SOCIALITY IN COBWEB SPIDERS (ANELOSIMUS SPP.): EVOLUTIONARY CONSEQUENCES AND THE ROLE OF PRE-EXISTING TRAITS

by

Kieran Mikhail Samuk

H.B.Sc., University of Toronto, 2008

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF

THE REQUIREMENTS FOR THE DEGREE OF

MASTER OF SCIENCE

in

THE FACULTY OF GRADUATE STUDIES

(Zoology)

THE UNIVERSITY OF BRITISH COLUMBIA

(Vancouver)

July 2011

© Kieran Mikhail Samuk, 2011 Abstract

Sociality – cooperative group living – is ubiquitous in the natural world, yet our understanding of its evolution is still in its infancy. In this thesis, I explore two poorly understood aspects of the evolutionary origin and consequences of sociality using social cobweb spiders (Anelosimus spp.) as a model system. First, I examine how pre-exisiting traits have contributed to the evolution of alloparental care – the care of non-descendant offspring – in social cobweb spiders. I begin by showing alloparental care is extensive in wild social cobweb spider nests. I then test the hypothesis that alloparental care occurs as a result of a lack of discrimination against foreign egg sacs. In support of this hypothesis, I show that subsocial species from clades sister to the social species freely care for foreign egg sacs. This suggests that a lack of offspring discrimination is ancestral to sociality in cobweb spiders. and alloparental care likely emerged spontaneously along with group living. This may have facilitated the evolution of sociality by immediately providing the group-level benefits of alloparental care. Secondly, I examine how social life may have altered natural selection acting on social cobweb spiders. In social cobweb spider nests, the protection offered by a communal nest and the presence of alloparents may have relaxed natural selection on individual maternal care behaviour. Using a comparative approach, I test the hypothesis that sociality is associated with reduced maternal care behavioural phenotypes. I show that social species from independently derived social clades score significantly lower than their subsocial sister taxa on six different assays of maternal care, including the probability of repairing damaged egg sacs and of abandoning egg sacs in the face of simulated predation. Integrating a number of supporting facts, I interpret this result as suggestive of relaxed natural selection on maternal care behaviour as a consequence of sociality. Together, the two comparative studies I present reveal a key role for pre-existing traits in the origin of sociality and that the forces of evolution are likely altered in concert with the onset of social life.

! ii ! Preface

I played the lead role in the two projects described in Chapters 2 and 3 of this thesis.

Along with my supervisor Dr. Leticia Avilés, I conceived and designed the project described in Chapter 2. I performed all field and laboratory work, statistical analyses, preparation of figures, and writing. Gyan Harwood (University of British Columbia) assisted with field and laboratory work. Dr. Avilés helped review and provide comments on the resulting chapter. Chapter 3 was based on a project I had a lead role in conceiving and designing, along with my supervisor. I carried out all fieldwork, statistical analyses, writing and preparation of figures. Emily LeDue (Dalhousie University, second co-author of Chapter 3) and myself performed the laboratory work. Dr. Avilés again helped review and provide comments on the resulting chapter.

! iii ! Table of contents

Abstract ...... ii Preface ...... iii Table of contents ...... iv List of tables ...... v List of figures ...... vi Acknowledgments ...... vii Chapter 1: General introduction Introduction ...... 1 Contents and aims of this thesis ...... 5 Chapter 2: Evidence for the emergent origin of alloparental care in social cobweb spiders Introduction ...... 7 Methods ...... 11 Results ...... 16 Discussion ...... 17 Chapter 3: Sister clade comparisons reveal reduced maternal care behaviour in social cobweb spiders Introduction ...... 28 Methods ...... 31 Results ...... 36 Discussion ...... 37 Chapter 4: General conclusion Summary of findings ...... 50 Future studies ...... 53 Conclusion ...... 54 References ...... 56 Appendices Appendix A: Simulation of estimate accuracy in alloparental care assays ...... 75 Appendix B: The effect of paint marks on spider care behaviour ...... 82 Appendix C: Pairwise tests of maternal behaviour care contrasts ...... 84

! iv ! List of tables

Table 2.1 Effect of maternity on four measures of maternal care in Anelosimus ...... 24

Table 3.1 Comparisons of maternal care between subsocial and social Anelosimus .... 46

Table B.1 The effect of egg sac paint marks on maternal care in Anelosimus ...... 83

Table C.1 Pairwise tests of maternal care behaviour contrasts ...... 84

! v ! List of figures

Figure 2.1 Two pathways for the evolution of alloparental care ...... 25

Figure 2.2 The extent of alloparental care in six species of Anelosimus ...... 26

Figure 2.3 Plots of differences in the level of five Anelosimus care behaviours ...... 27

Figure 3.1 Maternal care measured in natural nests of six species of Anelosimus ...... 47

Figure 3.2 Probability of accepting and repairing egg sacs in six species of Anelosimus . 48

Figure 3.3 Probability of dropping or reclaiming egg sacs in six species of Anelosimus .. 49

Figure A.1 Simulation of estimated rates of egg-sac switching vs. number of females ... 77

!

! ! !

! vi ! Acknowledgements

This thesis could not have been completed without the help and support of many amazing people. First and foremost, I am grateful to my supervisor, Leticia Avilés. Her skillful guidance, knowledge and enthusiasm made working in her laboratory truly a pleasure. Secondly, my committee members Wayne Maddison and Dolph Schluter provided many helpful and insightful comments throughout the process that greatly improved this thesis. Many thanks also to members of the Avilés lab: Jennifer Guevara, Gyan Harwood, Jessica Purcell, Maxence Salomon and Ruth Sharpe, for (1) willingly reading and commenting on early versions of my manuscripts with little to no coercion and (2) being fantastic lab mates, field companions and friends.

I am also deeply grateful to all the incredible, intelligent and hilarious members of Zoology and Botany Departments at UBC. Doing science at UBC is a pleasure and a privilege, and I could not ask for better colleagues and friends. Thanks especially to Aleeza, Alana, Andrea, Brook, Dave, Gerald, Gina, Greg, Jasmine, Jon, JS, Kate, Kathyrn, Matt, Laura, Leithen, Rich, and Sam for many helpful scientific and statistical discussions and general awesomeness.

I am extremely thankful for the financial support provided to me during my degree by the Natural Sciences and Engineering Research Council of Canada (NSERC) and UBC. NSERC funded most of the research in this thesis through a Discovery Grant to Dr. Avilés, and a CGS-M to myself. I would further like to thank the staff of SIMBIOE and El Ministerio del Medio Ambiente de Ecuador for assistance with logistics and obtaining permits, and for the opportunity to do field work in beautiful Ecuador. The associates of Reserva Ecólogica Antisana, Estacion Biológica Jatun Sacha and Bellavista Cloudforest Reserve also all provided fantastic general assistance and field support. Special thanks also to Maurico Vega and Gabriel Iturralde, whose generous help with field work in Ecuador was invaluable to the success of this thesis.

Finally, and most importantly, thank you to my family for supporting and inspiring me always.

! vii ! Chapter 1

General introduction

Sociality, i.e. cooperative group living, has long intrigued biologists. This interest may be somewhat self-reflective: humans are, after all, social animals. That said, the ecological importance and success of social animals is undeniable. Social insects, for example, constitute around 75% of the world’s insect biomass (Holldobler & Wilson

1990). The evolutionary impact of sociality is also widely acknowledged. The transition to social life is often accompanied by profound morphological change, the evolution of novel behaviours, and the alteration of life histories (Keller 2009). In short, sociality has played an important role in shaping both organisms and the ecosystems they inhabit. Accordingly, understanding how sociality arises has become a major focus of evolutionary research over the past fifty years (Foster 2010).

Current approaches

Thus far, two complementary approaches have dominated the study of the evolution of sociality. The first of these approaches focuses on explaining how costly cooperative behaviours (~altruism) can evolve via natural selection. This is the purview of two related theoretical frameworks: kin selection theory and multilevel selection theory

(Price 1970, Hamilton 1972, Okasha 2006, Bshary and Bergmüller 2008). Kin selection theory, introduced by Hamilton (1972), posits that costly cooperative behaviour can evolve if the costs are sufficiently offset by increasing the success of relatives. Hence, kin selection predicts that high relatedness among social partners should in some way facilitate the evolution of sociality. Hundreds of studies have found support for this prediction, and

! 1 ! it is generally agreed upon that kin selection has played an important role in the evolution of social behaviour (Wenseleers et al. 2009).

In contrast, multilevel selection theory posits that individually costly social behaviours can be offset by fitness benefits at higher levels of organization - from gene to cell to individual to group and so on (Okasha 2006). The most commonly discussed (and formerly reviled) incarnation of multilevel selection is group selection: groups of cooperators outperforming groups of selfish individuals, in spite of individual costs

(Williams 1971, Okasha 2006). For example, in a recent empirical study, Kerr et al.

(2006) showed that reduced fecundity (a manifestly costly and cooperative trait) can evolve via group selection in bacteriophage if it prevents overexploitation of resources.

Multilevel selection has also been advanced to explain a variety of explicitly social phenomena, ranging from the evolution of division of labour in social insects, aspects of human sociality, and mating behaviour in water striders (Korb 2010, Wilson et al. 2009,

Eldakar et al. 2009 ). Indeed, it is now uncontroversial that multilevel phenomena have likely contributed to the evolution of cooperation and sociality across disparate taxa

(Eldakar & Wilson 2011, Foster 2010). Together, kin and multilevel selection theory have provided a solid understanding of how costly cooperative behaviours can evolve by way of natural selection.

The second broad approach to studying the evolution of sociality is the ecological approach, which posits that sociality evolves because it is beneficial in certain ecological contexts (Wilson 1975, Emlen 1982, Wilson and Holldobler 2005). In most cases, this approach is complementary, rather than at odds with kin and multilevel selection approaches. The scope of the ecological approach is quite broad, with ecology being invoked to explain evolution of both costly and mutualistic cooperation as well as group living per se (Wilson and Holldobler 2005, Korb 2008).

! 2 ! The ecological approach has been largely successful in explaining a wide variety of social behaviour in disparate taxa. For example, it has been found that birds living in environments where resources or nesting sites are scarce often live in groups and cooperatively breed (Rubenstein & Lovette 2007). The general explanation for this is that the difficulty and danger of independent breeding has caused selection to favour birds that do not disperse over those that do – i.e. selection for philopatry. This results in group living, which in turn provides the opportunity for the evolution of cooperative behaviour via a number of other mechanisms such as reciprocal altruism, kin selection, “pay-to-stay”, etc. (Emlen 1991, Rubenstein & Lovette 2007, Duffy 2010 ).

Another ecological context where sociality can be favoured are environments rich in resources that can only be obtained by a cooperative group (and poor in other resources). For example, socially-living cobweb spiders are able to cooperatively catch larger prey than single female spiders, allowing them to access new resources and persist in novel environments (Guevara & Avilés 2006, Powers & Avilés 2007). There are manifold other examples of sociality as a response to ecological pressure, too many to discuss here

(reviewed in Korb & Heinze 2008). Overall, this body of literature has taught us that ecology plays an important role in the evolution of sociality, separate and unique from the processes described by kin/multilevel selection theory (Wilson 2005). That said, it is generally agreed that the processes and conditions of kin selection, multilevel selection and ecology can act in concert to promote the evolution of sociality (Foster 2010).

Issues with current approaches

Both the kin/multilevel selection and ecological approaches are united in the question they seek to answer: how and when does natural selection favour cooperative behaviours? However, this framework leaves us with a number of unanswered questions.

! 3 ! For one, if it were the case that the evolution of sociality simply required the appropriate amount of relatedness between interacting individuals or the correct ecological pressures, we would expect sociality to predominate in certain environments and in species where reladness is enhanced (e.g. clonal or highly inbred species). In fact, sociality has a strangely patchy distribution with respect to relatedness and ecology. For example, sociality in clonal aphids – where genetic relatedness between individuals is 100% – is strangely rare, occurring in only ~1% of species (Pike and Foster 2008). Further, there are many cosmopolitan groups (e.g. frogs) inhabiting environments in which sociality is predicted to be highly adaptive that have apparently never evolved sociality (Wilson 1975, Duffy 2010).

What is the cause of these discrepancies?

One promising idea is that that certain pre-existing traits – those evolved prior to the evolution of sociality – can predispose animal lineages to later evolving full-blown sociality (Hunt 1999, Page 2002, Agnarsson et al. 2002). In other words, sociality may be historically contingent on certain precursory traits. For example, the extended maternal care of live offspring –“subsociality” – appears to be ancestral to sociality in various insect and mammal lineages (Hunt 1999, Wcislo 2000, Agnarsson et al. 2002, Foster 2010).

One reason for this may simply be that maternal care creates scope for the evolution of social interactions between siblings, which are later elaborated into advanced social systems (Foster 2010). Another possibility, suggested by some authors, is that many social behaviours, such as the cooperative care of young, evolve via the co-option of pre-existing mother-offspring behaviours (Toth and Robinson 2007). Nevertheless, compared to the two dominant approaches, investigation of the role of historical contingency in the evolution of sociality has been extremely limited.

Secondly, while there has been a proliferation of research into the evolutionary causes of sociality, the evolutionary consequences of sociality have been much less

! 4 ! explored. Can the establishment of social life alter the prime forces of evolution? It seems reasonable that this must be at least partly true, as sociality involves changes in population structure and density, rates of migration/inbreeding, predation risk and possibly even mutation rate per se (Wilson 1975, Broham & Leys 2005, Hollis et al. 2011). Indeed, there have been a number of theoretical studies that predict changes in the forces of evolution, especially natural selection, as a consequence of social life (Wolf et al. 1999,

McGlothlin et al. 2010). Yet, this topic has received relatively little attention from empiricists (Cheverud 2003). Considering the importance of sociality in the grand scheme of evolution, furthering our understanding of its consequences is essential.

Contents and aims of this thesis

In light of the two above issues, in this thesis I attempt to answer two broad questions. First, how do preexisting traits contribute to the evolution of sociality?

Secondly, how can sociality affect the evolutionary process? Throughout, I employ cobweb spiders in the genus Anelosimus Simon 1891 (Araneae: Theridiidae) as a model system.

This group of spiders is useful for answering the above questions for two reasons. First, multiple species in this genus have independently evolved sociality. Thus, analogous to tests of association between phenotype and environment (e.g. dense fur and cold temperatures), we can test if certain phenotypes convergently evolve after the establishment of sociality. This gives us information about how the forces of evolution may change as a consequence of sociality. Secondly, each social species has at least one extant subsocial sister species. Hence, the ancestral phenotypic states of social species can be approximated by studying their subsocial sister species. This allows us to examine how pre-existing traits may have contributed to the repeated evolution of sociality in this group.

! 5 ! I begin in Chapter 2 by exploring how pre-existing maternal care behaviour can predispose social cobweb spiders to perform alloparental care, i.e. the cooperative rearing of young. I show that an ancestral absence of offspring recognition probably resulted in the spontaneous emergence of alloparental care coincident with the evolution of sociality.

This is consistent with the idea that specific pre-existing traits are involved in the transition to sociality. In Chapter 3, I go on to explore how individual maternal care behaviours may have undergone further evolutionary change in social cobweb spiders as a consequence of sociality itself. I show that reduced levels of maternal care behaviours have likely convergently evolved in two independently derived groups of social species. I present evidence suggesting that this is likely a result of a relaxation of natural selection coincident with the evolution of sociality. I close in Chapter 4 with a general discussion of my findings, their contribution to the field, and future directions of research.

The general aim of this thesis is to expand our knowledge of the origin and consequences of sociality, both in cobweb spiders and in animals in general. The evolution of sociality has been a major event in the history of life. Yet, we are just beginning to appreciate the scope of its ecological and continued evolutionary impacts. By moving beyond the traditional approaches to studying this event, I hope to contribute to a more complete understanding of the role of sociality in the natural world.

! !

! ! !

!

!

! 6 ! Chapter 2

Evidence for the emergent origin of alloparental care in social cobweb spiders

Introduction

Alloparental care, i.e. caring for nondescendent offspring, occurs widely among social animals. It encompasses phenomena such as sibling-sibling care in cooperatively breeding birds, allosuckling in sperm whales, and sterile care-giving castes in eusocial insects

(Wilson 1975). Over the last 40 years, this intriguing behaviour has attracted the interest of biologists for two principal reasons. First, along with group living, alloparental care is considered one of the essential aspects of advanced sociality (Crespi & Yanega 1995,

Burda et al. 2000). Hence, understanding the evolution of alloparental care is a key component in understanding the evolution of cooperative living generally. Secondly, alloparental care presents a Darwinian puzzle – why do alloparents invest resources in another’s offspring when those resources could be allocated to their own offspring?

A common approach to studying alloparental care is to measure its adaptive value, i.e. how it is maintained by natural selection (Heinsohn and Legge 1999). This approach has been fruitful, and many authors have shown that the direct and indirect fitness benefits of alloparental care can be substantial (Emlen et al. 1991). For example, individuals can benefit directly from alloparenting by gaining parenting practice, ensuring later access to breeding territories (“pay-to-stay”), or receiving alloparental care for their own offspring in the future (reciprocal altruism) (Kokko et al. 2002, Heg et al. 2009) . On the other hand, alloparents can benefit indirectly by increasing the fitness of their relatives beyond

! 7 ! the fitness cost of alloparenting, i.e. via kin selection (Russell & Hatchwell 2001,

Hatchwell 2009). That said, some authors have failed to find any adaptive benefit of alloparental care whatsoever (e.g. Brown 1998, Dugdale et al. 2008).

However, focusing solely on why contemporary selection does or does not maintain alloparental care has told us little about its evolutionary origin. Specifically, it still unknown whether alloparental care generally originates de novo or is simply a consequence of pre-existing parental care coupled with a lack of discrimination against foreign offspring (outlined in Jamieson 1989). Distinguishing between these alternatives can only be achieved with an explicit phylogenetic approach, e.g. independent contrasts or comparisons of sister taxa, and not by measuring contemporary selection (Gould &

Lewontin 1979, Gould 1997). In spite of Jamieson & Craig (1987) and Emlen (1991) calling attention to this issue over twenty years ago, there have been surprisingly few studies that test hypotheses regarding the historical origin of alloparental care. Indeed, many studies continue to conflate the question of how alloparental care is maintained by selection with the question of how it originated.

Here, we explore the extent and origin of alloparental care in cobweb spiders,

Anelosimus spp. Simon 1891 (Araneae: Theridiidae). We focus on species with two types of social organization: subsocial and social. Subsocial Anelosimus adult females live alone in single-individual nests in which their offspring hatch, fledge and eventually disperse

(Avilés 1997). In contrast, social Anelosimus do not obligately disperse – they live in group nests as adults, cooperatively building, hunting and likely performing alloparental care

(Avilés 1997). Anelosimus are particularly useful for this investigation, as sociality

(putatively in concert with alloparental care) is thought to have independently evolved from a subsocial state at least four times in the genus (Agnarsson 2006; Agnarsson et al.

! 8 ! 2006; Agnarsson et al. 2007). Further, each social species has extant subsocial sister taxa.

We can hence employ the comparative method to test evolutionary hypotheses regarding the origin of alloparental care by contrasting social species with their extant subsocial sister groups (Agnarsson et al. 2007).

We had two objectives in this study. First, we set out to confirm the occurrence of alloparental care in social Anelosimus and quantitatively describe how it varies between species. While alloparental care is thought to occur in ~11 species of communal-living spiders, quantitative studies of this behaviour in the wild are absent from the literature, with the majority of reports being largely qualitative or anecdotal (Lubin & Bilde 2007 and references therin). This is also case for most studies of Anelosimus (Christenson 1984,

Furey 1998, but see Jones & Riechert 2007). Yet, asking questions about the evolutionary and ecological consequences of alloparental care minimally requires quantitative knowledge of its extent (e.g. in order to estimate resource allocation to own versus foreign offspring). Here, we specifically focus on the alloparental care of egg sacs – silken structures that contain the entire clutch of a single female (~a few-dozens eggs, depending on the species; Foelix 1996). We describe the extent of the alloparental care of egg sacs in

3 social species, and also the extent of maternal care of egg sacs in 3 subsocial species for comparison and as procedure controls.

Secondly, we set out to identify the evolutionary route by which alloparental care originated in Anelosimus. Alloparental care behaviour can originate via two distinct pathways, each resulting in a unique phylogenetic distribution of the trait (Figure 2.1; see

Jamieson 1989 and Emlen et al. 1991 for original non-phylogenetic formulation). First, alloparental care can arise via de novo behavioural modification after or in concert with the evolution of group living (Figure 2.1B). In this scenario, willingness to care for foreign

! 9 ! offspring evolves via modification of the existing parental-care stimulus response system to accept cues from foreign offspring. This could occur due to selection for the adaptive reasons previously mentioned or simply drift. Alternatively, alloparental care can emerge as a byproduct of parental care as in Figure 2.1A (Williams 1966, Coyne 1979, Jamieson

1989). In this scenario, the similarity in appearance, begging calls, etc. of foreign offspring to the care-giver’s own offspring elicits a care response, and no special behavioural modification is required for the expression of alloparental care. This is sometimes known as “alloparental care by mistaken identity” (Williams 1966, Coyne 1977, Kohn 2006).

This would class alloparental care as an exaptation – a previously evolved character

(maternal care) instantaneously co-opted for a new function (alloparental care) in a new context (group living) (Gould and Vrba 1982). Under this route, alloparental care behaviour and parental care behaviour are the same trait (i.e. alloparental care is an exaptation), and are basal to the evolution of grouping (Figure 2.1A).

We tested between the de novo and emergent/exaptive routes by comparing the degree of maternal care behaviour that sister pairs of subsocial and social cobweb spider species performed when given their own egg sac or that of another female. If alloparental care evolved as a byproduct of indiscriminate parental care, we expected subsocial and social spiders to provide the same amount of care for their own and foreign egg sacs

(Figure 2.1A). Alternatively, if alloparental evolved as a separate trait unique to social species, we expected subsocial species to provide more care for their own egg sac, and social species to provide the same amount of care for all egg sacs (Figure 2.1B).

! 10 ! Methods

Study species

We focused on a subclade of the genus Anelosimus, shown in Figure 2.1. For our assays of alloparental care, we collected data on all species shown in Figure 2.1. For our assays of egg sac treatment we focused on two pairs of sister groups: subsocial A. elegans and its social sister species A. guacamayos and subsocial A. baeza and a social species A. eximius from a clade sister to that of A. baeza.

Field sites

We performed our work at four field sites in eastern Ecuador. We studied A. eximius and A. domingo at the Jatun Sacha Biological Station (lowland rainforest, 400 m elevation, 1·06°S, 77·61°W), A. guacamayos and A. elegans in the Reserva Ecológica

Antisana and Parque Nacional Sumaco (cloudforest, 1840 m elevation, 0·64°S, 77·8°W),

A. baeza at the Yanayacu Biological Station (cloudforest, 2200 m elevation,

0·60°S ,77·89°W), and A. cf. oritoyacu at the Bellavista Cloudforest Reserve near

Tandayapa (2000 m elevation, 0.016°S, 78.68°W).

Field assays of alloparental care

Our first objective was to confirm the existence and quantitatively estimate the extent of alloparental care between Anelosimus species. We measured the extent of alloparental care as the probability of observing a different individual caring for an egg sac in a subsequent observation – hereafter referred to as “switching probability”. This also provides, by extension, the probability that the same individual will remain with the same egg sac between subsequent observations (“remaining probability”), also an informative quantity.

! 11 ! We located and marked colonies of each species that contained females clutching egg sacs. Due to methodological constraints, we could not mark every female in nests with more than ten females. We thus sub-sampled larger nests using a transect method: for each

10 cm mark along the longest, geometrically centred axis of the colony, we selected the closest female clutching an egg sac. We removed these females and their clutched egg sac from the nest using a drinking-straw aspirator, and marked them with matching acrylic paint marks. After the paint had dried, we replaced the female and her egg sac at the approximate site of their removal.

Following marking, we returned to each colony approximately nine times over three days, distributed equally over morning (0600-1200), afternoon (1300-1600) and evening (1800-2300) observation periods. For each marked egg we were able to locate, we recorded whether it was being clutched, and the identity of the female clutching it (her mark colour or “unmarked”). We calculated switching probability using only meaningful transitions in female identity between two consecutive observations, i.e. we excluded unmarked female to unmarked female transitions. According to simulation results, this method generally results in conservative estimates of the switching rate, but also eliminates a large portion of the data (see Appendix A for simulation details and discussion).

Note that we performed this assay in nests of social species in which there are multiple females and egg sacs, and nests of subsocial species where there is typically only a single female and egg sac. For subsocial species, this protocol results in a female's single egg sac being removed, painted, and returned. We did this for a number of reasons. First, switching between egg sacs is not completely out of the question for subsocial females nesting close to one another (e.g. on the same plant, which is sometimes the case).

Secondly, this acts as a procedure control for the effect of paint marks on the egg sac care behaviour of females (we also systematically tested for this effect, see Appendix B). Finally,

! 12 ! it allowed us to compare the extent of care provide to an egg sac in social and subsocial nests.

Egg sac preference

Our second objective was to test the phylogenetic predictions of the de novo vs. emergent/exaptive origin hypotheses for the evolution of alloparental care (Figure 2.1).

Specifically, we sought to determine whether social and subsocial spiders provide the same level of care for foreign egg sacs as they do for their own egg sacs. Because we needed to be certain of maternity, we housed captured gravid females of social species until they produced egg sacs.

Rearing protocol

We collected female spiders that appeared gravid (i.e. had swollen abdomens with darkened sagittal spots) or had egg sacs (subsocial species only) from nests of A. elegans, A. guacamayos, A. baeza and A. eximius. In total we collected the following number of females: 32 A. elegans, 52 A. guacamayos (approximately three females from each of 17 different colonies), 58 A. baeza and 40 A. eximius (approximately seven females from each of six different colonies). We deposited these spiders in cylindrical plastic containers with a height of 5.5 cm and a circumference of 11 cm. Each container contained a ~0.2 cm wide by 10 cm long twig placed diagonally, and two 9x3 cm leaves obtained from various plant species we observed Anelosimus spp. nesting on. Leaves were placed over and under the twig, forming a retreat and nesting substrate. Containers were topped with perforated plastic lids with approximately eight 0.1 cm holes/cm2.

We opened each container daily, cleared it of detritus and fungus, and administered a fine misting of water from a spray bottle. Every other day, we fed spiders insects sized approximately one-half their body length (0.2-0.8 cm, depending on the

! 13 ! species). Insects were primarily small dipterans, lepidopterans, orthopterans and homopterans. We captured insects either by blacklighting or beating. After ~30 days of feeding, the following number of females produced egg sacs: 21/32 A. elegans, 32/52 A. guacamayos, 24/58 A. baeza and 11/40 A. eximius. All of these females were entered into the experiment described below.

Preference assay

We began our assay by removing egg sacs from females entering the experiment. In most cases, this involved separating the egg sac from the spider by gently gripping it with

FeatherWeight forceps (BioQuip products, California) and jiggling it free of the spider’s mouthparts. After a ten minute cool-down period, we presented females with their own egg sac or the egg sac of a conspecific (separately collected from field colonies at least 500 m away). To control for possible treatment order effects, we randomly chose half of the females to receive their own egg sac first, and the other half to receive the egg sac of a conspecific first. We repeated the assay described below at least 48 hours later with the opposite type of egg sac (own or conspecific, whichever the female had not yet received).

Egg sacs were used only once for each experiment and returned to their nest of origin after use.

We presented egg sacs to the females by placing them in the geometric centre of the nest (the intersection point of the longest two axes). We regarded this as a suitably natural way for a female to encounter and choose to care for an egg sac. We returned to each container at 11 observation times: immediately after introduction, then at three half-hour- long intervals between 0.5-1.5 hours, six hour-long intervals between 2.5-7.5 hours and a final observation at 24 hours after introduction. Based on a pilot study, this observation schedule allowed us to best capture the time course of the care decision.

! 14 ! At each observation time we measured: the distance between the egg sac and female, whether the egg sac was under cover (not visible from directly above), and whether the female was caring for the egg sac (clutching it with her palps/chelicerae). These metrics and the combinations of them described below were designed to capture the general protective functions (anti-predatory/parasite/fungus and thermoregulatory) of spider egg sac care described by Foelix (1996).

We used a linear modeling approach to test for the effects of egg maternity on five dimensions of maternal care derived from our experiment: (i) probability of accepting the egg sac (0,1), (ii) latency to accepting the egg sac, (iii) mean egg sac-female distance, (iv) proportion of observations in which the female was exhibiting care behaviour and (v) proportion of observations in which the egg sac was under cover. We only analyzed responses ii-v for females who had accepted both their own and foreign egg sac presented to them in separate trials (~95% of the females). We transformed responses ii-v to ensure normality of model residuals (see Table 2.1 for transformations). Because the response for variable i was binary, we used a mixed-effects logistic regression in lieu of standard mixed- effects regression for this analysis.

For each response variable, we fit generalized linear mixed effects models using the lme and lmer functions in the lme4 package in R (R core team 2009, Bates et al. 2011).

Because it was not standardized per se in the experiment, initial egg to female distance was included as a statistical control in all models. We also included the effect of species as the first term in every model to account for between-species variance in the absolute amount of care provided to egg sacs. Finally, each model included two random effects: source nest identity (to address potential non-independence of related females from the same nest) and female identity nested in source nest identity to account for repeated measures. Thus, each model followed the general formula of: response = species + initial egg sac distance + egg

! 15 ! maternity + source nest ID/female ID (random effects). We used F-tests (responses ii-v) or likelihood ratio tests (response i) to assess the significance of the fixed effects of the various models.

Results

Alloparental care

In total, we observed 723 meaningful egg-sac transition events (Figure 2.2). The difference in sample sizes between species was primarily due to the difficulty of marking and observing egg sacs in the convoluted nests of A. eximius, and the general rarity of A. eximius in 2009-2010 at our field sites.

With the exception of one out of ten A. cf. oritoyacu females, all subsocial females

(A. elegans, A. baeza, A. cf. oritoyacu) and all single A. guacamayos females remained with their egg sacs during the entire observation period (Figure 2.2). A. guacamayos females in multi-female nests remained with their egg sac 90% of the time. In extreme contrast, A. eximius and A. domingo females only remained with the same egg sac in ~3% of meaningful observations (Figure 2.2).

We observed high probabilities of female egg-sac switching behaviour in all three social species (Figure 2.2). Interestingly, A. domingo and A. eximius females switched egg sacs approximately three (36%) and five (55%) times more often than A. guacamayos females in multi-female nests (12%, Figure 2.2). Surprisingly, egg sacs in A. eximius and A. domingo were found without females caring for them approximately 70 (42%) and 100

(60%) times more often than A. guacamayos females in multi-female nests (0.06%).

Egg sac preference

After accounting for between-species differences and the effect of initial egg sac to female distance, there was no significant difference between the level of care provided by

! 16 ! female Anelosimus to their own egg sac and foreign egg sacs (Table 2.1, Figure 2.3). This was true of all five measures of maternal care. No two-way interaction effects (species x treatment, species x initial distance, initial distance x treatment) were significant in any of the five models, and were hence removed prior to F-tests or likelihood ratio tests (Crawley

2002).

Discussion

In our field assays, we found that switching between egg sacs was common in all of the social Anelosimus species we examined. The extent of this behaviour differed substantially between species; A. eximius and A. domingo females switched between egg sacs very frequently, while group-living A. guacamayos females rarely switched egg sacs.

In our experimental assay of egg sac preference, we found no difference in female’s treatment of their own egg sac and that of a conspecific. Our results are thus consistent with the phylogenetic distribution of alloparental willingess predicted under the emergent/exaptive pathway for the origin of alloparental care (Figure 2.1A), and we can accordingly reject the alternate hypothesis of de novo behavioural evolution.

Alloparental care in Anelosimus

Our results, along with previous studies by Furey (1998) and Jones & Riechert

(2007) provide a comprehensive picture of alloparental care behaviour in Anelosimus.

This greatly furthers our understanding of alloparental care in spiders, as our study is the first to quantitatively estimate alloparental care in the wild, and include multiple social and related subsocial species. In contrast, previous studies have been mostly anecdotal and focused largely on single social species (reviewed in Lubin & Bilde 2007). A notable exception is the work of Salomon and Lubin (2007) who quantitatively assessed the effect of sacrificial alloparental care (matriphagy) on the reproductive success of female

! 17 ! Stegodyphus dumicola (Eresidae). That said, while this presumably occurs in the wild, its frequency is not known. Interestingly, Kurpick (2002) reported that, at least in the case of egg sacs, alloparental care does not occur in field colonies of S. dumicola.

These findings raise a number of interesting questions about the nature of alloparental care in Anelosimus. For one, why does the extent of alloparental care differ so greatly between A. guacamayos and A. eximius/A. domingo (Figure 2.2)? One possibility is that a lower rate of alloparenting is a corollary of a generally lower level of sociality in

A. guacamayos. Specifically, A. guacamayos females are known to be more territorial, more likely to establish solitary nests, and when in groups physically space themselves and their eggs farther apart than A. eximius and A. domingo (Avilés et al. 2007, K Samuk unpublished data). So, in spite of the willingness to care for foreign offspring, nesting and behavioural differences probably spatially limit the opportunities for alloparenting in A. guacamayos.

Another interesting pattern we observed was that social species often leave their egg sacs unguarded for large periods of time, whereas subsocial species generally do not

(Figure 2.2). This is consistent with our findings in another study that social species exhibit reduced individual maternal care behaviour relative to subsocial species (Chapter

3). We believe this is probably indicative of relaxed natural selection on maternal care behaviour in social species, likely caused by a reduction in the risk to egg sacs in social nests (see Chapter 3).

Our field assays of alloparental care should be interpreted with one important caveat: in social nests, we were unable to explicitly identify which egg sacs belonged to which females. Maternity for a given egg sac can only be definitively assigned if the female spins the sac in isolation (as in our experiment), or the egg sacs and females are destructively sampled for genetic analysis. However, we do not believe this gravely affects

! 18 ! the validity of our results. Because egg sacs contain the eggs of a single female, any degree of switching is directly indicative of alloparental care. That said, one dimension of alloparental care our assay did not capture is the relative allocation of time to female’s own egg sac and that of others. For example, we do not know if the small amount of switching in A. guacamayos multi-female nests (Figure 2.2) is due to females remaining with their own egg sacs for longer than others, or simply infrequent (but unbiased) egg sac switching.

Origin of alloparental care via mistaken identity

Our assays of alloparental willingness are consistent with the emergent/exaptive hypothesis for the origin of alloparental care. Hence, it appears that alloparental care behaviour, i.e. the willingness to care for foreign offspring given the opportunity to do so, is in fact basal to the evolution of sociality per se in cobweb spiders (Figure 2.1A). In other words, alloparental care in social cobweb spider nests probably required no special behavioural evolution and emerged instantaneously as an epiphenomenon coincident with the evolution of grouping behaviour.

This raises a number of fascinating prospects. For one, if the cooperative care of offspring is in any way advantageous, i.e. via increased inclusive fitness or a nest-wide reduction in risk to egg sacs, it would provide an instant benefit to spiders who have evolved grouping behaviour. This could in turn facilitate the evolution of sociality by ameliorating other costs of group life (e.g. increased competition). This could help explain the unusually frequent evolution of social behaviour in Anelosimus – 4-6 of 18 known independent origins of social behaviour in spiders have occurred in Anelosimus

(Agnarsson et al. 2006). Indeed, this is precisely consistent with the idea suggested by

Agnarsson (2002) that the extended maternal care behaviour exhibited by subsocial

! 19 ! cobweb spiders may have somehow potentiated Anelosimus lineages to later evolve social behaviour (similar to the original argument by Wheeler 1928). More generally, our results mesh well with the ideas discussed by Hunt (1999) that suites of pre-exisiting traits (and specifically maternal care behaviour) can predispose lineages to evolve complex social behaviour. For example, there is good evidence from studies of paper wasps that sterile

“nurse” workers behaviourally and neurochemically recapitulate previously evolved maternal care behaviours when caring for non-descendant offspring (Toth et al. 2007).

While Anelosimus do not have sterile worker castes, the conceptual similarities between this result and the present study are striking.

Our study is in line with a growing body of evidence for the frequent origin of alloparental care via exaptation in social organisms. Our results specifically expand on and mirror those of Furey (1998), who showed that when given a choice, A. studiosus females clutch own and foreign egg sacs with equal frequency. There are numerous examples of this in the literature. Some other examples in the literature include: Price et al. (1983) who suggest that alloparental care in Darwin’s finches is a result of misdirected parental care by parents responding to the begging calls of nearby chicks; Brown (1998) who showed that ring-billed gull alloparents help foreign chicks due to physical similarity to their own chicks; and Maniscalco et al. (2006) who show that Stellar sea lions allonurse foreign young indiscriminately, again ostensibly due to recognition errors. Similar results have been reported for rheas (Codenotti and Alvarez 1998), lesser kestrels (Tella et al.

1997), and kittiwakes (Roberts and Hatch 1994), polycheate worms (Premoli and Sella

1995), and numerous mammals (Roulin 2002). Together, these studies all suggest that alloparental care may commonly originate via exaptation.

That said, it may seem counter-intuitive that so many animals do not discriminate to some extent between their own and foreign offspring. However, we might actually

! 20 ! expect this to be the case in most solitary species (Wisenden 1999). This is because most solitary-living animals likely rarely encounter foreign offspring in the wild and hence there would simply be no selection for such fine-scale discrimination (Wisenden 1999). In fact, there should probably be selection against any trait that comes with even a small risk of rejecting your own offspring – potentially true of any rudimentary discrimination system

(Tella et al. 1997). Hence, we expect that a lack of strong offspring discrimination is probably common. Indeed, among spiders a lack of offspring discrimination appears to be the rule rather than the exception. This is especially true of the care of egg sacs. For example, Japyassu et al. (2003) found that female Loxosceles gaucho (Sicariidae) spiders care equally for their own and experimentally introduced egg conspecific egg sacs. In a similar study, Opell (2001) found that female Miagrammopes animotus (Uloboridae) spiders willingly accept and clutch egg sacs of not only other females, but also those of the congeneric M. pinopus. In an extreme case, Culley et al. 2010 observed that female

Pardosa milvina wolf spiders (Lycosidae) will accept and care for not only conspecific egg sacs, but various plastic orbs and polyhedra of approximately egg-sac size and mass.

Our experimental results have one important caveat: by design, we focused solely on female spiders who had laid egg sacs. Thus, our data are silent on the nature and origin of any alloparental care performed by pre-reproductive females. However, this does not affect the conclusion that alloparental care did not require any specific behavioural evolution in social Anelosimus, as there are almost always multiple reproductive females in any nest. Further, there is evidence that pre-reproductive female spiders generally do not provide care to spiderlings (Schneider 2002), which is consistent with our field and laboratory observations of egg sac care in Anelosimus (K Samuk & L Avilés, unpublished data). That said, it has also been shown that the sustained presence of spiderlings and egg sacs can induce care behaviour in pre-reproductive female night spiders (Amaurobiidae,

! 21 ! Roland et al. 1996). This raises the tantalizing possibility that along with alloparental care per se, group living could automatically elicit a shift in the ontogenetic timing of the expression of maternal care behaviour (emergent behavioural heterochrony). Obviously, this could also be caused by selection for earlier expression of maternal care behaviour.

Further investigation of this possibility in social Anelosimus is needed, especially given that alloparenting is known to be an adaptive strategy for non-breeders in other systems

(Emlen 1991).

Maintenance and origin of alloparental care

The origin of alloparental care via mistaken identity in any given system does not mean it is necessarily non-adaptive in that system – these are separate issues. Yet, in many studies of alloparental care, mistaken identity is often used as a null hypothesis to be tested against other competing adaptive hypotheses (e.g. kin selection, Bohn et al. 2007). This confuses levels of analysis by conflating the origin of alloparental care with its maintenance (Heinsohn and Legge 1999). Mistaken identity is one possible explanation for the origin of alloparental care, but does not preclude later adaptive modification of the behaviour, e.g. to preferentially alloparent relatives (Emlen 1991). In other words, demonstrating the adaptive benefit of alloparental care behaviour is not sufficient evidence to reject mistaken identity as an underlying explanation for the origin alloparental care.

One hypothesis for the role of mistaken identity in the maintenance of alloparental care could be that alloparental care via mistaken identity is maintained because the benefits of communal nesting outweigh the costs of alloparenting (as discussed in Tella et al. 1997,

Heinsohn and Legge 1999).

The specific adaptive value of alloparental care behaviour in three social species we studied is as of yet unknown. However, based on the results of Jones & Riechert (2007,

! 22 ! 2008) cooperative care of offspring by female Anelosimus is likely adaptive at the group level when there is a high probability of females dying before their brood is reared (the case at high latitudes, for example). In other words, groups that perform alloparental care outcompete individuals who do not, in certain circumstances (Jones & Riechery 2007). At the individual level, the case is less clear-cut. The energetic costs of performing egg-sac clutching behaviour are probably quite low and high-relatedness among colony members may buffer any loss of fitness resulting from the behaviour (i.e. it is converted into indirect fitness instead of being lost altogether). Hence, alloparental care is likely not very costly.

Indeed, female spiders may actually increase their inclusive fitness by alloparenting their sister’s egg sacs. Further manipulative studies are needed to directly examine the individual adaptive benefit (if any) of alloparental care in social Anelosimus.

Choice vs. no-choice

One potential criticism of this study is that our preference assay was a “no-choice” experiment, and is hence not compelling evidence that alloparental care emerged via mistaken identity in social Anelosimus. This would presumably be predicated on the possibility that a female Anelosimus may not care for a foreign egg sac if her own sac was present. If this was the case, alloparental care would need to evolve via elimination of this discrimination after the establishment of sociality (de novo behavioural modification).

However, even a strong preference for their own egg sag, e.g. 80% of a female’s care-time dedicated to caring for her own egg sac, would result in alloparental care as a byproduct. So, the only way that that a choice experiment could affect the conclusions we draw here is if subsocial females with egg sacs offer no care whatsoever to foreign egg sacs.

The existence of such extreme discriminatory behaviour seems unlikely. First, subsocial female cobweb spiders only ever have one egg sac and tend to live far apart from one

! 23 ! another. Thus, we would not expect extreme exclusionary egg sac discrimination behaviour to exist, as there would be no reason for it to be selected for (at least until sociality is established) and the costs of potentially rejecting one’s own offspring are extremely high. Secondly, there is direct evidence that subsocial female cobweb spiders with egg sacs do indeed willingly care for foreign egg sacs, to the point of stealing them from other females (Downes 1984).

Conclusion

We have shown that alloparental care in social Anelosimus is pervasive and variable between species. Between species variance in alloparental care may be a result of differences in the level of social behaviour (probability of solitary nesting; inter-individual spacing). We have also demonstrated that alloparental care behaviour in Anelosimus is likely a result of pre-exisiting parental care behaviour being elicited by foreign offspring

(“mistaken identity”). Thus, alloparental care in Anelosimus likely emerged instantaneously, coincident with the evolution of communal nesting. Taken in phylogenetic context, out results suggest that the unusually frequent evolution of social behaviour in Anelosimus may in part be explained by the co-optive potential of preexisting behavioural traits, in this case indiscriminate maternal care.

! !

! ! ! !

! 24 !

Table 2.1 Likelihood ratio tests (acceptance probability) and F-tests (all other responses) of fixed effects derived from generalized linear mixed effect models of four measures of maternal care in experimental Anelosimus females. Acceptance probability: probability of accepting an egg sac when presented with one; Acceptance latency: time female took to accept egg sac, given that she did; Distance: egg sac to female distance. Prop. time caring: proportion of observations female was clutching egg sac; Prop. time covered: proportion of observations in which the egg sac was under cover (not visible from a top-down view). Slashes between random effects indicate nestedness of the second term in the first. See text for analysis details.

Response: Acceptance probability Acceptance latency Distance Prop. time caring Prop. time covered

Transformation – Reciprocal Box-cox Arcsine √ Arcsine √ Distribution Binomial Gaussian Gaussian Gaussian Gaussian

Random effects: Source/Individual Source/Individual Source/Individual Source/Individual Source/Individual

Fixed effects df !2 p df F p df F p df F p df F p

Species 6 21.05 <0.001 44 114.47 <0.001 21 74.04 <0.001 21 254.68 <0.001 21 44.26 <0.001 Initial Distance 7 1.79 0.18 41 8.73 0.005 89 20.17 <0.001 89 10.12 0.002 89 2.08 0.15 Treatment 8 0.38 0.54 41 0.32 0.57 89 0.47 0.49 89 0.97 0.33 89 0.44 0.51

! ! !

!

! 24 !

Figure 2.1 Two pathways for the evolution of alloparental care in social cobweb spiders (Anelosimus spp.) and corresponding expected phylogenetic distributions of alloparental willingness (the willingness of an individual to care for offspring other than its own). Trait symbols on the phylogeny correspond to behaviours in the schematics (stars willingness, squares group living). Independent origins of group living

(squares) are inferred from the complete phylogeny (Agnarsson 2007). Under the emergence scenario, willingness is pre-existing (due to e.g. lack of discrimination against foreign offspring) and we expect alloparental willingness to exist in solitary-breeding as well as group-living individuals. Under the de novo evolution scenario, willingness is derived after (or in concert with) the evolution of group living.

Bold species indicate species assayed for willingness in the experimental section of the study.

! 25 !

Figure 2.2 The extent of alloparental care in six species of cobweb spiders (Anelosimus spp.). Stacked bars represent species-wide average probabilities of observing each indicated behaviour in subsequent observations of an egg sac.

Behaviours: “Remain” – the identity of the female caring for an egg sac remains the same; “Switch” – the identity of the female caring for an egg sac changes; “Depart” – the female previously caring for the egg sac departs and no other female takes her place (i.e. a lone egg sac). A. guacamayos data are divided into observations of single female nests and multi-female nests. Squares denote independent origins of sociality inferred from the complete phylogeny

(Agnarsson et al. 2006). Sample sizes indicate the total number of observations summed across all egg sacs, with the number of nests in parentheses.

! 26 !

Figure 2.3 Dot plot (A) and boxplots (B-D) of differences in the level of five care behaviours provided by Anelosimus females to their own versus foreign egg sacs. Dependent axes display the level of care behaviour females provided to their own egg sac minus the level provided to a foreign egg sac in a different trial. Horizontal dotted lines indicate the expected score when there is no difference in egg sac care. Species labels correspond to EL - A. elegans, GU - A. guacamayos, BA - A. baeza, EX - A. eximius. Error bars in A represent 95% confidence intervals. Dashed lines (A) and shaded boxed (B-E) indicate social species. Solid lines (A) and white boxes (B-E) indicate subsocial species.

! 27 ! Chapter 3

Sister clade comparisons reveal reduced maternal care behaviour in social cobweb spiders

Introduction

The evolution of group living and cooperation (sociality) carries with it manifold changes to the environmental context individuals are exposed to (Wilson 1975). These can include a higher frequency of interaction with conspecifics, reduced predation (Elgar 1989, Beauchamp

2008), buffering of temperature changes (Willis & Brigham 2008), as well as increased food availability (Sonerud et al. 2001), mating opportunities (Bijleveld et al. 2010) and resting time

(Pollard & Blumstein 2008). Whatever their specific nature, these changes can result in fundamentally different environments for individuals in species with different levels of sociality. This can in turn have profound consequences for the expression and evolution of their individual and collective behaviours. Indeed, many authors believe that differences in social context have the potential to explain a great deal of behavioural variation in nature – a major goal of both evolutionary biology and ecology (West-Eberhard 1983, Wcislo 2000,

Keller 2009).

Sociality can influence behaviour via two interacting processes: plasticity and genetic evolution. First, plastic behaviours that rely on cues modified by the social environment can be differentially expressed across social contexts. For example, it is well known that many animals will reduce their level of anti-predatory behaviour (e.g. vigilance) when in larger groups (Lima 1995). Similar facultative responses to changes in social context have been widely reported for various behaviours in disparate taxa, including mating behaviour in

! 28 ! primates and vinegar flies (Dufty et al. 2002, Krupp et al. 2008), foraging behaviour in salmonids (Grand and Dill 1999), calling frequency in frogs (Chu et al. 1998) and dispersal behaviour in sciurids (Toth and Robinson 2004).

Apart from affecting plastic traits, different levels of sociality can also alter selective regimes and permit the evolution of novel behavioural phenotypes. For example, it has been shown that communicative complexity and vocal repertoire size have evolved in concert with social complexity in primates (McComb and Semple 2002). The widely held evolutionary explanation for this pattern is that increased social complexity creates unique intragroup fitness challenges for primates for which certain novel behavioural phenotypes (in this case advanced communication and cognition abilities) are selectively advantageous (Dunbar and

Shultz 2007). Such selection-altering effects arising out of social life are known as “indirect genetic effects” and/or “social selection” (West-Eberhard 1983, Queller 1992, Wolf et al. 1999,

Wscilo 2000, Wolf et al. 2001, Bijima & Wade 2008, Wolf and Moore 2010).

Using the comparative method (Harvey & Pagel 1992), we explore the association between level of sociality and individual maternal care behaviours in social and subsocial cobweb spiders of the genus Anelosimus Simon 1891 (Araneae: Theridiidae). These spiders are an ideal system for studying this association as sociality (cooperative breeding, hunting and nest building) has independently evolved at least four times in the genus and each social species has extant subsocial sister taxa (Agnarsson 2006; Agnarsson et al. 2006; Agnarsson et al. 2007). This allows for multiple phylogenetically controlled comparisons of behaviour between social and subsocial species from sister clades. Additionally, differences in level of sociality between Anelosimus species likely translate into acute environmental differences for individuals within social groups. Social Anelosimus live in nests of up to hundreds of individuals, build and share a communal living space, cooperatively hunt and perform alloparental care (Avilés 1997; Lubin and Bilde 2007). In contrast, subsocial Anelosimus live

! 29 ! alone in single-individual nests in which their offspring hatch, fledge and eventually disperse

(Avilés 1997). This profound difference in social context creates the potential for the expression and/or evolution of novel behavioural phenotypes.

In a previous study (Chapter 2) we noticed that more social species appeared to exhibit lower levels of individual maternal care behaviours, an association that we here formally test.

Lower levels of maternal care behaviour might be associated with greater degrees of sociality because social species enjoy increased physical protection of egg sacs given their larger nests, dilution effects due to multiple egg sacs, group anti-predator behaviours, and extensive alloparental care (Avilés 1997, Uetz & Heiber 1997). Together, these social factors are likely to ameliorate risks to egg sacs in social nests relative to subsocial ones (Uetz & Heiber 1997), thus relaxing selection for individual care behaviours. As a consequence, the benefits of a high level of maternal care behaviour may be accordingly lower for social females than for subsocial females. Assuming maternal care is costly (energetically or otherwise), we thus expect a facultative and/or evolutionary attenuation of these behaviours in social species.

We observed maternal care behaviours relevant to egg sac defense from predators, parasitoids and parasites, protection from physical damage and moisture, and thermoregulation (Austin 1985, Fink 1987, Gillespie 1990, Ruttan 1991, Foelix 1996). We compared six measures of such behaviours among species in two pairs of independently derived social and subsocial sister clades of Anelosimus. Overall, we found a strong trend for reduced maternal care behaviour in social species relative to subsocial ones. We consider alternative explanations for this pattern, including relaxed natural selection, phenotypic plasticity, and an incidental association due to differences in the external environment of subsocial and social cobweb spiders. We argue that reduced individual maternal care behaviours in social spiders may result from changes in selective pressures emerging from larger nests , and colonies, as well as the presence of multiple caregivers.

! 30 ! Methods

Study species

We selected two pairs of sister taxa comprising six species of cobweb spiders for our study. The first pair was the social species Anelosimus guacamayos Agnarsson 2006 and its subsocial sister species Anelosimus elegans Agnarsson 2006. The second pair was a simultaneous comparison between two members of a social clade, Anelosimus domingo Levi

1963 and Anelosimus eximius Simon 1891, versus two subsocial members of their sister clade,

Anelosimus baeza Agnarsson 2006 and Anelosimus cf. oritoyacu, the latter being an undescribed subsocial species close to Anelosimus oritoyacu Agnarsson 2006. For the latter we compared four species as two groups because we cannot rule out the possibility that A. domingo and A. eximius share a single origin of sociality (Agnarsson 2006). Hence, independent comparisons for each social species would not be warranted.

We performed all laboratory and field measurements in June 2010 at the natural habitats of each of the six species at various sites in eastern Ecuador. In the Napo province, we studied A. domingo and A. eximius at the Estacion Biológica Jatun Sacha (1.06°S,

77.61°W, ~410 m elevation). Jatun Sacha consists of 2500 hectares of 70% primary rain forest, and 30% regenerated secondary rain forest (Guevara & Avilés 2007). In the Quijos

Canton of Napo, we studied A. baeza along the Tena-Quito road south of Baeza (0.46°S,

77.89°W, ~1 900 m), A. elegans and A. guacamayos in the Reserva Ecólogica Antisana and the Parque Nacional Sumaco (0.63–0.65°S, 77.8°W, ~1 800 m elevation). Both of these locations are situated in lower montane cloud forest (Neill 1999). Finally, in the Pichincha province, we studied A. cf. oritoyacu at the Bellavista Cloud Forest Reserve near Tandayapa

(0.016°S, 78.68°W, ~ 2 000 m elevation), which is composed mostly of undisturbed montane cloud forest.

! 31 ! Field measurements

In total, we located 61 A. elegans, 16 A. guacamayos, 38 A. baeza, 35 A. cf. oritoyacu,

5 A. eximius and 3 A. domingo nests. In each nest, we measured two aspects of egg sac attendance behaviour (a key component of maternal care, Gillespie 1990) for each of the six species in their natural nests. First, we measured the distances between egg sacs and the nearest adult female spider. In larger nests (>10 egg sacs), we selected a sample of egg sacs for this measurement using a transect method. To do this, we began by approximating a linear transect (calibrated with an measuring tape held outside the nest) along the longest axis of the nest, passing through the centre of the nest and the two axes orthogonal to it; we then sampled the egg sac closest to the line at each 10-centimeter interval.

Next, we counted the number of egg sacs being attended by females, i.e. in contact with the female’s body, and the total number of egg sacs. We used these values to calculate the proportion of egg sacs being attended in each nest. Note that because disturbing the nest usually causes spiders to abandon egg sacs (K Samuk, personal observation), we could not exhaustively search for every egg sac in the inner retreats of the nest. Hence, our estimates are somewhat biased toward visible egg sacs.

Laboratory assays

For our laboratory assays, we collected females clutching egg sacs from natural colonies using a drinking straw aspirator. In the case of A. eximius and A. domingo, who rarely clutch egg sacs, we collected females either clutching, in physical contact ,or within 1 cm of an egg sac, along with the egg sac itself. In total, we collected 36 A. baeza, 37 A. cf. oritoyacu, 34 A. eximius from five colonies, 37 A. domingo from three colonies, 33 A. elegans and 44 A. guacamayos from sixteen colonies. Note that although we had no way of assessing the reproductive status of the females, clutching and proximity are themselves indicative of the

! 32 ! post-egg sac laying stage (K Samuk personal observation, K Samuk, E LeDue & L Avilés unpublished data).

We placed each field-collected female and egg sac in a plastic container ~11 cm in circumference and 6 cm in height. These containers had a raised inner circle, 8 cm in circumference and 0.3 cm high inset in their bases and were sealed with a perforated lid. We misted each container with water daily using a spray bottle. Before all trials, we removed all residual webbing in the container to standardize assay arenas across replicates. We conducted every assay described below within 1-4 days of collecting the egg sac and female under semi- natural lighting conditions, between 0800-2000 hours. After experimentation, we returned subsocial females and their egg sacs to vegetation similar to their original location of collection. For social species, we placed the spiders and egg sacs back into their original nests.

Assay 1: egg sac acceptance

To test for a female’s willingness to accept an egg sac after being separated from it, we placed females and their egg sacs 1-2 cm apart. We then slowly rolled the egg sac toward the female using a paintbrush and recorded whether the female grabbed the egg sac using her mouthparts and/or pedipalps. We terminated the trial if the female did not immediately grasp the egg sac.

Assay 2: egg sac relinquishment

Our second assay was a test of the female’s willingness to relinquish her egg sac when disturbed. We pursued females clutching egg sacs with a small paintbrush, lightly prodding them and moving along with the spider as it retreated. We stopped when the female relinquished the egg sac or three minutes elapsed, whichever came first.

! 33 ! Assay 3: egg sac reclamation latency

Our third assay measured the latency to reclaiming an egg sac after being separated from it. This metric was meant to capture the extent of egg sac searching behaviour (Opell

2001). To begin, we placed the female and her egg sac six body lengths apart .We then made six successive measurements of the distance between the egg sac and the female’s cephalothorax at 15 min intervals followed by seven successive observations at hour intervals, for a maximum of 13 observations over 8.5 h. We stopped measurements after 8.5 hrs or when the female first clutched the egg sac, whichever came first.

Assay 4: egg sac repair probability

Our final assay measured a female’s willingness to repair a damaged egg sac. We first placed the female spiders in 2.0 mL self-standing tubes with o-ring caps (USA Scientific,

Ocala, Florida, USA) with a section of dried leaf. Separately, we damaged the spider’s egg sac by making a small hole in the outer casing and widening it with forceps. The final holes were typically ~3mm in diameter. We placed the ripped egg sac into the tubes with the female and leaf, and capped the tubes. After 8 hours, we recorded whether the female had made any visible attempt to repair the egg sac. We defined a “visible attempt” as any combination of binding the hole shut with silk, folding the outer casing in on the hole, binding the open end of the egg sac to the leaf and/or holding the egg sac shut using the mouthparts and chelicerae.

Statistical analyses

To compare assays between social and subsocial sister taxa, we used three different types of analyses (Table 3.1). First, for four of our six measures, we employed generalized linear mixed models (GLMMs). In each model, we included species as a fixed effect and colony identity as a random effect to control for the possibility of pseudoreplication. We fit

! 34 ! each GLMM via penalized quasi-likelihood using the glmmPQL function in the MASS package in R (Venables & Ripley 2002, R Core Team 2009).

For the egg sac relinquishment experiment, we performed a “right-censored” mixed effects survival regression of relinquishment time with species as a fixed effect and source colony as random effect using the survreg function in R (Therneau & Lumley 2009). Note that “right-censored” indicates a replicate in which the event of interest (the egg-sac being dropped by the female) did not happen during the duration of the experiment. For the egg sac reclamation experiment, which produced both interval censored (i.e. the event occurred some time during a given interval) and right censored data, we calculated nonparametric maximum likelihood estimates (NPLME) of reclamation time curves (Kaplan Meier curves) using the icfit function in the interval package for R (Fay 2009). Note that for the latter it was not possible to include the effect of colony in the model. We thus also repeated the analysis using standard survival regression (i.e. disregarding the interval structure), in which colony could be included as a random effect.

Using each analysis framework above, we performed the following comparisons: A. guacamayos versus A. elegans; A. eximius and A. domingo, versus A. baeza and A. cf. oritoyacu. We computed GLMM comparisons by fitting pair-wise models for each comparison, with one group modelled as the intercept term and the other group as a standard model parameter. For comparison of survival regression parameters, we similarly fit pair-wise survival models, with one group modelled as an intercept term and the other as a regression parameter. Finally, we compared NPLME estimates of survival time using the logrank test.

Unless otherwise stated, all comparisons were two-tailed and employed an ! of 0.05.

! 35 ! Results

Field measurements

As a group, the social species A. eximius and A. domingo had significantly higher average egg sac to nearest female distances than the subsocial species A. baeza and A. cf. oritoyacu (Figure 3.1A, Table 3.1). The social species A. guacamayos also had significantly higher average egg sac to nearest female distances than the subsocial species A. elegans (Figure

3.1A, Table 3.1).

There was a similar pattern for the related measure of proportion of egg sacs attended in the nest. Nests of A. eximius and A. domingo had significantly lower proportions of guarded egg sacs than nests of A. baeza and A. cf. oritoyacu (Figure 3.1B, Table 3.1), as did nests of A. guacamayos relative to those of A. elegans (Figure 3.1B, Table 3.1).

Egg sac acceptance

Females of A. eximius and A. domingo had significantly lower probabilities of accepting egg sacs versus A. baeza and A. cf. oritoyacu (Figure 3.2A, Table 3.1). Similarly, A. guacamayos females had significantly lower probabilities of accepting egg sacs versus A. elegans (Figure 3.2A, Table 3.1).

Egg sac relinquishment

When under simulated threat, females of A. eximius and A. domingo relinquished their egg sac significantly sooner on average compared to females of A. baeza and A. cf. oritoyacu

(Figure 3.3A, Table 3.1). A. guacamayos females also released their egg sacs significantly sooner on average than A. elegans females (Figure 3.3B, Table 3.1).

Egg sac reclamation latency

When separated from their egg sacs, A. eximius and A. domingo females took significantly longer than A. baeza and A. cf. oritoyacu females to reclaim their egg sacs (Figure

! 36 ! 3.3C, Table 3.1). In contrast, there was no significant difference in reclamation time between

A. guacamayos and A. elegans (Figure 3.3D, Table 3.1). We could not include colony as a random effect in these interval survival models, but found no significant effect of source colony identity on reclamation time (logrank k-sample test: A. eximius "2=7.58, p=0.11, n=36;

A. domingo "2=3.89, p=0.14, n=30; A. guacamayos "2=16.5,p=0.12, n=38). Using standard survival regression, which allows control of colony identity, we found the same pattern for A. eximius and A. domingo (Z=-5.93, n=70, p<0.0001), while, in contrast to the interval survival analyses, A. guacamayos also had significantly longer reclamation times than its subsocial sister species (Z=-2.56, p=0.01, n=70).

Egg sac repair probability

A. eximius and A. domingo females were significantly less likely to repair a damaged egg sac compared to A. baeza and A. cf. oritoyacu females (Figure 3.2B, Table 3.1). There was, however, no significant difference in the probability of egg sac repair between females of A. guacamayos and A. elegans (Figure 3.2B, Table 3.1).

Discussion

In our field and laboratory measurements, we found reduced levels of maternal care behaviour by females of every social species relative to females of their respective subsocial sister taxa.

This was true of all six measures for A. eximius and A. domingo, and four of six measures for

A. guacamayos. Hence, for these two independent comparisons, our data support the hypothesis that increased level of sociality is associated with reduced individual maternal care behaviour.

Potential causes of reduced maternal care

Perhaps the largest question our results raise is: what biological process is responsible for the reduction in maternal care we observed in social cobweb spider species? There are a

! 37 ! number of viable alternatives. First, our results could be indicative of a facultative reduction of maternal care in response to the presence of “helpers” (alloparents). This behaviour, dubbed

“load-lightening” by Brown (1978), has been widely reported for various cooperatively breeding birds (reviewed in Crick 1992, Heinsohn 2004). In the case of birds, parents with helpers spend less time caring for their offspring and more time self-feeding, assumingly increasing their survival and future breeding (Crick 1992, Cockburn 1999).

However, evidence suggests that load-lightening may not be the explanation for the reduction in maternal care we observed. In a previous study, we reared in isolation in the laboratory gravid subsocial and social females for a period of one month. Despite these spiders being free of social interactions and any possibility of “help stimulus”, we recovered the same pattern and magnitude of reduced care in the social species relative to their subsocial relatives

(Chapter 2). Thus, it seems that in the timeframe of our previous study, cobweb spiders do not facultatively adjust their maternal care in response to social context.

Apart from the social environment, maternal care behaviour could conceivably also be facultatively adjusted to match the level of predation/parasitism perceived by female cobweb spiders (a major source of egg sac loss, Foelix 1999). However, during both this study and our previous study we incidentally excluded all predators and parasites (except for e.g. airborne fungi) from the experimental females. This exclusion lasted 1-2 days in this study, and ~30 days in our previous study. Yet, we still recovered levels of maternal care behaviour similar to our field estimates for both subsocial and social females (Figure 3.1; Chapter 2). Hence, during adulthood, the maternal care behaviour of Anelosimus females appears to be insensitive to predation threat per se. These observations suggest that the observed reduction of maternal care behaviour in social Anelosimus is likely a fixed species trait, and not a result of plasticity.

If we accept that maternal care behaviour is not a plastic trait in Anelosimus, the pattern we observed could instead be the result of divergent natural selection between social

! 38 ! and subsocial species caused by different external environments. Such selection would need to somehow favour “low parenting” behavioural phenotypes in social species, but not in subsocial species. This could occur if, for example, social species tend to live in areas with lower levels of predation generally. However, there are two pieces of evidence that suggest this scenario is unlikely. First, species in one social-subsocial pairs, A. guacamayos and A. elegans, are currently sympatric and have likely been so in the past (Avilés et al. 2007). Hence, their comparison has been mostly controlled for differences in the external environment, currently and through evolutionary time. If the reduction in maternal care behaviour we observed was indeed the result of selection imposed by an extrinsic factor, we would have expected to observe similar levels of maternal care in A. guacamayos and A. elegans, which we did not

(Figures 3.1-3.3).

The case for the other group of species is more complex: A. eximus and A. domingo are both found in the lowland rainforest (~400 m elevation) whereas A. baeza and A. cf. oritoyacu are found in mid to high elevation cloudforest (~1600-2000 m elevation). Clearly, there are major environmental differences between these habitats. However, it is almost certain that nearly all of these differences would actually increase the risk to egg sacs of social species, and thus favour higher levels of maternal care in social females. For example, we know that in the lowland rainforest there is much greater potential for predation on social spiders by ants, birds and wasps (Olson 1994, Rabhek 1997, Purcell & Avilés 2007, Avilés set al. 2007).

Further, there is far more physical disturbance in the form of debris, rain and tree falls

(Rabhek 1997, Purcell & Avilés 2007). Indeed, in their transplant study of A. baeza to the lowland rainforest, Purcell & Avilés (2008) found that A. baeza failed to establish due to high rates of rainfall and intense predation by ants. Therefore, based on the protective functions of maternal care, we would actually expect to observe the opposite pattern, i.e. increased care behaviour in the social species found in the rainforest if external environmental factors were

! 39 ! driving this pattern. Thus, based on our knowledge of Anelosimus habitats, it seems overall unlikely that externally imposed selection explains the reduction in maternal care we observed.

Note that a similar argument could be made for externally driven facultative reductions in maternal care.

Relaxed natural selection

Given that external environmental pressures do not appear to explain this pattern, we suggest that a relaxation of natural selection on maternal care behaviour coincident with increased sociality is the most promising alternative explanation. By “relaxation” we refer to the scenario described in Lahti et al. (2009): an elimination or weakening of a source of selection that was formerly important for the maintenance of a trait, coincident with environmental change.

There are several lines of evidence that support this idea in Anelosimus. First, the reduction in maternal care we observed appears to have convergently evolved in each social species. Such independent convergent evolution of phenotypes under similar environments

(social environments in this case) is generally agreed to be preliminary evidence of selection

(Endler 1985). Secondly, studies of other spider species suggest there is strong potential for a relaxation of selection on maternal care in social groups. For example, work in colonial

Metepeira spp. and social Stegodyphus dumicola spiders suggests that egg sacs experience less predation in social versus subsocial nests (reviewed in Uetz & Heiber 1997, Henschel 1998).

The amount of maternal care egg sacs need to prevent mortality also appears to be far lower in group-living species than in solitary ones (Uetz & Heiber 1997, Henschel 1998). This reduction in egg sac risk is ostensibly due to dilution effects, a larger, denser nest and possibly group active-antipredatory behaviours in social nests (Uetz & Heiber 1997, Henschel 1998).

Social Anelosimus enjoy all these protective features, and are also known to extensively

! 40 ! alloparent (this thesis Chapter 2; Avilés 1997). So, by analogy, it seems reasonable to assume that egg sacs in social Anelosimus nests experience at least as much of a reduction in risk as do egg sacs in Metepeira and Stegodyhpus nests. Such a reduction in risk to egg sacs could easily result in a relaxation of natural selection on maternal care behaviour in social Anelosimus.

Interestingly, in a recent study Jones et al. (2010) found that social morphs of the socially polymorphic cobweb spider Anelosimus studiosus had lower reproductive success in single-female colonies than did subsocial morphs. They interpret this difference as evidence of a fitness cost associated with decreased activity level and aggressiveness in the social morphs when in solitary nests. This is further evidence that the strength of natural selection can be ameliorated by social context in cobweb spiders generally.

Naturally, more information is needed to assess whether socially-mediated relaxed natural selection in Anelosimus is indeed a cause of the pattern we observed. For one, analysis of more sister clade pairs is needed to further support this evolutionary association. Secondly, one could directly test if egg sacs are indeed at lower risk in social nests than in subsocial nests

(i.e. attempt to link sociality with fitness directly). Third, the effect of early-life social environment on later-life maternal care behaviour could be explored by repeating our assays on social and subsocial females reared from egg sacs in isolation in the lab. Together, these experiments would help flesh out the selective forces involved in driving the pattern, the differential costs and benefits of maternal care behaviour across social contexts and the role of plasticity in driving this pattern.

If relaxed natural selection is indeed the mechanism driving this pattern, our results have interesting implications. First, for each maternal care metric we studied, we found consistent reductions in mean trait values in social versus subsocial species, rather than, for example, an increase in trait variance with no shift in mean (“classic” relaxed natural selection, Lahti et al. 2009). This result is suggestive of one specific type of relaxed selection,

! 41 ! vestigialization (Fong et al. 1995, Lahti et al. 2009). This implies there may be an evolutionary trade-off involving maternal care behaviour in social species. One possibility is that in social nests, females who spend most of their time guarding are using up time that could be used for capturing prey (i.e. paying a lost opportunity cost). Hence, natural selection may favour social females who spend less time with their egg sacs, and more time foraging. This extra time spent foraging may then translate into increased fitness, either directly through production of a second egg sac or indirectly via fitness benefits to close kin. Thus, a partial emancipation from egg sac defence may be one of the manifold selective benefits of sociality, all else being equal.

Correlated behaviours

There is one further mechanism that could act in concert with relaxed natural selection and/or underlie the pattern we observed: among-behaviour trait correlations. It is widely appreciated that among-trait correlations arising via pleiotropy, genetic linkage, etc. can lead to correlated evolutionary responses (Lande & Arnold 1983, Brodie et al. 1995). Because the evolution of sociality typically involves changes in many different classes of behaviour, among- behaviour correlations have likely played a role in the behavioural evolution in social animals

(Wcislo 2000, Pruitt et al. 2010). In terms of the present study, it could be the case that reduced maternal care evolved as a correlated by-product of the evolution of another behaviour. There is some evidence that supports this idea: Pruitt et al. (2010) have shown that that social morphs of the socially polymorphic spider Anelosimus studiosus tend to have low levels both aggressiveness and activity per se. Conceivably, the level of maternal care behaviour provided by a spider with lower-activity levels generally will be less vigorous, and of lower quality. Indeed, this was borne out in a study by Jones et al. (2010). Interestingly, these authors go on to suggest that such behavioural correlations could constrain the evolution of social behaviour by creating adaptive valleys between subsocial and social phenotypes.

! 42 ! Significance

Along with increasing our knowledge of social and behavioural evolution in spiders, our findings point to the importance of changes in the social environment in the evolutionary process. Theoretical discussion of this topic is widespread (e.g. Wolf and Moore 2010) yet empirical examples are, to our knowledge, relatively rare (but see Blumstein and Armitage

1998, Devillard et al. 2004). This dearth is quite surprising, as increasing levels of social complexity are widely recognised as extremely important transitions in the history of life

(Maynard Smith and Szathmáry 1997). One of the most interesting aspects of this transition is that traits evolved in previous social contexts are thrust into what are often completely different selective landscapes. Along with the present study, examples of this phenomenon (in the broad sense) include the changes in cellular traits coincident with the evolution of multicellularity, and the exportation of genes from the mitochondrial to nuclear genome after endosymbiosis was established. Another intriguing parallel with our study is the “social brain” hypothesis from the primate literature, which suggests that the unique abilities of the primate

(and human) brain are attributable to adaptation to complex social environments, i.e. a change of selection in concert with sociality (Dubar and Shultz 2007). Whatever the system, understanding how traits evolve and interact during and after the transition to a higher level of complexity is fundamental to our understanding of the evolution of complex organisms generally. Further studies of this process are badly needed.

Future studies

While our results are highly suggestive of an association between sociality and reduced maternal care, there is insufficient phylogenetic replication (n=2 pairs in our study) to solidify this fact. Indeed, only 2/6 of our behavioural assays are significantly different when analyzed at the level of phylogenetic pairs (Appendix C, Table C.1). Further studies of social

! 43 ! Anelosimus and subsocial/social pairs of other animals would allow for more statistical power in testing for such associations.

Conclusion

In this study, we have shown that three social cobweb spiders display lower levels of maternal care behaviour relative to subsocial species from sister clades. Based on this and a number of supporting facts, this may be the result of relaxed natural selection on maternal care behaviour. This relaxation of selection could have occurred as a result of a reduced risk to egg sacs and amelioration of environmental stressors in the more highly social species. Our study represents a starting point for further investigation into the understudied role of social complexity in behavioural evolution.

! 44 !

Table 3.1 Statistical model information and results for contrasts between two pairs of social and subsocial sister clades of cobweb spiders. Each row contains model information for one of six measures of maternal care behaviour. Estimates and standard errors (SEs) correspond to parameters of the difference between social and subsocial species. Bold p-values signify significance at the !=0.05 level.

Contrast (H0: social - subsocial = 0)

(A. eximius, A. domingo) - A. guacamayos - A. elegans

(A. baeza, A. cf. oritoyacu)

Measurement Analysis Error function Estimate df1, df2 S.E. p-value Estimate df1, df2 S.E. p-value

Female distance GLMM Gamma (log) 0.43 80,80 0.09 <0.00001 0.42 46,64 0.16 0.011

Attendance GLMM Binomial (logit) -1.36 80, 80 0.40 0.001 -1.54 46, 64 0.47 0.002

Acceptance GLMM Binomial (logit) -3.79 82, 82 0.82 0.00002 -1.47 53, 28 0.50 0.007

Repair GLMM Binomial (logit) -1.80 73,73 0.77 0.02 0.19 45,45 0.49 0.70

Relinquishment survreg Weibull (identity) -170.77 82, 82 9.9 <0.00001 -55.93 38,32 10.39 <0.00001

Reclamation surv-nplme* N/A 5.66† 69,67 – <0.00001 1.69† 38,32 – 0.09

* nonparametric maximum likelihood estimate of survival

† logrank test Z-score.

! 46 !

!

Figure 3.1 Two metrics of maternal care measured in natural nests of six species of Anelosimus. A: Distance from egg sacs to the nearest adult female spider; B: proportion of egg sacs guarded by female spiders; White and grey bars represent subsocial and social species respectively. Bar heights and 95% confidence intervals are derived from back- transformed generalized linear model estimates. Braced horizontal lines above groups represent statistical comparisons performed as part of the generalized linear model contrast structure (see Table 3.1 for full details).

Significance codes: *: p<0.05, **: p<0.001, ***: p< 0.0001. Sample sizes are indicated below each species label.

! 47 !

Figure 3.2 A: probability female spiders of six Anelosimus species will accept an egg sac when presented with one under laboratory conditions; B: probability female spiders of six Anelosimus species will repair a damaged egg sac over 8 hours under laboratory conditions. White and grey bars represent subsocial and social species respectively.

Bar heights and 95% confidence intervals are derived from generalized linear model estimates, back-transformed into proportions or probabilities. Braced horizontal lines above groups represent generalized linear model parameter contrasts (see text for statistical details). Sample sizes for each species are indicated under their respective labels. Significance codes: *: p<0.05, **: p<0.005, ***: p< 0.0005. Sample sizes are indicated below each species label.

! 48 !

Figure 3.3 Survival plots of the results of two laboratory assays of maternal care for females of six species of Anelosimus. A,B: probability female spiders dropped their egg sac when being pursued over

180 seconds, C,D: probability female spiders located their egg sac over 510 minutes after being separated from it. Fitted lines represent smoothed Kaplan-Meier and grey outlines denote 95% confidence intervals. Note that observations in CD were made at intervals versus continuously in AB, which results in rectangular confidence intervals when fitted by nonparametric maximum likelihood estimation. Darker grey areas signify areas of overlap between confidence intervals. Survival plots have been inverted and separated into the two contrast groups for clarity.

! 49 ! Chapter 4

General conclusion

Summary of findings

In this thesis, I presented studies of cobweb spiders that address two poorly understood aspects of the evolution of sociality. First, in Chapter 2, I showed that a lack of offspring recognition likely predisposed cobweb spiders toward evolving alloparental care. This demonstrates that pre-existing traits played an important role in facilitating the evolution of sociality in this group of spiders. Secondly, in

Chapter 3, my co-authors and I presented evidence that reduced maternal care behaviour has convergently evolved in social species of cobweb spiders. Based on a number of supporting facts, we interpret a relaxation of natural selection on maternal care as a consequence of social life to be the most likely explanation for this pattern. This illustrates that the evolution of sociality and its associated ecological changes have likely had profound effects on the evolutionary fate of social cobweb spiders. In this chapter, I highlight the broader implications of these results and explore avenues for future studies.

The role of preexisting traits

Since the inception of sociobiology, biologists have theorized that behaviours evolved in solitary contexts are often co-opted into new social roles during the evolution of social behaviour (Evan & West-

Eberhard 1970, Wilson 1975). My results in Chapter 2 lend direct support to this idea. Interestingly, our results also mirror studies of the evolution of alloparental care in social hymenopterans in which (sterile) alloparenting female worker bees and wasps behaviourally and neurochemically recapitulate pre- exisiting maternal care behaviours (Amdam et al. 2006, Toth et al. 2007). This recapitulation is apparently due to stimulation of the maternal care response by offspring stimuli that are non-specific to individuals, i.e. alloparental care via mistaken identity (Amdam et al. 2006, Toth & Robinson 2008).

! 50 ! The repeated co-option of maternal care behaviour in two distantly related classes of animal (insects and arachnids) suggests that pre-existing traits may play a very important role in the evolution of sociality.

But what role do these traits play exactly? The discoveries of maternal care co-option in hymenopterans and our finding of the spontaneous occurrence of alloparental care via mistaken identity in cobweb spiders are prime examples of emergence – new group properties arising out of simple individual behaviours (Holland 1998). This lends support to the idea advanced by Bell (1985) and others that emergent behaviours and phenomena often are involved in major evolutionary transitions in general (i.e. changes in the degree of hierarchical biological organization Maynard-Smith and Szathmáry

1997). Specifically, it has been suggested that the potential to transition from one level of organization to another (e.g. solitary to social; unicellular to multicellular) is in some degree predicated on the existence of “emergable traits” – pre-existing individual traits that when placed in a group context cause some new (often serendipitously beneficial) group property (Bell 1985, Michod 1999). This can also be extended to include traits that have the potential to be modified to have emergent group beneficial properties, as well as those that have them from the onset (e.g. the extracellular matrix in Volocine algae,

Herron & Michod 2008). This idea may help to explain why sociality is so common in some clades and absent in others. However, this idea is still in its infancy and further studies, both phylogenetic and behavioural, are needed.

The complexity of social behaviour

Social behaviours are often thought to be a priori more complex than other behaviours and phenotypic traits (Williams 1966, Lahti 2006). Yet, while alloparental care is heralded as one of the essential behaviours of advanced animal societies (Wilson 1975, Crespi & Yanega 1995), our study in

Chapter 2 demonstrates that it (i) it can be extremely simple in form and (ii) can require surprisingly little behavioural evolution. In fact, in social cobweb spiders, no behavioural evolution was required for the occurrence of alloparental care (beyond a change in social context). This is an important finding for

! 51 ! a number of reasons. First and foremost the simplicity of alloparental care by mistaken identity is in complete opposition to the idea of social behaviours as “complex”. Secondly, it illustrates that the evolution of sociality require fewer evolutionary steps than once thought. This is an important point, as the evolution of “advanced” social behaviour has been a major topic of evolutionary research since

Darwin (1872). Finally, studies such as ours show that the evolution of social behaviours can be tractably studied using simple, straightforward methods (e.g. the comparative method). This is noteworthy, as the perceived complexity of social behaviours has likely discouraged many investigators from examining them as they would morphology or physiology.

The evolutionary impact of sociality

The role of the social environment in shaping the evolutionary process has long been acknowledged (Darwin, Haldane, Hamilton). Yet, apart from studies of parent-offspring interactions

(reviewed in Clutton-Brock 2002), surprisingly few empirical studies have explicitly examined the evolutionary impact of the broader social environment, i.e. non-familial conspecific interactions

((Blumstein & Armitage 1998 and Devillard et al. 2008 are notable exceptions). The study I presented in

Chapter 3 contributes to our understanding of this understudied phenomenon in a number of ways. First, it suggests that different social contexts can potentially alter the strength of selection, resulting in highly divergent phenotypes. While this seems intuitive, and has been discussed at length in the literature, there is a major dearth of empirical evidence for it. Secondly, our study highlights a promising natural study system for testing the predictions of an emerging body of theory concerning the evolutionary effects of social interactions (Wolf et al. 1999, McGlothlin et al. 2010). Given that there have been numerous calls for such a system, our results will hopefully stimulate further work. Thirdly, it may be the case that the physical changes in the size and architecture of cobweb spiders nest coindicident with the evolution of sociality underlie a relaxation of natural selection. If this is true, our results could be an example of the evolutionary consequences of one special class of social environmental effects known as “niche

! 52 ! construction” – physical and heritable alteration of an organism’s environment resulting in changes to organism interactions with other organisms and its environment (Odling-Smee et al. 2002). Given that niche construction has received little empirical attention, the Anelosimus system may be an ideal jumping off point for investigating its evolutionary implications.

Future studies

The evolution of non-reproductive helpers

Our study of alloparental care in Chapter 2 raises a number of important follow-up questions.

For one, while we examined alloparental care by post-reproductive females, the evolution of non- reproductive female helping behaviour (e.g. sterile castes, helpers at the nest) is also a major question in evolutionary biology and sociobiology (Darwin 1859, Wilson 1975, West-Eberhard 1987). The mechanism by which non-reproductive helping is thought to evolve in hymenopterans is a heterochronic shift in the expression of maternal care behaviour – from after insemination and egg laying to before

(Amdam et al. 2006, Page 2002, Toth & Robinson 2008). It would be worth investigating if this is also the case in social cobweb spiders for two reasons. First, it is likely that many females do not reproduce in social cobweb spider nests, so there is scope for selection for non-reproductive helping (Salomon et al.

2008). Secondly, social cobweb spiders have evolved sociality relatively recently (unlike social hymenopterans), and allow us the unique opportunity to examine the early stages of the evolution of non-reproductive helping (Agnarsson et al. 2007).

The adaptive significance of alloparental care

Given alloparental care arose as a byproduct of indiscriminate maternal care, it would be interesting to see what the adaptive value (if any) of alloparental care behaviour has for individual female spiders. Indeed, it is possible that at the level of individuals, alloparental care is a cost, rather than a benefit of sociality in cobweb spiders. This question could be approached in the following kin selection framework. Three things would need to be shown to demonstrate an individual adaptive

! 53 ! benefit of alloparental care: 1) alloparental care increases the fitness of the offspring receiving the care;

2) females are related to the offspring they alloparent; 3) The product of the alloparent-offspring relatedness and the fitness increase due to alloparenting outweighs the fitness costs of performing the behaviour. A conceptually similar multilevel selection approach could also examine increased group survival due to alloparenting (although this has been more or less achieved by Jones & Riechert 2007).

Relaxed natural selection

Our demonstration of reduced maternal care in social spider species opens a number of future avenues of investigation. Foremost would be more explicit tests of the relaxation of natural selection on the maternal care of egg sacs in social spider nests. One way to approach this could be to artificially establish social females in solitary nests in the field, and compare egg sac mortality between these nests and social nests. This would indicate whether egg sacs are indeed at lower risk in social nests. This could be followed by examining the costs and fitness effects of maternal egg sac care (e.g. energetic; lost opportunity costs). This would allow us to better understand why maternal care has been reduced

(vestigilized), rather than simply freed from selection per se. One way to measure the costs of maternal care could be to examine the correlation between the level of maternal care provided by a female and her lifetime fitness under natural field conditions. Together, these tests would give us better insight into the role of relaxed selection in driving the reduction in maternal care we observed in social cobweb spiders.

Conclusion

In this thesis, I sought to go beyond traditional approaches to studying the origin and consequences of sociality and further our understanding of sociality in cobweb spiders and organisms generally. The comparative studies I presented revealed a major role for pre-existing traits in the origin of sociality, and that the forces of evolution are likely altered in concert with the onset of social life.

These results open the door to a variety of important future studies in the Anelosimus system. Social life is ubiquitous, yet our understanding of its origin and consequences is still in it infancy. Given the major

! 54 ! role of social organisms in nearly every ecosystem, furthering our understanding of sociality is key to our understanding of the natural world.

! 55 ! References

Abbot P. 2009. On dispersal and altruism in social aphids. Evolution, 63: 2687-2696.

Agnarsson I, Aviles L, Coddington J, Maddison W. 2006. Sociality in Theridiid Spiders: Repeated

Origins of an Evolutionary Dead End. Evolution 60(11): 2342–2351.

Agnarsson I. 2002. Sharing a web—on the relation of sociality and kleptoparasitism in theridiid spiders

(Theridiidae, Araneae). Journal Of Arachnology 30:181-188.

Agnarsson, I, Kuntner, M, Coddington, JA, Blackledge, T. 2010. Shifting continents, not behaviours:

independent colonization of solitary and subsocial Anelosimus spider lineages on Madagascar

(Araneae, Theridiidae). Zoologica Scripta, 39(1), 75–87.

Agnarsson, I, Maddison, WP, & Avilés, L. 2007. Phylogeny of the social Anelosimus spiders (Araneae:

Theridiidae) inferred from six molecular loci and morphology. Molecular Phylogenetics and

Evolution 43: 833–851.

Agnarsson, I. 2006. A revision of the New World eximius lineage of Anelosimus (Araneae, Araneoidea,

Theridiidae). Zoological Journal of the Linnean Society 141: 447–626.

Amdam GV, Csondes A, Fondrk MK, Page RE. 2006. Complex social behaviour derived from maternal

reproductive traits. Nature 439:76-78.

Arnold, KE, Owens, IPF. 1999 Cooperative breeding in birds: the role of ecology. Behav. Ecol. 10, 465 –

471

Austin, A. 1985. The function of spider egg sacs in relation to parasitoids and predators, with special

! 56 ! reference to the Australian fauna. Journal of Natural History 19(2): 359–376.

Aviles L, Agnarsson I, Salazar PA, Purcell J, Iturralde G, Yip EC, Powers KS, Bukowski TC. 2007.

Altitudinal Patterns of Spider Sociality and the Biology of a New Midelevation Social Anelosimus

Species in Ecuador. Am Nat 170:783-792.

Avilés, L. 1997. Causes and consequences of cooperation and permanent–sociality in spiders. In: Choe, J

and Crespi, BJ, editors. Evolution of social behaviour in insects and arachnids. Cambridge

University Press: New York. pp. 476–498.

Bates, D, Maechler, M, Bolker, B. 2011. lme4: Linear mixed-effects models using S4 classes. R package

version 0.999375-38. http://CRAN.R-project.org/package=lme4

Bateson, P. 1988. The active role of behaviour in evolution, In Ho, MW & Fox, S, editors. Process and

Metaphors in the New Evolutionary Paradigm. Chichester: Wiley. pp. 191– 207.

Beauchamp, G. 2008. What is the magnitude of the group-size effect on vigilance? Behav Ecol 19:1361-

1368.

Bell, G. 1985. The origin and evolution of early germ cells as illustrated by the Volvocales. In The Origin

and Evolution of Sex (Halvorson, H.O. and Monroy, A., eds), pp. 221–256, Alan R. Liss

Bergmüller, R., Johnstone, R., Russell, A., Bshary, R., 2007. Integrating cooperative breeding into

theoretical concepts of cooperation. Behav. Process..

Bijleveld, AI, Egas, M, Van Gils, JA and Piersma, T. 2010. Beyond the information centre hypothesis:

communal roosting for information on food, predators, travel companions and mates? Oikos

119: 277-285.

! 57 ! Bijma, P, Wade, MJ. 2008. The joint effects of kin, multilevel selection and indirect genetic effects on

response to selection. Journal of Evolutionary Biology 21(5):1175-1188.

Blumstein, DT, Armitage, KB. 1998. Life history consequences of social complexity a comparative study

of ground-dwelling sciurids. Behavioural Ecology 9(1): 8-19.

Boness DJ, Craig MP, Honigman L, Austin S. 1998. Fostering behaviour and the effect of female density

in Hawaiian monk seals, Monachus schauinslandi. J Mammal 79:1060–1069.

Brodie, ED III, Moore, AJ and FJ Janzen. 1995. Visualizing and quantifying natural selection. Trends in

Ecology and Evolution 10: 313-318.

Bromham L, Leys, R. 2005. Sociality and the Rate of Molecular Evolution. Molecular Biology and

Evolution 22:1393-1402.

Brown K. 1998. Proximate and ultimate causes of adoption in ring-billed gulls. Animal Behaviour

56:1529-1543.

Brown, J, Dow, D, Brown, E, Brown, S. 1978. Effects of helpers on feeding of nestlings in the grey–

crowned babbler (Pomatostomus temporalis). Behavioural Ecology and Sociobiology 4(1): 43–

59.

Brown, JL. 1983. Cooperation—a biologist’s dilemma. Adv. Study Anim. Behav. 13, 1–37.

Brown, JL. 1987. Helping and Communal Breeding in Birds. Princeton University Press, Princeton, New

Jersey.

Bshary R, Bergmüller R. 2008. Distinguishing four fundamental approaches to the evolution of helping.

J Evolution Biol 21:405-420.

! 58 ! Burda H, Honeycutt RL, Begall S, Locker-Grutjen O, Scharff A . 2000. Are naked and common mole-

rats eusocial and if so, why?. Behavioural Ecology and Sociobiology 47(5): 293-303.

Canestrari, D, Marcos, J and Baglione, V. 2007. Costs of chick provisioning in cooperatively breeding

crows: an experimental study. Animal Behaviour 73(2): 349-357.

Carducci, J, Jakob, E. 2002. Rearing environment affects behaviour of jumping spiders. Animal

Behaviour 59(1): 39–46.

Cheverud, JM. 2003. Evolution in a genetically heritable social environment. PNAS 100(8): 4357-4359.

Christenson T. 1984. Behaviour of Colonial and Solitary Spiders of the Theridiid species Anelosimus

eximius. Animal Behaviour:725-734.

Chu J, Marler CA, Wilczynski W. 1998. The effects of arginine vasotocin on the calling behaviour of

male cricket frogs in changing social contexts. Hormones and Behaviour 34 (3): 248-261.

Clutton-Brock, TH. 1991. The Evolution of Parental Care. Princeton, NJ: Prince- ton Univ. Press

Clutton-Brock, TH. 2002. Breeding together: kin selection, reciprocity and mutualism in cooperative

animal societies. Nature 373: 209–216.

Cockburn, A. 1998. Evolution of helping behaviour in cooperatively breeding birds. Annual Review of

Ecology and Systematics 29(1): 141–177.

Cockburn, A. 1998. Evolution of helping behaviour in cooperatively breeding birds. Annual Review of

Ecology and Systematics, 29, 141–177.

Codenotti T, Alvarez F. 1998. Adoption of unrelated young by greater rheas. Journal of Field

Ornithology.

! 59 ! Crespi, BJ, Yanega, D. 1995. The Definition of Eusociality. Behavioural Ecology 6 (1): 109-115.

Crespi, BJ. 1994 Three conditions for the evolution of eusociality—are they sufficient? Insectes Sociaux

41, 395–400.

Crespi, BJ. 1994 Three conditions for the evolution of eusociality—are they sufficient? Insectes Sociaux

41, 395–400.

Crick, H. 1992. Load–lightening in cooperatively breeding birds and the cost of reproduction. Ibis

134(1): 56–61.

Culley T, Wiley JE, Persons MH. 2010. Proximate cues governing egg sac discrimination and

recognition in the wolf spider Pardosa milvina (Araneae: Lycosidae). Journal Of Arachnology

38:387-390.

Darwin, C. 1872. The expression of the emotions in man and animals. London: John Murray.

Devillard S, Allainé, D, Gaillard J, Pontier D. 2004. Does social complexity lead to sex-biased dispersal

in polygynous mammals? A test on ground-dwelling sciurids Behavioural Ecology 15(1): 83-87.

Dickinson JL, Koenig WD, Pitelka FA. 1996. Fitness consequences of helping behaviour in the western

bluebird. Behavioural Ecology (2): 168-177.

Downes, MF. 1984. Egg sac "theft" among Latrodectus hasselti females (Araenae, Theridiidae).

Duffy J, Morrison C, Ríos R. 2000. Multiple Origins of Eusociality Among Sponge-Dwelling Shrimps

(Synalpheus). Evolution 54:503-516.

Duffy JE, Macdonald KS. 2010. Kin structure, ecology and the evolution of social organization in

shrimp: a comparative analysis. P Roy Soc B-Biol Sci 277:575-584.

! 60 ! Dufty, AM Jr., Clobert J, Møller AP. 2002. Hormones, developmental plasticity and adaptation. Trends

in Ecology and Evolution 17:190-196.

Dugdale HL, Ellwood SA, Macdonald DW. 2010 Alloparental behaviour and long-term costs of

mothers tolerating other members of the group in a plurally breeding mammal. Anim Behav 80

(4): 721-735.

Dunbar, RIM, Shultz, S. 2007. Evolution in the Social Brain Science 317 (5843): 1344-1347.

Eldakar OT, Dlugos MJ, Pepper JW, Wilson DS. 2009. Population structure mediates sexual conflict in

water striders. Science 326: 816.

Eldakar OT, Wilson DS. 2011. Eight criticisms not to make about group selection. Evolution: advanced

online publication, 3 APR 2011.

Elgar, MA. 1989. Predatory vigilance and group size in mammals and birds: a critical review of the

empirical evidence. Biological Reviews 64: 13–33.

Emlen, ST, Reeve, HK, Sherman, PW, Wrege, PH, Ratnieks, FLW and J Shellman-Reeve. 1991.

Adaptive Versus Nonadaptive Explanations of Behaviour: The Case of Alloparental Helping. Am

Nat 131(1): 259-270.

Emlen, ST. 1991. The evolution of cooperative behaviour in birds and mammals. in Krebs, J and N.

Davis eds. Behavioural ecology: an evolutionary approach (3rd edition). Blackwell Scientific,

Oxford. Pages 301-337.

Emlen, ST. 1982. The evolution of helping. I. An ecological constraints model. Am. Nat. 119, 29–39.

! 61 ! Emlen, ST. 1991. Evolution of cooperative breeding in birds and mammals. In Behavioural ecology. An

evolutionary approach (eds J. R. Krebs & N. B. Davies), 3rd edn. pp. 301–335. Oxford, UK:

Blackwell.

Emlen, ST, Wrege, PH. 1991. Breeding biology of white-fronted bee-eaters at Nakuru: the influence of

helpers on breeder fitness. Journal of Animal Ecology 60:309-326.

Endler JA. 1986. Natural Selection in the Wild. Princeton Univ Press, Princeton.

Evans, HE and West-Eberhard, MJ. 1970. The Wasps, University of Michigan Press

Fay, MP. 2009. interval: Weighted Logrank Tests and NPMLE for interval censored data. R package

version 0.7–5.5. http://CRAN.R-project.org/package=interval

Fink, L. 1987. Green lynx spider egg sacs: sources of mortality and the function of female guarding

(Araneae, Oxyopidae). Journal of Arachnology 15(2): 231–239.

Flatz R, Gerber LR. 2011. First Evidence for Adoption in California Sea Lions. PLoS One 5 (11):

e13873.

Foelix, RF. 1996. The Biology of Spiders, 2nd Edition. Cambridge, Massachusetts: Harvard University

Press.

Fong, D, Kane, T, Culver, D. 1995. Vestigialization and loss of nonfunctional characters. Annual

Review of Ecology and Systematics 26(1): 249–268.

Foster KR. 2010. A Defense of Sociobiology. Cold Spring Harbor Symposia on Quantitative Biology

74:403-418.

! 62 ! Furey F. 1998. Two cooperatively social populations of the theridiid spider Anelosimus studiosus in a

temperate region. Anim Behav 55:727-735.

Gardner, A. and Foster, KR. 2008. 1–35 in Korb, J and Heinze, J, Editors. Ecology of Social Evolution.

Springer, Berlin, Heidelberg.

Gillespie, R. 1990. Costs and benefits of brood care in the Hawaiian happy face spider Theridion

grallator (Araneae, Theridiidae). American Midland Naturalist 123(2): 236-243

Gould, SJ, Vrba ES. 1982. Exaptation—a missing term in the science of form. Paleobiology 8:4–15.

Gould, SJ. 1997. The exaptive excellence of spandrels as a term and prototype. Proceedings of the

National Academy of Sciences of the United States of America 94(20):10750.

Grand, TC, Dill, LM. 1999. The effect of group size on the foraging behaviour of juvenile coho salmon:

reduction of predation risk or increased competition? Animal Behaviour 58 (2): 443-451.

Guevara, J, Avilés, L. 2007. Multiple techniques confirm elevational differences in insect size that may

influence spider sociality. Ecology 88: 2015–2023.

Hanson, KC., O’Connor, CM. and SJ Cooke. 2009. Paternal aggression towards a brood predator

during parental care in wild smallmouth bass is not correlated with circulating testosterone and

cortisol concentrations. Hormones and Behaviour 55 (4): 495-499.

Harvey, PH, Pagel MD. 1991. The comparative method in evolutionary biology. Oxford University Press,

Oxford.

Hatchwell BJ. 2009. The evolution of cooperative breeding in birds: kinship, dispersal and life history.

Philosophical Transactions Of The Royal Society B-Biological Sciences 364:3217-3227.

! 63 ! Hatchwell, BJ. 1999. Investment strategies of breeders in avian cooperative breeding systems. The

American Naturalist 154(2): 205-219

Hayes, LD. 2000. To nest communally or not to nest communally: a review of rodent communal nesting

and nursing. Anim Behav 59:677–688

Heg D, Jutzeler E, Mitchell JS, Hamilton IM. Helpful Female Subordinate Cichlids Are More Likely to

Reproduce. Plos One (2009) vol. 4 (5) pp. e5458.

Heinsohn R, Legge S. 1999. The cost of helping. Trends in Ecology & Evolution 14:53-57.

Heinsohn, RG. 2004. Parental care, load-lightening and costs. In: Koenig, WD, Dickinson JL, editors.

Ecology and Evolution of Cooperative Breeding in Birds, Cambridge University Press,

Cambridge. p. 67–80.

Henschel, JR. 1998. Predation on social and solitary individuals of the spider Stegodyphus dumicola

(Araneae, Eresidae). Journal of Arachnology, 26: 61–69.

Herron, MD and Michod, RE. 2008. Evolution of complexity in the volvocine algae: transitions in

individuality through Darwin’s eye. Evolution Int. J. Org. Evolution 62, 436–451.

Holland, J. 1998. Emergence. From chaos to order. New York: Basic books.

Hollis, SW, Fischman, BJ, Venkat, A, Hudson, ME, Varala, K, Cameron, SA, Clark, AG and GE

Robinson. 2011, Genes involved in convergent evolution of eusociality in bees. PNAS Early

Edition 1-6. www.pnas.org/cgi/doi/10.1073/pnas.1103457108.

Hunt, J.H. 1999. Trait mapping and salience in the evolution of eusocial vespid wasps. Evolution 53(1):

225:237.

! 64 ! Jamieson IG. 1986. The functional approach to behaviour: is it useful? Am. Natur. 127:195–208

Jamieson IG. 1989. Behavioural Heterochrony and the Evolution of Birds' Helping at the Nest: An

Unselected Consequences of Communal Breeding?. The American Naturalist 133(3): 394-406.

Jamieson IG. 1991. The unselected hypothesis for the evolution of helping behaviour: too much or too

little emphasis on natural selection? Am. Natur. 138:271– 82

Jamieson, IG, and JL Craig. 1987. Critique of helping behaviour in birds: a departure from functional

explanations. Pages 79-98 in P. Bateson and P. Klopferm, eds. Perspectives in Ethology, Vol. 7.

Plenum, New York.

Jones ,TC and P Parker. 2002. Delayed juvenile dispersal benefits both mother and offspring in the

cooperative spider Anelosimus studiosus (Araneae: Theridiidae). Behavioural Ecology 13(1):

142-148

Jones TC, Riechert SE, Dalrymple SE, Parker PG. 2007. Fostering model explains variation in levels of

sociality in a spider system. Anim Behav 73:195-204.

Jones, TC. and S. Riechert. 2008. Patterns of reproductive success associated with social structure and

microclimate in a spider system. Animal Behaviour 76: 2001-2019.

Kasumovic, MM, Bruce, MJ, Herberstein, ME, and Andrade, MCB. 2009. Evidence for developmental

plasticity in response to demographic variation in nature. Ecology 90:2287–2296.

Keller L. 2009. Adaptation and the genetics of social behaviour. Philos T R Soc B 364:3209-3216.

Kirkpatrick, M, Lande, R. 1989. The evolution of maternal characters. Evolution, 43(3), 485–503.

! 65 ! Koening, WD and FA Pitelka. 1981. Ecological factors and kin selection in the evolution of

cooperatively breeding birds. Pages 261-280 in RD Alexander and DW Tinkle. Natural selection

and social behaviour: recent research and new theory. Chiron, New York.

Kokko H, Johnstone R, Wright J. 2002. The evolution of parental and alloparental effort in

cooperatively breeding groups: when should helpers pay to stay? Behavioural Ecology 13:291-

300.

Konig, B. 1994. Communal nursing in mammals. Verh Dtsch Zool Ges 87:115–127

Korb J. and Heinze J. 2008. The ecology of social life: a synthesis. In Ecology of social evolution (eds

Korb J and Heinze J). pp 245-259.

Korb, J. 2010. Social insects, major evolutionary transitions and multilevel selection. Animal Behaviour:

Evolution and Mechanisms (2010).

Krupp, JJ, Kent, C, Billeter, JC, Azanchi, R, So, AKC, Schonfeld, JA, Smith, BP, Lucas, C, Levine, JC.

2008. Social Experience Modifies Pheromone Expression and Mating Behaviour in Male

Drosophila melanogaster. Current Biology 18(18): 1373-1383.

Lahti, D, Johnson, N. Ajie, B, Otto, S. Hendry, A, Blumstein, D, Coss, R, Donohue, K, Foster, S. 2009.

Relaxed selection in the wild. Trends in Ecology & Evolution, 24(9): 487–496.

Lande, R, SJ Arnold. 1983. The measurement of selection on correlated characters. Evolution 37: 1210-

1226.

Lazarus, J and I.R. Inglis. 1978. The breeding behaviour of the pink-footed goose: parental care and

vigilant behaviour during the fledging period. Behaviour 65(1): 62-88.

! 66 ! Lenski, R. 1999. A distinction between the origin and maintenance of sex - Commentary. J Evolution

Biol 12(6): 1034-1035.

Lima, SL. 1995. Back to the basics of anti-predatory vigilance: the group-size effect, Animal Behaviour

49(1): 11-20.

Lindström, J. 1999. Early development and fitness in birds and mammals. Trends in Ecology &

Evolution 14: 43–348.

Linksvayer TA, Wade MJ. 2009. Genes with Social Effects Are Expected to Harbor More Sequence

Variation Within and Between Species. Evolution 63:1685-1696.

Lubin, Y, Bilde, T. 2007. The Evolution of Sociality in Spiders. Advances in the Study of Behaviour 37:

83–145.

Lummaa, V, Clutton-Brock, TH. 2002. Early development, survival and reproduction in humans.

Trends in Ecology & Evolution 17: 141–147.

Lyon, B.E., Montgomerie, R.D., and L.D. Hamilton. 1987. Male parental care and monogamy in snow

buntings. Behavioural Ecology and Sociobiology 20 (5): 377-382.

Maynard Smith, J, Szathmáry, E. 1997. The Major Transitions in Evolution. New York: Oxford

University Press.

McComb, K, Semple, S. Coevolution of vocal communication and sociality in primates Biol Lett 2005 1:

381-385.

Mcglothlin JW, Moore AJ, Wolf JB, Brodie III ED. 2010. Interacting Phenotypes and the Evolutionary

Process. III. Social Evolution. Evolution:no-no.

! 67 ! Michod, R. 1999. Darwinian dynamics. Oxford: Oxford University Press.

Moore AJ, Brodie, ED, Wolf, JB. 1997. Interacting phenotypes and the evolutionary process: I. Direct

and indirect genetic effects of social interactions. Evolution 51(5): 1352–1362.

Mousseau, TA, Fox, CW. 1998. The adaptive significance of maternal effects, Trends in Ecology &

Evolution 13(10): 403–407.

Odling-Smee, FJ, Laland, KN and Feldman, MWL. 2003. Niche Construction: The Neglected Process of

Evolution, Princeton University Press, Princeton, NJ.

Okasha S. 2006. Evolution and the Levels of Selection. Oxford University Press: New York.

Olson, DM. 1994. The distribution of leaf-litter invertebrates along a neotropical altitudinal gradient.

Journal of Tropical Ecology, 10: 129–150.

Page, RE & Erber, J. Levels of behavioural organization and the evolution of

divisionoflabour.Naturwissenschaften89,91–-106(2002).

Pike, N, Foster, WA. 2008. The ecology of altruism in a clonal insect. In: Korb, J and Heinze, J, Editors.

Ecology of Social Evolution pp. 37–56. Springer-Verlag, Berlin.

Pollard, KA, Blumstein, DT. Time allocation and the evolution of group size, Animal Behaviour 76(5):

1683-1699.

Powers KS, Aviles L. 2007. The role of prey size and abundance in the geographical distribution of

spider sociality. J Anim Ecology 76:995-1003.

Premoli M, Sella G. 1995. Alloparental Egg Care in the Polychaete Worm Ophryotrocha-Diadema.

Ethology 101:177-186.

! 68 ! Price T, Millington S, Grant P. 1983. Helping at the nest in Darwin's finches as misdirected parental

care. The Auk 100:192-194.

Price, GR. 1970. Selection and covariance. Nature 227:520–521.

Pruitt, JN, Riechert, SE, Jones, TC. 2008. Behavioural syndromes and their fitness consequences in a

socially polymorphic spider, Anelosimus studiosus. Animal Behaviour 76: 871–879.

Pruitt, JN, Riechert, SE, Iturralde, G, Vega, M, Fitzpatrick, BM, Avilés, L. 2010. Population differences

in behaviour are explained by shared within-population trait correlations. Journal of

Evolutionary Biology 23: 748–756.

Purcell, J, Avilés L. 2007. Smaller colonies and more solitary living mark higher elevation populations of

a social spider. Journal of Animal Ecology 76(3): 590–597.

Purcell, J, Avilés, L. 2008. Gradients of precipitation and ant abundance may contribute to the

altitudinal range limit of subsocial spiders: insights from a transplant experiment. Proc. Roy. Soc.

B 275(1651): 2617-2625.

Queller, DC. 1992. Quantitative genetics, inclusive fitness and group selection. Am. Nat. 139: 540–558.

R Development Core Team. 2008. R: A language and environment for statistical computing. R

Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL

http://www.R-project.org.

Rahbek, C. 1997. The relationship among area, elevation, and regional species richness in neotropical

birds. The American Naturalist 149: 875–902.

Rauter CM, Moore, AJ. 2002. Evolutionary importance of parental care performance, food resources,

! 69 ! and direct and indirect genetic effects in a burying beetle. Journal of Evolutionary Biology 15:

404–417.

Riedman, M.L. 1982. The Evolution of Alloparental Care and Adoption in Mammals and Birds. Q Rev

Biol 57 (4): 405-435.

Roberts BD, Hatch SA. 1994. Nestling movements and adoption in a colony of black-legged

kittiwakes.Wilson Bull 106:289-298

Robinson, GE, Fernald, RD, Clayton, DF. 2008. Genes and Social behaviour. Science 7, 322 (5903):

896–900.

Roland C, Gundermann J, Horel A. 1996. Maternal state induction in female spiders by the young.

Behaviour 133:1125-1131.

Rubenstein, DR. & Lovette, IJ. 2007 Temporal environ- mental variability drives the evolution of

cooperative breeding in birds. Curr. Biol. 17, 1414–1419.

Russell AF, Hatchwell BJ. 2001. Experimental evidence for kin-biased helping in a cooperatively

breeding vertebrate. Proc R Soc Lond B Biol Sci. 268:2169–2174.

Ruttan, L. 1991. Effects of maternal presence on the growth and survival of subsocial spiderlings

(Araneae: Theridiidae). Journal of Insect Behaviour 4 (2): 251-256.

Salomon M, Lubin Y. 2007. Cooperative breeding increases reproductive success in the social spider

Stegodyphus dumicola (Araneae, Eresidae). Behavioural Ecology and Sociobiology 61:1743-

1750.

! 70 ! Salomon M, Mayntz D, Lubin Y. 2008. Colony nutrition skews reproduction in a social spider. Behav

Ecol 19:605-611.

Schneider J. 2002. Reproductive state and care giving in Stegodyphus (Araneae : Eresidae) and the

implications for the evolution of sociality. Animal Behaviour 63:649-658.

Schwarz MP, Bull NJ, Cooper SJB. 2003. Molecular Phylogenetics of Allodapine Bees, with Implications

for the Evolution of Sociality and Progressive Rearing. Systematic Biology 52:1-14.

Selander, RK. 1964 Speciation in wrens of the genus Campylorhynchus. Univ. Calif. Publ. Zool. 74, 1–

224.

Solomon, NG, French, JA. 1997. Cooperative breeding in mammals. Cambridge: Cambridge University

Press.

Sonerud GA, Smedshaug CA, Bråthen O. 2001 Ignorant hooded crows follow knowledgeable roost-

mates to food: support for the information centre hypothesis. Proc. R. Soc. B. 268: 827–831.

Stafford R. 2008. Exaptation and emergence as mechanisms to cross fitness valleys during evolution: an

example using simulated homing behaviour.

Tella J, Forero M, Donazar J, Negro J, Hiraldo F. 1997. Non-adaptive adoptions of nestlings in the

colonial lesser kestrel: Proximate causes and fitness consequences. Behavioural Ecology and

Sociobiology 40:253-260.

Therneau, T, Lumley, T. 2009. survival: Survival analysis, including penalised likelihood. R package

version 2.35-7. http://CRAN.R-project.org/package=survival

! 71 ! Toth AL, Robinson GE. 2007. Evo-devo and the evolution of social behaviour. Trends in Genetics

23(7): 334-341.

Toth AL, Robinson GE. 2009 Evo-devo and the evolution of social behaviour: brain gene expression

analyses in social insects. Cold Spring Harb. Symp. Quant. Biol. 74: 419–426.

Toth AL, Varala K, Newman TC, Miguez FE, Hutchison SK, Willoughby DA, Simons JF, Egholm M,

Hunt JH, Hudson ME. 2007. Wasp Gene Expression Supports an Evolutionary Link Between

Maternal Behaviour and Eusociality. Science 318:441-444.

Uetz, GW, Hieber, CS. 1997. Colonial web-building spiders: Balancing the costs and benefits of group-

living. In: Choe, J and Crespi, BJ, editors. Evolution of social behaviour in insects and arachnids.

Cambridge University Press: New York. p. 458–475.

Venables, WN. & Ripley, B.D. 2002. Modern Applied Statistics with S. Fourth Edition. Springer, New

York.

Vollrath F. 1986. Eusociality and extraordinary sex ratios in the spider Anelosimus eximius (Araneae:

Theridiidae). Behav Ecol Sociobiol.

Wcislo W.T. 2000. Environmental hierarchy, behavioural contexts, and social evolution in insects. In:

Martins, RP, Lewinsohn, TM and MS Barbeitos, Editors. Ecologia e Comportamento de Insectos.

Rio de Janeiro, Brasil. Oecol. Brasil. suppl. 8: 49–84.

Wcislo, WT. 1989. Behavioural Environments and Evolutionary Change. Annual Review of Ecology and

Systematics 20: 37–169.

Wcislo, WT. 2000. Environmental hierarchy, behavioural contexts and social evolution in insects. pp. 49-

84. In: Martins, RP, Lewinsohn, TM & Barbeitos, MS, editors. Ecologia e comporatamento de

! 72 ! Insetos. Series Oecologia Brasiliensis, vol. VIII. PPGE-UFRJ. Rio de Janeiro, Basil.

Wenseleers, T, Gardner, A, and KR Foster. 2009. Social evolution theory: a review of methods and

approaches. in Szekely,T, Komdeur, J and AJ Moore, editors. Social behaviour: genes, ecology

and evolution. Cambridge Univ. Press Cambridge, UK .

West-Eberhard, MJ. 1979. Sexual selection, social com- petition and evolution. Proceedings of the

American Philosophical Society 123:222–234.

West-Eberhard, MJ. 1983. Sexual Selection, Social Competition, and Speciation. The Quarterly Review

of Biology 58(2): 155-183.

West-Eberhard, MJ. Developmental Plasticity and Evolution. 2003. Oxford Univ. Press, New York.

West-Eberhard, MJ. 1987. Flexible strategy and social evolution. 35–51 in Ito, Y., Brown, J. L. &

Kikkawa, J., editors. Animal Societies: Theories and Fact. Japan Sci. Soc. Press, Tokyo.

Wheeler, WM. 1928. The Social Insects. Kegan Paul, London.

Williams, GC. (Ed.) 1971 Group Selection. Aldine·Atherton, Chicago, IL.

Willis CKR, Brigham RM. 2007. Social thermoregulation exerts more influence than microclimate on

forest roost preferences by a cavity-dwelling bat. Behav Ecol Sociobiol 62:97–108.

Wilson DS, O'Brien DT, Sesma A. 2009. Human prosociality from an evolutionary perspective:

variation and correlations at a city-wide scale. Evolution and Human Behaviour 30:190-200.

Wilson, EO. 1975. Sociobiology: The New Synthesis. Cambridge, MA: Harvard-Belknap.

Wilson, EO and Holldobler, B. 2005 Eusociality: origin and con- sequences. Proc. Natl Acad. Sci. USA

102, 13367-13371. (doi:10.1073/pnas.0505858102)

! 73 ! Wilson, EO. 2005 Kin selection as the key to altruism: its rise and fall. Soc. Res. 72, 159-66.

Wisenden B. 1999. Alloparental care in fishes. Reviews In Fish Biology And Fisheries 9:45-70.

Wolf JB, Brodie III ED, Cheverud JM, Moore AJ, Wade MJ. 1998. Evolutionary consequences of

indirect genetic effects. Trends Ecol. Evol. (Amst.) 13:64-69.

Wolf JB, Brodie III ED, Moore A. 1999. Interacting phenotypes and the evolutionary process. II.

Selection resulting from social interactions. Am Nat 153:254-266.

Wolf JB, MJ Wade & ED Brodie III. 2001. Evolution when the environment contains genes. In: Scheiner,

S, and DeWitt T, editors. Plasticity. Oxford University Press, Oxford.

Wolf, JB, Moore AJ. 2010. Interacting phenotypes and indirect genetic effects: a genetic perspective on

the evolution of social behaviour. In Westneat, DF and Fox CW, editors. Evolutionary

Behavioural Ecology. Oxford Univ. Press, New York.

Wright, J, Dingemanse, N. 1999. Parents and helpers compensate for experimental changes in the

provisioning.

! 74 !

Appendix A

Simulation of estimate accuracy in alloparental care assays

We simulated the effect of the number of females in a colony on the accuracy of our switching rate assay using an individual-based simulation in R (R core team 2011).

Simulation details

We simulated 20 “colonies” of spiders for each switching level selected (this is equivalent to simulating a single colony with 20 times the number of observations, but was included for interpretive similarity to our empirical results). Each simulated colony contained 5 – 800 females, and a maximum of 40 eggs

(note: the number of eggs and the ratio of eggs to females does not bias estimates of egg switching; data not shown). Before the simulation is run, a switching rate is chosen. This rate corresponds to the quantity we attempted to estimate in our study, i.e. the probability that the identity of a female caring for an egg sac changes in a subsequent observation.

The simulation then functions as follows:

1. The number of females is initialized. We simulated the following number of females: 5, 10–100

by 10s, 150, and 200–800 by 100s.

2. A matrix of marked egg sacs and females is constructed (always indexed as eggs 1-5 and females

1-5).

3. The switching round: for each egg sac, a random number is generated between 0 and 1. If this

number is less than or equal to the chosen switching probability, a new female is assigned to this

egg. Only females not already clutching eggs are assigned to new eggs.

4. The checking round: each egg is checked to see if the same female has remained with the egg

! 75 ! after the switching round. This is scored as a “0”; any other meaningful change (unmarked ->

marked, marked -> unmarked, marked -> marked) is scored as a “1” (this is the same scheme

used in our study in Chapter 2).

5. Steps 3-4 are iterated 100 times for each of 20 colonies.

6. Steps 2-5 are iterated for each number of females described in 1. The simulation is run once for

each manual parameter setting of switching rate.

7. The simulated dataset is formatted and the switching rate is estimated using a logistic regression

model with a probit error function.

8. Results are plotted using ggplot2.

Discussion of simulation results

We found that for simulated switching rates under 0.7, our sampling method produces consistent

underestimates of switching rate (Figure A.1). The degree of this underestimate increases with the

number of females in the colony. That is, in the absence of all other forces we should generally

expect estimated female switching rate to be underestimated in large colonies. Thus for the social

species we assayed, our estimates of alloparental care are conservative.

! 76 !

Figure A.1 The relationship between number of females and estimated switching probability in a simulation of egg-sac switching behaviour.

Red lines correspond to the true switching rate (set as a model parameter) and black lines and confidence intervals correspond to estimates

of this quantity derived from simulated data. Each value of “number of females” has 2000 simulated “switch” events (n= 58000 total).

! 77 ! #R Simulation of egg-switching sampling bias #Kieran Samuk ([email protected]) #Last modified April 07/2011

##SIMULATION##

#required libraries require(fields) require(reshape) require(ggplot2)

#loop counters ca<-numeric() b<-numeric() y<-numeric() i<-numeric() z<-numeric()

#initials dataset<-c() # the simulated dataset for ONE colony observation<-c() #observations made in ONE timestep. joined together to make the dataset. matchprops<-c() # the proportion of observations in the dataset in which the egg and female had the same identity. finaldataset<-c()# dataset for one whole run (one parameter setting of 'switchprob' below) plots<-c()#final dataset for used for plotting ('finaldataset' is serially appended to this variable)

#simulation parameters timesteps<-100 #number of observations made per colony (less = more variance) reps<-20 #number of colonies (as above) noeggs<-40 #number of eggs per colony(has no effect) eggs<-c() females<-c() switchprob<-0.3 # the “probability of switching” parameter. needs to be manually changed each run. nomarks<-5 #number of marked eggs/females

#main loop: one interation= one parameter setting for the number of females for(ca in c(5,10,20,30,40,50,60,70,80,90,100,150,200,300,400,500,600,700,800)){ nofems<-ca

#colony loop; one iteration = measurements of one simulated colony for (b in 1:reps){

carematrix<-c() dataset<-data.frame() observation<-c() markedeggs<-c(1:5) markedfems<-markedeggs carematrix<-data.frame(markedeggs,markedfems) names(carematrix)<-c("markedeggs","fems") number.eggs<-length(carematrix$markedeggs)

! 78 !

#timesteps loop; one iteration = one simulated observation of a colony for(i in 1:timesteps){

femsample<-sample(c(1:nofems),nofems,replace=F) switching<-runif(length(carematrix$fems))<=switchprob switched<-rownames(carematrix[switching,]) if(length(switched)>0){ carematrix[switched,]$fems<- femsample[1:length(carematrix[switched,]$fems)] }

#number.eggs loop; 1 loop = 1 observation of a single egg in a single colony for (y in 1:number.eggs){ if(any(markedfems==carematrix$fems[y])){ observation<- c(i,carematrix$markedeggs[y],carematrix$fems[y]) dataset<-rbind(dataset,observation) } else{ observation<-c(i,carematrix$markedeggs[y],0) dataset<-rbind(dataset,observation) }

}#end number of eggs loop }#end timesteps loop

check<-c()

#checks for matches (egg colour = female colour)

for(i in (nomarks+1):length(dataset[,1])){ if(dataset[(i-nomarks),3]==dataset[i,3]){ check<-append(check,0) } else{ check<-append(check,1) }

}

#initial dataset formatting dataset<-cbind(dataset[(nomarks+1):length(dataset[,1]),1:3],check) names(dataset)<-c("time","egg","female","switch") dataset<-subset(dataset,dataset$female!=0)

#dataset quality control and assembly. runs only if there are data to process. if(nrow(dataset)>0){ finaldataset<- cast(melt(with(dataset,table(egg,switch)),id=c("egg","switch")),egg~ switch)

#deals with situation where every observation was a "switch" event

! 79 ! if(!any(dataset$switch==0)){ tempswitch<-c(finaldataset$"1"/finaldataset$"1") finaldataset<-data.frame(finaldataset$egg,tempswitch,1 tempswitch,c(finaldataset$"1"),switchprob-tempswitch) #note line break }

#deals with situation where there are no switches else if(!any(dataset$switch==1)){ tempremain<-c(finaldataset$"0"/finaldataset$"0") finaldataset<-data.frame(finaldataset$egg,1- tempremain,tempremain,finaldataset$"0",(switchprob-(1- tempremain))) #note line break }

else{ tempswitch<- c(finaldataset$"1"/(finaldataset$"0"+finaldataset$"1")) finaldataset<-data.frame(finaldataset$egg,tempswitch,1- tempswitch,finaldataset$"0"+finaldataset$"1",switchprob- tempswitch) #note line break }

#assembles final data set (= data for one colony) names(finaldataset)<-c("egg","switch","remain","n","bias") colony<-rep(b,length(finaldataset$egg)) females<-rep(nofems,length(finaldataset$egg)) finaldataset<-cbind(colony,females,finaldataset) plots<-rbind(plots,finaldataset) # this is the “big” dataset everything is appended on to } # end dataset assembly } # end colony loop print(nofems) # this displays the progress of the simulation in the console window

}# end females loop

##END SIMULATION##

##PLOTTING##

#ggplot mappings (note these span multiple lines) p<- opts(panel.background=theme_blank(),panel.grid.minor=theme_blank(),axis.li ne=theme_segment(),axis.text.x=theme_text(colour='black',vjust=1,size=14), axis.text.y=theme_text(colour='black',hjust=1,size=14)) gp<-geom_point(colour=alpha('black',0.05),size=4) q<-scale_y_continuous(limits=c(-1,1)) q0<-scale_y_continuous(limits=c(0,1)) a<-aes(x=females/nomarks,y=-bias+1) h<-geom_hline(yintercept=yi,colour="red",linetype=2,size=2) sm<- stat_smooth(method=glm,family=binomial(link="probit"),size=2,color="black" )

#plot of 0.2 switching rate

! 80 ! smo<-ggplot(plots0.2,aes(x=females,y=switch)) yi<-0.2 smo+sm+gp+q0+p+geom_hline(yintercept=yi,colour="red",linetype=2,size=2)

#plot of 0.4 switching rate smo1<-ggplot(plots0.4,aes(x=females/nomarks,y=switch)) yi<-0.4 smo+sm+gp+q0+p+geom_hline(yintercept=yi,colour="red",linetype=2,size=2)

#plot of 0.6 switching rate smo<-ggplot(plots0.6,aes(x=females/nomarks,y=switch)) yi<-0.6 smo+sm+gp+q0+p+geom_hline(yintercept=yi,colour="red",linetype=2,size=2)

#plot of 0.8 switching rate smo<-ggplot(plots0.8,aes(x=females/nomarks,y=switch)) yi<-0.8 smo+sm+gp+q0+p+geom_hline(yintercept=yi,colour="red",linetype=2,size=2)

! 81 ! Appendix B

The effect of paint marks on spider care behaviour

We conducted an experiment to test for the effects of a i) paint mark on a female’s abdomen and ii) a paint mark on an egg sac on a female spider’s care behaviour.

Methods

After performing our study of egg sac preference described in Chapter 2, we set aside

20 female spiders with egg sacs. Ten of these were A. elegans, and ten were A. guacamayos.

We randomly chose 5 females from each species (10 total) to receive abdominal paint marks as well as have their egg sacs painted. Using the painted and unpainted spiders, we assayed the probability of accepting one’s own egg sac when presented with it, and the probability of clutching the egg sac 1 hour and 2 days after receiving the mark (or not).

Results

All twenty A. elegans and A. guacamayos females immediately accepted the egg sac presented to them, regardless of whether it had been marked (!2 = 0, df=1, p=1). Further, there was no significant difference in the probability of clutching egg sacs between painted and unpainted treatments at 30 minutes or 48 hours [Table B.1; 30 minutes: !2 = 0, df=1, p=1

(both species), 48 hours: !2 = 0.42, df=1, p=0.52 (A. elegans); !2= 0.62, df=1, p=0.43 (A. guacamayos)].

Discussion

These results suggest that female spiders do not alter their maternal care behaviour in response to paint marks on their egg sacs. This is consistent with our field assays of subsocial females (described in the text), in which we painted single females with egg sacs and observed clutching behaviour in nearly every case (See Figure 2.2).

! 82 ! Table B.1 Counts of number of marked and unmarked females of two species of cobweb spiders clutching egg sacs at two time points. Clutching was scored as the female holding the egg sac with her palps and/or chelicerae.

Time / Behaviour

30 minutes 48 hours

Species Treatment Clutching Not clutching Clutching Not clutching

Not painted 5 0 3 2 A. elegans Painted 4 1 3 2 Not painted 5 0 3 2 A. guacamayos Painted 5 0 5 0

! 83 ! Appendix C

Pairwise tests of maternal care behaviour contrasts

Table C.1 Paired t-tests of phylogenetically independent contrasts of six measures of maternal care behaviour between social and subsocial species of cobweb spiders. Each mean pairwise difference represents the mean of two differences in maternal care response between pairs of social and subsocial species. Bold values denote significant results at the α =0.05 level. See the caption of Table 3.1 for behaviour definitions and contrast units.

Mean pairwise difference Measurement n (pairs) SE t-value p-value (social – subsocial)

Female distance 1.52 2 0.007 200.6 0.003 Attendance -0.19 2 0.014 -13.74 0.046 Acceptance -0.10 2 0.082 -1.268 0.42 Repair -0.34 2 0.200 -1.7 0.34 Relinquishment -113.35 2 57.42 -1.974 0.299 Reclamation 3.7 2 1.98 1.85 0.32

! 84 !