SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

Although there is much interest in behavioral syndromes, very little is known about how syndromes are generated in wild populations. Here, we assess the roles of correlated selection and divergent rearing environments in generating a syndrome between foraging aggressiveness and boldness in the spider Agelenopsis pennsylvanica. We first tested for and confirmed the presence of a behavioral syndrome between boldness and foraging aggressiveness in wild penultimate A. pennsylvanica (r = 0.24). Then, to assess the effects of rearing environment on the boldness–aggressiveness syndrome, we compared the behavioral tendencies of field- vs. laboratory-reared spiders over the course of their development. The presence of the boldness–aggressiveness syndrome differed based on spiders' developmental stage and rearing environment: field-reared juveniles did not exhibit a syndrome between boldness and foraging aggressiveness, but field-reared penultimates did. In contrast, laboratory-reared spiders never exhibited a behavioral syndrome, regardless of their developmental stage. Thus, the boldness–aggressiveness syndrome in A. pennsylvanica manifests only when individuals are reared in the field. Selection data from a mark–recapture study failed to indicate any signature of correlated selection, despite our finding that at least one element of the syndrome (foraging aggressiveness) can respond to selection (Heritability h2 = 0.27, from mid-parent breeding study). Thus, contemporary correlated selection does not appear to be a major driver of the boldness–aggressiveness syndrome of A. pennsylvanica. Taken together, our data are consistent with the hypothesis that the boldness–aggressiveness syndrome exhibited by wild A. pennsylvanica develops as a result of environmentally induced phenotypic plasticity, and not correlated selection.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

Behavioral syndromes are defined as suites of correlated behavioral traits that are expressed across time, situation, or ecological context (Sih et al. 2004a,b). A growing body of literature is devoted to documenting the developmental (Stamps & Groothuis 2010a,b), physiological (Biro & Stamps 2010), and genetic (Bell & Aubin-Horth 2010; Bell & Robinson 2011) underpinnings to behavioral syndromes, while a second, complementary literature has focused on their ecological (Kortet et al. 2010; Sih et al. 2012) and evolutionary implications (Dingemanse & Reale 2005; Reale et al. 2007). A common inference from the behavioral syndromes literature is that syndromes imply limited plasticity, and many authors have suggested that behavioral correlations can generate performance trade-offs and/or act as an evolutionary constraint (Johnson & Sih 2005; Duckworth 2006; Pruitt et al. 2010). Under the constraint hypothesis, functionally dissimilar contexts of behavior are thought to be linked by shared physiological (e.g., hormones, metabolic rate, organ size) and/or genetic underpinnings (e.g., physical linkage, pleiotropy) (Careau et al. 2010; Reale et al. 2010; van Oers et al. 2011; K. E. McGhee, L. M. Pintor & A. M. Bell, in review). However, another body of literature has questioned the constraining potential of behavioral syndromes and suggests that syndromes are labile—appearing sensitive to variation both in the contemporary (Ruiz-Gomez et al. 2008, 2011; Korpela et al. 2011) and historic environment (Bell 2005; Dingemanse et al. 2007). Consequently, the view that syndromes can generate evolutionary constraints is still contentious: (1) because there is system-specific evidence both for and against the constraining potential of syndromes, and (2) because studies elucidating the relative influences of genetic vs. developmental effects on syndromes (in unified test systems) remain scare (Bell 2005; Pruitt & Riechert 2009; Kralj-Fiser & Schneider 2012).

Spiders have served as a long-standing model for behavioral syndromes research. Behavioral syndromes have been documented in at least seven families and twenty different species of spiders (Pruitt & Riechert 2012). Moreover, as compared to other taxa (Herczeg & Garamszegi 2012), behavioral syndromes are believed to be quite stable in spiders, because distantly related species (Sih et al. 2004b; Johnson & Sih 2007; Pruitt et al. 2011; Kralj-Fiser et al. 2012) and populations from ecologically divergent habitats (Riechert 1993; Riechert & Hedrick 1993; Pruitt et al. 2010) often exhibit similar syndromes. The most common syndrome observed in spiders is a positive correlation between individuals' boldness and their aggressiveness toward prey and/or mates (Riechert et al. 2001; Johnson & Sih 2005; Pruitt et al. 2008), where bold individuals are also more aggressive. However, the literature on spiders also suffers from a weakness that is shared by the majority of the behavioral syndromes literature: the role of development in generating behavioral syndromes has been underappreciated. Instead, investigations have, with two exceptions (Riechert & Hedrick 1993; Kralj-Fiser & Schneider 2012), focused on field-collected individuals to test for the presence of syndromes (Johnson & Sih 2007; Pruitt et al. 2008, 2010) and have generally focused their attention on a single life stage (typically mature individuals). Thus, the aggressive syndromes commonly documented in spiders could easily be the result of environmentally induced phenotypic plasticity in response to shared environmental cues/pressures (e.g., sharing a similar position in food webs). Studies that (1) track the behavioral tendencies of individuals over the course of their development in the field and (2) compare the syndromes of individuals reared in different environments will help to reveal the role of rearing environment in generating behavioral syndromes in spiders and other taxa (Stamps & Groothuis 2010a,b; DiRienzo et al. 2012). The predictions being that if environmentally induced phenotypic plasticity is a driver of behavioral syndromes, then we should see (1) behavioral syndromes that emerge over the course of individuals development (i.e., as they gain experience) and (2) that their expression should be dependent on individuals' rearing environment, because divergent rearing environments will confer divergent experiential effects. Once environmental and ontogenetic contingencies have been documented, experimental manipulations of rearing environment will help to elucidate the cues responsible for generating the syndrome.

Here, we investigate the role of environmentally induced phenotypic plasticity in the development of a behavioral syndrome in the spider Agelenopsis pennsylvanica (Araneae, Agelenidae). The genus Agelenopsis is a classic model for behavioral syndromes research, where Riechert and her collaborators have (1) identified a syndrome that encompasses virtually every aspect of this genus' behavior (e.g., foraging, antipredator, territoriality, locomotion, mating)( Riechert & Hedrick 1993; Riechert et al. 2001; Pruitt & Husak 2010); (2) characterized various context-specific trade-offs associated with the syndrome (Riechert et al. 2001; Riechert & Johns 2003); and (3) obtained evidence for a genetic basis to several facets of the syndrome (Hedrick & Riechert 1989; Riechert & Smith 1989). Interestingly, in spite of the sizable number of behavioral syndromes studies devoted to this genus and to spiders in general, no studies have successfully tracked the behavior of individuals over the course of their development in the laboratory vs. field environment in order to simultaneously discern the contributions of (1) rearing environment and (2) developmental contingencies in the formation of behavioral syndromes. In the two studies on spiders that have compared the syndromes of laboratory vs. field-reared spiders, one study found that the behavioral syndrome of field-collected individuals vanished in laboratory-reared F1s (Kralj-Fiser & Schneider 2012), whereas another found that field-collected spiders and laboratory-reared F1s exhibited similar syndrome structures (Riechert & Hedrick 1993). Thus, more studies on other species are needed to explore the generality of either outcome.

We ask the following questions: (1) Do field-collected penultimate A. pennsylvanica exhibit a behavioral syndrome between foraging aggressiveness and boldness, as seen in other species of spider?; (2) Are the syndromes of field-collected and laboratory-reared spiders different?; (3) Is correlated selection associated with the emergence of behavioral syndromes in the wild (i.e., where particular trait combinations yield higher survivorship/fitness, and individuals that fall off of this correlated axis are ‘weeded out’ by selection)?; (4) Are the components of the syndrome heritable, as estimated by offspring on mid-parent regression? The rational being, that in order for the syndrome to respond to selection it must be heritable (Boake 1994). Through a combination of behavioral assays, breeding experiments and a mark–recapture study in a wild population, our investigation is designed to probe the relative contributions of two non-mutually exclusive processes in generating a behavioral syndromes in a wild population: contemporary correlated selection and environmentally induced phenotypic plasticity.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

Collection and Laboratory Maintenance

Spiders of genus Agelenopsis build funnel-webs that are composed of two parts: a non-sticky sensory sheet used to sense prey and a funnel retreat. When foraging, spiders position themselves at the entrance to their funnel and use high-velocity movements to subdue prey that make contact with their webs (Riechert 1976; Pruitt & Husak 2010). Late instar A. pennsylvanica were collected along urban hedgerows throughout Pittsburgh, Pennsylvania (213 Females, 188 Males) in July–Aug 2011, to serve as the parental generation in our breeding study. Spiders were collected by dropping a cricket onto the edge of their sheet web and scooping up the spider as they initiated their prey-capture sequence. Placing the prey item on the edge of the sheet minimizes the damage to the web, leaving the majority of the structure intact. Spiders were transported to the laboratory and housed in circular, transparent containers (radius = 9 cm, depth 7.5 cm). Spiders were provided an ad libitum diet of two-week-old crickets once a week. To provide water, spiders' webs were misted once a week using a spray bottle containing tap water. These spiders, which constituted the parental generation in our breeding study (hereafter termed ‘field-collected individuals’), were run through foraging aggressiveness and boldness assays (described below) upon reaching their penultimate molt. To assess whether the syndrome present in field-collected individuals was maintained in laboratory-reared spiders, our field-collected spiders were mated randomly following the protocol described below. We randomly selected one individual per brood (135 broods) for inclusion in our syndrome analyses on the F1 offspring. We used only 54 of the 135 total broods in our heritability analysis, because we failed to assay the foraging aggressiveness and boldness of a subset of our field-collected males. All spiders were weighed and measured within one week of reaching their penultimate instar. We measured the cephalothorax length and width of all penultimate spiders to the nearest hundredth of a millimeter using Leica digital imaging software (LAS V3.8) and stereomicroscope (Leica M80).

Assessment of Foraging Aggressiveness: Latency of Attack

All spiders were run through a latency of attack assay to determine their aggressiveness toward prey. Assays occurred in spiders' home containers, three days after a routine feeding. Trials were initiated by removing the lid to the spiders' containers. We then provided 30 s of acclimation time before a single two-week-old cricket was dropped centrally within the spider's web. We recorded the time taken for spiders to make contact with the cricket, termed their ‘latency of attack’. Latency of attack trials are a common measure of aggressiveness in spiders (Hedrick & Riechert 1989; Pruitt et al. 2008; Kralj-Fiser & Schneider 2012), where spiders with lower latencies to attack prey are deemed more aggressive. Individual differences in latency of attack in field-collected A. pennsylvanica were shown to be repeatable in a previous study (r = 0.41)(Berning et al. 2012).

Assessment of Boldness: Latency to Resume Movement

To assess spiders' boldness toward predator cues, we use a ‘puff test’ following the protocol of Riechert & Hedrick (1993). Boldness assays occurred 24 h after latency of attack assays and occurred in spiders' home containers. Trials are initiated by lifting off the lid to spiders' containers and allowing 60 s acclimation time before the application of an aversive stimulus. We then ‘puffed’ the spider's anterior, dorsal side with two rapid jets of air using an infant ear-cleaning bulb. Spiders universally responded by fleeing and huddling at the base of their funnel, with their legs drawn toward their cephalothorax. Boldness is estimated as the time taken for spiders to resume movement following this cue. Spiders with shorter latencies to resume movement are deemed more ‘bold’, and spiders with longer latencies are termed more ‘shy’. To test whether boldness was repeatable, we ran a second pool of penultimates (n = 41) collected in 2012 through three boldness assays. Spiders were testing once every three days, for nine consecutive days (three measurements per individual). Spiders used to assess repeatability were not used in any other studies.

Breeding Design

In order for a trait to respond to selection, it must exhibit additive genetic variation (i.e., it must be heritable). To assess the heritability of our behavioral traits, we employed a mid-parent of offspring breeding design (Falconer & McKay 1996). Male–female encounters (N = 181) were staged in females' home containers. Male–female pairings were determined randomly, and males were not reused with multiple females. Males were entered onto the side of females' containers using an open-tipped syringe. The lids to females' containers were left ajar to provide males an escape route, should the female attack the male. We observed pairs for the next 6–8 h, to confirm the mating event. Mating events occurred between Aug–Oct 2011. Mating pairs were kept under standard laboratory fluorescent lighting and ambient temperatures (21.6–24.4 C). We estimated heritability using offspring on mid-parent regression, where heritability is estimated as the slope of the regression line (Boake 1994; Falconer & McKay 1996).

Rearing Protocol- F1s

Egg cases resulting from our male–female pairings were placed in circular, clear plastic containers (radius = 6 cm, height = 2 cm) and housed in an environmental chamber at 27 C and 70% humidity. Upon hatching, broods were fed and housed communally until their third instar. Following their second molt, individuals were removed from their natal web and housed individually in 59-ml plastic cups. F1 individuals were run through the foraging aggressiveness and boldness assays (described above) once per instar for instars 5–8 (4 measures). Instar 8 is the penultimate instar in this population of A. pennsylvanica, and corresponds to the developmental stage at which we assayed the parental generation. Assays were conducted 2–4 d after spiders completed their recent molt.

Mark–Recapture Study

To assess the contribution of correlation selection in generating the boldness–aggressiveness of A. pennsylvanica, we performed a mark–recapture study. We collected 61 juvenile (estimated instar 3–5) A. pennsylvanica in July 2011 using the baiting protocol described above. Spiders were then transported back to laboratory and run through the latency of attack and boldness trials using the same protocols used for laboratory-reared spiders, and their mass was recorded (Mettler-Toledo Microbalance XP205). Spiders were each assigned a unique series of three colored paint dots, placed atop their cephalothorax. Spiders were subsequently replaced atop their original webs in the field, and their persistence was tracked over the next 40 d. Spiders were checked daily for molts. Spiders that molted were collected and remarked. In most instances, the spider's molt, bearing its former markings, could be observed within the webs. We terminated the experiment when spiders reached their 7–8th instar. We successfully tracked 53 of the 61 individuals over the 40-day period and, in 49 of these instances, individuals remained on the same webs where they were first released. Another four individuals dispersed short distances (<3 m). Individuals missing were presumed to be dead (n = 4). In four other instances, we were able to document mortality as a result of parasitism by small-headed flies (Diptera, Acroceridae), which are common parasites of Eastern populations of Agelenopsis (J. N. Pruitt, A. Gallasso & S. E. Riechert, pers. obs.). The 53 individuals (32 females, 21 males) that we recovered at the end of our experiment were brought back to laboratory and run through the latency of attack and boldness assays described above to (1) decipher whether the population's behavioral tendencies had shifted over the course of their development, and (2) determine whether the syndrome of early-instar juveniles and penultimate spiders differed significantly. Spiders recovered at the end of our study had their mass recorded within 48 h of collection (Mettler-Toledo Microbalance XP205).

Statistical Methods

To determine the repeatability of our foraging aggressiveness and boldness assays in our field-collected and F1 laboratory-reared spiders, we used analysis of variance and partitioned the variance into within-individual vs. among-individual components. The resulting intraclass correlation coefficient was used as our estimate of repeatability (Boake 1989; Falconer & McKay 1996). To test for the presence of a behavioral syndrome between foraging aggressiveness and boldness, we ran Spearman's rank-transformed correlations. For F1 offspring, we ran four such analyses, one for each instar for instars 5–8 (Bonferroni corrected α = 0.0125). Separate analyses were run for males and females. We tested for ontogenetic shifts in boldness and foraging aggressiveness using two-way ANOVAs, with instar and sex as our class variables and boldness and aggressiveness as our response variables. We estimate heritability of boldness and foraging aggressiveness using offspring on mid-parent regression (54 broods)(after Boake 1994; Falconer & McKay 1996), where heritability is estimated as the slope of the resulting regression line. For this analysis, we compared the boldness and foraging aggressiveness of penultimate F1s to the behavior of their parents' phenotypes, when taken at the same developmental stage.

We used linear regression analysis to test for standardized directional, quadratic and correlated selection gradients on early-instar juveniles' boldness, latency of attack, body mass, sex, all two-way interaction terms, and three quadratic terms (boldness2, latency of attack2, body length2) (Lande & Arnold 1983). In this analysis, (1) regression coefficients of single traits (βi) reflect the intensity of directional selection, where extreme traits in either direction are favored; (2) the quadratic terms (γii) reflects the intensity of disruptive or stabilizing selection, where extreme values on either end of the distribution are either favored or disfavored, respectively; and (3) interaction terms (γij) reflect the importance of correlated selection, where particular trait combinations are favored (e.g., bold aggressive individuals and fearful non-aggressive individual enjoy highest fitness). We estimated the effects of these traits on three estimates of individual fitness: survival (1,0), change in absolute mass (g), and a composite metric (survival*change in mass). Because scores of survival were binary, we used logistic regression to test for significance of partial regression estimates that were obtained from multiple linear regression (Janzen & Stern 1998). Quadratic regression coefficients estimated were doubled, after Stinchcombe et al. (2008). All predictor variables were transformed into mean zero and unit variance prior to our analyses, and relative fitness was obtained by dividing individuals' estimate of fitness by the mean value of the population (Bell & Sih 2007; Calsbeek & Irschick 2007; Pruitt & Krauel 2010). In instances, where we detected a significant effect of a single term within a non-significant combined model, we performed a backward stepwise regression procedure to test whether the significance of the single term was retained within a reduced model. Under no circumstances were such reduced models significant.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

Repeatabilities & Behavioral Syndromes

We detected significant repeatabilities for both foraging aggressiveness and boldness in both our laboratory-reared spiders (foraging aggressiveness F160,463 = 1.27, p = 0.03, r = 0.25; boldness F159,445 = 1.50, p = 0.001, r = 0.34) and field-collected spiders (aggressiveness: r = 0.24 (Berning et al. 2012); boldness: F40,82 = 1.99, p = 0.004, r = 0.49), thus confirming the presence of stable individual differences in their behavior.

We detected a syndrome between foraging aggressiveness and boldness in our penultimate field-collected individuals used as the parental generation in our breeding study (Females: r = 0.30, p = 0.004, df = 90; Males: r = 0.56, p < 0.001, df = 39), where individuals that exhibited low latencies to attack prey also exhibited low latencies to resume movement following a predator cue. In other words, bold individuals also tended to be more aggressive toward prey. For spiders in our mark–recapture study, we failed to detect an association between boldness and latency of attack in juvenile spiders of either sex (Table 1); however, we detected a strong association between these behavioral traits when we resampled the same cohort of individuals 40 d (i.e., 3–5 instars) later (Fig. 1, Table 1). In contrast, an instar-by-instar analysis on laboratory-reared spiders revealed no significant association between individuals' latency of attack and their boldness during any point of the development in laboratory (Table 2), regardless of sex.

Table 1. Correlations between latency of attack and boldness in mark–recapture A. pennsylvanica at two moments during their development: middle instar juveniles (≈5th instar) and penultimate spiders (8th instar)
SexEstimated InstarCorrelation Coefficientdfp-value
Males
 5th0.29240.16
8th0.43210.04
Females
 5th−0.12360.54
8th0.5731<0.001
Table 2. Correlation coefficient estimates between foraging aggressiveness (latency of attack) and boldness (latency to resume movement following an aversive stimulus) in laboratory-reared A. pennsylvanica, subdivided by sex and instar
SexInstarCorrelation Coefficientdfp-value
  1. At no point did we detect a significant association between the two traits for either sex.

Males
 5th−0.12760.29
6th0.14760.21
7th−0.07760.52
8th0.01760.91
Females
 5th−0.11820.38
6th−0.11820.38
7th0.02820.88
8th−0.11820.34
image

Figure 1. Scatterplot depicting the relationship between latency of attack and boldness in field-reared A. pennsylvanica at two moments during their development: early-instar juveniles (estimated instars 3–5) and penultimate spiders (8th instar). Early-instar juveniles (a) failed to show a significant relationship between boldness and latency to attack, but penultimate spiders (b) exhibited a positive association. We failed to recover only eight of the original 61 spiders used in this study.

Download figure to PowerPoint

Given the low mortality of individuals over the course of our mark–recapture study (detailed under ‘Selection Analyses’ below), we conjectured that the syndrome that emerged in penultimate spiders was the result of shifts in individuals behavioral tendencies over the course of their development. To further explore this hypothesis, we tested for an association between latency of attack and boldness using only those juveniles that persisted for the duration of our mark–recapture study (N = 53). If the emergent syndrome were the result of developmental shifts, we predicted that we should fail to detect an association between boldness and foraging aggressiveness in this subset of individuals, when sampled at instars 3–5. As predicted, we failed to detect any association between boldness and latency of attack for this subset of individuals early in their development (Male: r = −0.18, df = 22, p = 0.27; Females: r = −0.15, df = 32, p = 0.39), however, the exact same individuals exhibited a highly significant association 40 d later (Males: r = 0.43; Females: r = 0.57 Table 1). A Z-test comparing the syndromes of juveniles vs. penultimate spiders in our mark–recapture study confirmed that the correlation between boldness and latency of attack changed over the course of individuals' development (Males: Z = 2.03, p = 0.04; Females: Z = 2.94, p = 0.003). However, a Z-test comparing the syndromes of laboratory-reared spiders over the same developmental period failed to detect any significant difference (Males: Z = 0.79, p = 0.42; Females: Z = 0.06 p = 0.95; Table 2). Finally, Z-tests comparing the syndromes of laboratory-reared and mark–recapture spiders at two moments in their development demonstrate that the syndromes of laboratory-reared and mark–recapture spiders did not differ early on in spiders' development (Males: Z = 1.73, p = 0.08; Females: Z = −1.13, p = 0.26), but had diverged significantly for females (but not males) by the time spiders had reached their penultimate instars (Males: Z = 1.91, p = 0.056; Females: Z = 3.50, p < 0.001). Taken together, the syndrome between boldness and foraging aggressiveness emerges over the course of individuals' development, but only when individuals are reared in particular environments. This environmentally contingent aspect to the syndrome's development strongly implies a role of phenotypic plasticity.

Heritability

Using offspring on mid-parent regression, we detected a moderate heritability of latency of attack (F1,53 = 10.02, p = 0.003, R2 = 0.26, h2 = 0.27, Fig. 2), but we failed to detect a heritability of boldness (F1,53 = 1.02, p = 0.34, R2 = 0.04, h2 = 0.10). Thus, the foraging aggressiveness of laboratory-reared penultimate resembles the foraging aggressiveness of their parents when taken at the same developmental stage. However, if we remove instances where spiders were unresponsive to prey (for latency of attack) or failed to resume movement following an aversive stimulus (for boldness), our heritability estimate for latency of attack increased dramatically (F1,36 = 41.04, p < 0.0001, R2 = 0.53, h2 = 0.51), whereas we still failed to detect a significant heritability after removing these individuals (F1,49 = 0.02, P 0.90, R2 < 0.01, h2 = −0.01).

image

Figure 2. Offspring on mid-parent regression of latency to attack in the spider A. pennsylvanica. Heritability is estimated as the slope of the resulting regression line (h2 = 0.27). Parents were collected as late instar juveniles from the field, and offspring were raised in standardized laboratory conditions. Both parents and offspring were measured at their penultimate instar, to control for ontogenetic shifts in behavioral tendencies.

Download figure to PowerPoint

Ontogenetic Shifts in Mean Trait Values

Our combined model testing the effects of sex, instar, and the interaction term sex*instar on foraging aggressiveness was highly significant (F7,620 = 4.37, p < 0.0001, r2 = 0.05). We detected a significant main effect of instar (F3,620 = 8.94, p < 0.0001, Fig. 3), whereas the effects of sex (F1,620 = 1.99, p = 0.16) and the interaction term were non-significant (F3,620 = 0.61, p = 0.61). Similarly, our combined model testing the effects of sex, instar, and their interaction term sex*instar on boldness was also highly significant (F7,620 = 7.32, p < 0.0001, r2 = 0.07). We detect a significant main effect of instar (F3,620 = 13.86, p < 0.0001, Fig. 3) and sex (F1,620 = 5.73, p = 0.02), but their interaction term was non-significant (F3,620 = 0.88, p = 0.45). Spiders in instars 5–7 exhibited similar levels of foraging aggressiveness and boldness, but showed less aggressive (Instar 3–7 = 67.91 ± 8.43 s, Instar 8 = 113.23 ± 10.22 s) and less bold behavioral tendencies as penultimates (Instar 3–7 = 127.59 ± 8.17, Instar 8 = 187.92 ± 9.72), and females (inline image = 132.95 ± 8.55) tended to exhibit slightly greater boldness than males (inline image = 153.04 ± 8.5).

image

Figure 3. Bar chart depicting the latency of attack (top) and boldness (bottom) of A. pennsylvanica over the course of their development when reared in laboratory. Spiders were measured once per instar, for instars 5–8. The center line marks the mean, and error bars represent the standard error of the mean. Bars not sharing a letter were significantly different at p < 0.05 using post hoc Tukey's tests, which correct for multiple comparisons. Spiders remained relatively bold and aggressive throughout their development, but shifted toward less aggressive and less bold behavior upon reaching their penultimate instar.

Download figure to PowerPoint

A population-level comparison of our mark–recapture spiders reveals that, as with laboratory-reared spiders, penultimate individuals exhibited longer latencies to attack prey than younger spiders (Combined model: F3,107 = 6.45, p = 0.008; Instar: F1,110 = 4.45, p = 0.037; Sex: F1,110 = 2.13, p = 0.13, Fig. 4), but no such shift was detect for boldness (Combined model: F3,107 = 1.85, p = 0.14; Instar: F1,110 = 3.35, p = 0.08; Sex: F1,110 = 2.02, p = 0.16, Fig. 4).

image

Figure 4. Bar chart depicting the latency of attack (top) and boldness (bottom) of field-reared A. pennsylvanica at two moments in their development: early-instar juveniles (estimated instars 3–5) and penultimate spiders (8th instar). The center line marks the mean, and error bars represent the standard error of the mean. Bars not sharing a letter were significantly different at p < 0.05 using post hoc Tukey's tests, which correct for multiple comparisons. Early-instar spiders were more aggressive than penultimate spiders.

Download figure to PowerPoint

Average Foraging Aggressiveness and Boldness of Field- vs. Laboratory-reared Spiders

We failed to detect a significant difference in latency of attack measures of early-instar juveniles reared in laboratory vs. the field (t = −0.60, df = 214, p = 0.55; Laboratory-reared = 58.42 ± 7.86 s, Field-collected = 67.04 ± 11.86 s). Similarly, we failed to detect a difference a difference in latency of attack measures for laboratory- vs. field-reared penultimates (t = 0.69, df = 211, p = 0.49; Laboratory-reared = 113.24 ± 10.36 s, Field-collected = 99.98 ± 12.00 s).

We also failed to detect a significant difference in the boldness of early-instar juveniles reared in laboratory vs. the field (t = −1.76, df = 219, p = 0.08; Laboratory-reared = 115.51 ± 8.56 s, Field-collected = 142.18 ± 12.31 s). However, we detected a significant difference in the boldness of penultimate spiders reared in the laboratory vs. the field (t = 4.43, df = 211, p < 0.001; Laboratory-reared = 187.93 ± 8.83 s, Field-collected = 113.83 ± 12.17 s), where penultimate spiders reared in the field were more bold than penultimate spiders reared in laboratory.

Selection Analyses

Our mark–recapture study revealed a very low incidence of mortality at our study site: we failed to recover only 8 of our original 61 spiders. In contrast, variation in growth rate (∆mass) was considerable, with the average individual growing eight times its original body mass over 40 d. Regardless of the fitness proxy, we failed to detect any evidence for correlated selection between foraging aggressiveness and boldness (Table 3); thus, the syndrome does not appear to be driven by the increased growth or survivorship of individuals bearing certain, advantageous trait combinations (e.g., bold and aggressive, fearful and non-aggressive, bold and non-aggressive, etc.). Our combined model predicting survivorship over the course of our study was non-significant, as were all individual terms therein (χ261 = 0.97, p = 0.65). Similarly, our combined model predicting growth (∆mass) was non-significant (F13,40 = 0.35, p = 0.97), and we detected no evidence for selection of any kind (p > 0.44)(Table 3). Finally, our combined model predicting our composite performance metric (survival*change in mass) was also non-significant (F13,48 = 0.60, p = 0.84, Table 3).

Table 3. Selection analyses testing for correlated selection on boldness and foraging aggressiveness in our mark–recapture experiment on A. pennsylvanica (N = 62)
Response VariableSourceβ/γSE(β/γ)tp-value
  1. Standardized directional (βi), quadratic (γii), and correlated (γij) selection gradients on survival (persistence in the field over 40 d: 1,0), growth (∆mass), and a composite performance metric (survival*change in mass) are presented.

  2. p-values calculated using logistic regression for selection gradients estimated from survival, and linear regression for selection gradients estimated from growth and the composite metric. β and γ are estimated using linear regression coefficients, and the estimates and SE of the quadratic terms (γii) are doubled (after Stinchcombe et al. 2008).

Survival
 Sex−0.0010.042−0.030.97
Mass0.0050.0460.120.90
Latency of Attack−0.1740.105−1.640.11
Boldness−0.0240.047−0.520.60
Sex*Mass−0.0390.042−0.940.35
Sex*Latency of Attack−0.0250.051−0.490.62
Sex*Rel Bold−0.0320.041−0.820.42
Mass*Latency of Attack−0.0990.047−1.130.14
Mass*Boldness−0.0190.039−0.50.62
Latency of Attack*Boldness0.0120.0490.250.81
Mass*Mass−0.2170.091−1.390.10
Latency of Attack*Latency of Attack0.2160.1231.750.09
Boldness*Boldness0.0710.0940.750.46
Growth
 Sex0.0440.1760.250.80
Mass−0.1670.214−0.780.44
Latency of Attack−0.2170.446−0.490.63
Boldness0.0030.1870.020.99
Sex*Mass−0.0470.175−0.270.79
Sex*Latency of Attack−0.1710.238−0.710.48
Sex*Rel Bold−0.0510.166−0.310.76
Mass*Latency of Attack0.0590.2590.230.82
Mass*Boldness0.0940.1860.490.63
Latency of Attack*Boldness−0.0060.196−0.030.98
Mass*Mass0.110.1930.570.57
Latency of Attack*Latency of Attack0.1210.2630.460.65
Boldness*Boldness0.0440.2010.220.83
Composite (Survival * Growth)
 Sex0.0420.1540.270.79
Mass−0.1130.169−0.670.51
Latency of Attack−0.4760.389−1.230.23
Boldness−0.0310.171−0.180.86
Sex*Mass−0.1150.155−0.740.46
Sex*Latency of Attack−0.1840.187−0.980.33
Sex*Rel Bold−0.1180.148−0.80.43
Mass*Latency of Attack−0.0790.171−0.460.65
Mass*Boldness0.0490.1450.340.73
Latency of Attack*Boldness0.0140.1820.080.94
Mass*Mass−0.0760.166−0.460.65
Latency of Attack*Latency of Attack0.2840.2261.260.22
Boldness*Boldness0.0790.1370.460.65

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

In this study, we tested for the presence of a behavioral syndrome in field-collected A. pennsylvanica of different instars and then compared these syndromes against those exhibited by laboratory-reared spiders from the same source population. We found that field-collected and laboratory-reared spiders exhibited divergent syndromes and that the behavioral syndrome of individuals in our mark–recapture study varied based on their developmental stage: penultimate spiders reared in the field exhibited a positive association between boldness and foraging aggressiveness, but juveniles collected from the same population lacked such an association. The syndrome exhibited by wild-caught penultimate A. pennsylvanica is similar to those observed in many other species of spider (Pruitt & Riechert 2012), where more aggressive individuals are also bolder. In contrast, laboratory-reared spiders never exhibited an association between boldness and foraging aggressiveness. Taken together, these findings suggest that the behavioral syndrome exhibited by field-collected A. pennsylvanica is the result of phenotypic plasticity, driven by environmental conditions which are present in the field but are lacking in the laboratory environment. Additionally, data from our mark–recapture experiment indicate spider mortality was low, and we failed to detect any evidence of correlated selection (Table 3): spiders experienced similar growth and survivorship regardless of their trait combinations (boldness and aggressive), and thus, the positive association between boldness and foraging aggressiveness that emerges in penultimate spiders does not appear to be drive by the ‘weeding out’ of individuals that fall off of this correlated axis. Notably, our ability to explore the influences of environmentally induced phenotypic plasticity vs. correlated selection in creating a behavioral syndrome in a wild population is a novel contribution to the behavioral syndromes literature; because, although there are numerous models of how behavioral syndromes could develop (McElreath & Strimling 2006; Luttbeg & Sih 2010; Stamps & Groothuis 2010b), remarkably little is known about how cross-contextual trait correlations are generated in wild populations.

Ontogenetic Shifts in Behavioral Type

Both mark–recapture and laboratory-reared spiders exhibited shifts in their behavioral tendencies over the course of their development. In laboratory-reared spiders, the majority of individuals exhibited a shift toward less aggressive, less bold behavior as they matured (Fig. 2). A similar pattern was observed for spiders in our mark–recapture study for foraging aggressiveness, but no significant shift was observed in boldness (Fig. 4). The fact that A. pennsylvanica generally exhibit less aggressive and less bold behavioral tendencies as their near maturity is consistent with asset protection theory (Clark 1993, 1994), where individuals that have greater resource holdings (i.e., those that have a lot to lose) are predicted to exhibit risk-averse behavioral tendencies. It is commonly proposed that individuals near maturity have less to gain from risky behavior (e.g., they have fewer instars remaining, the majority of lifelong energy intake has already transpired), and penultimate individuals certainly have more to lose, because they are nearer to the point of evolutionary payoff: reproduction (Pravosudov & Grubb 1998; Kotler et al. 2004). In contrast, the rewards of increased resource acquisition are likely greater for juveniles, because they need resources to complete their development sequence.

Shift in Syndrome Structure: Rearing Environment vs. Correlated Selection

Although numerous studies have demonstrated that behavioral syndromes vary across environments (Bell 2005; Dingemanse et al. 2007; Korpela et al. 2011), to date, very few studies have resolved whether population-specific syndromes are the result of phenotypic plasticity or correlated selection (Bell & Sih 2007). Population-level studies of three-spined and nine-spined sticklebacks have shown that populations with large fish predators typically exhibit behavioral syndromes between boldness and activity level, whereas populations that lack such predators typically lack such syndromes (Bell 2005; Dingemanse et al. 2007; Herczeg et al. 2009). Moreover, one can artificially induce behavioral syndromes in populations that normally lack them, merely by introducing large fish predators (Bell & Sih 2007). In these systems, the syndrome between boldness and activity level emerges as a result of two processes working in concert: correlated selection and phenotypic plasticity. In contrast, in our mark–recapture study in a wild population, we detected very low mortality, and there was no indication of correlated selection in this population. Yet, we observed a shift in the behavioral syndrome exhibited by A. pennsylvanica over the duration of our study, even when we restricted our analyses to only those spiders that persisted over the duration of our study: young spiderlings failed to exhibit a behavioral syndrome between foraging aggressiveness and boldness, whereas penultimate individuals exhibited a positive association—but only when spiders were raised in the field. Thus, the data available suggest that the observed boldness–aggressiveness syndrome (1) emerges only when individuals develop under certain environmental conditions (i.e., it exhibits plasticity), and (2), is not associated with a signature of contemporary correlated selection. Notably, our results must be taken with the caveat that we presently lack data on early-life selective pressures in A. pennsylvanica in the field (i.e., prior to the third instar), which could have weeded out individuals which were not predisposed to undergo the observed developmental shifts in behavior that eventually form the boldness–aggressiveness syndrome of field-reared spiders. Such selective pressures would likely be absent in our laboratory population, because survivorship to the third instar is likely much greater in laboratory.

Although our results are consistent with the hypothesis that the boldness–aggressiveness syndrome of A. pennsylvanica is driven by environmentally induced phenotypic plasticity over the course of individuals' development, our data cannot yet implicate the specific aspects of the environment that could drive the observed trend. Population density (Korpela et al. 2011), presence of predatory cues (Bell & Sih 2007; Niemela et al. 2012), sexual signals (Bailey & Zuk 2008; DiRienzo et al. 2012), and numerous other aspects of the environment could be responsible for the divergent syndromes of laboratory- vs. field-reared penultimate A. pennsylvanica. Ongoing work is aimed at deciphering the relative contributions of population density and predator cues using (1) variation among sites in population density and intensity of predation, and (2) laboratory experiments where we are manipulating specific aspects of the developmental environment. Although more work is required to identify the precise causes of the ontogenetically- and environmentally contingent syndrome observe here in A. pennsylvanica, our work adds to the growing body of literature implicating plasticity a possible driver of behavioral consistent and population-level divergence in behavioral syndromes. Deciphering (1) when personality traits are sensitive to environmental cues, and (2), the types of cues that beget particular trait combinations remain a notable challenge for future behavioral syndromes research.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

This research was funded by the University of Pittsburgh's Department of Biological Sciences and by several undergraduate research grants awarded by the University of Pittsburgh's Honors College and the School of Arts and Sciences (awarded to KS, DRM, AWB).

Literature Cited

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited