The Influence of Demography, Development and Death on Seasonal Labor Allocation in the Florida Harvester Ant (Pogonomyrmex Badius) Christina L

Florida State University Libraries

Electronic Theses, Treatises and Dissertations The Graduate School

2014 The Influence of Demography, Development and Death on Seasonal Labor Allocation in the Florida Harvester Ant (Pogonomyrmex Badius) Christina L. Kwapich

Follow this and additional works at the FSU Digital Library. For more information, please contact [email protected] FLORIDA STATE UNIVERSITY

COLLEGE OF ARTS AND SCIENCES

THE INFLUENCE OF DEMOGRAPHY, DEVELOPMENT AND DEATH

ON SEASONAL LABOR ALLOCATION IN THE FLORIDA HARVESTER ANT

(POGONOMYRMEX BADIUS)

CHRISTINA L. KWAPICH

A Dissertation submitted to the Department of Biological Science in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Degree Awarded: Fall Semester, 2014 Christina L. Kwapich defended this dissertation on October 2, 2014 The members of the supervisory committee were:

Walter R. Tschinkel Professor Directing Dissertation

Frederick R. Davis University Representative

Emily H. DuVal Committee Member

Lisa C. Lyons Committee Member

Jeanette L. Wulff Committee Member

The Graduate School has verified and approved the above-named committee members, and certifies that the dissertation has been approved in accordance with university requirements.

ACKNOWLEDGMENTS

Above all, I would like to thank my most treasured teacher and mentor, Walter Tschinkel. For each day spent together in the sandhills or over a cup of coffee, I count myself lucky. From identifying, dissecting, drawing and extracting fat from small creatures, to digging a square hole and crafting a sturdy table, I am better because of the countless techniques and concepts you taught me. I am also grateful for the silent things you imparted by example.You are a scientist, person and friend worth emulating, and it was an honor to be your final graduate student. I would also like to thank my committee members, Fritz Davis, Emily DuVal, Lisa Lyons and Janie Wulff, who encouraged me to see societies at other scales and the parallel lives of ants, patiently taught me a number of topics and how to teach, offered praise and insight from other disciplines, and inspired a fondness and reverence for the history of my field. I am indebted to my undergraduate advisor Susan C. Jones for her many years of support and kindness and to John Wenzel, for his mentorship and wisdom in matching me with my major professor. I would also like to recognize Josh King, Nicky Gallagher, Christine Johnson, Sean Collins, Joan Her- bers, George Keeney, Mark Gurevich, Laura Chisholm and Mark Deering. Each has been a friend to me. I am thankful for the funding I received from the National Science Foundation including the Integrated Training in Biology and Society grant, Doctoral Dissertation Improvement Grant (number IOS-1311473), and an RA through Walter Tschinkel’s grant (number IOS-1021632). The Department of Biological Science at FSU also offered funding, support and comfort during my stay. Virginia Carter, Jen Kennedy, Judy Bowers and Ben Miller each made me feel at home. My lab mate, Tyler Murdock, also provided 6 years of assistance, companionship and scientific discourse beneath the pines. Nicholas Hanley also cheerfully recaptured ants with me for many hours. I am grateful for the friendship of my cohort-mates, Bonnie Garcia, Caroline Stahala and Anna Strimaitis; Andrew Merwin and my other Pancrustacean colleagues, my fellowship-mates, Martha Lang, Abe Gibson and Chelse Prather; and for EERDG and SB2013. I am indebted to my family. When I was five, my mother filled a green binder with paper, drew a lady bug on the front and told me to take good field notes. She went on to find me every book on ants ever written and tolerated the poorly hidden menagerie I kept in my room and the dollhouse that she built for me. Sensing my deepest hopes, my father bought me my first microscope, which I used to visit many alien worlds. He also drove me to my favorite Cremato- gaster nest each Sunday, and urged me to dump an entire serving of French toast on top. He has been a patron of the ants for more than 20 years now and I would not be here without his loving support. My sister has been my best friend and most ardent fan (despite the tennis court inci- dent). Her imagination, cleverness, and constant companionship remind me what it’s all about. Above all she imparted a set of study habits that obviously work.

iii

TABLE OF CONTENTS

List of Tables ...... v List of Figures ...... vii Abstract ...... viii 1. INTRODUCTION ...... 1 2. DEMOGRAPHY, DEMAND AND DEATH ...... 4 2.1 Introduction ...... 4

2.2 Methods...... 6

2.3 Results ...... 12

2.4 Discussion ...... 24

3. LONGEVITY AND SOCIAL INHIBITION OF FORAGING ...... 31 3.1 Introduction ...... 31

3.2 Methods...... 33

3.3 Results ...... 39

3.4 Discussion ...... 51

4. CONCLUSION ...... 56 4.1 The value of field studies ...... 56

4.2 In support of the superorganism ...... 57

4.3 The annual cycle of a Florida harvester ant colony ...... 58

REFERENCES ...... 61

BIOGRAPHICAL SKETCH ...... 67

LIST OF TABLES

2.1 Important dates and events related to foraging across the annual cycle from 2009-2012...... 30

3.1 Results for quasibinomial Generalized Linear Models for the effect of range-limitation on forager survival in wild colonies ...... 41

3.2 Results for quasibinomial Generalized Linear Models of the effect of season on survival...... 41

LIST OF FIGURES

2.1 (a) A wire-marked P. badius forager...... 10

2.2 Foragers were always found in near-surface chambers...... 13

2.3 Foraging began in March or April of each focal year, reaching a proportional maximum between May and June before declining to 0%, prior to overwintering...... 14

2.4 Foraging began in early spring and reached a proportional maximum in early summer, before declining to zero in late autumn...... 15

2.5 The ratio of foragers to larvae (1.6 to 1) is not significantly related to date between May and mid-October of each year despite a dramatic change in the proportion of the colony allocated to foraging (gray, dotted curve and right Y-axis) ...... 16

2.6 The proportion foragers and larvae (as a fraction of all colony members) change in paral- . lel, so that their ratio is constant from May to October...... 17

2.7 At the time of excavation, workers of each cuticle-color category were marked with a different colored wire belt...... 19

2.8 The mean proportion of each colony represented by larvae, pupae, callows, middle or dark workers at four points during the annual cycle, averaged across years (means with standard error bars)...... 20

2.9 For each focal colony, forager number on each sample date was divided the May’s estimate to obtain a factor of change in forager number (means with standard error bars)...... 21

2.10 When half of the forager population was removed it was not replaced 7 days later (means with standard error bars, n=5 controls, 5 forager removal)...... 23

3.1 Range-limited nests were enclosed in 60cm X 60cm X 10cm screen-bottom, open- topped, aluminum boxes that prevented long-distance foraging...... 35

3.2 Forager survival over 20 days was significantly higher in range-limited colonies than in control colonies during each season (quasibinomial GLM, p<0.0004), but differed among seasons (asterisks denote significance relative to controls within sea sons, quasibinomial GLM, p<0.005)...... 40

3.3 Proportion of foragers surviving in control colonies was estimated by mark recapture at varying intervals after the initial marking event...... 42

3.4 Between May and August, complete turnover of the initial forager population occurred at

a mean of 27 day in control colonies, 44 days for fed-enclosed colonies and 41 days for starved-enclosed colonies...... 43

3.5 From May-August the loss of foragers was matched or exceeded by the addition of new workers (replacement), and forager populations either grew or maintained their size...... 44

3.6 For each colony, the observed forager population size on day 20 was divided by forager population size from day one to obtain a factor of change in size...... 45

3.7 In spring/summer control colonies, 69% of the forager population was composed of new workers on day 20, while enclosed-fed and enclosed-starved colonies averaged 43% and 46% respectively...... 46

3.8 Kaplan-meier estimate of survival probability at each week for treatments and months (n= 1,400 foragers from 3 to 4 colonies per season, divided between starved and fed treatments)...... 47

3.9 Percent body fat was greater in Sept.-Oct. than in May-Aug for control colonies ...... 49

3.10 Neighbor removal improved forager survival for colonies sampled in May-June (+30%) and Sept. – October, but not for colonies sampled in July-August (left panel) ...... …… 50

vii

ABSTRACT

Eusocial insect societies are analogous to organisms in that the demography, development and regulation of workers within are shaped by selection acting on whole colony characteristics. Just as relative investment varies across the lifetime and reproductive cycle of a traditional organism, adaptive patterns of worker allocation are expected to vary with colony development and need across each annual cycle. Despite these predictions, adaptive patterns of labor allocation remain un-described for most social insect societies. This dissertation identifies a seasonal pattern of forager allocation in colonies of the Florida harvester ant (Pogonomyrmex badius) and describes its relationship to colony demography, size, reproduction, worker development rate, death rate, longevity, and neighborhood dynamics. Aging P. badius workers progress through a sequence of interior labor roles before leaving the nest to forage. By marking and recapturing foragers, forager population size was estimated and foragers were identified as a discrete, age-correlated labor group that resides only in the top 12cm of nests that may be more than 200cm deep. Excavation and census of whole colonies revealed that foragers were present in a consistent ratio to the colony’s larval population from May through August, but that forager allocation was not a response to larval presence. Proportional allocation to foraging followed an annual pattern, shaped by the interaction of seasonal phases of colony growth and worker development rate. Forager allocation began in March or April and increased to a peak of approximately 40% of the colony in June, as colonies provisioned alates for mating flights in the days surrounding the summer solstice. In spring, proportion foraging increased due to an increase in forager number combined with a reduction in colony size. Beginning in late summer, proportional allocation to foraging decreased, as colonies grew through new worker birth and forager replacement declined. This annual pattern was shaped by a five-fold difference in the age of summer and autumn-born workers when they entered the forager population (43 vs. 200+ days). The chronological age of foragers was revealed by collecting whole colonies across two annual cycles, marking age cohorts with colored wire-belts, releasing each colony into a field nest created from melting buried ice chambers, then monitoring the forager population for the appearance of each marked cohort. Slow-developing workers, produced from late August until mid-October each year, dominated the forager population the following March through mid-July; while fast developing

viii

workers appeared in early June and developed rapidly to become foragers the following month, overlapping with their older sisters. While wild foragers of both types lived an average maximum of 27 days after entering the forager population, these same foragers were capable of surviving for hundreds of days in the laboratory. Likewise, restricting the foraging range of wild foragers increased forager longevity by 57%, demonstrating that foraging carries mortality risks and the observed age at death was not part of a developmental program involving senescence in P. badius. By removing neighboring colonies, this study also showed that interactions with conspecific neighbors can influence the labor thresholds of individual workers, and the demographic structure of whole colonies, as neighbors account for 30% of forager mortality in the spring. At the colony level, increased forager longevity suppressed the movement of new workers into the forager caste, increasing their time in earlier labor roles and promoting colony growth. In contrast, both removing 50% of the forager population and doubling the larval population did not induce forager replacement or increase the daily rate of new foragers added within seven days. Together, these results suggest a unidirectional control of labor allocation in P. badius, where the forager population size is not maintained by workers detecting colony need and filling vacancies, but by workers developing at a rate selected to allow forager replacement. In essence, the annual cycle of forager allocation emerges as P. badius workers ‘age’ into behavioral roles at environmentally appropriate times, in the same proportions, on nearly the same dates each year and experience a predictable death rate. This process allows colonies to di- vide a limited number of workers between competing functions without a leader.The findings of this study reinforce our understanding of the organism-like nature of social insect colonies. Like cells in a body, the thousands of individual insects in a P. badius colony are organized into functional labor groups, which are responsive to cycles of growth, reproduction and dormancy through self-regulating processes. The emergence of measurable, colony-level traits from the accumulation of thousands of transient individuals, from multiple generations is one of the most striking feats of social organization across taxa.

CHAPTER ONE INTRODUCTION

In eusocial insect colonies, behavioral castes are composed of workers that specialize on sets of non-reproductive tasks over varying time scales (Wilson, 1968). In order to optimize growth and reproduction, proportional membership in each caste must change in response to seasonal differences in colony need (Oster and Wilson, 1978; Fukuda, 1983). Although the maintenance of complex caste structure is a key feature of social evolution, whether annual patterns of labor allocation exist and how they arise from the self-organization of thousands of individuals remains unknown for almost all social insect species (Schmid-Hempel, 1992; Tschinkel, 2011). An abundance of theoretical work on the topic has emerged over the last forty years (Robinson, 1992; Schmid-Hempel, 1992; Beshers and Fewell, 2001).Yet, in practice, untangling the processes underlying caste switching has been difficult because worker age, experience, physiological condition, location in the nest and task-set are often correlated (Tofts, 1993; Franks and Tofts, 1994). In social hymenoptera colonies, aging workers typically move through a sequence of behavioral castes, beginning with brood care and culminating in risky labor roles outside of the nest (Lindauer, 1953). The “centrifugal” movement of workers away from the brood pile and to- wards developmentally demanding tasks, suggests that rate and timing of birth, development and death among workers may drive the seasonal redistribution of labor (Wilson, 1976a; Bonabeau et al., 1998; Page and Mitchell, 1998). Under these conditions, colony-level selection is expected to shape age-frequency distributions so that appropriately-aged workers are aligned with the temporal availability of resources, or other predictable events during the annual cycle (Oster and Wilson, 1978). Alternatively, workers have been modeled as flexible generalists, capable of moving between local behavioral roles fluidly when changes in colony need are detected (Gordon, 1996; Beshers and Fewell, 2001). In these models, correlations between worker age and task set arise from the sequential filling of labor gaps in structured nest space, instead of directly from meeting developmental thresholds. As individuals detect hallmarks of changing colony need and task occupancy, they ‘forage’ for work, moving ever farther from the brood pile (Sendova-Franks and Franks, 1993; Franks and Tofts, 1994).

In reality, laboratory observations and experiments examining behavioral flexibility have identified a range of colony strategies related to task switching and have found seemingly con- flicting frameworks operating in the same nest. In Pheidole dentata, for example, the large repertory size of old minors is correlated with increases in structure and volume of the lip region of the mushroom bodies, which govern olfaction (Seid et al., 2005). Although the task set of young minors is constrained by neural capacity, older individuals acquire an increasingly elaborate repertory-size, and can potentially switch roles as changes in colony need arise (Seid and Traniello, 2006). In honeybees, labor is discretized and attended by age classes with almost no overlap in task-set (Johnson, 2008). Yet, the ability of workers to fill missing labor roles in single-cohort colonies has been demonstrated repeatedly in observation hives and behavioral reversions are common following swarm events in natural populations (Huang and Robinson, 1992; Robinson et al., 1992).

The capacity of workers from many social insect species to accelerate, reverse or slow development has led to the popular consensus that social insects more often adopt strategies involving behavioral flexibility over adaptive schedules of birth and development (Meudec and Lenoir, 1982; McDonald and Topoff, 1985). However, when workers fill experimentally induced labor gaps, their capacity to perform the new task set as efficiently as their predecessors may be limited by their developmental age (Calderone, 1995). It is therefore possible that behavioral plasticity acts only as buffer during catastrophes for some species, as the ideal ratio of developmentally appropriate age-cohorts is slowly restored. For example, in colonies of Neoponera api- calis, workers from colony fragments composed of single age-classes were capable of filling missing labor roles, but when recombined with their parent nest, resumed their previous age- specific tasks (Lachaud and Fresneau, 1987). Therefore, although behavioral flexibility is likely a critical component of colony resiliency, it may be best considered within the framework of a colony’s age-caste structure (Robinson, 1992). This dissertation examines both the age-structure and responsiveness of ant colonies across the dynamic annual cycle.

Study species The Florida harvester (Pogonomyrmex badius) is a dimorphic, diurnal, seed harvesting ant characteristic of the coastal plain in the southeastern United States. Each colony contains a sin-

2 gle, multiply-mated queen and reaches maturity with a minimum adult population of approximately 700 workers, releasing its own sexual alates in June of each year thereafter (Golley and Gentry, 1964). In northern Florida, colonies are active between March and November of each year and may consist of as many as 11,000 workers that excavate nests more than 2.5 meters in depth (Tschinkel, 2004) . Within each nest, workers are vertically stratified by cuticle-color (a proxy for relative age), with brood and seeds appearing in proportion to colony size, at specific depths. Unlike colonies of western sister-species, which are nearly impossible to collect, a large P. badius nest can be neatly exhumed, with all colony members vacuumed and sorted by strata in 4 - 9 hours by one enthusiastic worker and a small shovel. The relative ease of collection, along with the existing body of sociometric data relating to the annual cycle of fat storage and brood production, make P. badius an ideal candidate for studies of labor allocation (Golley and Gentry, 1964; Tschinkel, 1998; 1999; Smith and Tschinkel, 2006).

CHAPTER TWO DEMOGRAPHY, DEMAND AND DEATH

2.1 Introduction

Division of labor in ants is achieved by morphological and age-correlated behavioral specialization and more rarely, dominance interactions among workers. The value of morphological caste ratios to colony functioning has been described by manipulating the proportions of worker size-frequencies and measuring resultant brood success (Wilson, 1984; Porter and Tschinkel, 1985; Beshers and Traniello, 1994). However, the natural ratios and functions of age-correlated labor groups remain almost entirely unknown among ant species. Most studies of non- morphological division of labor have focused on the behavioral repertories of individual ants or age-cohorts in a laboratory setting. While these studies have produced valuable ethograms that define behavioral caste boundaries, repertory sizes and sequences (Wilson, 1976a; b; Mirenda and Vinson, 1981; Pratt, 1994; Santos et al., 2005; Holbrook and Fewell, 2010). For ground nesting ants, laboratory studies also eliminate the spatial relationships of workers, forcing task switching to play out in single chambered, two-dimensional, soil-free observation nests (Bourke and Franks, 1995; Gordon et al., 2005). Under these conditions, the capacity of workers of different ages to perform tasks may be inconsistent with their chance of encountering them within the architecture of a natural nest (Tschinkel, 2004). Laboratory studies of labor allocation also force worker development to occur in the absence of annual temperature and food cycles and most importantly, natural worker death rates. If individuals ‘forage for work’ and are pushed or pulled into new roles based on current occupancy, the absence of externally induced mortality of old workers could increase forager tenure while delaying the entry of younger workers into that role (Franks and Tofts, 1994). Alternative- ly, if labor allocation is actually age-specific, the reduction of mortality among old workers could produce an ever-growing population of foragers, limited only by natural lifespan. In either case, worker nest-space interactions and age-task distributions are subject to distortion under laboratory conditions and may obscure an understanding of how selection has shaped patterns of adaptive demography across the annual cycle, with realized worker lifespan at the helm.

Field studies Within social insect colonies, striking differences in lifespan are realized by individuals of the same genotype. While queens may live decades, sterile workers, the “soma” of the colony, may live only a fraction of a year (Keeler, 1982; Tschinkel, 1987; Porter and Jorgensen, 1988; Wiernasz and Cole, 1995; Meyer et al., 2009). Alternative adult phenotypes and size-related differences in lifespan among sterile workers are most frequently the product of larval nutrition (Wilson, 1953; Calabi and Porter, 1989). Yet, unlike worker morphology, hallmarks of chronological age are subject to environmental influence in adults. Proxies for relative age within a nest such as cuticle color and mandibular wear are not accurate predictors of chronological age in the field because their rate of change varies with factors such as temperature, humidity and experience (Oettler and Johnson, 2009). As a result, tests of age-related division of labor in ground nesting ants must rely on permanently marking newly eclosed cohorts; a task which is often too destructive to attempt. In non-nomadic ants, only behavioral castes that appear on the nest surface can be accessed without destruction of parent nests. Several studies have used mark-recapture techniques to describe the population size of surface-castes (foragers, defenders, etc.) in the field, but only a handful have excavated, and performed whole colony censuses to determine the proportions of the total adult population represented (Golley and Gentry, 1964; Erickson, 1972; Porter and Jorgensen, 1980; Nobua-Behrmann et al., 2013). Fewer still have described annual or environmental changes in caste ratios, which offer more than a snapshot in time of the annual cycle of proportional allocation (Herbers et al., 1985; Calabi and Traniello, 1989; Sendova-Franks and Franks, 1993). Using estimates of forager population size and total colony size, Tschinkel (2011) was the first to identify seasonal and size-related differences in proportional allocation to foraging in an ant species. He demonstrated that small Solenopsis invicta nests contain proportionally more foragers in the autumn than their large counterparts, owed in part to differential investment in sexual and worker production in spring. Because small colonies invest in growth instead of reproduction in the spring, their large, autumn-time forager populations are capable of expanding territory boundaries at the expense of neighbors, increasing access to resources that will build alates the following spring (Tschinkel, 2011). Thus, it seems that, like morphological castes, behavioral caste-size serves an adaptive function with respect to annual patterns of growth and reproduction.

Allocation in the Florida harvester ant This study describes the proportion of workers foraging in each of 55 colonies of the Flori- da P. badius, sampled across four annual cycles of colony growth and reproduction and across a full range of colony sizes. This work is one of the first field-based, colony-level analyses of labor ratios that integrates season, colony size, worker death rate and chronological worker age. It represents a missing component in the current literature on adaptive demography and the organization of labor in social insects and connects individual worker traits to whole-colony (superorganism) characteristics.

2.2 Methods

Study site Studies were conducted from 2009 to 2012 in a 23 ha, sand hills habitat of the Apalachico- la National Forest, 16 Km southwest of Tallahassee, FL (latitude 30.35, longitude −84.41). The site was characterized by well-drained deep sand and an over story of 40-year-old long leaf pine (Pinus palustris), a midstory of turkey oak (Quercus laevis) and a ground covering that included dwarf huckleberry (Gaylussacia spp.), pricklypear (Opuntia spp.), beard grass (Andropogon spp.), gopher apple (Licania michauxii) and catbrier (Smilax spp). P. badius colonies appeared at an average density of one nest per 670 m2 alongside Trachymyrmex septentrionalis, Solenopsis geminata, Forelius pruinosus, Dorymyrmex bureni, Aphaenogaster floridana and a number of less common ant species. All P. badius colonies were returned to their original territories following census. Proportion foraging Allocation of labor with respect to season, demography and colony size was estimated as the proportion foraging in 48 active season and 7 winter colonies of the Florida harvester ant (P. badius). Colonies were sampled in all months and seasons and focal nests were selected haphazardly based on mound diameter so that a full range of colony sizes were represented across each annual cycle. Forager population size (N) was calculated using the Lincoln index mark-release- recapture method, where the proportion of the total forager population marked in an initial sample (m) was expected to be equivalent to the proportion of a recapture sample (n) that was marked (Lincoln, 1930). The Lincoln-Peterson correction was not applied to population esti-

6 mates, because it did not significantly alter the estimates, given the high proportion of workers recaptured in each sample (Bailey, 1952). Each estimate of forager number was followed by the immediate excavation and census of the focal colony, thereby allowing calculation of the proportion of the total adult population that foraged

M/N=m/n where M = number initially marked N= total population m= number marked in recapture sample n= number recaptured.

Mark-recapture procedure: On day one, a band of bird seed was positioned 150 cm from the margin of the focal nest mound so that the mound was completely encircled and all estab- lished trunk trails were intersected. Foragers were defined operationally as individuals that trav- eled 150 cm or more from the nest mound, collected bait and began a return trip to their nest. To encourage continued recruitment to baits, the foraging area was shaded by a beach umbrella and individuals were gathered at approximately 15 minute intervals until fewer than five were captured in 30 minutes. Following this three to six hour collection period, a perfume sprayer was used to apply two “spritzes” of 10% fluorescent printer’s ink in diethyl ether to all foragers simultaneously (Gan’s Ink, Supply Co., Los Angeles CA; Risk Reactor, Santa Ana, California; (Porter and Jorgensen, 1980). Individuals were not anesthetized for marking and resumed normal behavior immediately following mark application. The quality and uniformity of each mark was checked using an ultraviolet flashlight and foragers were released en masse onto their nest mound. For each colony, recapture followed the initial capture and marking by 24 hours to allow sufficient mixing of foragers and to reduce the effects of forager death and recruitment into the population. Each colony was baited in the same manner as on day one. However, because colonies were observed to reject the same bait on consecutive days, they were offered shortbread cookie crumbs instead of birdseed. Foragers attending baits were gathered by hand as described above and taken to the laboratory where they were placed in a tray and counted individually by

7 removal with an aspirator. Then, marked individuals were separated and counted under ultraviolet light. All foragers were returned to their nest mound later that same day.

Nest excavation On day three of each sampling event, a large pit was excavated adjacent to the focal nest and chambers were revealed by making thin, horizontal slices into the nest area. Exposed ants, seeds and brood were gently vacuumed from their chambers using a Dewalt D500 vacuum cleaner and sorted into separate boxes at every 20cm depth increment. The number of marked foragers, unmarked adults, larvae and pupae per stratum were counted individually by removal with an aspirator in the laboratory. In 2011 and 2012, the number of adult workers belonging to each of three, cuticle-color specific groups was also tallied. For consistency, worker cuticle col- or was scored against a PANTONETM color scale palette derived from the myPANTONE application for iphone (Pantone Inc., copyright 2012).The youngest workers were identified as PAN- TONE 602 C and PANTONE 121 C (light tan to butter yellow, with even coloration), middle- aged workers as PANTONE 144 C and PANTONE 1375C (orange to pale red, with uneven tan- ning), and older workers as PANTONE 490 C or darker (deep, reddish brown). Foragers were always PANTONE 490 C or darker. Every effort was made to excavate nests in the early morning before foragers departed, but in cases when less than 100% of marked foragers were recovered during nest excavation, an estimate of the number of marked and unmarked foragers afield was calculated from the proportion of marked workers absent at the time of excavation. The number of missing workers was then added to the total number of adults in each colony. To determine the proportion of adults allocated to foraging, the estimated forager population size was then divided by the total adult population size, including foragers absent during colony collection. Inter-annual patterns of forager allocation were assessed both by calendar date and by aligning the dates that colonies first foraged following winter dormancy. Following colony census, most colonies were released in their original territory where they re-excavated nests.

Change in forager number Seasonal variation in percent foraging may result from a change in the total number of adults in a colony or a change in the number of adults that forage, or both. For example, a gradu-

8 al decrease in proportion foraging over time could be a consequence of an increase in colony size and maintenance of forager number, maintenance of colony size and a decrease in forager number, or a decrease in both colony size and forager number. To determine how each population influenced proportion foraging, repeated monthly estimates of forager number were made for a set of 10 colonies in two size classes, throughout the annual cycle (five in 2010, five in 2011). Maturity was inferred from initial colony size estimates (> 800 workers) and confirmed by the appearance of alates during mating flights (Smith and Tschinkel, 2006). On each sample date between May and November, foragers were marked with fluorescent printer’s ink, released, recaptured after 24 hours and then returned to their mound following mark inspection. Though capture events were separated by a minimum of two weeks, one of three unique ink colors was applied to foragers so that surviving, previously marked individuals would not confuse the most recent estimate of population size. Weather permitting, each focal colony was sampled monthly between mid-April and late October. Foragers were baited with a millet-based, bird seed mixture and cookie crumbs. These items were chosen because unlike natural seeds which the ants cache, seeds in the commercial mixture germinated within days of collection (observed during excavation) and were discarded by the ants, while cookie crumbs had disappeared from nests within 24 hours. It is therefore un- likely that baiting colonies 6 hours per month significantly influenced larval production in these colonies. In order to compare foraging between years, each estimate of forager number was divided by May’s estimate and expressed as a proportion of May’s population size for the focal year. Estimates of number foraging were also divided by date-specific, proportion foraging estimates (see above) and used to approximate focal-colony size on each sample date. The expected proportion foraging on each sample date was determined by the neighbor-weighted interpolation of points between sample dates by Kriging (Beers and Kleijnen, 2004). In each year, mature and immature colony values for proportion foraging were Kriged separately to account for differences in forager number resulting from alate production. Focal colony growth was calculated by dividing estimated colony size in October by initial colony size in May of each year to obtain a factor by which colony size changed. For each year, the average change in colony size was ob- tained by summing these factors and dividing by the total number of colonies sampled.

Chronological forager age In 2012, six P. badius colonies were excavated and workers were divided into three demographic groups by using the aforementioned PANTONETM color scale values. Each demographic group was marked with a different color of 38 gauge copper wire and fluorescent printer’s ink and released (fig. 2.1). In order to detect seasonal differences in the length of time between worker eclosion and foraging, as well as the length of time required for each initial demographic group to appear in the forager population, colonies were collected, marked and released at three different periods during 2012: February (1 colony), following mating flights in June (3 colonies), and prior to winter dormancy in October and November (2 colonies).

Fig. 2.1 (a) A wire-marked P. badius forager. Whole colonies were sorted into three cuticle color groups according the PANTONETM color scale values (b) and given one of three wire belts colors depending on relative age.

Wire marking consisted of tightening an overhand knot of colored, 38 gauge copper wire around the petiole of each lightly anesthetized ant (Mirenda and Vinson, 1979), while ink was applied to the cuticle in the manner described above. Though more time consuming to apply, the permanence of wire-marks allowed for verification of the longevity of concurrently applied ink marks. Aside from an initial period of self-inspection, wire belts did not appear to alter the behavior of ants. During the wire-application process each colony was housed in a plaster laboratory nest for a maximum of 48 hours and offered frozen crickets (Acheta domesticus), water in cot-

10 ton-plugged tests tubes and access to their original seed cache. Mortality during this period was negligible.

Ice nests During colony collection, the chamber depth, number and area were recorded. To spare each colony energetic costs associated with re-excavation, a nest with similar features was con- structed out of ice, buried and allowed to melt in the location of the original nest. Ice-chambers of the appropriate dimensions were frozen in species-typical, copper molds, transported to the field on dry ice and sequentially packed in place in the nest pit (Tschinkel, 2013). Chambers were linked together by threading plastic tubing through semicircular grooves filed into the ice. After the ice nest was buried and had melted, the tubing was gently removed by pulling the free end at the surface. Following marking, ants and their original seed cache were poured into a 20 cm x 20 cm enclosure surrounding the new nest entrance. All colonies settled in readily and enclosures were removed within 48 hours of worker release. Colonies were re-visited twice monthly and the color of the wire and ink mark of each forager captured in a three hour period was recorded. The longevity of the fluorescent ink-mark was verified by noting its presence with the permanent wire-mark in subsequent months. Appearance of workers without a wire-mark indicated the movement of the first new cohort since marking into the role of forager. An additional five colonies, containing between 312 and 2,314 workers, were excavated between December and March of 2010. Each colony was censused and all adults were marked with 10% orange fluorescent printers ink in ether (Gan’s Ink, Supply Co., Los Angeles CA; Risk Reactor, Santa Ana, California). For each colony, a subsample of the lightest-colored (and presumably, the youngest) workers were marked with an additional coat of blue fluorescent printer’s ink in diethyl ether. Though naturally inactive during winter months, colonies were released on warm days with their seeds stores and successfully re-excavated nests in their original territory following mark application. Beginning in mid-April, samples of 20 foragers were taken approximately every 30 days and checked for mark presence and quality until no more marked ants appeared. Marks were scored visually by anesthetizing ants with ether and counting the number of ink blotches and dots per dorsum under a microscope outfitted with an ultraviolet light. Foragers were returned to their nests following mark assessment. The appearance of workers marked with blue ink indicated the

11 movement of the youngest cohort into the role of forager. Having calibrated the permanence of ink marks with the wire-mark study, these five ink-marked colonies were combined with the 6 wire-marked colonies for analysis of forager age.

Behavioral flexibility To determine if forager death drives the movement of workers from other behavioral castes into the role of forager, colony response to the experimental reduction in forager population size was measured in the field over 7 day increments throughout the annual cycle. Number foraging was estimated for 10 colonies using the Lincoln-index mark-release-recapture method. For half of the nests, 50% of the estimated forager population was removed. After one week, forager number was estimated again in both experimental and control colonies. Because this experiment was conducted at several different points in the annual cycle, each experimental nest was sampled concurrently with a single control colony. The change in forager population size for all colonies was determined by subtracting the initial forager number from the final forager number after seven days. The estimated proportion of the final population added per day was determined by dividing the number of new foragers by the number of days (7) from the initial mark in control colonies. This rate of forager addition was used to determine whether the entry of foragers in experimental colonies occurred at a higher daily average than control colonies.

Data analysis Data were analyzed using regression and T-tests in Statistic 7.0 or R (package GeoStats). Proportion data were arcsine-transformed to stabilize the variance.

2.3 Results

Location and catchability of foragers A full series of immature and mature colonies, ranging in size from 157 to 9,656 adults, were sampled throughout four annual cycles. In all colonies, foragers were found in the upper- most strata of the nest, showed fidelity to the role of forager on successive days and represented only a fraction of the total adult population in each colony. Sampled nests contained between 40 and 1,865 foragers, and up to 1,651 foragers were captured and marked for each colony. On day

12 one of each sampling event, as many as 93% and no less than 35% of a colony’s total estimated forager population was marked (one outlier at 23%). A mean of 44% (SD 15.5%) of foragers captured on the first day of each sampling event were captured foraging the following day, representing 59% (SD 16.5%) of the recapture sample (n = 48). Though foragers were afield from some nests during early-morning excavations, a mean of 83% (SD 16.7%) of marked foragers were recovered in their nests on the third day of each sampling event. Nests ranged from 40 cm to 270 cm in depth, with an average depth of 135.8 cm (SD 54.2). Foragers were not the only individual near the nest’s surface, and represented only 46% (SD 20.1%) of the ants found above 20 cm. For five nests sorted by chamber during excavation, marked foragers were found a maximum of 12 cm below the nest surface. In these and the remaining nests sorted in 20 cm increments (n = 42), 96% of marked foragers were recovered 0 to 20 cm below the nest surface; indicating that the forager population is a discrete group dis- tributed non-randomly within the nest and the larger adult population of the colony (Fig. 2.2).

Fig. 2.2 Foragers were always found in near-surface chambers. For 5 nests sorted by chamber, marked foragers were discovered at a maximum depth of 12cm (represented in red, on a nest 145 cm in depth). Pink shading indicates the maximum depth of 96% of marked foragers from 42 nests excavated in 20 cm increments (image modified from Tschinkel 2004).

Seasonal allocation to foraging The percent of each colony allocated to foraging was highly seasonal, and a distinct pattern of proportional labor allocation was conserved across years (Fig. 2.3). Foraging began in early March or April and increased to a maximum of 35 - 41% foraging mid-summer before declining to zero in December. In comparing across 4 years of this study, calendar date was a better predic- tor of percent foraging than the number of days from the start of foraging (cubic polynomial fit, 2 arcsine transformed adjusted R = 72.0%, F3,45 = 42.9, p<< 0.0001). In 2009, 2011 and 2012, foraging began within five days of March 1st, while in 2010 foraging did not commence until a full month later (Table 2.1). In 2010, larval production, mating flights and peak in proportion foraging were aligned with other years, but proportion foraging reached a higher maximum. This indicates that the survival and accumulation of old workers during the additional month of dormancy may have increased the number of available foragers during subsequent, active months.

Fig. 2.3 Foraging began in March or April of each focal year, reaching a proportional maximum between May and June before declining to 0%, prior to overwintering. Each point represents a single colony.

Although the general pattern of foraging was conserved regardless of colony size, immature colonies (< 800 workers) reached a higher maximum proportion foraging than large, mature colonies each year (mean residuals, mature= -0.019; SD 0.058; immature= 0.050; SD 0.092; t53= 3.31, P < 0.03; Figure 2.4) and a higher proportion of dark colored workers were foragers in immature colonies (mature mean = 0.35, immature mean = 0.60 T22= 4.05, P< 0.01). This differ-

14 ence likely resulted from investment in growth in lieu of reproduction by immature colonies in the early summer, so that a larger forager population and proportion foraging was present by mid-summer. These findings confirm that estimates of colony size based on forager population size should be avoided unless a patterned relationship between number of foragers and total number of adults is first identified across a full range of dates. For example, P. badius colonies composed of 10,000 workers may contain anywhere from 100 to 3,500 foragers during the active period, depending on the date.

Fig. 2.4 Foraging began in early spring and reached a proportional maximum in early summer, before declining to zero in late autumn. A higher proportion of adults foraged in immature colonies than mature colonies during mid-summer. The graph is based on 55 colonies of varying size, sampled from 2009 – 2012, with 95% confidence bands (cubic polynomial fit, Y= -0.5194 + 0.0109 x – 4.5466-5x2+5.2176-8x3, arcsine transformed adjusted R2 = 0.72, F3,45 = 42.9, p<0.0000).

Forager relationship with larvae Foragers are a colony’s only means of acquiring food for developing larvae. Despite the presence of cached seeds, percent larvae and percent foraging were positively correlated (r = 0.76), and changed in parallel from May until October of each annual cycle with a mean of 1.64 foragers per larva (SD 0.99; Figure 2.5), regardless of date and adult population size. Foragers preceded larvae by 30 to 40 days in all years and decreased in abundance more slowly than larvae in late autumn (Figure 2.6). The appearance of foragers more than a month before the year’s first larvae suggests that larval cues do not stimulate workers to forage. In 2011 and 2012, larvae continued to appear in nests well into autumn, while no pupae or new callows were discovered after October. In 2011, a record drought year, failure of pupation occurred a full month earlier than in the three surrounding years. The presence of these ‘doomed’ larvae in the midst of a healthy forager population and seed cache suggests that provisioning stops either due to environmental cues or lack of appropriate non-seed, food items before winter.

Fig. 2.5 The ratio of foragers to larvae (1.6 to 1) is not significantly related to date between May and mid-October of each year despite a dramatic change in the proportion of the colony allocated to foraging (gray, dotted curve and right Y-axis). Variability in the ratio of foragers to larvae results from low larval abundance just before and after winter dormancy (grey bands).

Fig. 2.6 The proportion foragers and larvae (as a fraction of all colony members) change in parallel, so that their ratio is constant from May to October. The years 2009 – 2010 differed from 2011 and are shown separately (lines are cubic polynomials).

Wire and mark results For workers from 11 wire and ink marked colonies, the age of first foraging ranged from 43 to more than 300 days. This dramatic difference in the chronological forager age was strongly related to birth month, and produced two distinct worker lifespans. The first cohorts of workers, which eclosed in June and July of each focal year, sclerotized and passed through other temporal castes quickly to appear as foragers at an average adult age of only 43 days (SD 1.78, n = 10 colonies, sampled at 15 day intervals). Conversely, autumn-born workers darkened and sclerotized over a period of months, overwintered and did not appear as foragers until 210 to 360 days of adult age (n= 9 colonies; Fig. 2.7). In addition to moving through behavioral roles at an advanced rate, adults that eclosed in early summer showed an increased rate of physical development relative to autumn-sisters. P. badius colonies do not overwinter with brood, as October of each year is the latest possible month of pupal eclosion (Tschinkel, 1998). When all years were pooled, sampled colonies contained an average of 4% callows (SD 4.64%) and 54% mid-colored (SD 18.9%) prior to overwintering (October and November, n=7). During the first two months of post-winter foraging (March), no workers were still in the callow-color phase, but 64% (SD 2.2%) were scored as mid-colored (n=5). Surprisingly, colonies excavated before the appearance of new callows in June, still contained more than 30% mid-colored workers, that could only have eclosed in No- vember or earlier the prior year (Fig. 2.8). This is an important observation because all foragers that eclose in June are fully darkened and sclerotized by the time of first foraging, which occurs only one month later, in July. The youngest workers in winter (December – February) appeared as foragers the following July, a full 8 months after the last observed eclosion date. The darkest-colored and oldest workers marked in colonies excavated during winter months began foraging in early spring. These slow developing individuals were proportionally dominant in the forager population until July and present in small proportions until September; eleven months after an October eclosion date. Workers that eclosed in early summer during the focal year became proportionally dominant in the forager population by late August. Individuals born the year prior were replaced by early September as successive cohorts graduated into the role of forager. Therefore, long-lived individuals born the year prior foraged alongside their short-lived, summer sisters, co-occupying the forager population following sexual alate production and release.

Fig. 2.7 At the time of excavation, workers of each cuticle-color category were marked with a different colored wire belt. Marked ants appeared as foragers in the order of their relative ages at the time of marking, as shown for five sample colonies marked throughout 2012 (labeled a-e with excavation dates). Workers that eclosed in June or July of the focal year began foraging at an average age of 43 days, while those eclosing in autumn were 210 or more days in age. Ants that eclosed from pupae after colonies were re-release (and were unmarked as foragers) are shown in blue.

Like their autumn-born sisters, workers that eclosed in June and July declined in number and disappeared from the forager population by September of the same year in all but one wire- marked colony (n=3). For colony #357 (marked on July 23rd, 2012), workers scored as “dark” and “mid-colored” during excavation still dominated the forager population by October and even appeared in low numbers the following April (Fig. 2.7). This was puzzling because the “dark” colored workers could only have entered adulthood in June of the focal year or in autumn of the previous year. Although this colony is an anomaly, it invites questions about natural and induced forager lifespan and whether local environment could influence longevity and replacement of foragers.

Fig. 2.8 The mean proportion of each colony represented by larvae, pupae, callows, middle or dark workers at four points during the annual cycle, averaged across years (means with standard error bars). The large cohort of middle-aged workers present in March, moves into the forager population in late spring and early summer to produce the highest annual proportion foraging

Forager population size and colony size Foraging began in March or April of each year. Overwintering-workers populated labor roles without replacement until June, when the first callows appeared. Therefore, as foragers were lost, colony size declined for four months following winter dormancy. Monthly estimates of forager population size revealed that the percent of each colony allocated to foraging increased in spring and early summer due in part to an increase in total forager number alongside a reduction in total colony size from forager death (Fig.2. 9). In mature colonies, a sharp decrease in number foraging followed in mid-summer and likely resulted from investment in sexual production earlier that same year. The production of sexual larvae in lieu of additional worker larvae in the spring could result in a demographic gap, reducing the number of workers of foraging age for a period the following month. This is a reasonable interpretation, because the year’s first cohort of workers enters the forager population at approximately 43 days of age (mid-July) and is reared concurrently with sexual brood.

Fig. 2.9 For each focal colony, forager number on each sample date was divided the May’s estimate to obtain a factor of change in forager number (means with standard error bars). In 2010, forager population size increased in immature colonies and decreased in mature colonies when the year’s first cohort of new workers entered the forager population following mating flights (July). In 2011, forager population size did not differ between May and August for mature or immature colonies.

In June of 2010, the bulk of colony growth took place from July to October following the period of stasis and loss of colony size. The decrease in percent foraging between July and No- vember resulted from an increase in colony size through new worker birth and ultimately, forager death without replacement prior to overwintering. Between May and October of 2010 colonies grew a mean of 4.64 (SD 1.93, n=5) times their original size. Surprisingly, the mean growth of colonies monitored between the same months in 2011 was 0.996 (SD 0.73 n=5), representing a maintenance of adult population size for that annual cycle (an extreme drought year; Fig. 2.9). In the same year, foraging also reached a lower proportional maximum, though it followed the same general pattern. Therefore, the decrease in percent foraging during autumn of 2011 was the result of maintenance of colony size and an ever-shrinking forager population size. The result for immature colonies would certainly be continued immaturity the following year, as total colony size did not reach the minimum 700-ant mark before winter.

Forager lifespan and death rate While the rate of entry into the forager population varied considerably across the annual cycle, marked groups of foragers sampled in 2011 were lost at a mean rate 3.35% (SD 1.32) per day. Foragers sampled between March and July of each year survived an average of 38 (SD 10.1) days in the forager population and represented individuals born the autumn prior. Con- versely, colonies sampled after July contained both short-lived summer workers and as well as autumn-born workers, which survived an average of only 26.9 (SD 7.93) days after marking (T- test, t12=2.28, P< 0.041). Without further subdivision of the mixed age caste that appears from July to September, it is impossible to know whether the slight reduction in forager longevity arises from a change in environmental conditions or a reduction in worker quality later in year. Because colonies did not grow in 2011, reduced forager survival may have had a measurable im- pact on larval survival. However, sampling across multiple years would be necessary to determine the typical range and forager longevity necessary for colony growth to occur.

Behavioral flexibility In experimental nests, forager number was estimated seven days after reducing the initial forager population by 50% .Colonies did not replace experimentally removed foragers by drawing from other castes and nearly half of the population was still absent at the end of the seven

22 day period (Mean proportion of original population absent = 0.48, SD 0.05) (Fig 2.10). The daily rate of addition to the forager population did not differ significantly between paired control and forager removal colonies, indicating that colonies did not draw workers from other castes to fill the induced labor gap (T-test, control mean = -0.010, experimental mean = 0.002, t6 = -1.42, P = 0.21). Seven days after forager removal, the ratio of foragers to larvae in experimental colonies also did not differ significantly from that of paired control colonies (Paired T-test, T3= 1.75, P = 0.18; mean difference between pairs = 0.24, SD 0.27). This indicates that half of the anticipated larval population was missing in nests where foragers were removed. These results demonstrate that 1) larvae depend on foragers for survival and 2) colonies do not replace workers in missing labor groups by drawing individuals from other behavioral castes. Therefore, P. badius colonies appear inflexible with respect to large-scale re-allocation of labor over a period of a week and the ‘foraging for work’ model can be rejected in this case (Franks and Tofts 1992).

Fig. 2.10 When half of the forager population was removed it was not replaced 7 days later (means with standard error bars, n=5 controls, 5 forager removal).

2.4 Discussion

Four important results emerge from this study. First, in P. badius, proportional allocation to foraging follows a seasonal schedule that is conserved across years. Second, there is a 5-fold difference in the age at first foraging, lifespan and rate of cuticle darkening between summer- born and autumn-born workers of identical body size. Third, when foragers are experimentally removed, colonies do not replace them by drawing workers from other castes, and larvae die in proportion to the number of foragers lost. Finally, seasonal allocation to foraging maximizes colony fitness and is correlated with colony age-structure. Fitness is maximized by colony age structure because the annual production of alate larvae, and the provisioning of adults during May and June coincides with the annual peak in forager and old worker number in mature colonies; generated by the absence of new worker births and forager deaths during the four months of winter dormancy. Because forager abundance is directly responsible for larval abundance, this seasonal age distribution aligns the greatest number of old workers (foragers) with alate production and provisioning, maximizing colony fitness. Although the same relationship could arise from behaviorally flexible workers responding to more numerous and nutritionally demanding larvae, this study demonstrated that allocation to foraging is not driven by larval cues. In all years of this study, foragers appeared more than a month before larvae and larval population size shrank in response to the removal of foragers. It is unclear whether the annual pattern of demography that aligns the maximum number of old-foragers and alates is a response to pressures that drive mating flights to occur in early summer, or an epiphenomenon that has shaped the temporal occurrence of mating flights. Though all mature colonies have a similar proportion of foragers at this time, larger colonies have a greater number foraging and therefore a higher reproductive potential (Mackay 1981; Tschinkel 1993; Cole and Wiernasz 2000). Investment in worker production from August to October increases colony size and forager population at the time of sexual production and during mating flights the following year. For immature colonies that do not produce alates in spring, springtime investment in worker larvae produces a relatively larger forager population later in the summer, which allows for colony growth through increased larval production. In either scenario, springtime production is owed in part to the sequestration of fat by overwintering, mid-aged workers, which

24 peak in abundance early-on during alate production (Tschinkel, 1999; Smith and Tschinkel, 2006). Behavioral plasticity and age to first foraging The dramatic difference in the age to first foraging, lifespan and development between summer-born and autumn-born P. badius workers is not unheard of among social insects. Sea- sonal development rates in larvae and pupae have been demonstrated to vary by orders of magnitude due both to exogenous factors (temperature, photoperiod) and apparent, endogenous factors (Kipyatkov & Lopatina 2003, 2009). In honeybees, behavioral plasticity is socially mediated, so that the removal of foragers stimulates precocious foraging in younger cohorts, and the presence of older bees inhibits movement into the forager caste; while seasonal development rate is modu- lated by the presence or absence of larvae (Huang and Robinson, 1996). For P. badius, proportion foraging decreased precipitously during summer months, while the number foraging (2010) changed little, suggesting that a stable caste size was maintained when the rapid development of summer-born workers facilitated replacement of lost autumn-born foragers. It appears that replacement is achieved by an adaptive annual colony cycle not a flexible response to worker demand. If caste membership were socially regulated in P. badius, then the decreased age to first foraging in summer-born workers could arise from an increased forager death rate or demand for food beginning in July. However, we found that death rate is not significantly higher in mid- summer and new foragers do not enter the population at an accelerated rate when 50% of the standing forager population is removed. Gentry (1974) also demonstrated that the prolonged removal of foragers from field nests did not result in a redistribution of labor, instead it caused colonies to become inactive as foraging ceased for a period of months. Though brood care is typically associated with very young workers, it is accomplished by P. badius workers no younger than 6 months of age following winter dormancy. The year’s first callow workers do not appear until late May or June, indicating that all alates and the year’s first cohorts of workers are reared by un-darkened, autumn-born workers. These workers come to dominate the forager population in July and still appear as foragers as late as September. As only a small difference in death rate exists between individuals once they begin foraging, autumn- born workers are not simply entering the forager population en masse and outliving summer- born sisters.

Instead, it appears that autumn-born workers remain in interior labor roles for an extended period of time, as summer-born sisters move through at an accelerated rate. The result is that behavioral castes are comprised of both veteran and transient individuals that develop on different schedules. By the time the final autumn-born workers are fully darkened and become foragers, they are met by hundreds or thousands of sisters born in the new calendar year. The occurrence of “mixed” behavioral castes invites questions about how experience and learning may influence task performance and the capacity of labor groups (Farris et al., 2001; Tripet and Nonacs, 2004; Ravary et al., 2007). More importantly, it indicates that response thresholds differ for workers in the same behavioral caste and are not likely to be driven by the absence of workers in a subsequent labor role.

Seasonal characteristics of foragers Although this study does not address the mechanisms underlying seasonal variability in age to foraging onset, several important correlates of colony investment could produce a caste structure that promotes colony growth, while inhibiting the replacement of foragers when loss exceeds a normal rate. The first factor is that all foragers, regardless of birth month, share a low body fat content. Tschinkel (1998) demonstrated that dark-colored (older) workers in P. badius colonies have the lowest relative fat content of all adults, averaging just 10% of their total body weight. Additionally, fat content in all ages of workers was shown to be lowest in July, following the production and release of sexual alates. If summer-born workers develop more quickly as adults to become foragers at 43 days, then it follows that their relative fat content is lower than that of autumn-born sisters of the same age. It has been suggested that physiological correlates of starvation may actually drive the rapid movement of summer-born workers into the role of forager (Toth and Robinson, 2003; Robinson, 2009). Likewise, In Temnothorax albipennis, worker fat content governs task allocation and though age and physiological condition may be correlated, task attendance can be induced by supplemental feeding or starvation of workers (Robinson et al., 2009). In the 48 colonies censused in this study, the ratio of foragers to larvae was conserved from May to October, indicating that there is no seasonal difference in the ability of foragers to provi- sion the colony. This relationship seems puzzling in light of the large seed caches housed in most colonies; but as Smith (2007) demonstrated, at the expense of larval production, colonies do not

26 tap into those stores when experimentally starved. It is possible that the age of foraging onset is driven by food quality or limited by the processing speed of seeds . Selective feeding of foods with a higher 15N signature occurs between alate and worker larvae in early summer (Smith and Suarez, 2010). Presumably insect protein is available throughout the annual cycle, but in the absence of sexual larvae late in the year, it is possible that autumn-born workers receive higher value food, which may contribute to their adult fat content, higher physiological “quality” and relatively longer lifespan. Colony growth and size depend in part on a queen’s oviposition rate and the continued survival of eggs into adulthood (Nakata, 1996). In Solenopsis invicta, egg volume and resultant worker size, which is correlated with longevity, are related to follicle residence time in queens (Tschinkel, 1988). The factors that stimulate egg laying in P. badius queens are still a mystery, but if forager death rate and larval hunger offer any cues, their reduction in autumn months could slow egg deposition rate, producing larger more well provisioned eggs and longer-lived adults, contributing to the observed patterns in worker lifespan and labor allocation.

Worker location within the nest P. badius foragers were found primarily in the top 12 cm of nests which can reach depths of 250 cm or more. The position of foragers near the nest entrance likely makes them available for recruitment by returning foragers while decreasing the probability that they interact with seed stores, brood or excavation tasks deeper in the nest. The spatial localization of foragers has been observed in other species, and implies that behavioral discretization may be influenced by access to task-specific nest space. Localization may also be adaptive because it diminishes the transmission of externally acquired pathogens (Wilson, 1980; Naug and Camazine, 2002). It also suggests that foragers deposit seeds in superficial chambers and do not shuttle them down to seed chambers, 40 or more centimeters below the surface themselves. In the present study of P. badius, the observed number of workers present in the top 20 cm of the nest far exceeded the estimated number of foragers in those strata, implying that additional castes reside there or are transient workers that transport seeds downward or move sand and trash upward from lower chambers. It follows that nest architecture may serve as an additional factor driving the discretization of tasks (Tschinkel, 2004) . In species that occupy cavities or shallow nests with a simple archi-

27 tecture, the rigid form of labor allocation demonstrated in P. badius, which produce nests in ex- cess of 2.5 meters, may not exist.

Death rate It has been suggested that environmental predictability is capable of driving specialization and discretization of castes and that colonies organized more heavily by “inflexible” processes, like absolute worker age, are likely to live in predictable environments (Seeley, 1989). If age to first foraging is not driven by a response to losses in the standing forager population, it is possible that entry into labor groups evolved to match pressure from sources of extrinsic mortality, which occur on a consistent, annual schedule. Colonies sampled in 2010 achieved forager replacement in summer, while many of those sampled during the extreme drought of 2011 failed to do so, resulting in an ever-decreasing forager population following mating flights. In both cases, the most important factor in determining colony growth was the ability of rapidly developing summer-born workers to, at minimum, replace lost autumn-born workers (present from March until September the year following their birth). This is so because forager number was shown to be directly related to larval abundance, with the experimental removal of foragers resulting in predictable larval mortality. If labor in P. badius colonies can be redistributed in response to need, one would expect forager replacement in 2011 to have increased to meet forager loss and prevent larval death; but it did not (Fig. 9). On warm mornings, P. badius foragers deploy on trunk trails and make multiple forays in pursuit of seeds, insects, fungus and ‘decorative’ charcoal. Foraging pauses only under excep- tionally hot, cloudless conditions or during heavy rain and concludes approximately 1.5 hours before sunset. Although the age at first foraging differs for summer-born and autumn-born workers, both die approximately one month after entering the forager population. This indicates that either foraging itself is a fatal role or death is programmed to occur (based on intrinsic factors) upon entry into the forager caste. Rueppell et al. (2007) demonstrated that the lifespan of foraging honeybees cannot be extended by a reduction in exposure, once individuals join the forager caste. The same may be true for P. badius workers, but an experimental reduction in extrinsic mortality factors would be necessary to find out. If death in P. badius is not programmed, it may be owed, but not limited to: heat and desiccation, policing during mating flights, predation by mound-visiting Apiomerus Reduviids and

28 most obviously, tussles with neighboring conspecifics (personal observations). It is not uncom- mon to see a forager with the decapitated head of a conspecific permanently clamped onto her petiole, as fatal encounters between neighboring nests occur frequently during the peak of foraging in spring and early summer. It is therefore possible that death rate is density dependent, so that populations in relatively empty ‘neighborhoods’ see a reduction in forager mortality, and an increase in forager lifespan.

Inflexible workers and allocation to labor Marked foragers were observed performing a variety of non-foraging tasks on the nest surface, such as trash and sand removal. Daily, weather and resource-related variation in foraging intensity were also observed, suggesting that “foragers” have a complex task repertoire and that task switching does occur locally within behavioral castes. Detailed observations of these fine- scale shifts in task attendance have been made for other seed harvesters, such as Aphaenogaster cockerelli, and Pogonomyrmex barbatus (Gordon, 1991; Sanders and Gordon, 2002; Schafer et al., 2006; Greene and Gordon, 2007; Pinter-Wollman et al., 2011). However, it is apparent that colony allocation to task-sets is limited by colony age structure and worker development rate in P. badius. This study shows that while workers may move fluidly between local tasks, individuals performing a task-set are drawn from a finite group, structured by worker development rate. Seasonal patterns in forager allocation unfold from an interaction between two seasonally distinct development rates, lifespan and cohort size, and individuals are not drawn from other castes once the forager population is exhausted. Seasonal differences in worker investment are probably common among ant species that produce alates from worker-fat following winter dormancy; and may be the basis for generating annual patterns of labor allocation. The occurrence of seasonally distinct development rates, which produce behavioral castes of mixed age, is probably not unique to P. badius and may have important implications for caste performance in ants. Future studies of labor allocation in the field will likely reveal that seasonal caste ratios, arising from predictable rates of birth, death and development, are elegantly intertwined with annual patterns of growth, reproduction and resource availability for many ant species.

CHAPTER THREE LONGEVITY AND SOCIAL INHIBITION OF FORAGING

3.1 Introduction

The life history of a social insect worker is unusual, because individual workers do not reproduce and are not exposed to extrinsic sources of mortality for the bulk of their adult lifespan. When aging workers transition from interior to exterior tasks, their risk of death from age- independent sources of mortality increases greatly. In theory, the risk imposed by these factors should be correlated with age-dependent mortality. However, developmental age is often decoupled from chronological age in social insects, and can be accelerated, reversed or arrested by labor saturation and whole colony nutritional status (Robinson et. al. 1994). Social transfer theory predicts that for group living organisms, selection on longevity should parallel the gradual depletion of personal resources, transferred to other group members over time (Lee, 2003; Munch and Amdam, 2010). The depletion of individual resources and developmental aging are thus an aspect of whole-colony reserves and whole colony demand (larval presence). Worker senescence may not be correlated with a particular chronological age as in traditional models of life history, but triggered by a threshold of depletion in body reserves, related to colony investment patterns and behavioral group membership (Amdam and Page 2005). To examine life history traits like senescence in social insect workers, both developmental and chronological age must be considered in the context of colony demand for labor. Likewise, the adaptive value of any worker life history trait is only realized at the colony level, through its bearing on colony reproduction. Each ant colony is a superorganism and the demography, development and regulating mechanisms are shaped by selection acting on whole colony characteristics (reviewed by Hölldobler and Wilson 2009). Therefore, mechanisms regulating worker longevity both emerge from and serve a larger function in the seasonal epic of reproduction for a colony. The case of the honeybee and the harvester ant Honeybee (Apis melifera) colonies are characterized by season-specific variation in worker development rate and the presence of discrete, age-correlated behavioral castes. Despite dramatic differences in the chronological age at first foraging, honeybee workers experience a func-

31 tional senescence within 18 days of entering the forager population, and their survival cannot be increased by the elimination of extrinsic sources of mortality (Rueppell, 2007; Dukas 2008). The scheduled transitions between honeybee castes are sensitive to social feedback, and both physiological and behavioral acceleration and reversion can be induced within 24 hours of manipulating colony demography (Huang and Robinson 1996). While the removal of foragers induces precocious foraging in young bees, the presence of a stable forager population inhibits new workers from transiting into the role of forager (Huang and Robinson 1996, Robinson et. al. 1994). Like honeybees, workers of the Florida harvester ant move through a sequence of labor roles as they age, culminating in foraging. Workers develop at two different rates and enter the forager population at approximately 40 or 200+ days of age depending on birth month. Both short-lived and long-lived workers die less than a month after entering the forager population, but what controls their age at first foraging and whether forager mortality results from age- dependent, or age-independent factors is unknown. The lack of influence of social control and P. badius forager membership was examined in a previous field experiment, which demonstrated that eliminating 50% of the forager population did not induce replacement or accelerate development in non-foragers after seven days (Kwapich and Tschinkel 2013). Across social insect species, responsiveness to forager mortality is particularly important in maintaining colony homeostasis. In light of the honeybee’s flexible strategy, the apparent inflex- ibility of P. badius colonies in redistributing labor is intriguing. If behavioral transitions are not socially regulated, then development and functional senescence may have evolved to match a predictable rate of forager loss in the habitat where these ants occur. If the rate of worker addition to the forager population is fixed in both directions, then experimentally increasing forager longevity should not influence the development of new foragers, despite the presence of an ex- cessively large forager population. This study addresses 1) how foraging and extrinsic sources of mortality influence forager longevity in both slow and fast-developing workers and 2), how forager longevity in turn influences young worker development and behavioral transitions in P. badius.

3.2 Methods

The influence of extrinsic mortality factors on forager longevity in P. badius were assessed by comparing the proportion of foragers surviving 20 days in control (unenclosed, n=14) and enclosed, foraging-range limited colonies ( enclosed, n=15). The effect of conspecific neighbors on forager survival was also assessed by removing all neighbors within a 30 meter radius of focal colonies (n=11) and monitoring focal forager survival relative to controls after 20 days (n=13). The influence of forager survival on forager population size was determined by comparing the change in forager number over 20 days and the age-structure of each colony’s forager population from May-August. Potential forager lifespan was also measured for short and fast developing forager groups in the laboratory under starved and fed conditions. Percent body fat was used to compare forager body condition and proximity to starvation in all experiments and treatments.

Study site and colony selection P. badius colonies were sampled from July of 2012 through November of 2013 in a 23-ha, sand hills habitat known as Ant Heaven, located in the Apalachicola National Forest, 16 km southwest of Tallahassee, FL (latitude 30.35, longitude −84.41). The site was characterized by an over story of 40-year-old long leaf pine (Pinus palustris), a midstory of turkey oak (Quercus laevis), and ground covering composed of dwarf huckleberry (Gaylussacia spp.), pricklypear (Opuntia spp.), beard grass (Andropogon spp.), gopher apple (Licania michauxii ), and catbrier (Smilax spp.). P. badius colonies occurred at an average density of one nest per 670 m2 alongside Solenopsis geminata, Trachymyrmex septentrionalis, Forelius pruinosus , Dorymyrmex bureni , Aphaenogaster floridana, and numerous less common ant species. Of these, S. geminata is the only species that also consistently forages for seeds in the area. In P. badius, colony size and maturity are correlated with nest disc area (Tschinkel 1999). In order to compare mature colonies of similar size, only colonies with nest discs between 25- 40cm in diameter were selected for use in this study.

Estimating forager population size Forager population size was estimated using the Lincoln index mark-recapture method (Lincoln 1930). Mark-release-recapture-release events took place over a two day period at both

33 the beginning of each sample period, and again after 20 days. Foragers were defined as individuals that collected seed or cookie-crumb bait at a distance of 150 cm or more from the margin of their nest mound before beginning a return trip to the nest. On day one of each sampling event, foragers were gathered in 15 minute intervals until fewer than five were captured in 30 minutes, then marked with two spritzes of 10% fluorescent printer’s ink in diethyl ether and released en mass onto their nest mound (Gan's Ink, Supply Co., Los Angeles, CA; Risk Reactor, Santa Ana, CA; method, Porter and Jorgensen 1980).The following day, foragers were collected in the same manner, counted, checked for a fluorescent mark under UV-light and released. The number marked on day one was then multiplied by the number captured on day two and divided by the number of marked individuals recovered in day two’s sample to yield an estimate of the number foraging (for more detailed methods, see Kwapich and Tschinkel 2013). Following recapture, all unmarked foragers in the second sample were also marked with the same color fluorescent ink, to maximize the proportion of the forager population that could be identified at the end of the 20 day trial. The mean proportion of the initial population thus marked per nest was 0.70 (SD 0.10). All foragers were then returned to their nest.

Experiment 1: Limiting foraging range Following the initial estimate of forager population size, colonies were divided into “reduced foraging range” and control treatments. Fifteen range-limited colonies were enclosed in 60cm X 60cm X 10cm screen-bottom, open-topped, aluminum boxes that prevented long- distance foraging; while 14 control colonies were divided between natural foraging and mock enclosures, which were similar to experimental enclosures but allowed foragers to exit through four, 20cm flaps cut into the sides (Fig. 3.1). At the center of each enclosure, the nest mound was exposed by cutting out a disc of screen flooring. Hanger wire was used to pin the screening to the ground and the screen floor was weighted with a layer of sand and local debris to prevent es- capes by tunneling. A strip of plastic mesh was affixed to the top of the enclosure to produce a band of shade that projected 5 cm inward from the margins of the enclosure and prevented the unnatural desiccation of foragers thwarted from departing on normal trails by the enclosure’s wall. A band of slippery fluon (liquid Teflon that prevents climbing) was applied to the inner and outer walls of each enclosure to prevent colony members from crawling out and other insects from crawling in.

Treatments To untangle the influence of foraging and feeding on body condition and survival, enclosed colonies were divided into fed (n=7) and starved (n=8) treatments. Fed colonies were offered chopped mealworms and native seeds daily, while starved colonies were denied access to new food for the entirety of the 20 day focal period. Although starved nests lost foraging access, all colonies had access to seeds stored in the nest at the time they were enclosed. Starved and fed colonies were matched with mock-enclosure controls (n= 6) and naturally foraging control colonies (n= 8) of similar nest-disc diameter and location in the forest and sampled simultaneously as replicate sets. To prevent changes in forager lifespan that could result from reducing interactions with conspecifics, control nests never had their neighbors or next-nearest neighbors within 60 meters enclosed.

Fig. 3.1Range-limited nests were enclosed in 60cm X 60cm X 10cm screen-bottom, open- topped, aluminum boxes that prevented long-distance foraging. A strip of plastic mesh was affixed the top of the enclosure to shade foragers that gathered near the enclosure’s edge.

Assessing survival and population change At the end of the 20 day focal period, enclosures were removed and colonies were allowed to forage naturally. Beginning the morning of enclosure removal, each colony’s new forager population size was estimated by mark-recapture. A new ink color was used, so that the proportion of the initial forager population surviving could also be estimated by recapturing and double-marking individuals from that population. Due to weather, some colonies were sampled after more than 20 days and their survival estimates were corrected assuming equal daily loss between marks one and two. In autumn, foraging activity was low after 20 days in 4 enclosed colonies, so the top 30cm of each nest was excavated (Foragers only reside in the top 12cm) and old-forager survival was estimated by counting the actual number of surviving, marked workers and using this number to estimate the total proportion of the initial forager population that survived. For 30 naturally foraging colonies (including controls) the proportion of foragers surviving was compared to the number of days between estimates, which ranged from 2 to 31 days after marking in 2012-2013 (n=30). The slope of the line identified the rate at which control foragers were lost per day during the course of the study, and provided an estimate of the number of days until complete forager turnover. This value was the same as the mean calculated by assuming an equal fraction died daily in each control nest. Therefore, the number of days until complete forager turnover in each experimental nest was also estimated assuming an equal fraction of the total workers lost per day, despite only having 1 and 20 day samples.

Change in population size For each colony sampled May - August, the factor by which forager population size changed over 20 day was determined by dividing the final number of foragers on day 20 by the initial forager population on day one. Additionally, an expected population size for experimental nests (E) was calculated by multiplying the mean proportion of workers replaced by new foragers (r) over 20 days in control colonies, by the initial forager population size (i) for each experimental nest. The number of initial workers that would have been replaced under natural conditions was then added to the number of old workers that actually survived (s) 20 days in each experimental nest. This value was divided by the complete, initial forager population size to estimate the factor by which forager population size would change over the month, if new worker addition occurred at the control rate and old worker survival occurred at the enclosed-colony rate

(E = ((r x i) + s)/ i). The observed and expected factors for forager population size were then compared to 1 (replacement) and the mean control value with single mean T-tests. Estimates were calculated for spring and summer colonies only and one replicate-set was removed as an outlier (n =17).

Body condition in field foragers To analyze the influence of forager age, tenure, and feeding on body condition, 8 to 15 old foragers (bearing the original ink mark) and 15 new foragers (bearing only the second mark or no mark at all) were retained for fat extraction following the 20 day mark-recapture event Estimates of “new forager” body condition are conservative because never was 100% of the initial forager population marked, and a small portion of individuals labeled as “new foragers” likely represented old foragers that were not previously captured. In many control colonies, few individuals marked on day 1 of the study still survived to be captured after 20 days. In these instances, the top 20cm (where foragers reside) of each nest was exposed to search for marked individuals among uncaptured foragers. To determine percent body fat, individuals were killed by freezing, then dried at 55 o C in an oven for 48 hours, weighed on a microbalance, placed in individually labeled gelatin capsules, inserted into a Soxhlet extractor and washed repeatedly with ether (per Smith and Tschinkel 2009). The dry, fat free weight of each ant was then measured and divided by initial dry weight to determine percent body fat (energetic reserves) at the time of collection.

Lab survival and body condition To reveal the potential lifespan of workers, wild foragers from 4 colonies were captured and monitored under optimal or complete starvation conditions in the laboratory. To account for differences in longevity that could result from the 5-fold difference in age at first foraging between autumn-born and summer-born workers, 220 foragers were collected from each of 3 nests in both May and September of the same year, and from the fourth nest in September only. For- agers captured in May eclosed the previous autumn and were a minimum of 240 days old (long- lived workers). Foragers captured in September eclosed in July/August the same year and were 40-50 days old (short-lived workers) (Kwapich and Tschinkel 2013). To serve as a baseline for body condition, twenty foragers from each colony were killed by freezing on their date of collec-

37 tion and stored for later fat extraction. Unfortunately, most of these initial samples were lost in an oven fire, so season-specific means for other wild, control colonies were substituted. The remaining foragers were taken to the lab and divided into two treatments groups: fed and starved (n= 1400 foragers). Each group of 100 foragers was housed at 26 o C in a plaster laboratory nest and separate tray. Foragers belonging to fragments in the starved treatment were offered cotton-plugged test tubes containing water only, while those in the fed treatment were offered water, mealworm pieces, 20% sucrose water, and ad lib native seeds (gathered from field nests). For all treatments, the number of surviving foragers was recorded approximately weekly (initially), then less frequently up to 39 weeks. To determine body condition at the time of death from starvation, 10-15 ants were collected after their death, from each starved colony in each season and stored at -40 degrees C prior to fat extraction (sample n=80 starved workers).

Experiment two: Effect of neighbor removal For 11 focal colonies, the effect of conspecific neighbors on forager survival was assessed by removing each neighboring colony within a 30 meter radius and monitoring focal forager survival over 20 days. Each focal colony was paired and sampled simultaneously with at least one control colony (n=13) of the same disc diameter. In order to detect the potential seasonality of neighbor interactions, sampling took place from May-September in both 2013 and 2014. The forager population size of each neighboring colony was also estimated by mark recapture before neighbors were removed (temporarily enclosed for the duration of the experiment). This allowed a comparison of focal forager survival and the number of neighboring foragers removed for the local foraging area. Three neighbor-removal colonies were excavated and censused 23-25 days following the removal of conspecific neighbors. Three control colonies were also excavated simultaneously and the ratio of larvae, pupae and callows to foragers was compared. Excavations took place between June 4th and 6th, 2014, following the emergence of the year’s first new workers. While these data are a preliminary portion of a larger data set, they are presented here to aid in assessment of risk to foragers from extrinsic sources of mortality. Analysis Seasonal, colony and treatment related differences in forager survival were assessed using Generalized Linear Models with a quasibinomial error distribution to account for overdispersion

(proportion data, R v. 2.15.2, package, lmer). In the simplest model for experiment 1, control types (natural foraging vs. mock enclosure) were combined and replicate sets were merged in three separate seasons due to similarity. Season and treatment related effects on percent body fat were analyzed using beta regression (proportion data, R v. 2.15.2, package, betareg). For foragers retained in the laboratory, the influence of birth month, starvation and feeding on weekly survival were described by a Kaplan–Meier survival function. Differences in the weekly risk of death were compared using the Cox proportional hazards model to account for right censored data (observation stopped at 39 weeks), and the continuous change in the number of nestmates already dead per 100 ant group, as a covariate (Statistica v.12, StatSoft 2013). The model assumed that ants in the fed treatment represented the baseline hazard function, while ants in the starved treatment had a hazard proportional to that baseline.

3.3 Results

Forager survival Effect of foraging: The proportion of foragers surviving in range-limited colonies (55%± 4.4, n=14) was significantly higher than in controls (27% ± 2.6, n=12); demonstrating that longevity is not fixed when individuals enter the forager population, but reduced by foraging itself (quasibinomial GLM, t= 4.08, p<0.0004, Table 3.1). This difference likely resulted from both a reduction in exposure to extrinsic sources of mortality and a reduction in activity in enclosed colonies. Effect of starvation: When averaged across all months, forager survival was a significant, 28 % higher in both fed- enclosed and starved-enclosed colonies than controls; with a mean 55% ± 3.5 and 55%± 7.9 foragers surviving respectively (Fed t= 4.0, p<0.0006; starved t= 3.05, p<0.005, Table 3.1). Despite the apparent similarity between feeding treatments, important differences within treatments occurred seasonally. Effect of season: For all colonies, survival was lowest in May and June compared to other months (Table 3.2; Fig.3.2). May-June and July foragers were chronologically the oldest workers in the study, having eclosed the previous autumn (240+ days prior) and overwintered as adults (Kwapich and Tschinkel 2013). These individuals reared both the annual pulse of alates and the first new cohort of workers from stored reserves accumulated the previous year (Tschinkel 1998).

While survival in fed-enclosed colonies was uniform after spring, survival in starved- enclosed colonies improved by 51% from May to the final sample date in October. In Sept- Oc- tober, starved-enclosed forager survival reached its maximum at 82%, which was both the highest among all treatments and seasons of the study, and 50% higher than control survival in the same season. Forager survival in control colonies increased a month later than in starved colonies, from 30% in Sept-Oct to 68% November. The increase in autumn survival for both starved-enclosed and control colonies was related to a reduction in the number of days colonies opened their nests during the 20 day focal period (observation). Why foraging naturally stops in winter, despite the presence of foragers and suitable weather, remains a mystery. A lack of successful foraging trips may be one factor that induces winter dormancy in colonies, and could help explain the inability of larvae in typical nests to reach pupation in October (described in our previous study, Kwapich and Tschinkel 2013).

Fig. 3.2 Forager survival over 20 days was significantly higher in range-limited colonies than in control colonies during all seasons (quasibinomial GLM, enclosed colonies t=4.08, p<0.0004), but differed among seasons (asterisks denote significance relative to controls within seasons, quasibinomial GLM, May-June t= -3.05, p<0.005). Survival increased for starved colonies from May –October, and increased for control colonies in November, prior to dormancy.

Table 3.1 Results for quasibinomial Generalized Linear Models for the effect of range-limitation on forager survival in wild colonies Estimate Std. t P Mean. % Error value surviving std. error

Effect of range-limitation on survival (df= 26) 1.72e- Intercept -1.13 0.21 -5.30 05*** 55% ± 4.4 Enclosed (all) 1.08 0.26 4.08 0.0004***

Effect of treatment on survival (df = 26)

Intercept -1.13 0.21 -5.28 2.05e-05***

55% ± 3.5 Enclosed-fed 1..24 0.31 4.00 0.0006***

55%± 7.9 Enclosed-starved 0.93 0.30 3.05 0.005**

Table 3.2 Results for quasibinomial Generalized Linear Mod- els of the effect of season on survival Estimate Std. t P

Error value

Effect of season on survival (df = 26)

Intercept 0.013 0.25 0.054 0.96

May-June 0.005** (vs. Sept- Oct) -0.96 0.31 -3.05

July – Aug 0.79 (vs. Sept- Oct) 0.10 0.38 0.27

Death rate, forager turnover & chronological age Between May and October, control colonies lost 3.6% of their initial forager population per day (Fig. 3.3). Complete turnover of the initial forager population took an estimated 27± 1.39 days in controls, and significantly longer in both enclosed-fed colonies (44 ± 3.71), and enclosed-starved colonies (41± 6.24) during the same period. This 57% increase in forager tenure likely allowed the same group of foragers to forage for multiple cohorts of larvae (Fig. 3.4). In autumn, slower forager turnover was related to a reduction in the number of days colonies opened their nests to forage (observation). While fed-enclosed forager turnover did not differ from previous month, foragers from starved-enclosed and control colonies were estimated to turnover at 112±6.74 and 65±14.7 days respectively. Survival time as a forager was not dependent on the chronological age of workers entering the forager population. Foragers sampled from May-August ranged from 43 (summer-born) to more than 270 days (born the autumn prior) of age at the time of collection. If each were to live an additional 27 days as a forager (control rate), foraging would represent 39% and 9% of each lifespan, respectively.

Fig. 3.3 Proportion of foragers surviving in control colonies was estimated by mark recapture at varying intervals after the initial marking event. Marked foragers were lost at approximately 3.6% per day (n=30), and complete forager turnover occurred at approximately 27 days.

Fig 3.4 Between May and August, complete turnover of the initial forager population occurred at a mean of 27 day in control colonies, 44 days for fed-enclosed colonies and 41 days for starved-enclosed colonies. Turnover increased significantly in Sept - Oct. for starved nests (to 112 days) and in November for control nests (to 65 days).

Seasonal nature of forager replacement For control colonies, death rate was relatively consistent from May to October, with 62% to 81% of marked foragers lost by day 20. From May-August, these losses were matched or exceeded by the addition of new workers, and forager populations either grew or maintained their size (Fig. 3.5). Yet, in early autumn, while death rate remained constant, new worker addition slowed and less than 40% of foragers were replaced. By November, death rate decreased by half and replacement of dead workers continued to fall to approximately 20%, causing the forager population to decline to zero before overwintering. The disparity between death -rate and replacement in Sept.-Oct. confirms the experimental finding that entry rate into the forager population is intrinsically-metered, and not induced by losses in the forager population itself (Kwapich and Tschinkel 2013). In Autumn, the thousands of adult workers present in nests are not destined to forage until the following spring, and remain both physically and developmentally immature for months to come. Thus, failure to replace

43 foragers in the autumn results from worker developmental status rather than a lack of adult workers.

Fig 3.5 From May-August the loss of foragers was matched or exceeded by the addition of new workers (replacement), and forager populations either grew or maintained their size. However, while forager death rate remained constant in early autumn, replacement dropped, due to a demographic change created by newly-eclosed, slow-developing workers. By November death rate also decreased (likely due to a reduction in activity) and replacement of dead workers fell to approximately 20%.

Maintenance of forager population size For colonies sampled between May and August, there was no significant change in forager number between each initial and 20 day sample (reference constant =1, sample mean= 0.97, t= - 0.78, df = 16, p= 0.45, Fig. 3.6). Despite differences in forager survival, both enclosed and control colonies retained their initial forager population size over 20 days (control mean= 0.95, t- test, fed mean= 0.97, t= -0.22, df = 10, P=0.83, starved mean = 1.01, t=-0.64, df=10, p=0.54). In contrast, the calculated, expected population size for enclosed nests was significantly higher than the observed value (fed mean=1.21, t=-2.41, df= 8, p=0.04; starved mean= 1.19, t=-2.25, df= 8, p=0.05, Fig. 3.6). Thus, control colonies not only lost relatively more workers, but they added more new foragers than range-limited colonies to replace them.

Fig. 3.6 For each colony, the observed forager population size on day 20 was divided by forager population size from day one to obtain a factor of change in size (May – Aug.). En- closed colony means (bars are standard error) were not significantly different from controls (t= -0.22, p=0.83) and not significantly different than 1 (t= 0.78, p=0.45), indicating that on average, lost foragers were replaced in all treatments. The expected population size for experimental nests, calculated using the mean proportion replaced in control colonies, was significantly higher (single mean t-test fed, t = 2.21, p = 0.04; starved t=2.25, p =0.05) than the observed value, demonstrating the higher relative rate of new worker addition in control colonies.

Age-structure of forager population While fed-enclosed, starved-enclosed and control colonies all maintained their initial forager population size across the 20 day focal period, the age composition of the forager population differed markedly between controls and enclosed colonies (Fig 3.7). At 20 days, 69% ±4.2 of foragers were new in control colonies, while only 43%±8.7 and 46% ±5.9 were new in fed- enclosed and starved-enclosed nests. The difference in age structure can be attributed to relative difference in survival and rate of replacement in controls and enclosed nests.

Inhibition of Foraging Two interpretations could explain how both control and range-limited colonies maintained a constant forager caste size despite dramatic differences in forager loss over the focal period (Fig. 3.2). Either forager death stimulated forager replacement by younger workers or forager survival inhibited the entry of younger workers into the population. Previous work (Kwapich and Tschinkel 2013) demonstrated that when 50% of foragers were removed, colonies did not draw workers from other labor groups to replace them even after 7 days. Additionally, a 5-fold seasonal difference in adult development rate proceeded independently of forager loss rate, which was constant. Taken together; these results suggest that entry of new recruits into the forager population was inhibited by the presence of surviving foragers in enclosed nests. Contrary to our predictions, significant increases in forager survival did not produce increasingly large forager castes, instead forager population was regulated from the top down. Thus, there is a unidirectional, social control of forager population size in P. badius; maintained by increases in forager longevity but not increases in forager loss.

Fig. 3.7 In spring/summer control colonies, 69% of the forager population was composed of new workers on day 20, while enclosed-fed and enclosed-starved colonies average 43% and 46% respectively.

Laboratory lifespan and starvation Risk of mortality for foragers in starved laboratory groups was 5.6 times greater than for those in the fed groups (hazard ratio= 5.6, P < 0.0000 ). When fed, 10% of foragers still survived at 180 days and complete turnover did not occur until more than 250 days in several groups. In contrast, only 10% of laboratory-starved foragers still remained at 35 days, but a few were able to survive to 50 days on body reserves alone. Foragers that died of starvation in the laboratory had a mean fat content of 3.4% ±0.24 (median = 3.23%, n= 80, Fig. 3.8). Despite a 5-fold difference in chronological age when collected, lifespan was not significantly different for summer- born and autumn-born foragers in either laboratory treatment (hazard ratio=0.93, P = 0.18).

Figure 3.8 Kaplan-meier estimate of survival probability at each week for treatments and months (n= 1,400 foragers from 3 to 4 colonies per season, divided between starved and fed treatments). Starved foragers had a mortality rate that was more than 5 times higher than fed foragers (hazard ratio= 5.6, P < 0.0000), and there was no difference in the risk of mortality within treatments, between seasons (hazard ratio=0.93, P = 0.18). The red line represents the mean observed, forager lifespan from 30 wild, control colonies.

Wild forager fat content by treatment Effect of season: Season, forager tenure and treatment all influenced the body condition of wild foragers. Tschinkel (1998) demonstrated that percent fat increases across all age-classes and castes as winter approaches in P. badius colonies. Seasonal comparisons between control colonies confirmed that while young forager fat content was not significantly different than old forager fat content within seasons (prsq=0.34 estimate = 0.34, SE =0.19 , df = 3, z = 1.80, p=0.08 ), autumn foragers had slightly more body reserves than spring/summer foragers (prsq= 0.34, estimate = 0.48, SE =0.21, df = 3, z = 0.02, p= 0.02 ). Percent fat also did not differ between controls and enclosed colonies in autumn (Fed = prsq= 0.17, estimate = 0.31, SE =0.21, df = 5, z = 1.47, p=0.142, starved= prsq= 0.57, estimate = -0.01, SE =0.24, df = 5, z = -0.05, p=0.96). Effect of time foraging: Within individual control colonies, comparisons between young and old foragers revealed that body fat declined by 4% ± over 20 days (May-August colonies). If foragers were to lose the same amount every 20 days, it would take them approximately 40 days to die from starvation. Thus, the lifespan predicted by a gradual depletion of energetic reserves exceeds the maximum (27 day) realized lifespan of foragers in that season. Similarly, the mean percent body fat in old foragers from wild control nests was 2 to 6 times that of workers that died of starvation in the laboratory. Both the rate of decline in fat and difference from starvation suggest that the realized lifespan of natural foragers is shorter than their potential lifespan based on energetic reserves alone. However, it should be noted that laboratory foragers are likely to be less physically active than field foragers. Young foragers from both enclosed treatments had significantly more body fat than young foragers in controls in May-August (Fed: prsq= 0.57, estimate = 0.91, SE =0.27, df = 4, z = 3.34, p<0.0001, starved: prsq= 0.57, estimate = 0.65, SE =0.28, df = 4, z = 2.34, p<0.019). Despite a considerable difference in the proportion surviving 20 days, old foragers in starved-enclosed colonies had as much body fat as young workers in control colonies, suggesting that body reserves are not depleted unless foraging activity occurs. Although access to food was reduced, young foragers from starved-enclosed colonies had even more fat than their older sisters; perhaps due to inactivity (Fig. 3.9). Effect of feeding: Although a forager’s relative body fat content may be the lowest among colony members in any season, foraging behavior was not correlated with a specific percentage of body fat. Forager fat reached 36% of total body weight for some workers in fed-enclosed col-

48 onies. This value is well within the range recorded for mid-aged workers deeper in the nest, by Tschinkel (1998). Most notably, old foragers from fed-enclosed colonies had significantly more body fat than old and young foragers from wild control colonies in spring/summer (prsq=0.63, estimate = 0.72, SE =0.23, df =3, z =3.12, p<0.002); indicating that body condition was reversed and that foragers forming the initial population fed opportunistically. When body fat content was heterogeneous among foragers of different ages in fed-enclosed colonies, behavioral/labor role reversion did not occur and workers remained foragers. Likewise, starving and restricting foraging in colonies also did not influence forager membership despite improving body condition. Both results suggest that membership in the forager population is not dependent on a minimum body condition.

Fig. 3.9 Percent body fat was greater in Sept.-Oct. than in May-Aug for control colonies. In May-Aug, young foragers from enclosed colonies had significantly more fat than controls (beta regression, Fed z = 3.34, p<0.0001, starved z = 2.34, p<0.019). Old, fed-enclosed foragers from May-August also had more body fat than young workers from controls, demonstrating that they improved their body condition during the focal period, but did not revert to previous labor roles.

Neighbors as a source of extrinsic mortality The influence of neighbors was seasonal in P. badius, and accounted for nearly 30% of old, autumn-born forager mortality during the spring-time. In paired control colonies (2013-14), a

49 mean 13% ±2.8 (n=7) of foragers survived 20 days, while 43% ±4.7 (n=5) survived when neighbors were removed. This difference suggests that neighbors are large source of forager mortality early in the year. Data collection and analysis of neighbor interactions in other seasons is ongo- ing, but these preliminary data are presented here to aid in a discussion of extrinsic and age- dependent mortality. Following the 20 day neighbor-exclusion period, three focal nests were excavated and censused (n=3). Three focal, control colonies were also excavated simultaneously. Excavations took place following the predicted emergence of the first callow workers of the year on June 4-6th and revealed that all colonies contained larvae, pupae and very young callow workers. Compared to control colonies, the ratio of brood and callows to foragers was higher in 2 out of 3 neighbor- removed nests, which may indicate that increased forger survival in the absence of neighbors also increased per-forager productivity (Fig. 3.10).

Figure 3.10 Neighbor removal improved forager survival for colonies sampled in May-June (+30%) and Sept. – October, but not for colonies sampled in July-August (left panel). Three colonies were excavated and censused 23-25 days following the removal of conspecific neighbors. Three controls colonies were collected simultaneously. Compared to control colonies, the ratio of brood and callows to foragers was higher in 2 out of 3 neighbor- removed nests, which may indicate that increased forger survival in the absence of neighbors also increased per-forager productivity (right panel).

3.4 Discussion

Several important results emerge from this study of longevity in P. badius. First, the potential lifespan of foragers was shortened by foraging itself, as evidenced by a 57% increase in forager longevity relative to controls in response to reduced foraging range. Second, neighbors were one source of extrinsic mortality and directly or indirectly accounted for a mean of 30% of forager mortality each spring. Third, foragers of vastly different chronological ages had the same realized lifespan in the field, and the same potential lifespan under starved and fed conditions in the laboratory. Fourth, body condition was decoupled from task allocation and foragers did not return to interior labor roles despite regaining body reserves equivalent to those of mid-aged workers (Tschinkel 1998). Finally, increased longevity of foragers inhibited the scheduled movement of new workers into the forager population in wild colonies. Under natural conditions, opportunistic increases in forager longevity are likely to translate into colony growth by allowing the same group of foragers to feed multiple cohorts of worker larvae or promote the survival of more larvae, and by pre- venting the exposure of new workers to external sources of mortality. Likewise, the inability of workers to accelerate development when foragers are lost, may allow colonies to avoid worker depletion during longer stretches of unfavorable conditions (Gentry 1974, Kwapich and Tschinkel 2013). Among the many paradigms of division of labor for social insects, P. badius labor dynamics conform best to response threshold and social inhibition models designed to describe honeybee labor dynamics. Like honeybees, the behavioral response thresholds of harvester ant workers change at an intrinsic rate with increasing age, but this rate can be mediated by worker interactions, from the top down (Beshers et. al. 2001, Beshers and Fewell 2001). In honeybees, the suppression of new foragers is controlled directly by existing foragers, which distribute an inhibitor, ethyl oleate, to hive bees via trophallaxis (Leoncini et al., 2004). Trophallaxis does not occur between harvester ant adults, suggesting that the inhibitory mechanism may be tactile, relating to crowding or direct contact, or even pheromonal in nature. P. badius differs in another important way, in that the control of labor allocation is unidirectional and behavioral development is not accelerated when foragers are removed.

A disposable caste? Among the Pognomyrmex species that have been sampled to date, a realized forager lifespan of approximately one month is the rule. Gordon and Hölldobler (1987) described the disappearance of marked, exterior workers by 30 days in P. rugosus and P. barbatus, while Oet- tler and Johnson (2008) demonstrated that turnover in marked foragers and recruits, occurred at approximately 18 and 35 days, respectively. Forager lifespan in Pognomyrmex owyheei was as brief as 14 days in the field and approximated the 18 day survival of foragers starved in the laboratory (Porter and Jorgenson 1981). In light of the high risk of externally induced mortality that these foragers face (largely from vertebrates), the necessity of collecting high-value seeds, and the proportionally small number of foragers (10% when sampled), the authors suggest that the relatively poor body reserves of foragers single them out as a disposable caste. The perception of foragers in social insects as physiologically depleted and disposable is further supported by the relative differences in the lipid reserves of ‘corpulent’ nurse workers and lean foragers for numerous species of ants, wasps and bees (Tschinkel 1987, Blanchard et. al. 2000). Indeed, social transfer theory predicts that colonies should invest very little in individuals that will be subjected to intense extrinsic sources of mortality (Lee 2003). Given the consistently short lifespan and lean constitution of other social insect foragers, the question that remains is whether increased longevity in range-limited P. badius foragers arose from reduced exposure to extrinsic sources of mortality or reduced depletion of body reserves from a restricted foraging range. Although the estimated maximum lifespan observed in control colonies was 27 days, twenty day old foragers were scarcely closer to starvation than new recruits in the forager population. At 20 days, those that survived maintained body fat more than twice that of workers that died from starvation in the laboratory. Therefore, there is no evidence that field foragers would de- plete more than 50% of their reserves in 7 days, after having maintained them for at least 20. Based on the starvation criterion, mortality occurs earlier than predicted by the gradual depletion of body reserves alone. Yet, depletion of body reserves during forager tenure may not be particularly important in predicting age-dependent mortality initiated by dwindling ‘social transfers’. In honeybees, the number of days spent foraging does not correspond to a detectable decrease in the initial lipid stores after the onset of foraging. Toth and Robinson (2005) demonstrated that access to carbo-

52 hydrates was a more important nutritional factor affecting the ability of bees to sustain powered flight. Although harvester ants do not store sugary products in their nests or engage in adult trophallaxis, it is possible that they too rely on carbohydrate food source to supplement energetic expenditure when afield, or even another form of metabolic reserves (Tschinkel 1998). What is clear, is that P. badius foragers will feed, and are capable of restoring body reserves identical to those of younger workers in the top third of the nest (Tschinkel 1998). Though, under natural conditions, colony nutrition and subsequent opportunities for foragers feeding may be minimal. When foragers are removed, larvae die despite the presence of thousands of stored seeds and other adult colony members (Kwapich and Tschinkel, 2013). This evidence suggests that colonies are limited by insect protein in particular, or some other factor that foragers collect but cannot store.

Sources of extrinsic mortality at home and afield By removing neighboring colonies, this study demonstrated that interactions with conspecific neighbors influence the labor thresholds of individual workers, and the demographic structure of whole colonies. The influence of neighbors was seasonal in P. badius, and accounted for 30% of forager mortality during the spring-time production of alates and workers by old, autumn-born foragers. Among ants, spring-time conflicts with conspecifics are common, and often, very visible (McCook 1904). Forager population size in P. badius nests is highest in spring (Kwapich and Tschinkel 2013), suggesting that the increased density of foragers on the ground may increase the encounter-rate of neighboring foragers, or even reduce the availability of acces- sible seeds and insects. Extrinsic sources of forager mortality were demonstrated to include conspecifics from neighboring nests, but could also include extreme ground temperatures, desiccation, navigational errors and exposure to disease and predators, to name a few. Restricting foraging range protected foragers from some of these sources of mortality, which in turn significantly increased their longevity. Yet, while enclosing nests increased survival and body condition for a large fraction of foragers, many enclosed foragers still died during the focal period. It is possible that enclosures themselves introduced a source of mortality for workers. Many enclosed foragers still attempted to forager despite meeting a wall in every direction. Though the margins of the box were shaded, exhaustion or desiccation may have resulted. Fur-

53 thermore, various insects and arachnids invaded the boxes from above and below. Each invader was removed, but in the end, enclosures could not protect foragers from most threats localized to the mound. When starved foragers stopped opening their nest entrances in October, survival sky- rocketed to 82%; suggesting that stepping onto the surface alone may pose a significant risk to workers.

Potential and realized lifespan If longevity is typically programmed to match extrinsic mortality, why P. badius workers enter a fatal role so early in their potential lifespan remains a mystery. As the leanest colony members, foragers are the most expendable relative to other adults. However, they carry enough reserves with them to double their realized lifespans and are capable of surviving more than 250 days in the laboratory when fed. Several possibilities exist to explain the disparity. 1), investing resources in workers that can survive longer than necessary may not pose a significant cost to the colony, 2) A reduction of function may occur at body reserve-levels above starvation. 3) Body condition and potential lifespan could simply be emergent products of colony nutrition and the equal dispersion of resources. In a study of P. badius colony nutrition, Smith and Suarez (2010) noted that the Ant Heaven population (used in our study later), workers fed from a higher trophic level than neighboring populations and likely consumed relatively more insect protein. The authors also observed that the burden or benefit of experimental feeding and starvation is shared by all colony members, though old workers remained the leanest across age classes. 4) Lifespan may be the product of a mean response to an environment that is unpredictable or conversely, 5) a slow evolutionary response to increasingly powerful extrinsic factors. The density of colonies in Ant Heaven is particularly high relative to populations found in neighboring flatwoods sites (King and Tschinkel 2008). Indeed, conspecific neighbors and perhaps competition for insect protein may prove to be the greatest source of extrinsic mortality within the population. Finally, this study suggests that 6), foragers that have enough body reserves to live longer, may benefit colonies by promoting colony growth through the suppression of new foragers when opportunistically freed from extrinsic factors. The most parsimonious explanation for the difference in realized lifespan and colony investment may be a combination of several of the

54 aforementioned possibilities. Further examination of age-dependent sources of mortality is necessary.

Foraging in the Florida harvester ant Previous work has demonstrated that aging P. badius workers progress through a sequence of labor roles before leaving the nest to forage. The forager population is a discrete, age- correlated labor group that resides in the top 12cm of each nest, but represents only a fraction of the ants that occur at that depth. Foragers maintain a consistent ratio with the colony’s larval population from May to September, though forager allocation is not a response to larval presence. Instead, proportional allocation to foraging follows an annual pattern, shaped by the interaction of seasonal phases of colony growth and reproduction, with a five-fold difference in the age of summer and autumn-born workers entering the forager population, (43 vs. 200+ days). This study reveals that forager survival time is consistent from May-September but can be increased with a reduction in foraging or extrinsic sources of mortality (including neighbors of the same species). Increased forager survival results in the inhibition of new forager development, while increases in forager mortality do not stimulate accelerated development. Instead, workers age into the role of forager at a rate selected to, at a minimum, allow forager replacement. These findings suggest that opportunistic increases in forager longevity can promote colony growth by slowing the scheduled, behavioral transitions of younger workers, adding an additional layer of detail to the annual cycle labor allocation in the Florida harvester ant.

CHAPTER FOUR CONCLUSION

2.1 The value of field studies

Like long-lived unitary organisms, social insect societies adhere to annual and ontogenet- ic schedules. Personal resources are divided to promote growth, through worker production and expansion of territory, and reproduction through alate production and synchronization with conspecifics (Tschinkel 2011). While labor groups must change in size to meet changes in colony demand, this dissertation demonstrates that in the absence of flexible workers, the value of demographic labor groups can only interpreted in the context of the environment that shaped them. Most studies of labor groups endeavor to describe their adaptive function, but fail to consider the dynamic nature of both seasonal cycles and colony ontogeny. By connecting seasonality, neighborhood dynamics, labor allocation, worker longevity, demography and physiology; this dissertation focuses on the organism-like nature of whole, social insect societies in the environment that shaped them. In the absence of seasons, extrinsic sources of mortality and nest architecture, the adaptive value of colony age-structure is lost. Yet, studies of age-correlated division of labor in soil- dwelling ant species have traditionally taken place in single chambered, two-dimensional, soil- free, observation nests; often under constant light, food, and temperature conditions without in- corporating photoperiod or natural forager death rates (see Ch. 1) (Gordon et al., 2005; Mersch et al., 2013).While these laboratory studies describe general task-sets and the relative tendencies of old and young workers to perform interior or exterior jobs, they do not take in to consideration the three-dimensional architecture of natural nests, and the important connections of labor groups in physical space. This dissertation reinforces the observation that workers of different developmental stag- es do not mix haphazardly within the architecture of a natural nests and that additional task opportunities exist in field colonies, which are difficult to reproduce in the laboratory (Tschinkel, 2004). For example, the food items deposited by P. badius foragers in superficial chambers (12cm) are likely to be moved throughout the nest by workers belonging to a distinct labor

56 group. In a two dimensional laboratory nest, workers that move objects, and those that excavate soil, would be scored as inactive or otherwise, because their tasks are simply non-existent. Though birth and death are a central focus of models of temporal division of labor, laboratory studies on the subject also fail to integrate development rate and realized forager lifespan. Because of the incredible and unnatural longevity of foragers taken into the laboratory (see chapter 3), any studies of division of labor in ideal-laboratory conditions are subject to a less mean- ingful interpretation of labor ratios, as young workers accumulate and old workers continue to survive. Interpretations of the correlation between age and task are especially dangerous, as inhibition of behavioral development may arise due to a loss of extrinsic sources of mortality in terminal castes (chapter 3). Studies of division of labor in honeybees have been successful because observation hives are outdoors, workers are allowed to forage naturally and nest architecture does not preclude observation. Labor allocation in soil nesting ants should be studied with the same attention to au- thenticity, so that the seasonally distinct development rates related to predictable cycles of nutrition or temperature may be parsed out. Future studies of adaptive labor allocation in soil-nesting ant species should either take place in the field, or incorporate daily forager losses, seasonal diet, below-ground gradients, photoperiod and three-dimensional nest space.

4.2 In support of the superorganism

In this dissertation, worker development and colony allocation were investigated by com- bining whole colony censuses with single or multiple mark-recapture estimates of labor group size. Although simple, these tools revealed the tendency of certain component parts to change in synchrony within colonies, populations and species, providing a basis for countless experiments and comparative studies. By simply marking, recapturing and censusing ants, annual cycles of forager membership, chronological worker age, the longevity of workers, the proportion foraging, the social inhibition of worker development, the number of interacting foragers in neighborhoods and the change in the ratio of temporal castes to other demographic groups were revealed for P. badius. Within the focal population, relative labor group size, worker demography, worker death rate and seasonal worker lifespan showed little variance between colonies; demonstrating the conserved nature of colony-level traits.

Comparing the allocation of workers and reserves between competing functions is crucial to understanding the organism-like nature of colonies, yet most studies of social insect traits still focus on fine-scale worker characteristics at snapshots in time, without connecting them to the larger, colony reproductive cycle (Gordon, 1991; Sanders and Gordon, 2002; Schafer et al., 2006; Greene and Gordon, 2007; Pinter-Wollman et al., 2011, Mersch et al., 2013). Studies that follow individual worker transitions also fail to express worker tendancies as a proportion of the whole colony, dismissing the overall effect of such transitions on colony functioning. Like a traditional organism, a superorganism is composed of a hierarchy of specialized parts, which can only be understood once fully dissected. Just as a cell only has meaning in the context of an organ and that organ’s function in the body, the behaviors and histories of workers in highly integrated ant societies are best studied as components of a larger whole interacting with the environment (Reeve and Hölldobler, 2007; Hölldobler and Wilson, 2008) In 1989, Sober and Wilson identified groups as potential targets of selection in a multi-level framework, and in conjunction with Seeley (1989), redefined and demystified the idea of the superorganism. The de-emphasis on worker fitness tradeoffs in favor of colony-level selection, was met by the demonstration that species with genetically heterogeneous workers castes may out perform their more homogenous conspecifics, agonistic worker interactions may promote whole colony reproduction and that intercolonial competition could poise colonies as targets of selection (Powell and Tschinkel 1999; Reeve and Hölldobler, 2007; Wiernasz et. al 2008). The publication of The Super- oganism (Hölldobler and Wilson 2008), echoed a consensus among social insect biologists that consideration of colonies as analogous to organisms provides real, scientific value for comparative studies and contextualizing individual behavior. This dissertation provides further, unequivical evidence that each ant colony is analogous to an organism and the demography, development and regulating mechanisms of the workers within are shaped by selection acting on whole colony characteristics. The labor cycle of a harvester ant colony is a complex sequence of dependent phases and a striking example of how the interactions of thousands of transient individuals across generations coalesce to produce a single entity with measurable traits. The annual epic of a single Pogonomyrmex badius colony revealed by this dissertation and summarized below, represents just one of the countless and idi- osyncratic stories of contingency in social insect societies.

4.3 The annual cycle of a Florida harvester ant colony

Each March, chronologically old workers (200+ days) emerge from winter dormancy and begin to forage. During this period, agonistic interactions with neighboring colonies may account for as much as 30% of forager mortality. A pulse of sexual and worker brood appears in April and by early June, the first new workers of the year eclose. Although the physical and behavioral development rate of these workers is five times faster than that of the sisters that reared them, both die within 27 days of entering the forager force. In late June, alates emerge from the nest in synchrony with neighbors. Newly-mated queens struggle in trios, duos or alone to start colonies that will survive the coming winter, and if lucky, more than a decade. By mid-July, when foragers are most depleted of reserves (Tschinkel 1998), summer- born workers enter the forager population and replace them. For the remainder of the summer, colonies enter a period of growth and young workers make up the majority of the adult population. As weeks pass, forager mortality remains constant, but fewer and fewer foragers are replaced. The young workers produced in this season are only slightly larger than summer-born sisters, but their rate of cuticle darkening and behavioral development is five times slower. The number of these autumn-born workers amassed and their assimilated metabolic stores will determine the magnitude of colony reproduction the following year, when they enter the forager population for the first time (Tschinkel 1998). In October, brood production slows, and the ces- sation of foraging follows. During the span of a single year, a P. badius colony may move three times and re- excavate a nest more than two meters in depth, reorganizing a kilogram of seeds and thousands of larvae, pupae and charcoal bits in the process (Tschinkel 2013). Each nest may house a small army of kleptoparasitic beetles, collembola, mites, spiders and silverfish (Porter 1985, Kwapich, unpublished). Some workers will become infected with mermithid worms (more than 3cm long, unpublished) or fall prey to reduviids that wait eagerly on their mound. Inevitably, these small struggles draw to a close by December as workers migrate to the depths of their nests to enter a three month period of dormancy. The following March, new foragers will huddle on nest mounds, antennae waving, as some are touched by light for the first time. For a period of days they will gather and withdraw,

59 reluctant to depart. Then, moved by something yet to be described, each will step onto unfamiliar sand and the cycle will begin again.

REFERENCES

Bailey, N.T.J., 1952. Improvements in the interpretation of recapture data. Journal of Animal Ecology 21, 120-127.

Beers, W.C.M.v., Kleijnen, J.P.C., 2004. Kriging Interpolation in Simulation: A Survey Proceedings of the 2004 Winter Simulation Conference.

Beshers, S.N., Fewell, J.H., 2001. Models of division of labor in social insects. Annu Rev Entomol 46, 413-440.

Beshers, S.N., Traniello, J.F.A., 1994. The adaptiveness of worker demography in the attine ant Trachymyrmex septentrionalis. Ecology 75, 763-775.

Bonabeau, Theraulaz, G., Deneubour, J.L., 1998. Fixed response thresholds and the regulation of division of labor in insect societies. Bulletin of Mathematical Biology 60, 753 - 807

Bourke, A.F.G., Franks, N.R., 1995. Social evolution in ants. Princeton University Press, Princeton, N.J.

Calabi, P., Porter, S.D., 1989. Worker longevity in the fire ant Solenopsis invicta: ergonomic considerations of correlations between temperature, size and metabolic rates. J. Insect Physiol. 35, 643-649.

Calabi, P., Traniello, J.F.A., 1989. Social organization in the ant Pheidole dentata: Physical and temporal caste ratios lack ecological correlates. Behavioral Ecology and Sociobiology 24, 69-78.

Calderone, N.W., 1995. Temporal division of labor in the honey bee, Apis mellifera: a developmental process or the result of environmental influences? Canadian Journal of Zoology 73, 1410-1416.

Erickson, J.M., 1972. Mark-recapture techniques for population estimates of Pogonomyrmex ant colonies: an evaluation of the 32P technique. Ann. Entomol. Soc. Am. 65, 57-61.

Farris, S.M., Robinson, G.E., Fahrbach, S.E., 2001. Experience- and Age-Related Outgrowth of Intrinsic Neurons in the Mushroom Bodies of the Adult Worker Honeybee. 55, 1328 - 1338.

Franks, N.R., Tofts, C., 1994. Foraging for work: how tasks allocate workers. Anim. Behav. 48, 470-472.

Fukuda, H., 1983. The relationship between work efficiency and population size in a honeybee colony. Population Ecology 25, 249 - 263.

Gentry, J.B., 1974. Response to predation by colonies of the Florida harvester ant, Pogonomyrmex badius. Ecology 55, 1328-1338.

Golley, F.B., Gentry, J.B., 1964. Bioenergetics of the southern harvester ant, Pogonomyrmex badius. Ecology 45, 217-225.

Gordon, D.M., 1991. Behavioral flexibility and the foraging ecology of seed-eating ants. Am. Nat. 138, 379-411.

Gordon, D.M., 1996. The organization of work in social insect colonies. Nature 380, 121-124.

Gordon, D.M., Chu, J., Lillie, A., Tissot, M., Pinter, N., 2005. Variation in the transition from inside to outside work in the red harvester ant Pogonomyrmex barbatus. Insect. Soc. 52, 212- 217.

Greene, M.J., Gordon, D.M., 2007. Interaction rate informs harvester ant task decisions. Behav. Ecol. 18, 451-455.

Herbers, J.M., Adamowicz, S.C., Helms, S.D., 1985. Seasonal changes in social organization of Aphaenogaster rudis (Hymenoptera: Formicidae). Sociobiology 10, 1-15.

Holbrook, C.T., Fewell, J.H., 2010. Division of labor scales with colony size in the seed- harvester ant Pogonomyrmex californicus. [abstract], p. 251, In: Abstracts for the XVI Congress of the International Union for the Study of Social Insects, IUSSI, Copenhagen, Denmark, 8-13 August 2010.

Hölldobler, B., Wilson, E., 2008. The Superorganism: The Beauty, Elegance, and Strangeness of Insect Societies. W.W. Norton \\& Company.

Huang, Z., Robinson, G.E., 1996. Regulation of honey bee division of labor by colony age demography. Behavioral Ecology and Sociobiology 39, 147 - 158.

Huang, Z.Y., Robinson, G.E., 1992. Honeybee colony integration: worker-worker interactions mediate hormonally regulated plasticity in division of labor. Proceedings of the national academy of science USA 89, 11726 - 11172.

Johnson, B.R., 2008. Within-nest temporal polyethism in the honey bee. Behavioral Ecology and Sociobiology 62, 777 - 784.

Keeler, K.H., 1982. Preliminary Report of Colony Survivorship in the Western Harvester Ant (Pogonomyrmex occidentalis) in Western Nebraska. Southwest. Nat. 27, 245-246.

Kwapich, C., Tschinkel, W., 2013. Demography, demand, death, and the seasonal allocation of labor in the Florida harvester ant (Pogonomyrmex badius). Behavioral Ecology and Sociobiology 67, 2011-2027.

Lachaud, J.P., Fresneau, D., 1987. Social regulation in Ponerine ants, in: Pasteels, J.M., Deneubourg, J.L. (Eds.), From individual to collective behavior in social insects: les Treilles Workshop, Birkhauser, Basel, pp. 197-217.

Lee, R.D., 2003. Rethinking the evolutionary theory of aging: Transfers, not births, shape senescence in social species. Proceedings of the National Academy of Sciences 100, 9637-9642.

Leoncini, I., Le Conte, Y., Costagliola, G., Plettner, E., Toth, A.L., Wang, M., Huang, Z., Bécard, J.-M., Crauser, D., Slessor, K.N., Robinson, G.E., 2004. Regulation of behavioral maturation by a primer pheromone produced by adult worker honey bees. Proceedings of the National Academy of Sciences of the United States of America 101, 17559-17564.

Lincoln, F.C., 1930. Calculating Waterfowl Abundance on the Basis of Banding Returns. United States Department of Agriculture Circular 118, 1-4.

Lindauer, M., 1953. Division of labour in the honeybee colony. Bee World 34, 63-90.

McDonald, P., Topoff, H., 1985. Social regulation of behavioral development in the ant, Novomessor albisetosus (Mayr). J. Comp. Psychol. 99, 3-14.

Mersch, D.P., Crespi, A., Keller, L., 2013. Tracking Individuals Shows Spatial Fidelity Is a Key Regulator of Ant Social Organization. Science 340, 1090-1093.

Meudec, M., Lenoir, A., 1982. Social responses to variation in food supply and nest suitability in ants (Tapinoma erraticum). Anim. Behav. 30, 284-292.

Meyer, S.T., Leal, I.R., Wirth, R., 2009. Persisting hyper-abundance of leaf-cutting ants (Atta spp.) at the edge of an old Atlantic forest fragment. Biotropica 41, 711-716.

Mirenda, J.T., Vinson, S.B., 1979. A marking technique for adults of the red imported fire ant (Hymenoptera: Formicidae). Florida Entomol. 62, 279-281.

Mirenda, J.T., Vinson, S.B., 1981. Division of labor and specification of castes in the red imported fire ant Solenopsis invicta Buren. Anim. Behav. 29, 410-420.

Munch, D., Amdam, G.V., 2010. The curious case of aging plasticity in honey bees. FEBS Lett 584, 2496-2503.

Nakata, K., 1996. Does behavioral flexibility compensate or constrain colony productivity? Relationship among age structure, labor allocation, and production of workers in ant colonies. J. Insect Behav. 9, 557-569.

Naug, D., Camazine, S., 2002. The role of colony organization on pathogen transmission in social insects. J. Theor. Biol. 215, 427-439.

Nobua-Behrmann, B.E., Casenave, J.L., Milesi, F.A., Pavan, B., 2013. Forager abundance and its relationship with colony activity level in three species of South American Pogonomyrmex harvester ants. Insectes Sociaux 60, 243- 249.

Oettler, J., Johnson, R.A., 2009. The old ladies of the seed harvester ant Pogonomyrmex rugosus: Foraging performed by two groups of workers. J. Insect Behav. 22, 217-226.

Oster, G.F., Wilson, E.O., 1978. Caste and ecology in the social insects. Princeton University Press, Princeton.

Page, R.E., Mitchell, S.D., 1998. Self-organization and the evolution of division of labor. Apidologie 29, 171.

Pinter-Wollman, N., Wollman, R., Guetz, A., Holmes, S., Gordon, D.M., 2011. The effect of individual variation on the structure and function of interaction networks in harvester ants. J. R. Soc. Interface 8, 1562-1573.

Porter, S.D., Jorgensen, C.D., 1980. Recapture studies of the harvester ant, Pogonomyrmex owyheei Cole, using a fluorescent marking technique. Ecol. Entomol. 5, 263-269.

Porter, S.D., Jorgensen, C.D., 1988. Longevity of harvester ant colonies in southern Idaho. J. Range Manage. 41, 104-107.

Porter, S.D., Tschinkel, W.R., 1985. Fire ant polymorphism: the ergonomics of brood production. Behav. Ecol. Sociobiol. 16, 323-336.

Pratt, S.C., 1994. Ecology and behavior of Gnamptogenys horni (Formicidae, Ponerinae). Insect. Soc. 41, 255-262.

Ravary, F., Lecoutey, E., Kaminski, G., Châline, N., Jaisson, P., 2007. Individual experience alone can generate lasting division of labor in ants. Curr. Biol. 17, 1308-1312.

Reeve, H.K., Hölldobler, B., 2007. The emergence of a superorganism through intergroup competition. Proceedings of the National Academy of Sciences 104, 9736-9740.

Robinson, E.J.H., 2009. Physiology as a caste-defining feature. Insect. Soc. 56, 1-6.

Robinson, E.J.H., Feinerman, O., Franks, N.R., 2009. Flexible task allocation and the organization of work in ants. Proc. R. Soc. London B Biol. Sci. 276, 4373-4380.

Robinson, G.E., 1992. Regulation of Division of Labor in Insect Societies. Annual Review of Entomology 37, 637-665.

Robinson, G.E., Page, R.E., Strambi, C., Strambi, A., 1992. Colony Integration in Honey Bees: Mechanisms of Behavioral Reversion. Ethology 90, 336-348.

Rueppell, O., Bachelier, C., Fondrk, M.K., Page, R.E., Jr., 2007. Regulation of life history determines lifespan of worker honey bees (Apis mellifera L.). Exp Gerontol 42, 1020-1032.

Sanders, N.J., Gordon, D.M., 2002. Resources and the flexible allocation of work in the desert ant, Aphaenogaster cockerelli. Insect. Soc. 49, 371-379.

Santos, J.C., Yamamoto, M., Oliveira, F.R., Del Claro, K., 2005. Behavioral repertory of the weaver ant Camponotus (Myrmobrachys) senex (Hymenoptera: Formicidae). Sociobiology 46, 27-37.

Schafer, R.J., Holmes, S., Gordon, D.M., 2006. Forager activation and food availability in harvester ants. Anim. Behav. 71, 815-822.

Schmid-Hempel, P., 1992. Worker castes and adaptative demography. J. Evol. Biol. 5, 1-12.

Seeley, T.D., 1989. The honey bee colony as a superorganism. American Naturalist 150, 522 - 541.

Seid, M.A., Harris, K.M., Traniello, J.F.A., 2005. Age-related changes in the number and structure of synapses in the lip region of the mushroom bodies in the ant Pheidole dentata. J. Comp. Neurol. 488, 269-277.

Seid, M.A., Traniello, J.F.A., 2006. Age-related repertoire expansion and division of labor in Pheidole dentata (Hymenoptera: Formicidae): a new perspective on temporal polyethism and behavioral plasticity in ants. Behav. Ecol. Sociobiol. 60, 631-644.

Sendova-Franks, A., Franks, N.R., 1993. Task allocation in ant colonies within variable environments (A study of temporal polyethism, experimental). Bull. Math. Biol. 55, 75-96.

Smith, C.R., 2007. Energy use and allocation in the Florida harvester ant, Pogonomyrmex badius: are stored seeds a buffer. Behav. Ecol. Sociobiol. 61, 1479-1487.

Smith, C.R., Suarez, A.V., 2010. The trophic ecology of castes in harvester ant colonies. Funct. Ecol. 24, 122-130.

Smith, C.R., Tschinkel, W.R., 2006. The sociometry and sociogenesis of reproduction in the Florida harvester ant, Pogonomyrmex badius. J Insect Sci 6, 1-11.

Tofts, C., 1993. Algorithms for task allocation in ants. (A study of temporal polyethism: theory). Bull. Math. Biol. 55, 891-918.

Toth, A.L., Robinson, G.E., 2003. Worker nutrition and division of labour in honeybees. Animal Behaviour 69, 427 - 435.

Tripet, F., Nonacs, P., 2004. Foraging for work and age-based polyethism: the roles of age and previous experience on task choice in ants. Ethology 110, 863-877.

Tschinkel, W.R., 1987. Fire ant queen longevity and age: estimation by sperm depletion. Ann. Entomol. Soc. Am. 80, 263-266.

Tschinkel, W.R., 1988. Social control of egg-laying rate in queens of the fire ant, Solenopsis invicta. Physiol. Entomol. 13, 327-350.

Tschinkel, W.R., 1998. Sociometry and sociogenesis of colonies of the harvester ant, Pogonomyrmex badius: worker characteristics in relation to colony size and season. Insect. Soc. 45, 385-410.

Tschinkel, W.R., 1999. Sociometry and sociogenesis of colonies of the harvester ant, Pogonomyrmex badius: distribution of workers, brood and seeds within the nest in relation to colony size and season. Ecol. Entomol. 24, 222-237.

Tschinkel, W.R., 2004. The nest architecture of the Florida harvester ant, Pogonomyrmex badius. J Insect Sci 4, 21.

Tschinkel, W.R., 2011. The organization of foraging in the fire ant, Solenopsis invicta. J Insect Sci 11, 26.

Tschinkel, W.R., 2013. A method for using ice to construct subterranean ant nests (Hymenoptera: Formicidae) and other soil cavities. Myrmecological News 18, 99 - 102.

Wiernasz, D.C., Cole, B.J., 1995. Spatial distribution of Pogonomyrmex occidentalis: Recruitment, mortality and overdispersion. J. Anim. Ecol. 64, 519-527.

Wilson, E.O., 1953. The origin and evolution of polymorphism in ants. Q. Rev. Biol. 28, 136- 156.

Wilson, E.O., 1968. The ergonomics of caste in the social insects. Am. Nat. 102, 41-66.

Wilson, E.O., 1976a. Behavioral discretization and number of castes in an ant species. Behav. Ecol. Sociobiol. 1, 141-154.

Wilson, E.O., 1976b. A social ethogram of the Neotropical arboreal ant Zacryptocerus varians (Fr. Smith). Anim. Behav. 24, 354-363.

Wilson, E.O., 1980. Caste and division of labor in leaf-cutter ants (Hymenoptera: Formicidae: Atta). I. The overall pattern in Atta sexdens. Behav. Ecol. Sociobiol. 7, 143-156.

Wilson, E.O., 1984. The relation between caste ratios and division of labor in the ant genus Pheidole (Hymenoptera: Formicidae). Behav. Ecol. Sociobiol. 16, 89-98.

BIOGRAPHICAL SKETCH

Christina L. Kwapich

Education 2008 – 2014 PhD, Biological Science, Florida State University Advisor, Walter R. Tschinkel

2003- 2007 B.S., Entomology, Ohio State University Advisor, Susan C. Jones

2009 California Academy of Sciences Ant Course, Southwestern Research Station, AZ 2012 Sable System’s Respirometry Course, Las Vegas, NV 2014 Apalachee Beekeepers Association Beekeeping Short Course, Tallahassee, FL

Awards and Academic Honors 2014 Inducted, FSU Society of Fellows 2013 NSF Doctoral Dissertation Improvement Grant (DDIG) 2010 First Place, Behavioral Ecology, Student Competition for the President’s Prize (Ten-Minute Paper), National meeting of the Entomological Society of America, San Diego, CA 2010 NSF Integrated Training in Biology and Society fellowship (1 year of support) 2010 FSU Outstanding Teaching Assistant Award nominee 2009 FSU Department of Biological Science, Robert B. Short Scholarship in Zoology 2006, 2007 OSU Undergraduate Conference Travel Grant recipient 2004 NSF Research Experience for Undergraduates (REU) summer grant 2003 OSU National Buckeye Plus Scholarship (4 years of support) 2003 OSU Biological Sciences Scholar’s Community Scholarship

University Teaching assistantships Ohio State University - Biology 101: Introductory Biology for non- majors Florida State University - Insect Biology, Animal Behavior, Experimental Biology Guest lectures Insect Biology, Animal Behavior, Ecology Lab, Environmental Science

Publications Kwapich, C.L., Tschinkel, W.R. (2014), Foraging kills, staying home increases longevity and inhibits new forager development in the Florida harvester ant (Pogonomyrmex badius) (In review).

Kwapich, C.L., Tschinkel, W.R. (2013). Demography, demand, death and the seasonal allocation of labor in the Florida harvester ant (Pogonomyrmex badius). Behavioral Ecology and Sociobi- ology, 67:12.

Gibson, A. H., Kwapich, C.L., Lang, M. (2013). The Roots of Multilevel Selection Theory: Con- cepts of Biological Individuality in the Early Twentieth Century. History and Philosophy of the Life Sciences; 35:4.

Rink, W.J., Dunbar, J.S., Tschinkel, W.R., Kwapich, C.L., Repp, A., Stanon, W., Thulman, D.K. (2013). Subterranean transport and deposition of quartz by ants in sandy sites relevant to age overestimation in optical luminescence dating. Journal of Archaeological Science 40:4, 2217- 2226.

Tschinkel, W.R., Murdock, T, King, J.R, Kwapich, C.L. (2012). Ant distribution in relation to ground water in north Florida pine flatwoods. Journal of Insect Science 12:114.

In preparation Kwapich, C. L. (2014) Neighbor density influences forager longevity, production and colony growth in the seeder harvesting ant Pogonomyrmex badius.

Tschinkel, W.R., Kwapich, C. L., Murdock, T. (2014) Group size influences energetics and success in pleometrotic foundresses of the Florida harvester ant (Pogonomyrmex badius) Kwapich, C. L., Gibson, A. H., Lang, M. (2015) Whole colony connectivity, forager and egg ex- change in a polydomous carpenter ant (Camponotus socius)

Kwapich, C.L. (2015) A curious case of Alleculid kleptoparasitism in colonies of the Florida Harvester Ant (Pogonomyrmex badius)

Kwapich, C.L. (2015) A novel description of the association of an mimicking spider with its model, the Florida harvester ant (Pogonomyrmex badius)

Presentations at Meetings/Symposia . Christina Kwapich (2014). Neighbor removal increases forager longevity, slows progression through temporal castes (P. badius). Integrated analyses of division of labor, world congress of the IUSSI,Australia

Christina Kwapich (2014). Development, death, density, demand division of labor in a seed harvesting ant. Arizona State University, School of Life Sciences, invited seminar

Christina Kwapich, Walter Tschinkel (2013). Meddling neighbors induce an untimely end for foragers of the Florida harvester ant, Pogonomyrmex badius. Entomological Society of America Meeting, Austin, TX.

Christina Kwapich (2013). How to assemble a Pogonomyrmex badius colony from the bottom up, cookie shovel and wire required. Natural History as Insight and Inspiration Symposium, Tal- lahassee, FL

Christina Kwapich, Walter R. Tschinkel (2012). The Influence of Demand, Demography and Death on Labor Economics in the Florida Harvester Ant (Pogonomyrmex badius). Interna- tional Union for the Study of Social Insects NAS meeting, Greensboro, NC.

Christina Kwapich, Walter R. Tschinkel (2011). Seasonal worker demography shapes colony- level labor allocation in the Florida harvester ant (Pogonomyrmex badius). Symposium: Insect Demography, emerging concepts and applications. Entomological Society of America Meeting, Reno, NV.

Christina Kwapich, Walter R. Tschinkel (2010). Annual patterns of forager allocation in the Florida harvester ant (Pogonomyrmex badius). Entomological Society of America Meeting, Stu- dent Competition (first prize), San Diego, CA.

Christina Kwapich, Walter Tschinkel (2009). The organization and allocation of foragers in the Florida harvester ant (Pogonomyrmex badius). Entomological Society of America Meeting, Indi- anapolis, IN.

Christina Kwapich, Susan C. Jones (2006). Termite (Isoptera) Caste differentation in response to spatial separation from the reproductive female. Denman Undergraduate Research Forum, Co- lumbus, OH.

Christina Kwapich, Susan C. Jones, Nicola T. Gallagher (2006). Spatial dynamics of neotenics of Reticulitermes flavipes (Isoptera: Rhinotermitidae): male preference and ideal females. Entomo- logical Society of America Meeting, Indianapolis, IN.

Joan M. Herbers, Christina Kwapich (2004). Dysfunctional families in the insect world. Coali- tion for National Science Funding , Washington, D.C.