Submitted 21 December 2017Accepted 1 March 2018Published 21 March 2018

Corresponding authorSusan R Kennedysusanrkennedygmailcom

Academic editorLuiz Martinelli

Additional Information andDeclarations can be found onpage 13

DOI 107717peerj4527

Copyright2018 Kennedy et al

Distributed underCreative Commons CC-BY 40

OPEN ACCESS

Stable isotopes of Hawaiian spidersreflect substrate properties along achronosequenceSusan R Kennedy1 Todd E Dawson12 and Rosemary G Gillespie1

1Department of Environmental Science Policy and Management University of California BerkeleyBerkeley CA United States of America

2Department of Integrative Biology University of California Berkeley Berkeley CA United States of America

ABSTRACTThe Hawaiian Islands offer a unique opportunity to test how changes in the propertiesof an isolated ecosystem are propagated through the organisms that occur within thatecosystem The age-structured arrangement of volcanic-derived substrates follows aregular progression over space and by inference time We test how well documentedsuccessional changes in soil chemistry and associated vegetation are reflected inorganisms at higher trophic levelsmdashspecifically predatory arthropods (spiders)mdashacross a range of functional groups We focus on three separate spider lineages onethat builds capture webs one that hunts actively and one that specializes on eatingother spiders We analyze spiders from three sites across the Hawaiian chronosequencewith substrate ages ranging from 200 to 20000 years To measure the extent to whichchemical signatures of terrestrial substrates are propagated through higher trophiclevels we use standard stable isotope analyses of nitrogen and carbon with plantleaves included as a baseline The target taxa show the expected shift in isotope ratiosof δ15N with trophic level from plants to cursorial spiders to web-builders to spidereaters Remarkably organisms at all trophic levels also precisely reflect the successionalchanges in the soil stoichiometry of the island chronosequence demonstrating how thebiogeochemistry of the entire food web is determined by ecosystem succession of thesubstrates on which the organisms have evolved

Subjects Biochemistry Biogeography Ecology Biosphere InteractionsKeywords Nutrient cycling Stable isotopes Chronosequence Arthropods Spiders NitrogenTrophic niche

INTRODUCTIONEvolutionary processes are determined in large part by the ecosystems within which theytake place While connecting the processes of evolutionary biology and ecology remainsa critical frontier in biological sciences (Matthews et al 2011) there are few studies thatdemonstrate how mechanisms driving processes of evolutionary biology and ecosystemscience are linked The current study seeks to understand how organismal diversity mayreflect successional shifts in soil chemistry by testing the extent to which organisms atdifferent trophic levels reflect the properties of the substrates on which they occur

The Hawaiian Archipelago presents a highly suitable system for studying the linkbetween evolutionary processes and ecosystem properties The current high islands of

How to cite this article Kennedy et al (2018) Stable isotopes of Hawaiian spiders reflect substrate properties along a chronosequencePeerJ 6e4527 DOI 107717peerj4527

Hawaii are arranged sequentially from oldest to youngest with Kauai at 51 million yearsin the far northwest and Hawairsquoi Island at lt1 million years in the southeast (Carson ampClague 1995) This sequential order is a consequence of the archipelago being located on avolcanic hot spot where magma upwelling from the earthrsquos mantle has formed into largeshield volcanoes At the same time the tectonic plate on which the islands are situatedis moving toward the northwest such that each newly emerging island has appeared tothe southeast of its next-oldest neighbor The resultant nearly-linear age gradient makesHawaii an ideal chronosequence a temporally varied system in which the ecosystems of theyounger sites are currently developing in a manner assumed to reflect the developmentalhistory of the older sites (Walker et al 2010) Given fairly precise information on theage of formation and subsequent history across a chronosequence these systems canprovide unprecedented insights into ecosystem development Thus chronosequences haveadded significantly to our understanding of how nutrients change over time (Vitousek2004) and the impact of changes in soil nutrient availability on plants (Wardle et al2008) decomposers (Williamson Wardle amp Yeates 2005 Doblas-Miranda et al 2008)above-ground and below-ground ecosystem processes (Wardle Walker amp Bardgett 2004)and entire arthropod communities (Gruner 2007)

Hawaii has served as a chronosequence for detailed studies on the ways in whichecosystem properties and functions change over extended time (Vitousek 2004) Nutrientflow and its impacts on primary producers (trees) have been well characterized in thissystem Studies have examined the evolution of soils on substrates of different surficialage (300 ymdash41 Mya) controlled for elevation climate land use history and canopyvegetation (Metrosideros polymorpha) with all minerals derived from volcanic ash Animportant finding of this work was that soil nitrogen foliar nutrient availability andproductivity start off very low increase rapidly with substrate age peak on substrates ofintermediate age (ca 20000 y) on the youngest island and then decline rapidly on olderislands before all but disappearing on the oldest (Vitousek Shearer amp Kohl 1989 VitousekTurner amp Kitayama 1995 Vitousek et al 1997) A more recent study found that tree heightand canopy nitrogen also peak on intermediate-aged (20000 y) substrates on the youngestisland (Vitousek et al 2009) Nitrogen isotopes follow a similar pattern with foliar δ15Nvery low at the youngest sites increasing with substrate age and highest at a 67000 y site(Vitousek Shearer amp Kohl 1989)

Geologic history and nutrient flow evidently have important effects on the lowesttrophic levelmdashplantsmdashbut little is known about how these effects might be propagatedthrough higher trophic levels ie higher-level consumers At the same time work onthe effects of substrate age on above-ground systems including whole communitieshas shown that community traits such as population species diversity (Gillespie ampBaldwin 2010Lim ampMarshall 2017) genetic structure (Roderick et al 2012) andnetworkmodularity (Rominger et al 2016) change in a predictable manner across the Hawaiianchronosequence However although these community-level studies have markedlyenhanced our understanding of the changes in community ecology over time there hasas yet been no attempt to link chemical changes during the evolution and development ofsoils (and associated ecosystem properties) with the abundance diversity and evolutionary

Kennedy et al (2018) PeerJ DOI 107717peerj4527 218

histories of above-ground organisms The current study begins to address this question bytesting the effects of substrate age on the biochemistrymdashisotopic signaturesmdashof secondaryconsumers (predators) across a Hawaiian chronosequence

The application of stable isotope information has revolutionized studies of nutrientflow and niche ecology in a wide range of organisms (eg Fry 1988 Hobson amp Welch1992 Muschick Indermaur amp Salzburger 2012) In particular nitrogen and carbon stableisotopes have been found to reflect trophic position both δ15N and δ13C tend to increasein a predictable manner with each successive trophic level (Post 2002 but see De Vries etal 2015) Stable isotopes have also been used to track nutrient flow climatic shifts andmigration patterns in a variety of ecological systems (Best amp Schell 1996 Chamberlain etal 1997 Iacumin Davanzo amp Nikolaev 2005 McMahon et al 2016) The current studyuses stable isotopes of N and C to assess the extent to which entire food webs are influencedby the chemistry of their habitats We chose to focus on spiders because they are mobilegeneralist predators and encompass multiple trophic levels from feeding on herbivorousinsects to specializing on the highest predator levels among Hawaiian arthropods (otherspiders) This substantial variation in trophic ecology allows us to test how functionaland trophic differences are reflected in isotopic signatures and the extent to which thebiogeochemistry of a food web is determined by the chemistry of the immediate substrateThe use of N and C stable isotopes is especially well suited to this study because of thepredictable manner in which both elements can reflect trophic position and because ofthe importance of nitrogen in ecosystem development (Boring et al 1988)

We analyzed Hawaiian spiders belonging to two lineages within the adaptive radiationof long-jawed orb-weavers (Tetragnatha Tetragnathidae) and one lineage within the stickspiders (Ariamnes Theridiidae) The Tetragnatha radiation includes ca 60 species whichdisplay a spectacular array of colors shapes sizes behaviors and ecological affinities notobserved elsewhere in the range of this genus (Blackledge amp Gillespie 2004 Gillespie 2004Gillespie 2015) The radiation consists of two major clades one that spins webs for preycapture (lsquolsquoweb buildersrsquorsquo) and another that has lost the web-spinning behavior and insteadhunts actively (the lsquolsquoSpiny Legrsquorsquo clade Gillespie 1991 Gillespie 2002) Observational dataindicate that both web-building and Spiny Leg Tetragnatha feed on a mixture of insectherbivores and predators (Binford 2001) although the exact composition of these spidersrsquodiets has not yet been fully characterized The Hawaiian Ariamnes currently representedby 11 known species across the Hawaiian Islands (Gillespie amp Rivera 2007) are alsoecologically diverse and largely araneophagic (ie preying on other spiders) (Gillespie etal 2018) Like Tetragnatha these spiders are exclusively nocturnal and like the Spiny LegTetragnatha they hunt without the use of a web

Given the contrasting hunting strategies (web-building versus active hunting) andtrophic positions (generalist versus araneophagic) across these spider lineages thethree groups vary predictably in their position in the food web from largely feedingon primary consumers (ie insect herbivores) to exclusively feeding on secondary andhigher consumers (ie spiders) Within this system we tested the hypothesis that isotopicsignatures of spiders should reflect the biogeochemistry of their respective habitats fromyoung to older in the Hawaiian chronosequence Thus not only should the different spider

Kennedy et al (2018) PeerJ DOI 107717peerj4527 318

lineages illustrate now-standard expectations for isotope signatures associated with risingtrophic levels but the trophic ecology of the entire food web should reflect changes in soilchemistry across the chronosequence In particular given that δ15N in the soil increasesduring the building phase of the Hawaiian ecosystems (Vitousek et al 1997) we expectδ15N to be lowest in the spiders at the youngest (200ndash750 y) site and highest in the spidersat the oldest (20000 y) site

METHODSStudy sitesHawaiirsquos sequential age structure has made it an ideal system for previous studies on soilevolution and nutrient cycling wherein the Long Substrate Age Gradient (LSAG) wasestablished (Crews et al 1995 Vitousek 2004) This study focuses on Hawairsquoi Island theyoungest in the archipelago because the largest possible range of N availabilities is expectedto be found there soil nitrogen is lowest in the youngest substrates and peaks in the oldersubstrates on Hawairsquoi Island before declining on the older islands (Vitousek Turner ampKitayama 1995)

Specimens were collected on the windward side of Hawairsquoi Island under permits from theState of Hawaii Department of Land and Natural Resources (endorsement FHM14-349)and the National Park Service (study HAVO-00425) Three sites of different substrate agechosen for their comparable elevations and climates as well as the overlap of two of the sites(Olaa and Laupahoehoe) with those characterized based on soils (Vitousek et al 1997)were sampled (see Fig 1) Substrate ages were determined based on data from the UnitedStates Geological Survey (Sherrod et al 2007) and from the LSAG (Crews et al 1995Vitousek 2004) All three sites are wetmesic forest dominated byMetrosideros polymorphaand Acacia koa in the canopy with Cibotium spp dominating the understory The sitesrange from 1180 to 1390 m in elevation with mean annual temperatures of 139 to 154degrees Celsius and mean annual rainfall of 3035 to 3090 mm (Giambelluca et al 2014)

Upper Waiakea is a very young site on a 200- to 750-year-old lava flow in a stratifiedmatrix of differently-aged substrates within the Upper Waiakea Forest Reserve off ofStainback Highway on Mauna Loa Olaa Forest is on an older lava flow on Kilaueasituated within Hawairsquoi Volcanoes National Park The trees in Olaa are rooted in a thicklayer of tephra of approximately 2100 years old (Vitousek 2004) beneath which is an olderflow of 5000ndash11000 y (see Fig 1) Although the United States Geological Survey (USGS)classifies this substrate as 5000ndash11000 y we consider its biota to be influenced by thechemical properties of the 2100-year-old tephra in which the forest is rooted Olaa istherefore approximately one order of magnitude older than Upper Waiakea yet is locatedjust 115 km S of the younger site This proximity makes the two sites especially useful formeasuring effects of habitat age on a small geographical scale The oldest site in this studyis Laupahoehoe located in the Laupahoehoe Experimental Forest Unit on Mauna KeaWhile USGS data (Sherrod et al 2007) estimate the lava flow age at 5000ndash11000 y thesampling locality overlaps with an LSAG site which has been studied in great detail andwhose forest is rooted in a layer of soil believed to be approximately 20000 y old (Vitousek

Kennedy et al (2018) PeerJ DOI 107717peerj4527 418

Figure 1 Map showing field sites where samples were collected Colors represent geology and lava flowage determined by United States Geological Survey (Sherrod et al 2007) Where applicable substrate ageclassifications determined by the Long Substrate Age Gradient (LSAG Crews et al 1995) are included

Full-size DOI 107717peerj4527fig-1

2004 Vitousek et al 2009) We therefore follow the LSAG classification of 20000 y forLaupahoehoe

Collections at each site were centered at the following coordinates with searchesextending up to 100 m in any direction

bull Upper Waiakea 19562N 155272Wbull Olaa 19462N 155248Wbull Laupahoehoe 19922N 155301W

Specimen collectionSix species of Tetragnatha spiders (Tetragnatha anuenue T brevignatha T hawaiensis Tperkinsi T quasimodo and the undescribed species T lsquolsquogolden domersquorsquo) and two species ofAriamnes spiders (Ariamnes hiwa and A waikula) were collected in the field from 11Marchto 18 April 2015 and from 6 to 15 February 2016 All study species are morphologicallydistinct and can be easily identified in the field Plant samples were leaves of dominantor common forest vegetation (Metrosideros polymorpha Cibotium spp and Cheirodendronspp) all of which are also easily identified on sight After a preliminary analysis it wasdetermined that leaf litter should be added to the study in order to help explain differencesin carbon isotope values Unfortunately permitting and time constraints only allowed forleaf litter to be collected from the youngest site (Upper Waiakea) on 15 October 2016

Spiders were individually hand-captured into clean plastic snap-cap vials which werelabeled on the outside with unique identifiers while plant leaves were clipped off withscissors and stored in labeled paper envelopes In Upper Waiakea leaf litter was alsocollected from the bases of M polymorpha trees and placed in paper envelopes Leaves

Kennedy et al (2018) PeerJ DOI 107717peerj4527 518

were air-dried in their envelopes in a sealed container of silica for three weeks prior toprocessing Leaf litter was dried in a 60 C oven overnight

Spiders were photographed up close using a Nikon D5200 with an AF-S DX Micro-NIKKOR 40 mm f28 g lens and a Speedlight SB-400 flash from both dorsal and lateralangles This created a photographic voucher and allowed for visual identification of speciessex and maturity Spiders were killed overnight in a freezer and air-dried in separate snapcap vials each with one clean bead of silica gel before being transferred to individual 2-mLcentrifuge tubes

Stable isotope analysisIndividual dried spiders were weighed into 9 times 5 mm pressed tin capsules for isotopicanalysis To optimize N content for isotopic analysis a 15-mg mass was recommendedfor each sample (S Mambelli UC Berkeley Center for Stable Isotope Biogeochemistrypers comm 2014) Due to the spidersrsquo small body size (ranging from ca 04 to 5 mg dryweight for the majority of specimens) it was not feasible to obtain sufficient material fromindividual body parts although it has been found that different spider tissues can undergodifferent isotopic turnover rates (Belivanov amp Hambaumlck 2015) Therefore in order tocontrol for possible variation in isotopic turnover among tissue types the spidersrsquo entirebodies were analyzed When spiders exceeded 25 mg dry weight they were homogenized(powdered and mixed) with a mortar and pestle and a 15-mg sample of the homogenizedtissue was used For smaller spiders (lt25 mg) the whole intact body was packed intothe tin capsule in order to avoid excessive loss of material Plant leaves were individuallyhomogenized in a Mini-Beadbeater (BioSpec Model 8) in 7-mL tubes with stainless steelball bearings then weighed into 9times 5 mm tin capsules Due to the relatively low NC ratioof plants 6 mg of material was weighed out for each leaf sample Leaf litter was processedin the same manner as plant leaves

Samples were analyzed for nitrogen and carbon content ( dry weight) and nitrogenand carbon stable isotope ratios via elemental analyzercontinuous flow isotope ratiomass spectrometry using a CHNOS Elemental Analyzer (model Vario ISOTOPE cubeElementar Hanau Germany) coupled with an IsoPrime 100 mass spectrometer (IsoprimeLtd Cheadle UK) The isotope ratio is expressed in lsquolsquodeltarsquorsquo notation (in parts per thousandor h units) The isotopic composition of a material relative to that of a standard on aper mill deviation basis is given by δ15N (or δ13C) = (RsampleRstandardminus1) times 1000 whereR is the molecular ratio of heavy to light isotopes The standard for nitrogen is air Thestandard for carbon is V-PDB The reference material NIST SMR 1547 (peach leaves)was used as calibration standard (long-term precision [since 2000] using this standardis plusmn007h for both N and C isotope analyses) All isotope analyses were conducted atthe Center for Stable Isotope Biogeochemistry at the University of California BerkeleyLong-term external precision based on reference material NIST SMR 1577b (bovine liver)is 015h and 010h respectively for N and C isotope analyses

Data analysisResults from the isotopic analysis were categorized into the following functional groupslsquolsquoplantrsquorsquo (foliar samples of the genera Metrosideros Cibotium and Cheirodendron) lsquolsquoSpiny

Kennedy et al (2018) PeerJ DOI 107717peerj4527 618

Table 1 2-way ANOVA results Results of 2-way ANOVA testing for effects of site functional group andsitetimes functional group interaction on stable isotopes of samples Significant effects are indicated in bold

Isotope Effect F df p-value

Site 6921 2 lt0001Functional group 1136 4 lt0001δ15N

Sitefunctional group 8615 6 lt0001Site 5551 2 lt0001Functional group 9515 4 lt0001δ13C

Sitefunctional group 1841 6 0092

Legrsquorsquo (Spiny Leg Tetragnatha species T anuenue T brevignatha and T quasimodo) lsquolsquowebrsquorsquo(Tetragnatha hawaiensis T perkinsi and the undescribed species nicknamed T lsquolsquogoldendomersquorsquo) and lsquolsquoAriamnesrsquorsquo (Ariamnes hiwa and A waikula the two species found on HawairsquoiIsland (Gillespie amp Rivera 2007)) Although all spider species were initially analyzedseparately the results showed that the data grouped taxa together largely in accordancewith functional group Because members of one group (eg web-builders) were closer toone another than to other groups (eg Spiny Leg) we chose to focus on these broaderecological categoriesmdashfunctional groupsmdashrather than species

Effects of site and functional group on δ15N and δ13C were tested using a 2-way Anovaallowing for interaction (with site and functional group as factors) on R statistical software(version 322 64-bit) Main effects were then analyzed separately we tested (1) effect ofsite within each functional group and (2) effect of functional group within each site Whensignificant differences were found pairwise comparisons were made using Tukeyrsquos honestsignificant difference test (Tukey 1949)

RESULTSWe found a significant interaction between site and functional group for δ15N (F = 8615plt 0001) but not for δ13C (F = 1841 p= 0092 Table 1) Significant main effects werefound for all variables tested site for δ15N (F = 6921 plt 0001) functional group forδ15N (F = 1136 plt 0001) site for δ13C (F = 5551 plt 0001) and functional group forδ13C (F = 9515 plt 0001)

Main effects siteFor δ15N a significant site effect was found within all functional groups (plants F = 7874plt 0001 Spiny Leg F = 4469 plt 0001 web-builders F = 2166 plt 0001 AriamnesF = 8087 plt 0001 Table 2 and Fig 2) We performed a Tukeyrsquos HSD test to revealpairwise differences among sites δ15N showed a clear pattern of stepwise increase withsubstrate age (lowest in Upper Waiakea (200ndash750 y) intermediate in Olaa (2100 y) andhighest in Laupahoehoe (20000 y)) This pattern held true for every functional group theonly comparison not found to be significant was Olaa vs Laupahoehoe within Ariamnes(Tukeyrsquos adjusted p= 0203)

For δ13C a significant site effect was found within every functional group except forplants (plants F = 07997 p= 0482 Spiny Leg F = 5681 p= 0005 web-builders

Kennedy et al (2018) PeerJ DOI 107717peerj4527 718

Table 2 Main effects of site within functional groupMain effects of site (substrate ages Upper Wa-iakea 200ndash750 y Olaa 2100 y Laupahoehoe 20000 y) within each functional group of spiders andplants Site was found to have a significant effect on both C and N isotope ratios of every functional groupwith the exception of δ13C in plants

Isotope Comparison F df p-value

Plants 7874 2 lt0001Spiny Leg 4469 2 lt0001Web-builders 2166 2 lt0001

δ15N

Ariamnes 8087 2 lt0001Plants 07997 2 0482Spiny Leg 5681 2 0005Web-builders 3191 2 lt0001

δ13C

Ariamnes 3662 2 lt0001

minus10

minus50

5

δ15N

Upper Waiakea 200-750 y

lsquoOlarsquoa2100 y

Laupahoehoe 20000 y

plantsleaf litterSpiny Legweb-buildersAriamnes

Figure 2 Boxplots showing nitrogen isotope ratios of functional groups across sitesNitrogen isotoperatio (δ15N in h units) of plant leaves (green) leaf litter (purple) Spiny Leg (brown) web-building (yel-low) and Ariamnes (blue) spiders across sites of different ages

Full-size DOI 107717peerj4527fig-2

F = 3191 plt 0001 Ariamnes F = 3662 plt 0001 Table 2 and Fig 3) Among the threegroups of spiders δ13C was significantly lower in Laupahoehoe (20000 y) than in Olaa(2100 y)

Main effects functional groupA significant functional group effect was found in all sites for δ15N (Upper WaiakeaF = 6838 plt 0001 Olaa F = 3423 plt 0001 Laupahoehoe F = 2890 plt 0001Table 3 and Fig 2) A Tukeyrsquos HSD test found significant differences among every pairof groups except for the following pairs in Upper Waiakea web-builders vs Ariamnes(Tukeyrsquos adjusted p-value = 0071) web-builders vs leaf litter (Tukeyrsquos adjusted p-value= 0998) and Ariamnes vs leaf litter (Tukeyrsquos adjusted p-value = 0512)

For δ13C a significant functional group effect was found at all three sites (UpperWaiakea F = 3642 plt 0001 Olaa F = 4129 plt 0001 Laupahoehoe F = 4148

Kennedy et al (2018) PeerJ DOI 107717peerj4527 818

minus32

minus30

minus28

minus26

minus24

δ C13

Upper Waiakea 200-750 y

lsquoOlarsquoa2100 y

Laupahoehoe 20000 y

plantsleaf litterSpiny Legweb-buildersAriamnes

Figure 3 Boxplots showing carbon isotope ratios of functional groups across sites Carbon isotope ra-tio (δ13C in h units) of plant leaves (green) leaf litter (purple) Spiny Leg (brown) web-building (yel-low) and Ariamnes (blue) spiders across sites of different ages

Full-size DOI 107717peerj4527fig-3

Table 3 Main effects of functional group within site Functional groups were found to differ signifi-cantly from one another in their isotope ratios of both C and N at all three sites

Isotope Site F df p-value

Upper Waiakea (200ndash750 y) 6838 4 lt0001Olaa (2100 y) 3423 3 lt0001δ15N

Laupahoehoe (20000 y) 2890 3 lt0001Upper Waiakea (200ndash750 y) 3642 4 lt0001Olaa (2100 y) 4129 3 lt0001δ13C

Laupahoehoe (20000 y) 4148 3 lt0001

plt 0001 Table 3 and Fig 3) At all sites plants were significantly lower in δ13C thanthe next-lowest trophic level Spiny Leg spiders (consumers) In Olaa (2100 y) SpinyLeg spiders had significantly lower δ13C than either web-builders or Ariamnes (Tukeyrsquosadjusted p-valuelt 0001) InUpperWaiakea (200ndash750 y) although there was no significantdifference between Ariamnes and either of the Tetragnatha functional groups the Spiny LegTetragnatha were significantly lower in δ13C than web-builders (Tukeyrsquos adjusted p-valuelt 0001) In Laupahoehoe (20000 y) no significant difference was found between any pairof spider groups

DISCUSSIONThe results from this study provide a novel perspective on how changes in the substratechemistry of the terrestrial land surfaces across a chronosequence of developing ecosystemsare propagated up through an entire food web To our knowledge this is also the firststudy to characterize the isotopic signatures of different ecological groups represented byexemplary adaptive radiations of spiders in Hawaii

Kennedy et al (2018) PeerJ DOI 107717peerj4527 918

Parallel shifts in isotopic signature across the chronosequence ofecosystem developmentPrevious work on the ecological characteristics of forests across the chronosequencehas detailed the evolution of Hawaiian ecosystems in the context of soil properties andvegetation (Vitousek Shearer amp Kohl 1989 Vitousek Turner amp Kitayama 1995 Vitouseket al 1997) Our results show that the chemical signatures of nutrient availability thatcharacterize a given site are borne all the way up to the highest trophic levelsmdashtoppredatorsmdashon a Hawaiian chronosequence Indeed the δ15N values of spiders perfectlymatch expectations for their respective habitats Where nitrogen is most limitedmdashat theyoungest site (Upper Waiakea)mdashspiders have the lowest values of δ15N as substrateage and nitrogen availability increase so too does the δ15N of spiders This makes sensephysiologically because when nitrogen is very limited the lighter isotope (14N) is notas easily lost in reactions and is instead retained at a greater rate in an organismrsquostissues (Austin amp Vitousek 1998) Conversely when biologically available nitrogen is veryabundant 14N is readily lost (eg in excretion) leaving behind a greater proportion ofthe heavier 15N in the organismrsquos tissues Thus our results support the hypothesis that theisotopic signatures of the spidersmdashas well as the plantsmdashtrack the changes in the geologicalage of the islands (Sherrod et al 2007) and the associated changes in nitrogen in soils acrossthe Hawaiian chronosequence measured by Vitousek et al (1997)

While it might not be surprising that the increase in δ15N in the spiders tracks theincreases in plants across the geological gradient the fact that the relationship is sotight is remarkable as it suggests that even higher-level consumers (predators) reflectthe δ15N of the immediate site This result is especially notable because it was found incursorial animals which because of their mobility might be expected to show only a weakassociation with the substrates on which they were collected Instead the spiders carry clearsignatures of their immediate ecosystem Given that the sites that were sampled are in veryclose proximity (115 km between Olaa and Upper Waiakea) and not separated by anysignificant physical barrier the results suggest an extraordinary level of isolation amongspider populations This has implications for the mechanisms by which the Tetragnathaadaptive radiation may have arisen Isolation between populations separated by shortdistances can serve as a crucible for evolution (Carson Lockwood amp Craddock 1990)Perhaps the same mechanisms that are currently at work on Hawaiirsquos youngest islandalso led to the rise of the approximately 60 endemic Tetragnatha species found acrossthe archipelago today Results for Ariamnes were similar all higher than the other spiderlineages and values increasing with substrate age (though the increase from Olaa toLaupahoehoe was not significant presumably due to a relatively small sample size (n= 9)at these sites) Ariamnes like the Tetragnatha has undergone a substantial though smallerat 11ndash16 species adaptive radiation across the islands (Gillespie et al 2018)

The carbon isotope data show a less clear pattern than nitrogen but neverthelessindicate site-specific differences among the spiders Notably the plant samples did notdiffer significantly in δ13C among the three sites suggesting that perhaps δ13C does notaccurately reflect nutrient differences among the substrates By contrast δ15N appears tostrongly reflect nutrient availability at the different sites However spiders did consistently

Kennedy et al (2018) PeerJ DOI 107717peerj4527 1018

show higher δ13C in Olaa (2100 y) than in Laupahoehoe (20000 y) Thus the relationshipbetween baseline (plant) signatures and higher predator (spider) signatures is weaker incarbon than in nitrogen This suggests that spiders at the three sites may not be consumingexactly the same assemblages of prey perhaps due to variations in the availability ofdifferent insect (potential prey) species at different sites A detailed study of the precisecompositions of these spidersrsquo diets using either molecular gut content analysis (egKrehenwinkel et al 2017) or an isotopic mixing model with robust sampling of the entirearthropod community could greatly enhance our understanding of the processes thataccount for the differences in δ13C among spider populations

Trophic positionsOur stable nitrogen isotope data reflect the different functional roles and trophic positions ofthe Hawaiian spiders Our results are consistent with the enrichment of the heavier isotope15N at higher trophic levels with plants having the lowest values of δ15N Tetragnathahaving intermediate values and the spider-eatingAriamneshaving the highest Additionallywe found that the δ15N of the Spiny Leg (cursorial) spiders was consistently lower than thatof web-buildingTetragnatha A simple explanation for the difference in δ15N in cursorial vsweb-building spiders may be that the different functional groups consume different preyCursorial spiders are likely to interact with abundant insect herbivores while web-buildersmay trap a larger proportion of flying insects at higher trophic levels such as hymenopteranor dipteran parasitoids decomposers or predators This dietary difference has been usedto explain the phenomenon of a higher δ15N in web-builders compared to cursorial spidersin a forest hedge community (Sanders Vogel amp Knop 2015) Another possible explanationis that the difference is due to the manufacturing of the orb web itself Given that websare an excretory product and that excretion tends to favor the lighter 14N it may be thatthe higher levels of lsquolsquoexcretionrsquorsquo lead to an enrichment in 15N in the bodies of web-builderscompared with cursorial spiders A number of previous studies suggest such an effect Forexample across a community of web-building riparian spiders lower δ15N was found inMiagrammopes (Uloboridae Kelly Cuevas amp Ramiacuterez 2015) a genus characterized by areduced capture web (often just a single line Lubin Eberhard amp Montgomery 1978) thanin other spiders Likewise a study of niche width across a guild of spiders showed thatcursorial spiders consistently had the lowest δ15N while orb web spiders had the highest(Sanders Vogel amp Knop 2015) In each of these systems the cursorial spiders showed thehighest levels of intraguild predation (ie feeding on other spiders) indicating that trophicposition itself is insufficient to explain the lower δ15N of the cursorial spiders relative to theweb spinners This observation raises the possibility again that the web spinning processitself leads to δ15N enrichment However further data are clearly needed to determinewhich of these explanations best accounts for the now recurring pattern of higher δ15N inweb-builders compared with cursorial taxa

The patternswe found in δ13Cwere less dramatic than in δ15N butmatched expectationsFoliar samples consistently had the lowest δ13C of all functional groups Spider values weresubstantially offset from leaf valuesmdashapproximately 4ndash5 per mill higher at all sitesmdashwhichsuggests a complex food chain consisting of many trophic levels below the spiders This is

Kennedy et al (2018) PeerJ DOI 107717peerj4527 1118

plausible given that spiders are obligate predators (secondary consumers) and thereforemust be trophically removed from plants by at least two levels The conventional wisdomwith δ13C is lsquolsquoyou are what you eatrsquorsquo (Hobson Barnett-Johnson amp Cerling 2010) meaningthat most organisms are only slightly enriched in 13C relative to their diets with standardpublished offsets of less than 1 per mill for each successive trophic level (Post 2002)Indeed meta-analysis of isotopic studies has found an average discrimination factor ofsim03 per mill from one trophic level to the next within invertebrates (Caut Angulo ampCourchamp 2009) although it should be noted that many of the invertebrates included inthat meta-analysis are aquatic and no lsquolsquostandardrsquorsquo δ13C offset for spiders or other terrestrialarthropods has yet been established In an effort to fill the large gap between plants andspiders we added samples of leaf litter from Upper Waiakea The δ13C of leaf litter fitneatly between leaves and spiders This is to be expected given that the lighter 12C is lostas respired 12CO2 that is produced at a greater rate during decomposition leaving theremaining litter relatively 13C enriched (Dawson et al 2002) Furthermore this findingfits with previous studies of Tetragnatha trophic ecology wherein it was observed thattipulid flies comprise a large proportion of the diet of Tetragnatha on Maui (Binford 2001Blackledge Binford amp Gillespie 2003) Because tipulid larvae often feed on decomposingleaves (Williams 1942) it is reasonable to surmise that tipulidsrsquo own δ13C values fall closeto those of the leaf litter and that spiders on Hawairsquoi Island become relatively enriched intheir δ13C composition when feeding on these insects

Link between ecosystem properties and evolutionary processesThis study demonstrates that organisms at multiple tropic levels reflect the stoichiometricchanges in soil across the geological chronosequence of the island from very young (200ndash750 y) to older (20000 y) The importance of this result is that it shows that the evolutionaryprocesses associated with diversification are intimately linked to a landscape that itselfchanges through time The detailed work of Vitousek Shearer amp Kohl (1989) VitousekTurner amp Kitayama (1995) Vitousek et al (1997) and Vitousek et al (2009) documents thepattern of change in soil chemistry over extended time periods Nitrogen and phosphorusincrease almost linearly with time in the early stages of substrate development (up to 20000y for nitrogen and 150000 y for phosphorus) this increase then levels off and declines onthe oldest islands (4 my)

At the same time it is now well established that organismal diversity increases over timeduring the early stages of formation of an island archipelago (Whittaker Triantis amp Ladle2008 Lim ampMarshall 2017) and that higher trophic levels depend on lower levels in islandcommunity assembly (Simberloff amp Wilson 1970) yet explanations for such patterns haveas yet considered only area and age of the landscape The documentation of peaks ofdiversity on middle aged islands of the Hawaiian Archipelago has been explained variouslybased on the interaction between age and area (Gillespie amp Baldwin 2010 Lim ampMarshall2017) Notably missing from these studies is a link between evolutionary processes ofdiversification and shifts in nutrient availability associated with ecosystem succession Thatorganisms inHawaii are intimately reflective of the ecosystem properties of their immediatehabitat demonstrates that changes in nutrients associated with the island chronosequence

Kennedy et al (2018) PeerJ DOI 107717peerj4527 1218

are propagated through trophic and functional groups of entire biological communitiesWhile initial work has begun to address the ecosystem consequences of evolutionarychange (Elser et al 2003 Laiolo et al 2015) this study provides preliminary insights intohow ecosystem change may affect processes of evolution

CONCLUSIONSVariation in δ15N data indicates that different spider lineages reflect their differentfunctional roles and trophic positions in Hawaiian food webs from those feeding largelyon primary consumers to those feeding exclusively on secondary and higher consumersImportantly the relationships between these groups in terms of their δ15N remain strongacross the chronosequence Not only do the spidersrsquo relative values of δ15N show the samepattern at each site but their isotopic signatures also reflect the availability of nitrogenat different sites from younger to older ecosystems The tight relationship between Navailability plant isotopic values and spider isotopic values strongly suggests that thespiders are dispersal-limited and their populations are isolated from one another evenacross short distances Such isolation may be an important mechanism of speciation withinthe Tetragnatha adaptive radiation This study shows that these evolving lineages of spidersare intimately associated with the properties of their ecosystem which is also changing Thetight connection between the organisms and the characteristics of their substrate highlightsthe importance of considering the role of soil properties particularly chemistry in additionto age and area to understand how biodiversity accumulates over time

ACKNOWLEDGEMENTSWe thank Stefania Mambelli and Wenbo Yang for their help with the stable isotopeanalyses We also thank our colleagues on the Dimensions team for invaluable assistancewith logistics in the field the staffs of the Hawaii National Park Service and Natural AreaReserve System for help with permits Charles Griswold and Alexandra Rueda Estebanfor assisting with sample collection and Natalie Graham for collecting leaf litter We aredeeply grateful to the reviewers and editors who provided helpful feedback on an earliersubmission of this paper

ADDITIONAL INFORMATION AND DECLARATIONS

FundingThis material is based upon work supported by the National Science Foundation GraduateResearch Fellowship under Grant No DGE 1106400 This research was supported in partby the Margaret C Walker Fund for teaching and research in systematic entomologyAdditional funding was provided by the Robert L Usinger Memorial Award and theHarvey I Magy Memorial Scholarship Award There was no additional external fundingreceived for this study The funders had no role in study design data collection and analysisdecision to publish or preparation of the manuscript

Kennedy et al (2018) PeerJ DOI 107717peerj4527 1318

Grant DisclosuresThe following grant information was disclosed by the authorsNational Science Foundation Graduate Research Fellowship DGE 1106400Margaret C Walker FundRobert L Usinger Memorial AwardHarvey I Magy Memorial Scholarship Award

Competing InterestsThe authors declare there are no competing interests

Author Contributionsbull Susan R Kennedy conceived and designed the experiments performed the experimentsanalyzed the data prepared figures andor tables authored or reviewed drafts of thepaper approved the final draft

bull Todd E Dawson conceived and designed the experiments contributed reagentsmate-rialsanalysis tools authored or reviewed drafts of the paper approved the final draft

bull Rosemary G Gillespie conceived and designed the experiments contributedreagentsmaterialsanalysis tools authored or reviewed drafts of the paper approved thefinal draft

Field Study PermissionsThe following information was supplied relating to field study approvals (ie approvingbody and any reference numbers)

Field work was approved by the State of Hawaii Department of Land and NaturalResources (endorsement number FHM14-349) and the National Park Service (studynumber HAVO-00425)

Data AvailabilityData will be available at UC Berkeleyrsquos Essig Musuem database (httpsessigdbberkeleyedu) as well as at Kennedy Susan R Dawson Todd E amp Gillespie Rosemary G(2018) Raw data Stable isotopes of Hawaiian spiders reflect substrate properties along achronosequence [Data set] PeerJ Zenodo httpdoiorg105281zenodo1194482

Supplemental InformationSupplemental information for this article can be found online at httpdxdoiorg107717peerj4527supplemental-information

REFERENCESAustin AT Vitousek PM 1998 Nutrient dynamics on a precipitation gradient in

Hawairsquoi Oecologia 113519ndash529 DOI 101007s004420050405Belivanov YK Hambaumlck PA 2015 The time scale of isotope signals in spiders molting

the remains of a previous diet Entomologia Experimentalis et Applicata 156271ndash278DOI 101111eea12328

Kennedy et al (2018) PeerJ DOI 107717peerj4527 1418

Best PB Schell DM 1996 Stable isotopes in southern right whale (Eubalaena australis)baleen as indicators of seasonal movements feeding and growthMarine Biology124483ndash494 DOI 101007BF00351030

Binford GJ 2001 Differences in venom composition between orb-weaving and wan-dering Hawaiian Tetragnatha (Araneae) Biological Journal of the Linnean Society74581ndash595 DOI 101111j1095-83122001tb01415x

Blackledge TA Binford GJ Gillespie RG 2003 Resource use within a community ofHawaiian spiders (Araneae Tetragnathidae) Annales Zoologici Fennici 40293ndash303

Blackledge TA Gillespie RG 2004 Convergent evolution of behavior in an adaptive ra-diation of Hawaiian web-building spiders Proceedings of the National Academy of Sci-ences of the United States of America 10116228ndash16233 DOI 101073pnas0407395101

Boring LR SwankWTWaide JB Henderson GS 1988 Sources fates and impactsof nitrogen inputs to terrestrial ecosystems review and synthesis Biogeochemistry6119ndash159 DOI 101007BF00003034

Carson HL Clague DA 1995 Geology and biogeography of the Hawaiian IslandsIn Wagner WL Funk VA eds Hawaiian biogeography evolution on a hot spotarchipelago Washington DC Smithsonian Institution Press 14ndash29

Carson HL Lockwood JP Craddock EM 1990 Extinction and recolonization of localpopulations on a growing shield volcano Proceedings of the National Academy ofSciences of the United States of America 877055ndash7057 DOI 101073pnas87187055

Caut S Angulo E Courchamp F 2009 Variation in discrimination factors (115N and113C) the effect of diet isotopic values and applications for diet reconstructionJournal of Applied Ecology 46443ndash453 DOI 101111j1365-2664200901620x

Chamberlain CP Blum JD Holmes RT Feng X Sherry TW Graves GR 1997 Theuse of isotope tracers for identifying populations of migratory birds Oecologia109132ndash141 DOI 101007s004420050067

Crews TE Kitayama K Fownes JH Riley RH Herbert DA Mueller-DomboisD Vitousek PM 1995 Changes in soil phosphorus fractions and ecosystemdynamics across a long chronosequence in Hawaii Ecology 761407ndash1424DOI 1023071938144

Dawson TE Mambelli S Plamboeck AH Templer PH Tu KP 2002 Stable iso-topes in plant ecology Annual Review of Ecology and Systematics 33507ndash559DOI 101146annurevecolsys33020602095451

De Vries MS Del Rio CM Tunstall TS Dawson TE 2015 Isotopic incorporation ratesand discrimination factors in mantis shrimp crustaceans PLOS ONE 10e0122334DOI 101371journalpone0122334

Doblas-Miranda EWardle DA Peltzer DA Yeates GW 2008 Changes in thecommunity structure and diversity of soil invertebrates across the FranzJosef Glacier chronosequence Soil Biology and Biochemistry 401069ndash1081DOI 101016jsoilbio200711026

Elser JJ Acharya K Kyle M Cotner J MakinoWMarkow TWatts T Hobbie S FaganW Schade J Hood J Sterner RW 2003 Growth ratemdashstoichiometry couplings indiverse biota Ecology Letters 6936ndash943 DOI 101046j1461-0248200300518x

Kennedy et al (2018) PeerJ DOI 107717peerj4527 1518

Fry B 1988 Food web structure on Georges Bank from stable C N and S isotopiccompositions Limnology and Oceanography 331182ndash1190DOI 104319lo19883351182

Giambelluca TW Shuai X Barnes ML Alliss RJ Longman RJ Miura T Chen QFrazier AG Mudd RG Cuo L Businger AD 2014 Evapotranspiration of HawailsquoiFinal report submitted to the US Army Corps of EngineersmdashHonolulu District andCommission on Water Resource Management State of Hawailsquoi

Gillespie RG 1991Hawaiian spiders of the genus Tetragnatha I Spiny leg clade Journalof Arachnology 19174ndash209

Gillespie RG 2002Hawaiian Spiders of the Genus Tetragnatha IV new small species inthe spiny leg clade Journal of Arachnology 30159ndash172DOI 1016360161-8202(2002)030[0159HSOTGT]20CO2

Gillespie RG 2004 Community assembly through adaptive radiation in Hawaiianspiders Science 303356ndash359 DOI 101126science1091875

Gillespie RG 2015 Island time and the interplay between ecology and evolution inspecies diversification Evolutionary Applications 953ndash73 DOI 101111eva12302

Gillespie RG Baldwin BG 2010 Island biogeography of remote archipelagoes In LososJB Ricklefs RE eds The theory of island biogeography revisited Princeton PrincetonUniversity Press 358ndash387

Gillespie RG Benjamin SP Brewer MS Rivera MAJ Roderick GK 2018 Repeateddiversification of ecomorphs in Hawaiian stick spiders Current Biology 281ndash7DOI 101016jcub201801083

Gillespie R Rivera M 2007 Free-living spiders of the genus Ariamnes (AraneaeTheridiidae) in Hawaii Journal of Arachnology 3511ndash37 DOI 101636H04-051

Gruner DS 2007 Geological age ecosystem development and local resource constraintson arthropod community structure in the Hawaiian Islands Biological Journal of theLinnean Society 90551ndash570 DOI 101111j1095-8312200700748x

Hobson KA Barnett-Johnson R Cerling T 2010 Using isoscapes to track animalmigration In West JB et al eds Isoscapes understanding movement pattern andprocess on Earth through isotope mapping Dordrecht Springer 273ndash298

Hobson KAWelch HE 1992 Determination of trophic relationships within a highArctic marine food web using delta-13-C and delta-15-N analysisMarine EcologyProgress Series 849ndash18 DOI 103354meps084009

Iacumin P Davanzo S Nikolaev V 2005 Short-term climatic changes recorded bymammoth hair in the Arctic environment Palaeogeography PalaeoclimatologyPalaeoecology 218317ndash324 DOI 101016jpalaeo200412021

Kelly SP Cuevas E Ramiacuterez A 2015 Stable isotope analyses of web-spinningspider assemblages along a headwater stream in Puerto Rico PeerJ 3e1324DOI 107717peerj1324

Krehenwinkel H Kennedy SR Pekaacuter S Gillespie RG 2017 A cost-efficient and simpleprotocol to enrich prey DNA from extractions of predatory arthropods for large-scale gut content analysis by Illumina sequencingMethods in Ecology and Evolution8126ndash134 DOI 1011112041-210X12647

Kennedy et al (2018) PeerJ DOI 107717peerj4527 1618

Laiolo P Illera JC Meleacutendez L Segura A Obeso JR 2015 Abiotic biotic and evolu-tionary control of the distribution of C and N isotopes in food webs The AmericanNaturalis 185169ndash182 DOI 101086679348

Lim JY Marshall CR 2017 The true tempo of evolutionary radiation and decline re-vealed on the Hawaiian archipelago Nature 543710ndash713 DOI 101038nature21675

Lubin YD EberhardWGMontgomery GG 1978Webs ofMiagrammopes (AraneaeUloboridae) in the Neotropics Psyche 851ndash23 DOI 101155197872579

Matthews B Narwani A Hausch S Nonaka E Peter H Yamamichi M Sullam KEBird KC ThomasMK Hanley TC Turner CB 2011 Toward an integrationof evolutionary biology and ecosystem science Ecology Letters 14690ndash701DOI 101111j1461-0248201101627x

McMahon KW Thorrold SR Houghton LA BerumenML 2016 Tracing carbon flowthrough coral reef food webs using a compound-specific stable isotope approachOecologia 180809ndash821 DOI 101007s00442-015-3475-3

MuschickM Indermaur A SalzburgerW 2012 Convergent evolution within anadaptive radiation of cichlid fishes Current Biology 222362ndash2368DOI 101016jcub201210048

Post DM 2002 Using stable isotopes to estimate trophic position models methods andassumptions Ecology 83703ndash718DOI 1018900012-9658(2002)083[0703USITET]20CO2

Roderick GK Croucher PJP Vandergast AG Gillespie RG 2012 Species differen-tiation on a dynamic landscape shifts in metapopulation genetic structure usingthe chronology of the Hawaiian archipelago Evolutionary Biology 39192ndash206DOI 101007s11692-012-9184-5

Rominger AJ Goodman KR Lim JY Armstrong EE Becking LE Bennett GM BrewerMS Cotoras DD Ewing CP Harte J Martinez ND OrsquoGrady PM Percy DMPrice DK Roderick GK Shaw KL Valdovinos FS Gruner DS Gillespie RG 2016Community assembly on isolated islands macroecology meets evolution GlobalEcology and Biogeography 25769ndash780 DOI 101111geb12341

Sanders D Vogel E Knop E 2015 Individual and species-specific traits explain nichesize and functional role in spiders as generalist predators Journal of Animal Ecology84134ndash142 DOI 1011111365-265612271

Sherrod DR Sinton JMWatkins SE Brunt KM 2007 Geologic map of the state ofHawaii US geological survey open-file report 2007ndash1089 Available at http pubsusgsgovof 20071089

Simberloff DSWilson EO 1970 Experimental zoogeography of islands A two-yearrecord of colonization Ecology 51934ndash937

Tukey JW 1949 Comparing individual means in the analysis of variance Biometrics599ndash114 DOI 1023073001913

Vitousek PM 2004Nutrient cycling and limitation Hawairsquoi as a model system PrincetonPrinceton University Press

Kennedy et al (2018) PeerJ DOI 107717peerj4527 1718

Vitousek PM Asner GP Chadwick OA Hotchkiss S 2009 Landscape-level variationin forest structure and biogeochemistry across a substrate age gradient in HawaiiEcology 903074ndash3086 DOI 10189008-08131

Vitousek PM Chadwick OA Crews T Fownes J Hendricks DM Herbert D 1997 Soiland ecosystem development across the Hawaiian Islands GSA Today 71ndash9

Vitousek PM Shearer G Kohl DH 1989 Foliar N-15 natural abundance in Hawai-ian rainforestmdashpatterns and possible mechanisms Oecologia 78383ndash388DOI 101007BF00379113

Vitousek PM Turner DR Kitayama K 1995 Foliar nutrients during long-termsoil development in Hawaiian montane rain forest Ecology 76712ndash720DOI 1023071939338

Walker LRWardle D Bardgett RD Clarkson BD 2010 The use of chronosequences instudies of ecological succession and soil development Journal of Ecology 98725ndash736DOI 101111j1365-2745201001664x

Wardle DA Bardgett RDWalker LR Peltzer DA Lagerstroumlm A 2008 The responseof plant diversity to ecosystem retrogression evidence from contrasting long-termchronosequences Oikos 11793ndash103 DOI 101111j20070030-129916130x

Wardle DAWalker LR Bardgett RD 2004 Ecosystem properties and forestdecline in contrasting long-term chronosequences Science 305509ndash513DOI 101126science1098778

Whittaker RJ Triantis KA Ladle RJ 2008 A general dynamic theory of oceanic islandbiogeography Journal of Biogeography 35977ndash994DOI 101111j1365-2699200801892x

Williams FX 1942 Biological studies in Hawaiian water-loving insectsmdashPart IIImdashDiptera or fliesmdashC Tipulidae and Psychodidae Proceedings of the HawaiianEntomological Society 11313ndash338

WilliamsonWMWardle DA Yeates GW 2005 Changes in soil microbial and nema-tode communities during ecosystem decline across a long-term chronosequence SoilBiology and Biochemistry 371289ndash1301 DOI 101016jsoilbio200411025

Kennedy et al (2018) PeerJ DOI 107717peerj4527 1818