKNOWLEDGEMENTS

We would like to thank Lenin Arias-Rodriguez the Universidad

Juarez Autonoma de Tabasco and the Centro de Investigacion E

Innovacion Para la Ense~nanza y El Aprendizaje for their continued

collaboration enabling our work on the P mexicana system Funding

for this review article has been provided by the National Science

Foundation (IOS-1557860 IOS-1557795) and the US Army

Research Office (W911NF-15-1-0175)

AUTHOR CONTRIBUTIONS

MT wrote the first draft of the paper all authors contributed text

and comments

ORCID

Michael Tobler httporcidorg0000-0002-0326-0890

Joanna L Kelley httporcidorg0000-0002-7731-605X

Reuroudiger Riesch httporcidorg0000-0002-0223-1254

REFERENCES

Alberti M (2015) Eco-evolutionary dynamics in an urbanizing planet

Trends in Ecology amp Evolution 30 114ndash126 httpsdoiorg101016j

tree201411007

Alda F Reina R G Doadrio I amp Bermingham E (2013) Phylogeny

and biogeography of the Poecilia sphenops species complex

(Actinopterygii Poeciliidae) in Central America Molecular Phylogenet-

ics and Evolution 66 1011ndash1026 httpsdoiorg101016jympev

201212012

Alvarez del Villar J (1948) Descripcion de una nueva especie de Mol-

lienisia capturada en Ba~nos del Azufre Tabasco (Pisces Poeciliidae)

Anales de la Escuela Nacional de Ciencias Biologicas 5 275ndash281

Arnold S J (1983) Morphology performance and fitness American Zool-

ogist 23 347ndash361 httpsdoiorg101093icb232347

Aspiras A C Rohner N Martineau B Borowsky R L amp Clifford C J

(2015) Melanocortin 4 receptor mutations contribute to the adapta-

tion of cavefish to nutrient-poor conditions Proceedings of the

National Academy of Sciences USA 112 9968ndash9973

Ast G (2004) How did alternative splicing evolve Nature Reviews

Genetics 5 773ndash782 httpsdoiorg101038nrg1451

Bagarinao T (1992) Sulfide as an environmental factor and toxicant

Tolerance and adaptations in aquatic organisms Aquatic Toxicology

24 21ndash62 httpsdoiorg1010160166-445X(92)90015-F

Barton N H (1998) The effect of hitch-hiking on neutral genealogies

Genetic Research 72 123ndash133 httpsdoiorg101017

S0016672398003462

Barts N Greenway R Passow C N Arias Rodriguez L Kelley J L amp

Tobler M (2018) Expression and molecular evolution of oxygen

transport genes in livebearing fishes (Poeciliidae) from hydrogen sul-

fide rich springs Genome httpsdoiorg101139gen-2017-0051

[Epub ahead of print]

Beauchamp R O Bus J S Popp C S Boreiko C J amp Andjelkovich

D A (1984) A critical review of the literature on hydrogen sulfide

CRC Critical Reviews in Toxicology 13 25ndash97 httpsdoiorg10

310910408448409029321

Bell E M (2012) Life at extremes Environments organisms and strategies

for survival Oxfordshire CABI httpsdoiorg101079

97818459381470000

Bierbach D Arias Rodriguez L amp Plath M (2018) Intrasexual competi-

tion enhances reproductive isolation between locally adapted popula-

tions Current Zoology httpsdoiorg101093czzox071 [Epub

ahead of print]

Bierbach D Klein M Sassmannshausen V Schlupp I Riesch R

Parzefall J amp Plath M (2012) Divergent evolution of male aggres-

sive behavior Another reproductive isolation barrier in extremophile

poeciliid fishes International Journal of Evolutionary Biology 2012

148745

Blackburn D G (1999) Viviparity and oviparity Evolution and reproduc-

tive strategies In T E Knobil amp J D Neill (Eds) Encyclopedia of

reproduction Vol 4 (pp 994ndash1003) New York NY Academic Press

Bouzat J L (2010) Conservation genetics of population bottlenecks

The role of chance selection and history Conservation Genetics 11

463ndash478

Brown A P Greenway R Morgan S Quackenbush C R Giordani

L Arias-Rodriguez L Kelley J L (2017) Genome-scale data

reveals that endemic Poecilia populations from small sulfide spring

display no evidence of inbreeding Molecular Ecology 26 4920ndash

4934

Brown J H Marquet P A amp Taper M L (1993) Evolution of body

size Consequences of an energetic definition of fitness American

Naturalist 142 573ndash584 httpsdoiorg101086285558

Burton R S Pereira R J amp Barreto F S (2013) Cytonuclear genomic

interactions and hybrid breakdown Annual Review of Ecology

TOBLER ET AL | 855

between populations from sulphidic and nonsulphidic environments

Many of these phenotypic differences have evolved in convergence

across replicated sulphide spring lineages in different river drainages

Examples include modifications of morphological (body shape and

organ size Tobler et al 2011 Tobler Scharnweber et al 2015

Schulz-Mirbach et al 2016) physiological (tolerance and metabolic

rates Tobler et al 2011 Passow Arias-Rodriguez amp Tobler 2017)

behavioural (aggression and personality Riesch et al 2009 Bierbach

et al 2012 Sommer-Trembo et al 2016 Bierbach Arias Rodriguez

amp Plath 2018) and life history traits (fecundity and offspring size

Riesch Plath Garcia de Leon amp Schlupp 2010 Riesch et al 2014)

Hence proximate populations in different habitat types have

diverged in complex phenotypes that span many different suites of

traits and a key question is how divergence in multivariate pheno-

types arises In theory H2S could be a key source of selection driv-

ing multivariate phenotypic divergence either because it directly

exerts selection on a multitude of traits that are independently modi-

fied to reach new fitness peaks or because selection on a few fit-

ness-relevant traits causes simultaneous changes in correlated

characteristics Neither explanation seems sufficient because many

Abiotic environmentOxygen concentrations

Ionic compositionTemperature

Structure and flow

Biotic environmentTrophic resources

DemographyPredationParasitism

H2S

Genetic variation

Gene expression

Biochemistry

Physiologicalfunction

Organismal performance

Primary fitness components

Survivalgrowth

reproduction

Organismal function Environment

F IGURE 2 Understanding how organisms function in the context of their environment is a core question in evolutionary biology yet thecomplexities associated with organismal function and environmental variation provide significant challenges High-throughput sequencing hasallowed for the identification of candidate genes subject to selection and differential expression during adaptation to H2S The next step is tobridge the genotype-to-phenotype gap by quantifying the integrated responses to H2S across the continuum of organismal organization includingbiochemical phenotypes physiological functions whole organism performance and ultimately fitness In addition understanding adaptation toH2S-rich environments will also require acknowledging H2Srsquos relation to other abiotic and biotic environmental variables which can exert selectionand modulate aspects of organismal function This will allow for the establishment of cause-and-effect relationships between specific sources ofselection and evolutionary change in specific traits and contribute to our understanding of organismal function in natural systems

850 | TOBLER ET AL

of the divergent traits between populations from sulphidic and non-

sulphidic environments have no clearmdashor at best tenuousmdashlinks to

H2S toxicity and detoxification and there is only limited evidence

for within-population correlations between different traits groups

(Riesch et al 2016)

Evolutionary analyses frequently focus on seemingly simple

sources of selection juxtaposing populations that are assigned to dis-

crete habitat types to understand the evolution of specific traits

(Kaeuffer Peichel Bolnick amp Hendry 2012) But as Darwin (1859)

already observed in his ldquotangled bankrdquo allegory selection in natural

systems is rarely univariate but multifarious selective regimes con-

tribute to trait divergence even among populations that occur along

apparently simple environmental gradients Unquantified environmen-

tal variation that is subsumed in the categorization of habitat types

can profoundly affect trait variation in natural systems (Stuart et al

2017) highlighting the importance of considering and testing alterna-

tive hypotheses about the cause-and-effect relationship between

specific sources of selection and evolution of specific traits Thus

accurate natural history observations that consider organisms as holis-

tic entities consisting of interacting traits in the context of the abiotic

and biotic factors comprising their environment are critical to inform

the design and interpretation of empirical research in natural systems

(Greene 2005) Indeed adjacent sulphidic and nonsulphidic habitat

types do not merely differ in the presence or absence of H2S but also

in a variety of other abiotic and biotic factors (Figure 2) Unlike nonsul-

phidic habitats sulphide springs in southern Mexico are characterized

by extreme hypoxia elevated salinity and lower pH (Rosales Lagarde

2012 Tobler et al 2011) In addition the presence of physiochemical

stressors affects the biotic communities in these sulphide springs Spe-

cies richness of aquatic invertebrates and vertebrates is drastically

reduced as a consequence of H2S toxicity (Greenway Arias-Rodriguez

Diaz amp Tobler 2014) and primary production appears to be driven in

part by chemoautotrophic bacteria that use H2S as an energy source

rather than just by autochthonous and allochthonous photosynthetic

organisms (Roach Tobler amp Winemiller 2011) These changes in the

biotic environment affect the relative importance of intra-vs inter-

specific competition trophic resource use and exposure to predators

and parasites (Riesch Oranth et al 2010 Tobler et al 2014 Tobler

Scharnweber et al 2015)

31 | Alternative drivers of trait evolution

Acknowledging the complexity of selective regimes in sulphide

springs allows for a more nuanced understanding of trait evolution

and for the development of alternative hypotheses about the ulti-

mate mechanisms that cause divergence in specific genetic and phe-

notypic traits In some instances sources of selection other than H2S

can plausibly explain trait divergence between populations from sul-

phidic and nonsulphidic habitats For example hypoxia likely drives

increases in head size and gill surface area (Tobler et al 2011)

changes in the ionic composition of the water can explain differential

expression of genes associated with osmoregulation (Kelley et al

2016) and shifts in trophic resource use are associated with

modifications of the gastrointestinal system (Tobler Scharnweber

et al 2015) In other cases disentangling alternative hypotheses is

not trivial because selection mediated by H2S and other environmen-

tal factors are predicted to affect trait modifications in similar ways

Shifts towards anaerobic metabolism in sulphide spring fish could be

the consequence of inhibition of aerobic ATP production by H2S or

by constraints in oxygen availability Similarly reproductive life his-

tory shifts towards the production of larger but fewer offspring could

be driven by H2S (decreasing offspringrsquos surfacevolume ratio to mini-

mize the diffusion of H2S into the body Powell 1989) by habitat dif-

ferences in the quality and quantity of predation that can affect size-

specific rates of extrinsic mortality (Jorgensen Auer amp Reznick

2011) or by energetic constraints (see below) While comparisons

across different selective regimes can shed light into the ultimate dri-

vers of trait evolution (Moore Riesch amp Martin 2016) isolating the

effects of different sources of selection may not be possible because

they could act in concert and facilitate increases of organismal perfor-

mance in multiple contexts (Ghalambor Walker amp Reznick 2003)

When multifarious selection shapes multivariate phenotypic evo-

lution it is important to also consider the role of trade-offs both in

terms of trait expression (increased investment in one trait may

impose constraints on the expression of other traits) and in terms of

trait function (optimizing the trait for one function may impair the

traitrsquos utility for other functions) For example selection on large off-

spring size in sulphide spring fishes has inadvertent consequences

for fecundity (Riesch et al 2014) and selection for increased gill

surface areas due to hypoxia simultaneously increases exposure to

environmental H2S (Tobler et al 2011) Thus the outcome of phe-

notypic evolution is ultimately a result of balancing competing needs

in a manner that optimizes trait expression and organismal function

in multiple contexts to maximize fitness (Arnold 1983) and the pres-

ence of trade-offs can have cascading indirect effects with selection

on one or a few traits leading to the modification of a multitude of

apparently unrelated characteristics

In this context bioenergetics may represent a nexus in the evo-

lution of complex traits observed in sulphide spring fish and other

extreme environments Living in and adapting to sulphide springs

imposes constraints on organismal energy availability (Plath et al

2007) with according trade-offs for energy allocation Energy limita-

tion in sulphide springs does not arise from a scarcity of food

resources (chemoautotrophic bacteria eaten by the fish are abundant

in sulphide springs Tobler Scharnweber et al 2015) but from

behavioural and physiological adaptation A primary survival strategy

for fish in the toxic and hypoxic springs is to engage in aquatic sur-

face respiration where individuals breathe directly at the water sur-

face to maximize oxygen uptake and minimize H2S exposure This

behaviour mediates short-term survival (Plath et al 2007) but it

also constrains energy acquisition for benthic foragers like P mexi-

cana (Tobler Riesch Tobler amp Plath 2009) and increases susceptibil-

ity to bird predation (Riesch Oranth et al 2010) Costly investments

into H2S detoxification and tissue repair as well as a shift from aero-

bic to less efficient anaerobic metabolism further constrain energy

allocation to other functions (Passow Greenway Arias-Rodriguez

TOBLER ET AL | 851

Jeyasingh amp Tobler 2015) Theory predicts that energetic con-

straints ndash whether mediated by limitations in the rate of resource

acquisition or in the rate of energy allocation to reproduction (as

opposed to maintenance or growth) ndash should exert selection for a

reduction in organismal energy budgets which allows for the maxi-

mization of residual energy allocation for the production of offspring

(Brown Marquet amp Taper 1993) Several lines of evidence support

the hypothesis that energy limitation plays a critical role in pheno-

typic evolution of sulphide spring fish Compared to individuals in

nonsulphidic environments fish in sulphide springs have significantly

lower routine energy demands which is mediated both by reduc-

tions in body size and mass-specific metabolism (Passow Arias-

Rodriguez et al 2017) Reductions in other traits such as energeti-

cally costly behaviours (Bierbach et al 2012 Plath 2008) and the

size of energy-demanding organs (Schulz-Mirbach et al 2016) may

also contribute to the reduced energy demands and shifts in repro-

ductive life history traits (Riesch et al 2014) are consistent with

selection from energy limitation (Reznick amp Yang 1993)

32 | Ensuing challenges Disentanglingenvironmental and organismal complexities

Future research needs to empirically address how multifarious selec-

tive regimes shape the evolution of organisms with complex pheno-

types We have made great advances in our understanding of how

specific traits evolve and mediate adaptation (eg morphology Chan

et al 2010 coloration Dasmahapatra et al 2012 sensory organs

Cartwright Schwartz Merry amp Howell 2017) and we have a thor-

ough understanding of how single environmental factors can shape

organismal responses (eg climatic conditions Johnston amp Bennett

1996 predation Langerhans 2007 competition Pfennig amp Pfennig

2012) In contrast our understanding of phenotypic evolution in

response to multifarious selection is comparatively limited especially

because interactive effects of different environmental factors can lead

to organismal and evolutionary responses that are unpredictable based

on experiments that manipulate single environmental variables in iso-

lation (Gunderson Armstrong amp Stillman 2016 Todgham amp Stillman

2013) Future research thus needs to acknowledge the complexities of

environments and organismal design and address them explicitly in

empirical investigations to gain a robust understanding of biological

diversification (Culumber amp Tobler 2017 Stuart et al 2017)

4 | LINKING ADAPTATION TOSPECIATION

So far we primarily explored the causes of adaptive evolution but

what are its consequences Theory predicts that adaptation to local

environmental conditions can lead to speciation because reproduc-

tive isolation emerges incidentally as a by-product (ecological specia-

tion Nosil 2012) Population genetic analyses of Poecilia spp in

sulphide springs and adjacent populations in nonsulphidic environ-

ments have indicated strong genetic differentiation across small

spatial scales (in some instances lt100 m) and in the absence of

physical barriers that would prevent fish movement (Plath et al

2013) In addition population genetic structuring has been shown to

be stable even in the face of major environmental disturbance

BOX 1 Speciation in sulphide springs and conservation

biology

Adaptation to H2S-rich environments in poeciliids coincides

with ecological speciation giving rise to phenotypically dis-

tinct and reproductively isolated lineages In some

instances this has led to the evolution of highly distinct

micro-endemic species including Poecilia sulphuraria and

P thermalis Limia sulphurophila and Gambusia eurystoma

(Figure 3) Except for P sulphuraria which is found in two

spring complexes of the Rıo Pichucalco drainage of Mexico

(Tobler amp Plath 2009b) the distributions of these sulphide

spring endemics are restricted to single localities that

include only small areas of suitable habitat For example

the largest sulphide spring complex (the Ba~nos del Azufre

in southern Mexico which harbours P sulphuraria and G

eurystoma) includes ~18000 m2 of suitable habitat (equiva-

lent to lt35 football fields) while one of the smallest (La

Esperanza springs harbouring P thermalis) merely covers

~600 m2 (just over a tenth of a football field Brown et al

2017) Accordingly some sulphide spring endemics are

considered critically endangered (IUCN 2012) and pro-

tected by law (SEDESOL 1994)

Many threatened species with restricted ranges exhibit

low genetic diversity and high incidences of inbreeding

(Bouzat 2010 Willi Van Buskirk amp Hoffmann 2006) In

contrast at least P sulphuraria and P thermalismdashthe only

species investigated to datemdashactually exhibit relatively high

effective population sizes (Ne = 1200ndash3800 depending on

the population) and no evidence for inbreeding (FIS lt 0)

(Brown et al 2017) This is likely a consequence of the

long-term stability of sulphide springs and the high pro-

ductivity and low levels of interspecific competition also

support unusually high fish densities and thus population

census sizes (Brown et al 2017 Culumber et al 2016)

Nonetheless sulphide spring endemics are threatened by

habitat destruction which is mediated primarily by the

modification of springs for recreational purposes defor-

estation and conversion of land for agriculture and most

recently high-intensity palm oil plantations (Tobler amp Plath

2009ab) Despite the official protection of sulphide spring

endemics there have been no tangible conservation mea-

sures that foster the sustainable development of land

around sulphide springs with the well-being of the endemic

fishes in mind

852 | TOBLER ET AL

(catastrophic flood event Plath et al 2010) Population genetic dif-

ferentiation between adjacent sulphide-adapted and nonadapted

populations is indicative of ongoing ecological speciation (Box 1) but

different population pairs from sulphidic and nonsulphidic waters

have made variable progress along the speciation continuum towards

complete reproductive isolation (Plath et al 2013) This prompts

questions about the mechanisms involved in reducing gene flow

along the environmental gradient connecting the two habitat types

41 | Prezygotic mechanisms of reproductiveisolation

The simplest form of reproductive isolation between locally adapted

populations arises from selection against migrants between habitat

types (Nosil Vines amp Funk 2005) Translocation experiments in nat-

ural habitats have indicated strong survival selection against

migrants especially those translocated from nonsulphidic to sulphidic

habitats (Plath et al 2013) Fish translocated from nonsulphidic

environments face high mortality within short periods of time

(lt24 hr) which is likely caused by H2S toxicity and hypoxia in sul-

phide springs Hence low gene flow from populations inhabiting

nonsulphidic waters into populations inhabiting sulphidic habitats is

simply explained by a lack of adaptation to the extreme environmen-

tal conditions At least in some drainages there is also appreciable

mortality for sulphide spring fish translocated to nonsulphidic habi-

tats although mechanisms causing inviability are unclear and the

strength of selection appears to be lower (Plath et al 2013) If sur-

vival selection against immigrants was the sole source of reproduc-

tive isolation this should result in unidirectional gene flow from

H2S-adapted to nonadapted populations but population genetic

analyses provided no evidence for this implying that additional

mechanisms are at work (Plath et al 2013)

One such mechanism could be intersexual selection and mate

choice experiments have indicated that migrants from sulphidic to

nonsulphidic habitats are at a significant disadvantage with fish from

nonsulphidic habitats avoiding potential mating partners from sul-

phidic ones (Plath et al 2013 Sommer-Trembo et al 2016) Such

assortative mating has been found to be critical during ecological spe-

ciation because it counteracts the homogenizing effects of gene flow

(van Doorn Edelaar amp Weissing 2009) yet how assortative mating

patterns arise often remains unclear In theory assortative mating can

arise through a variety of processes (Nosil 2012) ldquoMagic traitrdquo mecha-

nisms where mating preferences evolve as a by-product of divergent

selection on ecological traits that serve as cues have recently garnered

much attention (Servedio van Doorn Kopp Frame amp Nosil 2011)

Consistent with magic trait mechanisms assortative mating in P mexi-

cana appears to be linked to adaptive population differences in body

shape which serve as cues to distinguish potential mating partners

from different habitats (Greenway Drexler Arias-Rodriguez amp Tobler

2016) However neither sulphide spring fish (which are unlikely to

encounter migrants from nonsulphidic habitats due to H2S toxicity)

nor allopatric P mexicana from nonsulphidic waters (which have never

experienced migrants from sulphide spring populations) exhibit a

preference for mating partners from their own ecotype suggesting

that assortative mating in this system is a product of reinforcement

(ie direct selection for assortative mating Greenway et al 2016)

This is different from classic magic traits mechanisms because assorta-

tive mating is not a mere by-product of adaptation but adaptive traits

have instead been co-opted as mating signals in response to selection

on mating preferences (Maan amp Seehausen 2012) Moreover males of

some H2S-adapted populations have reduced aggressive behaviour

(Bierbach et al 2012 2018) In staged contests of size-matched males

from different ecotypes intruders had a significantly lower likelihood

to establish dominance than resident males independent of whether

or not they showed reduced aggressiveness overall Hence intrasexual

selection may represent another mechanism mediating reproductive

isolation between locally adapted ecotypes Overall H2S-adapted poe-

ciliids allow disentangling the relative contributions of different forms

of ecological and sexual selectionmdashand their interdependencymdashduring

ecological speciation

42 | Ensuing challenges quantifying postzygoticmechanisms of reproductive isolation

In contrast to prezygotic mechanisms little is known about postzy-

gotic mechanisms of reproductive isolation that potentially act to

reduce gene flow between adjacent populations in sulphidic and

nonsulphidic habitats Putative hybrids between sulphidic and non-

sulphidic ecotypes have occasionally been observed in nonsulphidic

habitats immediately adjacent to confluences with sulphide springs

(R Riesch M Plath and M Tobler personal observations) but it

remains unclear whether and through what mechanisms such hybrids

exhibit reduced fitness compared to parentals

Direct selection of mitochondrially encoded genes in the pres-

ence of H2S (Pfenninger et al 2014) should be accompanied by co-

evolution at nuclear loci that directly interact with mitochondrial

gene products in the formation of protein complexes involved in the

OXPHOS pathway (Sunnucks Morales Lamb Pavlova amp Greening

2017 Willett amp Burton 2004) Disruption of co-evolved gene com-

plexes in hybrids should therefore lead to intrinsic fitness reductions

through mito-nuclear incompatibilities (Burton Pereira amp Barreto

2013 Wolff Ladoukakis Enrıquez amp Dowling 2014) which have

been hypothesized to be a key driver in speciation (Chou amp Leu

2010 Gershoni Templeton amp Mishmar 2009) Empirical tests of

these ideas remain to be conducted in sulphide spring fishes just

like studies on potential ecological and sexual selection against

hybrids that arise from interpopulation mating events

5 | PERSPECTIVES LIFE IN SULPHIDICENVIRONMENTS BEYOND P MEXICANA

Research on sulphide spring fishes of the P mexicana species com-

plex in Mexico has provided insights into mechanisms underlying

adaptation and the links between adaptation and speciation Sul-

phide springs do not only occur in Mexico however but are found

TOBLER ET AL | 853

worldwide and have been colonized by 24 evolutionarily indepen-

dent fish lineages in North America the Neotropics and the Middle

East (Greenway et al 2014) The vast majority of lineages belong to

the order Cyprinodontiformes (95) and livebearing members of the

family Poeciliidae in particular (80) This potentially indicates that

internal gestation of young has facilitated the colonization of

extreme environments because it shields developing embryos from

physiochemical stressors (Blackburn 1999) Among the Poeciliidae

sulphide spring fishes include members of the genera Brachyrhaphis

Gambusia Poeciliopsis Priapichthys Pseudoxiphophorus Xiphophorus

and multiple subgenera of Poecilia (Figure 3 Greenway et al 2014)

spanning over 40 million years of evolution (based on Reznick Fur-

ness Meredith amp Springer 2017)

A handful of studies have started to address molecular and phe-

notypic evolution across a broader taxon sampling of sulphide spring

fishes finding evidence for convergent evolution in morphology

(Riesch et al 2016 Tobler amp Hastings 2011) life history traits

(Riesch Plath et al 2010 Riesch et al 2014) and oxygen transport

genes (Barts et al 2018) In addition ecological speciation in

response to adaptation to H2S-rich environments has also been doc-

umented in sulphide spring poeciliids other than P mexicana (Riesch

et al 2016) These studies have laid a fruitful foundation for the

investigation of mechanisms of adaptation to H2S-rich springs from

genes to phenotypic traits linking micro-evolutionary processes to

macro-evolutionary patterns

The availability of broad-scale evolutionary replication of popula-

tions that have invaded habitats with the same physiochemical stressors

allows for addressing some of the ensuing challenges discussed above

and investigating the factors that give rise to predictable evolutionary

outcomes Convergent evolution is a commonmdashbut not omnipresentmdash

theme in adaptive evolution and multiple factors can determine the

degree of convergence among replicated lineages exposed to the same

sources of selection (Kaeuffer et al 2012 Landry amp Bernatchez 2010

Oke Rolshausen LeBlond amp Hendry 2017 Rosenblum amp Harmon

2011) Thus leveraging replicated lineages that have colonized sulphide

springs to assess molecular mechanisms underlying convergent pheno-

typic changes analyse how environmental heterogeneity among sul-

phide spring habitats causes lineage-specific patterns of trait evolution

and test whether and by what mechanisms adaptation to extreme envi-

ronments is linked to speciation will provide insights into how ecologi-

cal genetic and functional factors contribute to evolutionary

convergence at different levels of organismal organization (Rosenblum

et al 2014) An expanded taxon sampling with population pairs that

span almost the full range of the speciation continuum from panmixia to

complete reproductive isolation (Riesch et al 2016) will also help

addressing questions about temporal aspects of H2S adaptation In par-

ticular comparing younger to older lineages will provide insights about

the order at which genetic and phenotypic modifications accrue during

the adaptation process

Animals have also invaded H2S-rich habitats in other aquatic

ecosystems including cold seeps and deep-sea hydrothermal vents

(Tobler et al 2016) Adaptation to H2S in the P mexicana species com-

plex has involved modification of physiological pathways associated

with H2S toxicity (OXPHOS pathway) and detoxification (sulphide

F IGURE 3 Phylogeny of the family Poeciliidae (adopted from Culumber amp Tobler 2017) Various lineages in the genera Poecilia (subgeneraAcanthophacelus Limia and Mollienesia) Priapichthys Brachyrhaphis Poeciliopsis Xiphophorus Pseudoxiphophorus and Gambusia have colonizedH2S-rich springs in different parts of the Americas (see Greenway et al 2014) Lineages highlighted in yellow represent endemic speciesrestricted to H2S-rich springs while red lineages represent species in which some populations have invaded sulphidic habitats Pictures depict arepresentative of each focal genussubgenus

854 | TOBLER ET AL

quinone oxidoreductase pathway) which have been highly conserved

during the diversification of life on Earth (Saraste 1999 Shahak amp

Hauska 2008) The high replicability of H2S as an environmental factor

that shaped animal evolution across vastly different ecological contexts

consequently offers the opportunity to investigate whether selection

mediated by the presence of H2S has repeatedly modified these con-

served physiological pathways across animal phyla that have diverged

during the Cambrian explosion spanning over 500 million years of evo-

lution Such comparisons would provide unprecedented insights into

the effects of a single physiochemical stressor on evolutionary change

irrespective of ecological or phylogenetic contexts

6 | CONCLUSIONS

H2S-rich environments and their inhabitants represent useful models

for research in organismal biology ecology and evolution illustrating

both opportunities (eg strong and explicit source of selection evolu-

tionary replication) and challenges (eg complexity of environments

and organismal responses) in asking basic question about the origins

of biodiversity Understanding how organisms thrive in apparently

inhospitable environments will provide us with insights into lifersquos

capacities and limitations to adapt to novel environments (Waterman

1999) This is particularly relevant considering that human activities

substantially contribute to the exacerbation of physiochemical stres-

sors at local regional and global scales impacting ecological and evo-

lutionary dynamics (Alberti 2015 Norberg Urban Vellend

Klausmeier amp Loeuille 2012) Thus a focus on ecological and evolu-

tionary consequences of natural stressors that push organisms to the

brink of their physiological capacities may be instrumental to predict

consequences of anthropogenic environmental change

ACKNOWLEDGEMENTS

We would like to thank Lenin Arias-Rodriguez the Universidad

Juarez Autonoma de Tabasco and the Centro de Investigacion E

Innovacion Para la Ense~nanza y El Aprendizaje for their continued

collaboration enabling our work on the P mexicana system Funding

for this review article has been provided by the National Science

Foundation (IOS-1557860 IOS-1557795) and the US Army

Research Office (W911NF-15-1-0175)

AUTHOR CONTRIBUTIONS

MT wrote the first draft of the paper all authors contributed text

and comments

ORCID

Michael Tobler httporcidorg0000-0002-0326-0890

Joanna L Kelley httporcidorg0000-0002-7731-605X

Reuroudiger Riesch httporcidorg0000-0002-0223-1254

REFERENCES

Alberti M (2015) Eco-evolutionary dynamics in an urbanizing planet

Trends in Ecology amp Evolution 30 114ndash126 httpsdoiorg101016j

tree201411007

Alda F Reina R G Doadrio I amp Bermingham E (2013) Phylogeny

and biogeography of the Poecilia sphenops species complex

(Actinopterygii Poeciliidae) in Central America Molecular Phylogenet-

ics and Evolution 66 1011ndash1026 httpsdoiorg101016jympev

201212012

Alvarez del Villar J (1948) Descripcion de una nueva especie de Mol-

lienisia capturada en Ba~nos del Azufre Tabasco (Pisces Poeciliidae)

Anales de la Escuela Nacional de Ciencias Biologicas 5 275ndash281

Arnold S J (1983) Morphology performance and fitness American Zool-

ogist 23 347ndash361 httpsdoiorg101093icb232347

Aspiras A C Rohner N Martineau B Borowsky R L amp Clifford C J

(2015) Melanocortin 4 receptor mutations contribute to the adapta-

tion of cavefish to nutrient-poor conditions Proceedings of the

National Academy of Sciences USA 112 9968ndash9973

Ast G (2004) How did alternative splicing evolve Nature Reviews

Genetics 5 773ndash782 httpsdoiorg101038nrg1451

Bagarinao T (1992) Sulfide as an environmental factor and toxicant

Tolerance and adaptations in aquatic organisms Aquatic Toxicology

24 21ndash62 httpsdoiorg1010160166-445X(92)90015-F

Barton N H (1998) The effect of hitch-hiking on neutral genealogies

Genetic Research 72 123ndash133 httpsdoiorg101017

S0016672398003462

Barts N Greenway R Passow C N Arias Rodriguez L Kelley J L amp

Tobler M (2018) Expression and molecular evolution of oxygen

transport genes in livebearing fishes (Poeciliidae) from hydrogen sul-

fide rich springs Genome httpsdoiorg101139gen-2017-0051

[Epub ahead of print]

Beauchamp R O Bus J S Popp C S Boreiko C J amp Andjelkovich

D A (1984) A critical review of the literature on hydrogen sulfide

CRC Critical Reviews in Toxicology 13 25ndash97 httpsdoiorg10

310910408448409029321

Bell E M (2012) Life at extremes Environments organisms and strategies

for survival Oxfordshire CABI httpsdoiorg101079

97818459381470000

Bierbach D Arias Rodriguez L amp Plath M (2018) Intrasexual competi-

tion enhances reproductive isolation between locally adapted popula-

tions Current Zoology httpsdoiorg101093czzox071 [Epub

ahead of print]

Bierbach D Klein M Sassmannshausen V Schlupp I Riesch R

Parzefall J amp Plath M (2012) Divergent evolution of male aggres-

sive behavior Another reproductive isolation barrier in extremophile

poeciliid fishes International Journal of Evolutionary Biology 2012

148745

Blackburn D G (1999) Viviparity and oviparity Evolution and reproduc-

tive strategies In T E Knobil amp J D Neill (Eds) Encyclopedia of

reproduction Vol 4 (pp 994ndash1003) New York NY Academic Press

Bouzat J L (2010) Conservation genetics of population bottlenecks

The role of chance selection and history Conservation Genetics 11

463ndash478

Brown A P Greenway R Morgan S Quackenbush C R Giordani

L Arias-Rodriguez L Kelley J L (2017) Genome-scale data

reveals that endemic Poecilia populations from small sulfide spring

display no evidence of inbreeding Molecular Ecology 26 4920ndash

4934

Brown J H Marquet P A amp Taper M L (1993) Evolution of body

size Consequences of an energetic definition of fitness American

Naturalist 142 573ndash584 httpsdoiorg101086285558

Burton R S Pereira R J amp Barreto F S (2013) Cytonuclear genomic

interactions and hybrid breakdown Annual Review of Ecology

TOBLER ET AL | 855

of the divergent traits between populations from sulphidic and non-

sulphidic environments have no clearmdashor at best tenuousmdashlinks to

H2S toxicity and detoxification and there is only limited evidence

for within-population correlations between different traits groups

(Riesch et al 2016)

Evolutionary analyses frequently focus on seemingly simple

sources of selection juxtaposing populations that are assigned to dis-

crete habitat types to understand the evolution of specific traits

(Kaeuffer Peichel Bolnick amp Hendry 2012) But as Darwin (1859)

already observed in his ldquotangled bankrdquo allegory selection in natural

systems is rarely univariate but multifarious selective regimes con-

tribute to trait divergence even among populations that occur along

apparently simple environmental gradients Unquantified environmen-

tal variation that is subsumed in the categorization of habitat types

can profoundly affect trait variation in natural systems (Stuart et al

2017) highlighting the importance of considering and testing alterna-

tive hypotheses about the cause-and-effect relationship between

specific sources of selection and evolution of specific traits Thus

accurate natural history observations that consider organisms as holis-

tic entities consisting of interacting traits in the context of the abiotic

and biotic factors comprising their environment are critical to inform

the design and interpretation of empirical research in natural systems

(Greene 2005) Indeed adjacent sulphidic and nonsulphidic habitat

types do not merely differ in the presence or absence of H2S but also

in a variety of other abiotic and biotic factors (Figure 2) Unlike nonsul-

phidic habitats sulphide springs in southern Mexico are characterized

by extreme hypoxia elevated salinity and lower pH (Rosales Lagarde

2012 Tobler et al 2011) In addition the presence of physiochemical

stressors affects the biotic communities in these sulphide springs Spe-

cies richness of aquatic invertebrates and vertebrates is drastically

reduced as a consequence of H2S toxicity (Greenway Arias-Rodriguez

Diaz amp Tobler 2014) and primary production appears to be driven in

part by chemoautotrophic bacteria that use H2S as an energy source

rather than just by autochthonous and allochthonous photosynthetic

organisms (Roach Tobler amp Winemiller 2011) These changes in the

biotic environment affect the relative importance of intra-vs inter-

specific competition trophic resource use and exposure to predators

and parasites (Riesch Oranth et al 2010 Tobler et al 2014 Tobler

Scharnweber et al 2015)

31 | Alternative drivers of trait evolution

Acknowledging the complexity of selective regimes in sulphide

springs allows for a more nuanced understanding of trait evolution

and for the development of alternative hypotheses about the ulti-

mate mechanisms that cause divergence in specific genetic and phe-

notypic traits In some instances sources of selection other than H2S

can plausibly explain trait divergence between populations from sul-

phidic and nonsulphidic habitats For example hypoxia likely drives

increases in head size and gill surface area (Tobler et al 2011)

changes in the ionic composition of the water can explain differential

expression of genes associated with osmoregulation (Kelley et al

2016) and shifts in trophic resource use are associated with

modifications of the gastrointestinal system (Tobler Scharnweber

et al 2015) In other cases disentangling alternative hypotheses is

not trivial because selection mediated by H2S and other environmen-

tal factors are predicted to affect trait modifications in similar ways

Shifts towards anaerobic metabolism in sulphide spring fish could be

the consequence of inhibition of aerobic ATP production by H2S or

by constraints in oxygen availability Similarly reproductive life his-

tory shifts towards the production of larger but fewer offspring could

be driven by H2S (decreasing offspringrsquos surfacevolume ratio to mini-

mize the diffusion of H2S into the body Powell 1989) by habitat dif-

ferences in the quality and quantity of predation that can affect size-

specific rates of extrinsic mortality (Jorgensen Auer amp Reznick

2011) or by energetic constraints (see below) While comparisons

across different selective regimes can shed light into the ultimate dri-

vers of trait evolution (Moore Riesch amp Martin 2016) isolating the

effects of different sources of selection may not be possible because

they could act in concert and facilitate increases of organismal perfor-

mance in multiple contexts (Ghalambor Walker amp Reznick 2003)

When multifarious selection shapes multivariate phenotypic evo-

lution it is important to also consider the role of trade-offs both in

terms of trait expression (increased investment in one trait may

impose constraints on the expression of other traits) and in terms of

trait function (optimizing the trait for one function may impair the

traitrsquos utility for other functions) For example selection on large off-

spring size in sulphide spring fishes has inadvertent consequences

for fecundity (Riesch et al 2014) and selection for increased gill

surface areas due to hypoxia simultaneously increases exposure to

environmental H2S (Tobler et al 2011) Thus the outcome of phe-

notypic evolution is ultimately a result of balancing competing needs

in a manner that optimizes trait expression and organismal function

in multiple contexts to maximize fitness (Arnold 1983) and the pres-

ence of trade-offs can have cascading indirect effects with selection

on one or a few traits leading to the modification of a multitude of

apparently unrelated characteristics

In this context bioenergetics may represent a nexus in the evo-

lution of complex traits observed in sulphide spring fish and other

extreme environments Living in and adapting to sulphide springs

imposes constraints on organismal energy availability (Plath et al

2007) with according trade-offs for energy allocation Energy limita-

tion in sulphide springs does not arise from a scarcity of food

resources (chemoautotrophic bacteria eaten by the fish are abundant

in sulphide springs Tobler Scharnweber et al 2015) but from

behavioural and physiological adaptation A primary survival strategy

for fish in the toxic and hypoxic springs is to engage in aquatic sur-

face respiration where individuals breathe directly at the water sur-

face to maximize oxygen uptake and minimize H2S exposure This

behaviour mediates short-term survival (Plath et al 2007) but it

also constrains energy acquisition for benthic foragers like P mexi-

cana (Tobler Riesch Tobler amp Plath 2009) and increases susceptibil-

ity to bird predation (Riesch Oranth et al 2010) Costly investments

into H2S detoxification and tissue repair as well as a shift from aero-

bic to less efficient anaerobic metabolism further constrain energy

allocation to other functions (Passow Greenway Arias-Rodriguez

TOBLER ET AL | 851

Jeyasingh amp Tobler 2015) Theory predicts that energetic con-

straints ndash whether mediated by limitations in the rate of resource

acquisition or in the rate of energy allocation to reproduction (as

opposed to maintenance or growth) ndash should exert selection for a

reduction in organismal energy budgets which allows for the maxi-

mization of residual energy allocation for the production of offspring

(Brown Marquet amp Taper 1993) Several lines of evidence support

the hypothesis that energy limitation plays a critical role in pheno-

typic evolution of sulphide spring fish Compared to individuals in

nonsulphidic environments fish in sulphide springs have significantly

lower routine energy demands which is mediated both by reduc-

tions in body size and mass-specific metabolism (Passow Arias-

Rodriguez et al 2017) Reductions in other traits such as energeti-

cally costly behaviours (Bierbach et al 2012 Plath 2008) and the

size of energy-demanding organs (Schulz-Mirbach et al 2016) may

also contribute to the reduced energy demands and shifts in repro-

ductive life history traits (Riesch et al 2014) are consistent with

selection from energy limitation (Reznick amp Yang 1993)

32 | Ensuing challenges Disentanglingenvironmental and organismal complexities

Future research needs to empirically address how multifarious selec-

tive regimes shape the evolution of organisms with complex pheno-

types We have made great advances in our understanding of how

specific traits evolve and mediate adaptation (eg morphology Chan

et al 2010 coloration Dasmahapatra et al 2012 sensory organs

Cartwright Schwartz Merry amp Howell 2017) and we have a thor-

ough understanding of how single environmental factors can shape

organismal responses (eg climatic conditions Johnston amp Bennett

1996 predation Langerhans 2007 competition Pfennig amp Pfennig

2012) In contrast our understanding of phenotypic evolution in

response to multifarious selection is comparatively limited especially

because interactive effects of different environmental factors can lead

to organismal and evolutionary responses that are unpredictable based

on experiments that manipulate single environmental variables in iso-

lation (Gunderson Armstrong amp Stillman 2016 Todgham amp Stillman

2013) Future research thus needs to acknowledge the complexities of

environments and organismal design and address them explicitly in

empirical investigations to gain a robust understanding of biological

diversification (Culumber amp Tobler 2017 Stuart et al 2017)

4 | LINKING ADAPTATION TOSPECIATION

So far we primarily explored the causes of adaptive evolution but

what are its consequences Theory predicts that adaptation to local

environmental conditions can lead to speciation because reproduc-

tive isolation emerges incidentally as a by-product (ecological specia-

tion Nosil 2012) Population genetic analyses of Poecilia spp in

sulphide springs and adjacent populations in nonsulphidic environ-

ments have indicated strong genetic differentiation across small

spatial scales (in some instances lt100 m) and in the absence of

physical barriers that would prevent fish movement (Plath et al

2013) In addition population genetic structuring has been shown to

be stable even in the face of major environmental disturbance

BOX 1 Speciation in sulphide springs and conservation

biology

Adaptation to H2S-rich environments in poeciliids coincides

with ecological speciation giving rise to phenotypically dis-

tinct and reproductively isolated lineages In some

instances this has led to the evolution of highly distinct

micro-endemic species including Poecilia sulphuraria and

P thermalis Limia sulphurophila and Gambusia eurystoma

(Figure 3) Except for P sulphuraria which is found in two

spring complexes of the Rıo Pichucalco drainage of Mexico

(Tobler amp Plath 2009b) the distributions of these sulphide

spring endemics are restricted to single localities that

include only small areas of suitable habitat For example

the largest sulphide spring complex (the Ba~nos del Azufre

in southern Mexico which harbours P sulphuraria and G

eurystoma) includes ~18000 m2 of suitable habitat (equiva-

lent to lt35 football fields) while one of the smallest (La

Esperanza springs harbouring P thermalis) merely covers

~600 m2 (just over a tenth of a football field Brown et al

2017) Accordingly some sulphide spring endemics are

considered critically endangered (IUCN 2012) and pro-

tected by law (SEDESOL 1994)

Many threatened species with restricted ranges exhibit

low genetic diversity and high incidences of inbreeding

(Bouzat 2010 Willi Van Buskirk amp Hoffmann 2006) In

contrast at least P sulphuraria and P thermalismdashthe only

species investigated to datemdashactually exhibit relatively high

effective population sizes (Ne = 1200ndash3800 depending on

the population) and no evidence for inbreeding (FIS lt 0)

(Brown et al 2017) This is likely a consequence of the

long-term stability of sulphide springs and the high pro-

ductivity and low levels of interspecific competition also

support unusually high fish densities and thus population

census sizes (Brown et al 2017 Culumber et al 2016)

Nonetheless sulphide spring endemics are threatened by

habitat destruction which is mediated primarily by the

modification of springs for recreational purposes defor-

estation and conversion of land for agriculture and most

recently high-intensity palm oil plantations (Tobler amp Plath

2009ab) Despite the official protection of sulphide spring

endemics there have been no tangible conservation mea-

sures that foster the sustainable development of land

around sulphide springs with the well-being of the endemic

fishes in mind

852 | TOBLER ET AL

(catastrophic flood event Plath et al 2010) Population genetic dif-

ferentiation between adjacent sulphide-adapted and nonadapted

populations is indicative of ongoing ecological speciation (Box 1) but

different population pairs from sulphidic and nonsulphidic waters

have made variable progress along the speciation continuum towards

complete reproductive isolation (Plath et al 2013) This prompts

questions about the mechanisms involved in reducing gene flow

along the environmental gradient connecting the two habitat types

41 | Prezygotic mechanisms of reproductiveisolation

The simplest form of reproductive isolation between locally adapted

populations arises from selection against migrants between habitat

types (Nosil Vines amp Funk 2005) Translocation experiments in nat-

ural habitats have indicated strong survival selection against

migrants especially those translocated from nonsulphidic to sulphidic

habitats (Plath et al 2013) Fish translocated from nonsulphidic

environments face high mortality within short periods of time

(lt24 hr) which is likely caused by H2S toxicity and hypoxia in sul-

phide springs Hence low gene flow from populations inhabiting

nonsulphidic waters into populations inhabiting sulphidic habitats is

simply explained by a lack of adaptation to the extreme environmen-

tal conditions At least in some drainages there is also appreciable

mortality for sulphide spring fish translocated to nonsulphidic habi-

tats although mechanisms causing inviability are unclear and the

strength of selection appears to be lower (Plath et al 2013) If sur-

vival selection against immigrants was the sole source of reproduc-

tive isolation this should result in unidirectional gene flow from

H2S-adapted to nonadapted populations but population genetic

analyses provided no evidence for this implying that additional

mechanisms are at work (Plath et al 2013)

One such mechanism could be intersexual selection and mate

choice experiments have indicated that migrants from sulphidic to

nonsulphidic habitats are at a significant disadvantage with fish from

nonsulphidic habitats avoiding potential mating partners from sul-

phidic ones (Plath et al 2013 Sommer-Trembo et al 2016) Such

assortative mating has been found to be critical during ecological spe-

ciation because it counteracts the homogenizing effects of gene flow

(van Doorn Edelaar amp Weissing 2009) yet how assortative mating

patterns arise often remains unclear In theory assortative mating can

arise through a variety of processes (Nosil 2012) ldquoMagic traitrdquo mecha-

nisms where mating preferences evolve as a by-product of divergent

selection on ecological traits that serve as cues have recently garnered

much attention (Servedio van Doorn Kopp Frame amp Nosil 2011)

Consistent with magic trait mechanisms assortative mating in P mexi-

cana appears to be linked to adaptive population differences in body

shape which serve as cues to distinguish potential mating partners

from different habitats (Greenway Drexler Arias-Rodriguez amp Tobler

2016) However neither sulphide spring fish (which are unlikely to

encounter migrants from nonsulphidic habitats due to H2S toxicity)

nor allopatric P mexicana from nonsulphidic waters (which have never

experienced migrants from sulphide spring populations) exhibit a

preference for mating partners from their own ecotype suggesting

that assortative mating in this system is a product of reinforcement

(ie direct selection for assortative mating Greenway et al 2016)

This is different from classic magic traits mechanisms because assorta-

tive mating is not a mere by-product of adaptation but adaptive traits

have instead been co-opted as mating signals in response to selection

on mating preferences (Maan amp Seehausen 2012) Moreover males of

some H2S-adapted populations have reduced aggressive behaviour

(Bierbach et al 2012 2018) In staged contests of size-matched males

from different ecotypes intruders had a significantly lower likelihood

to establish dominance than resident males independent of whether

or not they showed reduced aggressiveness overall Hence intrasexual

selection may represent another mechanism mediating reproductive

isolation between locally adapted ecotypes Overall H2S-adapted poe-

ciliids allow disentangling the relative contributions of different forms

of ecological and sexual selectionmdashand their interdependencymdashduring

ecological speciation

42 | Ensuing challenges quantifying postzygoticmechanisms of reproductive isolation

In contrast to prezygotic mechanisms little is known about postzy-

gotic mechanisms of reproductive isolation that potentially act to

reduce gene flow between adjacent populations in sulphidic and

nonsulphidic habitats Putative hybrids between sulphidic and non-

sulphidic ecotypes have occasionally been observed in nonsulphidic

habitats immediately adjacent to confluences with sulphide springs

(R Riesch M Plath and M Tobler personal observations) but it

remains unclear whether and through what mechanisms such hybrids

exhibit reduced fitness compared to parentals

Direct selection of mitochondrially encoded genes in the pres-

ence of H2S (Pfenninger et al 2014) should be accompanied by co-

evolution at nuclear loci that directly interact with mitochondrial

gene products in the formation of protein complexes involved in the

OXPHOS pathway (Sunnucks Morales Lamb Pavlova amp Greening

2017 Willett amp Burton 2004) Disruption of co-evolved gene com-

plexes in hybrids should therefore lead to intrinsic fitness reductions

through mito-nuclear incompatibilities (Burton Pereira amp Barreto

2013 Wolff Ladoukakis Enrıquez amp Dowling 2014) which have

been hypothesized to be a key driver in speciation (Chou amp Leu

2010 Gershoni Templeton amp Mishmar 2009) Empirical tests of

these ideas remain to be conducted in sulphide spring fishes just

like studies on potential ecological and sexual selection against

hybrids that arise from interpopulation mating events

5 | PERSPECTIVES LIFE IN SULPHIDICENVIRONMENTS BEYOND P MEXICANA

Research on sulphide spring fishes of the P mexicana species com-

plex in Mexico has provided insights into mechanisms underlying

adaptation and the links between adaptation and speciation Sul-

phide springs do not only occur in Mexico however but are found

TOBLER ET AL | 853

worldwide and have been colonized by 24 evolutionarily indepen-

dent fish lineages in North America the Neotropics and the Middle

East (Greenway et al 2014) The vast majority of lineages belong to

the order Cyprinodontiformes (95) and livebearing members of the

family Poeciliidae in particular (80) This potentially indicates that

internal gestation of young has facilitated the colonization of

extreme environments because it shields developing embryos from

physiochemical stressors (Blackburn 1999) Among the Poeciliidae

sulphide spring fishes include members of the genera Brachyrhaphis

Gambusia Poeciliopsis Priapichthys Pseudoxiphophorus Xiphophorus

and multiple subgenera of Poecilia (Figure 3 Greenway et al 2014)

spanning over 40 million years of evolution (based on Reznick Fur-

ness Meredith amp Springer 2017)

A handful of studies have started to address molecular and phe-

notypic evolution across a broader taxon sampling of sulphide spring

fishes finding evidence for convergent evolution in morphology

(Riesch et al 2016 Tobler amp Hastings 2011) life history traits

(Riesch Plath et al 2010 Riesch et al 2014) and oxygen transport

genes (Barts et al 2018) In addition ecological speciation in

response to adaptation to H2S-rich environments has also been doc-

umented in sulphide spring poeciliids other than P mexicana (Riesch

et al 2016) These studies have laid a fruitful foundation for the

investigation of mechanisms of adaptation to H2S-rich springs from

genes to phenotypic traits linking micro-evolutionary processes to

macro-evolutionary patterns

The availability of broad-scale evolutionary replication of popula-

tions that have invaded habitats with the same physiochemical stressors

allows for addressing some of the ensuing challenges discussed above

and investigating the factors that give rise to predictable evolutionary

outcomes Convergent evolution is a commonmdashbut not omnipresentmdash

theme in adaptive evolution and multiple factors can determine the

degree of convergence among replicated lineages exposed to the same

sources of selection (Kaeuffer et al 2012 Landry amp Bernatchez 2010

Oke Rolshausen LeBlond amp Hendry 2017 Rosenblum amp Harmon

2011) Thus leveraging replicated lineages that have colonized sulphide

springs to assess molecular mechanisms underlying convergent pheno-

typic changes analyse how environmental heterogeneity among sul-

phide spring habitats causes lineage-specific patterns of trait evolution

and test whether and by what mechanisms adaptation to extreme envi-

ronments is linked to speciation will provide insights into how ecologi-

cal genetic and functional factors contribute to evolutionary

convergence at different levels of organismal organization (Rosenblum

et al 2014) An expanded taxon sampling with population pairs that

span almost the full range of the speciation continuum from panmixia to

complete reproductive isolation (Riesch et al 2016) will also help

addressing questions about temporal aspects of H2S adaptation In par-

ticular comparing younger to older lineages will provide insights about

the order at which genetic and phenotypic modifications accrue during

the adaptation process

Animals have also invaded H2S-rich habitats in other aquatic

ecosystems including cold seeps and deep-sea hydrothermal vents

(Tobler et al 2016) Adaptation to H2S in the P mexicana species com-

plex has involved modification of physiological pathways associated

with H2S toxicity (OXPHOS pathway) and detoxification (sulphide

F IGURE 3 Phylogeny of the family Poeciliidae (adopted from Culumber amp Tobler 2017) Various lineages in the genera Poecilia (subgeneraAcanthophacelus Limia and Mollienesia) Priapichthys Brachyrhaphis Poeciliopsis Xiphophorus Pseudoxiphophorus and Gambusia have colonizedH2S-rich springs in different parts of the Americas (see Greenway et al 2014) Lineages highlighted in yellow represent endemic speciesrestricted to H2S-rich springs while red lineages represent species in which some populations have invaded sulphidic habitats Pictures depict arepresentative of each focal genussubgenus

854 | TOBLER ET AL

quinone oxidoreductase pathway) which have been highly conserved

during the diversification of life on Earth (Saraste 1999 Shahak amp

Hauska 2008) The high replicability of H2S as an environmental factor

that shaped animal evolution across vastly different ecological contexts

consequently offers the opportunity to investigate whether selection

mediated by the presence of H2S has repeatedly modified these con-

served physiological pathways across animal phyla that have diverged

during the Cambrian explosion spanning over 500 million years of evo-

lution Such comparisons would provide unprecedented insights into

the effects of a single physiochemical stressor on evolutionary change

irrespective of ecological or phylogenetic contexts

6 | CONCLUSIONS

H2S-rich environments and their inhabitants represent useful models

for research in organismal biology ecology and evolution illustrating

both opportunities (eg strong and explicit source of selection evolu-

tionary replication) and challenges (eg complexity of environments

and organismal responses) in asking basic question about the origins

of biodiversity Understanding how organisms thrive in apparently

inhospitable environments will provide us with insights into lifersquos

capacities and limitations to adapt to novel environments (Waterman

1999) This is particularly relevant considering that human activities

substantially contribute to the exacerbation of physiochemical stres-

sors at local regional and global scales impacting ecological and evo-

lutionary dynamics (Alberti 2015 Norberg Urban Vellend

Klausmeier amp Loeuille 2012) Thus a focus on ecological and evolu-

tionary consequences of natural stressors that push organisms to the

brink of their physiological capacities may be instrumental to predict

consequences of anthropogenic environmental change

ACKNOWLEDGEMENTS

We would like to thank Lenin Arias-Rodriguez the Universidad

Juarez Autonoma de Tabasco and the Centro de Investigacion E

Innovacion Para la Ense~nanza y El Aprendizaje for their continued

collaboration enabling our work on the P mexicana system Funding

for this review article has been provided by the National Science

Foundation (IOS-1557860 IOS-1557795) and the US Army

Research Office (W911NF-15-1-0175)

AUTHOR CONTRIBUTIONS

MT wrote the first draft of the paper all authors contributed text

and comments

ORCID

Michael Tobler httporcidorg0000-0002-0326-0890

Joanna L Kelley httporcidorg0000-0002-7731-605X

Reuroudiger Riesch httporcidorg0000-0002-0223-1254

REFERENCES

Alberti M (2015) Eco-evolutionary dynamics in an urbanizing planet

Trends in Ecology amp Evolution 30 114ndash126 httpsdoiorg101016j

tree201411007

Alda F Reina R G Doadrio I amp Bermingham E (2013) Phylogeny

and biogeography of the Poecilia sphenops species complex

(Actinopterygii Poeciliidae) in Central America Molecular Phylogenet-

ics and Evolution 66 1011ndash1026 httpsdoiorg101016jympev

201212012

Alvarez del Villar J (1948) Descripcion de una nueva especie de Mol-

lienisia capturada en Ba~nos del Azufre Tabasco (Pisces Poeciliidae)

Anales de la Escuela Nacional de Ciencias Biologicas 5 275ndash281

Arnold S J (1983) Morphology performance and fitness American Zool-

ogist 23 347ndash361 httpsdoiorg101093icb232347

Aspiras A C Rohner N Martineau B Borowsky R L amp Clifford C J

(2015) Melanocortin 4 receptor mutations contribute to the adapta-

tion of cavefish to nutrient-poor conditions Proceedings of the

National Academy of Sciences USA 112 9968ndash9973

Ast G (2004) How did alternative splicing evolve Nature Reviews

Genetics 5 773ndash782 httpsdoiorg101038nrg1451

Bagarinao T (1992) Sulfide as an environmental factor and toxicant

Tolerance and adaptations in aquatic organisms Aquatic Toxicology

24 21ndash62 httpsdoiorg1010160166-445X(92)90015-F

Barton N H (1998) The effect of hitch-hiking on neutral genealogies

Genetic Research 72 123ndash133 httpsdoiorg101017

S0016672398003462

Barts N Greenway R Passow C N Arias Rodriguez L Kelley J L amp

Tobler M (2018) Expression and molecular evolution of oxygen

transport genes in livebearing fishes (Poeciliidae) from hydrogen sul-

fide rich springs Genome httpsdoiorg101139gen-2017-0051

[Epub ahead of print]

Beauchamp R O Bus J S Popp C S Boreiko C J amp Andjelkovich

D A (1984) A critical review of the literature on hydrogen sulfide

CRC Critical Reviews in Toxicology 13 25ndash97 httpsdoiorg10

310910408448409029321

Bell E M (2012) Life at extremes Environments organisms and strategies

for survival Oxfordshire CABI httpsdoiorg101079

97818459381470000

Bierbach D Arias Rodriguez L amp Plath M (2018) Intrasexual competi-

tion enhances reproductive isolation between locally adapted popula-

tions Current Zoology httpsdoiorg101093czzox071 [Epub

ahead of print]

Bierbach D Klein M Sassmannshausen V Schlupp I Riesch R

Parzefall J amp Plath M (2012) Divergent evolution of male aggres-

sive behavior Another reproductive isolation barrier in extremophile

poeciliid fishes International Journal of Evolutionary Biology 2012

148745

Blackburn D G (1999) Viviparity and oviparity Evolution and reproduc-

tive strategies In T E Knobil amp J D Neill (Eds) Encyclopedia of

reproduction Vol 4 (pp 994ndash1003) New York NY Academic Press

Bouzat J L (2010) Conservation genetics of population bottlenecks

The role of chance selection and history Conservation Genetics 11

463ndash478

Brown A P Greenway R Morgan S Quackenbush C R Giordani

L Arias-Rodriguez L Kelley J L (2017) Genome-scale data

reveals that endemic Poecilia populations from small sulfide spring

display no evidence of inbreeding Molecular Ecology 26 4920ndash

4934

Brown J H Marquet P A amp Taper M L (1993) Evolution of body

size Consequences of an energetic definition of fitness American

Naturalist 142 573ndash584 httpsdoiorg101086285558

Burton R S Pereira R J amp Barreto F S (2013) Cytonuclear genomic

interactions and hybrid breakdown Annual Review of Ecology

TOBLER ET AL | 855

Jeyasingh amp Tobler 2015) Theory predicts that energetic con-

straints ndash whether mediated by limitations in the rate of resource

acquisition or in the rate of energy allocation to reproduction (as

opposed to maintenance or growth) ndash should exert selection for a

reduction in organismal energy budgets which allows for the maxi-

mization of residual energy allocation for the production of offspring

(Brown Marquet amp Taper 1993) Several lines of evidence support

the hypothesis that energy limitation plays a critical role in pheno-

typic evolution of sulphide spring fish Compared to individuals in

nonsulphidic environments fish in sulphide springs have significantly

lower routine energy demands which is mediated both by reduc-

tions in body size and mass-specific metabolism (Passow Arias-

Rodriguez et al 2017) Reductions in other traits such as energeti-

cally costly behaviours (Bierbach et al 2012 Plath 2008) and the

size of energy-demanding organs (Schulz-Mirbach et al 2016) may

also contribute to the reduced energy demands and shifts in repro-

ductive life history traits (Riesch et al 2014) are consistent with

selection from energy limitation (Reznick amp Yang 1993)

32 | Ensuing challenges Disentanglingenvironmental and organismal complexities

Future research needs to empirically address how multifarious selec-

tive regimes shape the evolution of organisms with complex pheno-

types We have made great advances in our understanding of how

specific traits evolve and mediate adaptation (eg morphology Chan

et al 2010 coloration Dasmahapatra et al 2012 sensory organs

Cartwright Schwartz Merry amp Howell 2017) and we have a thor-

ough understanding of how single environmental factors can shape

organismal responses (eg climatic conditions Johnston amp Bennett

1996 predation Langerhans 2007 competition Pfennig amp Pfennig

2012) In contrast our understanding of phenotypic evolution in

response to multifarious selection is comparatively limited especially

because interactive effects of different environmental factors can lead

to organismal and evolutionary responses that are unpredictable based

on experiments that manipulate single environmental variables in iso-

lation (Gunderson Armstrong amp Stillman 2016 Todgham amp Stillman

2013) Future research thus needs to acknowledge the complexities of

environments and organismal design and address them explicitly in

empirical investigations to gain a robust understanding of biological

diversification (Culumber amp Tobler 2017 Stuart et al 2017)

4 | LINKING ADAPTATION TOSPECIATION

So far we primarily explored the causes of adaptive evolution but

what are its consequences Theory predicts that adaptation to local

environmental conditions can lead to speciation because reproduc-

tive isolation emerges incidentally as a by-product (ecological specia-

tion Nosil 2012) Population genetic analyses of Poecilia spp in

sulphide springs and adjacent populations in nonsulphidic environ-

ments have indicated strong genetic differentiation across small

spatial scales (in some instances lt100 m) and in the absence of

physical barriers that would prevent fish movement (Plath et al

2013) In addition population genetic structuring has been shown to

be stable even in the face of major environmental disturbance

BOX 1 Speciation in sulphide springs and conservation

biology

Adaptation to H2S-rich environments in poeciliids coincides

with ecological speciation giving rise to phenotypically dis-

tinct and reproductively isolated lineages In some

instances this has led to the evolution of highly distinct

micro-endemic species including Poecilia sulphuraria and

P thermalis Limia sulphurophila and Gambusia eurystoma

(Figure 3) Except for P sulphuraria which is found in two

spring complexes of the Rıo Pichucalco drainage of Mexico

(Tobler amp Plath 2009b) the distributions of these sulphide

spring endemics are restricted to single localities that

include only small areas of suitable habitat For example

the largest sulphide spring complex (the Ba~nos del Azufre

in southern Mexico which harbours P sulphuraria and G

eurystoma) includes ~18000 m2 of suitable habitat (equiva-

lent to lt35 football fields) while one of the smallest (La

Esperanza springs harbouring P thermalis) merely covers

~600 m2 (just over a tenth of a football field Brown et al

2017) Accordingly some sulphide spring endemics are

considered critically endangered (IUCN 2012) and pro-

tected by law (SEDESOL 1994)

Many threatened species with restricted ranges exhibit

low genetic diversity and high incidences of inbreeding

(Bouzat 2010 Willi Van Buskirk amp Hoffmann 2006) In

contrast at least P sulphuraria and P thermalismdashthe only

species investigated to datemdashactually exhibit relatively high

effective population sizes (Ne = 1200ndash3800 depending on

the population) and no evidence for inbreeding (FIS lt 0)

(Brown et al 2017) This is likely a consequence of the

long-term stability of sulphide springs and the high pro-

ductivity and low levels of interspecific competition also

support unusually high fish densities and thus population

census sizes (Brown et al 2017 Culumber et al 2016)

Nonetheless sulphide spring endemics are threatened by

habitat destruction which is mediated primarily by the

modification of springs for recreational purposes defor-

estation and conversion of land for agriculture and most

recently high-intensity palm oil plantations (Tobler amp Plath

2009ab) Despite the official protection of sulphide spring

endemics there have been no tangible conservation mea-

sures that foster the sustainable development of land

around sulphide springs with the well-being of the endemic

fishes in mind

852 | TOBLER ET AL

(catastrophic flood event Plath et al 2010) Population genetic dif-

ferentiation between adjacent sulphide-adapted and nonadapted

populations is indicative of ongoing ecological speciation (Box 1) but

different population pairs from sulphidic and nonsulphidic waters

have made variable progress along the speciation continuum towards

complete reproductive isolation (Plath et al 2013) This prompts

questions about the mechanisms involved in reducing gene flow

along the environmental gradient connecting the two habitat types

41 | Prezygotic mechanisms of reproductiveisolation

The simplest form of reproductive isolation between locally adapted

populations arises from selection against migrants between habitat

types (Nosil Vines amp Funk 2005) Translocation experiments in nat-

ural habitats have indicated strong survival selection against

migrants especially those translocated from nonsulphidic to sulphidic

habitats (Plath et al 2013) Fish translocated from nonsulphidic

environments face high mortality within short periods of time

(lt24 hr) which is likely caused by H2S toxicity and hypoxia in sul-

phide springs Hence low gene flow from populations inhabiting

nonsulphidic waters into populations inhabiting sulphidic habitats is

simply explained by a lack of adaptation to the extreme environmen-

tal conditions At least in some drainages there is also appreciable

mortality for sulphide spring fish translocated to nonsulphidic habi-

tats although mechanisms causing inviability are unclear and the

strength of selection appears to be lower (Plath et al 2013) If sur-

vival selection against immigrants was the sole source of reproduc-

tive isolation this should result in unidirectional gene flow from

H2S-adapted to nonadapted populations but population genetic

analyses provided no evidence for this implying that additional

mechanisms are at work (Plath et al 2013)

One such mechanism could be intersexual selection and mate

choice experiments have indicated that migrants from sulphidic to

nonsulphidic habitats are at a significant disadvantage with fish from

nonsulphidic habitats avoiding potential mating partners from sul-

phidic ones (Plath et al 2013 Sommer-Trembo et al 2016) Such

assortative mating has been found to be critical during ecological spe-

ciation because it counteracts the homogenizing effects of gene flow

(van Doorn Edelaar amp Weissing 2009) yet how assortative mating

patterns arise often remains unclear In theory assortative mating can

arise through a variety of processes (Nosil 2012) ldquoMagic traitrdquo mecha-

nisms where mating preferences evolve as a by-product of divergent

selection on ecological traits that serve as cues have recently garnered

much attention (Servedio van Doorn Kopp Frame amp Nosil 2011)

Consistent with magic trait mechanisms assortative mating in P mexi-

cana appears to be linked to adaptive population differences in body

shape which serve as cues to distinguish potential mating partners

from different habitats (Greenway Drexler Arias-Rodriguez amp Tobler

2016) However neither sulphide spring fish (which are unlikely to

encounter migrants from nonsulphidic habitats due to H2S toxicity)

nor allopatric P mexicana from nonsulphidic waters (which have never

experienced migrants from sulphide spring populations) exhibit a

preference for mating partners from their own ecotype suggesting

that assortative mating in this system is a product of reinforcement

(ie direct selection for assortative mating Greenway et al 2016)

This is different from classic magic traits mechanisms because assorta-

tive mating is not a mere by-product of adaptation but adaptive traits

have instead been co-opted as mating signals in response to selection

on mating preferences (Maan amp Seehausen 2012) Moreover males of

some H2S-adapted populations have reduced aggressive behaviour

(Bierbach et al 2012 2018) In staged contests of size-matched males

from different ecotypes intruders had a significantly lower likelihood

to establish dominance than resident males independent of whether

or not they showed reduced aggressiveness overall Hence intrasexual

selection may represent another mechanism mediating reproductive

isolation between locally adapted ecotypes Overall H2S-adapted poe-

ciliids allow disentangling the relative contributions of different forms

of ecological and sexual selectionmdashand their interdependencymdashduring

ecological speciation

42 | Ensuing challenges quantifying postzygoticmechanisms of reproductive isolation

In contrast to prezygotic mechanisms little is known about postzy-

gotic mechanisms of reproductive isolation that potentially act to

reduce gene flow between adjacent populations in sulphidic and

nonsulphidic habitats Putative hybrids between sulphidic and non-

sulphidic ecotypes have occasionally been observed in nonsulphidic

habitats immediately adjacent to confluences with sulphide springs

(R Riesch M Plath and M Tobler personal observations) but it

remains unclear whether and through what mechanisms such hybrids

exhibit reduced fitness compared to parentals

Direct selection of mitochondrially encoded genes in the pres-

ence of H2S (Pfenninger et al 2014) should be accompanied by co-

evolution at nuclear loci that directly interact with mitochondrial

gene products in the formation of protein complexes involved in the

OXPHOS pathway (Sunnucks Morales Lamb Pavlova amp Greening

2017 Willett amp Burton 2004) Disruption of co-evolved gene com-

plexes in hybrids should therefore lead to intrinsic fitness reductions

through mito-nuclear incompatibilities (Burton Pereira amp Barreto

2013 Wolff Ladoukakis Enrıquez amp Dowling 2014) which have

been hypothesized to be a key driver in speciation (Chou amp Leu

2010 Gershoni Templeton amp Mishmar 2009) Empirical tests of

these ideas remain to be conducted in sulphide spring fishes just

like studies on potential ecological and sexual selection against

hybrids that arise from interpopulation mating events

5 | PERSPECTIVES LIFE IN SULPHIDICENVIRONMENTS BEYOND P MEXICANA

Research on sulphide spring fishes of the P mexicana species com-

plex in Mexico has provided insights into mechanisms underlying

adaptation and the links between adaptation and speciation Sul-

phide springs do not only occur in Mexico however but are found

TOBLER ET AL | 853

worldwide and have been colonized by 24 evolutionarily indepen-

dent fish lineages in North America the Neotropics and the Middle

East (Greenway et al 2014) The vast majority of lineages belong to

the order Cyprinodontiformes (95) and livebearing members of the

family Poeciliidae in particular (80) This potentially indicates that

internal gestation of young has facilitated the colonization of

extreme environments because it shields developing embryos from

physiochemical stressors (Blackburn 1999) Among the Poeciliidae

sulphide spring fishes include members of the genera Brachyrhaphis

Gambusia Poeciliopsis Priapichthys Pseudoxiphophorus Xiphophorus

and multiple subgenera of Poecilia (Figure 3 Greenway et al 2014)

spanning over 40 million years of evolution (based on Reznick Fur-

ness Meredith amp Springer 2017)

A handful of studies have started to address molecular and phe-

notypic evolution across a broader taxon sampling of sulphide spring

fishes finding evidence for convergent evolution in morphology

(Riesch et al 2016 Tobler amp Hastings 2011) life history traits

(Riesch Plath et al 2010 Riesch et al 2014) and oxygen transport

genes (Barts et al 2018) In addition ecological speciation in

response to adaptation to H2S-rich environments has also been doc-

umented in sulphide spring poeciliids other than P mexicana (Riesch

et al 2016) These studies have laid a fruitful foundation for the

investigation of mechanisms of adaptation to H2S-rich springs from

genes to phenotypic traits linking micro-evolutionary processes to

macro-evolutionary patterns

The availability of broad-scale evolutionary replication of popula-

tions that have invaded habitats with the same physiochemical stressors

allows for addressing some of the ensuing challenges discussed above

and investigating the factors that give rise to predictable evolutionary

outcomes Convergent evolution is a commonmdashbut not omnipresentmdash

theme in adaptive evolution and multiple factors can determine the

degree of convergence among replicated lineages exposed to the same

sources of selection (Kaeuffer et al 2012 Landry amp Bernatchez 2010

Oke Rolshausen LeBlond amp Hendry 2017 Rosenblum amp Harmon

2011) Thus leveraging replicated lineages that have colonized sulphide

springs to assess molecular mechanisms underlying convergent pheno-

typic changes analyse how environmental heterogeneity among sul-

phide spring habitats causes lineage-specific patterns of trait evolution

and test whether and by what mechanisms adaptation to extreme envi-

ronments is linked to speciation will provide insights into how ecologi-

cal genetic and functional factors contribute to evolutionary

convergence at different levels of organismal organization (Rosenblum

et al 2014) An expanded taxon sampling with population pairs that

span almost the full range of the speciation continuum from panmixia to

complete reproductive isolation (Riesch et al 2016) will also help

addressing questions about temporal aspects of H2S adaptation In par-

ticular comparing younger to older lineages will provide insights about

the order at which genetic and phenotypic modifications accrue during

the adaptation process

Animals have also invaded H2S-rich habitats in other aquatic

ecosystems including cold seeps and deep-sea hydrothermal vents

(Tobler et al 2016) Adaptation to H2S in the P mexicana species com-

plex has involved modification of physiological pathways associated

with H2S toxicity (OXPHOS pathway) and detoxification (sulphide

F IGURE 3 Phylogeny of the family Poeciliidae (adopted from Culumber amp Tobler 2017) Various lineages in the genera Poecilia (subgeneraAcanthophacelus Limia and Mollienesia) Priapichthys Brachyrhaphis Poeciliopsis Xiphophorus Pseudoxiphophorus and Gambusia have colonizedH2S-rich springs in different parts of the Americas (see Greenway et al 2014) Lineages highlighted in yellow represent endemic speciesrestricted to H2S-rich springs while red lineages represent species in which some populations have invaded sulphidic habitats Pictures depict arepresentative of each focal genussubgenus

854 | TOBLER ET AL

quinone oxidoreductase pathway) which have been highly conserved

during the diversification of life on Earth (Saraste 1999 Shahak amp

Hauska 2008) The high replicability of H2S as an environmental factor

that shaped animal evolution across vastly different ecological contexts

consequently offers the opportunity to investigate whether selection

mediated by the presence of H2S has repeatedly modified these con-

served physiological pathways across animal phyla that have diverged

during the Cambrian explosion spanning over 500 million years of evo-

lution Such comparisons would provide unprecedented insights into

the effects of a single physiochemical stressor on evolutionary change

irrespective of ecological or phylogenetic contexts

6 | CONCLUSIONS

H2S-rich environments and their inhabitants represent useful models

for research in organismal biology ecology and evolution illustrating

both opportunities (eg strong and explicit source of selection evolu-

tionary replication) and challenges (eg complexity of environments

and organismal responses) in asking basic question about the origins

of biodiversity Understanding how organisms thrive in apparently

inhospitable environments will provide us with insights into lifersquos

capacities and limitations to adapt to novel environments (Waterman

1999) This is particularly relevant considering that human activities

substantially contribute to the exacerbation of physiochemical stres-

sors at local regional and global scales impacting ecological and evo-

lutionary dynamics (Alberti 2015 Norberg Urban Vellend

Klausmeier amp Loeuille 2012) Thus a focus on ecological and evolu-

tionary consequences of natural stressors that push organisms to the

brink of their physiological capacities may be instrumental to predict

consequences of anthropogenic environmental change

ACKNOWLEDGEMENTS

We would like to thank Lenin Arias-Rodriguez the Universidad

Juarez Autonoma de Tabasco and the Centro de Investigacion E

Innovacion Para la Ense~nanza y El Aprendizaje for their continued

collaboration enabling our work on the P mexicana system Funding

for this review article has been provided by the National Science

Foundation (IOS-1557860 IOS-1557795) and the US Army

Research Office (W911NF-15-1-0175)

AUTHOR CONTRIBUTIONS

MT wrote the first draft of the paper all authors contributed text

and comments

ORCID

Michael Tobler httporcidorg0000-0002-0326-0890

Joanna L Kelley httporcidorg0000-0002-7731-605X

Reuroudiger Riesch httporcidorg0000-0002-0223-1254

REFERENCES

Alberti M (2015) Eco-evolutionary dynamics in an urbanizing planet

Trends in Ecology amp Evolution 30 114ndash126 httpsdoiorg101016j

tree201411007

Alda F Reina R G Doadrio I amp Bermingham E (2013) Phylogeny

and biogeography of the Poecilia sphenops species complex

(Actinopterygii Poeciliidae) in Central America Molecular Phylogenet-

ics and Evolution 66 1011ndash1026 httpsdoiorg101016jympev

201212012

Alvarez del Villar J (1948) Descripcion de una nueva especie de Mol-

lienisia capturada en Ba~nos del Azufre Tabasco (Pisces Poeciliidae)

Anales de la Escuela Nacional de Ciencias Biologicas 5 275ndash281

Arnold S J (1983) Morphology performance and fitness American Zool-

ogist 23 347ndash361 httpsdoiorg101093icb232347

Aspiras A C Rohner N Martineau B Borowsky R L amp Clifford C J

(2015) Melanocortin 4 receptor mutations contribute to the adapta-

tion of cavefish to nutrient-poor conditions Proceedings of the

National Academy of Sciences USA 112 9968ndash9973

Ast G (2004) How did alternative splicing evolve Nature Reviews

Genetics 5 773ndash782 httpsdoiorg101038nrg1451

Bagarinao T (1992) Sulfide as an environmental factor and toxicant

Tolerance and adaptations in aquatic organisms Aquatic Toxicology

24 21ndash62 httpsdoiorg1010160166-445X(92)90015-F

Barton N H (1998) The effect of hitch-hiking on neutral genealogies

Genetic Research 72 123ndash133 httpsdoiorg101017

S0016672398003462

Barts N Greenway R Passow C N Arias Rodriguez L Kelley J L amp

Tobler M (2018) Expression and molecular evolution of oxygen

transport genes in livebearing fishes (Poeciliidae) from hydrogen sul-

fide rich springs Genome httpsdoiorg101139gen-2017-0051

[Epub ahead of print]

Beauchamp R O Bus J S Popp C S Boreiko C J amp Andjelkovich

D A (1984) A critical review of the literature on hydrogen sulfide

CRC Critical Reviews in Toxicology 13 25ndash97 httpsdoiorg10

310910408448409029321

Bell E M (2012) Life at extremes Environments organisms and strategies

for survival Oxfordshire CABI httpsdoiorg101079

97818459381470000

Bierbach D Arias Rodriguez L amp Plath M (2018) Intrasexual competi-

tion enhances reproductive isolation between locally adapted popula-

tions Current Zoology httpsdoiorg101093czzox071 [Epub

ahead of print]

Bierbach D Klein M Sassmannshausen V Schlupp I Riesch R

Parzefall J amp Plath M (2012) Divergent evolution of male aggres-

sive behavior Another reproductive isolation barrier in extremophile

poeciliid fishes International Journal of Evolutionary Biology 2012

148745

Blackburn D G (1999) Viviparity and oviparity Evolution and reproduc-

tive strategies In T E Knobil amp J D Neill (Eds) Encyclopedia of

reproduction Vol 4 (pp 994ndash1003) New York NY Academic Press

Bouzat J L (2010) Conservation genetics of population bottlenecks

The role of chance selection and history Conservation Genetics 11

463ndash478

Brown A P Greenway R Morgan S Quackenbush C R Giordani

L Arias-Rodriguez L Kelley J L (2017) Genome-scale data

reveals that endemic Poecilia populations from small sulfide spring

display no evidence of inbreeding Molecular Ecology 26 4920ndash

4934

Brown J H Marquet P A amp Taper M L (1993) Evolution of body

size Consequences of an energetic definition of fitness American

Naturalist 142 573ndash584 httpsdoiorg101086285558

Burton R S Pereira R J amp Barreto F S (2013) Cytonuclear genomic

interactions and hybrid breakdown Annual Review of Ecology

TOBLER ET AL | 855

(catastrophic flood event Plath et al 2010) Population genetic dif-

ferentiation between adjacent sulphide-adapted and nonadapted

populations is indicative of ongoing ecological speciation (Box 1) but

different population pairs from sulphidic and nonsulphidic waters

have made variable progress along the speciation continuum towards

complete reproductive isolation (Plath et al 2013) This prompts

questions about the mechanisms involved in reducing gene flow

along the environmental gradient connecting the two habitat types

41 | Prezygotic mechanisms of reproductiveisolation

The simplest form of reproductive isolation between locally adapted

populations arises from selection against migrants between habitat

types (Nosil Vines amp Funk 2005) Translocation experiments in nat-

ural habitats have indicated strong survival selection against

migrants especially those translocated from nonsulphidic to sulphidic

habitats (Plath et al 2013) Fish translocated from nonsulphidic

environments face high mortality within short periods of time

(lt24 hr) which is likely caused by H2S toxicity and hypoxia in sul-

phide springs Hence low gene flow from populations inhabiting

nonsulphidic waters into populations inhabiting sulphidic habitats is

simply explained by a lack of adaptation to the extreme environmen-

tal conditions At least in some drainages there is also appreciable

mortality for sulphide spring fish translocated to nonsulphidic habi-

tats although mechanisms causing inviability are unclear and the

strength of selection appears to be lower (Plath et al 2013) If sur-

vival selection against immigrants was the sole source of reproduc-

tive isolation this should result in unidirectional gene flow from

H2S-adapted to nonadapted populations but population genetic

analyses provided no evidence for this implying that additional

mechanisms are at work (Plath et al 2013)

One such mechanism could be intersexual selection and mate

choice experiments have indicated that migrants from sulphidic to

nonsulphidic habitats are at a significant disadvantage with fish from

nonsulphidic habitats avoiding potential mating partners from sul-

phidic ones (Plath et al 2013 Sommer-Trembo et al 2016) Such

assortative mating has been found to be critical during ecological spe-

ciation because it counteracts the homogenizing effects of gene flow

(van Doorn Edelaar amp Weissing 2009) yet how assortative mating

patterns arise often remains unclear In theory assortative mating can

arise through a variety of processes (Nosil 2012) ldquoMagic traitrdquo mecha-

nisms where mating preferences evolve as a by-product of divergent

selection on ecological traits that serve as cues have recently garnered

much attention (Servedio van Doorn Kopp Frame amp Nosil 2011)

Consistent with magic trait mechanisms assortative mating in P mexi-

cana appears to be linked to adaptive population differences in body

shape which serve as cues to distinguish potential mating partners

from different habitats (Greenway Drexler Arias-Rodriguez amp Tobler

2016) However neither sulphide spring fish (which are unlikely to

encounter migrants from nonsulphidic habitats due to H2S toxicity)

nor allopatric P mexicana from nonsulphidic waters (which have never

experienced migrants from sulphide spring populations) exhibit a

preference for mating partners from their own ecotype suggesting

that assortative mating in this system is a product of reinforcement

(ie direct selection for assortative mating Greenway et al 2016)

This is different from classic magic traits mechanisms because assorta-

tive mating is not a mere by-product of adaptation but adaptive traits

have instead been co-opted as mating signals in response to selection

on mating preferences (Maan amp Seehausen 2012) Moreover males of

some H2S-adapted populations have reduced aggressive behaviour

(Bierbach et al 2012 2018) In staged contests of size-matched males

from different ecotypes intruders had a significantly lower likelihood

to establish dominance than resident males independent of whether

or not they showed reduced aggressiveness overall Hence intrasexual

selection may represent another mechanism mediating reproductive

isolation between locally adapted ecotypes Overall H2S-adapted poe-

ciliids allow disentangling the relative contributions of different forms

of ecological and sexual selectionmdashand their interdependencymdashduring

ecological speciation

42 | Ensuing challenges quantifying postzygoticmechanisms of reproductive isolation

In contrast to prezygotic mechanisms little is known about postzy-

gotic mechanisms of reproductive isolation that potentially act to

reduce gene flow between adjacent populations in sulphidic and

nonsulphidic habitats Putative hybrids between sulphidic and non-

sulphidic ecotypes have occasionally been observed in nonsulphidic

habitats immediately adjacent to confluences with sulphide springs

(R Riesch M Plath and M Tobler personal observations) but it

remains unclear whether and through what mechanisms such hybrids

exhibit reduced fitness compared to parentals

Direct selection of mitochondrially encoded genes in the pres-

ence of H2S (Pfenninger et al 2014) should be accompanied by co-

evolution at nuclear loci that directly interact with mitochondrial

gene products in the formation of protein complexes involved in the

OXPHOS pathway (Sunnucks Morales Lamb Pavlova amp Greening

2017 Willett amp Burton 2004) Disruption of co-evolved gene com-

plexes in hybrids should therefore lead to intrinsic fitness reductions

through mito-nuclear incompatibilities (Burton Pereira amp Barreto

2013 Wolff Ladoukakis Enrıquez amp Dowling 2014) which have

been hypothesized to be a key driver in speciation (Chou amp Leu

2010 Gershoni Templeton amp Mishmar 2009) Empirical tests of

these ideas remain to be conducted in sulphide spring fishes just

like studies on potential ecological and sexual selection against

hybrids that arise from interpopulation mating events

5 | PERSPECTIVES LIFE IN SULPHIDICENVIRONMENTS BEYOND P MEXICANA

Research on sulphide spring fishes of the P mexicana species com-

plex in Mexico has provided insights into mechanisms underlying

adaptation and the links between adaptation and speciation Sul-

phide springs do not only occur in Mexico however but are found

TOBLER ET AL | 853

worldwide and have been colonized by 24 evolutionarily indepen-

dent fish lineages in North America the Neotropics and the Middle

East (Greenway et al 2014) The vast majority of lineages belong to

the order Cyprinodontiformes (95) and livebearing members of the

family Poeciliidae in particular (80) This potentially indicates that

internal gestation of young has facilitated the colonization of

extreme environments because it shields developing embryos from

physiochemical stressors (Blackburn 1999) Among the Poeciliidae

sulphide spring fishes include members of the genera Brachyrhaphis

Gambusia Poeciliopsis Priapichthys Pseudoxiphophorus Xiphophorus

and multiple subgenera of Poecilia (Figure 3 Greenway et al 2014)

spanning over 40 million years of evolution (based on Reznick Fur-

ness Meredith amp Springer 2017)

A handful of studies have started to address molecular and phe-

notypic evolution across a broader taxon sampling of sulphide spring

fishes finding evidence for convergent evolution in morphology

(Riesch et al 2016 Tobler amp Hastings 2011) life history traits

(Riesch Plath et al 2010 Riesch et al 2014) and oxygen transport

genes (Barts et al 2018) In addition ecological speciation in

response to adaptation to H2S-rich environments has also been doc-

umented in sulphide spring poeciliids other than P mexicana (Riesch

et al 2016) These studies have laid a fruitful foundation for the

investigation of mechanisms of adaptation to H2S-rich springs from

genes to phenotypic traits linking micro-evolutionary processes to

macro-evolutionary patterns

The availability of broad-scale evolutionary replication of popula-

tions that have invaded habitats with the same physiochemical stressors

allows for addressing some of the ensuing challenges discussed above

and investigating the factors that give rise to predictable evolutionary

outcomes Convergent evolution is a commonmdashbut not omnipresentmdash

theme in adaptive evolution and multiple factors can determine the

degree of convergence among replicated lineages exposed to the same

sources of selection (Kaeuffer et al 2012 Landry amp Bernatchez 2010

Oke Rolshausen LeBlond amp Hendry 2017 Rosenblum amp Harmon

2011) Thus leveraging replicated lineages that have colonized sulphide

springs to assess molecular mechanisms underlying convergent pheno-

typic changes analyse how environmental heterogeneity among sul-

phide spring habitats causes lineage-specific patterns of trait evolution

and test whether and by what mechanisms adaptation to extreme envi-

ronments is linked to speciation will provide insights into how ecologi-

cal genetic and functional factors contribute to evolutionary

convergence at different levels of organismal organization (Rosenblum

et al 2014) An expanded taxon sampling with population pairs that

span almost the full range of the speciation continuum from panmixia to

complete reproductive isolation (Riesch et al 2016) will also help

addressing questions about temporal aspects of H2S adaptation In par-

ticular comparing younger to older lineages will provide insights about

the order at which genetic and phenotypic modifications accrue during

the adaptation process

Animals have also invaded H2S-rich habitats in other aquatic

ecosystems including cold seeps and deep-sea hydrothermal vents

(Tobler et al 2016) Adaptation to H2S in the P mexicana species com-

plex has involved modification of physiological pathways associated

with H2S toxicity (OXPHOS pathway) and detoxification (sulphide

F IGURE 3 Phylogeny of the family Poeciliidae (adopted from Culumber amp Tobler 2017) Various lineages in the genera Poecilia (subgeneraAcanthophacelus Limia and Mollienesia) Priapichthys Brachyrhaphis Poeciliopsis Xiphophorus Pseudoxiphophorus and Gambusia have colonizedH2S-rich springs in different parts of the Americas (see Greenway et al 2014) Lineages highlighted in yellow represent endemic speciesrestricted to H2S-rich springs while red lineages represent species in which some populations have invaded sulphidic habitats Pictures depict arepresentative of each focal genussubgenus

854 | TOBLER ET AL

quinone oxidoreductase pathway) which have been highly conserved

during the diversification of life on Earth (Saraste 1999 Shahak amp

Hauska 2008) The high replicability of H2S as an environmental factor

that shaped animal evolution across vastly different ecological contexts

consequently offers the opportunity to investigate whether selection

mediated by the presence of H2S has repeatedly modified these con-

served physiological pathways across animal phyla that have diverged

during the Cambrian explosion spanning over 500 million years of evo-

lution Such comparisons would provide unprecedented insights into

the effects of a single physiochemical stressor on evolutionary change

irrespective of ecological or phylogenetic contexts

6 | CONCLUSIONS

H2S-rich environments and their inhabitants represent useful models

for research in organismal biology ecology and evolution illustrating

both opportunities (eg strong and explicit source of selection evolu-

tionary replication) and challenges (eg complexity of environments

and organismal responses) in asking basic question about the origins

of biodiversity Understanding how organisms thrive in apparently

inhospitable environments will provide us with insights into lifersquos

capacities and limitations to adapt to novel environments (Waterman

1999) This is particularly relevant considering that human activities

substantially contribute to the exacerbation of physiochemical stres-

sors at local regional and global scales impacting ecological and evo-

lutionary dynamics (Alberti 2015 Norberg Urban Vellend

Klausmeier amp Loeuille 2012) Thus a focus on ecological and evolu-

tionary consequences of natural stressors that push organisms to the

brink of their physiological capacities may be instrumental to predict

consequences of anthropogenic environmental change

ACKNOWLEDGEMENTS

We would like to thank Lenin Arias-Rodriguez the Universidad

Juarez Autonoma de Tabasco and the Centro de Investigacion E

Innovacion Para la Ense~nanza y El Aprendizaje for their continued

collaboration enabling our work on the P mexicana system Funding

for this review article has been provided by the National Science

Foundation (IOS-1557860 IOS-1557795) and the US Army

Research Office (W911NF-15-1-0175)

AUTHOR CONTRIBUTIONS

MT wrote the first draft of the paper all authors contributed text

and comments

ORCID

Michael Tobler httporcidorg0000-0002-0326-0890

Joanna L Kelley httporcidorg0000-0002-7731-605X

Reuroudiger Riesch httporcidorg0000-0002-0223-1254

REFERENCES

Alberti M (2015) Eco-evolutionary dynamics in an urbanizing planet

Trends in Ecology amp Evolution 30 114ndash126 httpsdoiorg101016j

tree201411007

Alda F Reina R G Doadrio I amp Bermingham E (2013) Phylogeny

and biogeography of the Poecilia sphenops species complex

(Actinopterygii Poeciliidae) in Central America Molecular Phylogenet-

ics and Evolution 66 1011ndash1026 httpsdoiorg101016jympev

201212012

Alvarez del Villar J (1948) Descripcion de una nueva especie de Mol-

lienisia capturada en Ba~nos del Azufre Tabasco (Pisces Poeciliidae)

Anales de la Escuela Nacional de Ciencias Biologicas 5 275ndash281

Arnold S J (1983) Morphology performance and fitness American Zool-

ogist 23 347ndash361 httpsdoiorg101093icb232347

Aspiras A C Rohner N Martineau B Borowsky R L amp Clifford C J

(2015) Melanocortin 4 receptor mutations contribute to the adapta-

tion of cavefish to nutrient-poor conditions Proceedings of the

National Academy of Sciences USA 112 9968ndash9973

Ast G (2004) How did alternative splicing evolve Nature Reviews

Genetics 5 773ndash782 httpsdoiorg101038nrg1451

Bagarinao T (1992) Sulfide as an environmental factor and toxicant

Tolerance and adaptations in aquatic organisms Aquatic Toxicology

24 21ndash62 httpsdoiorg1010160166-445X(92)90015-F

Barton N H (1998) The effect of hitch-hiking on neutral genealogies

Genetic Research 72 123ndash133 httpsdoiorg101017

S0016672398003462

Barts N Greenway R Passow C N Arias Rodriguez L Kelley J L amp

Tobler M (2018) Expression and molecular evolution of oxygen

transport genes in livebearing fishes (Poeciliidae) from hydrogen sul-

fide rich springs Genome httpsdoiorg101139gen-2017-0051

[Epub ahead of print]

Beauchamp R O Bus J S Popp C S Boreiko C J amp Andjelkovich

D A (1984) A critical review of the literature on hydrogen sulfide

CRC Critical Reviews in Toxicology 13 25ndash97 httpsdoiorg10

310910408448409029321

Bell E M (2012) Life at extremes Environments organisms and strategies

for survival Oxfordshire CABI httpsdoiorg101079

97818459381470000

Bierbach D Arias Rodriguez L amp Plath M (2018) Intrasexual competi-

tion enhances reproductive isolation between locally adapted popula-

tions Current Zoology httpsdoiorg101093czzox071 [Epub

ahead of print]

Bierbach D Klein M Sassmannshausen V Schlupp I Riesch R

Parzefall J amp Plath M (2012) Divergent evolution of male aggres-

sive behavior Another reproductive isolation barrier in extremophile

poeciliid fishes International Journal of Evolutionary Biology 2012

148745

Blackburn D G (1999) Viviparity and oviparity Evolution and reproduc-

tive strategies In T E Knobil amp J D Neill (Eds) Encyclopedia of

reproduction Vol 4 (pp 994ndash1003) New York NY Academic Press

Bouzat J L (2010) Conservation genetics of population bottlenecks

The role of chance selection and history Conservation Genetics 11

463ndash478

Brown A P Greenway R Morgan S Quackenbush C R Giordani

L Arias-Rodriguez L Kelley J L (2017) Genome-scale data

reveals that endemic Poecilia populations from small sulfide spring

display no evidence of inbreeding Molecular Ecology 26 4920ndash

4934

Brown J H Marquet P A amp Taper M L (1993) Evolution of body

size Consequences of an energetic definition of fitness American

Naturalist 142 573ndash584 httpsdoiorg101086285558

Burton R S Pereira R J amp Barreto F S (2013) Cytonuclear genomic

interactions and hybrid breakdown Annual Review of Ecology

TOBLER ET AL | 855

worldwide and have been colonized by 24 evolutionarily indepen-

dent fish lineages in North America the Neotropics and the Middle

East (Greenway et al 2014) The vast majority of lineages belong to

the order Cyprinodontiformes (95) and livebearing members of the

family Poeciliidae in particular (80) This potentially indicates that

internal gestation of young has facilitated the colonization of

extreme environments because it shields developing embryos from

physiochemical stressors (Blackburn 1999) Among the Poeciliidae

sulphide spring fishes include members of the genera Brachyrhaphis

Gambusia Poeciliopsis Priapichthys Pseudoxiphophorus Xiphophorus

and multiple subgenera of Poecilia (Figure 3 Greenway et al 2014)

spanning over 40 million years of evolution (based on Reznick Fur-

ness Meredith amp Springer 2017)

A handful of studies have started to address molecular and phe-

notypic evolution across a broader taxon sampling of sulphide spring

fishes finding evidence for convergent evolution in morphology

(Riesch et al 2016 Tobler amp Hastings 2011) life history traits

(Riesch Plath et al 2010 Riesch et al 2014) and oxygen transport

genes (Barts et al 2018) In addition ecological speciation in

response to adaptation to H2S-rich environments has also been doc-

umented in sulphide spring poeciliids other than P mexicana (Riesch

et al 2016) These studies have laid a fruitful foundation for the

investigation of mechanisms of adaptation to H2S-rich springs from

genes to phenotypic traits linking micro-evolutionary processes to

macro-evolutionary patterns

The availability of broad-scale evolutionary replication of popula-

tions that have invaded habitats with the same physiochemical stressors

allows for addressing some of the ensuing challenges discussed above

and investigating the factors that give rise to predictable evolutionary

outcomes Convergent evolution is a commonmdashbut not omnipresentmdash

theme in adaptive evolution and multiple factors can determine the

degree of convergence among replicated lineages exposed to the same

sources of selection (Kaeuffer et al 2012 Landry amp Bernatchez 2010

Oke Rolshausen LeBlond amp Hendry 2017 Rosenblum amp Harmon

2011) Thus leveraging replicated lineages that have colonized sulphide

springs to assess molecular mechanisms underlying convergent pheno-

typic changes analyse how environmental heterogeneity among sul-

phide spring habitats causes lineage-specific patterns of trait evolution

and test whether and by what mechanisms adaptation to extreme envi-

ronments is linked to speciation will provide insights into how ecologi-

cal genetic and functional factors contribute to evolutionary

convergence at different levels of organismal organization (Rosenblum

et al 2014) An expanded taxon sampling with population pairs that

span almost the full range of the speciation continuum from panmixia to

complete reproductive isolation (Riesch et al 2016) will also help

addressing questions about temporal aspects of H2S adaptation In par-

ticular comparing younger to older lineages will provide insights about

the order at which genetic and phenotypic modifications accrue during

the adaptation process

Animals have also invaded H2S-rich habitats in other aquatic

ecosystems including cold seeps and deep-sea hydrothermal vents

(Tobler et al 2016) Adaptation to H2S in the P mexicana species com-

plex has involved modification of physiological pathways associated

with H2S toxicity (OXPHOS pathway) and detoxification (sulphide

F IGURE 3 Phylogeny of the family Poeciliidae (adopted from Culumber amp Tobler 2017) Various lineages in the genera Poecilia (subgeneraAcanthophacelus Limia and Mollienesia) Priapichthys Brachyrhaphis Poeciliopsis Xiphophorus Pseudoxiphophorus and Gambusia have colonizedH2S-rich springs in different parts of the Americas (see Greenway et al 2014) Lineages highlighted in yellow represent endemic speciesrestricted to H2S-rich springs while red lineages represent species in which some populations have invaded sulphidic habitats Pictures depict arepresentative of each focal genussubgenus

854 | TOBLER ET AL

quinone oxidoreductase pathway) which have been highly conserved

during the diversification of life on Earth (Saraste 1999 Shahak amp

Hauska 2008) The high replicability of H2S as an environmental factor

that shaped animal evolution across vastly different ecological contexts

consequently offers the opportunity to investigate whether selection

mediated by the presence of H2S has repeatedly modified these con-

served physiological pathways across animal phyla that have diverged

during the Cambrian explosion spanning over 500 million years of evo-

lution Such comparisons would provide unprecedented insights into

the effects of a single physiochemical stressor on evolutionary change

irrespective of ecological or phylogenetic contexts

6 | CONCLUSIONS

H2S-rich environments and their inhabitants represent useful models

for research in organismal biology ecology and evolution illustrating

both opportunities (eg strong and explicit source of selection evolu-

tionary replication) and challenges (eg complexity of environments

and organismal responses) in asking basic question about the origins

of biodiversity Understanding how organisms thrive in apparently

inhospitable environments will provide us with insights into lifersquos

capacities and limitations to adapt to novel environments (Waterman

1999) This is particularly relevant considering that human activities

substantially contribute to the exacerbation of physiochemical stres-

sors at local regional and global scales impacting ecological and evo-

lutionary dynamics (Alberti 2015 Norberg Urban Vellend

Klausmeier amp Loeuille 2012) Thus a focus on ecological and evolu-

tionary consequences of natural stressors that push organisms to the

brink of their physiological capacities may be instrumental to predict

consequences of anthropogenic environmental change

ACKNOWLEDGEMENTS

We would like to thank Lenin Arias-Rodriguez the Universidad

Juarez Autonoma de Tabasco and the Centro de Investigacion E

Innovacion Para la Ense~nanza y El Aprendizaje for their continued

collaboration enabling our work on the P mexicana system Funding

for this review article has been provided by the National Science

Foundation (IOS-1557860 IOS-1557795) and the US Army

Research Office (W911NF-15-1-0175)

AUTHOR CONTRIBUTIONS

MT wrote the first draft of the paper all authors contributed text

and comments

ORCID

Michael Tobler httporcidorg0000-0002-0326-0890

Joanna L Kelley httporcidorg0000-0002-7731-605X

Reuroudiger Riesch httporcidorg0000-0002-0223-1254

REFERENCES

Alberti M (2015) Eco-evolutionary dynamics in an urbanizing planet

Trends in Ecology amp Evolution 30 114ndash126 httpsdoiorg101016j

tree201411007

Alda F Reina R G Doadrio I amp Bermingham E (2013) Phylogeny

and biogeography of the Poecilia sphenops species complex

(Actinopterygii Poeciliidae) in Central America Molecular Phylogenet-

ics and Evolution 66 1011ndash1026 httpsdoiorg101016jympev

201212012

Alvarez del Villar J (1948) Descripcion de una nueva especie de Mol-

lienisia capturada en Ba~nos del Azufre Tabasco (Pisces Poeciliidae)

Anales de la Escuela Nacional de Ciencias Biologicas 5 275ndash281

Arnold S J (1983) Morphology performance and fitness American Zool-

ogist 23 347ndash361 httpsdoiorg101093icb232347

Aspiras A C Rohner N Martineau B Borowsky R L amp Clifford C J

(2015) Melanocortin 4 receptor mutations contribute to the adapta-

tion of cavefish to nutrient-poor conditions Proceedings of the

National Academy of Sciences USA 112 9968ndash9973

Ast G (2004) How did alternative splicing evolve Nature Reviews

Genetics 5 773ndash782 httpsdoiorg101038nrg1451

Bagarinao T (1992) Sulfide as an environmental factor and toxicant

Tolerance and adaptations in aquatic organisms Aquatic Toxicology

24 21ndash62 httpsdoiorg1010160166-445X(92)90015-F

Barton N H (1998) The effect of hitch-hiking on neutral genealogies

Genetic Research 72 123ndash133 httpsdoiorg101017

S0016672398003462

Barts N Greenway R Passow C N Arias Rodriguez L Kelley J L amp

Tobler M (2018) Expression and molecular evolution of oxygen

transport genes in livebearing fishes (Poeciliidae) from hydrogen sul-

fide rich springs Genome httpsdoiorg101139gen-2017-0051

[Epub ahead of print]

Beauchamp R O Bus J S Popp C S Boreiko C J amp Andjelkovich

D A (1984) A critical review of the literature on hydrogen sulfide

CRC Critical Reviews in Toxicology 13 25ndash97 httpsdoiorg10

310910408448409029321

Bell E M (2012) Life at extremes Environments organisms and strategies

for survival Oxfordshire CABI httpsdoiorg101079

97818459381470000

Bierbach D Arias Rodriguez L amp Plath M (2018) Intrasexual competi-

tion enhances reproductive isolation between locally adapted popula-

tions Current Zoology httpsdoiorg101093czzox071 [Epub

ahead of print]

Bierbach D Klein M Sassmannshausen V Schlupp I Riesch R

Parzefall J amp Plath M (2012) Divergent evolution of male aggres-

sive behavior Another reproductive isolation barrier in extremophile

poeciliid fishes International Journal of Evolutionary Biology 2012

148745

Blackburn D G (1999) Viviparity and oviparity Evolution and reproduc-

tive strategies In T E Knobil amp J D Neill (Eds) Encyclopedia of

reproduction Vol 4 (pp 994ndash1003) New York NY Academic Press

Bouzat J L (2010) Conservation genetics of population bottlenecks

The role of chance selection and history Conservation Genetics 11

463ndash478

Brown A P Greenway R Morgan S Quackenbush C R Giordani

L Arias-Rodriguez L Kelley J L (2017) Genome-scale data

reveals that endemic Poecilia populations from small sulfide spring

display no evidence of inbreeding Molecular Ecology 26 4920ndash

4934

Brown J H Marquet P A amp Taper M L (1993) Evolution of body

size Consequences of an energetic definition of fitness American

Naturalist 142 573ndash584 httpsdoiorg101086285558

Burton R S Pereira R J amp Barreto F S (2013) Cytonuclear genomic

interactions and hybrid breakdown Annual Review of Ecology

TOBLER ET AL | 855

quinone oxidoreductase pathway) which have been highly conserved

during the diversification of life on Earth (Saraste 1999 Shahak amp

Hauska 2008) The high replicability of H2S as an environmental factor

that shaped animal evolution across vastly different ecological contexts

consequently offers the opportunity to investigate whether selection

mediated by the presence of H2S has repeatedly modified these con-

served physiological pathways across animal phyla that have diverged

during the Cambrian explosion spanning over 500 million years of evo-

lution Such comparisons would provide unprecedented insights into

the effects of a single physiochemical stressor on evolutionary change

irrespective of ecological or phylogenetic contexts

6 | CONCLUSIONS

H2S-rich environments and their inhabitants represent useful models

for research in organismal biology ecology and evolution illustrating

both opportunities (eg strong and explicit source of selection evolu-

tionary replication) and challenges (eg complexity of environments

and organismal responses) in asking basic question about the origins

of biodiversity Understanding how organisms thrive in apparently

inhospitable environments will provide us with insights into lifersquos

capacities and limitations to adapt to novel environments (Waterman

1999) This is particularly relevant considering that human activities

substantially contribute to the exacerbation of physiochemical stres-

sors at local regional and global scales impacting ecological and evo-

lutionary dynamics (Alberti 2015 Norberg Urban Vellend

Klausmeier amp Loeuille 2012) Thus a focus on ecological and evolu-

tionary consequences of natural stressors that push organisms to the

brink of their physiological capacities may be instrumental to predict

consequences of anthropogenic environmental change

ACKNOWLEDGEMENTS

We would like to thank Lenin Arias-Rodriguez the Universidad

Juarez Autonoma de Tabasco and the Centro de Investigacion E

Innovacion Para la Ense~nanza y El Aprendizaje for their continued

collaboration enabling our work on the P mexicana system Funding

for this review article has been provided by the National Science

Foundation (IOS-1557860 IOS-1557795) and the US Army

Research Office (W911NF-15-1-0175)

AUTHOR CONTRIBUTIONS

MT wrote the first draft of the paper all authors contributed text

and comments

ORCID

Michael Tobler httporcidorg0000-0002-0326-0890

Joanna L Kelley httporcidorg0000-0002-7731-605X

Reuroudiger Riesch httporcidorg0000-0002-0223-1254

REFERENCES

Alberti M (2015) Eco-evolutionary dynamics in an urbanizing planet

Trends in Ecology amp Evolution 30 114ndash126 httpsdoiorg101016j

tree201411007

Alda F Reina R G Doadrio I amp Bermingham E (2013) Phylogeny

and biogeography of the Poecilia sphenops species complex

(Actinopterygii Poeciliidae) in Central America Molecular Phylogenet-

ics and Evolution 66 1011ndash1026 httpsdoiorg101016jympev

201212012

Alvarez del Villar J (1948) Descripcion de una nueva especie de Mol-

lienisia capturada en Ba~nos del Azufre Tabasco (Pisces Poeciliidae)

Anales de la Escuela Nacional de Ciencias Biologicas 5 275ndash281

Arnold S J (1983) Morphology performance and fitness American Zool-

ogist 23 347ndash361 httpsdoiorg101093icb232347

Aspiras A C Rohner N Martineau B Borowsky R L amp Clifford C J

(2015) Melanocortin 4 receptor mutations contribute to the adapta-

tion of cavefish to nutrient-poor conditions Proceedings of the

National Academy of Sciences USA 112 9968ndash9973

Ast G (2004) How did alternative splicing evolve Nature Reviews

Genetics 5 773ndash782 httpsdoiorg101038nrg1451

Bagarinao T (1992) Sulfide as an environmental factor and toxicant

Tolerance and adaptations in aquatic organisms Aquatic Toxicology

24 21ndash62 httpsdoiorg1010160166-445X(92)90015-F

Barton N H (1998) The effect of hitch-hiking on neutral genealogies

Genetic Research 72 123ndash133 httpsdoiorg101017

S0016672398003462

Barts N Greenway R Passow C N Arias Rodriguez L Kelley J L amp

Tobler M (2018) Expression and molecular evolution of oxygen

transport genes in livebearing fishes (Poeciliidae) from hydrogen sul-

fide rich springs Genome httpsdoiorg101139gen-2017-0051

[Epub ahead of print]

Beauchamp R O Bus J S Popp C S Boreiko C J amp Andjelkovich

D A (1984) A critical review of the literature on hydrogen sulfide

CRC Critical Reviews in Toxicology 13 25ndash97 httpsdoiorg10

310910408448409029321

Bell E M (2012) Life at extremes Environments organisms and strategies

for survival Oxfordshire CABI httpsdoiorg101079

97818459381470000

Bierbach D Arias Rodriguez L amp Plath M (2018) Intrasexual competi-

tion enhances reproductive isolation between locally adapted popula-

tions Current Zoology httpsdoiorg101093czzox071 [Epub

ahead of print]

Bierbach D Klein M Sassmannshausen V Schlupp I Riesch R

Parzefall J amp Plath M (2012) Divergent evolution of male aggres-

sive behavior Another reproductive isolation barrier in extremophile

poeciliid fishes International Journal of Evolutionary Biology 2012

148745

Blackburn D G (1999) Viviparity and oviparity Evolution and reproduc-

tive strategies In T E Knobil amp J D Neill (Eds) Encyclopedia of

reproduction Vol 4 (pp 994ndash1003) New York NY Academic Press

Bouzat J L (2010) Conservation genetics of population bottlenecks

The role of chance selection and history Conservation Genetics 11

463ndash478

Brown A P Greenway R Morgan S Quackenbush C R Giordani

L Arias-Rodriguez L Kelley J L (2017) Genome-scale data

reveals that endemic Poecilia populations from small sulfide spring

display no evidence of inbreeding Molecular Ecology 26 4920ndash

4934

Brown J H Marquet P A amp Taper M L (1993) Evolution of body

size Consequences of an energetic definition of fitness American

Naturalist 142 573ndash584 httpsdoiorg101086285558

Burton R S Pereira R J amp Barreto F S (2013) Cytonuclear genomic

interactions and hybrid breakdown Annual Review of Ecology

TOBLER ET AL | 855

Evolution and Systematics 44 281ndash302 httpsdoiorg101146an

nurev-ecolsys-110512-135758

Bush S J Chen L Tovar-Corona J M amp Urritia A O (2017) Alterna-

tive splicing and the evolution of phenotypic novelty Philosophical

Transactions of the Royal Society B 372 20150474 httpsdoiorg

101098rstb20150474

Carroll S P Jorgensen P S Kinnison M T Bergstrom C T Denison

R F Gluckman P Tabashnik B E (2014) Applying evolutionary

biology to address global challenges Science 346 313 httpsdoi

org101364FIO2014JW3A13

Cartwright R A Schwartz R S Merry A L amp Howell M M (2017)

The importance of selection in the evolution of blindness in cavefish

Bmc Evolutionary Biology 17 45 httpsdoiorg101186s12862-

017-0876-4

Chan Y F Marks M E Jones F C Villarreal G Shapiro M D Brady

S D Kingsley D M (2010) Adaptive evolution of pelvic reduc-

tion in sticklebacks by recurrent deletion of Pitx1 enhancer Science

327 302ndash305 httpsdoiorg101126science1182213

Chou J Y amp Leu J Y (2010) Speciation through cytonuclear incompat-

ibility Insights from yeast and implications for higher eukaryotes

BioEssays 32 401ndash411 httpsdoiorg101002bies200900162

Colosimo P F Hosemann K E Balabhadra S Villarreal G Dickson

M Grimwood J Kingsley D M (2005) Widespread parallel evo-

lution in sticklebacks by repeated fixation of ectodysplasin alleles

Science 307 1928ndash1933 httpsdoiorg101126science1107239

Cooper C E amp Brown G C (2008) The inhibition of mitochondrial

cytochrome oxidase by the gases carbon monoxide nitric oxide

hydrogen cyanide and hydrogen sulfide Chemical mechanism and

physiological significance Journal of Bioenergy and Biomembranes 40

533ndash539 httpsdoiorg101007s10863-008-9166-6

Culumber Z W Hopper G W Barts N Passow C N Morgan S Brown

A Tobler M (2016) Habitat use of two extremophile highly ende-

mic and critically endangered fish species (Gambusia eurystoma and Poe-

cilia sulphurophila Poeciliidae) Aquatic Conservation-Marine and

Freshwater Ecosystems 26 1155ndash1167 httpsdoiorg101002aqc

2640

Culumber Z W amp Tobler M (2017) Sex-specific evolution during the

diversification of live-bearing fishes Nature Ecology amp Evolution 1

1185ndash1191

Dalziel A C Rogers S M amp Schulte P M (2009) Linking genotypes

to phenotypes and fitness How mechanistic biology can inform

molecular ecology Molecular Ecology 18 4997ndash5017 httpsdoi

org101111j1365-294X200904427x

Darwin C (1859) On the origin of species based on natural selection or

the preservation of favoured races in the struggle of life 6th ed Lon-

don John Murray

Dasmahapatra K K Walters J R Briscoe A D Davey J W Whibley

A Nadeau N J Jiggins C D (2012) Butterfly genome reveals

promiscuous exchange of mimicry adaptations among species Nature

487 94ndash98 httpsdoiorg101038nature11041

Doudna J A amp Charpentier E (2014) The new frontier of genome edit-

ing with CRISPR-Cas9 Science 346 1258096 httpsdoiorg10

1126science1258096

Dubilier N Giere O amp Grieshaber M (2005) Concomitant effects of sul-

fide and hypoxia on the aerobic metabolism of the marine oligochaete

Tubificoides benedii Journal of Experimental Zoology 269 287ndash298

Eales K L Hollinshead K E R amp Tennant D A (2016) Hypoxia and

metabolic adaptation of cancer cells Oncogenesis 5 e190 httpsd

oiorg101038oncsis201550

Eghbal M A Pennefather P S amp OrsquoBrien P J (2004) H2S cytotoxicity

mechanism involves reactive oxygen species formation and mito-

chondrial depolarisation Toxicology 203 69ndash76 httpsdoiorg10

1016jtox200405020

Feldman C R Brodie Jr E D Brodie III E D amp Pfrender M E (2012)

Constraint shapes convergence in tetrodotoxin-resistant sodium

channels of snakes Proceedings of the National Academy of Sciences

USA 109 4556ndash4561 httpsdoiorg101073pnas1113468109

Flores K B Wolschin F amp Amdam G V (2013) The role of methyla-

tion of DNA in environmental adaptation Integrative and Comparative

Biology 53 359ndash372 httpsdoiorg101093icbict019

Fradin H Kiontke K Zegar C Gutwein M Lucas J Kovtun M

Piano F (2017) Genome architecture and evolution of a unichromo-

somal asexual nematode Current Biology 27(2928ndash2939) e2926

Gershoni M Templeton A amp Mishmar D (2009) Mitochondrial bioen-

ergetics as a major motive force of speciation BioEssays 31 642ndash

650 httpsdoiorg101002bies200800139

Ghalambor C K Walker J A amp Reznick D N (2003) Multi-trait selec-

tion adaptation and constraints on the evolution of burst swimming

performance Integrative and Comparative Biology 43 431ndash438

httpsdoiorg101093icb433431

Gibert J amp Deharveng L (2002) Subterranean ecosystems A truncated

functional biodiversity BioScience 52 473ndash481 httpsdoiorg10

16410006-3568(2002)052[0473SEATFB]20CO2

Gjuvsland A B Vik J O Beard D A Hunter P J amp Omholt S W

(2013) Bridging the genotype-phenotype gap What does it take

Journal of Physiology 591 2055ndash2066 httpsdoiorg101113jphys

iol2012248864

Grant P R Grant B R Markert J A Keller L F amp Petren K (2004)

Convergent evolution of Darwinrsquos finches caused by introgressive

hybridization and selection Evolution 58 1588ndash1599 httpsdoi

org101111j0014-38202004tb01738x

Greene H W (2005) Organisms in nature as a central focus for biology

Trends in Ecology and Evolution 20 23ndash27 httpsdoiorg101016j

tree200411005

Greenway R Arias-Rodriguez L Diaz P amp Tobler M (2014) Patterns

of macroinvertebrate and fish diversity in freshwater sulphide springs

Diversity 6 597ndash632 httpsdoiorg103390d6030597

Greenway R Drexler S Arias-Rodriguez L amp Tobler M (2016) Adap-

tive but not condition-dependent body shape differences contribute

to assortative mating preferences during ecological speciation Evolu-

tion 70 2809ndash2822 httpsdoiorg101111evo13087

Gunderson A R Armstrong E J amp Stillman J H (2016) Multiple stres-

sors in a changing world The need for an improved perspective on

physiological responses to the dynamic marine environment Annual

Review of Marine Science 8 357ndash378 httpsdoiorg101146an

nurev-marine-122414-033953

Hildebrandt T M amp Grieshaber M (2008) Three enzymatic activities

catalyze the oxidation of sulfide to thiosulfate in mammalian and

invertebrate mitochondria FEBS Journal 275 3352ndash3361 httpsd

oiorg101111j1742-4658200806482x

Houle D Govindaraju D R amp Omholt S (2010) Phenomics The next

challenge Nature Reviews Genetics 11 855ndash866 httpsdoiorg10

1038nrg2897

IUCN (2012) Red List of threatened species version 20132 www

iucnredlistorg

Johnston I A amp Bennett A F (1996) Animals and temperature Pheno-

typic and evolutionary adaptation Cambridge Cambridge University

Press httpsdoiorg101017CBO9780511721854

Jorgensen C Auer S K amp Reznick D N (2011) A model for optimal

offspring size in fish including live-bearing and parental effects Amer-

ican Naturalist 177 E119ndashE135 httpsdoiorg101086659622

Kaeuffer R Peichel C L Bolnick D I amp Hendry A P (2012) Parallel

and nonparallel aspects of ecological phenotypic and genetic diver-

gence across replicate population pairs of lake and stream stickle-

back Evolution 66 402ndash418 httpsdoiorg101111j1558-5646

201101440x

Kell D B amp Oliver S G (2004) Here is the evidence now what is the

hypothesis The complementary roles of inductive and hypothesis-

driven science in the post-genomic era BioEssays 26 99ndash105

httpsdoiorg101002(ISSN)1521-1878

856 | TOBLER ET AL

Kelley J L Arias-Rodriguez L Patacsil Martin D Yee M C Busta-

mante C amp Tobler M (2016) Mechanisms underlying adaptation to

life in hydrogen sulfide rich environments Molecular Biology and Evo-

lution 33 1419ndash1434 httpsdoiorg101093molbevmsw020

Kelley J L Passow C Plath M Arias-Rodriguez L Yee M C amp

Tobler M (2012) Genomic resources for a model in adaptation and

speciation research The characterization of the Poecilia mexicana

transcriptome BMC Genomics 13 652 httpsdoiorg101186

1471-2164-13-652

Klerks P L Xie L amp Levinton J S (2011) Quantitative genetics

approaches to study evolutionary processes in ecotoxicology a per-

spective from research on the evolution of resistance Ecotoxicology

20 513ndash523 httpsdoiorg101007s10646-011-0640-2

Kump L R Pavlov A amp Arthur M A (2005) Massive release of hydrogen

sulfide to the surface ocean and atmosphere during intervals of oceanic

anoxia Geology 33 397ndash400 httpsdoiorg101130G212951

Keuroustner A Hoffmann M Fraser B A Kottler V A Sharma E Wei-

gel D amp Dreyer C (2016) The genome of the Trinidadian guppy

Poecilia reticulata and variation in the guanapo population PLoS ONE

11 e0169087 httpsdoiorg101371journalpone0169087

Landry L amp Bernatchez L (2010) Role of epibenthic resource opportu-

nities in the parallel evolution of lake whitefish species pairs (Core-

gonus sp) Journal of Evolutionary Biology 23 2602ndash2613 httpsdoi

org101111j1420-9101201002121x

Langerhans R B (2007) Evolutionary consequences of predation Avoid-

ance escape reproduction and diversification In A M T Elewa

(Ed) Predation in organisms Berlin Springer

Li L Rose P amp Moore P K (2011) Hydrogen sulfide and cell signaling

Annual Review of Pharmacology and Toxicology 51 169ndash187

httpsdoiorg101146annurev-pharmtox-010510-100505

Lin Q Fan S Zhang Y Xu M Zhang H Yang Y Gunter H M

(2016) The seahorse genome and the evolution of its specialized mor-

phology Nature 540 395ndash399 httpsdoiorg101038nature20595

Liu Y Ma S Wang X Chang J Gao J Shi R Zhao P (2014)

Highly efficient multiplex targeted mutagenesis and genomic struc-

ture variation in Bombyx mori cells using CRISPRCas9 Insect bio-

chemistry and molecular biology 49 35ndash42 httpsdoiorg101016j

ibmb201403010

Lopez-Maury L Marguerat S amp Beuroahler J (2008) Tuning gene expres-

sion to changing environments From rapid responses to evolutionary

adaptation Nature Reviews Genetics 9 583ndash593 httpsdoiorg10

1038nrg2398

Maan M E amp Seehausen O (2012) Magic traits versus magic prefer-

ences in speciation Evolutionary Ecology Research 14 779ndash785

Martyn U Weigel D amp Dreyer C (2006) In vitro culture of embryos

of the guppy Poecilia reticulata Developmental Dynamics 235 617ndash

622 httpsdoiorg101002(ISSN)1097-0177

Metzker M L (2010) Sequencing technologies The next generation

Nature Reviews Genetics 11 31ndash46 httpsdoiorg101038nrg2626

Moore M P Riesch R amp Martin R A (2016) The predictability and

magnitude of life-history divergence to ecological agents of selection

A meta-analysis in livebearing fishes Ecology Letters 19 435ndash442

httpsdoiorg101111ele12576

Mustafa A K Gadalla M M Sen N Kim S Mu W Gazi S K

Snyder S H (2009) H2S signals through protein S-sulfhydration

Science Signaling 2 ra72

Nesse R M amp Stearns S C (2008) The great opportunity Evolutionary

applications to medicine and public health Evolutionary Applications

1 28ndash48 httpsdoiorg101111j1752-4571200700006x

Nevo E (2011) Evolution under environmental stress at macro and

microscales Genome Biology and Evolution 3 1039ndash1052 httpsdoi

org101093gbeevr052

Norberg J Urban M C Vellend M Klausmeier C A amp Loeuille N (2012)

Eco-evolutionary responses of biodiversity to climate change Nature Cli-

mate Change 2 747ndash751 httpsdoiorg101038nclimate1588

Nosil P (2012) Ecological speciation Oxford UK Oxford University

Press httpsdoiorg101093acprofosobl9780199587100001

0001

Nosil P Vines T H amp Funk D J (2005) Perspective Reproductive

isolation caused by natural selection against immigrants from diver-

gent habitats Evolution 59 705ndash719

Oke K B Rolshausen G LeBlond C amp Hendry A P (2017) How par-

allel is parallel evolution A comparative analysis in fishes American

Naturalist 190 1ndash16 httpsdoiorg101086691989

Olson K R (2011) The therapeutic potential of hydrogen sulfide

Separating hype from hope American Journal of Physiology-Regulatory

Integrative and Comparative Physiology 301 R297ndashR312 httpsdoi

org101152ajpregu000452011

Olson K R Donald J A Dombkowski R A amp Perry S F (2012) Evo-

lutionary and comparative aspects of nitric oxide carbon monoxide

and hydrogen sulfide Respiratory Physiology amp Neurobiology 184

117ndash129 httpsdoiorg101016jresp201204004

Olson K R amp Straub K D (2015) The role of hydrogen sulfide in evo-

lution and the evolution of hydrogen sulfide in metabolism and sig-

naling Physiology 31 60ndash72

Palacios M Arias-Rodriguez L Plath M Eifert C Lerp H Lamboj A

Tobler M (2013) The rediscovery of a long described species

reveals additional complexity in speciation patterns of poeciliid fishes

in sulfide springs PLoS ONE 8 e71069 httpsdoiorg101371jour

nalpone0071069

Palacios M Voelker G Arias-Rodriguez L Mateos M amp Tobler M

(2016) Phylogenetic analyses of the subgenus Mollienesia (Poecilia

Poeciliidae Teleostei) reveal taxonomic inconsistencies cryptic biodi-

versity and spatio-temporal aspects of diversification in Middle

America Molecular Phylogenetics and Evolution 103 230ndash244

httpsdoiorg101016jympev201607025

Passow C Arias-Rodriguez L amp Tobler M (2017) Convergent evolu-

tion of reduced energy demands in extremophile fish PLoS ONE 12

e0186935 httpsdoiorg101371journalpone0186935

Passow C N Brown A Arias-Rodriguez L Yee M-C Sockell A

Schartl M Tobler M (2017) Complexities of gene expression

patterns in natural populations of an extremophile fish (Poecilia mexi-

cana Poeciliidae) Molecular Ecology 26 4211ndash4225 httpsdoiorg

101111mec14198

Passow C Greenway R Arias-Rodriguez L Jeyasingh P D amp Tobler

M (2015) Reduction of energetic demands through modification of

body size and metabolic rates in locally adapted extremophile fishes

Physiological and Biochemical Zoology 88 371ndash383 httpsdoiorg

101086681053

Passow C N Henpita C Shaw J H Quakenbush C Warren W C

Schartl M Tobler M (2017) The roles of plasticity and evolu-

tionary change in shaping gene expression variation in natural popu-

lations of extremophile fish Molecular Ecology 26 6384ndash6399

Pavey S A Bernatchez L Aubin-Horth N amp Landry C R (2012)

What is needed for next-generation ecological and evolutionary

genomics Trends in Ecology amp Evolution 27 673ndash678 httpsdoi

org101016jtree201207014

Peers C Bauer C Boyle J P Scragg J L amp Dallas M L (2012)

Modulation of ion channels by hydrogen sulfide Antioxidants amp Redox

Signaling 17 95ndash105 httpsdoiorg101089ars20114359

Pfennig D W amp Pfennig K S (2012) Evolutionrsquos wedge competition and

the origins of diversity Berkeley CA University of California Press

httpsdoiorg101525california97805202741810010001

Pfenninger M Lerp H Tobler M Passow C Kelley J L Funke E

Plath M (2014) Parallel evolution of cox genes in H2S-tolerant

fish as key adaptation to a toxic environment Nature Communica-

tions 5 3873

Pfenninger M Patel S Arias-Rodriguez L Feldmeyer B Riesch R amp

Plath M (2015) Unique evolutionary trajectories in repeated adapta-

tion to hydrogen sulphide-toxic habitats of a neotropical fish (Poecilia

TOBLER ET AL | 857

mexicana) Molecular Ecology 24 5446ndash5459 httpsdoiorg10

1111mec13397

Pietri R Roman-Morales E amp Lopez-Garriga J (2011) Hydrogen sul-

fide and heme proteins Knowledge and mysteries Antioxidants amp

Redox Signaling 15 393ndash404 httpsdoiorg101089ars20103698

Plath M (2008) Male mating behavior and costs of sexual harassment

for females in cavernicolous and extremophile populations of Atlantic

mollies (Poecilia mexicana) Behaviour 145 73ndash98 httpsdoiorg10

1163156853908782687241

Plath M Hermann C Schreurooder R Riesch R Tobler M Garcia de

Leon F J Tiedemann R (2010) Locally adapted fish populations

maintain small-scale genetic differentiation despite perturbation by a

catastrophic flood event BMC Evolutionary Biology 10 256 httpsd

oiorg1011861471-2148-10-256

Plath M Pfenninger M Lerp H Riesch R Eschenbrenner C Slattery

P Tobler M (2013) Genetic differentiation and selection against

migrants in evolutionarily replicated extreme environments Evolution

67 2647ndash2661 httpsdoiorg101111evo12133

Plath M Tobler M Riesch R Garcia de Leon F J Giere O amp

Schlupp I (2007) Survival in an extreme habitat The role of beha-

viour and energy limitation Naturwissenschaften 94 991ndash996

httpsdoiorg101007s00114-007-0279-2

Powell E (1989) Oxygen sulfide and diffusion Why thiobiotic meiofauna

must be sulfide-insensitive first-order respirers Journal of Marine

Research 47 887ndash932 httpsdoiorg101357002224089785076082

Ramirez-Soriano A Ramos-Onsins S E Rozas J Calafell F amp Navarro A

(2008) Statistical power analysis of neutrality tests under demographic

expansions contractions and bottlenecks with recombination Genetics

179 555ndash567 httpsdoiorg101534genetics107083006

Reiffenstein R Hulbert W amp Roth S (1992) Toxicology of hydrogen

sulfide Annual Reviews of Pharmacology and Toxicology 1992 109ndash

134 httpsdoiorg101146annurevpa32040192000545

Reznick D N Furness A I Meredith R W amp Springer M S (2017)

The origin and biogeographic diversification of fishes in the family

Poeciliidae PLoS ONE 12 e0172546 httpsdoiorg101371journa

lpone0172546

Reznick D amp Yang A P (1993) The influence of fluctuating resources

on life history Patterns of allocation and plasticity in female guppies

Ecology 74 2011ndash2019 httpsdoiorg1023071940844

Riesch R Duwe V Herrmann N Padur L Ramm A Scharnweber K

Plath M (2009) Variation along the shy-bold continuum in extre-

mophile fishes (Poecilia mexicana Poecilia sulphuraria) Behavioral Ecol-

ogy and Sociobiology 63 1515ndash1526 httpsdoiorg101007

s00265-009-0780-z

Riesch R Oranth A Dzienko J Karau N Schiessl A Stadler S

Plath M (2010) Extreme habitats are not refuges Poeciliids suffer

from increased aerial predation risk in sulphidic southern Mexican

habitats Biological Journal of the Linnean Society 101 417ndash426

httpsdoiorg101111j1095-8312201001522x

Riesch R Plath M Garcia de Leon F J amp Schlupp I (2010) Conver-

gent life-history shifts Toxic environments result in big babies in two

clades of poeciliids Naturwissenschaften 97 133ndash141 httpsdoi

org101007s00114-009-0613-y

Riesch R Plath M Schlupp I Tobler M amp Langerhans R B (2014)

Colonization of toxic springs drives predictable life-history shift in

livebearing fishes (Poeciliidae) Ecology Letters 17 65ndash71 httpsdoi

org101111ele12209

Riesch R Tobler M Lerp H Jourdan J Doumas L T Nosil P Plath

M (2016) Extremophile Poeciliidae Multivariate insights into the com-

plexity of speciation along replicated ecological gradients BMC Evolu-

tionary Biology 16 136 httpsdoiorg101186s12862-016-0705-1

Riesch R Tobler M amp Plath M (2015) Extremophile fishes Ecology

evolution and physiology of teleosts in extreme environments p 290

Heidelberg Germany Springer httpsdoiorg101007978-3-319-

13362-1

Ritchie M E Holzinger E R Li R Pendergrass S A amp Kim S (2015)

Methods of integrating data to uncover genotype-phenotype interac-

tions Nature Reviews Genetics 16 85ndash97 httpsdoiorg101038

nrg3868

Roach K Tobler M amp Winemiller K O (2011) Hydrogen sulfide bac-

teria and fish A unique subterranean food chain Ecology 92 2056ndash

2062 httpsdoiorg10189011-02761

Rohner N Jarosz D F Kowalko J E Yoshizawa M Jeffery W R

Borowsky R L Tabin C J (2013) Cryptic variation in morpho-

logical evolution HSP90 as a capacitor for loss of eyes in cavefish

Science 342 1372ndash1375 httpsdoiorg101126science1240276

Rosales Lagarde L (2012) Investigation of karst brackish-sulfidic springs

and their role in the hydrogeology subsurface water-rock interactions

and speleogenesis at northern Sierra de Chiapas New Mexico Institute

of Mining and Technology Mexico

Rosenblum E B amp Harmon L J (2011) ldquoSame but differentrdquo Repli-

cated ecological speciation at White Sands Evolution 65 946ndash960

httpsdoiorg101111j1558-5646201001190x

Rosenblum E B Parent C E amp Brandt E E (2014) The molecular

basis of convergent evolution Annual Review of Ecology Evolution and

Systematics 45 203ndash226 httpsdoiorg101146annurev-ecolsys-

120213-091851

Saltz J B Hessel F C amp Kelly M W (2017) Trait correlations in the

genomics era Trends in Ecology amp Evolution 32 279ndash290 httpsdoi

org101016jtree201612008

Saraste M (1999) Oxidative phosphorylation at the fin de siecle Science

283 1488ndash1493 httpsdoiorg101126science28354071488

Schartl M (2014) Beyond the zebrafish Diverse fish species for model-

ing human disease Disease Models and Mechanisms 7 181ndash192

httpsdoiorg101242dmm012245

Schartl M Walter R B Shen Y Garcia T Catchen J Amores A

Warren W C (2013) The genome of the platyfish Xiphophorus mac-

ulatus provides insights into evolutionary adaptation and several

complex traits Nature Genetics 45 567ndash572 httpsdoiorg10

1038ng2604

Schluter D (2009) Evidence for ecological speciation and its alternative

Science 323 737ndash741 httpsdoiorg101126science1160006

Schulz-Mirbach T Eifert C Riesch R Farnworth M S Zimmer C

Bierbach D Plath M (2016) Toxic hydrogen sulphide shapes

brain anatomy A comparative study of sulphide-adapted ecotypes in

the Poecilia mexicana complex Journal of Zoology 300 163ndash176

httpsdoiorg101111jzo12366

SEDESOL (1994) NORMA Oficial Mexicana NOM-059-ECOL-1994 que

determina las especies y subspecies de flora y fauna silvestres ter-

restres y acuaticas en peligro de extincion amenazadas raras y las

sujetas a proteccion especial y que establece especificaciones para

su proteccion Diario Oficial 16 May 1994 488 2ndash60

Servedio M R van Doorn G S Kopp M Frame A M amp Nosil P

(2011) Magic traits in speciation lsquomagicrsquo but not rare Trends in Ecology

amp Evolution 26 389ndash397 httpsdoiorg101016jtree201104005

Shahak Y amp Hauska G (2008) Sulfide oxidation from cyanobacteria to

humans Sulfide-quinone oxidoreductase (SQR) In R Hell C Dahl D

B Knaff amp Th Leustek (Eds) Advances in photosynthesis and respira-

tion (pp 319ndash335) Heidelberg Springer

Shalem O Sanjana N E amp Zhang F (2015) High-throughput func-

tional genomics using CRISPR-Cas9 Nature Reviews Genetics 16

299ndash311 httpsdoiorg101038nrg3899

Silva G J J Bye A el Azzouzi H amp Wisloslashff U (2017) MicroRNAs as

important regulators of exercise adaptation Progress in Cardiovascular

Diseases 60 130ndash151 httpsdoiorg101016jpcad201706003

Simola D F Wissler L Donahue G Waterhouse R M Helmkampf

M Roux J Viljakainen L (2013) Social insect genomes exhibit

dramatic evolution in gene composition and regulation while preserv-

ing regulatory features linked to sociality Genome Research 23

1235ndash1247 httpsdoiorg101101gr155408113

858 | TOBLER ET AL

Sommer-Trembo C Bierbach D Arias Rodriguez L Verel Y Jourdan

J Zimmer C Plath M (2016) Does personality affect premating

isolation between locally-adapted populations BMC Evolutionary Biol-

ogy 16 138 httpsdoiorg101186s12862-016-0712-2

Steindachner F (1863) Beitreuroage zur Kentniss der Sciaenoiden Brasiliens

und der Cyprinodonten Mejicos Sitzungsberichte der Mathematisch-

Naturwissenschaftlichen Classe der Kaiserlichen Akademie der Wis-

senschaften 48 162ndash185

Stinchcombe J R amp Hoekstra H E (2008) Combining population geno-

mics and quantitative genetics Finding genes underlying ecologically

important traits Heredity 100 158ndash170 httpsdoiorg101038sj

hdy6800937

Stipanuk M H (2004) Sulfur amino acid metabolism Pathways for pro-

duction and removal of homocysteine and cysteine Annual Review of

Nutrition 24 539ndash577 httpsdoiorg101146annurevnutr24

012003132418

Stipanuk M H amp Ueki I (2011) Dealing with methioninehomocysteine

sulfur Cysteine metabolism to taurine and inorganic sulfur Journal of

Inherited Metabolic Disease 34 17ndash32 httpsdoiorg101007

s10545-009-9006-9

Stockinger B Di Meglio P Gialitakis M amp Duarte J H (2014) The

aryl hydrocarbon receptor Multitasking in the immune system

Annual Review of Immunology 32 403ndash432 httpsdoiorg101146

annurev-immunol-032713-120245

Storz J F amp Wheat C W (2010) Integrating evolutionary and

functional approaches to infer adaptation at specific loci Evolution

64 2489ndash2509 httpsdoiorg101111j1558-5646201001044x

Stuart Y E Veen T Weber J N Hanson D Ravinet M Lohman B

K Bolnick D I (2017) Contrasting effects of environment and

genetics generate a continuum of parallel evolution Nature Ecology amp

Evolution 1 0158 httpsdoiorg101038s41559-017-0158

Sunnucks P Morales H E Lamb A M Pavlova A amp Greening C

(2017) Integrative approaches for studying mitochondrial and nuclear

genome co-evolution in oxidative phosphorylation Frontiers in Genet-

ics 8 25

Tobler M amp Hastings L (2011) Convergent patterns of body shape dif-

ferentiation in four different clades of poeciliid fishes inhabiting sul-

fide springs Evolutionary Biology 38 412ndash421 httpsdoiorg10

1007s11692-011-9129-4

Tobler M Palacios M Chapman L J Mitrofanov I Bierbach D

Plath M Mateos M (2011) Evolution in extreme environments

Replicated phenotypic differentiation in livebearing fish inhabiting

sulfidic springs Evolution 65 2213ndash2228 httpsdoiorg101111j

1558-5646201101298x

Tobler M Passow C N Greenway R Kelley J L amp Shaw J H

(2016) The evolutionary ecology of animals inhabiting hydrogen sul-

fide-rich environments Annual Review of Ecology Evolution and Sys-

tematics 47 239ndash262 httpsdoiorg101146annurev-ecolsys-

121415-032418

Tobler M amp Plath M (2009a) Threatened fishes of the world Gambu-

sia eurystoma Miller 1975 (Poeciliidae) Environmental Biology of

Fishes 85 251 httpsdoiorg101007s10641-009-9481-8

Tobler M amp Plath M (2009b) Threatened fishes of the world Poecilia

sulphuraria (Alvarez 1948) (Poeciliidae) Environmental Biology of

Fishes 85 333ndash334 httpsdoiorg101007s10641-009-9506-3

Tobler M Plath M Riesch R Schlupp I Grasse A Munimanda G

Moodley Y (2014) Selection from parasites favors immunogenic diver-

sity but not divergence among locally adapted host populations Journal

of Evolutionary Biology 27 960ndash974 httpsdoiorg101111jeb12370

Tobler M Riesch R amp Plath M (2015) Extremophile fishes An inte-

grative synthesis In R Riesch M Tobler amp M Plath (Eds) Extremo-

phile fishes Ecology evolution and physiology of teleosts in extreme

environments (pp 279ndash296) Heidelberg Germany Springer

Tobler M Riesch R Tobler C M amp Plath M (2009) Compensatory

behaviour in response to sulfide-induced hypoxia affects time

budgets feeding efficiency and predation risk Evolutionary Ecology

Research 11 935ndash948

Tobler M Scharnweber K Greenway R Passow C Arias-Rodriguez

L amp Garcia-De-Leon F J (2015) Convergent changes in the

trophic ecology if extremophile fish along replicated environmental

gradients Freshwater Biology 60 768ndash780 httpsdoiorg10

1111fwb12530

Todgham A E amp Stillman J H (2013) Physiological responses to shifts

in multiple environmental stressors Relevance in a changing worlds

Integrative and Comparative Biology 53 539ndash544 httpsdoiorg10

1093icbict086

van Doorn G S Edelaar P amp Weissing F J (2009) On the origin of

species by natural selection and sexual selection Science 326 1704ndash

1707 httpsdoiorg101126science1181661

Wall D H amp Virginia R A (1999) Controls on soil biodiversity Insights

from extreme environments Applied Soil Ecology 13 137ndash150

httpsdoiorg101016S0929-1393(99)00029-3

Wallace J L amp Wang R (2015) Hydrogen-sulfide based therapeutics

Exploiting a unique but ubiquitous gasotransmitter Nature Reviews

Drug Discovery 14 329ndash345 httpsdoiorg101038nrd4433

Wang R (2012) Physiological implications of hydrogen sulfide A whiff

exploration that blossomed Physiological Reviews 92 791ndash896

httpsdoiorg101152physrev000172011

Wang Z Gerstein M amp Snyder M (2009) RNA-seq A revolutionary

tool for transcriptomics Nature Reviews Genetics 10 57ndash63

httpsdoiorg101038nrg2484

Waterman T H (1999) The evolutionary challenges of extreme environ-

ments (part 1) Journal of Experimental Zoology 285 326ndash359

httpsdoiorg101002(SICI)1097-010X(19991215)2854amplt326

AID-JEZ3ampgt30CO2-T

Werren J H Richards S Desjardins C A Niehuis O Gadau J Col-

bourne J K amp Group N G W (2010) Functional and evolutionary

insights from the genomes of three parasitoid Nasonia species

Science 327 343ndash348 httpsdoiorg101126science1178028

Whitehead A (2012) Comparative genomics in ecological physiology

Toward a more nuanced understanding of acclimation and adapta-

tion Journal of Experimental Biology 215 884ndash891 httpsdoiorg

101242jeb058735

Whitehead A amp Crawford D L (2006) Neutral and adaptive variation

in gene expression Proceedings of the National Academy of Sciences

USA 103 5425ndash5430 httpsdoiorg101073pnas0507648103

Willett C S amp Burton R S (2004) Evolution of interacting proteins in

the mitochondrial electron transport system in a marine copepod

Molecular Biology and Evolution 21 443ndash453 httpsdoiorg10

1093molbevmsh031

Willi Y Van Buskirk J amp Hoffmann A (2006) Limits to the adaptive

potential of small populations Annual Review of Ecology Evolution

and Systematics 37 433ndash458 httpsdoiorg101146annurevecol

sys37091305110145

Wolff J N Ladoukakis E D Enrıquez J A amp Dowling D K (2014)

Mitonuclear interactions Evolutionary consequences over multiple

biological scales Philosophical Transactions of the Royal Society B 369

20130443 httpsdoiorg101098rstb20130443

How to cite this article Tobler M Kelley JL Plath M Riesch

R Extreme environments and the origins of biodiversity

Adaptation and speciation in sulphide spring fishes Mol Ecol

201827843ndash859 httpsdoiorg101111mec14497

TOBLER ET AL | 859