The effects of nectar-robbing on a plant- pollinator mutualism and the evolution of nectar-robbing and sociality in bees

Item Type text; Dissertation-Reproduction (electronic)

Authors Richardson, Sarah Claire

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 05/10/2021 07:26:04

Link to Item http://hdl.handle.net/10150/290257 INFORMATION TO USERS

This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.

In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand corner and continuing from left to right in equal sections with small overiaps.

Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.

ProQuest Information and Learning 300 North Zeeb Road, Ann Arbor. Ml 48106-1346 USA 800-521-0600

THE EFFECTS OF NECTAR-ROBBING ON A PLANT-POLLINATOR

MUTUALISM AND THE EVOLUTION OF NECTAR-ROBBING AND

SOCIALITY IN BEES

by

Sarah Claire Richardson

A Dissertation Submitted to the Faoxlty of the

DEPARTMENT OF ECOLOGY AND EVOLUTIONARY BIOLOGY

In Partial Fulfillment of the Requirements For the Degree of

DOCTOR OF PHILOSOPHY

In the Graduate College

THE UNIVERSITY OF ARIZONA

2001 UMI Number: 3016495

UMI ®

UMI Microform 3016495 Copyright 2001 by Bell & Howell Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.

Bell & Howell Information and Leaming Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 2

THE UNIVERSITY OF ARIZONA ® GRADUATE COLLEGE

As members of the Final Examination Committee, we certify that we have

read the dissertation prepared by Sarah niairp Richard'^nn

entitled The Effects of Nectar-Robbino on a PIant-Pnllinatnr

Mutualism and the Evolution of Nectar-Robbino and

Sociality in Bees

and recommend that it be accepted as fulfilling the dissertation

requirement for the Degree of Doctor of Philosophy

ri Tlan 3i, 2QQ1 BronsteiJ Date QO / UaTniel R. Papaj Date ^ ^CO ^ Nancy A. Nora^

Lucinaa Date

ucnmann Date

Final approval and acceptance of this dissertation is contingent upon the candidate's submission of the final copy of the dissertation to the Graduate College.

I hereby certify that I have read this dissertation prepared under my direction and recommend that it be accepted as fulfilling the dissertation requirement.

CUi/tlth. / ^/ny!/2tSAy

STATEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is deposited in the Uruversity Library to be made available to borrowers imder niles of the Library.

Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author. 4

ACKNOWLEDGMENTS

I am grateful to Dr. Judith Bronstein, Dr. Lucinda McDade, Dr. Daniel Papaj, Dr. Nancy Moran, Dr. Stephen Buchmarm, Dr. Francisco Omelas, and Dr. Richard Hudson for their time and their valuable suggestions for improving this dissertation. I would especially like to thank Dr. Judith Bronstein for helpful discussions of experimental design, stimulating discussions of cheating in mutualisms, and precise comments on drafts of all chapters in this dissertation. Dr. Stephen Buchmaim generously provided instruments, dye, greenhouse space, and advice on techniques for this research. Nicole Plotkin and Joyesha Chesnick were valuable field assistants; both had incredible stamina in the desert heat and useful suggestions for research. Sandy Adondakis, Eric Dyreson, and Maria Clauss gave suggestions on statistical analysis. Amy Faivre and Jill Miller gave advice on techniques for photography of pollen and pollen tubes. Sandy McGraw, Peter Murphy, David Banks, Nancy Compton, Kevin Toal, Katherina Spoms, Aaron Gove, and Mary Brown, who were volimteers from the Southwestern Research Station, helped with fieldwork investigating the effects of nectar-robbing bees.

Fieldwork was supported by two grants from the Theodore Roosevelt Memorial Fimd of the American Musevim of Natural History, by the 1994 Cranwell Smith Pollination Award from the University of Arizona, by the Southwestern Research Station Student Support Fund, and by grants from the Department of Ecology and Evolutionary Biology at the University of Arizona. A fellowship that gave me time to conduct phylogenetic research was generously provided by the Biological Diversification Research Trairiing Grant from the University of Arizona.

The Knight family of Paradise, AZ, the Ritchins families of Cotton City, N. M. and Animas, N. M., John Caron of Crown Dancer Estates, Myrtle BCraft of Portal, and Ann and John Dominic of Portal were kind in allowing research to be conducted on their land and in inviting me into their homes. 5

DEDICATION

I dedicate this dissertation to three scientists who showed me the joy of being a biologist. Dr. Myra Niemeier began our first day together by asking me what I wanted to leam; the list is still growing. Dr. Paul Johnsgard took me to see the mating dances of prairie chickens and whooping cranes, two of the most memorable sights in my life. Dr. John Janovy, Jr. gave me a net and let me discover the most beautiful parasites in the world for myself.

Finally, I dedicate this dissertation to my parents, Mr. and Mrs. Wallace and Martha Richardson, who have always encouraged me. 6

TABLE OF CONTENTS

ABSTRACT 7

CHAPTER 1: INTRODUCTION 9

Effects of Nectar-Robbing on Pollinator Behavior and Plant Reproduction 11

Comparison of Costs and Benefits from Mutualists and Antagonists 12

Evolution of Sociality, Nectar-robbing, and Diet Breadth in Bees 12

CHAPTER 2: PRESENT STUDY 15

APPENDIX A: TEMPORAL VARIATION IN POLLINATOR ABUNDANCE AND POLLINATOR FORAGING BEHAVIOR IN RESPONSE TO ROBBED FLOWERS 20

APPENDIX B: BENEFITS AND COSTS OF FLORAL VISITORS TO CHILOPSIS LINEARIS: POLLEN DEPOSTHON AND STIGMA CLOSURE 38

APPENDIX C: THE EFFECTS OF NECTAR-ROBBERS ON MUTUALISTS-COSTLY OR BENEHCLAL? 73

APPENDIX D: WERE DIET BREADTH AND NECTAR-ROBBEMG FACTORS IN THE EVOLUTION OF SOCIALITY IN BEES? 114

APPENDIX E: SUMMARY OF LEVEL OF SOCIALITY AND LEVEL OF NECTAR-ROBBING FOR GENERA OF SHORT-TONGUED AND LONG- TONGUED BEES 164 7

ABSTRACT

How will the intrusion of other species that remove rewards without providing reciprocal services affect the interaction between mutualists? How do costs and benefits from these "cheaters" compare to costs and benefits from potentially mutualistic visitors? Finally, did nectar-robbing, one kind of cheating, promote the evolution of complex levels of sociaKty by allowing bees access to a wider range of resources? I investigated these questions in the research described below. I found that pollinators visiting Chilopsis linearis (Bignoniaceae) spent less time visiting robbed flowers than visiting unrobbed flowers, and did not visit them as often as expected. Thus, robbing appeared to have a negative effect on pollinators visiting Chilopsis linearis. I compared costs and benefits of floral visitors to Chilopsis linearis (desert willow). Chilopsis had serrsitive stigmas that closed immediately upon touch and may have reopened later. I foimd that the probability of stigma reopening depended on the source and number of pollen grairis deposited. I compared visitors by number of pollen grains deposited, viability of pollen that they deposited, and their effect on stigmas. Nectar-robbers did not benefit plants by pollen deposition, but they also did not cost plants by causing stigmas to close without adequate pollen having been deposited. I investigated the effects of robbing on poUinator behavior and plant reproductive success. Nectar volumes were lower in robbed flowers than in unrobbed flowers. However, the most effective pollinators, bumblebees, did not avoid robbed flowers. In investigating male reproductive success, I foimd 8 that on some days, dye mimicking pollen traveled farther from robbed flowers, indicating that robbing may sometimes be beneficial to plants. In investigating female reproductive success, I found that there was no difference in pollen tube number between robbed and unrobbed flowers. Thus, a negative effect on one mutualist may not affect the other mutualist. I hypothesized that the evolution of robbing in bees was associated with a broad diet breadth and the evolution of complex sociality. Using phylogenetically independent contrasts for taxa within three geographical regions, I found that in some cases, a broad diet breadth was associated with sociality and robbing. 9

CHAPTER 1 INTRODUCTION

Recently, researchers have become interested in the balance between conflict and cooperation in mutualisms. Recent studies are showing that mutucdists may only benefit each other imder certain, conditions (Bronstein 1994) or may have antagonistic parts of their life history (Addicott 1986, Templeton 1985). Mutualists may also switch back and forth from antagonism to mutualism over ecological or evolutionary time (PeUmyr et al. 1996, Omelas 1994). Given that mutualists can both cooperate and exploit each other, how will the intrusion of other species affect the interaction? For example, how will a species that removes a reward from one of the mutualistic partners without providing a reciprocal service affect both partners? In addition, can this "cheater" on a mutualism provide any direct benefits to either mutualist? Theories of mutualism predict that fitnesses, measured by growth rates or density, of mutualistic partners should be linked, with the effect that an increase in the fitness of one mutualist should cause a decrease in the fitness of the other mutualist (Dean 1983, Pierce and Young 1986). A cheater theoretically should decrease the fitness of the mutualist that it competes with, such that it negatively affects both mutualists. In pollination biology, it is very difficult to measxire fitnesses of pollinators. However, one can determine their effectivenesses eind behaviors in response to cheating and then determine the effect of changes in their behaviors on plants. 1 0

One assumption that needs to be tested is that visitors that enter flowers, or "legitimate" visitors, are more effective pollinators than are nectar- robbers, the cheaters. Also, since there are multiple visitors to plants and since they may differ in effectiveness, their changing abimdances may change the effects of robbing for plants. Finally, does cheating provide benefits to the cheater? Two more assumptions that should be tested are that cheaters benefit from their behaviors, and that they either originally evolved from mutualists, or have the option in the present of behaving as muttialists but choose to cheat because of the relative benefits. In the Chilopsis linearis pollination mutualism, I have observed that at the beginning of each season, Xylocopa califomica, forage for the first few days as legitimate visitors and then switch to robbing the same flowers. In addition, these nectar-robbers spend less time visiting flowers than do legitimate visitors, so presumably they do benefit more from robbing than they do from entering flowers legitimately. Nectar-robbing may also benefit bees by allowing them access to flowers that have narrow tubes, giving the bees more resources to support their nests over long seasons. In turn, a broad diet breadth may be a prerequisite for the evolution of complex forms of sociality. A broad diet breadth may favor the lengthening of foundress lifespan and the seasonal activity of workers, which appear to be necessary components of generational overlap. Generational overlap, by definition, is a component of complex sociality. In this way, nectar-robbing could have played a significant part in the life history of different lineages of bees. 11

Effects of Nectar-Robbing on Pollinator Behavior and Plant Reproduction Theories of the evolution of mutualism predict that this "cheater" on a mutualism, also known as an "aprovechado" (Soberon and Martinez del Rio 1985), will cause a mutualism to become extinct (BuU and Rice 1991, Trivers 1971). This extinction will occur because the cheater removes rewards without inciirring the cost of providing a service or reciprocal reward to the mutualistic partner. By not reciprocating, a cheater has a competitive advantage over a mutualist Because the fitnesses of mutualistic partners are theoretically correlated (Keeler 1985, Addicott 1981), the expectation is that a cost to one partner will negatively affect the other partner. For this reason, mutualisms have often been predicted to be evolutionarily unstable (Poulin 1995, Boucher 1982). These theories assume that cheaters have a negative effect on one mutualist by competing with the other mutualist for rewards and by reducing the services provided. Alternatively, some mearis of pimishing the cheater (Mesterton-Gibbons and Dugatkin 1992) or controlling the cheater without excluding mutualists (Schwartz and Hoeksema 1998, Pellmyr and Huth 1994, Moe and Hammerstein 1994) may evolve. However, a different outcome has been predicted for one kind of cheater, nectar-robbing bees. As cheaters on the plant-pollinator mutualism, primary nectar-robbers take nectar from a flower by cutting a slit in the corolla and removing nectar, usually without pollinating. It has been suggested that nectar-robbers may have a positive effect on a plant if they reduce average levels of nectar (Zimmeiman and Cook 1985, Heinrich and Raven 1972) or increase variability between patches, because pollinators in search of nectar must carry pollen farther between plants. A pollinator may travel farther either because it must visit more flowers to collect the same amount of nectar, or because it searches for unrobbed flowers after it has encountered several robbed ones. Longer flight distances by pollinators are likely to carry poUen away from close relatives of the plants, thus decreasing inbreeding (Inouye 1983, Gentry 1978). In this way, nectar-robbing may indirectly benefit a plant. This hypothesis leads to another: the benefit of robbing should also increase, because variability makes it Ukely that pollinators will search for less robbed patches. The necessity to search for nectar depends on amoimts of nectar left in robbed flowers and the energetic costs of pollinators. However, if the intensity of robbing is even over the site, pollinators should not search for less robbed patches. In this scenario, cheaters affect one mutualist (pollinators) negatively by competing with them for limited rewards (nectar), but benefit the other mutualist (the plant) indirectly by changing pollinators' behavior.

Comparison of Costs and Benefits from Mutualists and Antagonists In pollination mutualisms, nectar-robbers are usually considered antagorusts; visitors that enter flowers (legitimate visitors) are usually considered mutualists. However, nectar-robbers may provide some benefits to plants, whereas legitimate visitors may inflict some costs. I compared costs and benefits of floral visitors to Chilopsis linearis (desert willow).

Evolution of Sociality. Nectar-robbing, and Diet Breadth in Bees The evolutionary factors that led to existence of social irisects with sterile castes mystified Darwin (1859). Eusociality, traditionally defined to 1 3 include cooperative brood care, presence of generation overlap, and reproductive division of labor (Crespi 1996; Michener 1974,1969) may have several factors necessary for its evolution. Hamilton (1964) proposed that kin selection explained workers' apparent altruism in caring for sisters. Through inclusive fitness, haplodiploidy was understood to be a critical factor in the evolution of sociality in Hymenoptera (ants, bees, and wasps). However, though haplodiploidy may have been a prerequisite for the evolution of sociality in Hymenoptera, it is not sufficient to explain the origin of sociality. Though all Hymenoptera are haplodiploid, only a few families out of more than 80 are eusocial; sociality apparentiy had multiple independent origins in Hjmienoptera and required more preadaptatiorts than just haplodiploidy (Hunt 1999). The evolutionary pathway from solitary to complex sociality, in which eusodality is the most complex level (Chapter 4, Table 1), has attracted intense interest because it may aid in explaining factors that led to the evolution of sociality (reviewed in Michener 2000). By definition, generational overlap is a key condition for existence of complex levels of socicdity. It occurs in all subsocial and eusocial Hymenoptera and in some parasocial bees (Danforth and Eickwort 1997). Generational overlap may occur through either multivoltinism (multiple broods per year) or long lifespan of females that initiate nests (Schwartz et al. 1997), or both. For both multivoltinism or longevity of females, activity of bees over long seasons must have been a prerequisite. Thus, pre-social bees must have had a broad diet breadth (large number of plants visited for resources) because the length of blooming season of plant taxa would not have been long enough to support social bees. 14

I investigated three hypotheses; 1) that a broad diet breadth was associated with the evolution of sociality in bees, 2) that a novel foraging behavior, nectar-robbing, was associated with the evolution of a broader diet breadth, and 3) that nectar-robbing also was associated with the evolution of sociality. Nectar-robbing bees make sHts or holes in corollas of flowers and remove nectar without entering corollas. In a comparison of the nectar-robbing bumblebees with non-nectar-robbing bumblebees at a site in Colorado, nectar- robbing bumblebees foraged on more species of plants than did non-robbers (Inouye 1983). The nectar-robbing bees visited plants that were inaccessible to other species of bees, such as hummingbird-visited Ipomopsis flowers with tubular morphologies, in addition to normally bee-pollinated flowers witl. open coroUas. For this reason, I predicted that the evolution of nectar-robbing in bees was associated with a broad diet breadth. If nectar-robbing were correlated with sociality, it may have facilitated more generalized plant use by social bees. 1 5

CHAPTER 2 PRESENT STUDY

The methods, results, and conclusions of this study are presented in the papers appended to this dissertation. The following is a summary of the most important findings in these papers. In Appendix A, I describe a study in which I found that pollinators visiting Chilopsis linearis (Bignoniaceae) spent less time visiting robbed flowers than visiting unrobbed flowers, and did not visit them as often as expected. Nectar was produced in buds in the late afternoon before they opened, so robbing bees gained access to the resource earlier in the life of the flower than did pollinating bees. Because pollinating bees searched for ururobbed flowers and because they found less nectar in robbed flowers than unrobbed flowers, robbing appeared to have a negative effect on pollinators visiting Chilopsis linearis. Appendix B presents research in which I investigated further whether pollinator behavior is affected by the presence of nectar-robbers and determined whether nectar-robbing had a costly, neutral, or positive indirect effect on the other partners in the mutualism, the plants. Nectar-robbers affected levels of reward available to mutualists; unrobbed and robbed flowers had the same probability of containing nectar, but nectar volumes were lower in robbed flowers. However, robbing may not have had an indirect negative effect on plant fitness for three reasons. First, though honeybees (Apis mellifera) spent less time in robbed flowers than in unrobbed ones and also avoided robbed flowers, honeybees are not effective 1 6 pollinators, so the effect of robbing on their behavior may mean little to the plant. Second, even though bumblebees {Bombus sonorus) also spent less time in robbed flowers, the length of time that they spent in robbed flowers was not correlated with pollen deposition, so the effect of robbing on their behavior may also mean little to the plant. Finally, imlike honeybees, bumblebees did not avoid robbed flowers and were highly effective poUinators. I investigated the indirect effects of robbing on the plant's male reproductive success by examining whether dye mimicking pollen traveled farther from robbed flowers than from imrobbed flowers. On some days, dye traveled farther from robbed flowers, indicating that robbing may sometimes be beneficial to plants. I also determined whether female reproductive success, measured by the number of pollen tubes growing in styles, was affected by robbing. I foimd that there was no difference in pollen tube number between robbed and unrobbed flowers, which means that robbers do not cost plants directly by damaging flowers or indirectly by reducing their female reproductive success. Overall, this study suggests that a negative effect on one mutualist, competition for resources by an aprovechado, may not affect the other mutualist. The presence of another species may even lead to a benefit in some cases. This suggests that fitnesses of mutualists are not linked as closely as has been assimied. Appendix C presents research in which I compared costs and benefits of floral visitors to Chilopsis linearis (desert willow). This plant was self- incompatible and poUen-Umited, so the source of pollen (self vs. outcross) and number of pollen grairxs deposited affected reproductive success. Chilopsis had sensitive stigmas that closed immediately upon touch and may have reopened 17 later. Stigma closure may have been costly to plants if too few pollen grains were deposited to set fruit. I found that the probability of stigma reopening depended on the source and number of pollen grains deposited. I compared visitors by number of pollen grains deposited, viability of pollen that they deposited, and their effect on stigmas. Nectar-robbers did not deliver benefits to plants by depositing pollen grairis, but they also did not irifUct costs on plants by causing stigmas to close without adequate pollen having been deposited. Some legitimate visitors imposed costs on plants by reducing pollen viability; nectar-robbers did not. Overall, legitimate visitors usually conferred greater benefits on plants than nectar-robbers did. However, a few legitimate visitors inflicted greater costs than nectar-robbers did, by causing stigmas to close without adequate pollen deposited. In this way, legitimate visitors could inflict some costs on plants that nectar-robbers could not. In Appendix D, I present work in which I hypothesized that the evolution of nectar-robbing in bees was associated with a broad diet breadth. If nectar-robbing were correlated with sociality, it may have facilitated more generalized plant use by social bees. To test these hypotheses, I combined four published phylogenies of bees (Roig-Alsina and Michener 1993, Alexander and Michener 1995, Danforth and Eickwort 1997, Peseriko 1999). The combined phylogerues encompass a monophyletic group composed of 146 genera, a broad sample of the approx. 600 genera of bees in the world (Michener 1979). With certain additions and deletions described below, I included 89 genera in my analysis. I surveyed these genera for degree of sociality (maximal social development) and presence of nectar-robbing (Appendix 5). 1 8

I used the combined phylogeny to determine the minimum number of origins of sociality in bees, the presence of reversals to solitary behavior, and whether the pathway to eusociality led from parasociality, subsociality, or just solitary behavior. I mapped maximal social development and presence of nectar-robbing onto the combined phylogeny in order to determine their degree of correlation using MacClade's concentrated changes test (Maddison and Maddison 1992). To determine the order of each character, I counted the number of times that each preceded the other. To test the hj^othesis that a broad diet breadth was associated with the evolution of sociality, I compared diet breadth of bees within three geographical regions: the Eastern United States, Guanacaste Province of Costa Rica, and California, (Mitchell 1960 and 1962, Heithaus 1979, and Moldenke and Neff 1974, respectively). For 345 species in the Eastern U.S., 135 species in Costa Rica, and 1163 species in California, I determined the median and maximum number of plants visited by all species in each genus, i.e., the diet breadth of the genus. I used ANOVAs to test the hypotheses that median and maximum diet breadth were a function of maximal social development within each genus and of geographical location. To take into account correlatioris between genera because of shared history, for each region I reconstructed the phylogeny of bees using only the genera occurring in each region. I used partial correlations of phylogenetically independent contrasts to determine whether correlations of diet breadth with maximal social development and with presence of nectar-robbing were significant. Over a phylogeny of bees made from four published phylogenetic analyses, I found that sociality has evolved at least 11 times and up to 14 times. 19 in the families Apidae and Halictidae. Eusociality, the most complex form of sociality, has originated at least 4 to 5 times in bees. Computer simulations showed that sociality and nectar-robbing were correlated on the phylogeny of bees. Comparing diet breadth of bee genera within three geographic regions, I foimd that diet breadth varied as a function of maximal social development among bee genera. I used phylogeneticaUy independent contrasts for taxa within each region to test for correlations between sociality, diet breadth, and nectar-robbing. In some cases, I found that a broad diet breadth was associated with sociality and nectar-robbing. 20

APPENDIX A

Title: Temporal Variation in Pollinator Abxindance and PoUinator Foraging Behavior in Resporise to Robbed Flowers ^

Abstract.—Xy/ocopiz californica robs some species of flowers by slitting open the sides of flowers to steal nectar, presumably without pollinating the flowers. PoUinators visiting Chilopsis linearis (Bignoniaceae) spend less time on robbed flowers than on unrobbed flowers, and do not visit them as often as expected. Of 228 flowers marked over the season in 1993, only 13, or 6%, produced fruits. Of those, 10 were robbed before maturing into fruits. Nectar was produced in buds in the late afternoon before they opened, so robbing bees gained access to the resource earlier in the life of the flower than did pollinating bees. Because pollinating bees must search for urvrobbed flowers and because they find less nectar in robbed flowers than unrobbed flowers, robbing appears to have a negative effect on pollinators visiting Chilopsis linearis. However, robbing may have no effect on the plants.

1 Published in: DeBano, L. F., FfoUiott, P. F., Ortega-Rubio, A., Gottfried, G. J., Hamre, R. H., and C. B. Edmirister. 1995. Biodiversity and Management of the Madrean Archipelago: the Sky Islands of Southwestern United States and Northwestern Mexico, Tucson, Arizona, U.S. Department of Agriculture, Forest Service, Fort Collins, CO. Reprinted with permission. 20

INTRODUCTION

In the plant-pollinator mutualism, pollinators, such as Apis mellifera, visit flowers in search of a reward provided by the plant, typically nectar or pollen. At the same time, the pollinators benefit the plant by carrying pollen between flowers. Other visitors to flowers are cheaters on the plant-pollinator mutualism. Certain cheaters, nectar-robbing bees, make holes or slits on flowers and rob the nectar, usually without touching the anthers or stigma (Fig. 1). Thus, nectar-robbing bees take the reward that the plant offers to mutualists, without providing the reciprocal service. When these bees visit flowers, they leave behind a record of their visit in the form of a slit or hole in a flower. Because this damage remains after the nectar-robbing bees have left the flowers, it is possible that the nectar-robbers influence the flower-visiting decisions made by legitimate pollinators of the plant even after the nectar- robbers have removed nectar and left the flower. Several researchers have investigated the effect of nectar-robbing bees on one or both partners in the mutualism—either plant or pollinators. When flowers are robbed, seed set or pod number of plants can increase (Hawkins 1961, Roubik et al. 1985), decrease (McDade and Kinsman 1980, Roubik 1982), or remain imchanged (Rust 1979, Newton and Hill 1983). A few researchers have shown that floral visitors are attracted to robbed flowers because they gain access to nectar through holes left by robbers (Hawkins 1961, Inouye 1983). Floral visitors that gain access to normally urureachable flowers may 2 1

increase seed set, either from their own visits (Hawkins 1961) or by causing other pollinators to visit more flowers when nectar has been depleted (Heinrich and Raven 1972). In contrast, other studies have shown that pollinators leave plants that have a high proportion of robbed flowers, presumably because the nectar rewards have been depleted (McDade and BCinsman 1980, Roubik 1982, Zimmerman and Cook 1985). The effect of nectar-robbing bees on plants should vary over time, depending on floral visitors' abundances and the behavior of pollinators in response to robbed flowers. In this study, I measured abimdances of pollinators visiting Chilopsis linearis (Bignoniaceae) during two flowering seasoris. I also investigated the foraging behavior of the two most common pollinators. Apis mellifera (Apidae), the honeybee, and Bombus sonorus (Apidae), the bvimblebee, on plants with robbed and xmrobbed flowers. The cheater on the mutualism between Chilopsis linearis and its pollinators is Xylocopa califomica (Apidae), a nectar-robbing carpenter bee. Future studies will address the effect of robbing and pollinator behavior on plants. I expected that Bombus sonorus would avoid less-rewarding robbed flowers and Apis mellifera would not. Individuals oiApis may not have evolved the ability to identify and avoid robbed flowers of Chilopsis, since Apis was recently (by evolutionary time) introduced into the New World tropics by setflers from Europe (Michener 1979). The historical distribution of A. mellifera does not overlap with the distribution of Chilopsis linearis (throughout the Southwestern U.S. and northern Mexico, Kearney and Peebles 1960), although the distribution of B. sonorus does. 22

METHODS

Chilopsis linearis blooms profusely for a few weeks in the "cornucopia" phenology described by Gentry (1974). It produces sympetalous, tnxmpet- shaped flowers that are visited primarily by bees, but also by himimingbirds and butterflies at the study site. Nectar is produced as flowers open in the late afternoon and again in the early morning (Brown et al. 1981 and Richardson, unpubl. data). The 2-5 m tall trees grow along normally dry washes in the Southwestern U.S. and northern Mexico. I observed floral visitors to Chilopsis linearis during the flowering season from May 20-Jime 7,1993 and May 12-May 31,1994. About 28 trees were located in a dry wash near the Chiricahua Mountains, 5.6 mi. N. of Portal on the San Simon Road, Cochise Co., Arizona. Common plants in the surrounding desert scrub are mesquite {Prosopis sp.). Acacia Greggii, and creosote-bush (Larrea tridentata). Patches of flowers on a single tree that varied in size from 16 to 207 flowers were marked before each observation period. Ail flowers within viewing distance were marked with colored thread to distinguish robbed flowers from unrobbed flowers. Visitors to the plants were observed in 19 half-hour observation periods at nine trees in 1993. The identity of visitors and number of marked robbed or unrobbed flowers visited by each individual within the patch were recorded in 1993. In 1994, numbers of visitors at seven trees were counted during four half-hour observation periods, 3.5 hours of general observations, and four hours of observations during other experiments. Observation periods were conducted at different times of the 23 day over both years, but were concentrated from. 6:00 a.m. to 11:00 a.m., when rates of visitation to plants were highest. The frequency of visitation to imrobbed and robbed flowers by individual visitors was determined by coimting the number of marked unrobbed and robbed flowers visited during a foraging bout. The niimbers of unrobbed and robbed flowers visited were compared with the nimibers of unrobbed and robbed flowers available within the marked patch. Using a sign test, the number of tmrobbed flowers visited by an individual pollinator was compared with the expected number that would have been visited if the pollinator were choosing flowers within the patch randomly. I assumed that if pollinators were neither avoiding nor choosing robbed flowers, the proportion of visits to lonrobbed flowers during a foraging bout would be the same as the proportion available in a patch. Differences between Apis and Bombus in number of individuals that chose more unrobbed flowers than robbed flowers were analyzed using a Mann-Whitney U test.

RESULTS

Relative abimdances of visitors to Chilopsis linearis are shown in Table 1. In both years, the total number of visitors observed was similar and the number of observation hours was equal, so the overall abimdance of visitors was similar in 1993 and 1994. The greatest change in relative abimdance was a decrease in Apis mellifera from 51.2% of the visitors to flowers in 1993 to 9.1% in 1994. Visits by Xylocopa califomica increased from 12.5% in 1993 to 41.8% in 24

1994. The relative abundance of Bombus sonorus at flowers increased from 8.3% in 1993 to 30.0% in 1994. Coefficients of variation shown in Table 1 indicate the variation in abundance of each species over 1993 and 1994. The three most frequent visitors. Apis, Bombus, and Xylocopa, were more variable between years than the two least frequent groups of visitors, which were hummingbirds and several bee species in the family Anthophoridae. However, bees in the family Megachilidae were the most variable between years, since none was observed at the study site in 1994. The percentage of unrobbed flowers available duxing observation periods ranged firom 3.9% to 92.2% in 1993. Apis visited significantly more unrobbed flowers than expected during each foraging bout (sign test: C = 10, P < 0.002, N = 33). In contrast, the number of visits to unrobbed flowers by Bombus was not significantly different than the number predicted by the proportion of unrobbed flowers available (sign test: C = 0, P > 0.50, N = 7). Apis visited more imrobbed flowers than Bombus did (Mann-Whitney U test: U = 192.5, P < 0.005, Ni = 33, N2 = 7): in observations during 1993,87.9% of Apis individuals visited more unrobbed than robbed flowers within a foraging bout, as compared with 42.7% of Bombus (Fig. 2).

DISCUSSION

The numbers of individual Bombus sonorus visiting C. linearis increased from 1993 to 1994 at the same time as visits by Apis mellifera decreased. Visits by Xylocopa califomica, nectar-robbing bees, to plants also declined from 1993 25 to 1994. In general, more common visitors varied more in abundances between years than did the least common visitors, with the exception of bees in the family Megachilidae. In addition to these changes in relative abimdance, I foimd that individuals of A. mellifera avoid robbed flowers, while individuals of B. sonoriis do not. These observations of behavior are opposite to my original prediction that individuals of Apis would not avoid robbed Chilopsis flowers because the species have a short evolutionary history together. Though A. mellifera has not come into contact with C. linearis until recently, its behavior in response to robbed flowers of C. linearis may be similar to its behavior in response to robbed flowers of Old World species. First, I will discuss the implications of variation in relative abimdance of the floral visitors over time. Second, I will consider the significance of the difference between Apis and Bombus in foraging behavior in response to robbed flowers. Bombus sonoriis increased from 1993 to 1994 at Chilopsis, simultaneously with a decrease in Apis mellifera. Several workers have hypothesized that the honeybee. Apis mellifera, has a negative impact on the abundance of native pollinators by outcompeting them for food (Schaffer et al. 1979, Wolda and Roubik 1986, Roubik 1983). However, in one study, Roubik (1983) foimd no competition for food measured by brood production between Apis mellifera and native Melipona after introduction of Apis into French Guiana in 1982. Since competition was measured early after the introduction, it may have been too soon to see an effect of the introduction. My data are consistent with the hypothesis that Apis competes with native polUnators. However, the reciprocal changes in relative abundance of Apis and Bombus could be caused 26 by another factor, such as a difference in precipitation between 1993 and 1994 that affected Apis cind native visitors differently. Interestingly, high variation in numbers of common visitors to Chilopsis linearis conflicts with results reported from other bees in the tropics. Roubik and Ackerman (1987) found that relatively scarce euglossine bees collected from traps at three sites in Panama had more variable populations than more abimdant bees did. In the present study, however, the most abundant bees were most variable between years. Differences may be caused by differences in stability of species' population sizes between the tropics and temperate zone or by differences in sampling methods. Perhaps the most abundant species are the fastest to switch to a more productive nectar producer as the season progresses. However, Chilopsis appeared to be the only abundant soxirce of nectar in the area during the present study imtil Yucca elata bloomed in mid-June. Correlations between species abxmdance and speed of resource switching would not be observed by trapping bees (as in the Panamanian study) because resource switching by bees would not affect the likelihood of collection.

Xylocopa califomica, the nectar-robbing bee, also changed in relative abundance over the two years of this study. Although more Xylocopa califomica individuals were observed at plants during observation periods in 1994 than in 1993, many of their foraging bouts were very brief. In 1994, Xylocopa often robbed just one or two flowers at the top of a tree while flying across the wash, instead of carrying out the sustained foraging bouts observed in 1993. A possible explanation for this change in behavior is that nest provisioning in 1993 occurred much earlier than in. 1994. In 1994, 27 provisioning of nearby nests coincided with the Chilopsis flowering season (unpubl. data). To provision nests, a female Xylocopa collects nectar and pollen, and shapes it into a ball. She places the ball of provisions in a tunnel inside a hollowed-out yucca or agave stalk. Before sealing the ball of provisions in the tunnel, she lays an egg on it. In 1994, female X. californica may have been concentrating foraging efforts on provisioning nests with pollen from nearby pollen-producing acacia and mesquite, instead of collecting nectar from Chilopsis. Even though more X. californica were observed in 1994 than in 1993, fewer flowers on the trees were robbed (unpubl. data). In addition to changing reciprocally in abimdance. Apis avoided robbed flowers during foraging bouts within a plant more than Bombus sononis did. A previous study with conflicting results has been conducted on foraging behavior by Bombus vagans in response to robbed flowers. B. vagans departs from patches of Impatiens capensis with artificially robbed flowers using a wider angle of flight than from imrobbed patches (Zimmerman and Cook 1985). Measurement of angle of flight after leaving a patch is an indicator of floral visitors' avoidance of similar patches; a wide angle suggests that a bee is leaving the area in search of patches that are different from the one just visited. Three possible explanations for the discrepancy between studies exist. 1) Boitibus sononis may simply behave differently than Bombus vagans. 2) Avoidance of a patch of robbed flowers involves a different level of decision­ making for bees than avoidance of robbed flowers within a patch. 3) B. sononis is restricted to robbed flowers at Chilopsis because of competition with A. mellifera, whereas pollinators visiting Impatiens did not partition resources because A. mellifera was not a frequent visitor. 28

Resource partitioning (the third possibility) may occur if Bombus sonoriis visits robbed flowers because Apis cannot obtain the nectar remaining after a visit by nectar-robbing bees, leaving the nectar available to B. sononis. Brown et al. (1981) observed that peak foraging by B. sononis on Chilopsis linearis occurred later in the day than the peak of foraging by Apis mellifera. Whitham (1977) showed that some of the nectar in Chilopsis is contained in deep grooves in the corolla. As a result. Brown et al. (1981) hypothesized that Apis cannot reach the nectar in the grooves, whereas Bombus can. Brown et al. (1981) suggested that this difference between Apis and B. sonorus might explain the fact that foraging by Bombus peaks later in the day than does that of Apis. For the same reason. Apis may avoid robbed flowers because they are not able to reach nectar in grooves of flowers. Several alternative explanations remain for the difference between Apis and Bombus in the degree to which they avoid robbed flowers. These explanations are not mutually exclusive; one or more may be valid. First, Apis mellifera have greater metabolic reqmrements than Bombus, such that for Bombus, but not Apis, a visit to a robbed flower may supply enough nectar to be worth the visit. At rest at 25^C, a Bombus vosnesenskii queen consumes less than 10 ml/g/hr oxygen (Heiiirich 1979). In contrast, individuals of Apis mellifera use more oxygen, about 15 ml/g/hr oxygen at 250C (Moritz and Southwick 1992). Individuals of Apis expend more metabolic energy than those of B. vosnesenskii up to about SO^C (Heinrich 1979, Moritz and Southwick 1992). Apis' metabolic costs during early morning foraging should be greater than that of Bombus sonorus (assximing that Bombus sonorus has a metabolism similar to B. vosnesenskii), since the peak of foraging by 29 individuals of Apis mellifera at Chilopsis linearis occurs in the morning, before the temperature reaches 30° C (Brown et al. 1981, and impubl. data). Apis may also avoid robbed flowers more readily than Bombus because they can sense marking scents that Xi/locopa califomica, the nectar-robbing bees, leave behind. X. califomica avoids revisiting some robbed flowers in the greerihouse and in the field (unpubl. data). A related bee, Xylocopa virginica, avoids revisiting Passiflora incamata flowers that it pollinates, possibly because females use scent marks to avoid revisiting flowers (Frankie and Vinson 1977). Only marking behavior on legitimate pollinating visits by X. virginica has been researched, but Xylocopa califomica may use scent-marking to avoid revisiting robbed flowers as well. Other research indicates that two species of pollinators in the present study. Apis mellifera and Bombus sonorus, avoid flowers marked by other individuals within their own species. A. mellifera has been shown to avoid flowers scent-marked by other individuals in the same species (Giurfa and Nunez 1992). It has also been suggested that B. sonorus avoids scent- marked flowers (Schmitt and Bertsch 1990). In all three cases, the identity of the chemical marking is unknown and the effects of marking on individuals of other species is also imknown. However, as suggested above, it is possible that Apis mellifera can sense the mark left behind from a robbing visit by X. califomica and chooses not to visit, while B. sonorus cannot. Finally, Apis may avoid robbed flowers while Bombus do not because Bombus have lower or slower learning abilities. Several studies have investigated aspects of the learning abilities of Apis and Bombus separately (e. g.. Dyer 1991, CoUett et al. 1993, Greggers and Menzel 1993, Gould 1993, Dukas and Real 1993), but none has yet compared them directly on the same 30 tasks. Duffield et al. (1993) have found that workers of Apis discriminate between flowers based on morphology (inflorescence length and number of bracts). One could test whether Apis leam to prefer flowers based on morphology or markings more quickly than Bombus do. Whatever the reason for the difference in Apis and Bombus behavior in response to robbed flowers, that difference may cause the effect of robbing on Chilopsis linearis to vary as relative abtmdances of visitors change. The effect of robbing also depends on the pollination efficiency of the pollinators, the prevalence of robbing, the level of self-compatibility of C. linearis and the amount of damage done to a flower during a robbing visit, all of which are being currently investigated. For example, if Apis is the least efficient pollinator, taking the reward provided by the plant without providing a reciprocal pollination benefit, the effect of robbing may be positive in years in which Apis is common. Willmer et al. (1994) have found that several species of Bombus transfer more pollen between Rubus idaeus flowers than Apis does, but the results may not apply directly to this study because Chilopsis flowers are shaped differently than Rubus flowers. If Bombus is more effective than Apis in pollinating Chilopsis, the effect of robbing may not be significant when Bombus predominates, because Bombus do not avoid robbed flowers. When pollinators vary in behavior and abundance, the overall effect of cheaters on the evolution of mutualisms may vary over time.

ACKNOWLEDGMENTS 3 1

Thanks to Stephen Buchmann and Judith Bronstein for assistance and Judith Bronstein, Luanda McDade, and J. Francisco Omelas for comments on the manuscript. This study was supported by two grants from the Theodore Roosevelt Memorial Fund of the American Museum of Natural History, by the 1994 Cranwell Smith Pollination Award from the University of Arizona, and by grants from the Department of Ecology and Evolutionary Biology at the University of Arizona. Sandy McGraw, Peter Murphy, and David Banks, volunteers from the Southwestern Research Station, helped with this part of the study of nectar-robbing bees. George Ferguson contributed the photograph of the nectar-robbing bee. The Knight family of Paradise, AZ, and the Ritchins families of Cotton City, N. M. and Animas, N. M. were kind in allowing the research to be conducted on their land. 32

LITERATURE CITED

Brown, J. H., A. Kodric-Brown, T. G. Whitham, and H. W. Bond. 1981. Competition between hummingbirds and insects for the nectar of two species of shrubs. Southwest. Nat. 26:133-145.

CoUett, T. S., S. N. Fry, and R. Wehner. 1993. Sequence learning by honeybees. J. Comp. Physiol. A. 172: 693-706.

Duffield, G. E., R. C. Gibson, P. M. Gilhooly, A. J. Hesse, C. R. Inkley, F. S. Gilbert, and C. J. Barnard. 1993. Choice of flowers by foraging honey bees (Apis mellifera): possible morphological cues. Ecol. Entom. 18:191-197.

Dukas, R. and L. A. Real. 1993. Learning constraints and floral choice behaviour in bumble bees. Anim. Behav. 46: 637-644.

Dyer, F. C. 1991. Bees acquire route-based memories but not cognitive maps in a familiar landscape. Anim. Behav. 41: 239-246.

Frankie, G. W. and S. B. Vinson. 1997. Scent marking of passion flowers in Texas by females of Xylocopa virginica texana (Hymenoptera: Anthophoridae). J. Kans. Ent. Soc. 50: 613-625. 3 3

Gentry, A. H. 1974. Coevolutionary patterns in Central American Bignoniaceae. Ann. Mo. Bot. Gard. 61: 728-759.

Giurfa, M. and J. A. Nunez. 1992. Honeybees mark with scent and reject recently visited flowers. Oecologia 89:113-117.

Gould, J. L. 1993. Ethological and comparative perspectives on honey bee learning. Pages 18-50 in D. Papaj and A. Lewis, editors. Insect learning. Chapman and Hall, New York, N. Y.

Greggers, U. and R. Menzel. 1993. Memory dynamics and foraging strategies of honeybees. Behav. Ecol. Sociobiol. 32: 17-29.

Hawkins, R. P. 1961. Observations of the pollination of red clover by bees. I. The yield of seed in relation to the nixtnbers and kinds of pollinators. Arm. Appl. Biol. 49: 55-65.

Heinrich, B. 1979. Biomblebee economics. Harvard University Press, Cambridge, Mass.

Heinrich, B. and P. H. Raven. 1979. Energetics and pollination ecology. Science 187: 597-602. 34

Inouye, D. W. 1983. The ecology of nectar robbing. Pages 153-173. in B. Bentley and T. Elias, editors. The biology of nectaries. Columbia University Press, New York, N. Y.

Kearney, T. H. and R. H. Peebles. 1960. Arizona flora. University of California Press, Berkeley, CA.

McDade, L. A. and S. Kinsman. 1980. The impact of floral parasitism in two neotropical hnmmingbrrd-poUinated plant species. Evolution 34: 944-958.

Michener, C. D. 1979. Biogeography of the bees. Arm. Mo. Bot. Card. 66: 277- 347.

Moritz, R. F. A. and E. E. Southwick. 1992. Bees as superorganisms. Springer- Verlag. Berlin, Germany.

Newton, S. D. and G. D. Hill. 1983. Robbing of field bean flowers by the short- tongued bumble bee Bombus terrestris L. J. Apic. Res. 22:124-129.

Roubik, D. W. 1982. The ecological impact of nectar-robbing bees and pollinating hummingbirds on a tropical shrub. Ecology 63: 354-360.

Roubik, D. W. 1983. Experimental commimity studies: time-series tests of competition between African and Neotropical bees. Ecology 64: 971-978. Roubik, D. W. and J. D. Ackerman. 1987. Long-term ecology of euglossine orchid-bees (Apidae: Euglossini) in Panama. Oecologia 73: 321-333.

Roubik, D. W., N. M. Holbrook, and G. Parra V. 1985. Roles of nectar robbers in reproduction of the tropical treelet Quassia amara (Simaroubaceae). Oecologia 66:161-167.

Rust, R. W. 1979. Pollination of Impatiens capensis: pollinators and nectar robbers. 1979. J. Kans. Ent. Soc. 52: 297-308.

Schmitt, U. and A. Bertsch. 1990. Do foraging bumblebees scent-mark food soLirces and does it matter? Oecologia 82:137-144.

Whitham, T. G. 1977. Coevolution of foraging in Bombiis and nectar dispensing in Chilopsis: a last dreg theory. Science 197: 593-596.

Willmer, P. G., A. A. M. Bataw, and J. P. Hughes. The superiority of bumblebees to honeybees as pollinators: insect visits to raspberry flowers. Ecol. Entomol. 19: 271-284.

Wolda, H. and D. W. Roubik. 1986. Nocturnal bee abundance and seasonal bee activity in a Panamanian forest. Ecology 67: 426-433.

Zimmerman, M. and S. Cook. 1985. Pollinator foraging, experimental nectar- robbing and plant fitness in Impatiens capensis. Am. Midi. Nat. 113: 84-91. 36 3 7

60 -1

50 -

APIS BOMBUS

Figure 2. Visits to unrobbed flowers in a foraging bout (mean ± 1 SE, n = 33 for Apis, 7 for Bombus) 3 8

APPENDIX B

Title: Benefits and costs of floral visitors to Chilopsis li7iearis: poUen deposition and stigma closure

Abstract—In pollination mutualisms, nectar-robbers are usually considered antagonists; visitors that enter flowers (legitimate visitors) are usually considered mutualists. However, nectar-robbers may provide some benefits to plants, whereas legitimate visitors may inflict some costs. I compared costs and benefits of floral visitors to Chilopsis linearis (desert willow). This plant was self-incompatible and pollen-limited, so the source of pollen (self vs. outcross) and number of pollen grains deposited affected reproductive success. Chilopsis had sensitive stigmas that closed immediately upon touch and may have reopened later. Stigma closiire may have been costly to plants if too few pollen grains were deposited to set fruit. I found that the probability of stigma reopening depended on the source and niimber of pollen grains deposited. I compared visitors by number of pollen grains deposited, viability of pollen that they deposited, and their effect on stigmas. Nectar-robbers did not deliver benefits to plants by depositing pollen grains, but they also did not inflict costs on plants by causing stigmas to close without adequate pollen having been deposited. Some legitimate visitors imposed costs on plants by reducing pollen viability; nectar-robbers did not. Overall, legitimate visitors usually conferred greater benefits on plants than nectar- robbers did. However, a few legitimate visitors inflicted greater costs than nectar-robbers did, by causing stigmas to close without adequate pollen 39 deposited. In this way, legitimate visitors could inflict some costs on plants that nectar-robbers could not.

Ke5mrords: pollinator effectiveness, mutualism, stigma closure, pollen viability, nectar-robbing 40

Introduction

Mutualisms have been seen as exchanges of benefits, which antagonists can exploit by using or removing without providing reciprocal benefits (Bronstein 1994). However, though mutualists confer net benefits and have a net positive effect on a partner's fitness, mutualists can also inflict costs on partners (Janzen 1979; Keeler 1985; Templeton and Gilbert 1985; Bronstein 2000). Conversely, antagonists may provide some benefits along with the costs they inflict. Comparing costs and benefits of both mutualists and antagorusts may elucidate conditioris under which mutualism may have the potential to switch to antagonism and vice versa, the evolution of specificity or generality within mutualisms (Trivers 1971; Schemske 1983; Bronstein 1994), and the stability of mutualism (Keeler 1985; Bull and Rice 1991). Comparing costs and benefits may also reveal conflicts between mutualists' interests and inform predictions about how mutualism evolves. For example, selection should act not only to increase benefits given to a mutualist, but also to reduce the costs inflicted by its partners. However, an evolutionary arms race may result because there is a conflict between mutualists' interests; a cost to one mutualist may translate into a benefit for the other. In addition, whether costs can be reduced by mutualists may depend on whether costs are correlated with benefits or whether the visitors that impose the greatest costs are the same visitors that provide the greatest benefits. Costs for plants in pollination mutualisms include resources spent providing attraction and rewards, but also include any effects by a floral visitor that reduce the chance of pollination by another visitor (Table 1). For 41 example, visitors may inflict a cost on a plant by reducing the viability of poUen that they touch and transfer to conspecific stigmas. Ants and wasps are known to reduce viability of pollen placed on their thoraxes (Harriss and Beattie 1991). However, whether floral visitors differ in their effect on viability of poUen moved to stigmas is unknown. Deposition of inviable pollen may be costly if it reduces the likelihood of future pollination. Costs and benefits are usually determined differently for different kinds of floral visitors, because floral visitors are typically divided a priori by behaviors. Visitors are usually considered to be mutualistic if they enter flowers (i.e., are "legitimate visitors") and appeeir to contact reproductive parts. Other visitors, such as nectar-robbing bees, appear to be antagonists because they remove nectar from flowers via slits in corollas, without contacting reproductive structures. However, if we wish to compare legitimate visitors to nectar-robbers in terms of costs that they inflict and benefits that they provide, both costs cind benefits of participants in an interaction should be determined regardless of their behaviors at flowers. For example, a nectar- robber may be strongly beneficial if it touches anthers and stigmas of a flower that it robs (Waser 1979). For a self-compatible plant, a nectar-robber may also be beneficial if it causes pollen transfer within flowers by its movement on corollas. Conversely, a legitimate visitor may fail to touch stigmas of flowers that it enters, though it removes nectar. Thus, nectar-robbers could provide xmexpected benefits to plants and legitimate visitors could impose unexpected costs. In this research, I compared legitimate visitors (apparent mutualists) to nectar-robbers (apparent antagorusts) in their benefits and costs to one plant 42 species, Chilopsis linearis, during visits to flowers. I compared different genera of visitors in terms of the benefits delivered through a component of female reproductive success, pollen deposition (number of grains and source). I also compared visitors with regard to two costs: (1) reduction of pollen viability of deposited pollen, and (2) stigma closure after low-quality visits. The selective impact of each genus on the plant also depends on the visitation rate for each genus, not only the costs and benefits from individual visits. (Stebbins 1970 in Schemske 1983, Addicott 1985). Hence, I also determined the floral visitation rates of common visitor genera over several flowering seasons. Stigmas of Chilopsis flowers have 2 lobes that close upon contact. The phenomenon has not been well investigated, but stigmas of some species are known to reopen with time (Newcombe 1924; Bertin 1982). In order to quantify the cost of visitors depositing litfle pollen on stigmas, I examined factors that affect the likelihood of stigma closure and reopening, and the breeding system of the plant. I hypothesized that stigmas that close upon receiving a load of pollen may have different probabilities of reopening depending on amount, viability, and source (self vs. outcross) of that pollen. If stigmas close permanently only when they have received enough poUen to set fruits, then stigma closure may indicate that a visit has been beneficial. Alternatively, stigma closure may be costly if the stigma holds little viable pollen, because closure of the stigma prevents further pollination. 43

Methods

The plant and the sites Chilopsis linearis (Bignoniaceae), the desert willow tree, grows along washes and roads in the Sonoran Desert from 0 to 1800 m (Turner et al. 1995). Chilopsis blooms from May to August, with a large pulse early in the season for 2-3 weeks at the sites in this study. During the pulse, a tree may have himdreds of flowers open simultaneously. After the pulse, only about 1-4 flowers are open at a time on a tree. This mass-flowering plant (sensu Gentry 1974) produces showy, purpKsh-white flowers on racemes. Flowers are zygomorphic and sympetalous, with a funnel-shaped, two-lipped corolla. Anthers dehisce and anthesis ocoirs in the late eiftemoon to early evening (day 0). At anthesis, flowers begin to produce nectar. Nectar production ends before dawn (Whitham 1977) and nectar is not renewed (unpubl. data). Flowers of C. linearis usually last 2 days; stigmas are imreceptive and closed on day 1. Fruit and seed production in the self-incompatible flowers (Peterson et al. 1982) have been found to be limited by pollen but not by resources at a site near Tucson, Arizona (Peterson et al. 1982). Fruit set at that site was 9 % (Peterson et al. 1982). Fruit set at 2 other sites near Tucson in other years was 1.9 % - 12.8 % (Pfister 1989). To determine the number of ovules and seeds in fruits, I collected fruits in 1996 that were undamaged by seed predation. I conducted most experiments at a site in a large wash, 6.9 km N of Portal and San Simon Roads in Portal, Arizona (desert site, "DS", 1387 m, 31° 57' N, 109° 07' W). A second site was located on Cave Creek, near Portal 44

("CC", 1463 m, 31° 54' N, 109° 08' W). Chilopsis bloomed in a pulse from May 21-Jime 24,1995 and May 15-June 7,1996, respectively, at these study sites. A third site was located at the St. Mary's Road wash, near the Tucson Mountains, Tucson AZ ("TM"), which flowered during Aug. 28-Sept. 11,1996. To determine visitation rates, I observed visitors at DS in 1993 and 1994 to large patches of flowers (16-285 fl.) for 17 and 7 half-hoior observation periods, respectively. In 1995 and 1996,1 calculated visitation rates at DS and CC based on 1/2 hr. observation periods, with the addition of observatioris of small patches of previously unvisited flowers (3-13 fl.) for 1/2 hr. - 2 hr. each day during pollination effectiveness experiments described below. (1995, DS: 4 1/2 hr. obs. periods and 28 d. of smaU patches; 1995, CC: 41/2 hrs. and 7 d. of small patches; 1996, DS: 51/2 hrs. and 5 d. of small patches; 1996, CC: 6 1/2 hrs. and 11 d. of small patches.) Data for two species of Megachile and for two species of Anthophora were combined for analyses. Male Anthophora were excluded, except for analysis of pollen deposition. I deposited representative insect specimens at the University of Kansas and University of Arizona. To determine rates of fnait set at DS, I marked all buds initiated below 3 m on 3 trees during 3 days in 1993 (N = 228 flowers) and 1 day in 1994 (N = 30). For each census, I rechecked buds twice each day for ovary or fruit abortion until corollas dropped. When corollas fell, I checked once a week for fruit abortion until collection of fruits 2-4 weeks after buds were marked.

Breeding system and stigma closure Stigma lobes opened 37-43 hrs after anthesis, during the late morning of day 2. The two-lobed stigmas closed with touch. If no pollen was 45 deposited when stigmas were closed by the touch of a brush, they reopened after about 10 min. Stigmas sometimes reopened within 30 min. cifter closure caused by a floral visitor. Hydrogen peroxide tests (Dafhi 1992) showed that stigmas were only receptive when lobes opened, and then orily the inner surfaces of lobes were receptive. To confirm degree of self-incompatibility and to determine the response of stigmas and ovaries to pollen from different sources, I conducted several experiments. To determine degree of self-incompatibiKty of Chilopsis, I conducted experiments beginning 25 May, 1993 at DS, 1 June, 1994 at DS, 9 Jime, 1995 at DS and 16 June, 1995 at CC. On day 0,1 emasculated and bagged buds to be hand-pollinated. I left another group of marked buds open to natural pollination. In addition, during hand-pollination experiments in 1995,1 bagged buds to be used as pollen donors before anthesis to prevent contamination by pollen from other plants. On day 2, stigmas became receptive. I hand-pollinated recipients with an excess of pollen from either the same plant (self) or a different plant (outcross); in both cases, I collected pollen from 12 anthers from 3 flowers on one plant. For outcross pollen, the tree I used as a pollen source was ca. 150 m from the recipient. I rebagged flowers to prevent visitation and subsequently observed them for fruit production or ovary abortion. In addition to these treatments, I marked 258 buds at DS in 1993-4 and observed them for natural pollination rates. I compared numbers of fruits set and aborted Ln selfed and outcrossed treatments from 1995 using Fisher's exact test. To compare probabilities of reopening of stigmas that were hand- pollinated with either self or outcross pollen, I conducted an experiment on 28 46

Aug.-ll Sept., 1996, at TM. I bagged buds on several trees from 1630 to 1830 on three afternoons (day 0). During the experiment, I removed all but target flowers from the bagged inflorescences. On day 2,1 applied an excess of pollen to recipients from first-day, male-phase flowers that were bagged and imvisited. Recipients were female-phase target flowers on the same and different trees. I used pollen from 12 anthers from 3 flowers from one tree at least 30 m away as the outcross pollen source. I rebagged the female-phase recipients. I returned to check for degree of stigma closure of each flower at approx. 10 min., 1 hr., and 2 hr. after hand-poUination, and at dusk of the same day, when coroUas were wilted and falling. For this paper, stigmas will be described as either "open", with lobes separated, or "closed", with lobes touching. I compared numbers of stigmas reopening or remaining closed after being treated with self or outcross pollen using a Fisher's exact test. From this experiment, I collected selfed and outcrossed styles from target flowers 32 hrs. after hand-pollination and observed pollen tube growth to determine whether plants have gametophytic or sporophytic incompatibility. I stained pollen tubes in the styles with aniline blue following methods in Dafni (1992). If plants have gametophytic incompatibility, incompatible pollen does not produce pollen tubes; Lf plants have gametophytic incompatibility, pollen tubes from incompatible pollen grow towards ovules but do not penetrate them.

Pollen deposition by visitors and viability of deposited pollen I compared different genera of visitors according to number and viability of pollen grains and the duration of closure of stigma lobes after 47 visits. I marked about 400 buds over 37 days in 1995 and 1996 with permanent pen on the pedicel and with colored yam at least 20 cm distant on the branch. Between 1600-1830 on the day before flowers opened, I placed floral visitor exclusion bags over buds. That evening when buds opened and the next morning (day 1), I removed all but one flower within each bag to prevent pollen passing to unvisited target flowers within bags. From 0630-1030 of day 2,1 removed all of the bags in the patch and observed visitation. Before visitation, all stigma lobes were open. After an insect visited a target flower, I caught the insect, identified it, checked the flower for stigma closure, and rebagged the flower. I rebagged flowers that were not visited and used them as control flowers. Using ICruskal-Wallis tests, I compared the number of poUen grains deposited by visitors. 2 to 5 hrs after visitation, I collected flowers and stained stigmas with Alexander's stain (Alexander 1980), which stains pollen differentially based on presence or absence of protoplasm, an indicator of viability. Using nonparametric multiple comparisons, I compared both number and viability of pollen grains deposited by visitors.

Stigma closure after visitation In the experiment described above, I examined stigmas within a minute after each visit to determine whether stigmas closed irutially. I rechecked two to four hrs after visitation to determine whether stigmas remained closed permanently. For all visitors combined, I used logistic regressions to determine whether the number of pollen grains or viability of pollen deposited by visitors affected the probability of stigmas remaining 48 permanently closed. For each genus of visitor, I used Fisher's exact test to compare the initial probability of stigma closure with the probability of closure of control stigmas described below. Using a contingency table, I determined whether genus of visitors had an effect on the percentage of stigmas remaining permanently closed. Control (i.e., imvisited) flowers were compared with visited flowers in order to estimate the amount of self pollen on stigmas that was not deposited by visitation. Under nattiral conditions, the amotint of self pollen on stigmas would probably have been less than observed on stigmas in these experiments because some or all pollen would have been removed by visitors from anthers before stigma lobes opened (day 1). Comparing of the rates of stigma closure of visited flowers to rates of closure of unvisited flowers therefore controlled for the effect of the wind or investigator jostling pollinator exclusion bags and causing stigmas to close.

Results

Breeding system and stigma closure Open-pollinated, unmanipulated flowers produced fruits from 6 % of flowers (N = 270) from 1993-5. In contrast, flowers hand-pollinated with outcross pollen produced fruits from 44 % of flowers (N = 44, Table 2), which suggests that Chilopsis at DS was pollen-Umited. During the same years, no fruits were set from hand-pollination by self pollen (N = 42 flowers), which indicates that Chilopsis was self-incompatible. Stained stigmas and styles from selfed and outcrossed flowers confirmed that Chilopsis is self-incompatible. 49

Incompatibility was gametophytic; pollen tubes from self poUen reached ovaries but did not fertilize ovules. Because Chilopsis was pollen-limited and self-incompatible, pollen deposition was needed to achieve female reproductive success; the magnitude of benefit conferred by a visitor depended on amoimt and source of pollen deposited. Both source and amoimt of pollen deposited affected the probability of stigmas remaining closed after initial closure from touch. In hand-pollination experiments, stigmas remained permanently closed significantly more often after receiving outcross pollen than after receiving self pollen (Table 3). In addition to source of pollen, the probability of stigmas of visited flowers remaining permanently closed was also affected by the number of pollen grairis deposited (Fig. 1). After a visit, stigmas held 0 - 824 pollen grains (mean 139, N = 188). For those stigmas that were closed irutially (indicating that they had been touched), for every 50 grains on the stigma, the probabiHty of reopening decreased by a factor of 2 (r^ = 0.33, p < 0.0001, N = 56; Fig lA). However, the probability of stigma reopening was not affected by deposition of inviable pollen by visitors (r^ = 0.0054, p < 0.57, N = 44; Fig IB.) These results indicate that the source and amount of pollen deposited on stigmas, but not viability of that poUen, affected the probabiKty of stigma reopening. Thus, stigmas were more likely to remain permanently closed when large amounts of outcross pollen had been deposited.

Visitation rate and rates of pollen deposition Chilopsis flowers were visited primarily by bees, but also by hummingbirds (Brown et al. 1981) and Lepidoptera. Bees, especially Bombus 50 sonorus and Apis mellifera, were the most common visitors entering flowers from 1993-1996 at DS and at CC (Table 4). One bee, Xylocopa califomica, robbed nectar. Visitation to flowers by both legitimate visitors and nectar- robbers peaked aroimd 0700-0930, depending on temperature. Visitors were almost completely absent from 1130-1630, and then returned in smaller numbers from 1630 to dusk. I did not see nocturnal visitors during several nights of observ'^ation (unpubL data). Among the most common visitor genera, the order of abundance changed among years. Apis were more common in 1993, whereas Bombiis were more common in 1994 and 1996. The nectar-robbing bees, Xylocopa, were present in all years. Rare visitors included Megachile bees, hiommingbirds {Archilochus colubris and Selasphorus platycercus), and halictid bees. The relative benefits conferred by legitimate visitors and nectar- robbers depended partly on the nixmber of pollen grains each genus deposited on Chilopsis stigmas. All flowers had poUen on stigmas. In the case of control flowers, these grains were presumably from anthers of the same flower (autogamous pollen). One legitimate visitor, Bombiis, deposited significantly more poUen during visits than was foimd on control flowers (Q =

3.77, N = 49, QO.05, II = 3.317, p < 0.05, Fig. 2). In contrast, stigmas of flowers visited by legitimate visitors Anthophora, Apis, Megachile, and nectar-robbing Xylocopa did not have significantly more pollen than was foimd on control stigmas (Q = 2.30, N = 14; Q = 0.163, N = 35; Q = 0.317, N = 15; Q = 2.16, N = 37;

QO.05, 11 = 3.317, Fig. 2). Thus, only one legitimate visitor was more beneficial than robbers in terms of pollen deposition per visit. 5 1

The relative benefits conferred by a given visit also depended on the number of ovules in the visited flower. Fruits collected in 1996 had 50 - 97 ovules (N=28) and a minimum of 18 seeds. Frequency distributions of pollen grains deposited per visit for the most common visitors (Fig. 3) show that visits within genera vary greatly in amounts of pollen deposited. Several flowers received few pollen grains when visited by Bombiis, a visitor that generally deposited many pollen grains. Conversely, several flowers visited by Apis received hundreds. Thus, individual visits from each genus of legitimate visitor ranged from strongly beneficial to weakly beneficial, depending on number of pollen grains deposited, although on average, Bombus benefited plants and Apis did not.

Stigma closure after visitation In 42.2% of all visits (n = 148), stigmas immediately closed, indicating that they had been touched by the visitor, in comparison to control stigmas, which closed at a rate of 2.2% (n = 36). Stigmas closed immediately upon visits by legitimate visitors Bombus, Anthophora, Megachile, and Apis significantly more often than stigmas of imvisited flowers (Fig. 4), indicating that these legitimate visitors often touched stigmas and so had the potential to deposit pollen. In contrast, Xylocopa, nectar-robbing bees, did not cause stigma closure any more than in control stigmas (Fig. 4), presumably because they removed nectar from slits in corollas. Because nectar-robbers did not contact stigmas, the few stigmas that were closed temporarily following those visits were likely due to wind or jarring of pollinator exclusion bags. 52

Some stigmas touched by visitors remained permanently closed after visitation (Fig. 4). In contrast, control stigmas and stigmas of robbed flowers never remained permanently closed (N = 36,38). All legitimate visitors caused some permanent stigma closure, even though only Bombus deposited significantly more pollen grains than fovmd on control stigmas (Fig. 2). Legitimate visitors differed in the probability that stigmas they touched remained closed (G^^ = 8.8, df = 3, p < 0.03); Bombus visits led to the largest percentage of permanent closure and Apis visits led to the smallest.

Viability of Deposited Pollen Because the likelihood that stigmas reopened was not affected by pollen viability (Table 5), deposition of inviable pollen may have infUcted a cost on the plant by leading to permanent stigma closiire and reducing the probability of later successful pollination. This cost could only have been inflicted by legitimate visitors, not nectar-robbers, because only legitimate visitors touched stigmas. Of the legitimate visitors, only Bombus and Anthophora deposited pollen of significantly reduced viability compared to pollen found on unvisited flowers (Table 5). Thus, Bombus, the visitor that conferred the greatest benefit to plants in terms of pollen deposition (Fig. 2), also reduced pollen viability the most. The costliness of deposition of inviable pollen depended on the cost of missed opportimities for later pollination, which was not measured. However, benefits probably outweighed costs for most Bombus visits: two Bombus reduced pollen viability by 64% in comparison to controls, but on average Bombus reduced viability by only 5%. 53

Discussion

Benefits and costs of legitimate visitors and nectar-robbers Understanding both benefits conferred and costs inflicted from all participants in a mutualism is critical for predicting selective forces acting on mutualisms. These forces may act on mutualists to reduce costs, to exclude antagorustic visitors, or to move towards generality or specificity. Costs and benefits in most mutualisms, besides pollinating seed predators such as joiccas and figs, are in different currencies, so it is difficult to determine net effects of visitors (i.e., gross benefits minus costs). Thus, in the Chilopsis pollination mutualism, the presence and magnitude of different costs and benefits from legitimate visitors and nectar-robbers were investigated. In this study, only legitimate visitors ever conferred benefits to plants, but legitimate visitors also inflicted some costs that nectar-robbers did not. Certain legitimate visitors conferred no greater benefits to plants than did nectar-robbers, in terms of pollen deposition. On average, only one species, Bombus, conferred more benefits to plants than did nectar-robbers, though pollen deposition varied greatly among visits of each legitimate visitor species. In contrast, stigmas of flowers visited by Apis usually held nearly the same number of pollen grains as those visited by nectar-robbers, though a few Apis deposited a large quantity of pollen. Whether visits in which Apis deposited few pollen grains resulted in a beneficial net effect to the plant depended on the cost of stigma closure, which depended on the breeding system of the plant. Because Chilopsis was self- 54 incompatible, only outcross pollen deposition yielded benefits for plants. Thus, a stigma must have been touched by a visitor to set a fruit. Rates of temporary closure identified genera that generally touched stigmas and thus had the potential to confer benefit to plants. Rates of permanent stigma closure m.ore precisely distinguished genera that generally deposited large amounts of pollen, such as Bombus, from genera that did not, such as Apis and nectar-robbers, because the likelihood of permanent closure depended on amount of pollen deposited. Though stigma closure generally indicated that a benefit had been given, after some visits stigma closure may have been costly. Legitimate visitors other than Bombus (Apis, Megachile, and Anthophora) did not deposit significantly more pollen grains than did nectar-robbers, although they did cause some permanent stigma closure. It is likely that permanent stigma closiire by ineffective legitimate visitors 'A^as caused by deposition of a few outcross pollen grains, because the likelihood of permanent stigma closxire depended on deposition of outcross grains (Fig. 1). In these cases, the benefit of pollen deposition probably was outweighed by the opportimity cost resulting from stigma closure, because closure prevented later successful pollination. Even temporary closure may have been costly because stigmas generally reopened two hrs. after closure, which may have prevented pollination during the daily period of high visitation to plants. In contrast, nectar-robbers did not cost plants by leading stigmas to close permanently. The cost of stigma closure depended on whether more grains had been deposited during a visit than the minimum number of pollen grains needed to set a fruit. Because a minimum of 18 seeds was found in collected fruits, there may have been a minimiun number of fertilized ovules necessary for a fruit to mature- If 18 outcross poUen grains were needed to set a fruit, then at least ca. 100 grains should be on a stigma for a finiit to mature (i.e., 18 + 81 autogamous foimd on control stigmas = 100), though larger pollen loads would probably have to be deposited to ensure that 18 grains are of outcross origin. Experiments would be needed to confirm the Vcdidity of these estimates. Of those stigmas that remained permanently closed (N = 52), 37% had received fewer than 100 viable pollen grains. Thus, under these assumptions, ail legitimate visitor species may have sometimes caused stigmas to close permanently with too few grains to set fruit (13% of visits by Bombus, 24% by Apis, and 11% by Megachile, and 9% by Anthophora ; sample sizes were too small to compare genera statistically). In contrast, no nectar-robbers inflicted this cost on plants, because visits by nectar-robbers did not lead to permanent stigma closure. Another cost that the legitimate visitors Bombiis and Anthophora inflicted that nectar-robbers did not was deposition of inviable pollen. Deposition of inviable pollen was costly for the plant because inviable poUen was as likely to lead to stigma closure as viable pollen. However, this effect was probably not biologically significant to the plant, because the reduction was so small; Bombus and Aiithophora only reduced pollen viability about 5%, on average. In other studies, contact with Apis has been found to decrease pollen viability significantly, particularly when the pollen is in the pollen baskets on the legs (Kubisova and Haslbachova 1989, Vaissiere et al. 1996). However, Chilopsis pollen was probably deposited by contact with the bees' thoraxes, whereas 56

pollen in pollen baskets would have permanently lost the potential to pollinate stigmas. The effect of visitors on plants depended on more than relative benefit or cost to a plant; it also depended on visitation rate. Because Bombus, Apis, Anthophora, and the nectar-robber Xylocopa all visited plants often, they potentially had a significant effect on the reproductive success of plants, depending on whether they delivered a benefit to plants by depositing a large amount of outcross pollen.

Stigma closure in pollination mutualisms In many pollination mutualisms, the effect of floral visitors may depend pardy on stigma closure. Stigmas with lobes that close from touch are known from all Bignoniaceae (Bertin 1982) and Mart)miaceae that have been tested. Sensitive stigmas are also known from some genera within Acanthaceae, Lentibulariaceae, and Scrophiolariaceae (Newcombe 1924). Benefits of stigma closure that may compensate for costs have been suggested, but rarely investigated. They include the following. 1) Stigma closure may keep an incoming load of outcross pollen concentrated at the flower on which it was deposited on (Thieret 1976; Bertin 1982). If outcross pollen is limiting, this may provide a benefit for plants. That function is likely to be especially important in a pollen-limited, self-incompatible plant such as Chilopsis, which may have a threshold number of pollen grains needed to set a fruit. Closure may 2) prevent male reproductive success from being reduced because pollen is lost on the plants' own stigmas (pollen discounting) or 3) cause the stigma to move away from anthers and thus to maximize pollen 57

removal from anthers (Webb and Lloyd 1986). This function is probably not relevant for Chilopsis, because anthesis occurs a day before stigmas open. Closure may also 4) aid pollen germination in dry habitats (Newcombe 1924). This h5^othesis may be valid for Chilopsis, because pollination of Chilopsis occurs during the summer in the Sonoran Desert. However, this latter h5^othesis was not supported in Mimulus growing in a dry habitat; pollen germination was not greater in closed stigmas than stigmas that were forced open (Fetscher and Kohn 1999). Alternatively, stigma closure could have evolved due to selection in the past but have no current function. The presence of sensitive stigmas in the families related to Bignoniaceae is likely to be a synapomorphy, such that the presence of stigma closure may be a result of phylogenetic constraint rather than a current adaptation. The effect of stigma closure on pollination mutucdisms depends on the mechanism of closure, because stigma closure and reopening partly determines whether visitors that deposit few grairis wQl be ultimately antagonistic as single visitors or mutualistic in combination with other visitors. The timing of stigma reopening suggested either that incompatible pollen was recognized in two locations, one at the stigma surface or upper stigma and another at the ovary, or that the first pollen tube from outcross pollen to penetrate an oviile sent a chemical signal to the stigma lobes to remain closed. At the stigma level, stigma lobes peaked in reopening two hours after they received a load of self pollen. However, it was hours later that pollen tubes from outcross pollen reached ovules and pollen tubes from selfed pollen failed to fertilize ovules. Because the timing of stigma closure is critical to 58

understanding costs and benefits of visitors in pollination mutualisms, stigma closure should be studied across plant families. If plant species with similar pollinators and environments converge on similar timing of stigma closure and reopening, it is likely that selection has been acting on the mechanism in resporise to costs and benefits of visitors.

Conclusion

Nectar-robbers in other pollination mutualisms can be antagonists of plants by destroying flowers (McDade and Kinsman 1980). A few nectar- robbers can also be mutualists, by pollinating flowers while they rob (Waser 1979). Although nectar-robbers did not provide benefits to Chilopsis by depositing pollen, they also inflicted no cost on the plant by leading stigmas to close permanently or by depositing inviable pollen, in contrast to legitimate visitors. Nectar-robbers potentially could have exacted an indirect cost from plants by removing nectar and causing pollinators to avoid flowers. However, in other work I have found that nectar removal did not affect the behavior of Bombiis (Richardson 1995, Appendix A), usually the most effective pollinator, whereas the ineffective Apis avoided robbed flowers. Thus, nectar- robbers did not inflict a cost on plants in female reproductive success either directly or indirectly, in contrast to some legitimate visitors. Thus, in Chilopsis, legitimate visitors both imposed the greatest costs on plants and provided the greatest benefits to plants. It has been suggested that in generalized mutualisms, mutualists can control costs of partners by excluding the most cintagonistic visitors (PeUmyr 1997). However, in Chilopsis, 59 selection may not be able to act to exclude antagonistic visitors because they are individuals of the same genera as mutualistic visitors, rather than differing in their behavior and morphology as nectar-robbers and legitimate visitors do. 60

Table 1. Benefits and costs of a single visitor to a flower in a pollination mutualism. Female Reproductive Success Male Reproductive Success Benefits Benefits poUen deposition pollen dispersal to another plant stigma closure, if adaptive pollen dispersal to self flowers (if self- compatible) Costs Costs deposition of rnviable pollen, if it reduction of pollen viability by reduces likelihood of contact successful pollination effects of a visit that reduce effects of a visit that reduce probability of another probability of another successful visit (i.e., successful visit (i.e., opportTonity cost)—pollen or opportimity cost)—stigma anther removal, tripping a closure, tripping a mechanism, mechanism, scent-marking, scent-marl^g, damage or damage or color change in color change in flower, flower clogging stigma with other species of pollen pollen dispersal to self flowers or same genot3^e if incompatible resources spent making flower and (pollen discounting) floral rewards resources spent making flower and floral rewards, including pollen eaten by visitor

pollen lost by pollinator grooming or transferred to different species of plant

seeds given to pollinating seed predators (i.e., yuccas, figs) Table 2. Percent fruit set, total buds marked in parentheses. 2-4 weeks after buds were initiated. Natural pollination rate was determined from buds marked and left unmanipulated. Selfed and outcrossed flowers were pollinated by hand; pollen donors were bagged in 1995. Treatment 1 was conducted at DS, treatment 2 at DS and CC. Treatments 1 vs. 3 and 2 vs. 3 were compared usin^ contingency tables, p < 0.Q5. Treatment 1993 1994 1995 Total

1) Natural pollination rate 4 (234) 6 (36) ~ 6 (270) 2) Selfed 0 (6) 0 (6) 0 (30) 0 (42) 3) Outcrossed 25 (8) 0 (8) 59 (32) 44 (48) 62

TABLE 3. Percent stigma closure after deposition by hand of 25-100 grains of self or cross pollen. Sample sizes are in parentheses. Pollen Source Stages Self Cross p-value % initial closure following hand pollination 100 (29) 100 (29) % remaining closed imtil corolla fallen 58.6 (29) 82.8 (29) p = 0.04 63

Reopened Remained Closed

Reopened Remained Closed

Fig 1. Comparison between stigmas that reopened after being closed by visitors and stigmas that remained permanently closed. (A) Mean numbers of pollen grains deposited by visitors (N = 23,30). (B) % viable pollen deposited by visitors (N = 17,25). Bars represent + 1 SE. TABLE 4. Summary of visitation rate (# visits by each taxon/fl-hr). Visitor 1993-DS 1994-DS 1995-DS 1995-CC 1996-DS 1996-CC Bombus 0.035 0.40 0.027 1.3 0.050 0.47 Anthophora 0 0.039 0.027 0.042 0 0.015 Megachile 0 0 0.076 0.048 0.011 0.0031 Apis 0.21 0.15 0.13 0.14 0.022 0.098 Xylocopa (nectar-robber) 0.029 0.23 0.36 0 0.043 0.0092 Other insects 0.0067 0.014 0.066 0.083 0.0036 0.018 Hummingbirds 0.0033 0.014 0 0 0.079 0.055

a\ 65

500

^ 200

2 100

-V^

Fig. 2. Summary of mean numbers of pollen grains found on stigmas after single visits to previously unvisited and bagged Chilopsis linearis flowers. N's above bars indicate number of observations of each species of visitor. Vertical bars indicate +1 SD around the mean. Asterisk above bar indicates species that deposits significantly more pollen grains than found on unvisited, control flowers, based on nonparametric multiple comparisons. 66

20 Control N = 30 15H o Median = 78 c o s-10 (D

5-j

0 0 100 200 300 400 500 600 700 800 Number of Pollen Grains

20 Bombus 15-1 N = 49 o Median = 231 c 0) g-10H o ul

200 300 400 500 600 700 800 Number of Pollen Grains

Fig. 3. Number of pollen grains on stigmas after single visits by most common visitors and imvisited control flowers. N = sample sizes. (Bombus and Apis were sigruficantly different from controls, and Xylocopa was not, in nonparametric multiple comparisons; Bombus: Q = 3.77, N = 49, Apis: Q = 0.163, N = 35, Xylocopa: Q = 2.16, N = 37; Qo.os, ii = 3.317.) 67

20 Apis 15 N = 34 o>> Median = 55 c CD i-10 (D LL 5

0 0 100 200 300 400 500 600 700 800 Number of Pollen Grains 20 Xylocopa 15 nectar-robber o N = 37 c CD Median = 20 g-10 ^ LL 5

0 0 100 200 300 400 500 600 Too ^ Number of Pollen Grains 68

100 • % initially closed ^ % of stigmas closed after 2-4 hr.

^ 16

^ 14

^ 31

-A •>$?

r ry \Q ^ ^ kzS^

Legitimate Visitors STectar- robber

Fig. 4. Percentage of stigmas closed initially by touch of visitors and 2-4 hrs. after visitation. Sample sizes of total # of stigmas closed are above bars. Differences between visited flowers and control flowers in rates of initial closure were compared using Fisher's exact test. Asterisks indicate genera significantly different from control in percentage of initial closure. 69

TABLE 5. Summary of mean percent viable pollen grains deposited on stigmas by a single visitor, ± 1 SD of the mean. Sample sizes of total # of stigmas closed are in (). Raised letters indicate similar groups based on nonparametric multiple comparisons, p < 0.05.

Visitor Mean % viable pollen Control 90.7 ± 11-4 (30) a Legitimate visitors Bombus 85-3 ± 21.7 (31) b Apis 91.4 ± 17.9 (26) a Anthophora 86.7 ± 19.7 (10) b Megachile 93.1 ± 9-9 (9)a,b Nectar-robber Xylocopa 93.2 ± 13.9 (35) a 70

References

Addicott JF (1985) On the population consequences of mutualism. Community Ecology. Case T and Diamond J (eds) Berkeley, University of California Press, pp 425-436.

Bertin RI (1982) Floral biology, hummingbird pollination, and fruit production of tnmipet creeper, Campsis radicans (Bignoiuaceae) Am J Bot 69:122- 134.

Bronstein in prep. The costs of mutualism. Amer Zool.

Bronstein JL (1994) Our current imderstanding of mutualism. Q Rev Biol 69:31- 51.

Brown JH, Kodric-Brown A, Whitham TG, Bond HW (1981) Competition between hummingbirds and insects for the nectar of two species of shrubs. Southwest Nat 26:133-145.

Bull JJ, Rice WR (1991) Distinguishing mechanisms for the evolution of cooperation. J ITieor Biol 149:63-74.

Cushman JH, Beattie AJ (1991) Mutualisms: assessing the benefits to hosts and visitors. Tr Ecol Evol 6:193-195.

Dafni A (1992) Pollination Ecology. New York, Oxford University Press.

Fetscher AE, Kohn AR (1999) Stigma behavior in Mimulus aurantiacus (Scrophulariaceae). Am J Bot 86:1130-1135.

Fleming TH, Sosa VJ (1994) Effects of nectarivorous and frugivorous mammals on reproductive success of plants. J Mammal 75:845-851.

Gaume L, McKey D (1999) An ant-plant mutualism and its host-specific parasite: activity rhythms, young leaf patrolling, and effects on herbivores of two specialist plant-ants inhabiting the same myrmecophyte. Oikos 84:130-144.

Gentry AH (1974) Coevolutionary patterns in Central American Bignoniaceae. MO Bot Gard Ann 61:728-759.

Harriss FCL, Beattie AJ (1991) Viability of pollen carried by Apis mellifera. Trigona carbonaria. and Vespula germanica. J Aust Ent Soc 30:45-47.

Huth CJ, Pelbnyr O. (2000) PoUen-mediated selective abortion in yuccas and its consequences for the plant-pollinator mutualism. Ecology 81:1100-1107. 7 1

Janzen DH (1979) How many babies do figs pay for babies? Biotropica 11:48- 50.

Keeler D (1985) Costrbenefit models of mutualism. In: Boucher DH (ed) The Biology of MutuaUsm. London, Croom Helm, pp 100-127.

Kubisova S, Haslbachova H (1990) Pollen grains adhering to bodies of honeybees after cole Brassica oleracea crop pollination. Apidologie 21:1- 10.

Letoumeau DK (1991) Parasitism of ant-plant mutualisms and the novel case of Piper. In: Huxley CR, Cutler DR. Ant-Plant Interactions. Oxford, Oxford University Press, pp 390-396.

McDade LA, Kinsman S (1980) The impact of floral parasitism in two neotropical hummingbird-pollinated plant species. Evolution 34:944- 958.

Newcombe FC (1924) Sigriificance of the behavior of sensitive stigmas n. Am J Bot 11:85-93.

PeUmyr O (1997) Stability of plant-animal mutualisms: keeping the benefactors at bay. Trends Plant Sci 2: 408-409.

Petersen C, Brown JH, Kodric-Brown A (1982) An experimental study of floral display and fruit set in Chilopsis linearis (Bignoniaceae). Oecologia 55:7- 11.

Pfister RW (1989) Effects of Nectar-robbing by Xylocopa califomica on Chilopsis linearis (Bignoniaceae). Dept. of Ecol. and Evol. Biol. Tucson, AZ, The Univ. of AZ.

Richardson SC (1995) Temporal variation in pollinator abimdance and pollinator foraging behavior in response to robbed flowers, in: Biodiversity and Management of the Madrean Archipelago: the Sky Islands of Southwestern United States and Northwestern Mexico, Tucson, Arizona, USDA Forest Service, pp 311-316.

Schemske SW (1983) Limits to specialization and coevolution in plant-animal mutualisms. In: Nitecki MH (ed) Coevolution. Chicago, Uruversity of Chicago Press, pp. 67-109.

Stebbins GL (1970) Adaptive radiation of reproductive characteristics in angiosperms, I. Pollination mecharusms. Ann Rev Ecol Syst 1:307-326. 72

Stephenson AG (1982) The role of the extrafloral nectaries of Catalpa speciosa in limiting herbivory and increasing fruit production. Ecology 63:663- 669.

Templeton AR, Gilbert LE (1985) Population genetics and the coevolution of mutualism. In: Boucher DH (ed) The Biology of Mutualism. London, Groom Helm, pp 128-144.

ThieretJW (1976) Floral biology of Proboscidea louisianica (Martyniaceael. Rhodora 78:169-179.

Trivers RL (1971) The evolution of reciprocal altruism. Q Rev Biol 46:35-57.

Turner RM, Bowers JE, Burgess TL (1995) Sonoran Desert Plants: An Ecological Atlas. Tucson, The University of Arizona Press.

Vaissiere BE, Malaboeuf F, Rodet G (1996) Viability of cantaloupe pollen carried by honeybees Apis mellifera varies with foraging behavior. Naturwiss 83:84-86.

Waser NM (1979) Pollinator availability as a determinant of flowering time in Ocotillo fEouquieria splendensl. Oecologia 39:107-121.

Webb CJ, Lloyd DG (1986) The avoidance of interference between the presentation of pollen and stigmas in angiosperms. EL. Herkogamy. NZ J Bot 24:163-178.

Whitham TG (1977) Coevolution of foraging in Bombus and nectar dispensing in Chilopsis: a last dreg theory. Science 197:593-596.

Wilson P, Thomson JD (1991) Heterogeneity among floral visitors leads to discordance between removal and deposition of pollen. Ecology 72:1503-1507. 73

APPENDIX C

Title; The Effects of Nectar-robbers on Mutualists—Costly or Beneficial?

Abstract. Mutualisms are beneficial interactions between species that also may exploit each other. Theories have predicted that the fitnesses of mutualists are tied together, so the presence of outside species that take rewards without providing reciprocal services, "cheaters", should have costs for both mutualistic partners. Cheaters in general are predicted to cost mutualists; however, theories specific to pollination biology predict that nectar-robbers, one kind of cheater, will benefit plants by causing pollinators to move quickly from robbed plants, carrying pollen to other plants. Nectar- robbing bees {Xylocopa californica), presumably have a negative effect on the behavior of pollinators visiting desert willow {Chilopsis linearis) by competing with them for nectar. I tested whether pollinator behavior is affected by the presence of nectar-robbers and whether nectar-robbing had a costly, neutral, or positive indirect effect on the other partners in the mutualism, the plants. Nectar-robbers reduced levels of reward available to mutualists; unrobbed and robbed flowers had the same probability of containing nectar, but nectar volumes were lower in robbed flowers. Apis mellifera, ineffective pollinators in terms of pollen deposition, avoided robbed flowers and spent less time in robbed flowers than in unrobbed flowers. Bombus sononis, effective pollinators, did not avoid robbed flowers. Bombus did spend less time in robbed flowers, but the time that they spent in flowers was not correlated with pollen deposition. Thus, nectar-robbers may have had a positive effect on plants by changing the behavior of the least effective pollinators. I investigated the indirect effects of robbing on the plant's male reproductive success by examining whether powder mimicking pollen dispersed farther from robbed flowers than from imrobbed flowers. On some days, powder dispersed farther from robbed flowers, indicating that robbing may sometimes be beneficial to plants. I also determined whether female reproductive success, measured by the number of pollen tubes growing in styles, was affected by robbing. I found that there was no difference in pollen tube number between robbed and imrobbed flowers, which means that robbers do not cost plants directly by damaging flowers or indirectly by reducing their female reproductive success. Overall, this study suggests that changes in behavior of one mutualist in response to cheating, presumably resulting from competition between mutualist cmd cheater, may sometimes benefit the other mutualist. 75

Introduction

Recently, researchers have become interested in the balance between conflict and cooperation in mutualisms. Mutualists may only benefit each other under certain conditions (Bronstein in prep.) or may have parts of their life histories in which they are antagonistic towards each other (Templeton 1985; Addicott 1986). Mutualists may also switch back and forth from antagonism to mutualism over ecological or evolutionary time (Omelas 1994; Pellmyr et al. 1996). Given that mutualists can both cooperate and exploit each other, how wiU the intrusion of other species affect the interaction? For example, how will mutualists be affected if another species removes a reward that one partner offers its mutualist? Some theories of the evolution of mutualism predict that this "cheater" on a mutualism (Bronstein in prep.), also known as an "aprovechado" (Soberon and Martinez del Rio 1985), "parasite" (Letoumeau 1991), "exploiter", or "antagonist" (Bronstein 1997), will cause a mutualism to become extinct (Trivers 1971; Bull and Rice 1991). This extinction will occtir because the cheater removes rewards without incurring the cost of providing a service or reciprocal reward to the mutualistic partner. By not reciprocating, a cheater may gain a competitive advantage over a mutualist. Because the fitnesses of mutualistic partners are correlated in theory (Addicott 1981; Keeler 1985), the expectation is that a cost to one partner will negatively affect the other partner. For this reason, mutualisms have often been predicted to be evolutionarily unstable (Boucher 1982; Poulin 1995). These theories assume that cheaters have a negative effect on one mutualist by competing with the 76 other mutualist for rewards and by reducmg the services provided. Altematively, some means of punishing the cheater (Mesterton-Gibbons and Dugatkin 1992) or controlling the cheater without excluding mutualists (Pellmyr and Huth 1994, Noe and Harrunerstein 1994; Schwartz and Hoeksema 1998) may evolve. However, a different outcome has been predicted for one kind of cheater, nectar-robbing bees. As cheaters on the plant-poUinator mutualism, nectar-robbers take nectar from a flower by cutting a slit ia the corolla and removing nectar, usually without pollinating. It has been suggested that nectar-robbers may have a positive effect on a plant if they reduce average levels of nectar (Heirmch and Raven 1972; Zimmerman and Cook 1985) or increase variability between patches, because pollinators in search of nectar must carry poUen farther between plants. A pollinator may disperse farther from robbed flowers either because it must visit more flowers to collect the same amotmt of nectar, or because it searches for unrobbed flowers after it has encountered several robbed ones. Longer flight distances by pollinators are likely to result in pollen moving further from close relatives of the plants, thus decreasing inbreeding (Gentry 1978; Inouye 1983). A pollinator encoimtering robbed flowers may also leave plants in search of imrobbed patches on other plants, which would benefit multiple-flowered and self- incompatible plants. In this way, nectar-robbing may indirectly benefit a plant. These hypotheses lead to another: the benefit of robbing should increase with variability between patches, because variability makes it likely that pollinators will search for less robbed patches. The necessity for poUinators to search for nectar depends on amounts of nectar left in robbed 77 flowers and energetic costs of pollinators. However, if the intensity of robbing is even over the site, pollinators should not search for less robbed patches. In this scenario, cheaters affect one mutualist (pollinators) negatively by competing with them for limited rewards (nectar), but benefit the other mutualist (the plant) indirectly by changing pollinators' behavior. In the Sonoran desert in Arizona, the abimdant carpenter bee, Xylocopa califomica, robs nectar from the desert willow, Chilopsis linearis (Bignoniaceae). This system is particularly well-suited to investigate the effects of nectar- robbing for several reasons: 1) the plant is pollen-limited (Appendix B), so pollen delivery by pollinators is important for its reproductive success; and 2) the plant is cornucopia-flowering, (sensu Gentry 1974), producing hundreds of flowers at one time, and self-incompatible (Petersen et al. 1982; Appendix B). Hence, there is a potential conflict between mutualists; it would benefit the plant if pollinators move between plants, but pollinators have the option of remaining at one plant foraging on nectar from numerous flowers. Benefits to the plant are greater if pollinators visit a few flowers and then leave the plant. For this reason, any effect that will cause pollinators to leave the plant with its pollen will potentially increase the male reproductive success of the plant. 3) Finally, Chilopsis does not possess an obvious mechanism to exclude or punish nectar-robbing bees. Gentry (1990) hj^othesized that other members of Bignoniaceae have clustered flowers or stiff calyces in order to protect flowers from nectar-robbers. However, calyces do not prevent robbing in Chilopsis and flowers are not tightly clustered. To determine the effect of the nectar-robbing bees on the pollination mutualism, I addressed four questions: 1) How does the removal of nectar by 78

nectar-robbers affect the behavior of "legitimate" pollinators (those visitors that enter flowers and potentially transfer pollen)? 2) Is the net effect of robbing to the plant costly or beneficial, in terms of male and female components of reproductive success? 3) If nectar-robbing benefits a plant by increasing pollen flow, does that benefit vary with either the intensity or variability of robbing? Finally, 4) If nectar-robbing is costly, does its effect change with the interisity of robbing? One might predict that a low intensity of exploitation may not measurably affect a mutualism, but a greater inter\sity wiU.

Methods

Background:

Chilopsis linearis (desert willow) is a cornucopia-flowering tree (Gentry 1974) that blooms in late May to early June in desert washes in the Southwestern U. S. and Mexico. This tree is foimd growing in washes alongside desert flats containing nests of Xylocopa, carpenter bees. The two sites used in this study were in S. Arizona, Cochise Co., near Portal, Arizona. At "Desert Site" (DS), located 6.9 km N of Portal on the San Simon Road, trees grew along the sides of a wash in the desert (elev. 1387 m) near the Chiricahua Mountains. All experiments were conducted at DS except where noted. The other site, "Cave Creek" (CC), was at slightly higher elevation (1463 m), about nine km from DS. Trees at CC bloomed slightly later than those at DS. 79

During 1-3 weeks, the trees produce hundreds of trumpet-shaped flowers. Trees bloomed at DS from May 20-June 7,1993, May 12- May 13,1994, May 18-June 16,1995, and May 15-May 22,1996. At CC, trees bloomed June 1-June 13,1994, June 13-June 23,1995, and May 21-June 1,1996. At both sites, the most common visitors to Chilopsis were bumblebees (Bombiis sonorus), honeybees {Apis mellifera), and nectar-robbing bees, Xylocopa califomica (Richardson 1995; Appendix A; Appendix B). Ruby-throated hummingbirds {Archilochus colubris) were occasionally seen at DS. Black-chinned {Archilochus alexandri) and broad-tailed huminingbirds (Selasphonis platycercus) occasionally visited flowers at CC. At both sites, little other floral nectar was available until Ipomopsis longiflora bloomed near the end of the flowering season of CC. Nectar in Chilopsis was produced as buds open late in the afternoon, during anthesis. Nectar did not refill (Whitham 1977). The protandrous flowers last approx. 2 days. Stigma lobes open and stigmas become receptive during the morning of the second day (Appendix B) When nectar-robbing bees {Xylocopa califomica) visited Chilopsis flowers, they landed on the base of the flower without touching anthers or stigmas. They slit the corolla with the sharp galeae that covered their tongues and removed nectar through the slit. Other visitors foraged "legitimately" through the entrance of the flower, with the exception of an occasional butterfly that visited slits made by nectar-robbers.

Intensity and Patchiness of Robbing 80

I evaluated changes in the frequency and between-tree variability of robbing over flowering seasons. To do this, I counted the percentage of robbed flowers every 2 days at 10:30 a.m. during 1995 and 1996 at DS. I measured the percentage of robbed flowers in a sample of approx. 100 haphazardly chosen flowers, usually divided among three trees that were judged to have the greatest number of flowers open at one time. At the end of the season at DS, I checked flowers on more trees to complete the survey of 100 flowers. In order to determine the between-tree variability of robbing, I calculated the coefficient of variation (cv) for each day of the survey on which intensity of robbing was determined using only three trees.

Measurement of Nectar

I compared amoimts of nectar remaining in robbed cmd unrobbed flowers by measuring nectar in both types of flowers with a micropipette at several times over the day. Even though the number of visits to each type of flower was unknown, I predicted that robbed flowers would have less nectar than unrobbed flowers measured at the same time of day, because robbed flowers must have been visited at least once, whereas unrobbed flowers could have been imvisited. Nectar in unvisited flowers was measured in order to determine whether nectar was secreted throughout mornings. I sampled nectar volixmes during the first morriing that flowers were open to visits by legitimate visitors (day 1) on six days in 1993,1994 and 1995 at DS. On two of those days, I also sampled nectar from bagged, imvisited flowers in order to compare the amount produced by the plant over the same period of time. It 8 1 was not possible to measure nectar concentration for most flowers because the water in samples evaporated immediately. Because many flowers had no nectar, testing for differences in amounts of nectar between robbed and unrobbed flowers was complicated. First, I compared the proportion of flowers that contained nectar in robbed and unrobbed flowers using a G-test for measurements sampled within 2-hr intervals. Second, for those flowers that contained nectar, I compared the volumes of nectar present in robbed and unrobbed flowers within 1-hr intervals in a nonparametric 2-way Kruskal-Wallis test. I used presence of robbing and time as factors. Finally, I measured nectar volume in imvisited, bagged flowers between 2-hr intervals to determine whether nectar was continually produced in the morning.

Pollinator behavior

From 1993-6 at DS and CC, I determined the behavior of pollinators at robbed flowers. First, because robbed flowers contained less nectar, I tested the hj^othesis that legitimate visitors spend less time in robbed flowers than in unrobbed ones. In the same experiment, I also tested whether these visitors avoided robbed flowers. In 34 half-hour observation periods, I used colored thread on flower pedicels to mark approx. 100-200 robbed and unrobbed flowers in one part of a tree (i.e., a patch). During observations, I stood on a 6' ladder about 2-3 m away. I observed legitimate visitors entering robbed and unrobbed flowers. I compared length of visits to robbed and unrobbed flowers using a Mann-Whitney U-test for each species of visitor. 82

Second, I hypothesized that legitimate visitors would avoid robbed flowers within patches. During observation periods, I counted the proportion of unrobbed flowers that visitors entered and compared that to the proportion available in observed part of the tree. I hypothesized that visitors avoided robbed flowers. I assumed that if visitors avoided robbed flowers, then the proportion of visits made to robbed flowers would be less than expected by chance, i.e., less than the proportion present on the tree. I tested this hypothesis using a log-likelihood ratio goodness of fit test for each species of visitor. For this hypothesis, I focused on honeybees and bumblebees, because they were the most common species of legitimate visitors during the years studied (Richardson 1995; Appendix A; Appendix B). In a third experiment, I investigated further the behavior of honeybees at robbed flowers. After I found that honeybees avoided robbed flowers (see Results), I tested whether they used the presence of slits in flowers as a cue to identify robbed flowers. On eight days over two seasons, I bagged about 100 buds in a patch on a tree and waited two days for all of the flowers to open. In the morning of day 2,1 imbagged the flowers and removed other flowers within the patch. I slit the flowers of one-half of the patch with a razor blade in order to simulate robbing. I touched all the flowers on the other half of the patch, but did not slit them. I did not remove nectar from any flowers. Over 1-hr observation periods on day 2,1 observed the number of flowers on each side of the patch that each honeybee visited. I hypothesized that if honeybees avoided robbed flowers using visual or odor cues from slits, then the proportion of pollinators that visited more slit (robbed) flowers would be less 83 than expected by chance alone. I tested this h5rpothesis using a Wilcoxon signed-rank test with each visitor as an observation. In a fourth experiment, I tested whether visit length was correlated with amount of pollen deposited by bumblebees. At CC, I bagged flowers as buds in order to prevent visitation, then imbagged them two days later when the stigma lobes were open and receptive. I recorded the length of time that bumblebees spent visiting these flowers. After each visit, I collected the stigma and stained it with Alexander's stain (Alexander 1980) in order to count pollen grains on the stigmas. I tested whether there was a correlation between visit length and pollen deposition using a Spearman rank correlation. Finally, in the fifth experiment, I tested one prediction from the hjrpothesis that robbing indirectly benefits a plant because pollinators depart after encoimtering robbed flowers, and thus robbing increases pollen dispersal away from that plant. One prediction of this hypothesis is that the last flower that a pollinator visits before leaving the plant should be a robbed flower more often than expected based on the number available on the plant. To test this prediction, I observed bumblebees foraging on Chilopsis trees at DS and CC during 17 1/2-hr observation periods. During each observation period, I observed departing bumblebees leaving a tree and determined whether the last flower visited was robbed or unrobbed. If bees' behavior were not affected by robbery, the proportion of bees departing from robbed flowers should equal the proportion of robbed flowers available on the plant. I tested this prediction after I foimd that bumblebees did not avoid robbed flowers within patches. I tested this prediction using a paired sign test that compared proportion of bees predicted to leave from robbed flowers with the 84 proportion, of bees observed to depart during each observation period. Thus, each observation period was a data point.

Ejfects of Robbing on Male Reproductive Success

If pollinators travel farther from robbed flowers than from unrobbed flowers in search of patches with more nectar, pollen should disperse farther from robbed than unxobbed flowers. I compared dispersal distances using fluorescent powders that mimic pollen in dispersal. In the mornings of eight days in 1995 and 1996,1 placed different colors of powder on anthers of equal numbers of robbed and unrobbed flowers on one tree (the "source"). I treated between 50-72 flowers each morning, during approx. two hrs. I selected robbed and unrobbed flowers to treat that were at similar heights and positions on branches. The flowers were left open to all visitors. After dusk, I used a black light to look for powder on stigmas and corollas of neighboring Chilopsis trees leading away from the source tree in both directions along the wash. I exhaustively searched every flower on all trees in both directions, stopping in each direction when I found a tree with no powder on any flower. Using this stopping point may have underestimated rare long-distance pollen dispersal. I used a different source tree each time I repeated this experiment. Dispersal of pollen, male reproductive success, was inferred by movement of powder. Powder may disperse shorter distances than pollen, but it does respond similarly to experimental manipulations (Thomson et al. 1986). Thus, it is likely that comparisons of powder movement from 85 unrobbed and robbed flowers will yield the same trends as actual dispersal of pollen.

Effects of Robbing on Female Reproductive Success

In order to investigate the effects of robbing on female reproductive success of the plant, I first examined whether robbed and vinrobbed flowers set fruits at the same rates. At DS and CC, I marked 565 buds over 17 days and returned to the flowers two times a day imtil corollas fell to observe whether they had been robbed. After corolla abscission, I returned weekly to determine whether the fruits had aborted. Because the rate of flower and fruit abortion was very high (96.1 % abortion in 1993 [N = 228], 80 % abortion in 1994 [N = 30]), I also determined the number of pollen tubes that grew in stigmas ia order to compare female reproductive success of robbed and unrobbed flowers at an earlier stage of reproduction. I marked buds over seven days and returned two times a day to determine whether they had been robbed. After the corollas fell, I collected the stigmas and placed them in either FAA or FPA (formalic acid). I followed the methods of Dafni (1992) to stain for pollen tubes. I fixed stigmas in FAA or FPA for 31 hrs., softened in NaOH for seven hours, rinsed in tap water for 15 hrs., and stained with aniline blue for 4.5 hrs. Under a UV microscope, I determined numbers of pollen tubes in each stained, squashed stigma at the junction of the stigma lobes. Using a Marm-Whitney U-test, I compared the number of pollen tubes growing in robbed and unrobbed flowers. 86

Results

Seasonal trends in intensity and patchiness of robbing

Xi/locopa califomica, the nectar-robbers, were common (Appendix B) and were one of the first visitors to arrive at plants each year during 1993-6. The intensity of robbing, calculated as the percentage of robbed flowers, increased quickly, except at CC in 1996 (Fig. 1). Variability among trees, measured by cv of robbing among trees, decreased over the season in 1995 (Fig. 1). Pollinators such as btimblebees arrived either at the same time as the nectar- robbers, or a few days later. As the season progressed, robbers began to rob buds, removing nectar before corollas opened in the late afternoon. Few other plants providing nectar were available at the sites; those with nectar were Ipomopsis longiflora and Cirsium spp. For these reasons, pollinators were not able to avoid robbed flowers by arriving before nectar-robbers or by switching species, because other plants with nectar were not available. However, honeybees in 1995 may have abandoned Chilopsis before the end of its flowering season on Jime 15; though they were rare visitors after June 4, they were commonly seen collecting pollen on nearby Acacia.

Nectar

The proportion of unvisited flowers that contained nectar did not increase over mornings (G2 = 1.25, p < 0.54, N = 36; Table 1); 17% of flowers 87 were empty over the 6-hr period (N = 36). The amount of nectar in unvisited flowers that contained nectar also did not increase (p < 0.25, N = 30; Table 1). The proportion of flowers containing nectar was the same for robbed and unrobbed flowers within 2-hr intervals (Table 1). 43% of unrobbed flowers were empty (N = 223) cind 50% of robbed flowers were empty (N = 16). However, among flowers that contained nectar, robbed flowers held less nectar than did unrobbed flowers (p < 0.04, N = 179).

Pollinator behavior

The two most common legitimate visitors to Chilopsis from 93-96 were honeybees and bumblebees (Appendix B). Because these two visitors accounted for most visits, I focused on their behavior in response to robbing. Honeybees spent less time in robbed flowers than in unrobbed flowers. There was a trend, though not significant, that bumblebees also spent less time in robbed flowers (p < 0.06 level. Fig. 2.) Because bumblebees were one of the most effective visitors in terms of pollen deposition (Appendix B), I tested the hypothesis that time spent visiting flowers was correlated with pollen deposition by bumblebees, which would have indicated that a reduction in time spent by bumblebees in flowers would have inflicted a cost on the plant in terms of reduced pollen deposition. I foimd that there was no correlation between the length of bimiblebees' visits and number of pollen grains they deposited (N = 12; Fig. 3). In addition to spending less time in robbed flowers, pollinators may respond to robbing by avoiding robbed flowers. Honeybees significantly 88

avoided robbed flowers, but bumblebees did not (Fig. 4). The presence of an artificial slit, with its accompanying visual and odor cues, was not sufficient to cause honeybees to avoid flowers (p < 0.61, N = 24). To svimmarize, among flowers with nectar, robbed flowers did contain less than imrobbed flowers at the same time of day. Nevertheless, bumblebees, the more effective pollen-depositing species, did not avoid robbed flowers. I followed up these results by testing whether the last flower that bumblebees visited on trees before they left were robbed flowers. Contrary to my prediction, biunblebees did not leave more often than expected from robbed flowers. In fact, there was a trend for bumblebees to leave plants more often than expected after a visit to an unrobbed flower (paired sign test, N = 17 observation periods, p < 0.33). Contrary to expectations, bumblebees often flew perpendicular to the site, not to another tree, when they departed a tree.

Effect of Robbing on Male Reproductive Success

Nectar-robbing may indirectly benefit the plant by causing bumblebees and other pollinators visiting desert willow to carry pollen from robbed flowers farther away than from unrobbed flowers. On two days, powder from robbed flowers dispersed sigiiificantly farther than powder from unrobbed flowers. On three other days, powder from robbed flowers tended to disperse farther and on two days, powder from unrobbed flowers tended to disperse farther. On two days, powder from robbed flowers reached significantly more recipients, whereas on three days, powder from unrobbed 89

flowers reached significantly more flowers. In addition, powder from robbed or unrobbed flowers tended to reach significantly more flowers on one and two days, respectively (Table 2). These results suggest that the effect of robbing was variable, but on some days benefited the plant through increased pollen dispersal.

Effect of Robbing on Female Reproductive Success

There was no significant difference between unrobbed and robbed in the number of pollen tubes in the styles (Fig. 5). In fact, robbed flowers tended to have more pollen tubes than did unrobbed flowers (p < 0.09, N = 64). Although some of those pollen tubes may originate from incompatible poUen (Appendix B), this indicates that robbing did not reduce female reproductive success.

Discussion

Effects of Nectar-robbers on Mutiialists—Costly or Beneficial?

Mutualisms have been described and modeled as if both partners are closely coupled; as the population or fitness of one partner increases, that of the other partner increases (Keeler 1985; Addicott 1981). Predictions have been made that when an antagorustic species has a negative effect on one mutualist by removing reward, the other mutualist should suffer a decrease in 90 fitness (Addicott 1981; Keeler 1985; BuU and Rice 1991; Poulin 1995). For this reason, nautualisms have often been predicted to be evolutionarily iinstable. However, in the Chilopsis-pollinatoT mutualism, reduction of nectar rewards by nectar-robbers does not appear to be linked to a decrease in fitness for plants; a negative effect on some pollinators did not affect the plant and even occasionally provided an indirect benefit to the plant. Because honeybees were not very effective at depositing pollen grains (Appendix B), any reduction of the lengths of their visits in resporise to robbing may have inflicted minimal cost on the plant. In contrast, bumblebees were one of the most effective pollinators in terms of pollen deposition (Appendix B). Bumblebees tended to spend less time visiting flowers that contained little nectar, but poUen deposition by bumblebees was not correlated with time spent in flowers. Thus, even if biomblebees spend less time in robbed flowers than in unrobbed flowers, the number of pollen grains tliat they deposit on stigmas, i.e. their potential benefit to the plant's female reproductive success, may not have been affected by robbing. Effects of cheaters on plants may not be negative for several reasons: 1) many plants are generalists—they have several potential partners, some more effective at providing a benefit than others. If the mutualists respond differently to nectar-robbing, the effect on the plant can be positive or negative depending on which mutualists are visiting. 2) Attraction of mutualists may not be equivalent to fitness. The effect of robbing may decrease the nectar available to mutualists, but not affect their fitness if their populations are limited by other factors—nest sites, pollen, etc. Nectar- robbing may have a short-term, but not long-term, negative effect on 9 1

pollinator populations. A long-term effect on fitness is what modelers are really modeling. 3) The benefit to plants by pollinators may not increase in step with attention by mutualist pollinators. Even though Chilopsis is dependent on pollinators for reproductive success, the number of pollen grains deposited by at least one poUinator (Bombus) is unrelated to the time that it spends in a flower when flowers have not previously been visited. Thus, a potentially negative effect on the pollinator, i. e. competition, may not translate into a negative effect on the plant. The benefit to the plant of pollinator visitation also depends on the trarislation of pollen grains deposited to seeds and frviits matured. In Chilopsis, there may be a threshold level of pollen grains needed to set a fruit, below which any benefit to the plant is lost (Appendix B). In addition, a level of pollen deposition above the number of ovxiles available may not benefit the plant. The number of grains needed will vary depending on the proportion of compatible pollen grains in a load of pollen. Details of the interactions between all participants matter when determining the effect of a species that removes a reward without providing a reciprocal service. For example, recent studies of the effect of nectar-robbing on plants have foimd that nectar-robbers cire costly, except when nectar- robbers pollinate while they rob (Scott 1993) or when they pollinate flowers at another time (Morris 1996; Higashi 1998). The cost of robbing can be direct, as when robbers damage ovaries (McDade 1980), or indirect as when pollinators abandon robbed plants (Irwin 1998). Nectar-robbers may also have little effect on plants if the frequency of robbing is low (Arizmendi 1996). 92

However, in this study, I found evidence that nectar-robbers can sometimes benefit plants indirectly by increasing male reproductive success. On some days, the pollen substitute dispersed farther or more often away from robbed flowers, which suggests that robbing can benefit a plant by increasing male reproductive success. This benefit may occur in an indirect way; the least effective pollinators, honeybees, avoided robbed flowers within a patch. The common and effective pollinators, bumblebees, did not, perhaps because robbed and unrobbed flowers had the same proportion of nectarless flowers. By changing the visitation behavior of less effective poUinators, such as honeybees, nectar-robbers may indirectly benefit the plant. At the same time, nectar-robbers did not cost plants directly by damaging flowers or indirectiy by decreasing attention from effective pollinators. Robbing may have little effect for several reasons. First, though bumblebees did spend less time in flowers in response to robbing, these bees did not avoid robbed flowers and the time that they spent visiting flowers was not correlated with pollen deposition. Second, robbing may also have little effect on the female reproductive success of the plant because robbers and pollinators share preferences for foraging at certain locations on trees. Any tendency by poUinators to avoid robbed flowers may be overcome by their tendency to choose flowers based on their location. In addition, the nectar that remains in robbed flowers may be sufficient reason for visitation by pollinators. Finally, robbing may sometimes occur after visitation by pollinators, so there would not be an opportimity to affect pollinators' behavior. These results indicate that robbing does not have an effect on female reproductive success in Chilopsis. 93

Factors That Affect the Cost of Robbing

By comparing studies of the effect of nectar-robbing, it appears that there are specific conditions under which nectar-robbing can be beneficial for plants. The effect of nectar-robbers depends on interactions between the breeding system of plants, morphology of flowers and pollinators, phenology of plants, and pollinator behavior. First, the breeding system and floral morphology of Chilopsis interact to make repeated visits to the plant beneficial to the plant and advantageous for pollinators. Chilopsis is pollen-limited and self-incompatible; a plant benefits most when poUinators leave it after a few visits. Because pollen must move between plcints, the amoimt of nectar available to pollinators must be enough so that they visit the plant, but they soon leave and continue foraging at other plants. In the case of Chilopsis, possibly because of the morphology of the flowers, nectar remains in flowers after they are robbed. Chilopsis flowers have "pool" nectar, which is easy to remove, and "groove" nectar, which pollinators do not remove easily. This morphology makes it advantageous for pollinators to return to flowers after legitimate visits (Whitham 1977). In a similar way, the floral morphology and energetics of nectar-robbers may interact to reduce the cost of nectar-robbing, because nectar-robbers leave nectar that pollinators can remove later. The importance of nectar removal by robbing depends on whether poUinators choose patches based on the proportion of flowers containing nectar or whether pollinators act based on the level of nectar in each individual flower. 94

The interaction between floral and poUinator morphologies also affects whether nectar-robbing is costly. Besides affecting whether nectar remains in flowers, floral morphology also affects the abUity of a nectar-robber to damage ovaries. In the case of Chilopsis, nectar-robbing Xylocopa slit corollas with sharp mouthparts, but does not destroy corollas as birds' beaks do. In addition, Chilopsis, has short, thick calyces that do not prevent robbing, but may keep robbers from damaging ovaries. So, nectar-robbing by Xylocopa might generally be less directly costly to plants than nectar-robbing by birds. Another factor, phenology, affects the outcome of interactions between mutualists and nectar-robbers. In the case of desert willow, nectar-robbers were among the first visitors to arrive at plants dxiring the season. Because they obtained nectar from buds, they often preceded legitimate visitors at flowers. Also, there were few other nectar-producing plants in the area. For these reasons, pollinators could not avoid visiting robbed plants. Hence, robbing could not cost the plant by driving away pollinators from the site. Pollinators may avoid individual trees that have a greater proportion of robbed flowers than others at the site. However, the frequency of robbed flowers became more evenly distributed over the site as it increased. Finally, pollinator behavior significantly affects the outcome of nectar robbing. The response of pollinators to patterns of nectar and cues left by nectar-robbers, and their "rules" of movement within plants and within the site all affect whether nectar-robbing is costly. The removal of nectar from robbed flowers may reduce the time that pollinators spend in flowers. In ttirn, a reduction in the time that a poUinator spends in a flower may be costly to the plant by reducing the number of pollen grains that a pollinator deposits 95 on a stigma. In this way, robbing could indirectly reduce the female reproductive fitness of a plant. However, I did not find evidence for this explanation in this study. Different pollinators may have different energetic requirements that affect their choices of flowers to visit (Richardson 1995; Appendix A) The interaction of pollinator behavior and the abimdance of different pollinators may affect the outcome of nectar-robbing. Different poUinators may have different "rules" that determine when they avoid flowers. Because honeybees avoided robbed flowers and bumblebees did not, honeybees may have been more influenced by the mean volume of flowers than by the proportion of flowers containing nectar. What signal honeybees used to identify robbed flowers in a patch is unclear. I h5^othesized that they would use either the visual cue of the sUt or the odor coming from the damage to the plant. However, honeybees may instead detect odors that nectar-robbers leave on robbed flowers. Xx/locopa virginica deposits scent on flowers while visiting legitimately (Frankie and Vinson 1977). Honeybees can sense conspecific odor, but it has not been shown that honeybees respond to scents deposited by other species. Bumblebees of other species have been found to sense odors deposited by conspecifics (Goulson 1998). However, on Chilopsis, bumblebees apparently are not influenced by odors deposited by nectar- robbers; whether they can detect them is unknown. Because pollinators differ in their responses to robbed flowers, the exact mix of pollinating species affects the outcome of robbing. Another aspect of "rules" of movement by pollinators may affect the outcome of robbing. Bombiis choose to leave patches of artificial flowers 96 based on the average volume of the last three flowers (Dukas and Real 1993). For this reason, I expected that bumblebees would leave trees more often after they visited robbed flowers than after they visited unrobbed flowers, because robbed flowers contain less nectar. However, if there was a trend, it was for bumblebees to leave plants more often after they visited imrobbed flowers. When bumblebees left trees, they often flew perpendicular to the linear axis of the wash in which the trees grew. A possible explanation for this behavior is that bumblebees fill up on nectar and leave for their nests more often from unrobbed flowers. This behavior may be costly to a plant, because pollen that is carried to a bee's nest usually is removed from circulation. In this way, being robbed may benefit a plant not by causing pollinators to leave a patch, but by causing pollinators to continue foraging. Bumblebees that encoimtered enough nectar to continue foraging, but not enough to stop foraging, should be more beneficial to the plant than bumblebees that left the site. In this way, nectar-robbing may indirectly benefit a plant, because pollen from imrobbed flowers may be more often removed from circulation when bumblebees return to their nests.

Effects of Cheaters, Ecological Parasites on Mutualisms

A comparison of two other examples of cheaters with nectar-robbers illustrates differences between nectar-robbers and other types of cheaters. In one case, "cheating" yucca-moths oviposit without pollen transfer. Two types of cheater moths exist in the obligate mutualism between yucca moths and yuccas; one facultative type of cheater appears to be individuals of the 97 pollinating species that do not tramfer pollen (Tyre and Addicott 1993). These moths are costly to plants, but are limited when yxicca plants abort fruits that have not been sufficiently pollinated (Pellmyr 1997; Wilson and Addicott 1998). Obligate cheating yucca moths that have evolved from pollinating ancestors also oviposit into fruits without pollinating (Pellmyr et al. 1996), but too late for plants to abort. These moths are costly and have existed for a long time evolutionarily (Pellmyr et al. 1996), but limitations to their population are unclear. In this example, mutualists are specialists and facultative cheaters are limited by the mutualist that they visit. In a second example, cheaters have been found in an obligate mutualism between Pheidole ants and Piper plants, which exchange defense against herbivory for food bodies and domatia. The cheaters are Tarsobaenus beetles that invade Piper plants, killing ants while causing plants to continue producing food bodies (Letoumeau 1991). The cheaters are very costly to plants; plants without ants are attacked by stem borers (Letoumeau 1998). These beetles are limited by competition with Pheidole ants; the ants remove beetle eggs from plants and kill invading beetle larvae. In contrast to these examples, nectar-robbers are not as costly to the mutualists that they visit, and may occasionally benefit them by the behavior of the other mutualists, pollinators. Effects of nectar-robbers on plants vary partly because the mutualists are generalists; different combinations of mutualists with different foraging behavior may be present. Finally, there is little evidence that nectar-robbers are limited by either mutualist. Pollinators do not actively defend flowers against robbers and flowers are not protected by morphological adaptations against cheating. Though they are not limited 98 by mutualists, the abundance of nectar-robbers may be limited by lack of nest sites (Minckley 1987). If their abundance were not limited, their cost to the mutualists might be greater. 99

uopi.BUBy\^ JO ;uapTjjao3

SJ0AiOij paqqo^ o/^

Fig. 1. Percentage of robbed flowers and coefficient of variation of three trees (N = 100 for each date). A. 1995, DS. B. 1996, DS. C. 1995, CC. D. 1996, CC. % Robbed Flowers K) CJl Ol o _1_ _1_

tt- Dd

v«0 ON % I o \ C/5 /c

"T" "T" I—' Un O Ol Coefficient of Variation % Robbed Flowers NJ CTN 00 O O O O 4 -I \ I n

\ VO \o Ol \ I

\ /o

7

G> % Robbed Flowers hO ON % \ %

\ \ TABLE 1. Proportions of flowers containing nectar and mean volume in flowers with nectar for unrobbed, robbed, and unvisited flowers. Proportions were compared using G-tests for each 2- hr interval. Mean mL nectar vol. in flowers Proportion of flowers containing nectar (n) containing nectar, ± s.d. (n) Time of unrobbed robbed unvisited P unrobbed robbed unvisited Day (unrobbed vs. robbed) 0500-0700 .76 (59) .75 (4) .83 (12) .95 1.9 ±2.1 0.7 ± 0.2 4.9 ± 3.0 (39) (3) (10) 0700-0900 .57 (102) .50 (8) .75 (12) .71 2.0 ± 3.2 1.0 ±1.7 7.2 ± 4.5 (58) (4) (9) 0900-1100 .40 (62) .25 (4) .92 (12) .53 0.7 ± 1.4 0.7 (1) 6.0 ± 5.8 (25) (11) 1 1 1 o m m Lo o cvi (Das) q:^Sua-[ ;tsta

Fig. 2. Comparison of lengths of time that Apis and Bombus spend visiting robbed and imrobbed flowers (medians ± SE). 105

surej^ ua^oj;

Fig. 3. Numbers of pollen grains deposited by Bombus visiting Chilopsis flowers compared with time that each bee spent in a flower. Flowers were not previously" visited. 106

u Cu

-a TJ

C/D PH

m o o

sio^TsiA JO uopiodoJcE

Fig. 4. Proportion of Apis and Bombus visiting more and fewer robbed flowers than expected during a foraging bout, based on the ratio of robbed flowers available in the observation area at that time. The null hypothesis is that number of visitors visiting more robbed flowers than expected in a bout = number of visitors visiting fewer robbed flowers = 0.50. Table 2. Distance and numbers of flowers reached by powder that mimics pollen. Different colors of powder were used to mark anthers of equal numbers of robbed and unrobbed source flowers. Distances that powder traveled were compared using Mann-Whitney U-tests for each date. Numbers of flowers that powder reached were compared by G-tests for each date. Significant differences are indicated by asterisks. Does dye Does powder from robbed from robbed flowers Maximum flowers travel to Distance Most travel Level of more Level of from Source Intensity of Common Date farther? Significance flowers? Significance (m) Robbing (%) Pollinators 5-21-95 NR > R p < 0.84 NR > R'*' P ^ 10.9 ~5 Z = -21 G = 3.40 Anthophom, few Apis

5-31-95 R>NR p<0.89 NR > R p < 0.0002 25.8 -50 Anthophom, Z = 0.14 G = 13.7 Apis

6-4-95 R>NR^ p < 0.02 R > NR''" P ^ 35.4 62 Anthophom, Z = -2.28 G = 3.68 Apis

6-8-95 R>NR p < 0.81 R > NR'*" P ^ 0.0001 42.1 73 Megachile, Z = -0.23 G = 52.4 Anthophom

6-12-95 R>NR'^ p < 0.0036 NR > R''" P ^ 49.1 70 Bombus, Z = 2.92 G = 3.65 Megnchile

5-17-96 NR>R p < 0.08 NR > R p < 0.13 25.4 ~6 Bombus Z = -1.77 G = 2.31 p = 0 4.71 ~ 44 Bombus, 5-20-96 no dye found NR>R''" G = 0 Megcichile from R p< 0.076 13.4 - 43 Bombus 5-23-96 R>NR p<0.14 R>NR Z = -1.46 G = 3.13 109

fC 80

C/!)

Unrobbed Robbed

Fig. 5. Ntimbers of pollen tubes in stained stigmas of unrobbed and imrobbed flowers (means ± SE). Analyzed using Mann-Whitney U-test, p < 0.09; N = 41, 1 10

Literature Cited

Addicott, J. F. 1981. Stability properties of 2-species models of mutualism: simulation studies. Oecologia 49:42-49.

Addicott, J. F. 1986. Variation in the costs and benefits of mutualism: the interaction between yuccas and jmcca moths. Oecologia 70:486-494.

Alexander, M. P. 1980. A versitile stain for pollen, fungi, yeast, and bacteria. Stain Tech 55:13-18.

Arizmendi, M. C., C. A. Dominguez, and Dirzo, R.. 1996. The role of an avian nectar robber and of hummingbird pollinators in the reproduction of two plant species. Functional Ecology 10:119-127.

Bao, T. and J. F. Addicott. 1998. Cheating in mutualism: defection of Yucca baccata against its yucca moths. Ecology Letters 1:155-159.

Boucher, D. H., S. James, and K. H. Keeler. 1982. The ecology of mutualism. Aimual Review of Ecology and Systematics. 13:315-47.

Bronstein, J. L. 1994. Conditional outcomes in mutualistic interactions. Trends in Ecology and Evolution. 9:214-217.

Bronstein, J. L. cind Y. Ziv. 1997. Costs of two non-mutualistic species in a 5nacca/yxicca moth mutualism. Oecologia 112:379-385.

Bull, J. J. and W. R. Rice. 1991. Distinguishing mechanisms for the evolution of cooperation. Journal of Theoretical Biology 149:63-74.

Dafni, A. 1992. Pollination Ecology. New York, Oxford University Press.

Dukas, R. and L. A. Real. 1993. Effects of recent experience on foraging decisions by bumble bees. Oecologia 94:244-246.

Frankie, G. W. and S. B. Vinson. 1977. Scent marking of passion flowers in Texas by females of Xylocopa virginica texana (Hymenoptera: Anthophoridae). Journal of the Kansas Entomological Society 50:613- 625.

Gentry, A. H. 1974. Coevolutionary patterns in Central American Bignoniaceae. Annals of the Missouri Botanical Gardens 61:728-759.

Gentry, A. H. 1978. Anti-pollinators for mass flowering plants? Biotropica 10:68-69. 111

Gentry, A. H. 1990. Evolutionary patterns in neotropical Bignoniaceae. Memoirs of the New York Botanical Garden. 55:118-129.

Goulson, H., S. A. Hawson, and J. C. Stout. 1998. Foraging bumblebees avoid flowers already visited by conspecifics or other bumblebee species. Animal Behavior 55:199-206.

Heinrich, B. and P. H. Raven. 1972. Energetics and pollination ecology. Science 187:597-602.

Higashi, S., M. Ohara, H. Arai, and K. Matsuo. 1988. Robber-like pollinators: overwintered queen bumblebees foraging on Corydalis ambigua. Ecological Entomology 13:411-418.

Inouye, D. W. 1983. The ecology of nectar robbing. Pages 153-173 in B. Bentley and T. Elias, eds. The Biology of Nectaries. Columbia University Press, New York.

Irwin, R. E. and A. K. Brody. 1998. Nectar robbing in Ipomopsis aggregata: effects on pollinator behavior and plant fitness. Oecologia 116:519-527.

Keeler, D. 1985. Cost:benefit models of mutualism. Pages 100-127 in D. H. Boucher, ed. The Biology of Mutualism. Croom Helm, London.

Letoumeau, D. K. 1991. Parasitism of ant-plant mutualisms and the novel case of Piper. Pages 390-396 in C. R. Huxley and D. F. Cutler, eds. Ant-Plant Interactions. Oxford University Press, Oxford.

Letoumeau, D. K. 1998. Ants, stem-borers, and fungal pathogens: experimental tests of a fitness advantage in Piper emt-plants. Ecology 79:593-603.

McDade, L. A. and S. Kinsman. 1980. The impact of florcil parasitism in two neotropical hummingbird-pollinated plant species. Evolution 34:944- 958.

Mesterton-Gibbons, M. and L. A. Dugatkin. 1992. Cooperation among unrelated individuals: evolutionary factors. The Quarterly Review of Biology 67:267-281.

Minckley, R. L. 1987. Aspects of the Reproductive Biology of Two Carpenter Bees (Genus Xylocopa) in Southern Arizona. Department of Entomology. Tucson, University of Arizona: 90. 1 12

Morris, W. F. 1996. Mutualism denied? Nectar-robbing bumble bees do not reduce female or male success of bluebells. Ecology 77:1451-1462.

Noe, R. and P. Hammerstein. 1994. Biological markets: supply and demand determine the effect of partner choice in cooperation, mutualism, and mating. Behavioral Ecology and Sociobiology 35:1-11.

Omelas, J. F. 1994. Serrate tomia: an adaptation for nectar robbing in hummingbirds. Auk 111:703-710.

Pellmyr, 0.1997. Stability of plant-ariimal mutualisms: keeping the benefactors at bay. Trends in Plant Science 2:408-409.

PeUmyr, O. and C. J. Huth. 1994. Evolutionary stability of mutualism between yuccas and yucca moths. Nature 372:257-260.

Pellmyr, O., J. LeebensMack, and C. J. Huth. 1996. Non-mutualistic yucca moths and their evolutionary consequences. Nature 380:155-156.

Petersen, C., J. H. Brown, and A. Kodric-Brown. 1982. An experimental study of floral display and fruit set in Chilopsis linearis (Bignoniaceae). Oecologia 55:7-11.

Poulin, R. and W. L. Vickery. 1995. Cleaning symbiosis as an evolutionary game: to cheat or not to cheat? Journal of Theoretical Biology 175:63-70.

Richardson, S. C. 1995. Temporal variation in poUinator abimdance and pollinator foraging behavior in response to robbed flowers. Pages 311- 316 in L. F. DeBano, P. F. FfoUiott, A. Ortega-Rubio, G. J. Gottfried, R. H. Hamre, and C. B. Edmirister, eds. Biodiversity and Management of the Madrean Archipelago: the Sky Islands of Southwestern United States and Northwestern Mexico. U.S. Department of Agriculture, Forest Service, Fort CoUins, CO.

Schwartz, M. W. and J. D. Hoeksema. 1998. Specialization and resource trade: biological markets as a model of mutualisms. Ecology 79:1029-1038.

Scott, P. E., S. Buchmann, and M. K. O'Rourke. 1993. Evidence for a mutualism between a flower-piercing carpenter bee and ocotillo: use of pollen and nectar by nesting bees. Ecological Entomology 18:234-240.

Soberon Mainero, J. and C. M. del Rio. 1985. Cheating and taking advantage in mutualistic associations. Pages 192-216 in D. H. Boucher, ed. The Biology of Mutualism. Oxford University Press, New York. 113

Templeton, A. R. and L. E. Gilbert. 1985. Population genetics and the coevolution of mutualism. Pages 128-144 in D. H. Boucher, ed. The Biology of Mutualism. Croom Helm, London.

Thomson, J. D., M. V. Price, N. M. Waser, and D. A. Stratton. 1986. Comparative studies of pollen and fluorescent dye transport by bumblebees visiting Erythronium grandiflonim. Oecologia 69:561-566.

Trivers, R. L. 1971. The evolution of reciprocal altruism. The Quarterly Review of Biology 46:35-57.

Tyre, A. J. and J. F. Addicott. 1993. Facultative non-mutualistic behaviour by an obligate mutualist: cheating by yucca moths. Oecologia 94:173-175.

Whitham, T. G. 1977. Coevolution of foraging in Bombus and nectar disperising in Chilopsis: a last dreg theory. Science 197:593-596.

Wilson, R. D. and J. F. Addicott. 1998. Regulation of mutualism between joiccas and yucca moths: is oviposition behavior resporisive to selective abscission of flowers? Oikos 81:109-118.

Zimmerman, M. and S. Cook. 1985. Pollinator foraging, experimental nectar- robbing and plant fitness in linpatiens capensis. The /^erican Midland Naturalist 113:84-91. 114

APPENDIX D

Title: Were Diet Breadth and Nectar-Robbing Factors in the Evolution of Sociality in Bees?

Abstract: Many factors may contribute to the evolution of sociality in Hymenoptera (ants, bees, and wasps). Most factors involve genetics or behavior in nests, including haplodiploidy, kin selection, and cooperative defense against predators and parasites. However, the role of foraging behavior in the evolution of sociality has not been investigated. Bees are useful for this investigation because they vary from solitary to eusocial; in addition, degrees of sociality and foraging behavior have been investigated in enough depth to make a phylogenetic comparison possible. Over a phylogeny of bees made from four published phylogenetic analyses, I found that sociality has evolved at least 11 times and up to 14 times, in the families Apidae and Halictidae. Eusociality, the most complex form of sociality, has originated at least 4 to 5 times in bees. I hypothesized that the evolution of sociality in bees may only be possible after the evolution of a broad diet breadth, because use of a diversity of resources would be needed to support multiple generations of bees over a season. I hypothesized that a novel foraging behavior, nectar-robbing, allowed bees to forage on plants that previously excluded them from nectaries with tubular corollas. For these reasons, I predicted that sociality, diet breadth, and nectar-robbing would be correlated in bees. Computer simulations showed that sociality and nectar- robbing were correlated on the phylogeny of bees. I also compared diet 1 15 breadth of bee genera within three geographic regioris: the Eastern United States, California, and Costa Rica. Diet breadth varied as a function of maximal social development among bee genera. I used phylogeneticaUy independent contrasts for taxa within each region to test for correlations between sociality, diet breadth, and nectar-robbing. In some cases, a broad diet breadth was associated with sociality and nectar-robbing. key words: sociality, generalization, nectar-robbiag, diet breadth, host plant 1 16

Introduction

The evolutionary factors that led to existence of social insects with sterile castes mystified Darwin (1859). There may be several factors that were necessary for the evolution of eusociality, traditionally defined to include cooperative brood care, presence of generation overlap, and reproductive division of labor (Michener 1974,1969; Crespi 1996). Hamilton (1964) proposed that kin selection explained workers' apparent altniism in caring for sisters. Through inclusive fitness, haplodiploidy was understood to be a critical factor in the evolution of sociality in Hymenoptera (ants, bees, and wasps). However, though haplodiploidy may have been a prerequisite for the evolution of sociality in Hymenoptera, it is not sufficient to explain the origin of sociality. Though all Hymenoptera are haplodiploid, only a few families out of more than 80 are eusocial; sociality apparently had multiple independent origins in Hymenoptera and required more preadaptations than just haplodiploidy (Hunt 1999). The evolutionary pathway from soUtary behavior to sociality, of which eusociality is the most complex level (Table 1), has attracted intense interest because it may aid in explaining factors that led to the evolution of sociality (reviewed in Michener 2000). In addition to haplodiploidy, factors that may favor the origin of sociality in Hymenoptera include nest permanence, cooperative defense against parasites or predators, long-lived reproductive females, and biased sex ratios (reviewed in Crozier and Pamilo 1996; Crespi 1996; Schwartz et al. 1997; Thome 1997). These factors concern either the genetics of insects or their 1 17 behavior at nests. However, generalization in resource use, i.e., having a broad diet breadth, and foraging behavior may also be important preadaptations to the evolution of sociality, though these factors have not been investigated. Generational overlap, by definition, is a key condition for existence of complex levels of sociality (Table 1). It occurs in all subsocial and eusocial Hymenoptera and in some parasocial bees (Danforth and Eickwort 1997). Generational overlap may occur through either multivoltinism (multiple broods per year), long lifespan of females that initiate nests (Schwartz et al. 1997), or both. For either multivoltirvism or longevity of females to have evolved, activity of bees over long seasons must have been a prerequisite. Thus, pre-social bees must have had a broad diet breadth because the length of blooming season of plant taxa would not have been long enough to support social bees. Hypotheses about the conditions necessary for sociality to evolve are typically tested using ecologiceil studies of variable species (reviewed in Crespi 1996, though see Himt 1999 for vespid wasps); however, phylogenetic methods are also valuable, because they can reveal the order of historical events. Because bees vary from solitary to eusocial, they are especially useful for understanding conditions that favor the evolution of sociality. In contrast, ants are almost completely eusocial (Crespi 1996); in wasps, sociality is less well-studied than it is in bees, so phylogenetic analysis for wasps may be premature. The evolution of sociality over a phylogeny representing all bees has not been investigated using modem phylogenetic methods, although origins 1 18

and reversals of sociality have been reconstructed for one tribe, the Augochlorini (Danforth and Eickwort 1997), and number of origins over cdl bees have been predicted (Michener 1969). Thus, the niimber of origins and ciny preadaptations associated with these origins remain unknown. In addition, though it is known that sociality tends to occur in bee families Apidae and Halictidae (Michener 1974,2000), no recent review of presence of sociality across bee genera has been conducted. I investigated three hypotheses: 1) that a broad diet breadth was associated with the evolution of sociality in bees, 2) that a novel foraging behavior, nectar-robbing, was associated with the evolution of a broader diet breadth, and 3) that nectar-robbing also was associated with the evolution of sociality. Nectar-robbing bees make slits or holes in corollas of flowers and remove nectar without entering coroUas. In a comparison of nectar-robbing bumblebees with non-nectar-robbing bumblebees at a site in Colorado, nectar-robbing bumblebees foraged on more species of plants than did non- robbers (Inouye 1983). The nectar-robbing bees visited plants that were inaccessible to other species of bees, such as hummingbird-poUinated flowers of Ipomopsis with narrowly tubular morphologies, in addition to normally bee- poUinated flowers with less narrow corollas. For this reason, I predicted that the evolution of nectar-robbing in bees was associated with a broad diet breadth. If nectar-robbing were correlated with generalized plcint use by social bees, it may have facilitated the evolution of sociality. To test these hjrpotheses, I used a phylogeny constructed from four published phylogenies of bees (Roig-Alsina and Michener 1993, Alexander and Michener 1995, Danforth and Eickwort 1997, Pesenko 1999). I surveyed these 1 19

89 genera for degree of sociality (mciximal social development) and presence of nectar-robbing (Appendix E). I mapped these traits onto the combined phylogeny in order to determine degree of correlation. To test the hypothesis that a broad diet breadth was associated with the evolution of sociality, I compared diet breadth of bees within three geographical regioris: the Eastern United States, Guanacaste Province of Costa Rica, and California, (Mitchell 1960 and 1962; Heithaus 1979a; and Moldenke and Neff 1974, respectively). For 345 species in the Eastern U.S., 135 species in Costa Rica, and 1163 species in California, I determined the number of plants visited by all species in each genus, i.e., the diet breadth of the genus. I tested the hj^otheses that diet breadth was a function of maximal social development within each genus and of geographical location. I used partial correlations of phylogenetically independent contrasts to determine whether correlations of diet breadth with maximal social development and with presence of nectar-robbing were significant. I used the combined phylogeny to determine the minimum number of origins of sociality in bees, the presence of reversals to solitary behavior, and whether the pathway to eusociaUty led from parasociality, subsociality, or just solitary behavior. To determine the order of each character, I determined the number of times that each preceded the other.

Methods

Evolution of Sociality 120

I determined the minimxim number of gains and losses of sociality by reconstructing character evolution using four combined phylogenies, one of "long-tongued (L-T) bees" and the other of "short-tongued (S-T) bees". To reconstruct the portion of a phylogeny representing L-T bees, I used the L-T bees' phylogeny from "Analysis D". This analysis was made using Hennig86 1.5 (Farris 1988) to analyze 126 adiilt morphological characters for exemplars from 54 genera (Roig-Alsina and Michener 1993). I removed cleptoparasitic genera because they cannot be social and because they gain resources by taking from other bees, and thus visit few plants for resources (Wcislo and Cane 1996). I also removed genera for which there is no information on nesting behavior (i.e. the level of sociality). Thus, relationships among 41 L-T bee genera were described in my combined phylogeny. The S-T bee phylogeny was constructed from ca. 109 morphological characters and includes adult S-T bees representing 48 genera (Alexander and Michener 1995). With the addition of the L-T bee phylogeny, the clade represented in this tree is monophyletic (Alexander and Michener 1995). The phylogenetic position of families differed significantly among Alexander and Michener's analyses, depending on the coding of just a few characters based on bee glossas (tongues). One analysis suggested that Halictidae was basal, whereas another analysis suggested that CoUetidae was basal. I used two different consensus trees for 36 S-T taxa in my analyses to determine whether the phylogenetic position of bee families affected outcomes of the analyses. For the former, I used Alexander and Michener's strict consensus tree for an island of 336 trees foxmd by NONA (Goloboff 1993) with equal weighting for 1 2 1

all characters. For the latter, I used Alexander and Michener's strict consensus tree for four trees from, analysis using impHed weights in PAUP Version 3.1.1 (Swofford 1993). In addition, I used two phylogenies to add five genera and clarify relationships among genera within the family Halictidae, which is within S-T bees (Danforth and Eickwort 1997; Pesenko 1999). To the phylogeny made from the above four phylogerues, I added three genera of nectar-robbing bees, Perdita, Trigona and Lasioglossiim, using taxonomic positions of genera from classification by Michener et al. 1994. The combined phylogenies encompass a monophyletic group composed of 146 genera, a broad sample of the ca. 600 genera of bees in the world (Michener 1979). With additions and deletions of taxa, I included 89 genera in my analysis. To determine the level of sociality of bee genera, I surveyed nesting behavior from journal articles published 1974-1999 and Michener (1974,1969) for pre-1974 studies (see Appendix E). Several species, particularly in Lasioglossiim and other genera in the family Halictidae, vary in the degree of sociality expressed depending on envirorunental factors such as length of season (Eickwort 1975; Eickwort et al. 1996), availability of resources, or permanence of nesting material (Schwartz et al. 1997). Thus, level of sociality should be viewed as the maximal social development that is recorded at present in a genus, given appropriate conditions. In addition, socicdity can be not only gained but lost within a genus (e.g. in Lasioglossiim [Wcislo and Danforth 1997] and Xi/locopa [Minckley 1987]). Thus, the numbers of gains and losses of sociality found in this study are minimum estimates. I categorized genera based on the most complex form of sociality attained by species in 122

categories defined by Michener (1974,2000) (Table 1). Though there are other defirutions of degrees of sodaKty (Sherman 1995), descriptions of nesting behavior of bees in particular usually follow Michener's definitions.

Correlation of sociality with nectar-robbing

I found recorded cases of nectar-robbing by bees (see Appendix E). Nectar-robbers are described as "primary robbers" when they make slits in corollas themselves and remove nectar without entering flowers. Nectar- robbers are described as "secondary robbers" when they visit flowers that have been cut by other species but do not cut flowers themselves (Inouye 1980). I hypothesized that primary robbers would have a broader diet breadth than secondary robbers, because primary robbers were not dependent on other floral visitors to cut flowers. I tested whether primary and secondary robbers differed in diet breadth using 2-way ANOVA on ranked, square-root transformed median or maximum diet breadth, with level of sociality and nectar-robbing as factors. For aU other tests of correlation of nectar-robbing with sociality and with host plant use, I combined primary and secondary robbers into a single category. I tested the hypothesis that maximal social development is correlated with nectar-robbing using MacClade's concentrated changes test (Maddison and Maddison 1992). The concentrated changes test determines whether the dependent variable, represented by parasociality or eusociality (Table 1), occurs more often by chance on branches that have changes in the independent variable, represented by nectar-robbing. Simulations scatter 123 changes randomly on the phylogeny and count the fraction of replicates in which gains and losses of sociality are located on branches in. which nectar- robbing has evolved. Because the concentrated-changes test cannot be used when there are polytomies, I resolved L-T polytomies based on Roig-Alsina and Michener's (1993) "Analysis A", which included cleptoparasitic taxa. Otherwise, I resolved polytomies based on classification of genera in Michener et al. 1994. Resolving polytomies had no effect on numbers of gains and losses in sociality or nectar-robbing, except that moving Megachile to be a sister-group to the Hoplitis-Trachiisa clade based on "Analysis A" reduced the number of gains in parasociality and increased the number of gains in nectar- robbing (Fig. 1-2). For the concentrated changes test, I simulated 10000 trees with 9 or 11 gains and either 0 or 2 losses of nectar-robbing using the option MINSTATE to account for possible bias in parsimony reconstruction.

Correlation of sociality zuith diet breadth

Diet breadth, as measured by number of host plants visited by species, may vary geographically because different resources are available in different regions. Thus, I compared visitation records for bee genera within three regions in which there were large collection records of bees on host plants: the Eastern U. S. (Mitchell 1962 and 1960, which included collection records from Robertson 1929), Costa Rica (Heithaus 1979a), and California (Moldenke and Neff 1974). Specimens representing approximately 1600 species of bees in the Eastern U. S. were collected over 38 years (Mitchell 1962 and 1960). For 124 each region, I compiled data ftrom those genera which were present in the combined phylogeny. From the Eastern U. S., I compiled data from 345 species in 24 genera. I removed Megachile from analyses because it was an outlier: the diet breath of the genus appeared to be artificially inflated because Mitchell focused on collecting Megachile. I also removed data on species that Mitchell described as "rare", species for which there were no host records, species that had distributions limited to southern Florida, and species that were known to be recent immigrants to the U. S.. Specimens of 330 species of bees in Guanacaste Province, Costa Rica were collected from three locations in tropical dry forest over one year (Heithaus 1979a); I compiled data on 135 species in 27 genera. Specimens representing 2,500 species of bees in California were collected over eight years along a transect crossing California, covering all habitats (Molderike and Neff 1974); I compiled data from 1163 species in 28 genera. For bees from California, I removed from analyses species that were described as "rare" by Moldkenke cmd Neff (1974) and those for which hosts were unknown. For each region, I determined the number of host plant species that each bee species was recorded to have visited and then found the median and maximum number visited by each bee genus. To be included in this analysis, a bee genus must have been present in one of these regions and in one of the phylogenetic analyses. Because independent contrasts cannot use the entire distribution of diet breadth for each genus, I used both medians and maxima in analyses in order to determine whether the choice of summary statistic for genera affected the outcome of analyses. The three studies did not differentiate between visitation to plants for nectar and visitation for pollen or 125 resins. However, I assumed that poUen and resin specialization did not contribute much to diet breadth because bees visit a wide range of plants for nectar (Moldenke 1976), even when the bees specialize on other resources such as poUen, resin, or oil (Wcislo and Cane 1996). To take into account correlations between genera because of shared history, for each region I reconstructed the phylogeny of bees using only the genera occirrring in each region. I used two methods to test the hypothesis that more complex social development was correlated with broader diet breadth across bee genera. First, I used ANOVAs to test the hypotheses that median and maximum diet breadth were a function of maximal social development within each genus and of geographical location. I compared effects without including phylogenetic relationships of genera. Second, I tested for a correlation between these factors using independent contrasts to remove possible correlations between genera from shared phylogenetic history. I used COMPARE 4.3 (Martins 1999) to compute sets of normally distributed phylogenetically independent contrasts (Felsenstein 1985). To compute contrasts, for each geographical region I reduced the four combined bee phylogenies to a phylogeny that only included the genera that were present in the region. I then used the contrasts in a partial correlation to determine whether the correlations among level of sociality, nectar-robbing, diet breadth cind number of species within each genus were significant when phylogenetic relationships among genera were considered. I conducted the partial correlations using both phylogenetic arrangements of S-T bee families described above. Including the effect of 126 phylogeny made this test for association between factors more conservative than the ANOVA.

Results

The combined phylogenetic reconstruction suggests in bees as a whole, sociality may have originated between 11 and 14 times. Sociality, including parasociality, subsociality, and eusociality originated in L-T bees at least 7 times with 2 losses, and up to 9 times with 0 losses (Figs 1-2). The estimate of number of gains and losses of sociality in S-T bees was not affected by the placement of the families within the clade; it originated at least 4 times with 1 loss, up to 5 times with no losses. Eusocialit}'" may have evolved a minimum of 4 to 5 times in bees; once in the Habropoda/Partamona-Bombus clade (Apidae), 1 to 2 times in the Xylocopa- Braunsapis clade (Apidae), and twice in the Lasioglossum-Augochlora clade (Halictidae) (Fig. 1-2). Most of these genera vary intraspecifically in social behavior from solitary to eusocial (Appendix E). Only within Apidae are there genera that have only eusocial species: Apfs, Bombus, Trigona, Melipona, Paratrigona, and Partomona. Genera in Halictidae that have the potential to be eusocial also contain semisocial species. Genera in Apidae that have the potential to be eusocial also contain subsocial, quasisocial, and semisocial members. In Apidae and Halictidae, genera that share ancestors with eusocial genera are solitary, not social. Because all members of these genera are eusocial and eusociality in these families did not evolve from aggregating or commimal behavior, the evidence suggests that the cincestors of eusocial bees 127 did not pass through each level of complexity of social development from solitary to eusodal behavior. It appears that nectar-robbing evolved at least 9 to 11 times and was lost from 0 to 2 times (Figs. 3-4). On the reconstructed phylogeny of L-T bees, nectar-robbing behavior originated either 4 times with 3 losses, or was gained 6 times. In S-T bees, nectar-robbing either originated 4 times or 3 times with one loss. Hypothesized numbers of gains and losses were not affected by the placement of S-T genera. Primary robbing may have preceded secondary robbing once, in the Melipona/Bombus clade, depending on whether the ancestor of that clade was not a robber or was a primary robber. Secondary robbing may have preceded primary robbing in the Augochloropsis/Augochlora clade. Both primary and secondary robbing appear to have originated from non-robbing foraging behavior. Sociality and nectar-robbing appear to have been correlated. The concentrated changes test indicated that sociality (parasociality, subsociality, or eusociality) evolved much more often than by chance on branches on which nectar-robbing was present (p < 0.0001). The observed pattern of gains of sociality on the phylogeny of bees was extremely imlikely to have arisen by chance alone; no simulations foimd distributions of gains as extreme as those observed. I investigated whether robbing preceded sociality, as predicted by the hypothesis that robbing allowed broader diet breadth and thus advanced sociality, or whether robbing followed sociality. There was only one case in which it is clear whether sociality preceded robbing or vice versa, and that is a case in which sociality originated first. On branches on which sociality and 128 nectar-robbing were correlated, in L-T bees, sociality may have preceded robbing twice in the Xylocopa/Braimsapis clade and possibly once more in the Eiifriesea/Bombus clade. In these clades, the alternative is that sociality and nectar-robbing originated simultaneously. In S-T bees, it appears that sociality preceded robbing once in the Corynura/Aiigochlora clade and either preceded robbing or occurred simultaneously in the ancestors of the Dieunomia/Augochlora clade. In S-T bees, sociality and robbing occurred simultaneously two more times, in Andrena and Perdita. Eusociality may have preceded robbing at least twice and robbing may have preceded eusociality at least once in bees. On branches on which sociality and nectar-robbing were correlated, in L-T bees eusociality preceded robbing in the Ceratina/Braunsapis clade, and either preceded robbing or originated simultaneously in the Melipona-Bombus clade. On branches on which sociality and nectar-robbing were correlated, in S-T bees robbing preceded eusociality in the Augochloropsis/Augochlora clade. Also in S-T bees, eusociality either preceded robbing or occurred simultaneously in the ancestor of Lasioglossum. The ANOVA indicated that sociality and geographical region significantly affected both the median and maximimi diet breadth of species within genera (Table 2). Subsocial and eusocial bees had the broadest diet breadth in all three regions (Figs. 5-7). There was no difference between primary and secondary nectar- robbers in median or maximum diet breadth of species within genera (p < 0.48, p < 0.22 in ANOVA using robbing, region, and robbing x region as 129 factors, n = 34). In ail three regions, nectar-robbers had a broader diet breadth than did non-robbers (Table 3). Partial correlatioris between pairs of variables using independent contrasts, with each region considered separately, indicated that sociality was significantly correlated with nectar-robbing in the Eastern U. S., but not significantly in the other regions (Table 4). Social species in the Eastern U. S. also visited significantly more host plants in the Eastern U. S. than did less social species, as measured by maximum diet breadth of species within each genus. Nectar-robbers, but not more social bees, visited significantly more plants in California, when diet breadth was measured by median diet breadth of species (Table 4). Nectar-robbers also used significantly more resources in Costa Rica, as shown by the maximim:\ diet breadth of species in genera. There were trends of positive correlations between sociality, nectar- robbing, and diet breadth for all locations and all permutations of phylogenies (Table 4). Phylogenetic arrangement of families of bees did not affect correlations between variables. However, whether diet breadth of bee genera was summarized by medians or by maximums did affect outcomes of analyses. One might predict that sociality, nectar-robbing, and broad diet breadth might be more likely to be foimd in larger genera, because even if these behaviors were distributed randomly among species, larger genera would have more opportunities for these behaviors to appear. However, only in California were nectar-robbers found in larger genera (Table 4). In CaUfomia, nectar-robbers also visited significantly more host plants, as measured by 130 median diet breadth; wh^ the effect of nectar-robbing was removed, species in. larger genera visited significantly fewer host plants than did species in smaller genera.

Discussion

Danforth and Eickwort (1997) hypothesized that activity of bees over long seasons may be a prerequisite for the evolution of sociality in bees. A corollary of this hypothesis is that pre-sodal bees must have had a wide range of host plants to use over the season to support multiple broods. A broad diet breadth may have been a precondition for the evolution of sociality either because it ciUowed a longer lifespan of females that initiate nests or because it was necessary for multivoltinism (multiple broods per year). Both long lifespan of nest foundresses and multivoltinism may have preceded generational overlap, a necessary condition for the existence of sociality (fig.

1). Generalization in host-plant use (i.e. broad diet breadth) may have been one necessary condition for the evolution of sociality. It is already known that social bees tend to be active for several months, whereas solitary bees often are active only a few weeks during the year, even in the tropics (Heithaus 1979b; Wcislo and Cane 1996). In addition, families of bees with social genera tend to have exceptionally long-lived females, in comparison to families without social genera. Within families with social genera, the evolution of long-lived females appears to have preceded sociality, because it 13 1 is present in closely related non-social groups (Michener 1969; Schwartz et al. 1997).

Association of sociality with nectar-robbing

A shift in foraging behavior to nectar-robbing was found to be correlated with the evolution of sociality in bees. By opening up nectar resources enclosed by tubular corollas, nectar-robbing may have allowed an increase in generalization, which may have been a necessary condition for sociality to evolve. If eusociality originated from subsociality, nectar-robbing may have favored the transition to subsociality, because a wide range of resources may have been necessary to sustain the multiple generations over a year that would occur when an adult female lived with her immature offspring. It is more difficult to produce a scenario in which expansion of foraging range promotes transitions to semisociality (cooperation among females of same age) to eusociality because cooperation within a generation may not depend on having a wide range of resources. However, a broad diet breadth may have been necessary to support a traiisition from semisociality to eusociality because it would have supported multiple broods per year and longevity of females. Thus, generalization may have been important at different stages in the evolution of eusociality from subsociality as compared with the evolution from semisociality. If eusocial behavior originated directly from solitary behavior, which is suggested at several locations in the phylogeny of bees, diet breadth may still have been an important 132 precondition; it may have been necessary to support long-lived breeding females and to support multiple generations living together over a season. Nectar-robbing is associated with sociality on the phylogeny. Bee genera that have a broader diet breadth are associated with more complex levels of sociality. However, the effect of these factors on the evolution of sociaKty disappears when phylogenetic effect is considered. These results indicate that genera that share sociality, the presence of nectar-robbing, and generalization also share conunon ancestors; there are a few evolutionary events that have affected miiltiple genera. Nectar-robbing may be associated with sociality in other groups of insects as well. Other nectar-robbing insects include eusocial wasps in the family Vespidae, Vespida (Rust 1979) and Polistes (Barrows 1980), and ants Camponotus (Rust 1979) and Pachycondyla (McDade and Kinsman 1980). In addition, there are no nectar-robbers among sphecid wasps, a group in which eusociality is rare. Finally, no robbers are found in Masarinae, the solitary pollen wasps, even though they visit flowers (Gess 1996). In three other groups of insects, a shift in foraging behavior is associated with sociality. First, in bark beetles, solitary beetles turmel only in dead trees, but a social clade of beetles attacks live trees, cooperating to overwhelm trees' defenses (Kirkendall 1997). Second, gregarious tent caterpillars cooperate and communicate by odor trails. These gregarious caterpLQars more effectively overcome plant defenses and when they forage as a group than when they forage alone (Fitzgerald 1995). Third, social spiders cooperate to capture and consume large prey that a solitary spider of the same species cannot subdue (Rypstra and Tirey 1991). In these three cases. 133 sociality is associated with foraging efficiency, not generalization. Sociality in these insects may be favored by the benefits of group foraging. Social bees may also benefit by group foraging when commimicating locations of resources with nestmates permits more efficient foraging. However, this commimication is only known from Apis, Bombus, and Trigona (Roubik 1989), so is unlikely to have preceded the evolution of sociality or been an important factor in its evolution.

The Origin of Bees and Routes to Eusocialiti/

Because most basal bees were solitary and one hypothesis suggests that the common ancestor of the sister-groups sphecid wasps and Apoidea (bees) was probably solitary, I coded the hypothetical ancestral bee as solitary. However, with more information on phylogenetic relationships among basal bees and sphecid wasps, this hypothesis could change. This hypothesis is supported by independent evidence that suggests that the most basal sphecid wasps are also solitary (Roig-Alsina and Michener 1993). Two hypotheses have been offered for which group in Sphecidae is most basal. One h5^othesis suggests that subfamilies Sphecinae and Ampulicinae are basal groups (Brockmann 1997; Alexander 1992). Because both subfamilies contain only solitary wasps, parsimony would reconstruct the ancestor of Apoidea and Sphecidae as solitary. Another hypothesis has placed Pemphredoninae, the group containing the only eusocial sphecid wasp Microstigmus, at the base of Sphecidae (Michener 1986; Matthews 1991). Though most of the species within this subfamily are solitary (Brockmann 1997), it is unknovm whether 134

the ancestral species within this subfamily was also likely to have been solitary. Understanding routes to eusociality in bees provides evidence about factors that favor the evolution of eusociality. It has been suggested that eusociality has evolved through two different routes in bees: through semisodaLity in Halictidae, and through subsodality in Apidae (Lin and Michener 1972; Michener 1974). In semisocial groups, imrelated females of the same age nest together; in subsodal groups, females nest with immature relatives. If eusocial bees evolved from semisocial bees, a factor such as group defense against predation or parasitism may have been important in selection for this transition. However, group defense is less likely to be an important factor in the evolution of eusociality from subsociality; when a single female is caring for immature offspring, kin selection is more likely to be a factor in the development of eusociality. A comparison of bee genera (Appendix E) shows that eusocial genera in Halictidae also vary to semisodality. Eusocial genera in Apidae also tend to be subsocial, quasisocial, or semisocial. Genera that share ancestors with the eusocial genera are solitary, not sodal (Fig. 1). That these groups are solitary suggests either that the potential to be eusocial arose within each eusocial genus soon after the evolution of less complex forms of sociality, or that less complex forms of sociality are not necessarily intermediate in the evolution from solitary to eusocial behavior.

Association of Eusociality, Nectar-robbing, and Diet Breadth 135

One difficulty" in using the type behavioral characters in this study is that they are dependent on number of observations of the taxa. I coded sociality and nectar-robbing as present when they were known, but the taxa in which sociality and nectar-robbing are absent may just be those in which these traits are yet to be discovered. Diet breadth of bees may also be dependent on observations because rare taxa are less likely to be collected on plants and thus may tend towards having a lower estimated diet breadth than do common taxa. However, estimates of diet breadth may be less affected by commonness of taxa because I excluded from analyses species that were known to be rare. For all three characters, socicility, nectar-robbing, and diet breadth, there will imdoubtedly be more observations and subsequent changes in coding from absent to present for the former characters and to increasing diet breadth in the latter. These changes may affect estimated numbers of gains and losses. However, they will probably not affect the hypothesized correlatioris between aU three because changes in codings of a few taxa will not much affect the outcome of analyses. In particular, the correlation between sociality and nectar-robbing is so strong that it is unlikely to change. An entire suite of traits—sociality, nectar-robbing, and diet breadth- may be connected through learning abilities of bees. Among several species of megachilid bees, bees that visit more pollen sources have a wider variety of methods of pollen uptake than do more specialized bees (Miiller 1996), which indicates that diet breadth is at least partly determined by flexibility in foraging behavior of bees. Nectar-robbing may originate from opportunistic foraging, perhaps at corollas damaged by herbivores rather than by other 136 bees, because secondary robbing appears to have evolved at least once without being preceded by primary robbing. Bees with this flexibility in behavior may have greater learning abilities than bees that are not robbers; sociality has also been suggested to be correlated with learning ability (Dukas and Real 1991). Having nestmates may also make robbing more energetically profitable, because nestmates may more easily find robbed flowers and use refilled nectar. 137

Table 1. Levels of social organization among bees (adapted from Michener 1974) Character state Character state in analysis of in analyses of correlation with correlation with Level Characteristics nectar-robbing diet breadth Solitary no castes, no progressive 0 0 feeding, no cooperative work on cells

Solitary and same as above, but nests aggregating are grouped

Par asocial a group of imrelated commxinal females of the same generation making own cells within a single nest

quasisodal a group of females cooperatively provisions cells; ovaries similar size in females; females may be unrelated

semisocial all females in a single generation cooperatively provision cells and divide labor; females differ in ovary sizes; females may be imrelated

Subsocial adult female and her immature offspring which are fed by adult

Eusodal several generations of adults; division of labor and castes; ovarian development differs between workers and queens 138 ]• Ancestral Apoid wasp r^aColletes inMourecotelles loCaupolicana jsCrawfordapis inCadeguala inDiphaglossa jMydrosoma CMomelitta jLeiproctus I r—in Eulbnchopria I Q "^nTrichocolIetes oLonchopria ] Scrapter Ampnyimvlaeus jnMerogiossa inHylaeus jffyleoides jCnilicola • Euryglossa • Euryglossula •—inSystropha 5n Dufourea nDieunomia iiiiNiiiiiiuim aNomia E Lasloglossum Agapostemon Corynura Augochloropsis Pseudaugochloropsis Augochlorella Augochlora _Anarena inMegandrena • Hypomacrotera jB Perdita inProtoxaea Maximal social development ]• Euherbstia unordered inAlocandrena I I solitary inMacropis I in Melitta V/y/A aggregating fp Lr—inPasvpoda HH communal-quasisocial-semisocial ^ '—lSd Hesperapis subsocial-eusocial I in Meganomia lSD LONG-TONGUED BEES equivocal Fig. 1. Phylogeny of the short-tongued bees with Colletidae basal, adapted from. Alexander and Michener 1995, Danforth and Eickwort 1997, and Pesenko 1999). The phylogeny shows the reconstructed evolution of social development. 139 ]SHORT-TONGUED BEES EremapiIS inParatefir;• Paratefrapedia iiim I BExojnalopsis • Melissodes • Svastra —inEucera '=nnPeponapis I Ancyloscelis ^•Piad^ipa 3DPtiolthnx inMelitoma inDiadasia lectra • letrapedia • Habropoda • Anthophora • Deltopma .•Centns ]• Epicjians • Eufhesea mmm Euglossa M^pona Partaniona Paratngona TngonafT< • O mbus

Ceratina Macrogalea • Braunsapis —inParafidelia 5nNe^delia IB Pararhophites inLithucge HopUti§ •J—in Anthidium n^sTrachusa Lr—in Megachile Osmia j-^nSystropha i_3n Dufourea n Dieunomia yiNoijiia, Maximal social development Lasioglossum unordered Agapostemon Corynura solitary Augochloropsis aggregating Pseudaugochloropsis Augochlorella communal-quasisocial-semisocial Augochlora subsocial-eusocial equivocal Fig. 2. Phylogeny of the long-tongued bees, adapted from Roig-Alsina and Michener 1993. The phylogeny shows the reconstructed evolution of social development. 140 J Ancestral Apoid wasp -—inColletes Mourecotelles •—in Caupolicana Crawfordapis jt -| gi_r—in Cadeguala I'—lSd Diphaglossa inMydrosoma inC^omelitta oLeiproctus • Tnchoco oLonchopria inScrapter ^n Amphylaeus '^n Meroglossa iHylaeus Hyleoides n Chilicola •—inEurvglossa 5n Euryglossula •—inSvstropha ^•Dufourea •—inDieimomia "^nNomia Lasioglossum ]• Agapostemon ]Corynura Augochloropsis Pseudaugocnloropsis Augochlorella Augochlora Andrena inMegandrena • Hypomacrotera M Perdita jnProtoxaea joEuherbstia inAlocandrena Nectar-robbing inMacropis unordered inMelitta Lf-—in Dasvpoda absent '—lSd Hesperapis primary anomia secondary G-TONGUED BEES

Fig. 3. Phylogeny of the short-tongued bees with Colletidae bascd, adapted from Alexander arid Michener 1995, Danforth and Eickwort 1997, and Pesenko 1999). The phylogeny shows the reconstructed evolution of nectar-robbing. 14 1

inSHORT-TONGUED BEES OtiremapisEremapis a Paratefrapedia nExopaalopsis 3]nMelisscxl& oSvastra • Eucera inAncyloscelis

^ n '—inMelitoma ^^='HDiadasia ]Ctenoplectra :i:]nTetrapedia ^•Habropoda ^•Anthophora Deltoplila jCentns _jEpipjiaris Eumesea "nn Euglossa iMelipona Partamona Paratrigona Trigona Apis Bombus Xylocopa ]• Manuelia [j=pigHCeratma ir^nMacrogalea Braunsapis ^-^•Par^defia J n_3n Neondelia ]• Pararhophites inLithur^e HopUti§ _ Anthidium • Trachusa • Megachile Osmia —in Systropha . *=!• Emfourea J—•• Dieunomia Nectar-robbing T_5nNoima unordered IBLasioglossum ]• Agapostemon absent inCorynura primary Augochloropsis Pseudaugochloropsis secondary AugochKrella equivocal Augochlora

Fig. 4. Phylogeny of the long-tongued bees, adapted from Roig-Alsina and Michener 1993. The phylogeny shows the reconstructed evolution of nectar- robbing. Table 2: ANOVAs of median and maximum diet breadth of species within genera as a function of maximal social development within each genus and of geographical region. Parameter Source df SS F P Median diet breadtht Sociality 3 3.643 10.876 <0.0001 Region 2 1.743 7.804 0.0009 Sociality x region 6 1.040 1.553 0.175 Error 67 7.481

Maximum diet breadtht Sociality 3 5.094 13.996 <0.0001 Region 2 7.435 30.639 <0.0001 Sociality x region 6 1.667 2.291 0.045 Error 67 8.129 8.129 t log-transformed 143

100 Eastern U. S.

Maximal Social Development within Genus

Fig. 5. Maximal social development and median diet breadth of bee genera in Eastern U. S.. See Table 1 for explanation of levels of sociality. These data were used in the ANOVAs that did not accoiint for phylogenetic relationships. 144

Costa Rica

12 3 Maximal Social Development within Genus

Fig. 6. Maximal social development and median diet breadth of bee genera in Costa Rica. See Table 1 for explanation of levels of sociality. These data were used in the ANOVAs that did not account for phylogenetic relationships. 145

125 California

25

0 Maximal Social Development within Genios

Fig. 7. Maximal social development and median diet breadth of bee genera in Califorrua. See Table 1 for explanation of levels of sociality. These data were used in the ANOVAs that did not account for phylogenetic relationships. Table 3: ANOVAs of median and maximum diet breadth of species within genera as a function of presence or absence of nectar-robbing within each genus and of geographical region. Parameter Source df SS F P Median diet breadtht Nectar-robbing 1 1.546 10.676 0.0017 Region 2 1.467 5.065 0.0087 Robbing x region 2 0.024 0.084 0.9193 Error 73 10.573

Maximum diet breadtht Nectar-robbing 1 3.623 23.233 <0.0001 Region 2 6.804 21.816 <0.0001 Robbing x region 2 0.068 0.220 0.803 Error 73 11.385 t log-transformed Table 4: Partial correlation coefficients for pairs of variables, using independent contrasts from COMPARE. Partial correlations determine the correlation between each pair of variables while holding constant the value of each of the other variables. Correlations for each region are analyzed separately. A-F. Correlations using median diet breadth of species in genera. G-L. Correlations using maximum diet breadth of species in genera. Number of contrasts = 23 for Eastern U. S., 27 for California, and 26 for Costa Rica. Variable Median diet breadth of Number of species in Sociality Nectar-robbing species in A. Eastern U.S.-Halictidae basal genera (0.1.2,3) (0.1) genera Median diet breadth of species in genera 0.42 -0.27 -0.10 Sociality (0,1,2,3) 0.42 0.51 0.17 Nectar-robbing (0,1) -0.27 0.51 0.21 Number of species in genera -0.10 0.17 0.21

B. Eastern U.S.-Colletidae basal Median diet breadth of species in genera 0.42 -0.27 0.10 Sociality (0,1,2,3) 0.42 0.51 0.18 Nectar-robbing (0,1) 0.27 0.51"^ 0.21 Number of species in genera 0.10 0.17 0.21

C. California-Halictidae basal Median diet breadth of species in genera 0.31 0.60 + • 0.76 ++ Sociality (0,1,2,3) 0.31 0.0013 0.34 Nectar-robbing (0,1) 0.60 + 0.0013 0.72 ++ Number of species in genera - 0.76 ++ 0.34 0.72 ++

D. California-Colletidae basal Median diet breadth of species in genera 0.31 0.65 ++ • 0,75 ++ Sociality (0,1,2,3) 0.31 0.00006 0.34 Nectar-robbing (0,1) 0.65 ++ 0.00006 0.74 ++ Number of species in genera • 0.75 ++ 0.34 0.74 ++ P< 0.05; P < 0.01; + P < 0.005; Variable Median diet breadth of Number of species in Sociality Nectar-robbing species in E. Costa Rica-Halictidae basal genera (0,1,2,3) (0,1) genera

Median diet breadth of species in genera — 0.020 0.24 0.016

Sociality (0,1,2,3) 0.020 — 0.26 - 0.029

Nectar-robbing (0,1) 0.24 0.26 — 0.23 Number of species in genera 0.02 - 0.029 0.23

F. Costa Rica-Colletidae basal

Median diet breadth of species in genera — 0.032 0.23 0.015

Sociality (0,1,2,3) 0.032 — 0.23 -0.045

Nectar-robbing (0,1) 0.23 0.23 — 0.23 Number of species in genera 0.015 - 0.045 0.23

Maximum diet breadth of Number of species in Sociality Nectar-robbing species in G. Eastern U.S.—Halictidae basal genera (0,1,2,3) (0,1) genera

Maximum diet breadth of species in — 0.52"^ 0.21 0.16 genera

Sociality (0,1,2,3) 0.52"^ — 0.18 - 0.0093

Nectar-robbing (0,1) 0.21 0.18 — 0.39 Number of species in genera 0.16 - 0.0093 0.39

H. Eastern U.S.-Colletidae basal

Maximum diet breadth of species in — 0.49"^ 0.17 0.23 genera

Sociality (0,1,2,3) 0.49'^ — 0.31 0.011

Nectar-robbing (0,1) 0.17 0.31 — 0.20

Number of species in genera 0.23 0.011 0.20 — P< 0.05; P < 0.01; + P < 0.005; ++ P < 0.001 Variable Maximum diet breadth of Number of species in Sociality Nectar-robbing species in I. California-Halictidae basal genera (0,1,2,3) (0,1) genera

Maximum diet breadth of species in — 0.080 0.33 -0.011 genera

Sociality (0,1,2,3) 0.080 — 0.21 0.17

Nectar-robbing (0,1) 0.33 0.21 — 0.48'^ Number of species in genera - 0.011 0.17 0.48"^

J. California-Colletidae basal

Maximum diet breadth of species in — 0.083 0.36 0.0097 genera

Sociality (0,1, 2,3) 0.083 — 0.23 0.16

Nectar-robbing (0,1) 0.36 0.23 — 0.46"^ Number of species in genera 0.0097 0.16 0.46"^

K. Costa Rica-Halictidae basal

Maximum diet breadth of species in — - 0.055 0.41 0.24 genera

Sociality (0,1,2,3) - 0.055 — 0.27 - 0.015

Nectar-robbing (0,1) 0.41"^ 0.27 — 0.12 Number of species in genera 0.24 - 0.015 0.12

L. Costa Rica-ColIetidae basal

Maximum diet breadth of species in — - 0.038 0.40"^ 0.23 genera

Sociality (0,1,2,3) - 0.038 — 0.24 - 0.035

Nectar-robbing (0,1) 0.40"^ 0.24 — 0.12 Number of species in genera 0.23 - 0.035 0.12 — P< 0.05; P < 0.01; + P < 0.005; ++ P < 0.001 150

References

Alexander, B. A. 1992. An exploratory analysis of cladistic relationships within the superfamily Apoidea, with special reference to sphecid wasps (Hymenoptera). J. Hym. Res. 1:25-61.

Alexander, B. A. and C. D. Michener 1995. Phylogenetic studies of the families of short-tongued bees (Hymenoptera: Apoidea). Univ. Kans. Sci. Bull. 55:377-424.

Brockmaim, H. J. 1997. Cooperative breeding in wasps and vertebrates: the role of ecological constraints. Pp. 347-371 in J. C. Choe and B. J. Crespi, eds. The Evolution of Social Behavior in Insects and Arachnids. Cambridge, Cambridge Univ. Press, Cambridge.

Crespi, B. J. 1996. Comparative analysis of the origiris and losses of eusociality: causal mosaics and historical uniqueness. Pp. 253-287 in E. P. Martins, ed. Phylogenies and the Comparative Method in Animal Behavior. Oxford Univ. Press, New York.

Crozier, R. H. and P. Pamilo. 1996. Evolution of social insect colonies : sex allocation and kin selection. Oxford, Oxford Univ. Press.

Danforth, B. N. and G. C. Eickwort. 1997. The evolution of social behavior in the augochlorine sweat bees (Hjnnenoptera: Halictidae) based on a phylogenetic analysis of the genera. Pp. 270-292 in J. C. Choe and B. J. Crespi, eds. The Evolution of Social Behavior in Insects and Arachnids. Cambridge Univ. Press, Cambridge.

Darwin, C. D. 1859,1998 Modem Library ed. The Origin of Species. New York.

Dukas, R. and L. A. Real. 1991. Learning foraging tasks by bees: a comparison between social and soKtary species. Arumal Behaviour 42:269-276.

Eickwort, G. C. 1975. Gregarious nesting of mason bee Hoplitis anthocopoides and evolution of parasitism and sociality among Megachilid bees. Evolution 29:142-150.

Eickwort, G. C., J. M. Eickwort, .Gordon, J.and M. A. Eickwort. 1996. Solitary behavior in a high-altitude population of the social sweat bee Halictus rubicundus. Behav. Ecol. Sociobiol. 38:227-233.

Farris, J. S. 1988. Hennig86: Version 1.5. Port Jefferson Station, N. Y., distributed by the author. 15 1

Felsenstein, J. 1985. Phylogenies and the comparative method. Am. Nat. 125:1- 15.

Gess, S. K. 1996. The PoUen Wasps: Ecology and Natural History of the Masarinae. Cambridge, Mass., Harvard Univ. Press.

Goloboff, P. A. 1993. NONA, Version 1.1. New York, N. Y., distributed by the author.

Hamilton, W. D. 1964. The genetical evolution of social behavior. E. J. Theor. Biol. 7:17-32.

Heithaus, E. R. 1979a. Flower visitation records and resource overlap of bees and wasps in northwest Costa Rica. Brenesia 16:9-52.

Heithaus, E. R. 1979b. Flower-feeding specialization in wild bee and wasp communities in seasonal neotropical habitats. Oecologia 42:179-194.

Inouye, D. W. 1980. The terminology of floral larceny. Ecology 61:1251-1253.

Inouye, D. W. 1983. The ecology of nectar robbing. Pp. 153-173 in B. Bentley and T. Elias, eds. The Biology of Nectaries. Columbia University Press, New York.

BCirkendaU, L. R., D. S. Kent, and K. F. Raffa. 1997. Interactions among males, females and offspring in bark cind ambrosia beetles: the significance of Uving in tunnels for the evolution of social behavior. Pp. 181-215 in J. C. Choe and B. J. Crespi, eds. The Evolution of Social Behavior in Insects and Arachnids. Cambridge Univ. Press, Cambridge.

Lin, N. and C. D. Michener. 1972. Evolution of sociality in insects. Quart. Rev. Biol. 47:131-159.

Maddison, W. P. and D. R. Maddison. 1992. MacClade: Analysis of Phylogeny and Character Evolution. Simderland, Mass., Sinauer and Assoc.

Martins, E. P. 1999. COMPARE: statistical analysis of comparative data, version 4.3. Distributed by the author. Department of Biology, University of Oregon, Eugene.

Matthews, R. W. 1991. Evolution of social behavior in sphecid wasps. The Social Biology of Wasps. K. G. Ross and R. W. Matthews. Ithaca, Comstock Publ. Assoc.: 570-602.

McDade, L. A. and S. Kinsman. 1980. The impact of floral parasitism in two neotropical hummingbird-poUinated plant species. Evolution 34:944-958. 152

Michener, C. D. 1969. Comparative social behavior of bees. Ann. Rev. Ent. 14:299-343.

Michener, C. D. 1974. The Social Behavior of the Bees: A Comparative Study. Cambridge, Mass, Belknap Press of Harvard Univ. Press.

Michener, C. D. 1979. Biogeography of the bees. Annals of the Missouri Botanical Garden 66:277-347.

Michener, C. D. 1986. Family-group names among bees. J. Kans. Ent. Soc. 59:219-234.

Michener, C. D., R. J. McGinley, and B. N. Danforth. 1994. The Bee Genera of North and Central America. Washington, Smithsonian Institution Press.

Michener, C. D. 2000. The Bees of the World. Baltimore, The Johns Hopkins University Press.

Minckley, R. L. 1987. Aspects of the Reproductive Biology of Two Carpenter Bees (Genus Xylocopa) in Southern Arizona. Dept. of Entomology. Tucson, University of Arizona.

Mitchell, T- B. 1960. Bees of the Eastern United States, Vol. 1, North Carolina Agricultural Experiment Station.

Mitchell, T. B. 1962. Bees of the Eastern United States, Vol n. North Carolina Agricultural Experiment Station.

Moldenke, A. R. and J. L. Neff. 1974. The bees of Califorrua: a catalogue with special reference to pollination and ecological research. International Biological Program Origin and Structure of Ecosystems Technical Reports 74-1 to 74-6. Parts I-IV. (not publicly circulated): 1073.

Moldenke, A. R. 1976. California pollination ecology and vegetation types. Phytologia 34:305-361.

MuUer, A. 1996. Host-plant specialization in western palearctic Anthidiine bees (Hymenoptera: Apoidea: Megachilidae). Ecol. Monogr. 662:235-257.

Pesenko, Y. A. 1999. Phylogeny and classification of the family Halictidae revised (Hymenoptera: Apoidea). J. Kans. Ent. Soc. 72:104-123.

Robertson, C. 1929. Phenology of oligolectic bees and favorite flowers. Psyche 36:112-118. 153

Roig-Alsina, A. and C. D. Michener. 1993. Studies of the phylogeny and classification of long-tongued bees (Hymenoptera: Apoidea). The University of Kansas Science Bulletin 55:123-173.

Roubik, D. W. 1989. Ecology and Natural History of Tropical Bees. Cambridge University Press, New York, N. Y.

Rust, R. W. 1979. Pollination of Impatiens capensis: pollinators and nectar robbers. Joiimal of the Kansas Entomological Society 52:297-308.

Rypstra, A. L. and R. S. Tirey. 1991. Prey size, prey perishability, and group foraging in a social spider. Oecologia 86:25-30.

Schwartz, M. P., L. X. Silberbauer, and P. S. Hurst. 1997. Intrinsic and extrinsic factors associated with social evolution in allodapine bees. Pp. 333-346 in J. C. Choe and B. J. Crespi, eds. The Evolution of Social Behavior in Irisects and Arachnids. Cambridge Uruv. Press, Cambridge.

Sherman, P. W. 1995. The eusodality continuum. Behav. Ecol. 1:102-108.

Swofford, D. L. 1993. PAUP: Phylogenetic Analysis Using Parsimony, Version 3.1. Champaign, Illinois, Illinois Natural History Survey.

Thome, B. L. 1997. Evolution of eusodality in termites. Arm. Rev. Ecol. Syst. 28:27-54.

Wcislo, W. T. and J. H. Cane. 1996. Floral resovurce utilization by solitary bees (Hymenoptera: Apoidea) and exploitation of their stored foods by natural enemies. Ann. Rev. Ent. 41:257-286.

Wcislo, W. T. and B. N. Danforth. 1997. Secondary solitary: the evolutionary loss of social behavior. Trends in Ecol. and Evol. 12:468-474. 154

References on Social Development of Bees

Abrams, J. and G. C. Eickwort. 1981. Nest switching and guarding by the communal sweat bee Agapostemon virescens (Hymenoptera, Halictidae). Insectes Sociaux 28:105-116.

Bullock, S. H., R. Ayala, G. Rodriguez-Gonzalez, R. Palados-Chavez, D. Ramos Zamora, D. L. Quiroz-Garcia, and M. D. Arreguin-Sanchez. 1991. Nest provision and pollen foraging in 3 Mexican species of solitary bees (Hjnnenoptera, Apoidea). Pan-Pac. Ent. 67:171-176.

Cameron, S. A., J. B. Whitfield, C. L. Hulslander, W. A. Cresko, S. B. Isenberg, and R. W. King. 1996. Nesting biology and foraging patterns of the solitary bee Melissodes nistica (Hymenoptera: Apidae) in northwest Arkansas. J. Kans. Ent. Soc. 69:260-273, Suppl. S.

Cane, J. H. 1994. Nesting biology and mating behavior of the southeastern blueberry bee, Habropoda laboriosa (Hymenoptera, Apoidea). J. Kans. Ent. Soc. 67:236-241.

Cane, J. H. 1995. Notes on the nesting biology of Svastra atripes atrimita (Laberge) (Hymenoptera, Apidae). J. Kans. Ent. Soc. 68:238-240.

Cane, J. H. 1996. Nesting resins obtained from larrea pollen host by an oligolectic bee, Trachusa larreae (CockereU) ''Hymenoptera: Megachilidae). J. Kans. Ent. Soc. 69:99-102.

Chmurzyriski, J. A., M. Kieruzel, A. Krzysztofiak, and L. Krzysztofiak. 1998. Long-distance homing ability in Dasypoda altercator (Hymenoptera, MeKttidae). Ethology 104:421-429.

Danforth, B. N. 1989. Nesting behavior of 4 species of Perdita (Alloperdita) bradleyi (Hymenoptera, Andrerudae). J. Kcms. Ent. Soc. 62:59-79.

Danforth, B. N. 1990. Provisioning behavior and the estimation of investment ratios in a solitary bee, Calliopsis (Hypomacrotera) persimilis (Cockerel!). Behav. Ecol. Soc. 27:159-168.

Danforth, B. N. 1991. Female foraging and intranest behavior of a commimal bee, Perdita portalis (H5anenoptera, Andrenidae). Ann. Ent. Soc. Am. 84:537-548.

Danforth, B. N. and G. C. Eickwort 1997. The evolution of social behavior in the augochlorine sweat bees (Hymenoptera: Halictidae) based on a 155

phylogenetic analysis of the genera. The Evolution of Social Behavior in Insects and Arachnids. J. C. Choe and B. J. Crespi. Cambridge, Cambridge Univ. Press: 270-292.

Danforth, B. N., J. L. Neff, and P. BarrettoKo. 1996. Nestmate relatedness in a communal bee, Perdita texana (Hymenoptera: Andrenidae), based on DNA fingerprinting. Evolution 50:276-284.

Eickwort, G. C. 1975. Gregarious nesting of mason bee Hoplitis anthocopoides and evolution of parasitism and sociality cimong Megachilid bees. Evolution 29:142-150.

Eickwort, G. C., P. F. Kukuk, and P. Barrettoko. 1986. The nesting biology of Diifourea novaeangliae (Hymenoptera, Halictidae) and the systematic position of the Dufoxireinae based on behavior and development. J. Kans. Ent. Soc. 59:103-120.

Eickwort, G. G. 1981. Aspects of the nesting biology of 5 Nearctic species of Agapostemon (Hymenoptera; Halictidae). J. Kans. Ent. Soc. 54:337-351.

Fauria, K. and R. Campan. 1998. Do solitary bees Osmia cornuta Latr. and Osmia lignaria Cresson use proximal visual cues to localize their nest? J. Iris. Behav. 11:649-669.

Giblin, R. M. and H. K. Kaya. 1980. The association of a nematode, Hiintaphelenchoides sp. (Aphelenchoididae) with the solitary soil dwelling bee, Anthophora bomboides stanfordiana (Anthophoridae Hymenoptera). J. Nematol. 12:221.

Heithaus, E. R. 1979. Flower visitation records and resource overlap of bees and wasps in northwest Costa Rica. Brenesia 16:9-52.

Hogendoom, K. and R. Leys. 1997. Life-cycle of Halictus nibicundus Christ (Hymenoptera : Halictidae) in the Netherlands: Comparison of two populations. J. Kans. Ent. Soc. 70:347-352.

Kadmon, R. and A. Shmida. 1992. Departure rules used by bees foraging for nectar~a field test. Evol. Ecol. 6:142-151.

Martins, R. P. and Y. Antoniru. 1994. The biology of Diadasina distincta (Holmberg, 1903) (Hymenoptera, Anthophoridae). Proc. Ent. Soc. Wash. 96:553-560.

Martins, R. P. and J. E. C. Figueira. 1992. Spatial distribution of nests in Diadasina distincta (Holmberg) (Hymenoptera, Anthophoridae). J. Ins. Behav. 5:527-529. 156

Martins, R. P., F. G. Guimaraes, and C. M. Dias. 1996. Nesting biology of Ptilothrix pliimata Smith, with a comparison to other species in the genus (Hymenoptera: Anthophoridae). J. Kans. Ent. Soc. 69:9-16.

McGinley, R. J. and J. G. Rozen. 1987. Nesting biology, immature stages, and phylogenetic placement of the Palaearctic bee Pararhophites (H5anenoptera: Apoidea). Amer. Mus. Novit. 2903:1-21.

Mcintosh, M. 1997. Nest-substrate preferences of the twig-nesters Cemtina acantha, Ceratina nanula (Apidae) and Pemphredon lenthifer (Sphecidae). J Kans Ent Soc 69(Suppl. 4):216-231.

Michener, C. D. 1974. The Social Behavior of the Bees: A Comparative Study. Cambridge, Mass, Belknap Press of Harvard Univ. Press.

Michener, C. D., R. J. McGinley, and B. N. Danforth. 1994. The Bee Genera of North and Central America. Washington, Smithsonian Institution Press.

Miliczky, E. R. and E. A. Osgood. 1995. Bionomics oiAndrena (Melandrena) vicina Smith in Maine and Washington, with new parasite records for A. (M.) regiilaris Malloch and a review of Melandrena biology. J. Kans. Ent. Soc. 68:51-66.

Minckley, R. L. 1987. Aspects of the Reproductive Biology of Two Carpenter Bees (Genus Xylocopa) in Southern Arizona. Dept. of Entomology. Tucson, University of Arizona: 90.

Minckley, R. L., W. T. Wcislo, D. Yanega, and S. L. Buchmarm. 1994. Behavior and phenology of a specialist bee {Diewiomia) and sunflower (Helianthiis) pollen availability. Ecology 75(5): 1406-1419.

Mitchell, T. B. 1960. Bees of the Eastern United States, Vol. 1, North CaroUna Agricultural Experiment Station.

Mitchell, T. B. 1962. Bees of the Eastern United States, Vol II, North Carolina Agricultural Experiment Station.

Moldenke, A. R. 1976. California pollination ecology and vegetation t)rpes. Phytologia 34:305-361.

MuUer, A., W. Topfl, and F. Amiet, 1996. Collection of extrafloral trichome secretions for nest wool impregnation in the solitary bee Anthidium manicatum. Naturwissenschaften 83:230-232. 157

Myers, J. and M. D. Loveless. 1976. Nesting aggregations of Euglossine bee Euplusia surinamensis (Hymenoptera, Apidae): individu^ interactions and advantage of living together. Can. Ent. 108:1-6.

Neff, J. L. 1984. Observations on the biology of Eremapis parviila Ogloblia an Anthophorid bee with a metasomal scopa. Pan-Pacific Ent. 60:155-162.

Neff, J. L. and B. B. Simpson. 1992. Partial bivoltinism in a ground-nesting bee: the biology of Diadasia rinconis in Texas (Hymenoptera, Anthophoridae). J. Kans. Ent. Soc. 65:377-392.

Neff, J. L. and B. B. Simpson. 1997. Nesting and foraging behavior of Andrena iCallandrena) rudbeckiae Robertson (Hymenoptera: Apoidea: Andrenidae) in Texas. J. Kans. Ent. Soc. 70:100-113.

Norden, B. B., K. V. Krombein, and S. W. T. Batra. 1994. Nests and enemies of Exomalopsis (Phanamalopsis) solani Cockerell (Hymenoptera, Apoidea, Mutillidae, Diptera, Asilidae). Proc. Ent. Soc. Wash. 96:350-356.

Ordway, E. 1987. The life history of Diadasia rinconis CockereE (Hymenoptera, Anthophoridae). J. Kans. Ent. Soc. 60:15-24.

Parker, E. D. 1975. Nests of mason bees Osmia tanneri sandhouse and Osmia longida Cresson with a description of female of Osmia tanneri (Hymenoptera Megachilidae). Pan-Pac. Ent. 51:179-183.

Parker, E. D. 1980. Nests of Osmia marginipennis Cresson with a description of the female (Hymenoptera, Megachilidae). Pan-Pac. Entomol. 56:38-42.

Parker, F. D. 1981. Nests and nest associates of a desert bee, Osmia marginata Michener (Hymenoptera, Megachilidae). Southwest. Entomol. 6:184-189.

Paxton, R. J. 1991. Profile of a solitary bee-Andrena fulva. Bee World 72:11-18.

Paxton, R. J., P. A. Thoren, J. Tengo, A. Estoup, and P. Pamilo. 1996. Mating structure and nestmate relatedness in a commimal bee, Andrena jacobi (Hymenoptera, Andrerudae), using microsateUites. Mol. Ecol. 5:511-519.

Paxton, R. J., I. Fries, N. J. Pieruazek, and J. Tengo. 1997. High incidence of infection of an imdescribed microsporidium (Microspora) in. the commimal bee Andrena scotica (Hymenoptera, Andrenidae). Apidologie 28:129-141.

Ramalho, M., V. L. Imperatriz-Fonseca, and T. C. Giannini. 1998. Within-colony size variation of foragers and pollen load capacity in the stingless bee 158

Melipona qiiadrifasciata anthidioides Lepeletier (Apidae, Hymenoptera). Apidologie 29:221-228.

Ramirez-Arriaga, E., J. 1. CuadrieUo-Aguilar, and E. M. Hernandez. 1996. Nest structure and parasite of Euglossa atroveneta Dressier (Apidae: Bombinae: Euglossini) at Union Juarez, Chiapas, Mexico. J. Kans. Ent. Soc. 69:144- 152.

Raw, A. 1972. Biology of solitary bee Osmia rufa (L) (Megachilidae). Trans. Roy. Ent. Soc. Lond. 124:213-229.

Richards, M. H. and L. Packer. 1998. Demography and relatedness in multiple- foundress nests of the social sweat bee, Halictus ligatus. Ins. Sociaux 45:97- 109.

Riddick, E. W. 1992. Nest distribution of the solitary bee Andrena macra Mitchell (Hymenoptea, Andrerudae), with observations on nest structure. Proc. Ent. Soc. Wash. 94:568-575.

Rozen, J. G. 1967. Review of the biology of the Panurgine bees, with observations on North American forms (Hymenoptera, Andrenidae). Am. Mus. Novit. 2297.

Rozen, J. G. 1983. Nesting biology of the bee Svastra sabinensis (Hymenoptera, Anthophoridae). N. Y. Ent. Soc. 91:264-268.

Rust, R. W., R. W. Thorp, and P. F. Torchio. 1974. Ecology of Osmia nigrifrons with a comparison to other Acanthosmioides. J. Nat. Hist. 8:29-47.

Sakagami, S. F. and Y. Maeta 1977. Some presumably pre-social habits of Japanese Ceratina bees, with notes on various social types in Hymenoptera. Insectes Sociaux 24:319-343.

Sakagami, S. F. and Y. Maeta 1984. Multifemale nests and rudimentary castes in the normally solitary bee Ceratina japonica (Hymenoptera, Xylocopinae). J. Kans. Ent. Soc. 57:639-656.

Scott, V. 1993. Use of communal nest entrances by Osmia simillima (Hymenoptera, Megachilidae). Great Lakes Entomol. 26:79-80.

Spangler, H. G. and S. L. Buchmarm. 1991. Effects of temperature on wingbeat frequency in the solitary bee Centris caesalpiniae (Anthophoridae, Hymenoptera). J. Kans. Ent. Soc. 64:107-109.

Spessa, A., P. Schwarz, and M. Adams. 2000. Sociality in Amphylaeus morosus (Hymenoptera: CoUetidae: Hylaeinae). Arm. Ent. Soc. Am. 93:684-692. 159

Stone, G. N. 1994. Activity patterns of females of the solitary bee Anthophora plumipes in relation to temperature, nectar supplies, and body size. EcoL Ent. 19:177-189.

Sugiura, N. 1994. Parental investment and offspring sex ratio in a solitary bee, Anthidium septemspinosum Lepeletier (Hymenoptera, Megachilidae). J. Ethol. 12:131-139.

Tepedino, V. J. and P. F. Torchio. 1982. Temporal variability in the sex ratio of a nonsocial bee, Ostnia lignaria propinqua: extrinsic determination or the tracking of an optumum. Oikos 38:177-182.

Torchio, P. F. 1974. Notes on the biology of Ancyloscelis armata Smith and comparisons with other Anthophorine bees (Hymenoptera; Anthophoridae). J. Kans. Ent. Soc. 47:54-55.

Torchio, P. F. and V. J. Tepedino. 1980. Sex ratio, body size, and seasonality in a solitary bee, Osmia lignaria propinqiia Cresson (Hjonenoptera, Megachilidae). Evolution 34:993-1003.

Triplett, D. C. and A. R. Gittins. 1988. Nesting, mating, and foraging habits of Melissodes (Melissodes) tepida tepida Cresson in Idaho. Proc. Ent. Soc. Wash. 90:462-470.

Vandenberg, J. D. 1995. Nesting preferences of the solitary bee Osmia sanrafaelae (Hymenoptera, Megachilidae). J. Econ. Entomol. 88:592-599.

Vicidomini, S. 1996. Biology of Xylocopa violacea (Hymenoptera): In-nest ethology. Ital. J. Zool. 63:237-242.

Vogel, M. E. and P. F. Kukuk 1994. Individual foraging effort in the facultatively social Halictid bee, Nomia (Austronomia) australica (Smith). J. Kans. Ent. Soc. 67:225-235.

Wcislo, W. T. 1992. Attraction and learning in mate-finding by solitary bees, Lasioglossum (Dialictiis) figiierisi Wcislo and Nomia triangidifera Vachal (Hymenoptera, Halictidae). Behav. Ecol. Socio. 31:139-148.

Wcislo, W. T. 1993. Commimal nesting in a North American Pearly-banded bee, Nomia tetrazonata, with notes on nesting behavior of Dieunomia heteropoda (Hymenoptera, Halictidae, Nomiinae). Arm. Ent. Soc. Am. 86:813-821.

Wcislo, W. T. and J. H. Cane. 1996. Floral resource utilization by solitary bees (Hjonenoptera: Apoidea) and exploitation of their stored foods by natural enemies. Arm. Rev. Ent. 41:257-286. 160

Wcislo, W. T., R. L. Minckley, R. A. B. Leschen, and S. Reyes. 1994. Rates of parasitism by natural enemies of a solitary bee, Dieunomia trianulifera (Hymenoptera, Coleoptera, and Diptera) in relation to phenologies. Sociobiology 23:265-273.

Willis, D. S. and P. G. Kevan. 1995. Foraging dynamics of Peponapis pruinosa (Hymenoptera, Anthophoridae) on pumpkin (Cucurbita pepo) in southern Ontario. Canad. Ent. 127:167-175.

Willmer, P., F. Gilbert, J. Ghazoul, S. Zalat, and F. Semida. 1994. A novel form of territoriality-daily paternal investment in an Anthophorid bee. Anim. Behav. 48:535-549.

Yanega, D. 1994. Beetles associated with bee nests (Hymenoptera, Apidae) in Chiapas, Mexico, with descriptions of the immature stages of Vanonus balteatus Werner (Coleoptera, Aderidae, Endomychidae, Meloidae). Coleo. Bull. 48:355-360. 161

References on Nectar-Robbing

Barrows, E. M. 1976. Nectar robbing and pollination of Lantana camara (Verbenaceae). Biotropica 8:132-135.

Barrows, E. M. 1980. Robbing of exotic plants by introduced carpenter and honey bees in Hawaii, with comparative notes. Biotropica 12: 23-29.

Barrows, E. M., M. R. Chabot, C. D. Michener, and T. P. Snyder. 1976. Foraging and mating behavior in Perdita texana. Journal of the Kansas Entomological Society 49:275-279.

Benedek, P., L. Blank, and J. Koml0di. 1973. Behaviour of wild bees (Hjncnenoptera: Apoidea) on hairy vetch (Vicia villosa Roth.) flowers. Zeitschrift fur Angewandte Entomologie 74: 80-85.

Brandenburgh, W. 1961. Broad beans, causes of poor crop yields sought. N. Z. Jl. Agric. 102: 277-280.

Darwin, C. D. 1859 (1998 Modem Library ed.). The Origin of Species. New York.

Eaton, G. W. and M. G. Stewart. 1969. Blueberry blossom damage caused by bumble-bees. Canad. Entomol. 101:149-150.

Free, J. B. 1975. The behaviour of Xi/locopa mordax foraging on Ipomoea nil. Bee World 56:121.

Hawkins, R. P. 1961. Observations on the pollination of red clover by bees. I. The yield of seed in relation to the numbers and kinds of pollinators. The Armals of Applied Biology 49:55-65.

Inouye, D. W. 1983. The ecology of nectar robbing. Pp. 153-173 in B. Bentley and T. Elias, eds. The Biology of Nectaries. Columbia University Press, New York.

Iwata, K. 1963. Miscellaneous biological notes on Aculeate Hymenoptera in Kagawa in the years of 1948 and 1949. Trans. Shikoku ent. Soc. 7:114- 118.

Janzen, D. H. 1964. Notes on the behaviour of four subspecies of the carpenter bee, Xylocopa (Notoxi/locopa) tabanifonnis in Mexico. Aim. ent. Soc. Am. 57:296-301. 162

Laroca, S. 1970. Contribuigao para o cohedmento das relagoes entre abelhas e flores: coleta de polen das anteras tubulares de certas Melastomataceae. Floresta 2:69-74.

Linsley, E. G., C. M. Rick, and S. G. Stepheris. 1966. Observations on the floral relationships of the Galapagos carpenter bee (Hymenoptera: Apidae). Pan-Pacif. Ent. 42:1-18.

McDade, L. A. and S. Bdnsman. 1980. The impact of floral parasitism, in two neotropical hurmningbird-poUinated plant species. Evolution 34: 944- 958.

Meidell, 0.1944. Notes on the pollination of Melampynim pratense and the honeystealing of humble-bees and bees. Bergens Museums Arbok No. 11:5-11.

Miyamoto, S. 1961. Flower-visiting habit of Xylocopa appendiculata circumvolaris Smith (Biological studies on Japanese bees XXni). Kontyu 29:4-13.

Olesen, J. M. 1996. From naivete to experience: Bumblebee queens (Bombus terrestris) foraging on Corydalis cava (Fumariaceae). J. Kans. Ent. Soc. 69:274-286.

Pijl, V. d., L. 1954. Xylocopa and flowers of the tropics. I. The bees as pollinators: Lists of the flowers visited. Proc. Koninkl. Nederl. Akad. Wet. (Series C) 57:413-423.

Richardson, L. L. 1999. Mutualist and Antagonist Flower Visitors: a Pollination Study of Two Species of Manzanita (Arctostaphylos: Ericaceae). Dept. of Ecology and Evol. Biol. Tucson, University of Arizona.

Richardson, S. C. 1995. Temporal variation in pollinator abimdance and pollinator foraging behavior in response to robbed flowers. Biodiversity and Management of the Madrean Archipelago: the Sky Islands of Southwestern United States and Northwestern Mexico, Tucson, Arizona, USDA Forest Service.

Roubik, D. W. 1979. Africariized honey bees, stingless bees, and the structure of tropical plant-poUinator commxmities. Proc. IV Int. Symp. Poll. Maryland Agric. Exp. Sta. Misc. Publ. 1. D. Caron: 403-417.

Roubik, D. W. 1982. The ecological impact of nectar-robbing bees and pollinating hummingbirds on a tropical shrub. Ecology 63:354-360. 163

Roubik, D. W. 1989. Ecology and Natural History of Tropical Bees. New York, N.Y., Cambridge University Press.

Rust, R. W. 1979. Pollination of Impatiens capensis: poUinators and nectar robbers. Journal of the Kansas Entomological Society 52:297-308. APPENDIX E

Summary of level of sociality and level of nectar-robbing for genera of short-tongued and long-tongued bees. Sources describe level of sociality. Nectar-robbing is categorized as 0 = non-robber, 1 = primary robber, 2 = secondary robber. Foraging references include studies of all aspects of foraging behavior, not only nectar- robbing. Median number of host plants visited by all species of bees within a genus is summarized for each geographic area. Data for host use were compiled from Michell 1960 and Mitchell 1962 for Eastern U. S., from Heithaus 1979 for Costa Rica, and from Moldenke and Neff 1974 from California. Median Number (Range) of Hosts Level of Visited in a Geographical Area Level of Nectar- Genus Sociality Sources Robbing Eastern U.S. Costa Rica California Colletes solitary (Michener 1974) 0 4 (2-7) 3 (1-39) Mourecotelles solitary (Michener 1974) 0 Caiipolicana solitary (Michener 1974) 0 Craivfordapis solitary (Michener 1974) 0 Cade^uala solitary (Michener 1974) 0 Diphci^lossa solitary (Michener 1974) 0 Mydrosoma solitary (Michener 1974) 0 Callomelitta solitary (Michener 1974) 0 Leiproctus solitary (Michener 1974) 0 Eulonchopria solitary (Michener 1974). 0 Trichocolletes solitary (Michener 1974) 0 Lonchopria solitary (Michener 1974) 0 Scrapter solitary (Michener 1974) 0 Amphylaeus communal or (Spessa et al. 2000; 0 quasisocial, Michener 1974) solitary Mero^lossn solitary (Michener 1974) 0 Hyleoides solitary (Michener 1974) 0 Hylaeus solitary (Michener 1974) 0 3 (1-13) 16 (1-104) Chilicola solitary (Michener 1974) 0 Eim/^lossa solitary (Michener 1974) 0 Eiin/^lossula solitary (Michener 1974) 0 Andrena communal, (Paxton 1991; Riddick 2 5 (1-46) 6 (1-61) solitary 1992; Miliczky and Osgood 1995; Paxton, Thoren et al. 1996; Neff and Simpson 1997; Paxton, Fries et al. 1997) Euherbstia solitary (Michener 1974) 0 Alocandrem solitary (Michener 1974) 0 Hypomacrotera/ aggregating, (Rozen 1967; Danforth 0 3(1-17) Colliopsis solitary 1990) Megandrem solitary (Michener 1974) 0 Protoxaea solitary (Michener 1974) 0 Dufoiirea solitary (Michener 1974; 0 2 (1-5) 3(1-32) Eickwort, Kukuk et al. 1986) Systropha solitary (Michener 1974) 0 Dieunomia solitary (Wcislo 1993; 0 Minckley, Wcislo et al. 1994; Wcislo, Minckley et al. 1994) Nomia quasisocial, (Michener 1974; Wcislo 0 11 (1-17) 34(18-43) communal, 1992; Wcislo 1993; solitary Vogel and Kukuk 1994)

ON Lasioglossum eusocial, (Michener 1974; 1 13 (1-148) 4 (1-21) 4 (1-132) semisocial, Sakagami and Maeta solitary 1984; Danforth 1991) Agapostemon communal, (Abrams and Eickwort 0 52 (20-89) 9(9) 67 (29-130) solitary 1981) AugochJora eusocial, (Danforth and 2 68 (68) 5.5 (1-18) semisocial, kckwort 1997) solitary Corynura communal or (Danforth and 0 semisocial, Eickwort 1997) solitary Augochlorella eusocial, (Danforth and 2 89 (18-142) 13 (2-23) 123 (123) semisocial, Eickwort 1997) solitary Augochloropsis semisocial, (Danforth and 2 16 (13-80) 23 (11-35) communal, Eickwort 1997) solitary Pseudaugochlom semisocial (Michener, 1974; 1 5(5) Sakagami and Maeta 1984) Macropis solitary (Michener 1974) 0 3.5 (1-5) Melitta solitary (Michener 1974) 0 2.5 (1-5) Dasypoda solitary (Michener 1974; 0 Chmurzynski, Kieruzel et al. 1998) Meganomia solitary (Michener 1974) 0 Haplomelitta solitary (Michener 1974) 0 Hesperapis solitary (Michener 1974) 0 7 (1-27) Eremapis aggregating (Neff 1984) 0

ON ON Exomalopsis communal (Michener 1974; 0 4.5 (MO) 4 (1-18) Norden, Krombein et al. 1994) Paratetrapedia aggregating (Michener 1974) 2 7 (1-22) Ancyloscelis aggregating, (Torchio 1974; Bullock, 0 4.5 (2-7) solitary Ayala et al. 1991; Yanega 1994) Melitoma solitary (Bullock, Ayala et al. 0 7(7) 1991; Yanega 1994) Diadasia solitary (Ordway 1987; Neff 0 8(1-23) and Simpson 1992; Wcislo and Cane 1996) Diadnsin solitary (Martins and Figueira 0 1992; Martins and Antonini 1994) PtiJothrix solitary (Martins, Guimaraes et 0 6(6) al. 1996) Telrapedin solitary (Michener 1974) 0 2(2) Eucera solitary (Kadmon and Shmida 0 1992) Melissodes aggregating, (Triplett and Gittins 0 15 (1-77) 7 (4-17) 5(1-89) solitary 1988; Cameron, Whitfield et al. 1996) Svastra communal (Rozen 1983; Cane 0 22 (7-50) 1(1) 1.5 (M6) 1995) Peponapis solitary (Michener 1974; Willis 0 8(8) 3 (1-5) and Kevan 1995) Anthophora aggregating, (Giblin and Kaya 1980; 0 12 (2-31) 1(1) 4 (1-56) solitary Stone 1994; Willmer, Gilbert et al. 1994) Hnbropoda solitary (Cane 1994) 0 2 (1-14) Deltoptik solitary (Michener 1974) 0 Centris solitary (Michener 1974; 0 3.5 (1-11) 4.5 (1-17) Spangler and Buchmann 1991; Yanega 1994) Epidmris solitary (Michener 1974) 0 2(1-3) Manuelia solitary ? (Michener 1974) 0 Xylocopn eusocial, (Minckley 1987; 1 19 (17-21) 4(2-11) 41 (20-50) subsocial, Danforth 1991; solitary Vicidomini 1996) Ceratim eusocial, (Sakagami and Maeta 2 31 (4-140) 8(1-14) 32 (3-97) quasisocial, 1977; Sakagami and semisocial, Maeta 1984; Danforth solitary 1991; Mcintosh 1997) Macro^alea social (Michener 1974) 0 Braunsapis eusocial, (Sakagami and Maeta 0 semisocial 1984; Danforth 1991) Bombus eusocial (Michener 1974) 2 14 (1-157) 1 (1-1) 36 (5-139) Apis eusocial (Michener 1974) 2 13 (13) Melipom eusocial (Ramalho, Imperatriz- 0 23 (23) Fonseca et al. 1998) Partamona eusocial (Michener, McGinley 1 et al. 1994) Euglossn parasocial, (Ramirez-Arriaga, 0 1(1) solitary Cuadriello-Aguilar et al. 1996) Eiifriese/ aggregating (Myers and Loveless 0 Euplusia 1976) Pararhopites communal (McGinley and Rozen 0 1987) Neofidelia solitary (Michener 1974) 0 Parafidelia solitary (Michener 1974) 0 Lithur^e solitary (Michener 1974) 0 3.5 (1-6) 1(1) 13 (13) Trachusa aggregating, (Michener 1974; Cane 0 5(1-12) solitary 1996) Anthidium solitary (Sugiura 1994; Muller, 2 7.5 (5-10) 5 (3-7) 13 (2-40) Topfl et al. 1996) Hoplitis communal, (Eickwort 1975) 0 10 (1-45) 2(1-61) solitary Osmin communal, (Raw 1972; Rust, 2 8 (1-51) 7(1-71) solitary Thorp et al. 1974; Parker 1975; Parker 1980; Torchio and Tepedino 1980; Parker 1981; Tepedino and Torchio 1982; Scott 1993; Vandenberg 1995; Fauria and Campan 1998) Megachile aggregating, (Michener 1974) 2 13 (2-102) 3 (1-7) 6 (1-116) solitary Ctenoplectra solitary (Michener 1974) 0 Partmnona eusocial (Michener 1974) 1 Paratri^om eusocial (Michener 1974) 1 Perditn communal, (Rozen 1967; Danforth 1 3 (1-8) 2(1-21) solitary 1989; Danforth 1991; Danforth, Neff et al. 1996) Tri^om eusocial (Michener 1974) 1 30 (3-95) 0\ VO 170

REFERENCES

Addicott, J. F. 1981. Stability properties of 2-species models of mutualism: simulation, studies. Oecologia 49:42-49.

Addicott, J. F. 1986. Variation in the costs and benefits of mutualism: the interaction between yuccas and yxicca moths. Oecologia 70:486-494.

Alexander, B. A. cind C. D. Michener. 1995. Phylogenetic studies of the families of short-tongued bees (Hymenoptera: Apoidea). Uruv Kans Sci Bull 55:377-424.

Boucher, D. H., S. James, and K. H. Keeler. 1982. The ecology of mutualism. Ann Rev Ecol Syst 13:315-47.

Bronstein, J. L. 1994. Conditional outcomes in mutualistic interactions. Trends Ecol Evol 9:214-217.

Bull, J. J. and W. R. Rice. 1991. Distinguishing mechanisms for the evolution of cooperation. J Theor Biol 149:63-74.

Crespi, B. J. 1996. Comparative analysis of the origins and losses of eusociaHty: causal mosaics and historical imiqueness. Pages 253-287 in E. P. Martins, ed. Phylogenies and the Comparative Method in Animal Behavior. Oxford Univ. Press. New York.

Demfortli, B. N. and G. C. Eickwort 1997. The evolution of social behavior in the augochlorine sweat bees (Hymenoptera: Halictidae) based on a phylogenetic analysis of the genera. Pages 270-292 in J. C. Choe and B. J. Crespi, eds. The Evolution of Social Behavior in Insects and Arachnids. Cambridge Uruv. Press, Cambridge.

Darwin, C. D. 1859 (1998, Modem Library ed.). The Origin of Species. New York.

Dean, A. M. 1983. A simple model of mutualism. Am Nat 121:488-417.

Gentry, A. H. 1978. Anti-pollinators for mass flowering plants? Biotropica 10:68-69.

Hamilton, W. D. 1964. The genetical evolution of social behavior. H. J Theor Biol 7:17-32.

Heinrich, B. and P. H. Raven. 1972. Energetics and pollination ecology. Science 187:597-602. 17 1

Heithaus, E. R. 1979. Flower visitation records and resource overlap of bees and wasps in northwest Costa Rica. Brenesia 16:9-52.

Hunt, J. H. 1999. Trait mapping and salience in the evolution of eusocial vespid wasps. Evolution 53:225-237.

Inouye, D. W. 1983. The ecology of nectar robbing. Pages 153-173 in B. Bentley and T. Elias, eds. The Biology of Nectaries. Columbia University Press, New York.

Keeler, D. 1985. Cost:benefit models of mutualism. Pages 100-127 in D. H. Boucher, ed. The Biology of Mutualism. Croom Helm, London

Maddison, W. P. and D. R. Maddison. 1992. MacClade: Analysis of Phylogeny and Character Evolution. Sinauer and Assoc, Sunderland, Mass..

Mesterton-Gibbons, M. and L. A. Dugatkin. 1992. Cooperation among unrelated individuals: evolutionary factors. Q Rev of Biol 67:267-281.

Michener, C. D. 1969. Comparative social behavior of bees. Arm Rev Ent 14:299-343.

Michener, C. D. 1974. The Social Behavior of the Bees: A Comparative Study. Belknap Press of Harvard Univ. Press, Cambridge, Mass.

Michener, C. D. 2000. The Bees of the World. The Johns Hopkins Uruversity Press, Baltimore.

Mitchell, T. B. 1960. Bees of the Eastern United States, Vol. 1, North Carolina Agricultural Experiment Station.

Mitchell, T. B. 1962. Bees of the Eastern United States, Vol 11, North Carolina Agricultural Experiment Station.

Moldenke, A. R. and J. L. Neff. 1974. The bees of Califomia: a catalogue with special reference to pollination and ecological research. International Biological Program Origin and Structixre of Ecosystems Technical Reports 74-1 to 74-6. Parts I-IV. (not publicly circulated).

Noe, R. and P. Hammerstein. 1994. Biological markets: supply and demand determine the effect of partner choice in cooperation, mutualism, and mating. Behav Ecol Sociobiol 35:1-11.

Omelas, J. F. 1994. Serrate tomia: an adaptation for nectar robbing in hummingbirds. Auk 111:703-710. 172

Pellmyr, O. and C. J. Huth. 1994. Evolutionary stability of mutualism between yxiccas and yucca moths. Nature 372:257-260.

Pellmyr, O., J. LeeberisMack, and C. J. Huth. 1996. Non-mutualistic )aicca moths and their evolutionary consequences. Nature 380:155-156.

Pesenko, Y. A. 1999. Phylogeny and classification of the family HaKctidae revised (Hymenoptera: Apoidea). J Kansas Entomol Soc 72:104-123.

Pierce, N. E. and W. R. Yoimg. 1986. Lycaenid butterflies and ants: two-species stable equilibria in mutualistic, commensal, and parasitic interactions. Am Nat 128:216-227.

PouHn, R. and W. L. Vickery. 1995. Cleaning symbiosis as an evolutionary game: to cheat or not to cheat? J Theor Biol 175:63-70.

Roig-Alsina, A. and C. D. Michener. 1993. Studies of the phylogeny and classification of long-tongued bees (Hymenoptera: Apoidea). The University of Kansas Science Bulletin 55:123-173.

Schwartz, M. P., L. X. SUberbauer, and P. S. Hurst. 1997. Intrinsic and extrinsic factors associated with social evolution in allodapine bees. Pages 333- 346 in J. C. Choe and B. J. Crespi, eds. The Evolution of Social Behavior in Insects and Arachnids. Cambridge Univ. Press, Cambridge.

Schwartz, M. W. and J. D. Hoeksema. 1998. Specialization and resource trade: biological markets as a model of mutualisms. Ecology 79:1029-1038.

Soberon Mainero, J. and C. Martinez del Rio. 1985. Cheating and taking advantage in mutualistic associations. Pages 192-216 in D. H. Boucher, ed. The Biology of Mutualism. Oxford University Press, New York.

Templeton, A. R. eind L. E. Gilbert. 1985. Population genetics and the coevolution of mutualism. Pages 128-144 in D. H. Boucher, ed.. The Biology of Mutualism. Croom Helm, London.

Trivers, R. L. 1971. The evolution of reciprocal altruism. Q Rev Biol 46:35-57.

Zimmerman, M. and S. Cook 1985. PoUinator foraging, experimental nectar- robbing and plant fitness in Impatiens capensis. Amer Midi Nat 113:84- 91.