the ecological consequences and … comparative study across asclepias series incarnatae ... 72...
TRANSCRIPT
THE ECOLOGICAL CONSEQUENCES AND ADAPTIVE FUNCTION
OF NECTAR SECONDARY METABOLITES
By
Jessamyn Sara Manson
A thesis submitted in conformity with the requirements
for the degree of Doctor of Philosophy
Graduate Department of Ecology and Evolutionary Biology
University of Toronto
© Copyright by Jessamyn Sara Manson, 2009
ii
The Ecological Consequences and Adaptive Function of Nectar Secondary Metabolites
Jessamyn Sara Manson
Doctor of Philosophy
Department of Ecology and Evolutionary Biology
University of Toronto
2009
ABSTRACT
Plants are under selection to simultaneously attract pollinators while deterring
herbivores. This dilemma can lead to tradeoffs in floral traits, which are traditionally
thought to be optimized for pollinators. My dissertation addresses the ecological costs
and putative functional significance of nectar secondary metabolites, a paradoxical but
widespread phenomenon in the angiosperms. I investigate this issue from the pollinator’s
perspective using a series of controlled laboratory investigations focused primarily on the
bumble bee Bombus impatiens and the nectar alkaloid gelsemine, from Gelsemium
sempervirens. I begin by demonstrating that nectar enriched with the alkaloid gelsemine
significantly deters visits from bumble bees at a range of natural alkaloid concentrations.
However, this aversion can be mitigated by increasing the sucrose concentration such
that the alkaloid-rich nectar is more rewarding than its alkaloid-free counterpart. I then
demonstrate that the consumption of gelsemine-rich nectar can inhibit oocyte
development and protein utilization in bees, but that this effect is limited to bees of
suboptimal condition. Continuous consumption of the nectar alkaloid gelsemine also
leads to a reduction in the pathogen load of bumble bees infected with Crithidia bombi,
iii
but direct interactions between the pathogen and the alkaloid have no impact on infection
intensity. Gelsemine also fails to inhibit floral yeast growth, suggesting that nectar
alkaloids may not be universally antimicrobial. Finally, I demonstrate that gross nectar
cardenolides from the genus Asclepias are strongly correlated with gross leaf
cardenolides and that the majority of individual cardenolides found in nectar are a subset
of those identified in leaves. This pattern suggests that nectar cardenolides are a
consequence of defense for Asclepias; however, they may not be a costly corollary
because bumble bees show an overall preference for nectar cardenolides at mean
concentrations. Altogether, my dissertation provides a new perspective on the role of
chemical defenses against herbivores in plant-pollinator interactions.
iv
ACKNOWLEDGEMENTS
Writing a dissertation is like raising a child – it takes a village! First of all, I’d
like to thank my supervisor, James Thomson, who willingly welcomed this stray graduate
student back in 2004. James understood that my interests lay in the field of chemical
ecology and encouraged me to pursue research questions beyond the expertise of the lab.
I am extraordinarily grateful to James for letting me take so many risks and teaching me
to be stubbornly independent. I would also like to thank to Spencer Barrett and Peter
Kotanen, the intrepid members of my supervisory committee, for their discussion and
advice during my doctorate. Because of the integrative nature of my work, I have had a
lot of help over the years. A huge thank you to three of my collaborators, Marc-André
Lachance, Mario Vallejo-Marìn and Anurag Agrawal, who have allowed my research to
grow in exciting new directions. I am very appreciative of all the time and energy that
others have invested so that I could complete my chemical analyses; thank you to Sergio
Rasmann, Rayko Halitchke, John Arnason and especially to Ammar Saleem, who spent
many hours at the HPLC with me. Thanks to Bruce Hall and Andrew Petrie for watering
my army of useless plants every time I was out of town. And thanks to Lynn Adler and
Rebecca Irwin for suggesting that I work on a plant that shares my name!
The members of the Thomson lab have been a wonderful source of guidance,
support and collaboration. Thank you to Robert Gegear, without whom I would not
know the joys of the flight cage, to Michael Otterstatter, who rekindled my passion for
pathogens and to James Burns, Jessica Forrest, Nathan Muchhala, Jane Ogilvie and
Alison Parker for being great friends and mentors. I have forged strong bonds with many
v
members of the graduate community. I’d especially like to recognize Kate Edwards,
Heather Coiner, Danielle Way, Danielle Marcos, Marc Johnson, Patrick Vogan, Sarah
Yakimowski, Jannice Friedman, Brechann McGoey, Anna Simonsen and Brandon
Campitelli for their friendship, humour and compassion throughout the course of my
degree.
I have also had an immense about of support from my family, particularly my
parents. Thank you mom for teaching me to overcome obstacles with a smile on my
face, thank you pa for instilling in me such a strong work ethic, and thank you dad for
always telling me to go outside! Finally, this would not have been possible without the
love and encouragement of my husband, Chris; your strength and generosity amaze me!
I would also like to acknowledge several presses for permitting me to include
previously published work in my thesis. Chapter two (R.J. Gegear, J.S. Manson and J.D.
Thomson. 2007. Ecological context influences pollinator deterrence by alkaloids in floral
nectar. Ecology Letters 10: 375-382) and chapter three (J.S. Manson and J.D. Thomson.
2009. Post-ingestive effects of nectar alkaloids depend on dominance status of bumble
bees. Ecological Entomology 34: 421-426) were reproduced with permission from
Wiley-Blackwell Publishing. Chapter four (J.S. Manson, M.C. Otterstatter and J.D.
Thomson. In Press. Consumption of a nectar alkaloid reduces pathogen load in bumble
bees. Oecologia DOI: 10.1007/s00442-009-1431-9) and appendix one (J.S. Manson,
M.A. Lachance and J.D. Thomson. 2007. Candida gelsemii sp. nov., a yeast of the
Metschnikowiaceae clade isolated from nectar of the poisonous Carolina Jessamine.
Antonie von Leeuwenhoeck 92: 37-42) were reproduced with permission from Springer
Science and Business Media.
vi
TABLE OF CONTENTS
ABSTRACT………………………………………………………………………...ii
ACKNOWLEDGEMENTS………………………………………………………... iv
TABLE OF CONTENTS…………………………………………………………... vi
LIST OF TABLES…………………………………………………………………. x
LIST OF FIGURES………………………………………………………………... xi
LIST OF APPENDICES…………………………………………………………... xiii
CHAPTER ONE – Introduction…………………………………………………… 1
A brief history of nectar secondary metabolites…………………………… 2
A spotlight on alkaloids……………………………………………………. 6
Principal experimental systems……………………………………………. 7
How toxic is “toxic” nectar? Addressing the ecological consequences
and adaptive functions of nectar secondary metabolites……………………9
CHAPTER TWO – Ecological context influences pollinator deterrence by
alkaloids in floral nectar……………………………………………………………12
Abstract…………………………………………………………………….. 12
Introduction………………………………………………………………… 13
Methods……………………………………………………………………. 16
Bees and flowers…………………………………………………………… 16
Experimental procedure…………………………………………………….17
Data analysis………………………………………………………………. 18
Results……………………………………………………………………... 20
Flower preference………………………………………………………….. 20
vii
Foraging proficiency………………………………………………………. 21
Discussion………………………………………………………………….. 21
Acknowledgements………………………………………………………… 26
CHAPTER THREE – Post-ingestive effects of nectar alkaloids depend on
dominance status of bumble bees………………………………………………….. 32
Abstract……………………………………………………………………. 32
Introduction………………………………………………………………… 33
Methods……………………………………………………………………. 36
Oocyte development………………………………………………………... 36
Haemolymph carbohydrates……………………………………………….. 39
Results……………………………………………………………………... 39
Protein metabolism………………………………………………………… 39
Carbohydrate concentrations……………………………………………… 40
Discussion………………………………………………………………….. 40
Acknowledgements………………………………………………………… 45
CHAPTER FOUR – Consumption of a nectar alkaloid reduces pathogen load
in bumble bees……………………………………………………………………... 50
Abstract…………………………………………………………………….. 50
Introduction………………………………………………………………… 51
Methods……………………………………………………………………. 54
Statistical analysis…………………………………………………………. 57
Results……………………………………………………………………... 58
Discussion…………………………………………………………………. 60
viii
Acknowledgements………………………………………………………… 65
CHAPTER FIVE – Cardenolide concentrations of nectar, leaves and flowers:
A comparative study across Asclepias series Incarnatae…………………………... 72
Abstract…………………………………………………………………….. 72
Introduction………………………………………………………………… 73
Methods……………………………………………………………………. 77
Study system………………………………………………………………... 77
Quantifying cardenolides…………………………………………………... 77
Pollination biology………………………………………………………….80
Statistical analysis…………………………………………………………. 83
Quantitative cardenolide analysis…………………………………………. 83
Qualitative cardenolide analysis…………………………………………... 84
Behaviour analysis…………………………………………………………. 86
Results……………………………………………………………………....87
Quantitative cardenolide analysis…………………………………………. 87
Qualitative cardenolide analysis…………………………………………... 88
Behaviour analysis…………………………………………………………. 90
Discussion………………………………………………………………….. 92
Acknowledgements………………………………………………………… 101
CHAPTER SIX – CONCLUDING DISCUSSION………………………………... 108
Ecological context is crucial……………………………………………….. 109
A subtle effect is still an effect…………………………………………….. 110
There is no such thing as a general adaptive hypothesis…………………... 111
ix
Consequence of defense is often assumed but rarely tested……………….. 112
Costs and benefits are not always obvious………………………………… 114
REFERENCES.……………………………………………………………………. 116
x
LIST OF TABLES
Table 2.1. Description of reward conditions for each behavioural assay…….……. 27
Table 2.2. Results of t-tests evaluating pollinator preference…………………........ 28
Table 2.3. Generalized linear model results for foraging proficiency……………... 29
Table 3.1. The effect of gelsemine on oocyte length and width …………………... 46
Table 3.2. Daily pollen consumption in microcolonies……………………………. 47
Table 4.1. Mixed model statistics describing the effects of continuous gelsemine
consumption on Crithidia bombi infections……………………………….. 66
Table 4.2. Mixed model statistics describing the direct effects of gelsemine on Crithidia
bombi infections……………………………………………………………. 67
Table A.1. Summary of yeasts recovered from the nectar of flowers in
Statesboro, Georgia……………………………………………………….... 152
Table A.2. Growth responses of Candida gelsemii that exhibit variation…………..153
xi
LIST OF FIGURES
Figure 2.1. Choice behaviour of individual bees over 80 consecutive visits………. 30
Figure 3.1. Mean oocyte size in dominant and subordinate bees at three
concentrations of gelsemine-rich nectar…………………………………… 48
Figure 3.2. Mean carbohydrate concentrations in bee haemeolymph 24 hours
after consuming gelsemine-rich nectar at three concentrations…………… 49
Figure 4.1. Diagram of experimental design for Crithidia bombi assays …………. 68
Figure 4.2. Effect of continuous alkaloid consumption on Crithidia bombi
infections in bumble bees………………………………………………….. 69
Figure 4.3. Relationship between Crithidia bombi infection intensity and
bumble bee body size………………………………………………………. 70
Figure 4.4. Effect of exposing Crithidia bombi cells to alkaloids prior to
bumble bee inoculation…………………………………………………….. 71
Figure 5.1. Average total cardenolide concentrations for the nectar, leaves and
flowers of twelve species of Asclepias series Incarnatae…………………...102
Figure 5.2. Correlations between total cardenolide concentrations in Asclepias
nectar and leaves …………………………………………………………... 103
Figure 5.3. NMDS two-dimensional ordination of individual cardenolide
concentrations of Asclepias nectar and leaves…………………………….. 104
Figure 5.4. Correlation between intensity of Crithidia bombi infections in bees
and their preference for blue flowers………………………………………. 105
xii
Figure 5.5. Boxplots of average visit length and mean foraging rate by bees to
flowers varying in nectar cardenolide concentration………………………. 106
Figure 6.1. Path diagrams summarizing interactions between secondary
metabolites, pollinators, herbivores and plant fitness……………………… 115
Figure A.1. Phylogram of Candida gelsemii and its closest relatives……………... 154
Figure A.2. Differentiation of Candida gelsemii into “pulcherrima” cells…………155
xiii
LIST OF APPENDICES
APPENDIX ONE – Candida gelsemii sp. nov., a yeast of the Metschnikowiaceae
clade isolated from nectar of the poisonous Carolina jessamine…………………... 144
Abstract…………………………………………………………………….. 144
Introduction………………………………………………………………… 145
Methods……………………………………………………………………. 146
Results and Discussion…………………………………………………...... 147
Species delineation, phylogenetic placement and phenotypic
variability………………………………………………………………….. 147
Ecology…………………………………………………………………...... 148
Description of Candida gelsemii Lachance sp. nov………………………... 150
Acknowledgements………………………………………………………… 151
APPENDIX TWO – Concentrations of cardenolides from the nectar, leaves and
flowers of twelve species of Asclepias series Incarnatae………………………...... 156
APPENDIX THREE – Raw visit data from behavioural assays testing the effect
of nectar cardenolides on bumble bee preference………………………………….. 160
1
CHAPTER ONE
Introduction
For outcrossing flowering plants, maximizing fitness frequently entails attracting
effective pollinators while deterring herbivores. Traits that are critical to fulfilling these
two mandates have been studied largely in isolation, but there is a growing body of
evidence indicating that many of these characteristics are intrinsically linked and
therefore under selection from both pollinators and herbivores (Armbruster 1997, Strauss
1997, Strauss et al. 1999, Strauss and Whittall 2006, Bronstein et al. 2007). In many
cases, selection by these two groups is antagonistic, forcing adaptive compromises (sensu
Strauss and Whittall 2006) between reproduction and defense.
Tradeoffs to accommodate pollination and herbivory are particularly evident
when they affect floral traits. Several studies have shown shifts in flower shape (Gomez
2003), size (Ashman et al. 2004) and flowering phenology (Pilson 2000) in response to
selection from herbivores that reduce the frequency, efficiency or likelihood of
pollination. In plants such as Raphanus sativus and Ipomoea purpurea, flower petal
colour is directly associated with the production of anthocyanins, a secondary compound
that chemically defends plants from herbivory (Fineblum and Rausher 1997, Irwin et al.
2003). Although tortoise beetle larvae have lower survivorship on the anthocyanin-rich
purple flowers of I. purpurea, the rarer white (anthocyanin-free) morph exports more
pollen per capita, but has lower visitation rates than its purple counterpart (Simms and
Bucher 1996). In the case of I. purpurea, being chemically defended does not
unequivocally lead to a cost in terms of pollination services. However, the effects of
2
secondary metabolites on pollination services may be more insidious; compounds such as
alkaloids, phenolics and glycosides have been detected in floral nectar (Adler 2000).
Could this so-called “toxic” nectar represent another tradeoff between pollinator
attraction and herbivore defense?
My dissertation investigates the ecological consequences of nectar secondary
metabolites. More specifically, I examine the effects of nectar alkaloids and nectar
cardenolides on bumble bee behaviour and physiology. I also research adaptive functions
for nectar alkaloids, focusing on their putative antimicrobial properties. Finally, I
investigate whether cardenolides in milkweed nectar are likely to have particular adaptive
roles in nectar, or are present systemically as a byproduct of the chemical defense of
foliage against herbivores. Taken together, my research provides novel insight into the
impact of herbivory on plant-pollinator interactions from the pollinator’s perspective and
is an important contribution to the field of plant-animal interactions as a whole.
A brief history of nectar secondary metabolites
Floral nectar is produced by plants to reward pollinators and has two principal
ingredients, water and sugar. In the mid-twentieth century, however, biologists began
finding that nectar was not just a simple syrup and that it contained a myriad of other
components in small concentrations (Baker and Baker 1983). Some of these additional
nectar constituents may increase the value of floral nectar as a reward. Amino acids,
which are found nearly universally in nectar (Baker and Baker 1973, Baker 1977, Baker
and Baker 1983), can make nectar more palatable for pollinators (Kim and Smith 2000)
and are particularly important to adult lepidoptera that have no other dietary source for
protein-building material (Baker and Baker 1973). Nectar may also contain lipids (Baker
3
and Baker 1975, Baker 1977), proteins (Baker and Baker 1975, Thornburg et al. 2003)
and antioxidants (Baker and Baker 1973), all of which may help to attract and retain
pollinators. However, many plants have nectar constituents that are unpalatable,
repellant, and sometimes poisonous to floral visitors. One of the most frequently
observed “toxic” nectar components are secondary metabolites, compounds that are
produced to defend plants against herbivores and include alkaloids, phenolics, tannins
and glycosides.
The first comprehensive work on nectar secondary metabolites was completed by
Baker and Baker in the 1970s. The Bakers adapted a series of simple colourimetric
assays to test for the presence of alkaloids (Baker and Baker 1975) and phenolics (Baker
1977) in the field. The surveys that they performed remain unrivalled in their breadth,
characterizing the nectar chemistry of over 1000 plant species in environments as diverse
as California, Costa Rica (Baker and Baker 1975) and the Rocky Mountains of Colorado
(Baker and Baker 1982). These studies detected nectar alkaloids in 0-12% and nectar
phenolics in 30-49% of samples, depending on the latitude or altitude of the site. Nectar
alkaloids and phenolics were most prevalent in the Costa Rican lowlands, and nectar
alkaloids were entirely absent from the alpine tundra of Colorado (Baker and Baker
1982).
The Bakers’ work on nectar constituents was motivated by their interest in plant-
pollinator coevolution, and they made several predictions regarding the role of nectar
secondary metabolites in plant-animal interactions. They were particularly intrigued by
the observation that nectar alkaloids were absent from plants pollinated by lepidoptera
but present in plants pollinated by bees (Baker and Baker 1975). The Bakers reasoned
4
that nectar alkaloids may deter lepidoptera, which are inconstant pollinators, while bees,
which exhibit strong floral constancy, are somehow resistant to the compounds; this
rationale became the basis for the pollinator fidelity hypothesis (see below). A few years
later, Rhoades and Bergdahl (1981) built upon this idea, suggesting that nectar secondary
metabolites are another means by which plants encourage specialist pollinators and
discourage ineffective floral visitors. While neither the Bakers nor Rhoades and Bergdahl
followed up their hypotheses for the function of “toxic” nectar with empirical evidence,
other researchers began testing their ideas. Stephenson, for example, found that nectar
robbers were deterred by iridoid glycosides while valid pollinators were unaffected
(1981, 1982). Masters determined that specialists will visit nectar with pyrrolizidine
alkaloids but generalists will not (1991); however, his results have been contested
(Landolt and Lenczewski 1993).
In 2000, Adler published the authoritative review on “toxic” nectar in which she
summarized all substantiated accounts of nectar secondary metabolites. Adler collected
33 reports of nectar that was toxic or aversive for animals, nectar where secondary
metabolites had been isolated and nectar with both identified compounds and deleterious
effects. The studies, which date back to 1933, find evidence for “toxic” nectar in at least
21 angiosperm families and specify the presence of alkaloids, phenolics, iridoid
glycosides and non-protein amino acids, along with some assorted rare compounds.
Animals affected by the noxious nectar include honey bees, ants, butterflies and humans,
while the responses range from undetectable to lethal. However, the greatest strength of
Adler’s review is her synopsis of the hypotheses for the functional significance of this
strange but widespread phenomenon.
5
The most compelling aspect of the nectar secondary metabolite question is the
why: Why does nectar have toxic constituents? To date, five potential reasons have been
proposed, all of which were succinctly summarized by Adler and which I will touch on
only briefly here. The pollinator fidelity hypothesis, proposed by Baker and Baker
(1975) but fleshed out by Rhoades and Bergdahl (1981), stipulates that nectar secondary
metabolites may encourage specialist pollinators while discouraging generalists by
promoting secondary metabolite tolerance in more frequent visitors. The nectar robber
hypothesis was suggested by both Janzen (1977) and Baker (1978) and states that
chemical defenses in nectar could defend plants from floral visitors that take resources
but do not pollinate. Hagler and Buchmann (1993) were the first to hypothesize that
nectar secondary metabolites could have an antimicrobial function, thereby keeping
nectar palatable for pollinators. The drunken pollinator hypothesis, from Ehlers and
Olesen (1997), is a tangential proposal suggesting that the ethanol produced from floral
yeasts can improve pollen transfer because intoxicated pollinators spend less time
grooming. This hypothesis can be expanded to include a “nectar as narcotic” hypothesis,
in which consuming very small doses of secondary metabolites such as caffeine and
nicotine could cause both intoxication and addiction (Singaravelan et al. 2005). Finally,
Adler herself (2000) suggests that secondary metabolites in nectar occur as a
consequence of a plant’s chemical defense system. This last hypothesis implies a non-
adaptive and potentially deleterious function for “toxic” nectar, but is not mutually
exclusive from the previous hypotheses. In fact, it is likely that nectar secondary
metabolites originated as a sort of chemical overflow, but may have gained new, nectar-
specific functions through “exaptation” (Armbruster 1997).
6
Since Adler’s review, interest in the field of “toxic” nectar, and the implications
of chemical defenses on plant mutualisms in general, has increased dramatically. Nectar
secondary metabolites have been identified in ten new plants, including almonds
(London-Shafir et al. 2003) and avocados (Afik et al. 2006). Work on the impact of
putatively “toxic” nectar on pollinators has expanded to include effects on solitary bees
(Elliott et al. 2008) and birds (Tadmor-Melamed et al. 2004, Johnson et al. 2006). Those
focusing on the honey bee are now looking at dose-dependent effects of nectar secondary
metabolites (Singaravelan et al. 2005) and mechanisms for tolerating toxic compounds
(Liu et al. 2005). Studies are addressing the hypothesized adaptive functions of noxious
compounds in nectar; there is now evidence from one system to suggest that nectar
alkaloids are ineffective at deterring nectar robbers (Adler and Irwin 2005), while recent
work demonstrating how floral yeasts degrade nectar quality (Herrera et al. 2008) has
prompted interest in exploring the antimicrobial properties of nectar secondary
metabolites. In addition, researchers are considering the role of secondary metabolites as
components (either directly or indirectly) of other floral traits like pollen (Praz et al.
2008) and floral scent (Raguso 2008). Perhaps most importantly, authors no longer
discount the influence of herbivory and herbivore protection on nectar, nectarivorous
animals and plant-pollinator interactions as a whole.
A spotlight on alkaloids
Alkaloids, which represent the largest and most diverse group of secondary
metabolites, are loosely defined by a cyclic structure and the presence of nitrogen in a
negative oxidation state (Hartmann 1992). Infamously toxic to vertebrates and
invertebrates (think strychnine and atropine), alkaloids can affect neuroreceptors,
7
structurally disrupt DNA, induce apoptosis or inhibit critical enzymes (Wink and
Schimmer 1999). However, alkaloids also have favorable properties at appropriate doses
and can be antimicrobial (e.g. quinine; Cowan 1999), therapeutic (e.g., morphine), or can
act as stimulants (e.g., caffeine). Alkaloids are found in 20% of flowering plants and are
considered a highly effective constitutive defense against herbivores (Hartmann 1992).
The first circumstantial evidence for alkaloids in nectar came from a study on
Sophora microphylla in 1972 (Clinch et al.). Baker and Baker (1983) subsequently
detected nectar alkaloids in 8% of the species they tested (75 of 910 species). Attention
quickly turned to the effects of nectar alkaloids on flower visitors; studies suggest that
responses vary depending on alkaloid concentration and type of visitor. Clinch et al.
(1972) found that honey bees died as a result of ingesting as little as 10 µL of S.
micorphylla nectar, while Singaravalen (2005) suggested that low concentrations of
nicotine and caffeine may cause addiction to nectar in honey bees. Nectarivorous birds
that consume nectar rich in nicotine are less able to assimilate sucrose (Tadmor-Melamed
et al. 2004), while specialist butterflies may (Masters 1991) or may not (Landolt and
Lenczewski 1993) find nectar alkaloids unattractive. Finally, high concentrations of
nectar in alkaloids reduce the transfer rate of a pollen analogue (Adler and Irwin 2005),
suggesting that nectar alkaloids can reduce plant fitness.
Principal experimental systems
The majority of my thesis focuses on the relationship between a single plant,
Gelsemium sempervirens, and one of its key pollinators, Bombus impatiens. When my
research questions required a plant system amenable to comparative studies, I had to look
8
beyond the tiny Gelsemium genus to the larger, more diverse genus Asclepias. Here I
describe the relevant natural history of my two principal experimental systems.
Gelsemium sempervirens, the Carolina jessamine, is a high-climbing or trailing
woody vine distributed throughout the southeastern United States and into Central
America (Radford et al. 1968, Ornduff 1970). The genus Gelsemium, part of the
Loganiaceae, contains only three species, two of which are native to North America,
while the third, G. elegans is found in Asia (Ornduff 1970, Wyatt et al. 1993). G.
sempervirens grows abundantly on the margins of tilled fields, along fence lines and in
ditches along the road. It flowers from mid-January until early April, depending on
longitude (Leege and Wolfe 2002, Pascarella 2007), but does not produce mature seeds
until late fall. G. sempervirens flowers are 1.7 – 3.3 cm in length, vivid yellow, tubular
and very fragrant. The flowers are distylous, with familiar pin and thrum morphology,
and are largely self-incompatible (Ornduff 1970) . G. sempervirens is pollinated by the
solitary bees Osmia lignaria (orchard mason bee) and Habropoda laboriosa (blueberry
bee), the social bees Apis mellifera (honey bee), Bombus impatiens and Bombus
bimaculatus (bumble bees) and the nectar robber Xylocopa virginica (carpenter bee)
(Ornduff 1970, Adler and Irwin 2005, Pascarella 2007). Floral visitors differ in their
pollen transfer ability, with Bombus spp. being among the most efficient (Adler and Irwin
2006).
Gelsemium sempervirens is heavily defended from herbivory by indole alkaloids,
with the plant’s roots and floral nectar being particularly noxious (Hardin and Arena
1969, Burrow and Tyrl 2001). The most abundant alkaloid, gelsemine, is found in high
concentrations in flowers (Blaw et al. 1979) and is the principal alkaloid detected in floral
9
nectar (Adler and Irwin 2005), with gelsemine concentrations in nectar ranging from 5.8
to 246.1 ng/µL (Adler and Irwin 2005; Manson, unpublished data). G. sempervirens is
highly toxic to mammals: the ingestion of five flowers rendered a young child semi-
comatose (Blaw et al. 1979), while livestock and chickens are also reportedly at risk
(Burrow and Tyrl 2001). The plant’s effect on invertebrates is less clear, with reports of
G. sempervirens nectar poisoning honey bees being unsubstantiated (Hardin and Arena
1969). However, G. sempervirens nectar artificially enriched with gelsemine did reduce
the length and number of flower visits made by several pollinators, including bumble
bees (Adler and Irwin 2005).
The bumble bee Bombus impatiens is native to eastern North America, with a
range from Ontario to Maine and south to Florida (Kearns and Thomson 2001). B.
impatiens has a moderate tongue length (5.8 +/- 0.8 mm; Harder and Barrett 1993), which
allows it to be a generalist pollinator on a large number of plants, including several
species known to have nectar secondary metabolites, such as milkweeds, rhododendrons
and Gelsemium sempervirens (Manson, personal observation). The foraging habits of
bumble bees, including B. impatiens, have been well-documented, and bumble bees have
been recognized as an excellent model system for behavioural assays (Heinrich 1979,
Real 1991, Chittka et al. 1999, Chittka and Thomson 2001). In addition, B. impatiens is
commercially available as a pollinator of greenhouse crops.
How toxic is “toxic” nectar? Addressing the ecological consequences and adaptive
functions of nectar secondary metabolites
My dissertation contributes to a growing body of literature addressing the
ecological consequences of chemical defense on plant-insect mutualisms. I have focused
10
primarily on the effects of a single nectar secondary metabolite, gelsemine, on a single
pollinator, Bombus impatiens. I broaden my scope to nectar cardenolides in the genus
Asclepias, adding a phylogenetic framework to a larger comparative study. My work
addresses many of the key questions raised by my peers and represents the most
comprehensive treatment of the impact of nectar secondary metabolites on bumble bees
to date.
Previous studies have reported that nectar alkaloids can be aversive, however a
rigorous examination of their impact on pollinator behaviour was lacking. In chapter
two, I describe the first laboratory study to address the effects of the nectar alkaloid
gelsemine on bumble bee behaviour under a range of ecologically relevant nectar
conditions. I trained workers to forage on artificial flower arrays with distinctly coloured
flowers that contrasted in alkaloid or sucrose concentration, or both, and examined the
effects of gelsemine on preference, foraging rate and flower handling time. In this
controlled environment, I also evaluated whether pollinators learn to avoid nectar
alkaloids over time.
In chapter three, I explore whether bees experience a physiological cost due to
nectar alkaloid consumption. This study, which took advantage of a unique
developmental process to induce queen-like behaviour in worker bees, focused on how
the constant consumption of gelsemine-rich nectar affected oocyte development and
carbohydrate metabolism in bees. These two traits relate to protein utilization and
energetics, respectively, and are therefore good metrics for sublethal costs of nectar
alkaloid ingestion. This is one of only a handful of studies dealing with post-
consumptive effects of nectar secondary metabolites.
11
One of the most interesting issues surrounding nectar secondary metabolites is
whether they have an adaptive function. The hypothesis that nectar alkaloids might be
antimicrobial is addressed in chapter four and appendix one. In appendix one I report the
discovery of a new floral yeast species found in Gelsemium sempervirens and its
susceptibility to gelsemine. In chapter four, I take the antimicrobial hypothesis further
and evaluate whether gelsemine-rich nectar can reduce infections of the gut pathogen
Crithidia bombi in bumble bees. This chapter turns the adaptive hypothesis on its head
by suggesting that consuming nectar alkaloids might be beneficial for pollinators.
Chapter five used high-performance liquid chromatography to survey
cardenolides from twelve species of the genus Asclepias. This survey looks at
correlations between cardenolide concentrations in nectar, leaves and flowers. I paid
particular attention to differences in the identity and chemical polarity of the individual
cardenolides found in nectar and leaves. I then used these data to evaluate whether nectar
cardenolides are a consequence of a plant’s systemic chemical defenses. This represents
the first study to identify and quantify nectar cardenolides and only the third to compare
nectar secondary metabolites to secondary metabolites in other plant parts.
Finally, chapter six is a general discussion that synthesizes my results and
identifies how the findings contribute to our understanding of plant-pollinator
interactions. In particular, I discuss themes that bridge my chapters such as the
importance of ecological context, the functional significance of nectar alkaloids and
nectar cardenolides and the subtle costs of nectar secondary metabolites on pollinators
and pollination.
12
CHAPTER TWO
Ecological context influences pollinator deterrence by alkaloids in floral nectar
Robert J. Gegear*, Jessamyn S. Manson*, and James D. Thomson
*equally contributing authors
R. J. Gegear and I contributed equally to the design and execution of this project and the writing of the
manuscript. J.D. Thomson provided substantial comments on the experiments and the paper, which is
published in Ecology Letters, 2007, 10: 375-382.
Abstract
Secondary metabolites may benefit plants by deterring herbivores, but the
presence of these defensive chemicals in floral nectar may also deter beneficial
pollinators. This tradeoff between sexual reproduction and defense has received minimal
study. We determined whether the pollinator-deterring effects of a nectar alkaloid found
in the perennial vine Gelsemium sempervirens depend on ecological context (i.e., the
availability of alternative nectar sources) by monitoring the behavioural response of
captive bumble bees (Bombus impatiens, an important pollinator of G. sempervirens in
nature) to nectar alkaloids in several ecologically-relevant scenarios. Although alkaloids
in floral nectar tended to deter visitation by bumble bees, the magnitude of that effect
depended greatly on the availability and nectar properties of alternative flowers.
Ecological context should thus be considered when assessing ecological costs of plant
13
defense in terms of pollination services. We consider adaptive strategies that would
enable plants to minimize pollinator deterrence due to defensive compounds in flowers.
Introduction
The interaction between plants and their animal pollinators has been a significant
force in the evolution of floral characters. However, plants also simultaneously interact
with other types of animal visitors, such as herbivores, which may affect pollination and
ultimately influence the evolution of floral traits. Although herbivory can influence
pollination through direct damage to reproductive tissues (e.g. pistils and stamens, Leege
and Wolfe 2002) or floral characters used to attract pollinators (e.g. corolla characters,
Strauss et al. 1996, for further review see McCall and Irwin 2006), it can also influence
pollination through more subtle, but nonetheless ecologically significant, mechanisms
(Agrawal et al. 1999, Strauss et al. 1999, Strauss et al. 2002). For instance, pollination
services to plants may be reduced when plant characters used to defend against
herbivores are linked to characters used to attract pollinators (Simms and Bucher 1996,
Strauss 1997). This integration of attractive and defensive traits presents plants with a
potential fitness tradeoff between the benefits of pollinator attraction and the costs of
reduced herbivore defense. Consequently, many pollination- and herbivory-related traits
that have traditionally been considered the result of selection by pollinators or herbivores
alone may actually be an evolutionary compromise between the contrasting selection
pressures exerted by both plant interactors together (Herrera et al. 2002). Despite the
important implications of interactions between pollination and herbivory for the ecology
and evolution of plant characters, there is little known about how pollinator visitation is
influenced by the concurrent presence of plant attractive and defensive traits. Here, we
14
describe a series of controlled experiments that were designed to examine the effects of
plant defensive compounds in flowers on the attractiveness of plants to pollinators.
From the pollinator’s perspective, the attractiveness of plants is determined
primarily by the perceived amount of beneficial compounds such as carbohydrates and
amino acids contained in floral nectar (Proctor et al. 1996). Paradoxically, floral nectar
of some plant species also contains secondary metabolites such as phenols and alkaloids
(Baker and Baker 1983) that occur in leaves, stems, and roots to defend against attack by
herbivores and microorganisms (Berenbaum 1995). Indeed, secondary metabolites have
been reported in the floral nectar of at least twenty-one angiosperm families (Adler
2000), indicating that this phenomenon is widespread. Although the presence of
secondary metabolites in floral nectar may provide reproductive benefits to plants in
some special cases (see Rhoades and Bergdahl 1981, Adler 2000 for reviews of
hypotheses), it is predicted to have detrimental effects if pollinators are deterred from
visiting flowers (Strauss et al. 1999, Adler and Irwin 2005). Although we are testing
adaptive hypotheses in other work, here we implicitly assume that defense chemicals
occur in nectar as an unavoidable byproduct of their production in other tissues; we
therefore consider the deterrence of pollinators as an ecological cost of defense (Strauss et
al. 1999, 2002).
Because the foraging decisions of pollinators are contingent on current floral
conditions and past floral experiences, the ecological cost of defense to plants with
secondary compounds in floral nectar likely depend on the ecological context in which
pollinators interact with flowers. For example, ecological costs may be reduced if
secondary compounds are present in nectar for short periods or increased if pollinators
15
have the option of visiting alternative plants with no secondary compounds in floral
nectar. How these costs vary with ecological context has not been rigorously examined,
presumably because of the difficulty in manipulating floral environments and tracking the
behaviour of individual pollinators under natural conditions.
Based on the pollination ecology of Gelsemium sempervirens (L.), we devised
laboratory choice experiments in which we used one of its major floral visitors and
pollinators, the bumble bee Bombus impatiens Cresson (Ornduff 1970; Manson
unpublished data), as a model system to investigate how ecological context influences the
effects of secondary compounds in floral nectar on pollinator choice behaviour and
foraging proficiency (flowers visited per minute and flower-handling time). Gelsemium
sempervirens is an obligate outcrosser and secretes the commercially available alkaloid
gelsemine in floral nectar. Previous work has shown that bumble bees (Bombus
bimaculatus, which is closely related to B. impatiens) spend less time on G. sempervirens
flowers and visit fewer flowers per plant when natural gelsemine concentrations are
increased in nectar (Adler and Irwin 2005, 2006), suggesting that nectar gelsemine
imposes an ecological cost on plants by altering visitation by bumble bees. We
determined whether the behavioural response of bumble bees to gelsemine-rich floral
nectar, and thus the ecological cost of defense to plants, depends on ecological context by
monitoring the choice behaviour, foraging rate and flower-handling time of freely
foraging bees on artificial floral arrays that simulated the following ecologically-relevant
scenarios: (1) G. sempervirens co-occurring and co-flowering with an equally rewarding
plant species without alkaloids in floral nectar, (2) G. sempervirens co-occurring and co-
flowering with less rewarding plant species without alkaloids in floral nectar, and (3) A
16
population of G. sempervirens in which plants have either a low or high level of
gelsemine in floral nectar. By comparing the behaviour of bees foraging under these
conditions, we were not only able to assess ecological costs of gelsemine in floral nectar
to G. sempervirens in terms of potential bumble bee pollination services, but provide new
perspectives on the adaptive significance of traits commonly observed in plant species
with secondary compounds in flowers.
Methods
Bees and flowers
Colonies of Bombus impatiens Cresson, each with 30-50 workers, were obtained
from Biobest Canada (Leamington, Ontario). Nest boxes were connected to a 2.2m x
2.2m x 2.4m flight cage by a gated tube so that we could control the number of bees
entering the flight cage. Prior to experiments, colonies were allowed to collect 30% w/w
sugar solution from feeders located in the center of the flight cage. Colonies were
supplied with pollen ad libitum. Workers that made regular foraging trips between the
colony and feeders were individually marked with coloured liquid paper.
Artificial flowers were constructed by removing the lids from 30 yellow and 30
blue 1.5 mL microcentrifuge tubes and adding 4.2 cm circles of blue and yellow
construction paper respectively around the mouth of the tubes. Yellow flowers resembled
the tubular yellow flowers of Gelsemium sempervirens. A blue-yellow colour
dimorphism was used to make it easier for bees to discriminate flowers based on
gelsemine and sucrose content of nectar. To access the test solution (hereafter referred to
as ‘nectar’), bees had to land on the surface of the paper corolla and crawl to the bottom
of the tube, much as they do in real G. sempervirens flowers. Flowers were presented to
17
bees by embedding them upright in a 1.26 x 0.79 x 0.032 m styrofoam board covered in
green paper. Flowers were positioned in a 67.5 cm by 56.0 cm grid so that bees had an
equidistant choice of each flower type upon departing any flower on the array (with the
exception of the two outer columns). Our artificial array was designed in this manner so
that we could create realistic floral environments for bees while controlling for the
availability, distribution and nectar properties of flowers.
The principal alkaloid found in the nectar of Gelsemium sempervirens is
gelsemine (Irwin and Adler 2006), with natural concentrations ranging across populations
from 5.8 ng/µL to 246.1 ng/µL (Adler and Irwin 2005). Sucrose concentrations in G.
sempervirens nectar reportedly range from 11 to 62% under natural conditions (Leege
and Wolfe 2002, Adler and Irwin 2005; Manson, personal observation). For our
behavioral assays, nectar containing both gelsemine and sucrose was created by adding
gelsemine hydrochloride (ChromaDex, Santa Ana, CA) to aqueous sucrose solutions
(either 30% or 50% w/w sucrose) until all gelsemine was dissolved (gelsemine
concentrations were 0, 5, 50, and 125 ng/µL sucrose solution; Table 2.1). Solutions were
refrigerated at 4°C when not in use and replaced every 3-5 days. For brevity, we refer to
the sucrose concentration of nectar as either S30 or S50 and the gelsemine concentration
of nectar as either G0, G5, G50, or G125.
Experimental procedure
We determined the behavioural response of bumble bee foragers to gelsemine in
nectar under the three ecological scenarios described above. Gelsemine and sucrose
concentrations used in each assay were selected based on values reported for G.
sempervirens under natural conditions (Table 2.1).
18
Marked bees were trained by allowing them to forage freely on an array of each
flower type (i.e., each nectar condition) in succession for three foraging trips. This
procedure ensured that bees had experienced the nectar condition associated with each
flower colour prior to testing. The flower colour associated with each nectar condition
was randomized among bees to control for the possibility that floral preference was
influenced directly by colour. Immediately following training, bees were individually
presented with a mixed array containing 30 flowers of each type, and we videotaped at
least 80 flower visits for later analysis. Flowers were filled with 3µL of nectar and
refilled quickly after being drained by bees. Flowers were replaced between bees. After
testing, bees were freeze-killed, and body size was estimated by measuring the length of
the radial cell on the right forewing (Harder 1982).
Data analysis
For each assay, we determined whether bees overall had a preference for flowers
with lower levels of gelsemine on the mixed array (i.e. visitation frequency was non-
random with respect to alkaloid level) by using a two-tailed one-sample t-test to compare
the mean proportion of visits to flowers with the lower concentration of nectar gelsemine
to the proportion of visits expected given the abundance of both flower types on the
mixed array (0.5 in all cases). Proportions were arcsin-transformed so that they
conformed more closely to a normal distribution. We then examined how the flower-
choice behaviour of individual bees changed as they gained foraging experience on the
mixed array by dividing the first 80 flower visits for each bee tested into four blocks of
20 consecutive visits. For each block, we determined whether individuals had a
preference for one of the available flower types by using a G-test of independence (Sokal
19
and Rohlf 1995) to compare the observed frequency of visits to low gelsemine flowers to
the frequency of visits expected given random flower selection (10 visits). An observed
visit frequency of 15 or greater indicated that the bee had a preference for low gelsemine
flowers whereas a visit frequency of 5 or less indicated a preference for high gelsemine
flowers. We then tested for changes in flower-choice behaviour of bees over time by
using a repeated measures ANOVA to compare visit frequency to low gelsemine flowers
among the four blocks, followed by Tukey’s multiple comparison test.
Because previous work has suggested that nectar alkaloids, including gelsemine,
may affect plant fitness by altering the behaviour of pollinators on flowers rather than
through pollinator deterrence (Strauss et al. 1999, Adler and Irwin 2005), we also
examined the effect of gelsemine on bee foraging proficiency. Here, we assess foraging
proficiency by measuring foraging rate (number of flowers visited per minute) and
flower-handling time (total time in seconds that the bee spends on a flower), which are
two components of bumble bee behavior that may affect how they collect and deposit
pollen and thus provide another measure of how nectar gelsemine may influence plant
reproductive success though behavioral alterations to bees. We used a generalized linear
model with radial cell length as a covariate (Proc Genmod; SAS version 8) to compare
foraging rates and flower-handling times of individuals that showed a preference (i.e.
visitation frequency was significantly biased in favour of the flower type) for S30G0
flowers (n=20) and those that showed a preference for S30G50 flowers (n=10). Measures
of foraging proficiency were calculated based on 10 consecutive flower visits taken
randomly between visits 50 and 70 and results are reported as likelihood ratio statistics
20
(G) (SAS Institute1999). All foraging proficiency measures were log-transformed to
meet the assumptions of normality and equal variance.
Results
Flower preference
Bumble bee choice behaviour was significantly influenced by the concentration of
gelsemine in the nectar of available flower types. Although bees as a group readily
collected floral nectar containing gelsemine on monotypic arrays during training, they
had a strong preference for nectar with equal sucrose rewards but no gelsemine (Assays
1A-C) or lower levels of gelsemine (Assay 3 - S30G50 vs S30G125) on mixed arrays
(Table 2.2). Bees showed no overall nectar preference when the sucrose concentration of
nectar with alkaloids was increased relative to an alkaloid-free nectar alternative (Assay 2
- S30G0 vs S50G50; Table 2.1). In fact, there was a significant decrease in the proportion
of visits to flowers with no nectar gelsemine between Assays 1B and 2 (t=6.37, df=16,
p<0.001), indicating that an increase in sucrose concentration of floral nectar containing
gelsemine relative to alkaloid-free nectar alternatives increased its attractiveness to bees.
At the individual level, there was a considerable amount of variation in flower-
choice behaviour of bees on the mixed array over time (Fig. 2.1). With the exception of
Assay 2, most individuals had a strong preference for flowers with nectar containing
sucrose only or low levels of gelsemine in the final visit block (percentage of bees with a
preference for flowers with no or low nectar gelsemine in the final visit block was: 92.3%
(Assay 1A), 63.6% (Assay 1B), 77.8% (Assay 1C), 36.4% (Assay 2), and 77.8% (Assay
3). Interestingly, a small percentage of bees showed a preference for flowers with higher
nectar gelsemine concentrations in the final visit block (7.7% (Assay 1A), 9.1% (Assay
21
1B), 11.1% (Assay 1C), 45.5% (Assay 2)). The mean proportion of visits to the flower
type on the mixed array with the lower nectar gelsemine concentration differed
significantly among visit blocks for Assay 1a (F3,12=12.32, P<0.0001), 1b (F3,10=7.09,
P=0.001), 1c (F3,8=6.98, P=0.0015, and 3 (F3.8=6.61, P=0.0021), but not for Assay 2
(F3,10=0.542, P=0.657). Pairwise comparisons showed that the proportion of flower visits
to low gelsemine flowers significantly increased between block 1 and 3 and block 1 and 4
for Assays 1A-C and Assay 3 and also between block 2 and 3 and block 2 and 4 for
Assay 1A, indicating that bees tended to sample both flower types on the mixed array
prior to developing a preference for the flower type with the lower level of nectar
gelsemine.
Foraging proficiency
Visitation to gelsemine-rich flowers had no significant effect on foraging rate or
mean handling time (Table 2.3). Bee size, determined from radial cell length, did
positively correlate with foraging rate (G1,27=5.01, P=0.03), but did not correlate with
mean handling time. There was no interaction between preference and size for any of the
three foraging efficiency measures, so the interaction term was removed from the model.
Discussion
Our study supports the hypothesis that defensive compounds in floral nectar
impose an ecological cost on plants in the form of reduced pollinator visitation, and
demonstrates for the first time that such ecological costs to plants depend heavily on the
ecological context in which pollinators make foraging decisions. Bumble bees readily
foraged on monotypic arrays of alkaloid-rich flowers regardless of nectar alkaloid level,
22
but quickly developed a strong aversion to them when flowers with lower levels of nectar
alkaloids were made available. These results suggest that nectar alkaloids would only be
a significant ecological cost to Gelsemium sempervirens plants when they must compete
for pollination services with alternate plants that have lower levels of nectar alkaloids.
Interestingly, G. sempervirens flowers very early in the spring (Pascarella, 2007), perhaps
because pollinator response to alkaloids in floral nectar was an important selective
pressure on flowering time. Nectar quality has been postulated to be a significant factor
in the evolution of flowering phenology (e.g. Mosquin 1971, Heinrich 1975, Brody
1997), with plant species of low nectar quality evolving earlier bloom times to escape
competition for pollinators with plant species of high nectar quality. Early bloom time
may thus be one adaptive mechanism, or ‘counteradaptation’ (Strauss et al. 1999), in
plants to mitigate the loss of pollination services due to nectar alkaloids.
Although most bees quickly learned to associate alkaloid concentration with
flower-colour cues, and avoided alkaloid-rich flowers when equally rewarding alkaloid-
free alternatives were available, the deterrent effect of the alkaloid was offset by higher
sugar concentrations. Thus, bees acted as if they were balancing economic gains (sugar
collection) against palatability (alkaloid concentration). Previous work on feeding
behaviour in herbivorous insects has shown that carbohydrates can counteract the
deterrent effects of many plant secondary compounds, including alkaloids, and do so
through a variety of complex physiological response mechanisms (Dethier 1982, Mitchell
and Sutcliffe 1984, Mitchell 1987, Dethier and Bowdan 1992, Shields and Mitchell 1995,
Glendinning et al. 2000). Similar mechanisms may mediate the behavioral response of
bumble bees, and other insect pollinators, to nectar alkaloids. For instance, the
23
unpleasant taste of the alkaloids may be ‘masked’ by higher sucrose concentrations
(Glendinning 2002), in the same way that a person can make a bitter food like chocolate
more palatable by adding sugar. Alternatively, alkaloids may interfere with the ability of
sucrose-sensitive receptor cells in the peripheral taste system to detect the correct sucrose
concentration of nectar (a process called sensory inhibition; Mitchell & Sutcliffe 1984).
In this view, higher sucrose concentrations are required for alkaloid-rich nectar to be
perceived as a profitable resource. Regardless of the behavioural mechanisms involved,
the combined effect of nectar alkaloids and sugars on floral attractiveness has important
implications for our understanding of how pollinators assess nectar quality and make
adaptive foraging decisions. For example, pollinators performing a “behavioral titration”
(Kotler and Blaustein 1995, Webster and Dill 2006) of sucrose and alkaloid uptake would
alter how they allocate foraging effort to available plant species.
The reduced deterrent effect of alkaloids caused by a relative increase in nectar
sucrose concentration suggests that plants with nectar alkaloids would minimize the loss
of pollination services to co-flowering plants with lower alkaloid levels by increasing the
caloric content of nectar. At present, there is little information on the relationship
between secondary metabolites and caloric content of nectar because past studies have
either held caloric content and secondary metabolite content constant (Stephenson 1982,
Masters 1991, Hagler and Buchmann 1993, Landolt and Lenczewski 1993, Singaravelan
et al. 2005), or not compared caloric content and secondary metabolite concentration
among plant species available to pollinators (Stephenson 1981, Adler and Irwin 2005).
We predict that plants with more alkaloids in floral nectar will contain higher caloric
rewards (concentration and possibly volume) relative to other available plants with lower
24
nectar alkaloid levels. This hypothesis for nectar alkaloids is akin to the “nutrient/toxin
titration” model proposed for the presence of toxins in fruit (Cipollini and Levey 1997b).
Our results indicate that ecological costs of alkaloids in floral nectar to G.
sempervirens are due to a reduction in the quantity (number of individuals visiting plants)
and not the quality (individual behaviour on flowers) of floral visitation by pollinators.
Bees that collected gelsemine-rich nectar spent the same amount of time on flowers and
visited the same number of flowers per minute as bees that collected gelsemine-free
nectar, indicating that there would be no reproductive cost to G. sempervirens in terms of
a reduction in the ability of pollinators to remove and deposit pollen. In contrast to our
results, Adler and Irwin (2005) found that increasing gelsemine concentrations in flowers
of natural G. sempervirens populations had no effect on initial attraction of pollinators to
plants, but reduced the amount of time that they spent per flower and the number of
flowers visited per plant. One likely explanation for the discrepancy between our results
and those of Adler and Irwin is that bees in our study visited flowers with different levels
of gelsemine in sucrose rewards for extended periods of time and were thus able to learn
the association between floral cue (colour in our case) and nectar properties. Indeed,
many bees in our study entered and quickly departed from flowers with gelsemine-rich
nectar while learning to associate flower colour with reward condition (see Fig. 2.1),
suggesting that ecological costs to plants increase with pollinator foraging experience.
This point underscores the importance of identifying and tracking individual pollinators
over ecologically relevant periods during field studies aimed at determining the effects of
pollinator behaviour on plant fitness.
25
Plasticity in the behavioural response of pollinators to secondary compounds in
floral nectar has important implications for the evolution of plant defenses against
herbivory. Optimal defense theory (ODT) predicts that constitutive defenses in valuable
tissues should be more advantageous to plants than induced defenses because such tissues
are protected prior to damage by herbivores (Rhoades 1979, McCall and Karban 2006).
Our study suggests that predictions of ODT for plants with defense compounds in flowers
need to incorporate potential tradeoffs between benefits of herbivore resistance and costs
of pollinator deterrence. Based on the behavioural response of bumble bees to nectar
alkaloids observed in our study, we expect constitutive defenses in floral tissues and
nectar (i.e. reproductive traits) of outcrossed plants to be favoured only when pollinators
have few other floral resources available or caloric rewards compensate for reduced
nectar palatability. In contrast, we expect induced defenses in flowers to be advantageous
when there is strong competition for pollinators, since floral attractiveness would be
reduced for short time periods subsequent to damage by herbivores. Moreover, induced
defenses would reduce the floral attractiveness of a small subset of plants in the
population (assuming that levels of herbivory are low), thereby decreasing the likelihood
that pollinators will learn to discriminate against all plants with a floral signal similar to
that of the defended plant (i.e. other plants in the population). Thus, inducible defenses
can benefit plants under many ecological conditions by allowing them to mount a strong
defense against herbivores while minimizing ecological costs due to pollinator
deterrence. Although there is growing evidence for induced defenses in flowers (e.g.
Adler et al. 2006, McCall 2006, McCall and Karban 2006), the costs and benefits of
induced versus constitutive defense strategies has not been considered in terms of
26
pollinator deterrence. In future field experiments, we plan to determine how plant
defense strategies affect pollination services, and thus plant fitness, in different ecological
contexts.
Acknowledgements
We would like to thank J. Forrest, J. Thaler, and four anonymous reviewers for helpful
comments on the manuscript. This study was funded by an NSERC grant to JDT.
27
Table 2.1. Descriptions of the reward conditions used in each behavioural assay.
Sucrose and gelsemine concentrations used in each assay were selected based on their
ecological significance. Concerning Assay 3, note that gelsemine production by
Gelsemium sempervirens has yet to be characterized as either induced or constitutive.
Nectar Condition
Assay
Gelsemine-poor Gelsemine-rich
Ecological Significance
1A 30% w/w
sucrose
30% w/w sucrose
with 50 ng/µL
gelsemine
50 ng/µL is in the middle range of natural
concentrations of gelsemine in G. sempervirens
nectar (Adler and Irwin 2005)
1B 50% sucrose
50% sucrose with
50 ng/µL gelsemine
Simulates high nectar concentrations found in the
field (Adler and Irwin 2005)
1C 30% sucrose 30% sucrose with
5 ng/µL gelsemine
5 ng/µL is the lowest concentration of gelsemine
found in G. sempervirens nectar (Adler and Irwin
2005)
2 30% sucrose
50% sucrose with
50 ng/µL gelsemine
Presents bees with a trade-off between palatability
and economic gain
3
30% sucrose
with 50 ng/µL
gelsemine
30% sucrose with
125 ng/µL
gelsemine
Simulates a possible induced chemical response to
herbivory that increases nectar alkaloid
concentrations or natural variation in nectar
alkaloid concentrations
28
Table 2.2. Results of one-sample t-test for Assays 1-3. For each assay, the mean
proportion of visits to the flower type with the lower level of nectar alkaloids on the
mixed array was compared to the proportion of visits to flowers with lower nectar
alkaloids expected given the abundance of both flower types on the mixed array (0.5 in
all cases). Means are given +/- SE.
Assay
Mean proportion of visits to
low gelsemine flowers t-value df P
1A 0.86 +/- 0.04 7.932 12 <0.0001
1B 0.76 +/- 0.09 2.54 10 0.029
1C 0.84 +/- 0.07 4.89 8 0.0012
2 0.50 +/- 0.10 -0.149 10 0.88
3
0.82 +/- 0.04 7.04 8 0.0001
29
Table 2.3. Generalized linear model results for bee foraging rate (flowers visited per
minute), and flower-handling time (in seconds). Data are from foragers in Assays 1A, 1C
and 3 that had a significant preference for either sucrose-only (n=20) or sucrose plus
gelsemine (n=10) nectar. All values are mean +/- SE and df=1, 27 for each analysis.
Foraging Proficiency
Measure
Sucrose Only Sucrose and Gelsemine G P
Foraging Rate
(flowers/minute) 9.38 +/- 0.36 8.91 +/- 0.29 1.82 0.18
Flower- Handling Time
(seconds)
4.76 +/- 0.6 5.77 +/- 0.42 3.2 0.07
30
0
25
50
75
100
1A: S30G0 vs S30G50
0
25
50
75
1002: S30G0 vs S50G50
0
25
50
75
100
3: S30G50 vs S30G125
0
25
50
75
1001B: S50G0 vs S50G50
1 2 3 4
0
25
50
75
100
1C: S30G0 vs S30G5
Visit Block
Per
cen
t o
f vis
its
to l
ow
alk
alo
id f
low
ers
31
Figure 2.1. (previous page) Choice behaviour of individual bees over four consecutive
blocks of 20 flower visits. Figure panel numbers correspond to assay numbers from
Table 2.1. Changes are displayed in percent preference for flowers with low or no
gelsemine in nectar, as indicated by the treatment in bold text, and each line represents
the preference trajectory of a single foraging bee. Dashed reference lines demarcate zones
of significance as determined by G-test values, with percent of visits to flowers with low
or no gelsemine in nectar above 71.6% indicating that the bee had a preference for
flowers with no or low levels of gelsemine in nectar and below 28.4% indicating that the
bee had a preference for flowers with gelsemine in nectar.
32
CHAPTER THREE
Post-ingestive effects of nectar alkaloids depend on dominance status of bumble bees
Jessamyn S. Manson and James D. Thomson
This project was designed in collaboration with J. D. Thomson. I completed the experiments and wrote the
paper with assistance from J. D. Thomson, which appears in Ecological Entomology, 2009, 34: 421-426.
Abstract
Secondary metabolites have acute or chronic post-ingestive effects on animals,
ranging from death to growth inhibition to reduced nutrient assimilation. Although
characterized as toxic, the nectar of Gelsemium sempervirens is not lethal to pollinators,
even when the concentration of the nectar alkaloid gelsemine is very high. However,
little is known about the sublethal costs of nectar alkaloids. Using a microcolony assay
and paired worker bumble bees we measured the effects of artificial nectar containing
gelsemine on oocyte development. Oocytes are a sensitive indicator of protein utilization
and general metabolic processes. We also calculated carbohydrate concentrations in the
haemolymph to examine energetic costs of gelsemine consumption. High concentrations
of gelsemine significantly reduced mean oocyte width in subordinate bees, while
dominant bees showed only a trend towards oocyte inhibition. Gelsemine consumption
did not reduce carbohydrate concentrations in haemolymph. The cost of ingesting
gelsemine may be due to direct toxicity of alkaloids or may be an expense associated with
detoxifying gelsemine. Detoxification of alkaloids can require reallocation of resources
33
away from essential metabolic functions like reproduction. The risks associated with
nectar alkaloid consumption are tied to both the social and nutritional status of the bee.
Introduction
Plant secondary metabolites are believed to have evolved as chemical defenses
against herbivorous animals (Whittaker and Feeny 1971, Janzen 1973, Feeny 1992,
Berenbaum 1995). Acute toxicity, resulting in death, is reported in many of the major
secondary metabolite families (e.g., alkaloids, phenolics, glycosides, Rosenthal and
Berenbaum 1991). Although lower concentrations of a secondary metabolite may not be
lethal, they can reduce the overall health and fitness of an animal (chronic toxicity,
Berenbaum et al. 1986).
Such chronic effects are subtle and highly variable. Typically, bioassays of
growth, development or reproduction are used. Zangerl and Berenbaum (1993) showed
decreased growth of parsnip webworm (Depressaria pastinacella) larvae when fed on
wild parsnip (Pastinaca sativa) umbels with high furanocoumarin levels. Similarly,
winter moth caterpillars (Operophtera brumata) fed on oak leaves with high tannin
concentrations had reduced larval and pupal weights, along with reduced adult emergence
(Feeny 1970). Tannic acid caused developmental malformations in tent caterpillar
(Malacosoma disstria) pupae (Karowe 1989), while phenolic glycoside concentrations
were negatively correlated with fecundity in gypsy moth, Lymantria dispar (Osier et al.
2000).
Nearly all studies on plant chemical defense focus on secondary metabolites in the
shoots or roots; however, these compounds are also paradoxically found in floral nectar.
Although the functional significance of nectar secondary metabolites is not fully
34
understood (but see Adler 2000 for a review of hypotheses), studies do suggest that this
so-called “toxic” nectar can have deleterious consequences for nectar-collecting floral
visitors. Honey bees have died after consuming artificial nectar containing very low
concentrations of alkaloids and glycosides (Detzel and Wink 1993); in other cases,
ingesting “toxic” nectar has less severe consequences. For example, Palestine sunbirds
consuming nectar containing pyridine alkaloids were less able to assimilate sugar from
their diet (Tadmor-Melamed et al. 2004). Secondary metabolites in nectar, particularly
alkaloids and phenolics, have been shown to deter pollinators and reduce number of
flower visits (Adler and Irwin 2005, Singaravelan et al. 2005, Johnson et al. 2006, Gegear
et al. 2007). Despite multiple reports of the distastefulness of “toxic” nectar, few studies
have correlated behavioural responses with effects of nectar secondary metabolite
consumption on floral visitors.
The chemical arsenal of Gelsemium sempervirens L. (Carolina jessamine)
includes the indole alkaloid gelsemine, a compound found in the roots, shoots, flowers
and floral nectar of the plant. G. sempervirens is a perennial vine native to the
southeastern United States; its fragrant yellow flowers open in the early spring and
attract several flower visitors (Adler and Irwin 2005, Pascarella 2007; Manson, personal
observation), including bumble bee queens and workers (Bombus bimaculatus, B.
impatiens), honey bees (Apis mellifera), carpenter bees (Xylocopa virginica) and solitary
bees (Osmia lignaria, Habropoda laboriosa). The consequences of ingesting flowers or
leaves from Gelsemium spp. are severe for mammals and include psychosis, respiratory
failure, severe convulsions and death (Blaw et al. 1979, Ott 1998, Rujjanawate et al.
2003, Fung et al. 2007). In contrast, adult bumble bees exposed to high levels of
35
gelsemine experience no acute effects, even when gelsemine concentrations are 20 times
higher than natural levels (Manson, personal observation). Similarly, Elliott et al. (2008)
reported no effect of gelsemine on the number or survivorship of offspring produced by
the megachilid solitary bee, Osmia lignaria. However, laboratory assays indicate that
bees prefer to feed on sucrose-only nectar rather than a solution of gelsemine and sucrose
(Gegear et al. 2007), while enriching G. sempervirens nectar with gelsemine deterred
visitors in nature (Adler and Irwin 2005), implying that there are consequences to the
ingestion of alkaloid-rich nectar.
Given the absence of acute effects, we tested for sublethal costs of nectar
alkaloids by feeding bumble bee workers (Bombus impatiens Cresson) artificial nectar
containing gelsemine and measuring development of their oocytes. Oocyte development
provides a good bioassay because it is a defined metabolic challenge that can be induced
by pairing worker bees without a queen (Cnaani et al. 2002, Cnaani et al. 2007).
Furthermore, it is a complex and costly physiological process which we predict to be
sensitive to toxins for several reasons. First, oocyte development is highly correlated
with protein utilization in worker bees (Lin and Winston 1998, Pernal and Currie 2000).
Although there are several ways for insects to cope with secondary metabolites, a
common mechanism is to detoxify these compounds into less hazardous ones (Slansky
1992). Detoxification requires the production of specific enzymes from dietary proteins.
If this mechanism is used by bumble bee workers, protein could be re-allocated towards
the construction of enzymes and away from oocyte production. Second, alkaloid
processing requires energy, so carbohydrates used for normal metabolic processes may be
redirected to alkaloid metabolism, leaving less energy for the formation of reproductive
36
structures. In addition, alkaloids may directly interfere with the absorption of nutrients
by inhibiting digestive enzymes or forming nutrient-allelochemical complexes (Slansky
1992). We therefore hypothesize that metabolic costs associated with nectar alkaloid
consumption will result in reduced oocyte development. We also evaluated whether
alkaloid processing directly reduces available carbohydrates by measuring carbohydrate
levels in bee haemeolymph 24 hours after ingestion. We discuss our findings with a
focus on possible mechanisms for alkaloid tolerance in pollinators.
Methods
Oocyte development
Oocyte development in Bombus spp. depends on social circumstances. If multiple
workers are kept in queenless colonies, one of them frequently assumes a queen-like role,
becoming the dominant worker and developing oocytes (Cnaani et al. 2002, Cnaani et al.
2007). When two bumble bee workers interact in a queenless colony, the dominant
worker will develop its ovaries at an optimal rate while suppressing the ovary
development rate of the subordinate worker. We took advantage of this developmental
strategy, building “microcolonies” from worker bees to assess how gelsemine
consumption affects ovary development under optimal and suboptimal conditions.
We obtained pupal clumps of Bombus impatiens from Biobest Canada Ltd.
(Leamington, ON). Bumble bee microcolonies were composed of two unfed callow
workers (<24 hours old). We created a size dichotomy in each container in an effort to
enhance differences between dominant and subordinate bees, as previous work suggests
that larger bees are more likely to be dominant (Ayasse et al. 1995). Each pair of bees
was housed in a closed 500 mL clear plastic food container, lined with paper to absorb
37
feces, and equipped with holes for ventilation (along the sides) and nectar access (on the
base). This container was nested in a second food container, which held a small petri
dish of artificial nectar, made accessible by a cotton wick that led up to the holes on the
base of the first container. This arrangement reduced spilling and contamination of the
nectar by preventing direct contact between the nectar and the bees. Individuals were
divided evenly among treatments so that we had a total of 28, 29, and 28 pairs of bees in
control, moderate and high gelsemine treatments, respectively, after three replicates,
which were run at three separate dates.
We used a 30% w/w solution of sucrose as artificial nectar, which fell well within
the range of natural sugar levels reported in Gelsemium sempervirens flowers (Leege and
Wolfe 2002, Adler and Irwin 2005), to which we added gelsemine hydrochloride
(hereafter referred to as gelsemine; Chromadex, Santa Ana, CA). We used three diet
treatments: control, composed of sucrose only; moderate gelsemine, a solution of sucrose
plus 50 ng/µL gelsemine; and high gelsemine, a solution of sucrose plus 250 ng/µL
gelsemine. The two gelsemine treatments simulate the mean and maximum
concentrations, respectively, of gelsemine found in the nectar of natural Gelsemium
sempervirens populations (Adler and Irwin 2005). Bees avoid nectar of both of these
concentrations if control nectar is available (Gegear et al. 2007). We supplied 1.5 mL of
artificial nectar daily, as well as commercially available pollen ad libitum. We provided
new pollen every day; pollen lumps were weighed before and after they were provided to
a container to determine daily pollen consumption by the pair of bees. Microcolonies
were maintained for six days under controlled environmental conditions (23-27ºC in the
dark, except for during feeding), which is the estimated time needed for B. impatiens
38
oocytes to mature (Cnaani et al. 2002, Cnaani et al. 2007). After six days, we froze the
bees and dissected them in distilled water to determine ovary development. Using a
scaled ocular, we measured the length and width of the largest oocyte in each of the two
paired ovaries using the average of the two in our analyses. Development in the two
tended to be symmetrical. We also recorded the length of the radial cell in the front right
wing as a proxy for bee size (Harder 1982).
We compared oocyte length and width between treatments with ANCOVA, using
radial cell length as a covariate. We chose to analyze length and width separately to
pinpoint the effects of gelsemine consumption on each of these size parameters. Previous
work on oocyte development has used either a subjective size ‘score’ (Pernal and Currie
2000) or measured length alone (Bloch and Hefetz 1999, Cnaani et al. 2007), which may
have overlooked possible variation in oocyte width. Dominance was assigned to the bee
within each pair with the larger oocytes, estimated as length times width, and we
analyzed dominant and subordinate bees separately. We pooled data from the three
experimental replicates, as the data did not significantly differ between replicates. We
removed four pairs of bees from the analysis because one of the pair died before the
experiment was completed, changing the social environment of the remaining bee. These
pairs were spread across treatments. We also removed two subordinate bee outliers with
extremely small oocytes.
To assess possible differences in protein intake between treatments, we analyzed
daily pollen consumption by microcolonies using a repeated measures ANOVA. When
necessary, data was transformed to meet assumptions for normality and homogeneity of
variance.
39
Haemolymph carbohydrates
We removed pupal clumps from individual commercial colonies and isolated
unfed callow bees (<24 hours old) in individual vials. We provided bees with 500 µL of
one of the three treatments: control (30% w/w sucrose), moderate gelsemine (50 ng/µL
gelsemine in 30% sucrose) or high gelsemine (250 ng/µL gelsemine in 30% sucrose).
After 24 hours, when nearly all the nectar was consumed, we refrigerated the bees and
then decapitated them. We took haemolymph samples by separating the ventral terga with
forceps and gently inserting a 5 µL microcapillary tube into the lower abdominal cavity.
We estimated the volume of each haemolymph sample and stored them individually in 1
mL of 80% ethanol. We analyzed carbohydrates in 18, 17, and 17 individuals in the
control, moderate and high gelsemine treatments, respectively.
We calculated carbohydrate concentrations, expressed as micrograms of trehalose
equivalents per microlitre of haemolymph, using the anthrone method (modified from
Siegert 1987). Because the carbohydrate data were not normally distributed, we tested
for differences in carbohydrate concentration between gelsemine treatments using a non-
parametric Kruskal-Wallis test.
All statistical analyses were performed in R (version 2.6.0).
Results
Protein metabolism
Nectar gelsemine concentration did not affect oocyte length of dominant or
subordinate bees (Table 3.1, Fig. 3.1). However, high levels of gelsemine did
significantly reduce oocyte width in subordinate bees (post-hoc Tukey tests using the
multcomp package in R, Fig. 3.1), while there was a trend towards smaller widths in
40
oocytes of dominant bees in the high gelsemine treatment. An unexpected element of the
experiment was that dominance was not reliably predicted based on bee size; that is to
say, the largest bees did not always have the largest oocytes. However, there was still a
positive relationship between oocyte size (LxW) and radial cell length (R2=0.2, F=42.36,
df=162, P<0.001). Therefore, radial cell was kept in the analyses and did contribute to
the explanatory power of each model.
The reduction in oocyte size was not correlated with reduced protein intake, as
pollen consumption did not differ between treatments (Table 3.2). Pollen consumption
did vary significantly between days within each treatment, with bees eating the most on
the second day of the assay, followed by a decline in appetite and a resurgence in pollen
consumption by day 6. There was no interaction between treatment and day.
Carbohydrate concentrations
Gelsemine did not affect the concentration of carbohydrates found in bee
haemolymph (Kruskal-Wallis test, χ2 =3.01, df=2, P=0.22). Although the data are highly
variable (Figure 3.2), removing outliers did not reveal differences between treatments
(analysis not shown).
Discussion
The nectar alkaloid gelsemine significantly inhibits oocyte development in
subordinate bees, but is only marginally effective at reducing oocyte size in dominant
bees. This effect was detectable at ecologically relevant concentrations, suggesting that
ingestion of nectar alkaloids can incur a cost to pollinators. The severity of this cost,
however, appears to depend on the condition of the bumble bee and the concentration of
41
the alkaloid. Overall, the mean concentration of gelsemine found in nature may be
largely innocuous to healthy bees. Under suboptimal circumstances, however, the
ingestion of nectar alkaloids might be chronically deleterious to pollinators.
Sublethal effects of nectar alkaloids on bumble bees may arise in various ways.
First, the alkaloids may not be toxic enough to kill bees, but they may be toxic enough to
compromise nutrient absorption, alter neurohormonal processes, or damage internal
organs (see Slansky 1992 for review). These outcomes may lead to protein excretion or
increased protein investment in immune function, reducing oocyte size. Another
explanation for inhibited oocyte development is that detoxifying alkaloids is costly.
Detoxification of secondary metabolites is a common process whereby compounds are
metabolized into less toxic components, and it is often accompanied by rapid excretion.
This process requires protein to build detoxification enzymes and energy to process the
deconstruction of the secondary metabolites; it is therefore assumed to be metabolically
expensive. The assumption of costliness is supported by the observation that many
detoxification mechanisms are induced only after the consumption of a secondary
metabolite. Inducibility is interpreted as an energy-saving strategy (Berenbaum and
Zangerl 1994). Although this explanation is attractive, empirical evidence is divided on
whether detoxification is a significant expense. Detoxification of alkaloids is reported to
reduce digestive efficiency in the southern armyworm, Spodoptera eridania (Cresswell et
al. 1992), while parsnip webworms (Depressaria pastinacella) shunt energy away from
growth to metabolize furanocoumarins (Berenbaum and Zangerl 1994). In contrast,
alkaloid detoxification in Helicoverpa zea required a negligible amount of energy
compared to that spent on regular metabolic activity (Neal 1987). Evidence of metabolic
42
costs due to “toxic” nectar ingestion are sparse; we know that Palestine sunbirds
experienced reduced sucrose assimilation after consuming nectar containing the alkaloids
nicotine and anabasine (Tadmor-Melamed et al. 2004), but whether this was the result of
reallocation to alkaloid detoxification was not identified. In our study we did not find
that gelsemine affected carbohydrate levels (see Fig. 3.2). However, carbohydrate levels
in insect haemolymph are reportedly highly variable and may lack the resolution
necessary to detect the energetic costs associated with alkaloid detoxification (Thompson
2003).
Secondary metabolites have reduced protein utilization in previous studies on
both vertebrates and invertebrates (reviewed in Duffey and Stout 1996). Oocyte size is a
sensitive measure of protein utilization (Duchateau and Velthuis 1989, Lin and Winston
1998, Pernal and Currie 2000) and smaller oocytes were reported in worker bumble bees
infected with the gut pathogen Crithidia bombi, suggesting that oocyte size can indeed be
an indicator of poor health (Shykoff and Schmid-Hempel 1991). We must therefore
conclude that protein metabolism and, consequently, fecundity in dominant bees is only
modestly affected by ingested nectar alkaloids. This conclusion is supported by work
done on Osmia lignaria (Elliott et al. 2008), which found that gelsemine did not reduce
the fecundity of healthy solitary bees. The significant reduction in the oocyte size of
subordinate bees in the high gelsemine treatment suggests nectar alkaloids may incur a
cost to protein metabolism in individuals of suboptimal condition. Whether the inhibition
of oocyte development due to gelsemine results in extended ovary development time or
smaller offspring is unknown, but both outcomes could affect fitness.
43
The microcolony assay was designed to test the direct effects of gelsemine on
dominant bees because previous studies have shown a predictable response on the ovary
development of the dominant worker under different social and nutritional environments
(Duchateau and Velthuis 1989, Cnaani et al. 2002, Cnaani et al. 2007). The role of the
subordinate bees in the microcolonies was simply to fulfill the social conditions required
for optimal ovary development in their dominant counterparts. However, the significant
treatment response by the subordinate bees, which experienced suboptimal conditions for
oocyte development, is an unexpected but important result. The response of subordinate
bees to the consumption of gelsemine is a complex effect that may involve both
metabolism and behaviour. Previous studies on worker oocyte development have
reported that bees exert dominance, in part, by monopolizing the pollen ball (Cnaani et al.
2007). This behaviour reduces the subordinate bee’s access to protein, which likely
explains why all subordinate bees have smaller oocytes. In addition to obtaining less
dietary protein to support oocyte development, these food-stressed bees may experience
heightened costs of detoxification. In fact, Wahl and Ulm (1983) demonstrated that the
cost of metabolizing pesticides increased when honey bee pollen intake was reduced.
Compensatory pollen feeding by the dominant bees in the high-gelsemine treatment
might further reduce the amount available to subordinates in those treatments, either
directly through consumption by the dominants, or indirectly because dominant bees
spend more time at the pollen ball and guard it more stringently. The possibility of
competition for access to pollen could also explain why larger bees tended to fare better
(significant effect of radial cell length Table 3.2). We found no effects of gelsemine on
pollen consumption (Table 3.2), but those data include pollen consumption by both bees.
44
We cannot determine whether the allocation of pollen to dominants and subordinates may
have differed among treatments. Furthermore, the hygroscopic nature of pollen, coupled
with the necessity of using fresh weights, renders the pollen consumption data only
approximate.
The effects of nectar alkaloids on pollinators must be interpreted with natural
plant-insect interactions in mind. In this study, we found that the inhibition of oocyte
development due to the consumption of gelsemine was related to a pollinator’s condition.
Despite its acute toxicity to mammals, gelsemine seems to be distasteful but largely
harmless to bees, except if they are consuming the highest natural concentrations, have
little other food to dilute the toxic effects, or are metabolically challenged. These
circumstances might apply to bumble bee queens foraging on the early spring flowers of
G. sempervirens. Queens that have recently emerged from hibernation are developing
their ovaries to begin nestmaking, and often have few other nectar and pollen resources to
choose from. They may be ingesting substantial amounts of gelsemine-rich nectar while
highly food-stressed and therefore vulnerable to the deleterious consequences of
gelsemine. Even a slight sublethal effect of nectar alkaloids may present a subtle but
significant impediment to pollinator fitness. Whether that impediment offsets the
positive value of the nectar sugars obtained would depend on the dietary choices
available. The role that nectar secondary metabolites play in plant-pollinator
communities will therefore be shaped by the composition of each community, and future
work needs to move beyond the interactions of a single plant and pollinator to include
more complex, community-level interactions. Although “toxic” nectar may be less
45
severe than its name suggests, its deleterious effects still have the potential for
widespread consequences to pollinators and the plants they visit.
Acknowledgements
We would like to thank Kate Edwards and Tamryn Ah-Long for their assistance, and two
anonymous reviewers for comments on the manuscript. This research was funded by
NSERC.
46
Table 3.1. The effect of zero, moderate (50 ng/µL) and high (250 ng/µL) gelsemine on oocyte length and width, on dominant and
subordinate bees. All analyses are ANCOVAs with Type III SS and radial cell length (a proxy for bee size) as a covariate. Significant
effects of the gelsemine treatment are in bold.
Oocyte Length
Oocyte Width
Source
df
SS F P df SS F P
Dominant Bees
Treatment 2 0.01 0.39 0.68 2 0.01 1.60 0.21
Radial Cell 1 0.09 4.98 0.03 1 0.02 4.87 0.03
Subordinate Bees
Treatment 2 0.04 1.57 0.22 2 0.03 4.80 0.01
Radial Cell
1 0.12 10.11 <0.001 1 0.02 6.92 0.01
47
Table 3.2. Daily pollen consumption for six day microcolony assay, compared between
the three gelsemine treatments using a repeated measures ANOVA. Note that because
bees were raised in pairs, each measurement represents pollen consumed for one
dominant and one subordinate bee.
Source
df SS F P
Between Subjects
Treatment 2 8.96 1.53 0.29
Day 1 0.41 0.14 0.72
Within Subjects
Treatment 2 7.48 3.74 0.25
Day 5 181.4 13.36 <0.001
Treatment x Day
10 20.10 0.74 0.69
48
Figure 3.1. Mean oocyte size, plotted as length against width, in dominant and
subordinate bees fed 0 ng/µL (closed circles), 50 ng/µL (open circles) or 250 ng/µL
(closed inverted triangle) gelsemine. Dominant bees have larger oocytes and clump
together in the upper right corner of the graph, while the smaller oocytes of subordinate
bees fall in the lower left corner. The graph indicates the SE of both length and width
measurements.
49
Figure 3.2. Median carbohydrate concentrations in haemolymph 24 hours after bees
consumed artificial nectar with 0 ng/µL, 50 ng/µL, or 250 ng/µL gelsemine.
Carbohydrates are expressed as micrograms of trehalose equivalents per microlitre
haemolymph. The boxes represent the 25th
and 75th
percentiles, the whiskers indicate
roughly two standard deviations and filled circles represent outliers.
50
CHAPTER FOUR
Consumption of a nectar alkaloid reduces pathogen load in bumble bees
Jessamyn S. Manson*, Michael C. Otterstatter* and James D. Thomson
*equally contributing authors
M.C. Otterstatter and I contributed equally to the design and execution of this project and the writing of the
manuscript. J.D. Thomson provided comments on the paper, which is in press at Oecologia.
Abstract
Floral nectar is produced to attract pollinators, but frequently contains secondary
metabolites that are often distasteful and toxic. One hypothesized function of this so-
called “toxic” nectar is that it has antimicrobial properties, which could prevent nectar
from spoiling. Microbicidal nectar may also benefit insect pollinators by reducing the
infection intensity of pathogens. We tested whether gelsemine, a nectar alkaloid of the
bee-pollinated plant Gelsemium sempervirens, could reduce pathogen loads in bumble
bees infected with the gut protozoan Crithidia bombi. In our first laboratory experiment,
artificially infected bees consumed a daily diet of gelsemine post-infection to simulate
continuous ingestion of alkaloid-rich nectar. In the second experiment, bees were
inoculated with C. bombi cells that were pre-exposed to gelsemine, simulating the direct
effects of nectar alkaloids on pathogen cells that are transmitted at flowers. Gelsemine
significantly reduced the fecal intensity of C. bombi seven days after infection when it
was consumed continuously by infected bees, whereas direct exposure of the pathogen to
51
gelsemine showed a non-significant trend towards reduced infection. Lighter pathogen
loads may relieve bees from the behavioural impairments associated with the infection,
thereby improving their foraging efficiency. If the collection of nectar secondary
metabolites by pollinators is done as a means of self-medication, pollinators may
selectively maintain this unusual trait in natural plant populations.
Introduction
At first glance, the function of plant-produced secondary metabolites seems
straightforward: these distasteful and often toxic compounds defend plants from
herbivores (Rosenthal and Berenbaum 1991). However, identical secondary metabolites
are also found in the floral nectar of plants, which is paradoxical given that floral nectar is
usually interpreted as attractive, not deterrent, to pollinators. Secondary metabolites,
including tannins, phenols, alkaloids and terpenes, have been found in floral nectar across
21 angiosperm families (Adler 2000). The prevalence and diversity of secondary
compounds across the angiosperms suggests that this so-called “toxic” nectar has some
adaptive function for plants. Hypothesized functions of secondary metabolites in nectar
include deterrence of nectar robbers, increased constancy of effective pollinators, or
protection against deleterious microbes (see Rhoades and Bergdahl 1981, Adler 2000 for
full review). Of these, the antimicrobial hypothesis is perhaps the most general because
microbes are ubiquitous and nectar is an otherwise ideal medium to support a variety of
microorganisms that can harm plants and deter pollinators. Although many secondary
metabolites have microbicidal properties (Cowan 1999), and diverse microorganisms
often occur in floral nectar (Ehlers and Olesen 1997, Golonka 2002, Brysch-Herzberg
52
2004), few studies have tested whether nectar secondary metabolites actually suppresses
microbes (but see Manson et al. 2007).
For plants that rely on animal vectors for pollination, microbial contamination of
nectar can be costly. Spoilage due to yeast, fungi, and bacteria can affect nectar
palatability, leading to changes in pollinator behaviour (Ehlers and Olesen 1997) and
potentially reducing pollen transfer. Microbes may also affect pollination by disrupting
processes like pollen tube growth (Kevan et al. 1989) or stigmatic receptivity (Buban and
Orosz-Kovacs 2003). In some cases, the pollinators themselves cause nectar
contamination by transporting plant venereal diseases that destructively infect floral
structures, reducing seed set through damaged pollen and ovules (Buban and Orosz-
Kovacs 2003, Antonovics 2005). Allocating secondary metabolites to nectar might
reduce microbial growth, thereby protecting plants from pathogens and maintaining the
quality of their rewards.
Antimicrobial properties of this so-called “toxic” nectar may also benefit nectar-
collecting pollinators. Gathering nectar exposes pollinators to a variety of pathogenic
microorganisms that can reduce their survival and foraging efficiency. For example,
bumble bees in Europe and North America frequently carry the intestinal protozoan
Crithidia bombi (Lipa and Triggiani 1988, Schmid-Hempel 2001, Colla et al. 2006),
which elevates their mortality rate under food stress (Brown et al. 2000) and impairs their
associative learning, flower handling, and foraging efficiency (Gegear et al. 2005,
Otterstatter et al. 2005, Gegear et al. 2006). Horizontal transmission of C. bombi occurs
at flowers, when infected bees deposit ‘free-living’ pathogen cells that are subsequently
ingested by susceptible foragers (Durrer and Schmid-Hempel 1994). Since C. bombi is
53
known to occur in the nectar of wild flowers (Durrer and Schmid-Hempel 1994), nectar
secondary metabolites may influence the survival and infectivity, and consequently the
transmission, of this pathogen. Furthermore, for gut pathogens such as C. bombi, host
diet can significantly effect the severity of infection by altering immunocompetence,
metabolic processes, or by limiting nutrient availability for the parasite (Wink and Theile
2002, Logan et al. 2005, Cory and Hoover 2006). Hence, in flowers or in the guts of
flower visitors, “toxic” nectar may benefit pollinators via antimicrobial action.
The nectar of the Carolina jessamine (Gelsemium sempervirens L.) contains the
indole alkaloid gelsemine, a secondary metabolite that is highly toxic to vertebrates
(Blaw et al. 1979). Gelsemine appears to have little effect on the fitness or physiology of
bees (Elliott et al. 2008, Manson and Thomson 2009) and no effect on non-pathogenic
floral yeasts (Manson et al. 2007). Although gelsemine-rich nectar can be distasteful and
deterrent to pollinators (Adler and Irwin 2005, Gegear et al. 2007), G. sempervirens
flowers consistently attract a number of floral visitors in nature, including bumble bees.
In the present study, we examined the putative antimicrobial properties of the
nectar alkaloid gelsemine on the bumble bee pathogen Crithidia bombi. First, we asked,
does consumption of alkaloid-rich nectar by bumble bees reduce the severity of intestinal
infections by C. bombi? Second, given that this pathogen is naturally transmitted at
flowers, we asked, does alkaloid-rich nectar directly reduce the infectivity of C. bombi
cells? We then discuss the ecological impact of nectar alkaloids on pollinator-pathogen
dynamics and the potential for pollinators to selectively maintain secondary metabolites
as a natural microbicide.
54
Methods
Our experimental protocol is illustrated in Figure 4.1. In both experiments, we
exposed the pathogen C. bombi to either a gelsemine solution (Alkaloid) or plain sucrose
solution (Control) and then compared the intensity of developing infections in inoculated
bumble bees (Bombus impatiens Cresson). In Experiment 1, bees were first inoculated
with C. bombi and then fed on a daily diet of Alkaloid or Control solution. In Experiment
2, we exposed C. bombi cells to Alkaloid or Control solutions for varying durations
before inoculating bees.
We made artificial alkaloid-rich ‘nectar’ by mixing gelsemine hydrochloride
(purchased from Chromadex, Irvine, CA, hereafter referred to as gelsemine) into a 30%
w/w aqueous sucrose solution. The concentration of gelsemine was 250 ng/µL , which
represents the highest naturally occurring concentration of gelsemine reported (Adler and
Irwin 2005). Alkaloid solutions were refrigerated at 4°C when not in use and stored for
up to two days, although they were usually prepared immediately before use.
We prepared pathogen inocula from the gut tracts of four ‘donor’ B. impatiens
workers from each of five hives infected by C. bombi (provided by a commercial rearing
company). Following the general protocol of Otterstatter and Thomson (2006), gut tracts
were excised and crushed in a microcentrifuge tube containing 300 µL of distilled water.
The mixture was allowed to settle at room temperature for three hours, after which the
supernatant was removed and mixed thoroughly. Supernatants were diluted to the
appropriate density of C. bombi cells (Neubauer haemocytometer counts) and sucrose
was added to a concentration of 30%. In each of the two experiments (described below),
we used 20 new donor bees from five new hives; thus, within experiments, all bees
55
received the same cocktail of C. bombi strains (genotypes), but between experiments,
inocula may have contained different pathogen genotypes.
We obtained susceptible ‘recipient’ B. impatiens workers from pupal clumps
originating from commercially reared hives (same supplier as above). Previous studies
have found that Crithidia infections are not acquired until workers emerge (Otterstatter
and Thomson 2007), making new workers naïve to Crithidia regardless of the infection
status of the source colony. Newly emerging (< 24 hours old) worker bees were placed
in containers according to their hive of origin and given 30% sucrose solution and pollen
ad libitum. After two days, workers were starved overnight, weighed (± 0.1 mg), and
then arbitrarily assigned to an experimental group. We ensured that each of the Alkaloid
and Control groups in both experiments contained recipient bees from at least three hives,
in roughly equal numbers.
In Experiment 1, ‘Continuous Exposure’, bees inoculated with C. bombi were
allowed to feed daily on gelsemine, simulating the continual ingestion of nectar
constituents by an infected foraging bee. Each bee was initially fed a 2 µL drop
containing 104 C. bombi in 30% sucrose solution and we monitored individuals until the
entire drop was consumed. This dose falls within the range of C. bombi cells shed in the
faeces of infected bees in previous studies (Schmid-Hempel and Schmid-Hempel 1993,
Logan et al. 2005), and therefore simulates cells available for transmission to naïve
individuals. Bees were reared in individual 15 mL vials and received either a 0.5 mL
solution of 250 ng/µL gelsemine in 30% sucrose (Alkaloid bees, n = 35) or 0.5 mL of
30% sucrose only (Control bees, n=35) along with a pollen lump daily for 10 days.
56
In Experiment 2, ‘Delayed Exposure’, C. bombi was exposed to gelsemine for
various durations prior to host ingestion, simulating direct exposure of the pathogen to
nectar in a flower. We placed 104 C. bombi (in 2 µL of 30% sucrose solution) into each
of 60 microcentrifuge tubes: 30 of these contained 8 µL of a 250 ng/µL solution of
gelsemine in 30% sucrose (Alkaloid), and 30 contained 8 µL of 30% sucrose only
(Control). In the ‘Immediate’ group, we fed the Alkaloid and Control pathogen mixtures
to recipient bees immediately; each bee was housed individually and received only one
dose, yielding 10 Alkaloid bees and 10 Control bees. In the ‘1 hr Delay’ and ‘2 hr Delay’
groups, we left the Alkaloid and Control pathogen mixtures at room temperature (~ 21-
24°C) under fluorescent lighting for one and two hours, respectively, before feeding them
to recipient bees (as before, 10 Alkaloid bees and 10 Control bees per group). We
selected these two delay treatments to simulate the time delay between the deposition of
C. bombi cells by infected bees and the next flower visit by a naïve bee. In this
experiment, we compensated for evaporative water loss by starting with more dilute
sucrose solutions that evaporated to a concentration of 30% sucrose after one or two
hours (dilutions calculated from a preliminary study). Following the inoculation with C.
bombi, bees were kept in individual vials and given 0.5 mL of 30% sucrose solution and a
fresh pollen lump daily.
In both experiments, we quantified infection intensities of all bees at day 7 and
day 10 post-inoculation, as these days delimit the period in which pathogen load is
saturated (Schmid-Hempel and Schmid-Hempel 1993, Otterstatter and Thomson 2006)
On day 7, all bees were transferred to clean vials without food and left until they
defecated. The density of C. bombi in each bee’s feces was determined with a
57
haemocytometer. On day 10, all bees were sacrificed and the total density of C. bombi in
their gut tracts was determined with a haemocytometer following Otterstatter and
Thomson (2006).
Statistical analysis
Our final sample sizes were lower than the original design due to mortality from
unknown causes (9% of bees died before day 10; subsequent examinations did not reveal
unusually intense C. bombi infections), missing fecal samples (17% of bees did not
produce enough feces for analysis on day 7 post-inoculation), and the failure of certain
bees to develop an infection (10% of bees remained uninfected throughout the
experiment). We excluded all of these bees from further analyses. Likelihood ratio (G)
tests showed that, in each case, the proportion of ‘excluded bees’ did not differ between
Control and Alkaloid groups (P > 0.20 in all cases), suggesting that these were not
serious sources of bias. In total, we analyzed the infection intensities of 76 bees for day 7
(fecal counts), and 102 bees for day 10 (gut counts).
Given their differing designs, Experiments 1 and 2 were analyzed separately. In
both cases, we used multiple regression analysis, with repeated measures on bees (to
account for the non-independence of observations on the same individual at day 7 and
day 10), to determine whether or not gelsemine reduced the intensity of gut infections. In
order to directly compare a bee’s intensity of infection at day 7 (measured as C. bombi
cells /µL host feces) and day 10 (C. bombi cells /µL of gut fluid) post-inoculation, fecal
counts were converted to estimated gut counts using the linear regression, gut count =
-6.3455 + 0.6955 x feces count (F1,41 = 268.93, P < 0.001, R2 = 0.93) based on data in
Otterstatter and Thomson (2007). We treated pathogen counts (square-root transformed
58
to satisfy the standard assumption of normally-distributed errors) as our dependent
variable, and whether or not C. bombi was exposed to gelsemine (‘Alkaloid’ or ‘Control’
group), time (day 7 or day 10 post-inoculation), and bee body mass, as explanatory
factors in our analyses. In Experiment 2, we also included ‘Delay’ as an explanatory
factor, i.e., the duration that C. bombi was exposed to gelsemine prior to host inoculation
(no delay, 1 hr delay, 2 hr delay). Preliminary analyses showed that infection intensity in
Control and Alkaloid groups did not meet the standard assumption of homoscedasticity
(F-test for equal variance, day 7: F = 3.96, P = 0.004; day 10: F = 2.07, P = 0.067); we
therefore used a heterogeneous variance model (Proc MIXED, SAS Institute 2006) to
account for this deviation. For both experiments, we began with a saturated model and
removed non-significant effects via backward stepwise elimination. Akaike information
criterion (AIC) values were used to compare candidate models; ultimately, the model
with the lowest AIC value was chosen as the best fit to the data. We used linear contrasts
(t-tests) in our regression models to compare the average infection intensities of Control
and Alkaloid bees at day 7 and day 10. Finally, we used Kolmogorov-Smirnov (K-S)
two-sample tests to compare the distributions of infection intensities between Control and
Alkaloid bees; P-values for K-S tests were computed using Monte Carlo estimation (Proc
NPAR1WAY, SAS Institute 2006).
Results
In Experiment 1, an alkaloid-rich diet reduced the intensity of C. bombi infections
in bumble bees. Our regression analysis revealed significant main effects of gelsemine
(Alkaloid or Control diet), time since inoculation, and bee body size on infection intensity
(Table 4.1). At 7 days post-inoculation, bees receiving dietary gelsemine had infections
59
that were, on average, 2.2 times less intense than bees receiving the control diet (t = 2.45,
df = 36, P = 0.019; Fig. 4.2). Indeed, gelsemine completely prevented heavy infections in
bees by day 7: whereas infections in the Control group ranged from 0 – 51500 cells/µL,
the most intense infection in the Alkaloid group was only 5300 cells/µL (significantly
different distributions of infection intensity, K-S test: D = 0.42, P = 0.02). Infection
intensities increased significantly from day 7 to day 10, and this effect did not differ
between Control and Alkaloid groups (non-significant ‘Gelsemine x Time’ effect, Table
4.1). At 10 days post-inoculation, although average infection intensities were similar in
Alkaloid and Control groups (t = 0.96, df = 36, P = 0.342; Fig. 4.2), the distribution of
infection intensities was skewed to significantly lighter infections among bees receiving
gelsemine compared to bees receiving the control diet (K-S test: D = 0.35, P = 0.02). For
example, while Alkaloid and Control bees exhibited similar ranges of infection intensity
at day 10 (150 – 39 188 cells/µL vs. 150 – 35 250 cells/µL, respectively), the median
infection intensity of Alkaloid bees was less than half that of Control bees (4775 cells/µL
vs. 10850 cells/µL, respectively). Overall, in Experiment 1, larger bodied bees developed
lighter infections than small bees, independently of alkaloid treatment (Table 4.1; Fig.
4.3).
Given that a continuous diet of gelsemine reduced infection intensity in bumble
bees, we asked in Experiment 2 if exposing C. bombi cells to gelsemine prior to host
inoculation would also reduce infections. Exposing C. bombi inocula to gelsemine did
not have a clear effect, however. Gelsemine did not significantly reduce average
infection intensity (non-significant ‘Gelsemine’ effect, Table 4.2), nor did the distribution
of infection intensities differ between Control and Alkaloid groups for any of the
60
treatments (K-S tests: P > 0.60 in all cases). Average infection intensity increased
significantly over time (from day 7 to day 10) when C. bombi was fed to bees
immediately (t = 3.40, df = 41, P = 0.002; Fig. 4.4a), but this effect decreased when
pathogen cells sat for one hour before inoculation (t = 1.91, df = 41, P = 0.064; Fig. 4.4b),
and disappeared when pathogen cells sat for two hours before inoculation (t = 1.60,
df = 41, P = 0.118; Fig. 4.4c) (significant ‘Delay x Time’ interaction, Table 4.2). There
was no indication that gelsemine affected this variation in infection intensity over time in
any of the three experimental treatments (non-significant ‘Gelsemine x Delay x Time’
interaction, Table 4.2).
Discussion
Insect pollinators regularly feed from flowers that contain alkaloid-rich nectar but
the consequences of such nectar for pollinators and plants remain unclear (Adler 2000).
Our results demonstrate for the first time that artificial nectar containing a naturally
occurring nectar alkaloid reduces the severity of gut infections in pollinators. Bumble
bees (Bombus impatiens) inoculated with the intestinal parasite Crithidia bombi
developed less intense infections when feeding on the alkaloid gelsemine for several days
(Fig. 4.2). In particular, the distribution of infections differed substantially between
treatments, with most gelsemine-consuming bees experiencing far lighter infections than
the control bees. However, the infectivity of C. bombi inocula was unaffected when
pathogen cells were exposed to gelsemine outside of the host. These results suggest that
alkaloid-rich nectar can act as a microbicide against a protozoan pathogen of pollinators
when ingested, but does not directly interfere with pathogen viability. Given that
C. bombi is deposited at flowers by infected foragers, and can spread between bees via
61
contaminated floral nectar, alkaloid-rich nectar could have substantial effects on the
transmission of this pathogen both within the hive and through bumble bee populations.
Our experiments were conducted under laboratory conditions, which facilitated
the careful manipulation of both alkaloid and pathogen. However, the artificiality of the
lab may also have limited certain aspects of our study. Natural floral nectar is rarely as
simple as the artificial nectar that we used; thus, our experiment may have eliminated
some of the subtle interactions between nectar components and Crithidia bombi.
Similarly, bees may not forage on a single nectar source continuously for ten days, as we
simulated in Experiment 1. Nevertheless, Gelsemium sempervirens flowers very early in
the spring (Pascarella 2007) and thus represents one of the few nectar sources for early
emerging bumble bees. Finally, by isolating bees in individual vials, we may have
disrupted important aspects of infection dynamics that naturally occur within hives, such
as the exchange of pathogen cells and strains among nestmates.
The effects of plant secondary metabolites on host-pathogen interaction are
understudied and poorly understood (Price et al. 1980, Cory and Hoover 2006). Plant-
derived alkaloids appear to have anti-protozoal properties that are effective against
human parasites, such as Trypanosoma brucei rhodesiense, the causative agent African
sleeping sickness (Freiburghaus et al. 1996). In bumble bees, the plant-derived alkaloid
gelsemine appears to have similar anti-protozoal effects on C. bombi, another
trypanosome parasite. Although the underlying mechanism is not yet clear, it may be that
when a host’s gut contains substantial concentrations of alkaloids, C. bombi cells suffer
reduced growth and replication because of costs associated with alkaloid tolerance.
Similar reductions in pathogen proliferation have been reported for secondary metabolite-
62
tolerant plant pathogens (Vanetten et al. 2001). Alternatively, the consumption of
alkaloids might alter the host’s gut environment, making it less hospitable for pathogen
cells. Logan et al. (2005) proposed this mechanism after pollen consumption altered the
rate at which C. bombi populations increase within hosts, perhaps by affecting their
adherence to the gut wall. Consumption of alkaloids may also increase gut pH, which
could be deleterious to pathogen cells (Stiles and Paschke 1980). Finally, an alkaloid-
rich diet may increase a bee’s excretion rate, effectively ‘flushing’ C. bombi cells from
the gut wall. Indeed, animals that consume secondary metabolites often deal with the
inherent toxicity through rapid excretion (Wink and Theile 2002, Despres et al. 2007),
and alkaloid-rich nectar in particular has been shown to increase excretion rates in a
nectarivorous bird (Tadmor-Melamed et al. 2004). Gelsemine does not, however, seem to
hinder a bee’s immunocompetence towards C. bombi, since this would result in a pattern
opposite to what we observed, i.e., higher levels of infection in the gelsemine-consuming
bees.
The impact of nectar secondary metabolites on pathogens could be ecologically
significant both to bumble bees and the plants they pollinate. Although C. bombi is often
considered a benign pathogen (Schmid-Hempel 1998), it renders foragers less able to
provide food for their colonies. For example, infected workers have reduced foraging
rates, a decreased capacity to learn floral cues, and difficulty manipulating complex
flowers (Gegear et al. 2005, Otterstatter et al. 2005, Gegear et al. 2006). The severity of
these impairments increases with infection intensity, so although dietary gelsemine does
not appear to cure C. bombi infections, it could curtail the adverse effects of the pathogen
on host behaviour by reducing infection intensity. Bumble bee queens might derive the
63
greatest benefit from “toxic” nectar. In the spring, queens that emerge from hibernation
harbouring C. bombi are less likely to found a colony than healthy queens (Brown et al.
2003). It is possible that a gelsemine-rich diet would suppress an infected queen’s
pathogen load to the extent that she could establish a viable colony. In the south-eastern
United States, Bombus impatiens and B. bimaculatus queens often collect alkaloid-rich
nectar from Gelsemium sempervirens in the spring (Manson, personal observation);
whether or not these queens receive a ‘medicinal’ benefit from this nectar is an important
topic for further study. The medicinal properties of “toxic” nectar might also have
consequences for plant communities, as parasitic infections are known to alter pollen
collection (Schmid-Hempel and Schmid-Hempel 1991) and plant species choice
(Schmid-Hempel and Stauffer 1998) in bumble bees.
Our demonstration that gelsemine can mitigate infections raises the possibility
that infected bees might actively self-medicate. There is mounting evidence that infected
insects alter their foraging strategies in order to fight pathogens. Some insects adjust
basic nutrient intake to improve their overall immune response (Lee et al. 2006), whereas
others actively seek compounds that have antimicrobial properties. The active collection
of non-nutritive secondary metabolites, or ‘pharmacophagy’ (Boppre et al. 2005), is often
associated with a significant shift in diet. For example, parasitoid-infested Platyprepia
virginalis caterpillars preferentially consume alkaloid-rich hemlock instead of lupine,
their primary host plant, in field choice experiments (Karban and English-Loeb 1997).
Parasitized Grammia geneura caterpillars also choose a mixed diet of plants rich in
secondary metabolites rather than a nutrient-rich, but toxin-poor, single-plant diet (Singer
et al. 2004). In fact, Singer et al. (2009) elegantly demonstrated that G. geneura self-
64
medicate with pyrrolizidine alkaloids to reduce parasite infections and increase caterpillar
survival, despite the fact that the alkaloid reduces fitness in unparasitized individuals. The
preferential consumption of secondary metabolites in parasitized G. geneura is caused by
an increase in the firing rates of the animal’s taste receptors, which results in increased
consumption of pyrrolizidine alkaloids (Bernays and Singer 2005), although the
generality of this mechanism is unknown. Several social insect species, including wood
ants and honey bees, are known to collect antimicrobial resins to prevent microbe growth
within their hives (Konig 1988, Marcucci 1995, Christe et al. 2003, Chapuisat et al.
2007). In the current study, we did not allow bees to choose their diet, so we were unable
to test for a gelsemine preference amongst infected individuals. We know of no reports
of self-medication by pollinators; the possibility warrants study.
Plants experience multidirectional selection on secondary metabolite
concentrations. Strong chemical defenses that reduce herbivory may also reduce
pollinator attraction, unless secondary compounds confer a fitness benefit to either plants
or pollinators via reduced pathogen loads. Indeed, Price et al. (1980) proposed that the
very multifunctional nature of plant defenses may shape the concentration of plant
secondary metabolites. If pollinators benefit from, and even seek out, nectar rich in
secondary metabolites, selection on plants to decrease alkaloid compounds in nectar may
be minimal, and potentially even countered by stabilizing selection from pollinators
(Clayton and Wolfe 1993). Few studies have attempted to tease apart the various forces
that select for or against “toxic” nectar in plants (but see Irwin et al. 2004); however, the
potential impact of this unusual trait on both plant and pollinator fitness suggests that it
merits further investigation.
65
Acknowledgements
We would like to thank Nathan Muchhala and Mario Vallejo-Marín for comments on the
manuscript. This study was supported by grants from the Natural Sciences and
Engineering Council (NSERC). All experiments complied with current laws of Canada.
66
Table 4.1. Experiment 1: Mixed model statistics describing the effect of an alkaloid-rich
diet on the intensity of Crithidia bombi infections in bumble bees. ‘Bee’ was included in
each model as a repeated factor to account for the non-independence of sequential
observations on individuals. Bees were inoculated with pathogen cells and then fed a
daily diet of either alkaloid or control solution (‘Gelsemine’). Pathogen counts were
done at 7 and 10 days post-inoculation (‘Time’). Numerator and denominator degrees of
freedom (df) are shown for each explanatory factor.
Explanatory Factor
F df P
Gelsemine 4.65 1,57 0.035
Time 32.97 1,36 <0.001
Bee Body Size 7.61 1,57 0.008
Gelsemine x Time
0.88 1,36 0.36
67
Table 4.2. Experiment 2: Mixed model statistics describing the effect of exposing
Crithidia bombi cells to gelsemine for varying durations prior to bumble bee inoculation
on infection intensity. ‘Bee’ was included in each model as a repeated factor to account
for the non-independence of sequential observations on individuals. Pathogen inocula
were mixed with either an alkaloid or control solution (‘Gelsemine’) and fed to bees
immediately, or after either a 1 hr or 2 hr delay (‘Delay’). Pathogen counts were done at
7 and 10 days post-inoculation (‘Time’). Numerator and denominator degrees of freedom
(df) are shown for each explanatory factor.
Explanatory Factor
F df P
Gelsemine 0.11 1,51 0.74
Delay 10.48 2,51 <0.001
Time 3.74 1,41 0.06
Gelsemine x Delay 0.1 2,51 0.91
Gelsemine x Time 1.1 1,41 0.3
Delay x Time 6.15 2,41 0.005
Gelsemine x Delay x Time
1.06 2,41 0.36
Bee body size was non-significant and excluded from the final model (F = 1.53, df =
1,51, P = 0.22)
68
Figure 4.1. Diagram of the experimental design. Throughout, ‘Alkaloid’ refers to a
30% sucrose solution containing 250 ng/µL of the alkaloid gelsemine and ‘Control’ refers
to a 30% sucrose-only solution. In Experiment 1, bees harbouring the gut pathogen
Crithidia bombi were fed a daily diet of Alkaloid or Control solution, whereas in
Experiment 2, bees were fed C.bombi cells that had bee pre-exposed to Alkaloid or
Control solutions for varying durations.
69
Figure 4.2. Experiment 1: Effect of an alkaloid-rich diet on the intensity of Crithidia
bombi infections in bumble bees. Bees were inoculated with a standard dose of pathogen
cells and then fed a daily diet of either a gelsemine or control solution. The lower and
upper edges of each box indicate the 25th
and 75th
percentiles, respectively, the solid and
dashed lines within a box indicate the median and mean values, respectively. Error bars,
where visible above and below a box, indicate the 90th and 10th percentiles, respectively.
Infection intensities have been square-root transformed.
70
Figure 4.3. Relationship between intensity of infection by Crithidia bombi and bumble
bee body size in Experiment 1. Each point represents a bee’s infection intensity (adjusted
value, from the statistical model shown in Table 4.1) at 7 or 10 days post-inoculation.
71
Figure 4.4. Experiment 2: Effect of exposing Crithidia bombi cells to the alkaloid
gelsemine for varying durations prior to bumble bee inoculation. Boxes are as described
in Figure 4.2. Pathogen inocula were mixed with either a gelsemine or control solution
and fed to bees (a) immediately, (b) after a 1 hr delay, or (c) after a 2 hr delay.
72
CHAPTER FIVE
Cardenolide concentrations of nectar, leaves and flowers:
A comparative study across Asclepias series Incarnatae
Jessamyn S. Manson, Sergio Rasmann, Rayko Halitschke, James D. Thomson
and Anurag A. Agrawal
This project was designed in collaboration with A.A. Agrawal. I completed the chemical analyses with the
assistance of S. Rasmann and R. Halitschke, while I executed the behavioural assays alone. I received
substantial statistical assistance from both A.A. Agrawal and J.D. Thomson, and J.D. Thomson also
assisted in writing this chapter.
Abstract
Secondary metabolites are usually thought to deter insects, so their occurrence in
floral nectar is surprising. These compounds may have been selected for particular roles
in nectar, or they may occur there non-adaptively as a passive consequence of systemic
chemical defense of other tissues. Finding the same relative abundances of the same
compounds in nectar and leaves would support the “consequence-of-defense” hypothesis,
whereas finding unique profiles of compounds in nectar would suggest some adaptive
role. We quantified cardenolides in the nectar, leaves and flowers of twelve species from
the Asclepias series Incarnatae to test this “consequence-of-defense” hypothesis. We
collected samples from a large greenhouse collection and used HPLC to determine
cardenolide concentrations. Using the nectar data, we developed behaviour assays to
73
assess the ecological consequences of a representative cardenolides on a bumble bee, an
important Asclepias pollinator. We found that nectar, leaf and flower cardenolides varied
substantially among Asclepias species, but within species there were positive correlations
in cardenolide concentration between the plant components. Nectar had fewer individual
cardenolides than leaves and, within species, nectar compounds were generally a subset
of leaf compounds, supporting our non-adaptive hypothesis. A multivariate analysis of
the concentrations of individual cardenolides also found that leaves and nectar of the
same species were similar and that this similarity was independent of evolutionary
history. Surprisingly, bumble bees had an overall preference for nectar with average
concentrations of the cardenolide digoxin, suggesting that nectar cardenolides may
contribute to rewarding and retaining pollinators. With some exceptions, our data
generally support the consequence-of-defense hypothesis, but imply that “toxic” nectar
may not deter pollinating bees. This study is the first to identify and quantify cardenolides
in Asclepias nectar and to explore their effects on pollinators.
Introduction
To defend themselves against herbivores, many plants produce noxious
compounds that are distasteful, deterrent and often deleterious (Rosenthal and
Berenbaum 1991). Although defensive secondary metabolites have been identified and
quantified in the leaves of a vast number of plant species, plants rarely limit the
distribution to of these compounds to leaves. Secondary metabolites are also common in
roots, stems and flowers (Van der Putten et al. 2001, McCall and Irwin 2006), where
defense against herbivores remains a plausible function, but their function in other plant
parts such as nectar remains uncertain (Adler 2000). An accurate characterization of
74
plant chemical defense requires a whole-plant perspective that compares concentrations
of secondary metabolites between plant “components”, by which we include both organs
and exudates such as floral nectar. Given that resources are finite, differences in the
allocation of secondary metabolites between plant components should relate to both the
relative importance and vulnerability of that structure to herbivory (McKey 1974).
Although this proposition has been addressed by recent work comparing above- and
below-ground chemical defenses (Van der Putten et al. 2001, Bezemer and van Dam
2005, Rasmann et al. In Press), substantially less work has been conducted on floral
structures and floral rewards. Reports of positive correlations between the concentrations
of secondary metabolites in leaves and floral nectar (Adler et al. 2006) provide the first
empirical evidence that so-called “toxic” nectar may be due to the systemic production of
defensive chemicals.
The presence of secondary metabolites in nectar is a widespread if paradoxical
phenomenon, reported in at least 21 angiosperm families (Adler 2000). When
investigators have assessed the consequences for animal pollinators of imbibing these
“toxic” nectars, the effects range from putative nicotine and caffeine addiction in honey
bee colonies (Singaravelan et al. 2005) to lethal poisoning of honey bee workers after the
consumption of a mere 10 µL of nectar from Sophora microphylla (Clinch et al. 1972).
Adaptive hypotheses for nectar-specific functional roles for secondary metabolites are
largely derived from the herbivory literature and include deterring inefficient pollinators
and nectar robbers (Rhoades and Bergdahl 1981, Adler 2000). Alternatively, secondary
metabolites in nectar may arise from leakage of compounds into the nectary during
vascular transport or from the systemic production of phytochemicals, making their
75
presence in nectar an undesirable consequence of defense (sensu Adler 2000). Finding
correlations between the concentration and identity of compounds in nectar and other
plant components would support this less adaptive explanation, whereas finding
compounds unique to nectar would suggest that nectar secondary metabolites have
functional significance (e.g. phylogenetic constraint vs. adaptive function; Rhoades and
Bergdahl 1981, Strauss et al. 1999, Adler et al. 2006). The best plant system for
evaluating these functional hypotheses would therefore possess defense chemicals that
have been identified and quantified across a number of plant structures as well as a
collection of pollinators that vary in efficiency and may thus select for or against nectar
secondary metabolites.
Milkweeds (Asclepias spp.) are a classic system for studying plant chemical
defenses and their effects on animals. Milkweeds have evolved a number of strategies to
protect themselves against a suite of generalist and specialist herbivores, including both
physical and chemical defenses (Agrawal and Fishbein 2006). The most potent
secondary metabolites found in milkweeds are cardenolides, bitter-tasting compounds
that elicit aversive or emetic responses in both vertebrates (Brower et al. 1968) and
invertebrates (Dussourd and Hoyle 2000). Cardenolides affect animals by inhibiting
sodium pumps, thereby altering sodium and potassium transportation in cells (Malcolm
1991). Cardenolides are often considered a “qualitative” defense, because they are
poisonous to generalists at very low doses, yet they can be tolerated and even co-opted by
specialists (Feeny 1976, Agrawal and Fishbein 2006). Herbivore damage can increase
cardenolide concentrations in some species of Asclepias, but not in others (Rasmann et
al. In Press).
76
The concentrations of cardenolides in leaves vary significantly across the genus
Asclepias (Agrawal and Fishbein 2006, Agrawal et al. 2008) and can also vary among
plant components such as the leaves, roots, pith and epidermis within the same plant
(Nelson et al. 1981, Fordyce and Malcolm 2000, Rasmann et al. In Press). Milkweeds
are pollinated by a diverse collection of generalists such as lepidoptera, honey bees and
bumble bees (Wyatt and Broyles 1994), but there are substantial differences in pollinator
efficiency between guilds (Fishbein and Venable 1996, Kephart and Theiss 2004). Given
that Asclepias nectar can apparently be toxic to bees (Pryce-Jones 1942), the milkweeds
are a good candidate for examining consequences of secondary metabolites in nectar.
In this study, we use the genus Asclepias to explore the functional significance of
nectar cardenolides. We quantified cardenolides in the nectar, leaves and flowers of
twelve species from the series Incarnatae, a monophyletic group that varies in a number
of key traits, including leaf cardenolide concentration (Agrawal et al. 2008). We then
used these data to evaluate the “consequence-of-defense” hypothesis. We interpret
similarity of cardenolide occurrence in different plant components as supporting a non-
adaptive hypothesis such as leakage or a systemically controlled chemical defense
strategy; differences in cardenolide profiles among plant components may indicate
compartmentalized regulation of cardenolide production, with the possibility of such
regulation being adaptive. We propose that similarities in individual cardenolides across
plant components are more informative than similarities in gross cardenolides, as
overlaps in cardenolide identity between leaves and nectar support a shared chemical
production site, while correlations in gross cardenolides may mask significant differences
in cardenolide composition. Thus, we specifically asked: 1) Are there positive
77
correlations between cardenolide concentrations in different plant components?, 2) Do
cardenolides in nectar have an independent source or are they instead a subset of the
cardenolides found in leaves?, 3) Are differences in the concentrations of cardenolides
between leaves and nectar constrained by evolutionary history in the series Incarnatae,
and 4) What are the ecological consequences of nectar cardenolides for bumble bees, a
common pollinator of Asclepias?
Methods
Study system
The genus Asclepias is a monophyletic group (Agrawal and Fishbein 2008)
composed of about 135 species found in the Americas (Woodson 1954, Agrawal and
Fishbein 2008, Fishbein et al. In Press). We selected species from the series Incarnatae,
as this is a monophyletic group with relatively well-resolved phylogenetic relationships
(Agrawal and Fishbein 2008) and a significant variation in leaf cardenolide
concentrations among species (Agrawal et al. 2008). The plants were part of two
independently established permanent greenhouse collections at Cornell University.
Quantifying cardenolides
We collected samples to analyze constitutive cardenolide concentrations on five
occasions between July 2007 and August 2008. The frequency and timing of nectar
collections depended on the flowering time of the different species; we collected nectar
from species that flowered abundantly over several sampling periods, whereas we were
only able to collect nectar from rarely flowering species on a single sampling day. We
also collected whole mature leaves from all twelve species. On the four occasions when
78
we did this, we sampled the leaves after taking nectar samples in order to avoid inducing
cardenolide production in nectar or leaves by the leaf removal. We also collected whole
flowers but were unable to gather samples from all twelve species; this was, in part, due
to the fact that flowers were often damaged during nectar collection and were therefore
not appropriate for evaluating constitutive cardenolide concentrations.
We extracted nectar from flowers using 5 µL graduated microcapillary tubes
primarily, although several species (e.g., Asclepias nivea) produced nectar so copiously
that we used a 200 µL microcapillary tubes for sample collection. We took every
precaution to ensure that we caused no damage to the nectary, as this could induce
cardenolides; fortunately, damage can be visually detected in Asclepias as it induces the
exudation of latex. On the few occasions where damage did occur, samples were
discarded. Because most plants produce very little nectar per flower, we pooled nectar
samples to improve our chances of detecting trace amounts of cardenolides. Samples
were pooled within days and individual plants, with each sample containing between 20
and 230 µL of nectar, representing tens to hundreds of individual flowers. After pooling
we had between 1 and 8 individual samples per species. Nectar was added to 500 µL of
70% ethanol (following Bluthgen et al. 2004) and stored at -80°C prior to analysis. We
harvested leaves and flowers from a subset of plants also used for nectar sampling and
froze them immediately at -80°C.
We used high performance liquid chromatography (HPLC) to quantify
cardenolides in nectar, leaf and flower samples, adapting a protocol from Zehnder and
Hunter (2007). We prepared nectar samples for extraction by drying down all water and
ethanol from the stored samples using a rotary evaporator (Labconco). We extracted the
79
residuum with 1 mL of 100% methanol and added 10 µL of a 0.2 g/L solution of the
cardenolide digitoxin as an internal standard, which allowed for the direct comparison of
unknown cardenolide peaks to a known cardenolide of predetermined quantity. We let
the samples shake gently for 24 hours (cardenolides are generally stable at room
temperature; Malcolm 1991) and then spun them down in the centrifuge for 30 minutes at
14000 rpm and 15°C, removing the supernatant. In some cases, preliminary analysis of
samples had very low cardenolide yield, so where necessary, we pooled samples yet
again to increase our power to detect cardenolides. For leaf and flower samples, we
ground wet tissue with liquid nitrogen, and added 10 µL of 2 g/L digitoxin to each
sample, which was around 100 mg of tissue (wet weight). We extracted the samples with
1 mL of 100% methanol using the FastPrep® homogenization system (MP Biomedicals)
to rapidly lyse cells set at 5.0 m/s for 30 seconds. Flower and leaf samples were spun
down as before, but we did not further concentrate the samples.
Cardenolides were analyzed on an Agilent 1100 series HPLC. We injected 15 µL
of the extract. Compounds were separated on a Gemini C18 reversed phase column
(3µm, 150 x 4.6 mm, Phenomenex, Torrance, CA) using the following solvent gradient
(solvent A: 0.25% phosphoric acid in water; solvent B: acetonitrile): 0-5 min 20% B, 20
min 70% B, 20-25 min 70% B, 30 min 95% B, 30-35 min 95 % B at a flow rate of 0.7
mL/min. UV absorption spectra were recorded from 200 to 400 nm and cardenolides
were quantified by integrating the peak area at 218 nm. Cardenolides are identifiable
from other compounds by a single symmetrical peak that absorbs at 218 nm (Zehnder and
Hunter 2007), and we confirmed this by checking the shape and absorption of the peak in
control samples containing only digitoxin. The amounts of cardenolides present were
80
then calculated relative to the peak area of the digitoxin internal standard, of which a
known concentration was added to the sample during extraction. This estimation
assumes that all cardenolide molecules have the same absorption characteristics as
digitoxin. Final cardenolide concentration estimates were calculated as nanograms of
cardenolides per microlitre of nectar or per microgram of fresh tissue collected.
Pollination biology
Milkweeds have a unique and complex pollination system: their pollen grains are
contained and transported in discrete pollinia. Pollinia become attached to the legs of
flower visitors and must then be oriented and inserted correctly into the stigmatic
chamber of the recipient flower to achieve pollination (Wyatt and Broyles 1994). Pollen
tubes travel down from the surface of pollinium, through the nectary to the ovary below
for successful fertilization to occur (Wyatt and Broyles 1994). Milkweed pollination
requires an animal pollinator, but despite their mechanically elaborate pollination system,
all milkweeds that have been studied to date are visited by a suite of generalist flower
visitors, frequently including bumble bees (Wyatt and Broyles 1994).
Nectar secondary metabolites can substantially reduce pollinator visitation
frequency in choice experiments against flowers containing sugar alone (Gegear et al.
2007), but can also reduce pollinator proficiency by increasing the length of time a
pollinator spends on a flower (Adler and Irwin 2005). While the number of visits clearly
correlates with reproductive success, the length of a floral visit can be tied to the number
of pollen grains removed and subsequently donated to the next flower (Thomson 1986).
Whether visit length affects pollinia removal and deposition rates in Asclepias spp.
remains untested (Morse 1982), but we can presume that longer visits should increase a
81
bees’ probability of collecting pollinia and inserting them in subsequence flowers. We
evaluated the effect of nectar cardenolides on both aspects of bumble bee behaviour by
using artificial nectar enriched with a cardenolide and testing pollinator preference on
artificial flower arrays. We chose bumble bees because they are generalists, important
natural pollinators of milkweeds and were commercially available to us. We created an
artificial nectar solution by mixing 30% w/w sucrose with the cardenolide digoxin (92%
HPLC grade, Sigma).
Digoxin, a cardenolide found in Digitalis spp., is a bitter cardenolide (Malcolm
1991) that causes 50% mortality at a concentration of 0.5% in honey bees (Detzel and
Wink 1993). The concentration range of cardenolides we observed was 0-109 ng/µL
across the twelve species found in the series Incarnatae, with a mean nectar cardenolide
concentration of 11.5 ng/µL (see Fig 5.1 and appendix two). With these data for
reference, we created artificial nectar concentrations of 0, 10 and 50 ng/µL digoxin,
simulating control, mean and moderate nectar cardenolide concentrations, respectively.
Our base artificial nectar was a 30% w/w sucrose solution, a reasonable simulation of
natural Asclepias nectar. We mixed nectar solutions every two days, refrigerating unused
portions at 4°C for no more than 24 hours.
We evaluated pollinator preference and flower-handling proficiency using
methods reported by Gegear et al. (2007). In short, marked worker bees were trained to
associate artificial flower colour (either blue or yellow) with one of two nectar conditions
(see descriptions below) by foraging freely on alternating monotypic arrays of each
flower type. The association between flower colour and nectar condition was randomized
among bees to control for any potential bias due to innate colour preferences.
82
Immediately following training, individual bees foraged on an array with 30 flowers of
each type for at least 80 flower visits. We filled flowers with 2 µL of nectar and refilled
each flower immediately after it was drained by a bee. We replaced flowers between
bees. All foraging bouts were videotaped for later analysis using JWatcher Video
Version 1.0 (Blumstein and Daniel 2007).
We completed three separate experiments to evaluate pollinator preference. In the
first, bees chose between control nectar (30% w/w sucrose only) and nectar simulating
the mean nectar cardenolide concentration found in the greenhouse populations of
Asclepias (10 ng/µL digoxin in 30% sucrose). In the second experiment, bees chose
between control nectar and nectar with a higher cardenolide concentration (50 ng/µL
digoxin); this concentration simulates the upper limit of cardenolides found in all species
save Asclepias pumila, which had an extremely high cardenolide concentration (see
appendix two). Finally, we evaluated whether bees reacted differently to the two
concentrations (10 ng/µL vs. 50 ng/µL digoxin). Training periods were consistent
between experiments except for the differences in nectar treatments. Individual bees
were used only once and then dissected for pathogen analysis.
Bumble bee foraging behaviour can be significantly affected by the gut protozoan
Crithidia bombi (Gegear et al. 2005, Otterstatter et al. 2005), but the ingestion of nectar
secondary metabolites can reduce pathogen loads in bumble bee workers (Manson et al.
In Press). To evaluate whether bumble bees infected with C. bombi preferred to forage
on “toxic” nectar as a means of self-medication, we evaluated each bee’s pathogen load
after each foraging assay. Each bee was refrigerated for 48 hours and then its gut was
excised and ground with 100 µL of distilled water in a microcentrifuge tube. We allowed
83
samples to settle for 24 hours at 4°C, then we removed the supernatant and counted C.
bombi cells using an improved Newbauer haemocytometer (Otterstatter and Thomson
2006). As an additional covariate potentially affecting foraging, we also estimated the
bee’s body size by measuring the length of the radial cell on the right forewing (Harder
1982).
Statistical analysis
Quantitative cardenolide analysis
Because many cardenolides did not appear in all samples, the data set contains
many zeros. The distribution of estimated concentrations could not be made normal via
data transformation, so we chose non-parametric analyses. Nectar was quantified by
volume during collection, but we converted these values to mass to facilitate comparisons
across plant components. To do this, we determined the mass of 2 µL of a 60% sucrose
solution in the lab (representing the average sucrose concentration of our nectar samples)
and converted microlitres to micrograms by a factor of 1.057 (data not shown). Prior to
any statistical analyses, we calculated average concentrations as follows: for average total
concentration, we summed all cardenolides in a sample and divided that by the mass of
the sample, giving us a total concentration in ng/µg and then we summed the
concentrations of all samples of that plant component for each species and divided by the
number of samples collected to achieve an average total concentration (see appendix two
for raw data); for average concentration of individual cardenolides, we divided each
compound by the sample mass to get a ng/µg measure, then averaged the concentration of
each compound across all samples within a species (see Fig. 5.1).
84
We began our analyses with a general comparison of the concentrations of
average total nectar cardenolides between the twelve species as well as between plant
components within each species using Kruskal-Wallis tests. This analysis was followed
by a sign test to examine consistent directional differences between nectar, leaves and
flowers, regardless of species identity. We then tested for positive correlations in total
cardenolide concentrations between leaves, nectar and flowers of the same species using
Kendall’s rank correlation.
Qualitative cardenolide analysis
Similarities in the suite of individual cardenolides found in different species and
plant structures can indicate whether cardenolides have a shared site of production. Here
we focused our analysis on differences between nectar and leaves only, as we have the
most complete data for these two plant components.
We compared the number of individual cardenolides in leaves and nectar between
all twelve species using a generalized linear model with a Poisson distribution. We then
examined differences in chemical polarity, a characteristic that affects the mobility and
absorbency of compounds. Highly polar cardenolides are poorly absorbed by animals
and less mobile within plants, while less polar cardenolides are absorbed quickly and are
highly mobile (Malcolm 1991); because reverse phase HPLC filters compounds based on
chemical polarity, retention time is a reasonable surrogate for that polarity, with highly
polar compounds having a short retention time and less polar compounds having longer
retention times. We examined the average retention time of samples with detectable
cardenolides, comparing differences across species and between plant parts. To calculate
the average retention time for a sample containing multiple compounds, we weighted the
85
retention times of individual compounds by their proportion of the sample’s total
cardenolide concentration (Fordyce and Malcolm 2000). We analyzed the data using a
generalized linear model with a Gaussian distribution. We also used a sign test to check
for directionality of changes in weighted averages between leaf and nectar samples.
Comparing the concentrations of individual cardenolides across species and plant
parts required a multivariate approach. Because of fundamental differences in the
physical construction of nectar and leaves, we converted raw cardenolide concentrations
into relative concentrations, or the proportion that the cardenolide contributed to the total
cardenolide concentration of the sample, thus reducing spurious associations driven by
absolute concentration differences in plant components. We used a two-dimensional
non-metric multidimensional scaling (NMDS) ordination to order nectar and leaf tissues
from each species (entities) by the similarity of their cardenolide profiles (attributes).
The ordination used a Bray-Curtis dissimilarity index and was conducted in R’s vegan
package (Oksanen 2009). We were required to remove samples that had no detectable
cardenolides, as their positions in an ordination are undefined; these samples included the
leaves and nectar of Asclepias angustifolia and A. fascicularis, but also the nectar
samples of A. curassavica, A. incarnate ssp. pulchra and A. texana, leaving some of the
leaf samples in the ordination unpaired. We examined contributions of each cardenolide
to the ordination using the envfit function in the R package vegan (Oksanen 2009)
followed by Bonferonni corrections to account for the multiple pairwise comparisons.
We performed Mantel tests to examine correlations between chemical
composition and phylogenic relationships. We did this by comparing our chemical
dissimilarity (distance) matrix, which identifies whether plant components within a
86
species have more compounds in common than plant components in different species, to
a phylogenetic distance matrix, which was constructed from pairwise distances based on
the molecular branch lengths from a pruned phylogenetic tree of the series Incarnatae (see
Agrawal and Fishbein 2008, Fishbein et al. In Press for complete phylogeny). We
analyzed nectar and leaf data separately to prevent spurious correlations between leaves
and nectar from the same species, which have a phylogenetic distance of zero. We
performed the analysis using the R package ape (Paradis 2009).
Behaviour analysis
We assessed pollinator preference for flowers with or without nectar cardenolides
using a G-test to evaluate the total goodness of fit of all bees within an assay, testing for
an expected visit frequency of 0.5 or random. We also examined whether the visitation
data from each bee was homogeneous across an assay. In addition, we tested for
correlations between Crithidia bombi infection intensity and pollinator preference.
We then examined foraging proficiency by comparing average flower handling
time and foraging rate (visits per minute) between treatments. To capture data that
demonstrated consistent foraging behaviour on each flower type, we calculated the
average length of a flower visit from ten consecutive visits to the same flower type found
between visits 45 and 75. We analyzed the extracted data using a generalized linear
model, with radial cell length and pathogen load as covariates. Data were transformed to
meet assumptions of normality when necessary. Analyses were performed in R (version
2.9.0).
87
Results
Quantitative cardenolide analysis
The concentration of average total cardenolides per microgram of Asclepias
nectar, calculated as the sum of total cardenolide concentrations within a species divided
by the number of samples, differed significantly among species (χ2=38.8, df=11,
P<0.001; Fig. 5.1), with five of the twelve species having no constitutive cardenolides in
their nectar at all. Species that had quantifiable nectar cardenolides showed a 30-fold
difference between the lowest and highest total cardenolide concentrations. A within-
species non-parametric analysis revealed that all but two species demonstrated no
difference in average total cardenolide concentrations between plant components
(analysis not shown), while the two species that did differ significantly, Asclepias
curassavica and Asclepias angustifolia, had no detectable nectar cardenolides.
The cardenolide concentration for nectar samples (12.17 +/- 3.22 ng/µg)
regardless of species, was more than twelve-fold higher than the average cardenolide
concentration in leaves (0.54 +/- 0.18 ng/µg) or flowers (0.76 +/- 0.41 ng/µg).
Nonetheless, it is difficult to quantitatively compare these values due to the inherent
physical differences between nectar and leaf tissue. Sign tests comparing nectar
cardenolides to leaf and flower cardenolides found that these differences were not
consistent between species (leaf: s=7, n=11, P=0.34; flower: s=2, n=6, P=0.69). It should
also be noted that cardenolide concentrations were entirely independent of initial mass
(analysis not shown), such that gross cardenolide mass (ng) in each leaf, nectar and leaf
sample is not correlated with the amount of material collected.
88
There was a significant positive correlation between average total cardenolide
concentrations in nectar and those in leaves in the series Incarnatae (n=12, τ =0.43,
P=0.03). Although this relationship is maintained between the average concentration of
cardenolides in leaves and flowers (n= 6, τ =0.73, P=0.03), there is a poor correlation
between nectar cardenolides and flower cardenolides (n=6, τ =0.15, P=0.34), likely due to
the limited flower sample size.
Qualitative cardenolide analysis
The number of individual cardenolides differed significantly across species
(F 11,74= 12.07, P<0.01) and we found a significant interaction between plant part and
species (F 10,74= 7.69, P<0.01). Furthermore, within-species analyses using Wilcoxon
rank-sum tests showed that this pattern was driven by two sister species, Asclepias
curassavica and Asclepias nivea (Agrawal and Fishbein 2008), which have significantly
fewer individual cardenolides in their nectar relative to their leaves (see appendix two).
The identities of cardenolides in leaves and nectar were somewhat different. In
all, we identified thirty individual cardenolides (see appendix two). There was a single
cardenolide that was unique to Asclepias nectar: both A. pumila and A. perennis had a
detectable peak at 14.1 minutes that was found only in their nectar samples. Conversely,
nearly one third (9 of 30) of the compounds found in leaves were not found in nectar
samples. The compounds that were missing from nectar were scattered throughout the
range of cardenolide retention times. The majority of individual cardenolides were found
in both leaf and nectar samples, although seven compounds were found in only a single
species, five of which originated from A. perennis.
89
The average weighted retention time, which estimates polarity, also differed
across species (F 9, 39=16.59, P<0.01). While there was no difference in retention time
between leaves and nectar (F 1, 38=0.18, P=0.68), there was a significant species by plant
component interaction in the model (F 6, 39=5.92, P<0.01). A sign test found no
directional differences in average weighted retention time between leaf and nectar
samples (n=12, s=6, P=0.75).
Our NMDS ordination analysis (Fig. 5.3) revealed that identity and concentration
of the cardenolides in nectar tended to match those of leaves, with the ordination rejecting
the notion that nectar cardenolides are unique or that their proportions are independently
controlled. Five of the seven species with leaf and nectar data have consistent
directionality, with nectar samples closer to the y-axis and leaf samples close to the x-
axis; this trend was driven by the reduction in number of individual compounds found in
nectar samples relative to leaf samples. A sixth species, Asclepias pumila, had the
opposite relationship, with its nectar and leaf samples sitting closer to the x- and y-axis,
respectively; this reversed trend was because both plant components have the same
number of compounds and nectar cardenolides were therefore not a subset of leaf
cardenolides in this case (see appendix two). The three points that fall near the edge of
the ordination are isolated by the fact that each species has only a single detectable
cardenolide. This single compound makes these samples much less complex than the
rest, but also means that each compound represents 100% of their relative cardenolides,
which is what drives the points to the ordination’s edge. Interestingly, the leaf samples
from Asclepias mexicana and its paired nectar sample differ in their single cardenolide,
but the compounds were not unique to the species.
90
Our Mantel tests indicated that there was no autocorrelation between chemical
differences and phylogenetic distances in nectar (z=12.12, P=0.67), or in leaves (z=27.82,
P=0.41), suggesting that the spatial associations in the ordination, which summarize
cardenolide similarity, are independent of the species’ evolutionary history. We also
found that no individual cardenolide had a significant effect on the organization of the
ordination (P>0.05 for correlations of all compounds with both axes of the ordination
after Bonferonni correction).
Behaviour analysis
Bees had different responses to the two different nectar cardenolide
concentrations. In assay 1, bees exhibited a preference for nectar with 10 ng/µL digoxin
rather than the sucrose-only control (G=45.49, df=1, P<0.01). Bees in the second assay
were generally indifferent to nectar with 50 ng/µL digoxin (G=0.87, df=1, P=0.35), while
bees in the third assay preferred to forage on nectar with 10 ng/µL digoxin significantly
more than nectar with 50 ng/µL digoxin (G=23.19, df=1, P<0.01). All three assays had
significant levels of heterogeneity between bees (assay 1 G=670.6, df=11; assay 2
G=843.2, df=10; assay 3 G=551.4, df=9, all P<0.01; see appendix three), indicating that
individual bees had variable preferences. The key result from these behavioural assays is
that low levels of cardenolides were attractive to bees, while intermediate cardenolide
levels neither attracted nor deterred foragers.
Despite these general trends in pollinator preference, individual bees had very
different responses to nectar cardenolides (see appendix three). In the first assay, three of
the twelve individual bees randomly visited flowers (bees 8, 9 and 11), while three bees
primarily visited flowers with sucrose-only nectar (bees 2, 3, 6), and six bees preferred
91
nectar with 10 ng/µL digoxin (bees 1, 4, 5, 7, 10, 12). Bees in the second assay had only
a single bee that foraged without preference (bee 21), while the remaining ten individuals
were evenly split between strong preferences for sucrose-only flowers and strong
preferences for flowers with moderate cardenolides. In assay 3, half the bees chose the
lower cardenolide concentration, while four bees chose higher cardenolides and a single
bee foraged without preference. We found no correlation between the proportion of
sucrose-only visits and intensity of Crithidia bombi infection detected in workers (n= 29,
τ=-0.02, P=0.55).
Further analysis revealed that bees in all three treatments had a significant overall
preference for blue flowers (G=23.09 for assay 1, G=225.4 for assay 2, G=300.3 for
assay 3, all df=1, P<0.01; see appendix three). However, we still saw significant
heterogeneity for this preference within the data (assay 1 G=693.0, df=11; assay 2
G=618.8 df=10; assay 3 G=261.9, df=9, all P<0.01). The majority of all individual bees,
twenty-one of thirty-three, preferred to forage on blue (appendix three). This apparent
blue bias may be associated with the colour of the first flower visited, which was blue
95%, 91% and 100% of the time for assays 1, 2 and 3, respectively. Interestingly, the
intensity of C. bombi in bees did significantly correlate with the proportion of visits to
blue flowers, (n= 29, τ=0.22, P=0.04). This pattern was driven by one particular bee with
a very high infection level and strong affinity for blue flower. However, removing this
bee from the analysis still shows a suggestive, if non-significant, trend (n= 28, τ=0.18,
P=0.08, Fig. 5.4).
The consumption of cardenolides had no effect on bumble bee foraging
proficiency. There was no difference in average visit length between bees foraging on
92
control or cardenolide-enriched nectar of either average or moderate concentration
(F2,24=1.79, P=0.19; Fig. 5.5a). We also found no difference in the number of visits per
minute between the three treatments (F2,24=0.14, P=0.87, Fig. 5.5b). Neither visit length
nor visitation rate was affected by either bee size or pathogen load.
Discussion
Nectar cardenolides in the Asclepias series Incarnatae vary in concentration,
identity, number and chemical polarity; the variation across species is substantially higher
than between nectar and other plant components within species. More importantly, the
comparative data generally support the consequence-of-defense hypothesis for nectar.
Positive correlations between the cardenolide concentrations of different plant
components suggest that the presence of nectar cardenolides is likely a consequence of
systemic chemical defense. Nectar appears to contain a simpler cocktail of secondary
metabolites than foliage, and the compounds are generally a subset of those found in
leaves (see appendix two). With one exception, the individual cardenolides found in
nectar can also be found in leaves, refuting the suggestion that nectar has evolved an
independent chemical arsenal. Finally, we found that bumble bees preferred to forage on
flowers with average cardenolide concentrations, which could imply that pollinator-
mediated selection could reinforce or stabilize the concentrations of cardenolides in
nectar. Our study, the first to detect and quantify nectar cardenolides, suggests that the
chemical complexity of Asclepias nectar may be a byproduct of systemic phytochemical
defenses.
The relationship between cardenolide concentrations in leaves and in other plant
components is an understudied area, but recent interest in the feedback between above-
93
and below-ground herbivory has provided some insight. Rasmann et al. (In Press) found
a significant correlation in total constitutive cardenolide levels between the roots and
leaves of twelve Asclepias species from the series Incarnatae (ten of which are shared
between this study and our own), a pattern which complements our findings in nectar.
Taken together, the correlations between total cardenolide concentrations in Asclepias
shoots, nectar and roots suggest that chemical defense concentrations in milkweeds are
systemically linked. This is further supported by evidence that total leaf cardenolides
also correlate with flavonoids, another foliar defensive chemical (Agrawal et al. 2009).
In contrast, leaf cardenolides are not correlated with physical defensive traits such as
trichome number and leaf thickness, refuting the concept of a complex plant defense
syndrome (sensu Agrawal and Fishbein 2006). Because our study is only the second to
compare nectar secondary metabolites to foliar secondary metabolites, we cannot infer a
general trend towards correlated chemical composition across these plant parts.
However, Adler et al.’s study on alkaloids in Nicotiana tabacum (2006) also found a
significant positive correlation between leaf and nectar secondary metabolites. Despite
the implications of these correlations, analyses that focus on total cardenolide
concentrations may mask differences in the concentrations of individual cardenolides,
differences which are critical to understanding cardenolide allocation within a plant.
Similarities in the identity and concentration of individual cardenolides found
in Asclepias nectar and leaves tend to support the consequence-of-defense hypothesis.
Regardless of species, of the twenty-one individual cardenolides found in nectar, twenty
are also found in leaves, so nectar cardenolides in general appear to be a subset of leaf
cardenolides and not a unique suite of compounds. However, when we compare leaf and
94
nectar cardenolides within species, we see that individual nectar cardenolides have
matching compounds in leaves only 63% of the time, on average, while the remaining
nectar compounds are not unique to Asclepias in general but may be unique to that
species (see appendix two). In other words, although the majority of nectar compounds
are nested within leaf compounds, there is still a substantial proportion that, while not
novel, are absent from the leaves of a species but present in its nectar. These
cardenolides, found in the nectar but absent from the leaves, can represent up to 40% of a
species’ total nectar cardenolide concentration. A simple nested relationship is further
refuted by the single unique nectar cardenolide (14.1 min), which is found in the nectar of
two species but has not been identified in leaves or flowers and may therefore have a
novel source and function. One explanation for this discrepancy between nectar and
leaves could be that the “missing” compounds are in very low concentrations in leaves
and are therefore not detected during analysis. Although the majority of nectar
cardenolides are a subset of leaf cardenolides, the inconsistencies in this putatively nested
relationship suggest that the systemic defense hypothesis does not explain every feature
of cardenolide distribution within a plant.
In contrast, our multivariate analysis indicates that the concentrations of
individual cardenolides in nectar and leaves from the same Asclepias species have more
in common than individual cardenolides of nectar from different species, reaffirming that
nectar cardenolides are largely a subset of leaf cardenolides. The lack of a phylogenetic
signal in the data suggests that the similarities in chemistry across species are
independent of evolutionary history. Two previous studies comparing total cardenolide
concentrations across species have found no phylogenetic conservatism in Asclepias leaf
95
cardenolides (Agrawal et al. 2008, Rasmann et al. In Press), while another did detect a
phylogenetic signal in total leaf cardenolides (Agrawal et al. 2009); all three studies used
substantially more species than ours and the species used in the comparisons were similar
but did not overlap completely. In addition, the study by Rasmann et al. (In Press) did
not find a phylogenetic signal in the total leaf or total root cardenolides from twelve
species within the series Incarnatae (ten of which overlap with our own study), although
the authors suggest this may be due the limited number of species examined. Overall,
similarities in cardenolide composition between leaves and nectar seem to provide some
evidence for a phylogenetically independent systemic chemical defense strategy.
Nearly a third of leaf cardenolides were undetected in nectar, but the reason for
their absence was not clear. Previous work comparing the chemistry of different
Asclepias plant parts has found that the chemical polarity of cardenolides can differ
significantly between leaves and roots (Nelson et al. 1981), and between leaves and stems
(Fordyce and Malcolm 2000), which led to us to question whether the cardenolides
missing from nectar might all have similar chemical polarities. The chemical polarity of
a cardenolide is linked to the compound’s rate of absorption in an animal’s gut:
compounds with low polarity are bitter-tasting and quickly absorbed after consumption,
causing rapid emetic responses but also potential systemic toxicity, whereas highly polar
compounds are harder to taste and are poorly absorbed, but can become cumulatively
toxic over time (Malcolm 1991). In addition, chemical polarity seems to be related to the
mobility or transportability of a compound within a plant, with compounds of low
polarity being much more portable than compounds of high polarity (Malcolm 1991).
Given these characteristics, we might expect that the absence of compounds with high
96
polarity from nectar would support the hypothesis that secondary metabolites leak into
nectar as a consequence of defense. However, if nectar secondary metabolites function to
deter inefficient pollinators, we would also expect that highly polar, and therefore less
toxic, cardenolides would be excluded from nectar. Unfortunately for this conjecture, we
can infer from the both the raw data (appendix two) and the weighted average retention
times that excluded compounds have a range of polarities. In addition, there is no
directionality in weighted average retention time, suggesting that there is no trend of an
increased proportion of cardenolides with low polarity in nectar. Finally, we did not find
that the absent compounds were of particularly low concentration in leaves, refuting the
possibility that missing compounds were simply too dilute to detect in nectar. We must
therefore conclude that neither chemical polarity nor reduced concentration explain the
exclusion of certain cardenolides from nectar.
The effects of nectar secondary metabolites on pollinator behaviour are
inconsistent, with some studies reporting that bees have a strong aversive reaction to
compounds such as alkaloids (Adler and Irwin 2005, Gegear et al. 2007), phenolics
(Hagler and Buchmann 1993) and cyanogenic glycosides (London-Shafir et al. 2003),
and others showing indifference or preference for secondary metabolites such as alkaloids
(Singaravelan et al. 2005). Although Pryce-Jones (1942) observed that Asclepias nectar
was toxic to bees, this is the first study to rigorously evaluate how pollinators respond to
a cardenolide that is believed to be in the nectar of Digitalis spp. (Detzel and Wink 1993).
Our analysis of bumble bee preference suggests that bumble bees find average
concentrations of nectar cardenolides more attractive than nectar containing only sugar,
and are indifferent to higher concentrations of nectar cardenolides. Although pollinator
97
attraction to nectar secondary metabolites is not without precedent (Singaravelan et al.
2005, Liu et al. 2007), the result was surprising in light of previous research that
demonstrates cardenolide toxicity in other animals (Malcolm 1991).
Despite their reputed toxicity, secondary metabolites may act as a post-
consumptive foraging cue, creating an association between the taste of the secondary
compound and the quality of the reward (Cipollini and Levey 1997a). This association
is, however, unlikely under our experimental conditions, which provided rewards of
equal caloric value and a clear pre-consumption colour cue to identify each nectar
treatment. We must therefore consider that cardenolides at low concentrations are
actually appealing to bumble bees; to our knowledge this is the first case of cardenolides
acting as a feeding stimulant to a generalist forager. In this study, bumble bees preferred
nectar with low cardenolides to nectar with high cardenolides or sucrose-only nectar,
which could suggest a dose-dependent association. Artificial nectar with low
concentrations of caffeine or nicotine are preferred by free-foraging honey bees over a
sugar-only solution (Singaravelan et al. 2005), whereas nectar with average
concentrations of either alkaloid deterred foragers. However, it is unclear whether the
preference for low-dose nicotine or caffeine is driven by a simple cue association or by a
dependence on these compounds, which are known to be addictive. There is currently no
evidence to suggest that cardenolides have addictive properties, although there are many
examples of cardenolide preferences in specialist herbivores (Malcolm 1991).
Despite these intriguing results, it should be noted that our behaviour assays may
not accurately reflect bumble bee foraging decisions in natural populations of Asclepias.
The cardenolide concentrations we used reflect the constitutive concentrations found in
98
our greenhouse grown collections, but concentrations of nectar cardenolides may be
significantly higher in natural Asclepias populations, where plants are under selection by
herbivores, and where concentrations may be elevated through induction or evaporation.
In addition, while many of our Asclepias nectar samples contain a number of different
cardenolides, we used only a single compound in our artificial nectar. We chose to use
digoxin to synthesize Asclepias nectar because it is a commercially available botanical
cardenolide that has lethal post-consumptive consequences for honey bees (Detzel and
Wink 1993). Digoxin also has low polarity and is therefore quite bitter and more rapidly
absorbed than its more polar counterparts (Malcolm 1991). We predicted that these
characteristics would enhance the likelihood of both aversive and emetic behaviour in
bees, which was clearly not the case. It is difficult to predict how nectar with a more
complex chemical profile might affect foraging behaviour, and the issue merits further
investigation.
The unexpected preference for blue flowers makes interpreting the results of the
behavioural assays more complicated. This trend may be driven simply by the
overwhelming preference for the first flower visited on the array to be blue (true in 94%
of the assays, despite randomizing the colour of the closest flower to the hive), which
may have been driven by an innate preference for blue (Heinrich et al. 1977, Gumbert
2000). Previous studies suggest that bees rapidly associate colour with reward (e.g.
Heinrich et al. 1977, Dukas and Real 1991) and that bees with foraging experience rely
on learned colour cues over innate colour bias (Gegear and Laverty 2004); this does not
appear to be the case in our study. Interestingly, we found that the infection intensity of
Crithidia bombi correlated significantly with the proportion of visits to blue flowers (Fig.
99
5.4). C. bombi is known to impair associative learning in foraging bees (Otterstatter et al.
2005, Gegear et al. 2006) and we hypothesize that, as C. bombi infection intensity
increased, bees became unable to associate floral colour with nectar treatment during
training, rendering the training period irrelevant. We may further hypothesize that this
innate preference for blue may have motivated bees to make their first foraging visit on
the mixed array to a blue artificial flower. Bees may then have chosen to become
constant visitors on blue flowers because they were adequately rewarding. Although the
blue bias complicates our understanding of the results, it actually reinforces the fact that
the nectar cardenolide treatments are not particularly deleterious for pollinators, as
digoxin at moderate concentrations did not discourage visitation to blue flowers.
Pre-existing infections of the protozoan pathogen Crithidia bombi within our
experimental colony gave us the opportunity to simultaneously test another adaptive
hypothesis of “toxic” nectar, that of its putative antimicrobial properties (Adler 2000).
Secondary metabolites are hypothesized to have antimicrobial activity to reduce
degredation of nectar quality by floral yeasts (Hagler and Buchmann 1993). However,
these antimicrobial properties may also provide benefits for foragers if they can reduce
pathogen loads (Manson et al. In Press). C. bombi is a protozoan pathogen that affects
pollinator foraging (Gegear et al. 2005, Otterstatter et al. 2005) and can reduce colony
fitness (Imhoof and Schmid-Hempel 1999). Previous work has found that leaf
cardenolides from A. curassavica can reduce infections of the protozoan pathogen
Ophryocystis elektroscirrha in monarch butterfly caterpillars (de Roode et al. 2008). In
addition, the nectar alkaloid gelsemine has been shown to reduce C. bombi loads in
bumble bees at natural concentrations (Manson et al. In Press). In our study, we assessed
100
standing pathogen loads only at the end of preference trials, so we cannot evaluate
whether cardenolides reduced C. bombi infections. It also seems unlikely that the
consumption of cardenolides for tens of minutes might be effective at mitigating
infection. Instead, we postulated that if cardenolides are therapeutic for ill bumble bees,
infected individuals may demonstrate a preference for cardenolide-rich nectar indicative
of self-medication (Singer et al. 2009). We found that this was not the case, nor was there
any indication that the greater the infection intensity, the more visits made to cardenolide-
rich flowers. However, given the tantalizing possibility of pollinators actively mitigating
pathogen levels with cardenolides, the issue warrants further investigation.
Finally, we consciously decided to sample plants free from herbivore damage in
order to restrict our analysis to constitutive cardenolides. However, herbivore-induced
chemical defenses can significantly alter the distribution and concentration of secondary
metabolites in a plant (Karban and Baldwin 1997). This is particularly true in Asclepias,
where differences in the concentration of leaf cardenolides between species can jump
from five-fold to forty-five fold after damage by monarch butterfly larvae (Rasmann et al.
In Press). In contrast, the same study found that foliar damage did not increase
cardenolide concentrations in roots (Rasmann et al. In Press), demonstrating that
induction may not be a systemic response in Asclepias. While foliar herbivory in
Nicotiana tobacum increased nectar alkaloids by 33% (Adler et al. 2006), no studies have
addressed the immediate consequences of herbivory on nectar cardenolide
concentrations. If nectar cardenolides are inducible, examining the concentrations of
individual compounds before and after damage could provide concrete evidence for the
101
consequence-of-defense hypothesis and may even illuminate the underlying mechanism
for the presence of secondary metabolites in nectar.
Many plants must balance the costs of simultaneously attracting pollinators and
deterring herbivores, and strategies that accommodate both factors may result in
suboptimal reproduction or defense. Our chemical survey of nectar cardenolides and
their relationship with leaf and flower cardenolides suggests that the presence of
secondary metabolites in nectar is likely a consequence of a systemic chemical defense
strategy. Determining the allocation and distribution of plant secondary metabolites,
along with their subsequent effects on both antagonists and mutualists, requires a wider
lens than most studies have employed.
Acknowledgements
We would like to thank A. Parachnowitsch, S. DeLeon, M. Stasny, S. Campbell, A.
Erwin and A. Hastings for their support during cardenolide collection and analysis.
Cardenolide analyses were conducted in the Cornell Chemical Ecology Core Facility,
with support from Paul Feeny, New Life Sciences Initiative, College of Agriculture and
Life Sciences, Center for a Sustainable Future, Boyce Thompson Institute, and
Departments of Ecology & Evolutionary Biology, Neurobiology & Behavior,
Entomology, Plant Biology, and Horticulture. The chapter was vastly improved by C.
Parsons. This research was supported by NSERC of Canada and US NSF-DEB
0447550.
102
Asclepias species
ang bar bol can cur fas inc mex niv per pum tex
Ca
rde
no
lide
Co
ncen
tra
tio
n (
ng
/µg
)
0
20
40
60
80
100
120
Figure 5.1. Average total cardenolide concentration for nectar, leaf and flower samples
of twelve species from the Asclepias series Incarnatae: Asclepias angustifolia, A.
barjoniifolia, A. boliviensis, A. candida, A. curassavica, A. fascicularis, A. incarnata
pulchra, A. mexicana, A. nivea, A. perennis, A. pumila and A. texana. Black bars
represent nectar, light grey represent leaves and dark grey represent flowers, all with
standard error bars. Flower samples were analyzed for the underlined species, while
those not underlined represent species for which flower samples were not collected.
Nectar concentrations have been converted to mass equivalents for comparative purposes
(see methods). All other missing peaks represent a sample concentration of zero. For
exact cardenolide concentrations, consult appendix two.
103
0 10 20 30 40 50 60
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
Leaf Cardenolide Concentration (ng/µg)
Ne
cta
r C
ard
en
olid
e C
on
ce
ntr
atio
n (
ng
/ µ
g)
Figure 5.2. Correlation between leaf and nectar cardenolide concentrations. Each point
represents paired concentration data for total cardenolide concentration of leaf and nectar
samples from a single plant. Note that one extreme outlier, Asclepias pumila (nectar
concentration: 110 ng/µg, leaf concentration: 0.17 ng/µg), has been omitted from the plot
for visual effect but is included in the analyses.
104
−1 0 1 2 3
−1
0
1
2
3
Dim1
Dim
2
BAR
bar
BOL
bol
CAN
can
cur
inc
MEX
mexNIV
niv
PER
per
PUM
pumtex
Figure 5.3. NMDS two-dimensional ordination of the individual carndenolide
concentrations of Asclepias nectar and leaves. Nectar samples are denoted by upper case
letters, while lower case denotes leaf samples (see Fig. 5.1 for species abbreviations).
Ellipses highlight nectar and leaves from the same species. Leaves and nectar with no
detectable cardenolides could not be included in the ordination (see Methods for more
information).
105
10000 20000 30000 40000 50000 60000
0.2
0.4
0.6
0.8
1.0
Crithidia (cells/µL)
Pro
po
rtio
n o
f V
isits t
o B
lue
Flo
we
rs
Figure 5.4. Correlation between intensity of Crithidia bombi infections in bees and their
preference for blue flowers (n=28; outlier excluded in plot).
106
0 10 50
5
10
15
20
Nectar Cardenolide Concentration (ng/µL)
Ave
rage V
isit L
ength
(s)
Figure 5.5a. Boxplots of average length of ten consecutive flowers visits to flowers with
no (0 ng/µL, n=6 bees), low (10 ng/µL, n=11 bees) or moderate (50 ng/µL, n=8 bees)
cardenolide concentrations. The boxes represent the 25th
and 75th
percentiles, the
whiskers are roughly two standard deviations and open circles represent outliers.
107
0 10 50
2
4
6
8
10
Nectar Cardenolide Concentration (ng/µL)
Fo
rag
ing
Ra
te (
vis
its/m
in)
Figure 5.5b. Boxplots of visitation rate to flowers with no (0 ng/µL, n=6 bees), low (10
ng/µL, n=11 bees) or moderate (50 ng/µL, n=8 bees) cardenolide concentrations.
The rate was calculated using ten consecutive visits to each nectar type and therefore
estimates visitation rate during a period of foraging constancy. Again, the boxes are the
25th
and 75th
percentiles, the whiskers represent approximately two standard deviations
and the open circle represents an outlier.
108
CHAPTER SIX
Concluding Discussion
The study of nectar secondary metabolites has emerged as an important
intersection between plant-herbivore and plant-pollinator studies; recognizing that floral
traits are shaped by selection from both mutualists and antagonists, researchers have
begun to integrate theory and methods from these historically separated fields. My
dissertation contributes to this new area of study by examining the consequences of
nectar secondary metabolites from a pollinator’s perspective. Specifically, I am
broadening our understanding of how pollinators respond both behaviourally and
physiologically to nectar alkaloids and nectar cardenolides. I also provide the first
evidence to support an antimicrobial function for nectar secondary metabolites and the
first survey of nectar cardenolides. In addition, my findings support the hypothesis that
nectar secondary metabolites are a byproduct of a plant’s defenses against herbivores.
The relationships between plant defense, pollination and herbivory can be
summarized using path diagrams to illustrate the direct and indirect effects of secondary
metabolites on plant fitness. Secondary metabolites mitigate herbivore damage and
therefore have a direct positive effect on plant fitness, but the direct effects of these
defensive chemicals are context dependent when one considers how secondary
metabolites affect pollinators (Fig 6.1a). The direction of these undefined arrows could
be shaped by both the identity of the secondary compound and the type of pollinator (e.g.
generalist or specialist, robber or legitimate pollinator); for example, results from
chapters two and three suggest that nectar alkaloids have a negative effect on pollinators,
109
while chapter five found that nectar cardenolides have a positive effect on pollinators.
When the path diagram is expanded to include the role of pollinator pathogens on plant
fitness (Fig 6.1b), the relationships depicted in the diagram become more defined. Based
on results from chapter four, I can predict that secondary metabolites will have a direct
positive effect on plant fitness mediated indirectly by their negative effect on pollinator
pathogens. While the empirical data necessary to determine the strength of each effect
has yet to be collected, these diagrams provide a framework for further evaluations of
chemically-mediated plant-herbivore-pollinator interactions.
My research addresses the costs and benefits of nectar secondary metabolites for
pollinators in a number of different ways. Here I synthesize my results and suggest their
broader implications.
Ecological context is crucial
Secondary metabolites are known to be toxic, unpalatable and aversive, yet
pollinators often forage on nectar containing these compounds. In chapter two of my
dissertation, I show that although the nectar alkaloid gelsemine deters bees at even the
lowest concentration tested, this response is mitigated when alkaloid-rich nectar has
higher sucrose concentration than alkaloid-free control nectar. This result not only
suggests a potential mechanism that plants may use to overcome the deterrent properties
of nectar alkaloids, but it also highlights the importance of context for bumble bee
foraging choices. Animals make foraging decisions by weighing their options and
selecting food that provides the most reward at the least cost (Sih 1980). Nectar with
secondary metabolites can fit this description if: a) it provides a better source of calories
for foragers, b) it is larger in volume or c) other nectar options are scarce. Several studies
110
have found that “toxic” nectar collection by honey bees is associated with a reduction in
alternative nectar availability (London-Shafir et al. 2003, Tan et al. 2007), and I have
observed the same patterns with bumble bees in the lab. We might therefore expect that
plants with nectar secondary metabolites may be under selection by pollinators to bloom
either early or late in the season, times when other floral resources are limited.
Gelsemium sempervirens, which flowers from January to April (Pascarella 2007), meets
this expectation; whether there is a broader pattern between plants with nectar secondary
metabolites and flower phenology is currently unknown. Pollinators that choose to
forage on “toxic” nectar do not make this decision blindly, and the effects of nectar
secondary metabolites on pollinator preference cannot be interpreted without considering
the ecological circumstances surrounding the animal’s decision.
A subtle effect is still an effect
Despite the dramatic reports of pollinator mortality due to nectar secondary
metabolites (Baker and Baker 1975, Rhoades and Bergdahl 1981, Adler 2000), the post-
consumptive effects of gelsemine on pollinators are less apparent but no less important.
In chapter three, I found that subordinate bees developed smaller oocytes after consuming
nectar enriched with high concentrations of gelsemine. This sublethal effect of nectar
alkaloids suggests that gelsemine can inhibit protein utilization and is potentially costly to
reproduction. It is important to recognize that this effect was detected only when a
preliminary study prompted me to measure the width of oocytes. Previous work on
Osmia lignaria revealed no detectable cost of consuming gelsemine at very high doses
during several developmental stages (Elliott et al. 2008). Given that costs in my study
were detectable only in bumble bees with limited access to food, I would suggest that
111
costs may have been masked by the generous provisions supplied to each O. lignaria by
Elliott et al. (2008) and that fitness effects could exist but are simply difficult to detect.
In contrast, the results of chapter four suggest that consuming nectar alkaloids
could have a subtle positive effect on bumble bees. Continuous ingestion of gelsemine-
enriched nectar significantly reduces the number of Crithidia bombi cells in foragers’
guts. C. bombi is not a lethal pathogen; instead, it impairs worker foraging efficiency
(Gegear et al. 2005, Otterstatter et al. 2005, Gegear et al. 2006), an effect that can
subsequently reduce colony fitness. Because the behavioural costs of a C. bombi
infection are correlated with infection intensity, we can infer that gelsemine consumption
should mitigate these costs and subsequently improve overall colony fitness. Detecting
improvements at the colony level is challenging due to natural variation in hive fitness
but does not negate that these improvements occur. The long-term costs and benefits of
nectar alkaloid consumption for pollinators are far from resolved, but future studies
should not discount the importance of an effect based on magnitude alone.
There is no such thing as a general adaptive hypothesis
In her review paper, Adler (2000) hypothesized that an antimicrobial function
would be the most general adaptive mechanism to explain nectar secondary metabolites.
Rather than being directed at a particular type of floral visitors, this hypothesis is based
on the ubiquity of microbes in nature and the fact that nectar is an excellent medium for
microbial growth. I first tested this hypothesis on floral nectar yeasts, which degrade
nectar and reduce its palatability for pollinators (Herrera et al. 2008). In appendix one, I
demonstrate that a number of floral yeasts from a variety of different plants were
112
unaffected when grown on gelsemine-enriched media, indicating that nectar alkaloids do
not confer “anti-yeast” properties to floral nectar. In chapter four, I report that nectar
alkaloids can have deleterious effects on the microbe Crithidia bombi, a bumble bee
pathogen that is transferred from infected to naïve individuals at the flower (Durrer and
Schmid-Hempel 1994). This experiment did not find a direct effect of nectar alkaloids on
C. bombi, but rather suggested that ingesting gelsemine can indirectly reduce the number
of C. bombi cells in the bees’ guts. These results call into question the generality of the
antimicrobial hypothesis for two reasons: first, the fact that some microbes are affected
while others aren’t indicates that the nectar alkaloid gelsemine is not universally
antimicrobial; second, I expected to see microbial inhibition in floral yeasts, which are
ubiquitous in flowers, but not inhibition of a pathogen that affects only bumble bees. In
other words, the more general microbe, which is predicted to reduce pollination
indiscriminately through nectar degradation, is immune to nectar alkaloids, while the
microbe that is deleterious to only a single genus of pollinators is inhibited by the
alkaloid. There is significant diversity in the characteristics of secondary metabolites
found in nectar, and the impact of these compounds on biotic interactions with microbes
and flower visitors is highly variable. It is probably simplistic to predict a general
mechanism to explain the presence of secondary compounds in nectar.
Consequence of defense is often assumed but rarely tested
Nectar secondary metabolites may or may not have an adaptive function, but it is
frequently suggested that they appear in nectar as a consequence of a plant’s need to
chemically defend itself (Adler 2000, Strauss and Whittall 2006). There are at least two
mechanisms that could explain nectar secondary metabolites as a byproduct of foliar
113
defense: first, secondary metabolites might leak into nectar during vascular transportation
from the site of production to the site of defense; second, if production of secondary
metabolites is controlled systemically but compounds are synthesized locally in the
nectary, a systemic augmentation in defense could result in the production of nectar
secondary metabolites. Both of these mechanisms would result in a correlation between
secondary metabolite concentrations in nectar and in leaves, but characterizing the
distribution of chemical defenses is difficult and therefore understudied. To my
knowledge, only two previous studies have addressed this issue, both finding positive
correlations between the total alkaloid concentrations of leaves and nectar in plants from
across the genus Nicotiana (Adler et al. 2006; Adler unpublished). In chapter five, I
report positive correlations between total cardenolide concentrations in the nectar and
leaves of the genus Asclepias series Incarnatae, providing support for the consequence-
of-defense hypothesis. The relationship between individual cardenolides in nectar and
those in leaves is not nearly as clear. Most nectar cardenolides represent a subset of those
found in leaves, which suggests vascular leakage; however, some of the cardenolides
isolated in nectar did not match those found in leaves, suggesting that at least some
cardenolides may be produced or regulated independently in nectar and leaves.
Evaluating the consequence-of-defense hypothesis is fundamental to the study of nectar
secondary metabolites because it establishes the link between herbivore defenses and
nectar toxicity. Future studies should compare chemical composition across plant
components in a number of different species and focus on investigating the physical
mechanism responsible for secondary metabolites in nectar.
114
Costs and benefits are not always obvious
My dissertation would have been substantially easier if the consumption of nectar
secondary metabolites killed bumble bees. Unfortunately, detecting the costs or benefits
of this paradoxical phenomenon is not that simple. In chapter three, detecting any
sublethal cost of nectar alkaloids required measuring a very small trait in a certain subset
of bees that could have easily been overlooked. The reduction in pathogen load found in
chapter four would have been impossible to detect through the anticipated improvements
of this reduction on pollinator foraging efficiency. The overall preference for nectar
cardenolides of average concentration in chapter five was not immediately obvious
because of the significantly heterogeneous behaviour of individual bees and was only
detected after careful statistical analyses. Finally, even though bumble bees have a strong
aversion to gelsemine-rich nectar in the lab, they readily collect nectar from Gelsemium
sempervirens in the field, suggesting that even significant behavioural patterns can be
obscured depending on the conditions of observation. Nectar secondary metabolites may
not always be lethal, but they may nevertheless have ecologically important effects on
pollinators which need to be taken into consideration when interpreting plant-pollinator
interactions.
Although often described as a simple mutualism, the relationship between plants
and their pollinators is anything but simple. In recognizing that this relationship is
shaped by both mutualists and antagonists, my dissertation enriches study of pollination
ecology and, more broadly, the study of plant-animal interactions as a whole.
115
Figure 6.1. Path diagrams of the relationships between secondary metabolites,
herbivores, pollinators and plant fitness, both without (A) and with (B) the inclusion of
pollinator pathogens. Solid lines represent positive effects while dashed lines represent
negative effects and both solid and dashed suggest a context-dependent outcome. See
text for explanation of models.
116
REFERENCES
Adler, L. S. 2000. The ecological significance of toxic nectar. Oikos 91:409-420.
Adler, L. S., and R. E. Irwin. 2005. Ecological costs and benefits of defenses in nectar.
Ecology 86:2968-2978.
Adler, L. S., and R. E. Irwin. 2006. Comparison of pollen transfer dynamics by multiple
floral visitors: experiments with pollen and fluorescent dye. Annals of Botany
97:141-150.
Adler, L. S., M. Wink, M. Distl, and A. J. Lentz. 2006. Leaf herbivory and nutrients
increase nectar alkaloids. Ecology Letters 9:960-967.
Afik, O., A. Dag, Z. Kerem, and S. Shafir. 2006. Analyses of avocado (Persea
americana) nectar properties and their perception by honey bees (Apis mellifera).
Journal of Chemical Ecology 32:1949-1963.
Agrawal, A. A., and M. Fishbein. 2006. Plant defense syndromes. Ecology 87:S132-
S149.
Agrawal, A. A., and M. Fishbein. 2008. Phylogenetic escalation and decline of plant
defense strategies. Proceedings of the National Academy of Sciences of the
United States of America 105:10057-10060.
Agrawal, A. A., M. J. Lajeunesse, and M. Fishbein. 2008. Evolution of latex and its
constituent defensive chemistry in milkweeds (Asclepias): a phylogenetic test of
plant defense escalation. Entomologia Experimentalis Et Applicata 128:126-138.
117
Agrawal, A. A., J. P. Salminen, and M. Fishbein. 2009. Phylogenetic trends in phenolic
metabolism of milkweeds (Asclepias): evidence for escalation. Evolution 63:663-
673.
Agrawal, A. A., S. Y. Strauss, and M. J. Stout. 1999. Costs of induced responses and
tolerance to herbivory in male and female fitness components of wild radish.
Evolution 53:1093-1104.
Antonovics, J. 2005. Plant venereal diseases: insights from a messy metaphor. New
Phytologist 165:71-80.
Armbruster, W. S. 1997. Exaptations link evolution of plant-herbivore and plant-
pollinator interactions: a phylogenetic inquiry. Ecology 78:1661-1672.
Ashman, T. L., D. H. Cole, and M. Bradburn. 2004. Sex-differential resistance and
tolerance to herbivory in a gynodioecious wild strawberry. Ecology 85:2550-
2559.
Ayasse, M., T. Marlovits, J. Tengo, T. Taghizadeh, and W. Francke. 1995. Are there
pheromonal dominance signals in the bumblebee Bombus hypnorum L
(Hymenoptera, Apidae). Apidologie 26:163-180.
Baker, H. G. 1977. Non-sugar chemical constituents of nectar. Apidologie 8:349-356.
Baker, H. G. 1978. Chemical aspects of the pollination biology of woody plants in the
tropics. Pages 57-82 in P. B. Thomlinson and M. H. Zimmerman, editors.
Tropical trees as living systems. Cambridge University Press, Cambridge.
118
Baker, H. G., and I. Baker. 1973. Amino acids in nectar and their evolutionary
significance. Nature 241:543-545.
Baker, H. G., and I. Baker. 1975. Studies of nectar-constitution and pollinator-plant
coevolution. Pages 100-140 in L. E. Gilbert and P. H. Raven, editors. Coevolution
of animals and plants. University of Texas Press, Austin.
Baker, H. G., and I. Baker. 1982. Chemical constituents of nectar in relation to
pollination mechanisms and phylogeny. University of Chicago Press, Chicago.
Baker, H. G., and I. Baker. 1983. A brief historical review of the chemistry of floral
nectar. Pages 126-152 in B. Bentley and T. Elias, editors. The Biology of
Nectaries. Columbia University Press, New York.
Berenbaum, M. R. 1995. The chemistry of defense - theory and practice. Proceedings of
the National Academy of Sciences of the United States of America 92:2-8.
Berenbaum, M. R., and A. R. Zangerl. 1994. Costs of inducible defense - protein
limitation, growth, and detoxification in parsnip webworms. Ecology 75:2311-
2317.
Berenbaum, M. R., A. R. Zangerl, and J. K. Nitao. 1986. Constraints on chemical
coevolution - wild parsnips and the parsnip webworm. Evolution 40:1215-1228.
Bernays, E. A., and M. S. Singer. 2005. Taste alteration and endoparasites. Nature
436:476-476.
119
Bezemer, T. M., and N. M. van Dam. 2005. Linking aboveground and belowground
interactions via induced plant defenses. Trends in Ecology & Evolution 20:617-
624.
Blaw, M. E., M. A. Adkisson, D. Levin, J. C. Garriott, and R. S. A. Tindall. 1979.
Poisoning with Carolina jessamine (Gelsemium sempervirens [L] Ait). Journal of
Pediatrics 94:998-1001.
Bloch, G., and A. Hefetz. 1999. Regulation of reproduction by dominant workers in
bumblebee (Bombus terrestris) queenright colonies. Behavioral Ecology and
Sociobiology 45:125-135.
Blumstein, D. T., and J. C. Daniel. 2007. Quantifying behaviour the JWatcher way.
Sinauer Associates, Inc., Sunderland.
Bluthgen, N., G. Gottsberger, and K. Fiedler. 2004. Sugar and amino acid composition of
ant-attended nectar and honeydew sources from an Australian rainforest. Austral
Ecology 29:418-429.
Boppre, M., S. M. Colegate, and J. A. Edgar. 2005. Pyrrolizidine alkaloids of Echium
vulgare honey found in pure pollen. Journal of Agricultural and Food Chemistry
53:594-600.
Brody, A. K. 1997. Effects of pollinators, herbivores, and seed predators on flowering
phenology. Ecology 78:1624-1631.
120
Bronstein, J. L., T. E. Huxman, and G. Davidowitz. 2007. Plant-mediated effects linking
herbivory and pollination. Pages 75-103 in T. Ohgushi, T. P. Craig, and P. W.
Price, editors. Ecological communities: plant mediation in indirect interaction
webs. Cambridge University Press, Cambridge.
Brower, L. P., W. N. Ryerson, L. Coppinger, and S. C. Glazier. 1968. Ecological
chemistry and palatability spectrum. Science 161:1349-1351.
Brown, M. J. F., R. Loosli, and P. Schmid-Hempel. 2000. Condition-dependent
expression of virulence in a trypanosome infecting bumblebees. Oikos 91:421-
427.
Brown, M. J. F., R. Schmid-Hempel, and P. Schmid-Hempel. 2003. Strong context-
dependent virulence in a host-parasite system: reconciling genetic evidence with
theory. Journal of Animal Ecology 72:994-1002.
Brysch-Herzberg, M. 2004. Ecology of yeasts in plant-bumblebee mutualism in Central
Europe. Fems Microbiology Ecology 50:87-100.
Buban, T., and Z. Orosz-Kovacs. 2003. The nectary as the primary site of infection by
Erwinia amylovora (Burr.) Winslow et al.: a mini review. Plant Systematics and
Evolution 238:183-194.
Burrow, G. E., and R. J. Tyrl. 2001. Toxic plants of North America, 1st edition. Iowa
State University Press, Ames.
121
Chapuisat, M., A. Oppliger, P. Magliano, and P. Christe. 2007. Wood ants use resin to
protect themselves against pathogens. Proceedings of the Royal Society B-
Biological Sciences 274:2013-2017.
Chittka, L., and J. D. Thomson. 2001. Cognitive ecology of pollination: animal behaviour
and floral evolution. Cambridge University Press, Cambridge.
Chittka, L., J. D. Thomson, and N. M. Waser. 1999. Flower constancy, insect
psychology, and plant evolution. Naturwissenschaften 86:361-377.
Christe, P., A. Oppliger, F. Bancala, G. Castella, and M. Chapuisat. 2003. Evidence for
collective medication in ants. Ecology Letters 6:19-22.
Cipollini, M. L., and D. J. Levey. 1997a. Secondary metabolites of fleshy vertebrate-
dispersed fruits: adaptive hypotheses and implications for seed dispersal.
American Naturalist 150:346-372.
Cipollini, M. L., and D. J. Levey. 1997b. Why are some fruits toxic? Glycoalkaloids in
Solanum and fruit choice by vertebrates. Ecology 78:782-798.
Clayton, D. H., and N. D. Wolfe. 1993. The adaptive significance of self-medication.
Trends in Ecology & Evolution 8:60-63.
Clinch, P. G., I. W. Forster, and Palmerjo.T. 1972. Effect on honey bees of nectar from
yellow kowhai (Sophora microphylla Ait). New Zealand Journal of Agricultural
Research 15:194-&.
122
Cnaani, J., R. Schmid-Hempel, and J. O. Schmidt. 2002. Colony development, larval
development and worker reproduction in Bombus impatiens Cresson. Insectes
Sociaux 49:164-170.
Cnaani, J., A. Wong, and J. D. Thomson. 2007. Effect of group size on ovarian
development in bumblebee workers (Hymenoptera : apidae : Bombus).
Entomologia Generalis 29:305-314.
Colla, S. R., M. C. Otterstatter, R. J. Gegear, and J. D. Thomson. 2006. Plight of the
bumble bee: pathogen spillover from commercial to wild populations. Biological
Conservation 129:461-467.
Cory, J. S., and K. Hoover. 2006. Plant-mediated effects in insect-pathogen interactions.
Trends in Ecology & Evolution 21:278-286.
Cowan, M. M. 1999. Plant products as antimicrobial agents. Clinical Microbiology
Reviews 12:564-582.
Cresswell, J. E., S. Z. Merritt, and M. M. Martin. 1992. The effect of dietary nicotine on
the allocation of assimilated food to energy metabolism and growth in fourth
instar larvae of the southern armyworm, Spodoptera eridania (Lepidoptera,
Noctuidae). Oecologia 89:449-453.
de Roode, J. C., A. B. Pedersen, M. D. Hunter, and S. Altizer. 2008. Host plant species
affects virulence in monarch butterfly parasites. Journal of Animal Ecology
77:120-126.
123
Despres, L., J. P. David, and C. Gallet. 2007. The evolutionary ecology of insect
resistance to plant chemicals. Trends in Ecology & Evolution 22:298-307.
Dethier, V. G. 1982. Mechanism of host-plant recognition. Entomologia Experimentalis
Et Applicata 31:49-56.
Dethier, V. G., and E. Bowdan. 1992. Effects of alkaloids on feeding by Phormia regina
confirm the critical role of sensory inhibition. Physiological Entomology 17:325-
330.
Detzel, A., and M. Wink. 1993. Attraction, deterrence or intoxication of bees (Apis
mellifera) by plant allelochemicals. Chemoecology 4:8-18.
Duchateau, M. J., and H. H. W. Velthuis. 1989. Ovarian development and egg-laying in
workers of Bombus terrestris. Entomologia Experimentalis Et Applicata 51:199-
213.
Duffey, S. S., and M. J. Stout. 1996. Antinutritive and toxic components of plant defense
against insects. Archives of Insect Biochemistry and Physiology 32:3-37.
Dukas, R., and L. A. Real. 1991. Learning foraging tasks by bees - a comparison between
social and solitary species. Animal Behaviour 42:269-276.
Durrer, S., and P. Schmid-Hempel. 1994. Shared use of flowers leads to horizontal
pathogen transmission. Proceedings of the Royal Society of London Series B-
Biological Sciences 258:299-302.
124
Dussourd, D. E., and A. M. Hoyle. 2000. Poisoned plusiines: toxicity of milkweed latex
and cardenolides to some generalist caterpillars. Chemoecology 10:11-16.
Ehlers, B. K., and J. M. Olesen. 1997. The fruit-wasp route to toxic nectar in Epipactis
orchids? Flora 192:223-229.
Elliott, S. E., R. E. Irwin, L. S. Adler, and N. M. Williams. 2008. The nectar alkaloid,
gelsemine, does not affect offspring performance of a native solitary bee, Osmia
lignaria (Megachilidae). Ecological Entomology 33:298-304.
Feeny, P. 1970. Seasonal changes in oak leaf tannins and nutrients as a cause of spring
feeding by winter moth caterpillars. Ecology 51:565-581.
Feeny, P. 1992. The evolution of chemical ecology: contributions from the study of
herbivorous insects, Vol. II: Ecological and evolutionary processes. Pages 1-44 in
G. A. Rosenthal and M. R. Berenbaum, editors. Herbivores: their interactions with
plant secondary metabolites. Academic Press, Inc., San Diego.
Feeny, P. P. 1976. Plant apparency and chemical defense. Pages 1-40 in J. W. Wallace
and R. L. Mansell, editors. Biochemical interactions between plants and insects.
Plenum, New York.
Fineblum, W. L., and M. D. Rausher. 1997. Do floral pigmentation genes also influence
resistance to enemies? The W locus in Ipomoea purpurea. Ecology 78:1646-1654.
125
Fishbein, M., D. Chuba, C. Ellison, R. Mason-Gamer, and S. P. Lynch. In Press.
Phylogenetic relationships of Asclepias (Apocynaceae) estimated from non-
coding cpDNA sequences. Systematic Botany.
Fishbein, M., and D. L. Venable. 1996. Diversity and temporal change in the effective
pollinators of Asclepias tuberosa. Ecology 77:1061-1073.
Fordyce, J. A., and S. B. Malcolm. 2000. Specialist weevil, Rhyssomatus lineaticollis,
does not spatially avoid cardenolide defenses of common milkweed by
ovipositing into pith tissue. Journal of Chemical Ecology 26:2857-2874.
Freiburghaus, F., R. Kaminsky, M. H. H. Nkunya, and R. Brun. 1996. Evaluation of
African medicinal plants for their in vitro trypanocidal activity. Journal of
Ethnopharmacology 55:1-11.
Fung, H. T., K. K. Lam, S. K. Lam, O. F. Wong, and S. K. Kam. 2007. Two cases of
Gelsemium elegans Benth. poisoning. Hong Kong Journal of Emergency
Medicine 14:221-224.
Gegear, R. J., and T. M. Laverty. 2004. Effect of a colour dimorphism on the flower
constancy of honey bees and bumble bees. Canadian Journal of Zoology-Revue
Canadienne De Zoologie 82:587-593.
Gegear, R. J., J. S. Manson, and J. D. Thomson. 2007. Ecological context influences
pollinator deterrence by alkaloids in floral nectar. Ecology Letters 10:375-382.
126
Gegear, R. J., M. C. Otterstatter, and J. D. Thomson. 2005. Does parasitic infection
impair the ability of bumblebees to learn flower-handling techniques? Animal
Behaviour 70:209-215.
Gegear, R. J., M. C. Otterstatter, and J. D. Thomson. 2006. Bumble-bee foragers infected
by a gut parasite have an impaired ability to utilize floral information.
Proceedings of the Royal Society B-Biological Sciences 273:1073-1078.
Gimenez-Jurado, G., C. P. Kurtzman, W. T. Starmer, and I. Spencer-Martins. 2003.
Metschnikowia vanudenii sp nov and Metschnikowia lachancei sp nov., from
flowers and associated insects in North America. International Journal of
Systematic and Evolutionary Microbiology 53:1665-1670.
Glendinning, J. I. 2002. How do herbivorous insects cope with noxious secondary plant
compounds in their diet? Entomologia Experimentalis Et Applicata 104:15-25.
Glendinning, J. I., N. M. Nelson, and E. A. Bernays. 2000. How do inositol and glucose
modulate feeding in Manduca sexta caterpillars? Journal of Experimental Biology
203:1299-1315.
Golonka, A. M. 2002. Nectar-inhabiting microorganisms and the dioecious plant species
Silene latifolia. PhD Thesis. Duke University, Durham.
Gomez, J. M. 2003. Herbivory reduces the strength of pollinator-mediated selection in
the Mediterranean herb Erysimum mediohispanicum: consequences for plant
specialization. American Naturalist 162:242-256.
127
Gumbert, A. 2000. Color choices by bumble bees (Bombus terrestris): innate preferences
and generalization after learning. Behavioral Ecology and Sociobiology 48:36-43.
Hagler, J. R., and S. L. Buchmann. 1993. Honey bee (Hymenoptera, Apidae) foraging
responses to phenolic-rich nectars. Journal of the Kansas Entomological Society
66:223-230.
Harder, L. D. 1982. Measurement and estimation of functional proboscis length in
bumblebees (Hymenoptera, Apidae). Canadian Journal of Zoology-Revue
Canadienne De Zoologie 60:1073-1079.
Harder, L. D., and S. C. H. Barrett. 1993. Pollen removal from tristylous Pontederia
cordata - effects of anther position and pollinator specialization. Ecology
74:1059-1072.
Hardin, J. W., and J. M. Arena. 1969. Human poisoning from native and cultivated
plants. Duke University Press, Durham.
Hartmann, T. 1992. Alkaloids. Pages 79-121 in G. A. Rosenthal and M. R. Berebaum,
editors. Herbivores: their interactions with secondary plant metabolites, Vol. I:
The chemical participants. Academic Press, Inc., San Diego.
Heinrich, B. 1975. Bee flowers - hypothesis on flower variety and blooming times.
Evolution 29:325-334.
Heinrich, B. 1979. Bumblebee economics. Harvard University Press, Cambridge.
128
Heinrich, B., P. R. Mudge, and P. G. Deringis. 1977. Laboratory analysis of flower
constancy in foraging bumblebees - Bombus ternarius and Bombus terricola.
Behavioral Ecology and Sociobiology 2:247-265.
Herrera, C. M., I. M. Garcia, and R. Perez. 2008. Invisible floral larcenies: microbial
communities degrade floral nectar of bumble bee-pollinated plants. Ecology
89:2369-2376.
Herrera, C. M., M. Medrano, P. J. Rey, A. M. Sanchez-Lafuente, M. B. Garcia, J.
Guitian, and A. J. Manzaneda. 2002. Interaction of pollinators and herbivores on
plant fitness suggests a pathway for correlated evolution of mutualism- and
antagonism-related traits. Proceedings of the National Academy of Sciences of the
United States of America 99:16823-16828.
Imhoof, B., and P. Schmid-Hempel. 1999. Colony success of the bumble bee, Bombus
terrestris, in relation to infections by two protozoan parasites, Crithidia bombi
and Nosema bombi. Insectes Sociaux 46:233-238.
Irwin, R. E., and L. S. Adler. 2006. Correlations among traits associated with herbivore
resistance and pollination: implications for pollination and nectar robbing in a
distylous plant. American Journal of Botany 93:64-72.
Irwin, R. E., L. S. Adler, and A. K. Brody. 2004. The dual role of floral traits: pollinator
attraction and plant defense. Ecology 85:1503-1511.
129
Irwin, R. E., S. Y. Strauss, S. Storz, A. Emerson, and G. Guibert. 2003. The role of
herbivores in the maintenance of a flower color polymorphism in wild radish.
Ecology 84:1733-1743.
Janzen, D. H. 1973. Community structure of secondary compounds in plants. Pure and
Applied Chemistry 34:529-538.
Janzen, D. H. 1977. Why don't ants visit flowers? Biotropica 9:252-252.
Johnson, S. D., A. L. Hargreaves, and M. Brown. 2006. Dark, bitter-tasting nectar
functions as a filter of flower visitors in a bird-pollinated plant. Ecology 87:2709-
2716.
Karban, R., and I. T. Baldwin. 1997. Induced responses to herbivory. University of
Chigaco Press, Chigaco.
Karban, R., and G. English-Loeb. 1997. Tachinid parasitoids affect host plant choice by
caterpillars to increase caterpillar survival. Ecology 78:603-611.
Karowe, D. N. 1989. Differential effect of tannic acid on 2 tree-feeding lepidoptera -
implications for theories of plant anti-herbivore chemistry. Oecologia 80:507-512.
Kearns, C. A., and J. D. Thomson. 2001. The natural history of bumblebees: a
sourcebook for investigations. University Press of Colorado, Boulder.
Kephart, S., and K. Theiss. 2004. Pollinator-mediated isolation in sympatric milkweeds
(Asclepias): do floral morphology and insect behavior influence species
boundaries? New Phytologist 161:265-277.
130
Kevan, P. G., D. Eisikowitch, and B. Rathwell. 1989. The role of nectar in the
germination of pollen in Asclepias syriaca-L. Botanical Gazette 150:266-270.
Kim, Y. S., and B. H. Smith. 2000. Effect of an amino acid on feeding preferences and
learning behavior in the honey bee, Apis mellifera. Journal of Insect Physiology
46:793-801.
Konig, B. 1988. The honeybee as pharmacophorus insect. Entomologia Generalis 14:145-
148.
Kotler, B. P., and L. Blaustein. 1995. Titrating food and safety in a heterogeneous
environment: when are the risky and safe patches of equal value? Oikos 74:251-
258.
Kurtzman, C. P., and C. J. Robnett. 1998. Identification and phylogeny of ascomycetous
yeasts from analysis of nuclear large subunit (26S) ribosomal DNA partial
sequences. Antonie Van Leeuwenhoek International Journal of General and
Molecular Microbiology 73:331-371.
Lachance, M. A. 1987. Approaches to yeast identification. Pages 33-51 in D. R. Berry, I.
Russell, and G. G. Stewart, editors. Yeast biotechnology. Allen & Unwin,
London.
Lachance, M. A., W. T. Starmer, C. A. Rosa, J. M. Bowles, J. S. F. Barker, and D. H.
Janzen. 2001. Biogeography of the yeasts of ephemeral flowers and their insects.
FEMs Yeast Research 1:1-8.
131
Landolt, P. J., and B. Lenczewski. 1993. Lack of evidence for the toxic nectar hypothesis
- a plant alkaloid did not deter nectar feeding by lepidoptera. Florida Entomologist
76:556-566.
Lee, K. P., J. S. Cory, K. Wilson, D. Raubenheimer, and S. J. Simpson. 2006. Flexible
diet choice offsets protein costs of pathogen resistance in a caterpillar.
Proceedings of the Royal Society B-Biological Sciences 273:823-829.
Leege, L. M., and L. M. Wolfe. 2002. Do floral herbivores respond to variation in flower
characteristics in Gelsemium sempervirens (Loganiaceae), a distylous vine?
American Journal of Botany 89:1270-1274.
Lin, H. R., and M. L. Winston. 1998. The role of nutrition and temperature in the ovarian
development of the worker honey bee (Apis mellifera). Canadian Entomologist
130:883-891.
Lipa, J. J., and O. Triggiani. 1988. Crithidia bombi sp n. a flagellated parasite of a
bumblebee Bombus terrestris L. (Hymenoptera, Apidae). Acta Protozoologica
27:287-290.
Liu, F., J. Chen, J. Chai, X. Zhang, X. Bai, D. He, and D. W. Roubik. 2007. Adaptive
functions of defensive plant phenolics and a non-linear bee response to nectar
components. Functional Ecology 21:96-100.
Liu, F. L., J. Z. He, and W. J. Fu. 2005. Highly controlled nest homeostasis of honey bees
helps deactivate phenolics in nectar. Naturwissenschaften 92:297-299.
132
Logan, A., M. X. Ruiz-Gonzalez, and M. J. F. Brown. 2005. The impact of host
starvation on parasite development and population dynamics in an intestinal
trypanosome parasite of bumble bees. Parasitology 130:637-642.
London-Shafir, I., S. Shafir, and D. Eisikowitch. 2003. Amygdalin in almond nectar and
pollen - facts and possible roles. Plant Systematics and Evolution 238:87-95.
Malcolm, S. B. 1991. Cardenolide-mediated interaction between plants and herbivores.
Pages 252-296 in G. A. Rosenthal and M. Berebaum, editors. Herbivores: their
interactions with secondary plant metabolites, Vol. II: Ecological and
evolutionary processes. Academic Press, San Diego.
Manson, J. S., M. A. Lachance, and J. D. Thomson. 2007. Candida gelsemii sp. nov., a
yeast of the Metschnikowiaceae clade isolated from nectar of the poisonous
Carolina jessamine. Antonie Van Leeuwenhoek International Journal of General
and Molecular Microbiology 92:37-42.
Manson, J. S., M. C. Otterstatter, and J. D. Thomson. In Press. Consumption of a nectar
alkaloid reduceds pathogen load in bumble bees. Oecologia.
Manson, J. S., and J. D. Thomson. 2009. Post-ingestive effects of nectar alkaloids depend
on dominance status of bumble bees. Ecological Entomology 34:421-426.
Marcucci, M. C. 1995. Propolis - chemical composition, biological properties and
therapeutic activity. Apidologie 26:83-99.
133
Masters, A. R. 1991. Dual role of pyrrolizidine alkaloids in nectar. Journal of Chemical
Ecology 17:195-205.
McCall, A. C. 2006. Natural and artificial floral damage induces resistance in Nemophila
menziesii (Hydrophyllaceae) flowers. Oikos 112:660-666.
McCall, A. C., and R. E. Irwin. 2006. Florivory: the intersection of pollination and
herbivory. Ecology Letters 9:1351-1365.
McCall, A. C., and R. Karban. 2006. Induced defense in Nicotiana attenuata
(Solanaceae) fruit and flowers. Oecologia 146:566-571.
McKey, D. 1974. Adaptive patterns in alkaloid physiology. American Naturalist 108:305-
320.
Miller, M. W., and H. J. Phaff. 1998. Metschnikowia Kamienski. Pages 256-267 in C. P.
Kurtzman and J. W. Fell, editors. The yeasts, a taxonomic study. Elsevier,
Amsterdam.
Mitchell, B. K. 1987. Interactions of alkaloids with galeal chemosensory cells of
Colorado potato beetle. Journal of Chemical Ecology 13:2009-2022.
Mitchell, B. K., and J. F. Sutcliffe. 1984. Sensory inhibition as a mechanism of feeding
deterrence - effects of 3 alkaloids on leaf beetle feeding. Physiological
Entomology 9:57-64.
Morse, D. H. 1982. The turnover of milkweed pollinia on bumble bees, and implications
for outcrossing. Oecologia 53:187-196.
134
Mosquin, T. 1971. Competition for pollinators as a stimulus for evolution of flowering
time. Oikos 22:398-402.
Neal, J. J. 1987. Metabolic costs of mixed-function oxidase induction in Heliothis zea.
Entomologia Experimentalis Et Applicata 43:175-179.
Nelson, C. J., J. N. Seiber, and L. P. Brower. 1981. Seasonal and intraplant variation of
cardenolide content in the California milkweed, Asclepias eriocarpa, and
implications for plant defense. Journal of Chemical Ecology 7:981-1010.
Oksanen, J. 2009. vegan: R functions for vegetation ecologists. URL:
http://cc.oulu.fi/~jarioksa/softhelp/vegan.html.
Ornduff, R. 1970. Systematics and breeding system of Gelsemium (Loganiaceae). Journal
of the Arnold Arboretum 51:1-17.
Osier, T. L., S. Y. Hwang, and R. L. Lindroth. 2000. Effects of phytochemical variation
in quaking aspen Populus tremuloides clones on gypsy moth Lymantria dispar
performance in the field and laboratory. Ecological Entomology 25:197-207.
Ott, J. 1998. The Delphic Bee: bees and toxic honeys as pointers to psychoactive and
other medicinal plants. Economic Botany 52:260-266.
Otterstatter, M. C., R. J. Gegear, S. R. Colla, and J. D. Thomson. 2005. Effects of
parasitic mites and protozoa on the flower constancy and foraging rate of bumble
bees. Behavioral Ecology and Sociobiology 58:383-389.
135
Otterstatter, M. C., and J. D. Thomson. 2006. Within-host dynamics of an intestinal
pathogen of bumble bees. Parasitology 133:749-761.
Otterstatter, M. C., and J. D. Thomson. 2007. Contact networks and transmission of an
intestinal pathogen in bumble bee (Bombus impatiens) colonies. Oecologia
154:411-421.
Paradis, E. 2009. ape: analyses of phylogenetics and evolution. URL:
http://ape.mpl.ird.fr/.
Pascarella, J. B. 2007. Mechanisms of prezygotic reproductive isolation between two
sympatric species, Gelsemium rankinii and G. sempervirens (Gelsemiaceae), in
the southeastern United States. American Journal of Botany 94:468-476.
Pernal, S. F., and R. W. Currie. 2000. Pollen quality of fresh and 1-year-old single pollen
diets for worker honey bees (Apis mellifera L.). Apidologie 31:387-409.
Pilson, D. 2000. Herbivory and natural selection on flowering phenology in wild
sunflower, Helianthus annus. Oecologia 122:72-82.
Praz, C. J., A. Mueller, and S. Dorn. 2008. Specialized bees fail to develop on non-host
pollen: do plants chemically protect their pollen? Ecology 89:795-804.
Price, P. W., C. E. Bouton, P. Gross, B. A. McPheron, J. N. Thompson, and A. E. Weis.
1980. Interactions among 3 trophic levels - influence of plants on interactions
between insect herbivores and natural enemies. Annual Review of Ecology and
Systematics 11:41-65.
136
Proctor, M., P. Yeo, and A. Lack. 1996. The natural history of pollination. Timber Press,
Portland.
Pryce-Jones, J. 1942. Some problems associated with nectar, pollen and honey.
Proceedings of the Linnean Society of London 155:129-174.
R Development Core Team. 2009. R: a language and environment for statistical
computing. Foundation for Statistical Computing, Vienna; URL: http://ww.R-
project.org.
Radford, A. E., H. E. Ahles, and C. R. Bell. 1968. Manual of the vascular flora of the
Carolinas. University of North Carolina Press, Chapel Hill.
Raguso, R. A. 2008. Wake up and smell the roses: the ecology and evolution of floral
scent. Annual Review of Ecology Evolution and Systematics 39:549-569.
Rasmann, S., A. A. Agrawal, S. C. Cook, and A. C. Erwin. In Press. Cardenolides,
induced responses, and interactions between above and belowground herbivores
in the milkweeds (Asclepias spp.). Ecology.
Real, L. A. 1991. Animal choice behavior and the evolution of cognitive architecture.
Science 253:980-986.
Rhoades, D. F. 1979. Evolution of plant chemical defense against herbivores. in G. A.
Rosenthal and D. H. Janzen, editors. Herbivores: their interaction with secondary
plant metabolites. Academic Press, San Diego.
137
Rhoades, D. F., and J. C. Bergdahl. 1981. Adaptive significance of toxic nectar.
American Naturalist 117:798-803.
Rosa, C. A., M. A. Lachance, J. O. C. Silva, A. C. P. Teixeira, M. M. Marini, Y.
Antonini, and R. P. Martins. 2003. Yeast communities associated with stingless
bees. FEMs Yeast Research 4:271-275.
Rosenthal, G. A., and M. R. Berenbaum. 1991. Herbivores: Their interactions with
secondary plant metabolites, Vol. I: The chemical participants, 2nd edition.
Academic Press, San Diego.
Rujjanawate, C., D. Kanjanapothi, and A. Panthong. 2003. Pharmacological effect and
toxicity of alkaloids from Gelsemium elegans Benth. Journal of
Ethnopharmacology 89:91-95.
SAS Institute. 1999. SAS user's guide. SAS Institute, Cary.
SAS Institute. 2006. SAS/STAT 9.1 user's guide. SAS Institute, Cary.
Schmid-Hempel, P. 1998. Parasites in social insects. Princeton University Press,
Princeton.
Schmid-Hempel, P. 2001. On the evolutionary ecology of host-parasite interactions:
addressing the question with regard to bumblebees and their parasites.
Naturwissenschaften 88:147-158.
138
Schmid-Hempel, P., and R. Schmid-Hempel. 1993. Transmission of a pathogen in
Bombus terrestris, with a note on division-of-labor in social insects. Behavioral
Ecology and Sociobiology 33:319-327.
Schmid-Hempel, P., and H. P. Stauffer. 1998. Parasites and flower choice of bumblebees.
Animal Behaviour 55:819-825.
Schmid-Hempel, R., and P. Schmid-Hempel. 1991. Endoparasitic flies, pollen-collection
by bumblebees and a potential host-parasite conflict. Oecologia 87:227-232.
Shields, V. D. C., and B. K. Mitchell. 1995. The effect of phagostimulant mixtures on
deterrent receptor(s) in 2 crucifer-feeding lepidopterous species. Philosophical
Transactions of the Royal Society of London Series B-Biological Sciences
347:459-464.
Shykoff, J. A., and P. Schmid-Hempel. 1991. Genetic relatedness and eusociality -
parasite-mediated selection on the genetic composition of groups. Behavioral
Ecology and Sociobiology 28:371-376.
Siegert, K. J. 1987. Carbohydrate metabolism in Manduca sexta during late larval
development. Journal of Insect Physiology 33:421-427.
Sih, A. 1980. Optimal behavior - can foragers balance 2 conflicting demands. Science
210:1041-1043.
Simms, E. L., and M. A. Bucher. 1996. Pleiotropic effects of flower-color intensity on
herbivore performance on Ipomoea purpurea. Evolution 50:957-963.
139
Singaravelan, N., G. Nee'man, M. Inbar, and I. Izhaki. 2005. Feeding responses of free-
flying honeybees to secondary compounds mimicking floral nectars. Journal of
Chemical Ecology 31:2791-2804.
Singer, M. S., Y. Carriere, C. Theuring, and T. Hartmann. 2004. Disentangling food
quality from resistance against parasitoids: diet choice by a generalist caterpillar.
American Naturalist 164:423-429.
Singer, M. S., K. C. Mace, and E. A. Bernays. 2009. Self-medication as adaptive
plasticity: increased ingestion of plant toxins by parasitized caterpillars. PLoS
ONE 4:e4796.
Slansky, F. 1992. Allelochemical-nutrient interactions in herbivore nutrient ecology, Vol.
II: Ecological and evolutionary processes. Pages 135-176 in G. A. Rosenthal and
M. R. Berenbaum, editors. Herbivores: their interactions with secondary plant
metabolites. Academic Press, San Diego.
Sokal, R. R., and F. J. Rohlf. 1995. Biometry, 3rd edition. W.H. Freeman and Company,
New York.
Stephenson, A. G. 1981. Toxic nectar deters nectar thieves of Catalpa speciosa.
American Midland Naturalist 105:381-383.
Stephenson, A. G. 1982. Iridoid glycosides in the nectar of Catalpa speciosa are
unpalatable to nectar thieves. Journal of Chemical Ecology 8:1025-1034.
140
Stiles, B., and J. D. Paschke. 1980. Midgut pH in different instars of 3 Aedes mosquito
species and the relation between pH and susceptibility of larvae to a nuclear
polyhedrosis virus. Journal of Invertebrate Pathology 35:58-64.
Strauss, S. Y. 1997. Floral characters link herbivores, pollinators, and plant fitness.
Ecology 78:1640-1645.
Strauss, S. Y., J. K. Conner, and S. L. Rush. 1996. Foliar herbivory affects floral
characters and plant attractiveness to pollinators: implications for male and female
plant fitness. American Naturalist 147:1098-1107.
Strauss, S. Y., J. A. Rudgers, J. A. Lau, and R. E. Irwin. 2002. Direct and ecological costs
of resistance to herbivory. Trends in Ecology & Evolution 17:278-285.
Strauss, S. Y., D. H. Siemens, M. B. Decher, and T. Mitchell-Olds. 1999. Ecological
costs of plant resistance to herbivores in the currency of pollination. Evolution
53:1105-1113.
Strauss, S. Y., and J. B. Whittall. 2006. Non-pollinator agents of selection on floral traits.
Pages 120-138 in L. D. Harder and S. C. H. Barrett, editors. Ecology and
evolution of flowers. Oxford University Press, Oxford.
Tadmor-Melamed, H., S. Markman, A. Arieli, M. Distl, M. Wink, and I. Izhaki. 2004.
Limited ability of Palestine sunbirds Nectarinia osea to cope with pyridine
alkaloids in nectar of tree tobacco Nicotiana glauca. Functional Ecology 18:844-
850.
141
Tan, K., Y. H. Guo, S. W. Nicolson, S. E. Radloff, Q. S. Song, and H. R. Hepburn. 2007.
Honeybee (Apis cerana) foraging responses to the toxic honey of Tripterygium
hypoglaucum (Celastraceae): changing threshold of nectar acceptability. Journal
of Chemical Ecology 33:2209-2217.
Thompson, S. N. 2003. Trehalose - The insect 'blood' sugar. Pages 205-285 in Advances
in Insect Physiology, Vol 31. Academic Press, San Diego.
Thomson, J. D. 1986. Pollen transport and deposition by bumble bees in Erythronium -
influences of floral nectar and bee grooming. Journal of Ecology 74:329-341.
Thornburg, R. W., C. Carter, A. Powell, R. Mittler, L. Rizhsky, and H. T. Horner. 2003.
A major function of the tobacco floral nectary is defense against microbial attack.
Plant Systematics and Evolution 238:211-218.
Van der Putten, W. H., L. E. M. Vet, J. A. Harvey, and F. L. Wackers. 2001. Linking
above- and belowground multitrophic interactions of plants, herbivores,
pathogens, and their antagonists. Trends in Ecology & Evolution 16:547-554.
Vanetten, H., E. Temporini, and C. Wasmann. 2001. Phytoalexin (and phytoanticipin)
tolerance as a virulence trait: why is it not required by all pathogens?
Physiological and Molecular Plant Pathology 59:83-93.
Wahl, O., and K. Ulm. 1983. Influence of pollen feeding and physiological condition on
pesticide sensitivity of the honey bee Apis mellifera Carnica. Oecologia 59:106-
128.
142
Webster, S. J., and L. M. Dill. 2006. The energetic equivalence of changing salinity and
temperature to juvenile salmon. Functional Ecology 20:621-629.
Whittaker, R. H., and P. P. Feeny. 1971. Allelochemics - chemical interactions between
species. Science 171:757-770.
Wink, M., and O. Schimmer. 1999. Modes of action of defensive secondary metabolites.
Pages 17-133 in M. Wink, editor. Functions of plant secondary metabolites and
their exploitation in biotechnology, Annual Plant Reviews Vol. III. Wiley-
Blackwell.
Wink, M., and V. Theile. 2002. Alkaloid tolerance in Manduca sexta and
phylogenetically related sphingids (Lepidoptera : Sphingidae). Chemoecology
12:29-46.
Woodson, R. E. 1954. The North American species of Asclepias L. Annals of the
Missouri Botanical Garden 41:1-211.
Wyatt, R., and S. B. Broyles. 1994. Ecology and evolution of reproduction in milkweeds.
Annual Review of Ecology and Systematics 25:423-441.
Wyatt, R., S. B. Broyles, J. L. Hamrick, and A. Stoneburner. 1993. Systematic
relationships within Gelsemium (Loganiaceae) - evidence from isozymes and
cladistics. Systematic Botany 18:345-355.
143
Yarrow, D. 1998. Methods for the isolation and identification of yeasts. Pages 77-100 in
C. P. Kurtzman and J. W. Fell, editors. The yeasts, a taxonomic study. Elsevier,
Amsterdam.
Zangerl, A. R., and M. R. Berenbaum. 1993. Plant chemistry, insect adaptations to plant
chemistry, and host plant utilization patterns. Ecology 74:47-54.
Zehnder, C. B., and M. D. Hunter. 2007. Interspecific variation within the genus
Asclepias in response to herbivory by a phloem-feeding insect herbivore. Journal
of Chemical Ecology 33:2044-2053.
Zhang, Z., S. Schwartz, L. Wagner, and W. Miller. 2000. A greedy algorithm for aligning
DNA sequences. Journal of Computational Biology 7:203-214.
144
APPENDIX ONE
Candida gelsemii sp. nov., a yeast of the Metshnikowiaceae clade isolated from
nectar of the poisonous Carolina jessamine
Jessamyn S. Manson, Marc-André Lachance and James D. Thomson
This project was conceived by me with guidance from both M.A. Lachance and J.D. Thomson. The work
was done in collaboration with M.A. Lachance, who also co-wrote the manuscript, now published in
Antonie van Leeuwenhoeck, 2007, 92: 37-42.
Abstract
A new yeast species, Candida gelsemii, is described to accommodate three
isolates recovered in Georgia, USA, from the toxic nectar of the Carolina jessamine
(Gelsemium sempervirens). The species resembles other members of the
Metschnikowiaceae clade that have been recovered from nectar, but differs in a number
of morphological and physiological characteristics. Analysis of rDNA sequences places
the new species well into the clade, but in a basal position with respect to a group of
Metschnikowia and Candida species known to occur in association with nectars, bees, as
well as marine invertebrates. The type is strain UWOPS 06-24.1T (CBS 10509
T,
NRRL Y-48212 T
).
145
Introduction
Floral nectars often contain yeasts that are vectored by pollinating and non-
pollinating insects (Lachance et al. 2001, Rosa et al. 2003, Brysch-Herzberg 2004). The
role of the yeasts in this ecosystem is poorly understood, but it is clear that the
composition of the yeast community is highly dependent on the types of insects involved.
Whereas nitidulid beetles carry yeasts with affinities in the genera Metschnikowia,
Kodamaea, and Wickerhamiella, bees tend to vector other yeasts related to the genera
Metschnikowia and Starmerella. Even within the Metschnikowiaceae clade, the species
associated with nitidulids and those associated with bees are not the same. It was
therefore of interest to examine the nectar of the Carolina jessamine (Gelsemium
sempervirens). This perennial vine is endemic to the southeastern United States. It is
also a distylous species, making it an obligate outcrosser. It blooms in the early spring,
producing a multitude of fragrant tubular yellow flowers which attract a diverse range of
pollinators including eusocial, solitary, and nectar-robbing bees (Ornduff 1970, Adler and
Irwin 2005, 2006). Most importantly, the nectar of this plant contains gelsemine, an
alkaloid that is highly toxic to vertebrates (Burrow and Tyrl 2001). The alkaloid,
presumed to be a deterrent for herbivores, is reported to deter pollinators and nectar
robbers at natural concentrations (Adler and Irwin 2005). Adaptive hypotheses for the
presence of alkaloids in floral nectar include providing the nectar with antimicrobial
properties to prevent the proliferation of organisms such as floral yeasts (Adler 2000).
In the course of determining whether the yeast community of the nectar is also
affected by gelsemine, we sampled nectars from G. sempervirens and also some
146
sympatric azaleas that appeared to have a similar array of bees foraging for nectar. At the
time of collecting, both plant species were visited by queens of the native bumble bees
Bombus impatiens and B. bimaculatus, as well as introduced honey bees (Apis mellifera).
In the process, strains of Metschnikowia reukaufii were recovered from azalea nectar,
whereas the jessamine nectars yielded strains of Candida rancoensis as well as a new
asexual species with metschnikowiaceous affinities, which we now describe as Candida
gelsemii.
Methods
Nectar samples were collected on March 30th
and 31st 2006 in four sites located
near the campus of Georgia Southern University in Statesboro, Georgia. Other isolation
details are given in Table A.1. In each case, approximately 2 µL of nectar was placed
onto a plate of YM agar supplemented with 100 mg/L chloramphenicol and the nectar
was streak-diluted with a sterile loop. Mould colonies were removed periodically with a
knife. The plates were returned to the laboratory (UWO) and colonies were picked for
purification and identification by rDNA partial sequencing (Kurtzman and Robnett 1998).
Sequence editing, alignment, and analysis were conducted with DNAMAN version 4.1.
The sequences were queried against the GenBank database, using the Megablast
algorithm of Zhang et al. (2000). Strain characterization followed standard methods
(Yarrow 1998). Growth responses were determined by replica plating as detailed by
Lachance (1987). Replica plating was also used to evaluate the effect of gelsemine on
yeast growth. The potential effect of gelsemine on yeasts was also assesses by agar
diffusion. An ethanol solution of gelsemine was added to sterile discs of Whatman 3mm
paper so that each disk contained 35, 3.5, and 0.35 µg, respectively. The air-dried discs
147
were then placed individually on the surface of YM agar plates inoculated with dilute
yeast suspensions. The plates were examined periodically for evidence of inhibition
zones.
Results and Discussion
Species delineation, phylogenetic placement and phenotypic variability
The three isolates of Candida gelsemii were similar but not identical in sequences,
morphology, and growth responses. The ribosomal internal transcribed spacer (rDNA
ITS) sequences of strains 06-17.1 and 06-24.1 differed by a single indel and that of strain
06-11.1 differed from the other two by 8 substitutions and one or two indels. However,
the nearly identical large subunit rDNA D1/D2 regions (strain 06-11.1 differs by a single
substitution in the D1 domain) and a comparison of morphology and physiology within
the greater context of other related yeast species supports assigning the isolates to a single
species. Extensive attempts to obtain evidence of a sexual cycle were not successful, and
so a biological species concept cannot be applied at present. A Megablast search of the
GenBank database using the D1/D2 domains identified the nearest known relative as
Metschnikowia bicuspidata, with a divergence of 57 substitutions. Figure A.1 shows that
the new species occupies a somewhat basal position with respect to that group, which
contains a number of species that are often isolated from nectars or other plant
components, as well as a small subclade of species thought to be parasitic on certain
aquatic invertebrates (i.e., Metschnikowia bicuspidata and allies; Miller & Phaff 1998).
Sisterhood of Candida gelsemii and Candida rancoensis is not well supported by the
data. However, addition of any other species to the sequence analysis did not alter the
148
presumed monophyly of the ingroup species included in Fig. A.1 (Metschnikowia
pulcherrima was the outgroup).
Considering the narrow distribution of the known isolates of Candida gelsemii,
their phenotypic variation is significant. As shown in Fig. A.2, our attempts to obtain
ascus formation on dilute V8 agar, although unsuccessful, demonstrated that the three
strains are distinct with respect to cell size and propensity to differentiate into
chlamydospores that are often referred to as “pulcherrima cells”, in reference to the
resting, pre-ascal cells formed by Metschnikowia pulcherrima (Miller and Phaff 1998).
The large lipid globules seen in strain 06-24.1 are most typical of this. The formation of
bilobate lipid globules by some of the cells is not typical, however. The variation
observed at the physiological level, detailed in Table A.2, was not obviously correlated
with the extent of divergence in ITS sequences.
Physiologically, Candida gelsemii superficially resembles phylogenetic congeners
such as Metschnikowia lachancei and M. vanudenii, and to a lesser extent M. bicuspidata
and M. gruesii, most of which have been isolated from floral nectars (Miller and Phaff
1998, Gimenez-Jurado et al. 2003). M. bicuspidata is of marine origin. The most
important differences from other nectar isolates were the weak or negative growth at
30ºC or in the presence of 50% glucose, and the rather weak fermentation. Such
differences would not constitute strong key characters for identification.
Ecology
The presence of a highly potent toxic alkaloid in the nectar of Gelsemium
sempervirens might constitute a selective factor that favours the presence of resistant
yeast species over the more frequently isolated nectar species such as Metschnikowia
149
reukaufii. To test the hypothesis that gelsemine might be such a niche determinant, we
tested its effect on the growth of the yeasts listed in Table A.1 as well as others, using
two approaches. Two concentrations of synthetic gelsemine, 100ng/µL and 250ng/µL,
were added to YM agar and used in the replica plate series used to characterize the
isolates as well as others that came from nectar of a tropical palm. These concentrations
simulate levels of gelsemine that occur in the nectar of natural G. sempervirens
populations and are known to deter several species of flower visitors, including the
bumble bee Bombus impatiens, an important pollinator of G. sempervirens (Adler and
Irwin 2005, 2006, Manson, personal observation). To ensure that these concentrations
were neither insufficient nor excessive, disks impregnated with gelsemine were also
applied to lawns of yeast on agar. The species tested included M. pulcherrima, M.
reukaufii, Debaryomyces melissophilus, and Starmerella bombicola. In all cases, no
significant effect was detected, indicating that neither yeasts from jessamine nectar nor
those from the nectar of other plants experienced reduced growth due to the presence of
gelsemine. The mechanism of tolerance to this alkaloid is unknown but appears to be
generalized within a broad range of yeasts and suggests that predictions of the toxicity of
nectar alkaloids to microbial communities may be incorrect (Adler 2000). We propose
that perhaps the yeasts found in jessamine nectar may instead act as a detoxifying agent
selectively carried by visitors to that plant as a co-evolved adaptation. We hope to test
that hypothesis in the future.
150
Description of Candida gelsemii Lachance sp. nov.
In 2% glucose 0.5% yeast extract after 3 days, the cells are ovoid, occur singly or
in bud-mother cell pairs, and measure 3-7 x 5-10 µm. Neither a ring nor a pellicle is
formed. On agar media, the colonies are low-convex, slightly umbonate with an entire
margin. The surface is glabrous and can be papillate or pitted. On Dalmau plates with
YCB agar supplemented with 0.01% ammonium sulfate, after 2 weeks, a few chains of
undifferentiated cells may be formed after two weeks. The cultures were examined
individually or mixed in pairs on YCB agar with 0.01% ammonium sulfate and dilute
(1/20) V8 agar. Mating or ascus formation were not observed. Resting cells with
conspicuous lipid globules may be formed on dilute V8 (Fig. A.2). Glucose is fermented
weakly. Glucose, sucrose, maltose, melezitose, cellobiose (variable or slow), salicin
(variable or weak), glycerol (weak), glucitol (weak), gluconic acid, glucono-∆-lactone
(weak), N-acetyl glucosamine, and hexadecane (weak) are assimilated, but not inulin,
raffinose, melibiose, galactose (sometimes weak), lactose, trehalose (sometimes weak), α-
methyl-D-glucoside, sorbose, rhamnose, xylose, L-arabinose, D-arabinose, ribose,
methanol, 1-propanol, 2-propanol, 1-butanol, erythritol, ribitol, xylitol, galactitol,
mannitol, inositol, lactic acid, succinic acid (sometimes weak), citric acid, malic acid, 2-
keto-gluconic acid, or glucosamine (occasionally weak). Ethylamine, L-lysine, and
cadaverine are utilized as sole nitrogen sources, but not nitrate or nitrite. Growth in
vitamin-free medium negative. Growth in amino acid-free medium positive. Growth at
4ºC weak, at 24ºC positive, at 30ºC negative or weak. Gelatin hydrolysis weak. Casein
hydrolysis positive. Tween 80 hydrolysis positive. Acid production on chalk agar
negative. Growth in YM agar with 10% NaCl positive or slow; 15% negative. Growth in
151
50% W/W glucose, 1% yeast extract agar negative. Growth in the presence of 10 mg/L
cycloheximide negative. Growth in the presence of 75 mg/L CTAB positive. Starch
production negative. Diazonium Blue B reaction negative. The habitat is nectar of
Gelsemium sempervirens in Georgia, USA. The type culture is strain UWOPS 06-24.1T
isolated from nectar. The type was deposited in the culture collection of the
Centraalbureau voor Schimmelculture, Utrecht, the Netherlands (CBS 10509, =
NRRL Y-48212 T
). Mycobank 46490.
Etymology: gel.se’mi.i, L. gen. sing. neut. n., gelsemii, of Gelsemium, referring to
the plant from which the isolates were obtained.
Acknowledgements
This work was funded by grants from the Natural Science and Engineering Research
Council of Canada (MAL and JDT). We thank Sheila Colla and Erin Willis for their
assistance during field collections, and Lissa Leege for directing us towards our field
sites.
152
Table A.1. Summary of yeasts recovered from nectar of flowers obtained in the vicinity
of Georgia Southern University campus in Statesboro, Georgia.
Yeast Species
Strain Plant Species (N) Locality
Candida gelsemii
06-11.1
25.5 km east of campus
06-17.1
25 km east of campus
06-24.1
Candida rancoensis
06-22.1
06-26.1
Gelsemium sempervirens (34)
1.6 km east of campus
Metschnikowia reukaufii
06-29.1
06-32.1
Horticultural azalea (6)
University campus
153
Table A.2. Growth responses that exhibit variation among the three known strains of
Candida gelsemii. Growth is scored from weakest to strongest response as follows: - = no
growth, w = weak growth, s= slow growth, + = successful growth.
Strain Growth Test
06-11.1
06-17.1
06-24.1T
Galactose w - -
Trehalose - - w
Cellobiose s - s
Salicin w - w
Succinic Acid - w -
Glucosamine w - -
30 ºC w - -
10% NaCl
s + s
154
Candida gelsemii UWOPS 06-24.1T (DQ988046)
Candida gelsemii UWOPS 06-17.1 (DQ988047)
64
Candida gelsemii UWOPS 06-11.1 (DQ988045)
100
Candida rancoensis CBS 8174T (AJ508580)
Metschnikowia zobellii NRRL Y-5387T (U44823)
Metschnikowia bicuspidata NRRL YB-4993NT (U44822)
Metschnikowia australis NRRL Y-17414T (U76526)80
78
Metschnikowia krissii NRRL Y-5389T (U45735)
60
Metschnikowia viticola NCAIM Y.01705T (AY626892)
79
Metschnikowia reukaufii NRRL Y-7112T (U44825)
Metschnikowia koreensis KCTC 7828T (AF257272)98
52
Metschnikowia vanudenii PYCC 4650T (AF017404)
Metschnikowia lachancei PYCC 4605T (AY080995)
Metschnikowia gruessii NRRL Y-17809T (U45737)
Metschnikowia pulcherrima NRRL Y-7111T (U45736)
0.05
Candida gelsemii UWOPS 06-24.1T (DQ988046)
Candida gelsemii UWOPS 06-17.1 (DQ988047)
64
Candida gelsemii UWOPS 06-11.1 (DQ988045)
100
Candida rancoensis CBS 8174T (AJ508580)
Metschnikowia zobellii NRRL Y-5387T (U44823)
Metschnikowia bicuspidata NRRL YB-4993NT (U44822)
Metschnikowia australis NRRL Y-17414T (U76526)80
78
Metschnikowia krissii NRRL Y-5389T (U45735)
60
Metschnikowia viticola NCAIM Y.01705T (AY626892)
79
Metschnikowia reukaufii NRRL Y-7112T (U44825)
Metschnikowia koreensis KCTC 7828T (AF257272)98
52
Metschnikowia vanudenii PYCC 4650T (AF017404)
Metschnikowia lachancei PYCC 4605T (AY080995)
Metschnikowia gruessii NRRL Y-17809T (U45737)
Metschnikowia pulcherrima NRRL Y-7111T (U45736)
0.05
Figure A.1. Phylogram of Candida gelsemii and closest relatives based on a neighbour-
joining analysis (K2P transform) of D1/D2 LSU rDNA sequences. Bootstrap values
(1000 pseudoreplications) of 50% or greater are shown. Strain numbers and sequence
accession numbers are given. The superscripts identify type (T) and neotype strains
(NT).
155
Figure A.2. Candida gelsemii after 1 month on dilute V8 agar (1/20) at 18ºC. Strains
06-11.1 (a), 06-17.1 (b), and 06-24.1 (c), showing various degrees of differentiation into
“pulcherrima cells”. Asci were not formed on these and other sporulation media.
156
APPENDIX TWO
Complete list of the concentration of cardenolides from the nectar, leaves and flowers of
twelve species of Asclepias from the series Incarnatae, as described in chapter five. Total
average concentrations represent the sum of the total cardenolide concentration of each
sample within a species divided by the number of samples, with all concentrations on a
per microgram basis. The average concentrations of all individual compounds, which are
identified by their unique retention times, are also listed. Note that because of variation
in the concentration of individual cardenolides within samples, the sum of individual
cardenolide concentrations differs from the average total cardenolide concentration. I’ve
indicated several trends throughout appendix two: first, columns with borders represent
compounds never found in nectar; second, dark grey cells represent nectar cardenolides
that are not matched with a cardenolide in the leaf of that species (see chapter five
Discussion); third, the column highlighted in light grey represents the single compound
found only in nectar.
(appendix two continued on the next three pages)
157
Avg Concentration for Individual Cardenolides (ng/µg)
Species
Plant Part
Total Avg (ng/µg) n 12.5 13.1 13.5 13.7 14.1 14.3 14.5 14.7 14.9 15.3 15.7
Asclepias angustifolia nectar 0 5 0 0 0 0 0 0 0 0 0 0 0
Asclepias angustifolia leaf 0 2 0 0 0 0 0 0 0 0 0 0 0
Asclepias angustifolia flower 0.06 2 0 0 0 0 0 0 0 0 0 0 0
Asclepias barjoniifolia nectar 12.26 3 0 0 0 0 0 0 0 0 0 0 0
Asclepias barjoniifolia leaf 0.37 2 0 0 0 0 0 0 0 0 0 0 0
Asclepias boliviensis nectar 4.83 6 0 0 0 0 0 0 0 0 0 2.21 0
Asclepias boliviensis leaf 0.45 2 0 0 0 0 0 0 0 0 0 0 0
Asclepias candida nectar 11.27 3 0 0 0 0 0 0 0 0 0 0 0
Asclepias candida leaf 0.21 2 0 0 0 0 0 0 0 0.05 0 0 0
Asclepias curassavica nectar 0 8 0 0 0 0 0 0 0 0 0 0 0
Asclepias curassavica leaf 1.96 2 0.08 0 0 0 0 0 0 0.26 0 0.11 0
Asclepias curassavica flower 2.71 2 0 0.46 0 0.15 0 0 0 0.49 0 0 0
Asclepias fascicularis nectar 0 7 0 0 0 0 0 0 0 0 0 0 0
Asclepias fascicularis leaf 0 1 0 0 0 0 0 0 0 0 0 0 0
Asclepias incarnata pulchra nectar 0 2 0 0 0 0 0 0 0 0 0 0 0
Asclepias incarnata pulchra leaf 0.12 1 0 0 0 0 0 0 0 0 0 0 0
Asclepias mexicana nectar 3.55 2 0 0 0 0 0 0 0 0 0 0 0
Asclepias mexicana leaf 0.02 2 0 0 0 0 0 0 0 0 0 0 0
Asclepias nivea nectar 32.81 7 0 0 0 0 0 0 0 0.54 0 0 0.3
Asclepias nivea leaf 0.62 2 0 0 0 0 0 0 0 0 0 0 0.08
Asclepias nivea flower 0.79 2 0 0.07 0 0 0 0 0 0 0.1 0 0.07
Asclepias perennis nectar 38.10 4 0 0 0 3.13 2.11 0 0 0 10.6 1.73 0
Asclepias perennis leaf 2.23 2 0.14 0 0.03 0.68 0 0.06 0.44 0 0.23 0.03 0
Asclepias perennis flower 0.79 2 0 0 0 0 0 0 0.14 0 0.31 0 0
Asclepias pumila nectar 109.97 1 0 8.4 0 0 15.9 0 0 0 35.9 0 0
Asclepias pumila leaf 0.17 2 0 0 0 0 0 0 0 0 0.08 0 0
Asclepias pumila flower 0.20 2 0 0 0 0 0 0 0 0 0 0 0
Asclepias texana nectar 0 4 0 0 0 0 0 0 0 0 0 0 0
Asclepias texana leaf 0.06 3 0 0 0 0 0 0 0 0 0 0 0
Asclepias texana flower 0.04 2 0 0 0 0 0 0 0 0 0 0 0
158
Avg Concentration (ng/µg)
Species
Plant Part 15.9 16.1 16.3 16.5 16.7 16.9 17.2 17.5 17.8 18.4 18.6 19 19.3 19.5 19.8
Asclepias angustifolia nectar 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Asclepias angustifolia leaf 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Asclepias angustifolia flower 0 0 0 0 0.04 0 0 0 0 0 0 0 0 0.01 0
Asclepias barjoniifolia nectar 0 0 0 0 0 4.47 0 0 0 0 3.18 0 0 0 0
Asclepias barjoniifolia leaf 0 0 0 0.09 0 0 0 0 0.07 0 0.09 0 0 0 0
Asclepias boliviensis nectar 0 0 0 0 0 0.35 0 0.16 0 0 0.96 0 0 0 2.43
Asclepias boliviensis leaf 0.09 0 0 0 0 0 0 0.11 0 0.01 0.03 0.07 0 0 0.13
Asclepias candida nectar 2.78 0 0 0 0 0 4.15 0 2.55 0 0 1.95 0 0 0
Asclepias candida leaf 0 0 0 0 0.04 0 0.02 0 0.02 0 0 0.05 0 0.01 0.02
Asclepias curassavica nectar 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Asclepias curassavica leaf 0 0.14 0 0.74 0 0 0 0.05 0.11 0 0.09 0 0 0.05 0
Asclepias curassavica flower 0 0.3 0 0.29 0 0.27 0 0.19 0.03 0 0.14 0 0 0.19 0
Asclepias fascicularis nectar 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Asclepias fascicularis leaf 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Asclepias incarnata pulchra nectar 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Asclepias incarnata pulchra leaf 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Asclepias mexicana nectar 0 0 0 0 0 0 0 0 0 0 0 0 7.51 0 0
Asclepias mexicana leaf 0 0 0 0 0 0 0 0 0 0.05 0 0 0 0 0
Asclepias nivea nectar 0 4.48 0 0.42 0 1.81 0 1.39 0 0 13.4 0.52 0.41 0 0
Asclepias nivea leaf 0 0.07 0 0.1 0.03 0 0 0.06 0.07 0 0.06 0.03 0.01 0.02 0
Asclepias nivea flower 0 0 0 0.13 0 0 0 0.02 0 0.01 0.18 0.02 0 0.06 0
Asclepias perennis nectar 0 0 5.54 0 0 0 0 1.84 5.64 0 10.2 0 0 0 0
Asclepias perennis leaf 0 0 0.1 0.08 0 0 0 0.06 0.11 0 0.08 0 0 0.02 0
Asclepias perennis flower 0 0.06 0 0.03 0 0 0.02 0.09 0 0 0.08 0 0 0.03 0
Asclepias pumila nectar 0 0 0 0 0 0 0 12.9 6.28 0 32.8 0 0 0 0
Asclepias pumila leaf 0 0.01 0 0 0 0 0 0.03 0.02 0 0.03 0 0 0 0
Asclepias pumila flower 0 0 0 0 0 0 0 0.12 0 0 0.05 0 0 0.02 0
Asclepias texana nectar 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Asclepias texana leaf 0 0 0 0 0 0 0 0 0.11 0 0.06 0 0 0 0
Asclepias texana flower 0 0 0 0 0.06 0 0 0 0 0 0 0 0 0 0
159
Avg Concentration (ng/µg)
Species
Plant Part 20 20.3 20.7 20.9 21.3
Asclepias angustifolia nectar 0 0 0 0 0
Asclepias angustifolia leaf 0 0 0 0 0
Asclepias angustifolia flower 0 0 0 0.01 0
Asclepias barjoniifolia nectar 2.45 0 0 2.86 0
Asclepias barjoniifolia leaf 0.06 0 0 0.06 0
Asclepias boliviensis nectar 0 0 0 0 0
Asclepias boliviensis leaf 0 0 0 0.01 0
Asclepias candida nectar 0 0 0 0 0
Asclepias candida leaf 0 0 0 0 0
Asclepias curassavica nectar 0 0 0 0 0
Asclepias curassavica leaf 0.19 0 0 0.14 0
Asclepias curassavica flower 0.09 0.04 0 0.06 0
Asclepias fascicularis nectar 0 0 0 0 0
Asclepias fascicularis leaf 0 0 0 0 0
Asclepias incarnata pulchra nectar 0 0 0 0 0
Asclepias incarnata pulchra leaf 0 0 0 0 0.12
Asclepias mexicana nectar 0 0 0 0 0
Asclepias mexicana leaf 0 0 0 0 0
Asclepias nivea nectar 2.08 0 0 16.4 0
Asclepias nivea leaf 0.02 0.01 0 0.06 0
Asclepias nivea flower 0 20.3 0.01 0.04 0
Asclepias perennis nectar 0 0 0 9.76 0
Asclepias perennis leaf 0.02 0.01 0 0.12 0
Asclepias perennis flower 0 0.02 0 0.02 0
Asclepias pumila nectar 0 0 0 4.07 0
Asclepias pumila leaf 0 0.01 0 0 0
Asclepias pumila flower 0 0.01 0 0 0
Asclepias texana nectar 0 0 0 0 0
Asclepias texana leaf 0 0 0 0 0
Asclepias texana flower 0 0 0 0 0
160
APPENDIX THREE
Raw visit data from the three behavioural assays testing pollinator preference for the nectar cardenolide digitoxin. Data represents the
number of visits to flowers of each treatment type. The grey cells indicate which nectar treatment was associated with blue flowers
for each bee.
Assay 1 Assay 2 Assay 3
Bee 10 ng/µl digitoxin 30% sucrose Bee 50 ng/µl digitoxin 30% sucrose Bee 50 ng/µl digitoxin 10 ng/µl digitoxin
1 80 57 13 100 17 24 1 102
2 24 142 14 3 84 25 51 41
3 47 102 15 6 94 26 108 7
4 95 9 16 51 148 27 44 9
5 99 7 17 104 3 28 41 94
6 20 127 18 87 18 29 17 92
7 80 24 19 2 106 30 4 101
8 88 60 20 104 1 31 70 29
9 65 66 21 60 79 32 38 67
10 143 7 22 18 98 33 66 1
11 91 77 23 114 35
12 121 3