Behavioral Syndrome Over the Boundaries of Life—Carryovers from Larvae to Adult Damselfly

Behavioral Ecology doi:10.1093/beheco/arn111 Advance Access publication 11 September 2008 Behavioral syndrome over the boundaries of life—carryovers from larvae to adult damselﬂy

Tomas Brodina,b aDepartment of Ecology and Environmental Science, Umea˚ University, S-90187 Umea˚, Sweden and bDepartment of Environmental Science and Policy, UC Davis, One Shields Avenue, 95616 Davis, CA, USA Downloaded from https://academic.oup.com/beheco/article/20/1/30/212812 by guest on 01 October 2021

Activity is an important behavioral trait that mediates a trade-off between obtaining food for growth and avoiding predation. Active individuals usually experience a higher encounter rate with food items and suffer higher predation pressure than less active individuals. I investigated how activity of the damselfly Lestes congener is affected by larval state and predator presence and if larval behavioral type (BT) can be used to predict larval boldness, foraging success, and adult BT. Activity level of individual larvae was studied without predator at 2 different physiological states (hungry and fed) and in 2 predator treatments: familiar predator cues and unfamiliar predator cues. Larvae did not adjust their activity depending on state or when subjected to unfamiliar predator cues, but a general reduction in activity was seen in the familiar predator treatment. Hence, active individuals remained active compared with their conspecifics, independent of state or predator treatment. Active individuals were also bolder and more efficient foragers than their less active conspecifics. Furthermore, both adult activity and boldness were correlated with larval BT. The results illustrate that BT of a larvae is carried over many different situations keeping active larvae active even in maladaptive situations, demonstrating how a behavioral syndrome may constrain behavioral plasticity. Furthermore, results showed that behavioral syndromes can carry over from larvae through metamorphosis and dictate the BT of the adult. Key words: activity, behavioral syndrome, behavioral type, boldness, Lestes, phenotypic plasticity. [Behav Ecol 20:30–37 (2009)]

ctivity is an important behavioral trait that influences food Bell, Johnson, and Ziemba 2004) or animal personalities (Dall Aintake and thereby energy gain. Higher activity usually et al. 2004; Groothuis and Carere 2005). A behavioral syn- results in more encounters with food and hence has the poten- drome exists if, even as individuals alter their behavior de- tial to result in higher energy gain that translates to higher pending on the context, some are consistently more active growth rate and larger size and/or shorter development than others across multiple contexts (e.g., in how they feed (Werner and Anholt 1993). However, a high activity also im- and cope with predators, but also in contests with conspe- plies a higher encounter rate with predators. Hence, individ- cifics, in how they mate and disperse) (Sih et al. 2003; Sih, uals are faced with a growth/predation risk trade-off (Lima Bell, and Johnson 2004; Sih, Bell, Johnson, and Ziemba 2004). 1998). Two types of situations can generate the growth/pre- Over the last few years, several evolutionary and/or ecological dation risk trade-off (Sih et al. 2003). The first situation is studies on animal personalities have been produced on mam- termed ‘‘within-situation conflicts’’ and is often associated mals (Reale et al. 2000), birds (Sol and Lefebvre 2000; Sol with a time budget conflict. Here individuals that spend more et al. 2002; Both et al. 2005; Carere et al. 2005; Mettke- time foraging actively gain more food and probably grow fast- Hofmann 2005; Quinn and Cresswell 2005; van Oers et al. er but suffer a higher predation risk. Active individuals may do 2005), lizards (Sinervo 2001; Stapley and Keogh 2005; Dame well in the absence of predators but suffer in the presence of and Petren 2006), salamanders (Sih et al. 2003), fish (Fraser predators. In contrast, the less active individuals should do et al. 2001; Overli et al. 2004; Bell 2005; Brown et al. 2005; well in the presence of predators, but poorly in the absence Magnhagen and Staffan 2005; Overli et al. 2006), spiders of predators. Continuing in this line of reasoning, one would (Johnson and Sih 2005), and insects (Hedrick 2000; Brodin expect it to be advantageous for animals to have uncon- and Johansson 2004; Slos and Stoks 2006). It has become clear strained plastic behavior, that is, being able to optimize behav- that consistent individual differences in behavior exist in ior according to the current environment to maximize a wide range of contexts and organisms from humans to in- foraging while minimizing predation risk. Yet, this is not the sects (Sih, Bell, Johnson, and Ziemba 2004). Note that if all norm, instead constrained behavioral plasticity is the dominat- personalities do well in some situations, but not others (i.e., ing pattern. One reason for such constraints is the other type no one is optimal everywhere), this trade-off can help to ex- of conflicts, termed ‘‘across-situation conflicts.’’ It involves plain the maintenance of variation in individual behavior. suites of correlated behaviors called behavioral syndromes Correlations among behaviors can also have important eco- (or personalities) that make individuals differ consistently in logical effects. For example, the possibility that dispersers their behavioral tendencies, and the behavior in one context tend to be bold and aggressive has potentially important ef- is correlated with behavior in many other contexts (Sih, Bell, fects on the dynamics of biological invasions and on metapo- Johnson, and Ziemba 2004). Such suites of correlated behav- pulation dynamics. iors are, in ecology, often called behavioral syndromes (Sih, Despite the surge of interest in this field, no study has yet investigated if personality carries over from early life stages Address correspondence to T. Brodin. E-mail: tomas.brodin@emg. in one environment (e.g., larvae) through a metamorphosis umu.se. into adulthood in a different environment. Intuitively, one Received 24 July 2007; revised 14 July 2008; accepted 30 could suggest at least 2 scenarios of how larval and adult behav- July 2008. iors are connected. First, because several behavioral traits are inherited, it could be expected that adults would adopt The Author 2008. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: [email protected] Brodin • Behavioral syndrome over the boundaries 31 a similar behavioral type (BT) as when they were larvae. This predator that coexists with Lestes in lakes and ponds in Pope seems like a plausible prediction, and similar results of behav- Valley (McCauley S, Brodin T, unpublished data). Tiger sala- ioral carryover from juvenile to adult have been shown in other mander, however, was labeled as ‘‘unfamiliar predator’’ be- organisms, such as spiders (Johnson and Sih 2005). Further- cause the species do not coexist with Lestes in the area. more, because behavioral traits have been shown to be in- In late April, 30 larvae (instar F-3, where F-0 denotes last inherited (Magurran 1990; Maurer and Sih 1996; O’Steen star, F-1 second last instar, and so on) were collected from et al. 2002; Drent et al. 2003; Sih et al. 2003), juvenile behavior a small fishless pond, Green pond, in the Wantrup Wildlife Re- should give an indication of what adult BTs that were prosper- serve, Pope Valley, CA, approximately 80 km northwest of ous last generation. Hence, due to the above water surface Davis. The dominant predators in Green pond are large drag- selection pressure on adults, larval behavior might be more onfly larvae (Aeshna), large dytiscid larvae, and belostomatids adapted to the above water surface selection pressure than to (McCauley S, Hammond J, Brodin T, unpublished data). The the selection pressures in the larval environment. However, Lestes larvae were kept in separate plastic containers (25 3 15 3 there are few behavioral transformations as dramatic as those 8 cm) during the entire study and fed approximately 30 observed during insect metamorphosis, when the neural and daphnia from laboratory cultures every second day. This food Downloaded from https://academic.oup.com/beheco/article/20/1/30/212812 by guest on 01 October 2021 motor systems are remodelled (Consoulas et al. 2000). For provisioning corresponds with an ad libitum approach. Re- example, in aquatic insects, typical larval behaviors are crawl- maining daphnia were removed prior to every feeding bout ing, feeding, and swimming, whereas adult behaviors include to eliminate the risk of unequal prey densities between con- flying and reproducing. The alternative scenario would be tainers. Late-instar Aeshna larvae were collected at Bear Creek, that the remodelling of the neural and motor systems could approximately 115 km northwest of Davis, and salamanders facilitate a decoupling of the larval and adult behavior, freeing were taken from a laboratory stock at UC Davis. The predators the adult from any constraints that a behavioral carryover were all fed either larval Chironomids (Diptera: Chironomidae) from larvae could impose (Moran 1994). It has been sug- or Pacific tree frog, Pseudacris regilla, tadpoles (1–3) daily. gested that selection should uncouple a behavioral syndrome through ontogeny when environmental conditions experi- Larval behavioral experiment enced by juveniles differ substantially from those experienced by adults (Sih, Bell, Johnson, and Ziemba 2004). This issue is, To determine individual BT of each larva, I ran several behav- to my knowledge, addressed for the first time in this paper. ioral trials where I monitored activity level of the larvae, in the In this study, I focus on individual activity correlations be- absence of food, at different physiological states and in several tween different states and predator environments and how this predator environments. The trials were all carried out between behavioral syndrome affects boldness and foraging success in 1230 and 1450 h on 10, 12, 15, and 16 May 2007. Each con- damselfly larvae. Furthermore, I investigate how these larval tainer (25 3 15 3 8 cm) was filled with 1.2 L aged well water BTs are pronounced in the adult damselfly. Damselflies have and had a coordinate grid drawn on the bottom with a grid size complex life cycles with an aquatic larval stage and a terrestrial of 1 3 1 cm. The larvae had all grown to instars F-3 or F-2 when adult stage wherein reproduction occurs. Because reproduc- used in the behavioral trials. I observed the containers once tive success in the adult stage depends on time of emergence every 10 min, for 140 min, and noted whether the larvae (De Block and Stoks 2005), fat content (Plaistow and Siva- had moved or not. A move was noted when the larvae had Jothy 1996, 1999), and adult size (Anholt et al. 1991; Sokolovska moved their head from one square to another. This procedure et al. 2000; De Block and Stoks 2005), activity in the larval generated an activity score for each larva ranging from 0 (inac- stage has the potential to affect fitness parameters in the adult tive) to 14 (very active). The first 2 trials focused on larval ac- stage. Based on current empirical knowledge and theory pre- tivity at different states and differed in larval hunger level. In sented above, I predict that 1) larvae will be more active when the first trial, larvae were hungry (starved for 5 days) when ob- starved than when well fed, 2) individuals should show behav- served, whereas in the second experiment, they were well fed ioral adjustments according to the growth/predation risk (fed the same morning). The nonpredator trials were run first trade-off (the within-situation conflict), 3) highly active indi- to avoid any risk of long-lasting antipredator response affecting viduals should stay relatively active (compared with less active subsequent experiments. In the third and fourth trials, well-fed ones) independent of state or predator environment, 4) active larvae were studied in the presence of nonlethal predator cues individuals should be bolder and forage more efficiently than from an unfamiliar salamander predator (tiger salamander) less active ones, and 5) larval BT will be positively correlated and a familiar dragonfly predator (Aeshna), respectively. with adult BT. One hour prior to both predator experiments, chemical cues (100 mL of water from predator containers) were added to prey containers. After each predator trial, prey containers MATERIALS AND METHODS were emptied and cleaned and aged well water was added to All experiments were run at the Aquatic Biology and Environ- minimize the risk of future behavioral antipredator responses mental Science facility on the University of California, Davis or stress caused by predator cues. All larvae that survived to (USA) campus. Experimental rooms were kept constant the 10th of May were used in the behavioral trial, adding up to at 22 C and at a light:dark regime of 14:10 h. My target spe- a total of 26 individuals. During the behavioral trials, 4 of the cies was the damselfly Lestes congener (Hagen) (Odonata: larvae moulted and were removed from the following trials Zygoptera), hereafter called Lestes. It is a very widespread spe- making the total number of studied larvae 22. After the initial cies with transcontinental range that is commonly found in activity study (starved larvae without predator cues), larvae ponds without fish (Stoks and McPeek 2003). This Lestes spe- were assigned to either of 2 BTs. The highly active BT had cies oviposits endophytically in the late summer and fall and initial activity scores of 10–14 (n ¼ 12), whereas the less active overwinters as eggs (Walker 1953; Sawchyn and Gillott 1974). BT had activity scores ranging from 0 to 7 (n ¼ 10); no larvae Eggs hatch in spring, and larvae grow and mature to breed in this experiment got activity scores of 8 or 9. This categori- during the flight period that, in the study area, begins in early zation was supported by a stepwise classical linear discrimi- June (Brodin T, personal observation). nant analysis (Wilks’s lambda ¼ 0.098, F ¼ 185.12, P tail , I used Aeshna larvae (Odonata: Anisoptera) and tiger sala- 0.0001). Furthermore, the categorization of BT was also mander, Ambystoma tigrinum, as predators. The Aeshna was la- supported by a stepwise classical linear discriminant analysis beled as ‘‘familiar predator’’ because Aeshna larva is a common of the principal components analysis (PCA) scores from all 32 Behavioral Ecology

4 behavioral trials (Wilk’s lambda ¼ 0.071, F ¼ 216.32, P tail , performing statistical tests on component scores rather than 0.0001). many different variables, problems of multiple comparisons In addition to the activity studies, I also measured individual can often be avoided. PC1 scores were not normally dis- larval boldness and foraging success. The boldness experiment tributed and were hence analyzed with Kruskal–Wallis test, was carried out after the initial activity assay but before the whereas the PC2 scores could be analyzed with a 1-way other 3 activity trials were done. Larvae were disturbed (touch- ANOVA. To ensure statistical accuracy, I also rank transformed ing larval lamellae with a prod) until they reacted with an PC1 scores and used them in a 1-way ANOVA that supported escape response and swam away. Time until larvae started the result from the nonparametric test. These monitored moving again after the initial swimming burst was measured. adult behaviors represent general activity and dispersal behav- After 2 min, the observation was terminated and any larvae iors related to foraging and antipredator behavior (Corbet that had not moved were assigned the maximum value of 120 1999). Admittedly, scoring adult behavioral data in an aquar- s. As suggested by Reale et al. (2007), boldness trials were ium is somewhat artificial compared with field observations carried out in the, for each larvae, familiar environment of and may increase the proportion of uncommon behaviors their holding container. This was done to reduce potential (i.e., walking). However, numbers of walking bouts recorded Downloaded from https://academic.oup.com/beheco/article/20/1/30/212812 by guest on 01 October 2021 effects that other behaviors (e.g., exploratory behavior) might here were still very few compared with other monitored behave on boldness. Boldness has repeatedly been reported of haviors. The difficulties of making behavioral measurements being positively correlated with activity level, and observing on adult damselflies, particularly repeated measurements on behavior after disturbance is a commonly used method to a single individual, argued for the use of a more controlled quantify the trait (Johnson and Sih 2005). However unlikely, but artificial context for measuring behavior. Future work I cannot exclude the possibility that a similar pattern of larval should certainly follow this up with observations in increas- boldness as shown here could be generated as a side effect of ingly realistic habitat settings. Adult boldness was analyzed individual activity because active larvae move more frequently with 1-way ANOVA and linear regression. All statistical analy- than less active larvae. In the foraging experiment, I added 30 ses were done using the statistical package SYSTAT 11 (2004). daphnia in each larval container and observed individual larva Larval activity (BT), foraging success, and boldness could be for 120 s while noting how many daphnia the larva captured. confounded by 3 factors: damselfly larvae size, time to emer- A Shapiro–Wilk normality test revealed that larval activity, gence, and damselfly gender. However, linear regressions be- boldness, and foraging data did not fit to a normal distribution. tween these behaviors and size and these behaviors and Such lack of normality is not unusual in behavioral data, espe- time to emergence showed no significant relationships (P . cially when sample sizes are fairly low. Boldness was ln trans- 0.45 in all regressions). Female and male damselflies did not formed to meet the assumptions of normal distribution, differ in activity, foraging success, or boldness (Kruskal–Wallis; whereas activity and foraging data were analyzed with Kruskal– all P . 0.23). The potential for adult size confounding the Wallis 1-way analysis of variance (ANOVA). In addition, to check behavior of the adults was also tested for and subsequently for differences in activity between trials, a Friedman 2-way rejected (linear regression; P . 0.18). ANOVA was used; it is a nonparametric analogue of repeated This research adhered to the Association for the Study of measures analysis. Animal Behavior/Animal Behavior Society Guidelines for the Use of Animals in Research, the legal requirements of USA, and all institutional guidelines. Adult behavioral experiment Sixteen adults successfully emerged between 2 June and 16 June 2007. After emerging, they were moved to individual glass RESULTS aquaria (75 3 30 3 30 cm) containing a wooden structure Larval activity (diameter ¼ 8mm,length¼ 50 cm) for perching, water, and potential prey (20 Drosophila sp.). Adults stayed in their aquaria Larvae of both the highly active BT (mean 6 standard error for 24 h before the behavioral trial began. Three different be- [SE] ¼ 13.08 6 0.23, starved) and the less active BT (mean 6 havioral measurements were made: 1) activity scores (0–14), SE ¼ 4.5 6 0.64, starved) all retained their respective BT in- using the same technique as for larvae (see above); 2) behav- dependent of hunger level or presence of predator cues and ioral profile, all adult damselflies were observed for 20 min and their activity levels consistently differed accordingly (Figure 1; every flying, walking, and ‘‘fluttering’’ behavior was noted; and Kruskal–Wallis test, all P , 0.001). This activity difference 3) boldness, perching adults were approached by a bird decoy between larvae was consistent even when larvae reduced ac- until they flew, the decoy was removed, and time until the adult tivity in response to the presence of Aeshna cues. A repeated returned to the perch was noted. Adults respond to risky situa- measures test confirmed that larval activity differed between tions with active flight behavior to escape risk, whereas larvae the behavioral trials (Friedman 2-way ANOVA, P ¼ 0.004). use inactivity to escape risk (see references in Corbet 1999). The differences found between trials were accounted for by the Hence, to get adequate measures of boldness for larvae activity reduction in the presence of Aeshna cues (Figure 1; (latency to move) and adults (latency to return to a perch), Kruskal–Wallis test, P ¼ 0.038), indicating a behavioral anti- different methods that are appropriate to the behaviors of the predator reaction to the familiar predator. Contrastingly, no respective life stage had to be used. Boldness is distinct from behavioral antipredator response was seen in the experiment activity because it reflects the time required after a disturbance when larvae were subjected to Ambystoma cues (Kruskal–Wallis to return to a potentially risky behavioral pattern following test, P ¼ 0.29). the behavior typically adopted when the risk of predation is encountered. The fluttering behavior was scored when a dam- Boldness and foraging success selfly fluttered its wings while remaining stationary. Activity scores turned out to be too crude a measurement of adult Larvae of the highly active BT were bolder than the less active activity because all but one adult got scores between 10 and larvae (mean 6 SE ¼ 38.5 6 11.1 and 81.2 6 13.1, respectively; 14, and hence, these data were not used in any analysis. In 1-way ANOVA, F1,20 ¼ 6.31, P ¼ 0.021). Active larvae also order to summarize adult behavioral profile, I used PCA on foraged more successfully than less active larvae (mean 6 a correlation matrix. There are advantages to use component SE ¼ 1.5 6 0.5 and 0.3 6 0.1, respectively; Kruskal–Wallis test, scores rather than single behaviors in analyses. Foremost, by P ¼ 0.035). All in accordance with my predictions based on Brodin • Behavioral syndrome over the boundaries 33 Downloaded from https://academic.oup.com/beheco/article/20/1/30/212812 by guest on 01 October 2021

Figure 1 Figure 2 Mean activity of larval BTs in 4 different environments. Low Active larvae become active adults. PCA score of larval activity is state—starved larvae, high state—well-fed larvae, unfamiliar positively correlated with the PCA score of the adult behavioral predator—salamander cues, and familiar predator—dragonfly cues. profile. Open circles denoted highly active BTs and filled circles less Open circles denoted highly active BTs and filled circles less active active BTs. BTs. Error bar denotes 61 SE. that, one benefit of being active is found to be an increase in DISCUSSION prey encounter and hence more opportunities to forage. My study is a longitudinal behavioral study that follows individual behavior of damselfly larva over multiple physiological Adult behavior states and in several predator environments and ultimately Adult behavior was monitored in 16 out of the 22 individuals examines how larval behavior is translated into adult behavior. because 6 larvae were unable to emerge successfully. I used I focus on how individual activity level responds to changes in a PCA to compress the adult behavioral data collected into the environment and how activity correlates to larval boldness 2 principal components that together explained 90.6% of and foraging success. In addition, I follow each larva into the variance (Table 1). PC1 explained 56.6% of the variation adulthood and survey their behavior as adults to see if behav- and was highly positively correlated with walking and flutter- ioral syndromes may cross the barrier of metamorphosis and ing. PC2 explained 34% of the variation and was strongly carry over from the aquatic larvae to the terrestrial adult. positively associated with flying. Larvae of the active BT The study shows that Lestes individuals differ consistently in emerged as more active adults than the less active larvae, a re- their activity across several situations and life stages and that sult supported by both PC1 (Kruskal–Wallis, P ¼ 0.009; 1-way more active larva are bolder and better foragers than less active larvae. This clearly illustrates the existence of an ANOVA, F1,14 ¼ 13.22, P ¼ 0.004, on ranked data) and PC2 (1- way ANOVA, F1,14 ¼ 5.92, P ¼ 0.029). In addition, a linear regression between PCA scores of larval activity and adult behavioral profile showed a strong positive correlation between active larvae and active adults (Figure 2; linear regression, F1,14 ¼ 12.13, P ¼ 0.004). Furthermore, was adult boldness positively correlated to larval boldness (Figure 3; linear regression, F1,14 ¼ 6.46, P ¼ 0.011); hence, bold/shy larvae gave rise to bold/shy adults (mean 6 SE ¼ 2.68 6 0.16 and 4.17 6 0.39, respectively) and the BTs differed accordingly (1-way ANOVA, F1,14 ¼ 11.38, P ¼ 0.005). These results are intriguing and clearly illustrate that BT is carried over from larvae, through a metamorphosis, to the adult damselfly.

Table 1 Results of PCAs on adult behavior

Behavior\loading PC1 PC2

Adult activity Walking 0.928 0.027 Figure 3 Fluttering 0.850 20.401 Bold larvae become bold adults. Individual boldness declines with Flying 0.341 0.927 increasing time on both axes because low time scores indicate high Cumulative variance explained 56.65% 34.00% boldness. Open circles denoted highly active BTs and filled circles less active BTs. 34 Behavioral Ecology activity–boldness behavioral syndrome in Lestes that not only and McPeek (2003) found. This lack of response to Ambystoma carry over larval situations but also carry over into the behav- cues might seem maladaptive; however, Lestes coexist with ior of the adults. The link between individual larval and adult Aeshna larvae and not with Ambystoma and accordingly reacted behavior in an organism that metamorphosis from one dis- strongly only to the presence of dragonflies by decreasing tinct larval habitat type to a completely different adult habitat activity. These predator results are interesting and offer empiri- shown here is novel and may have important implications for cal support for a model presented by McElreath and Strimling dispersal, metapopulation, and fitness issues. I found a positive (2006). Their model predicts that if information about the correlation between individual activity in different situations, immediate environment is uncertain, prey should continue illustrating a behavioral syndrome, and I also found a positive to behave the same way as before to avoid the risk of making correlation between activity/boldness and activity/foraging a mistake. Because the larvae never encountered cues from the rate, the latter illustrating a well-known benefit of being ac- unfamiliar predator, they might perceive the environment con- tive. These benefits may have important consequences for taining the cues as uncertain and to avoid making the wrong damselfly fitness because they affect size and time of emer- behavioral adjustment they keep behaving as before. gence, variables known to be important components of fitness My study shows consistently significant differences in activity Downloaded from https://academic.oup.com/beheco/article/20/1/30/212812 by guest on 01 October 2021 in these organisms (Fincke 1982; Banks and Thompson 1987; between larval BTs in 2 different predator treatments. The dif- De Block and Stoks 2005; however, see Fincke and Hadrys ferences in activity between larvae may cause the active larvae 2001). to grow faster than less active larvae, which in the absence of Individual activity level seemed unaffected by physiological predators would give the active larvae a fitness advantage. How- state of the larvae. This result was surprising and does not fol- ever, I found no correlation between activity and size or activity low the empirical and theoretical predictions of low state–high and development time. One plausible explanation for the lack activity and high state–low activity (Godin and Smith 1988; of apparent benefits of being active is that the larvae were fed Godin and Crossman 1994; Godfrey and Bryant 2000). One ad libitum. This implies that high food availability might over- explanation for this lack of state-dependent activity could be ride potential effects on growth and development that differ- that larvae, in the absence of predators, take advantage of the ences in activity otherwise would lead to. In the presence of predator-free environment and forage as active as they can predators, larvae from the active BTwill, due to higher encoun- independent of state. Another possible explanation for this ter rates with the predator, be subjected to an increased pre- result may be that larvae were foraging actively as a response dation pressure that might reverse the fitness advantage in to the risk of desiccation of Green pond. Response to pool favor of larvae from the less active BTs. Thus, larvae showed desiccation is common among amphibians (Laurila and a behavioral correlation across situations that Sih et al. Kujasalo 1999; Johansson et al. 2005) and increase of activity (2003) termed an across-situation conflict. The existence of has been shown in response to time constraints in L. congener interspecific variation in important traits like activity has long (Johansson and Rowe 1999). Hence, the generally high activ- been known and thoroughly studied. It has often been dem- ity caused by a time constraint is a very plausible explanation onstrated that the same traits that allow a species to be suc- here because Green pond dried up about 1 month after the cessful in some environments can exclude the species from larvae were collected. However, if all larvae were foraging (rel- other ones, leading to the development of distinct community atively) actively, this implies that the less active BTs were be- types across environmental gradients (e.g., McPeek 1990b; haviorally inhibited and that their activity-related plasticity was Schluter 1995; Tessier et al. 2000; Wellborn 2002; Stoks and constrained by a behavioral syndrome. McPeek 2003). Larvae reduced their activity in the presence of cues from The persistence of consistent behavioral variation between a familiar predator. Previous studies have shown that when pre- individuals in a population has long been an unexplained phe- dation risk increases, prey usually decrease the time spent ac- nomenon. Some have even claimed it to be an artifact because tive, leading to a lower encounter rate with predators and natural selection will favor individuals that do best and remove hence a lower mortality rate than for more active individuals less fit individuals from the population. Evolutionary theory (Lima and Dill 1990; Anholt and Werner 1998; Brodin and predicts that variation in traits that are connected to survival Johansson 2004). As a result of low activity, encounter rates or reproduction should, eventually, be lost from a population. with potential prey are also lower and hence a decreased However, several studies have shown that many behavioral growth rate that ultimately translates into a smaller size traits are both heritable (e.g., Magurran 1990; Maurer and and/or a longer time to maturation (Werner and Anholt Sih 1996; O’Steen et al. 2002; Drent et al. 2003; Sih et al. 1993; Anholt and Werner 1998; Brodin and Johansson 2003) and connected to fitness (Dingemanse et al. 2004; Both 2004). Any reduction in larval energy gain is predicted to in- et al. 2005). Recently Wolf et al. (2007) presented a model voke a negative influence on individual lifetime reproductive that uses the negative feedback of asset protection (Clark output. Despite this, I find a reduction in activity (i.e., energy 1994) to illustrate one possible mechanism for why evolution gain) in the presence of familiar predator cues. This reaction favors animal personalities (behavioral syndromes). They to predator presence illustrates an important trade-off that show that animals that have more to lose should be less bold most organisms face: ‘‘the growth/predation risk trade-off’’ and less aggressive rather than risk losing their assets. In con- (Werner and Anholt 1993; Lima 1998) or in terms of Sih trast, animals with little to lose should be bolder and more et al. (2003) the within-situation conflict. Similar decreases aggressive, and as long as assets do not change, animals in activity in the presence of a perceived predator have been should maintain a stable personality. This is an intriguing reported in Lestes spp. (Jeffries 1990; Stoks 1998; Stoks and and counterintuitive thought, whose generality already has Johansson 2000; Johansson et al. 2001; Stoks and McPeek been questioned and debated, primarily because the negative 2003) and other damselflies (Pierce et al. 1985; McPeek feedback mechanism can produce animal personality only un- 1990a; Stoks et al. 2003; Brodin et al. 2006) and are generally der restrictive circumstances (McElreath et al. 2007). A possi- expressed by many other organisms (review in Lima and Dill bly more general way of explaining behavioral consistency and 1990; Lima 1998). When larvae were subjected to familiar correlations is through positive feedback (McElreath et al. Aeshna cues, they reacted by decreasing activity, whereas in 2007). For example, if thorough explorers gain assets (energy, the presence of cues from the unfamiliar Ambystoma, larva size, and knowledge) that improve their abilities to escape did not change activity. These results are convergent with predators or win fights, then one might see positive correla- the lack of response to a non-coexisting predator that Stoks tions among exploration, boldness, and aggressiveness. Brodin • Behavioral syndrome over the boundaries 35

Additional behavior would have positive feedback on state, data of behavioral carryovers from one habitat, with certain se- maintaining differences in assets and BTs. lection pressures and constraints, to a totally different habitat Behavioral correlations may differ in stability depending on with different selection pressures completely uncoupled from what mechanism that underlies the correlation. Individual dif- the larval environment. An interesting continuation of this ferences in single traits of animal personalities (e.g., activity) work would be to compare survival and dispersal tendencies are usually moderately heritable and relatively stable over an of adults emerging from highly active and less active larvae individual’s life (Koolhaas et al. 1999; Bouchard and Loehlin in different predator environments. I predict that highly active 2001). Most stable over time (many generations) are pheno- larvae will emerge, mainly from low predator pressure sites, as typic correlations caused by strong genetic correlations (van highly active adults with high tendencies to disperse from their Oers et al. 2004; Bell 2005), such as pleiotropy, where the suitable larval sites. Contrastingly, less active larvae will emerge, correlated behaviors (e.g., boldness and activity) are governed mainly from high predator pressure sites, as less active adults by a shared genetic mechanism. Evidence for strong genetic with low tendencies to disperse. correlation between behavioral traits suggests that behavioral traits are often structured in personality traits because they are Downloaded from https://academic.oup.com/beheco/article/20/1/30/212812 by guest on 01 October 2021 controlled by the same hormones (Ketterson and Nolan 1999; Koolhaas et al. 1999) or genes (Sih, Bell, and Johnson 2004; FUNDING Sih, Bell, Johnson, and Ziemba 2004). Therefore, may com- Swedish Research Council Formas (2006-655) to T.B. ponents of personality be difficult to decouple (Ketterson and Nolan 1999). Another mechanism generating behavioral cor- Thanks to Andy Sih for great support and to Frank Johansson, Rob relations is linkage disequilibrium caused by correlational Brooks, and 2 anonymous reviewers for comments on previous versions selection. Correlational selection is selection for optimal trait of the manuscript. I am also grateful to the Land Trust of Napa Valley combinations (Barton and Turelli 1991), and theory predicts and to Shannon McCauley, John Hammond, and Joe Callizo. that correlational selection will eventually produce an adaptive genetic correlation between traits because alleles influencing one trait will be coinherited together with alleles influencing the other trait (Jones et al. 2003). The correlation REFERENCES between boldness and activity shown in this study could po- Anholt BR, Marden JH, Jenkins DM. 1991. Patterns of mass gain and tentially be explained by correlational selection favoring this sexual dimorphism in adult dragonflies (Insecta, Odonata). Can J particular combination of traits. Zool. 69:1156–1163. In this study, I show for the first time that larval behavior in Anholt BR, Werner EE. 1998. Predictable changes in predation mor- one environment may carry over metamorphosis into adult be- tality as a consequence of changes in food availability and predation havior in a completely different environment. Active and bold risk. Evol Ecol. 12:729–738. larvae gave rise to active and bold adults, illustrating a behav- Banks MJ, Thompson DJ. 1987. Lifetime reproductive success of ioral syndrome from an aquatic life stage to a terrestrial life females of the damselfly Coenagrion puella.JAnimEcol.56: stage. This novel result is very intriguing because it has been 815–832. Barton NH, Turelli M. 1991. Natural and sexual selection on many suggested that selection should uncouple a behavioral syn- loci. Genetics. 127:229–255. drome through ontogeny when environmental conditions ex- Bell AM. 2005. Behavioral differences between individuals and two perienced by juveniles differ substantially from those populations of stickleback (Gasterosteus aculeatus). J Evol Biol. 18: experienced by adults (Sih, Bell, Johnson, and Ziemba 464–473. 2004). For Lestes, it is highly likely that the selection pressures Both C, Dingemanse NJ, Drent PJ, Tinbergen JM. 2005. Pairs of ex- under and above the water surface differs. For example, treme avian personalities have highest reproductive success. J Anim a highly active larva will be favored in a low predation site, Ecol. 74:667–674. but a highly active adult might be more easily spotted by Bouchard TJ, Loehlin JC. 2001. Genes, evolution and personality. a predatory bird or will have a higher probability of ending Behav Genet. 31:243–273. its life in a spiderweb. Yet, the (potentially maladaptive) selec- Brodin T, Johansson F. 2004. Conflicting selection pressures on the growth/predation risk trade-off in a damselfly. Ecology. 85: tion pressure that the aquatic larva experienced dictates the 2927–2932. BT of the adult. Brodin T, Mikolajewski DJ, Johansson F. 2006. Behavioural and life Another interesting implication of this study is the possibil- history effects of predator diet cues during ontogeny in damselfly ity that larval behavior might govern adult dispersal. My larvae. Oecologia. 148:162–169. results suggest that less active larvae (commonly found in Brown C, Jones F, Braithwaite V. 2005. In situ examination of bold- sites with high predator pressure) will emerge as less active ness-shyness traits in the tropical poeciliid, Brachyraphis episcopi. and shy adults, whereas active larvae (commonly found in Anim Behav. 70:1003–1009. sites with low predation pressure) will emerge as active and Carere C, Drent PJ, Privitera L, Koolhaas JM, Groothuis TGG. 2005. bold adults. Following the reasoning of current knowledge Personalities in great tits, Parus major: stability and consistency. Anim Behav. 70:795–805. of behavioral syndromes and dispersal, the less active and Clark CW. 1994. Antipredator behaviour and the asset-protection prin- shyeradultswillhavelesstendencies to disperse compared ciple. Behav Ecol. 5:159–170. with highly active and bold adults (Dingemanse et al. Consoulas C, Duch C, Bayline RJ, Levine RB. 2000. Behavioural trans- 2003). Counterintuitively, this will lead to a high degree of formations during metamorphosis: remodeling of neural and mo- dispersal from sites with low predation pressure and low de- tor systems. Brain Res Bull. 53:571–583. gree of dispersal from high predation sites. It can be viewed Corbet PS. 1999. Dragonflies behavior and ecology of odonata. Ithaca as if predators, by eating active individuals, are locking the (NY): Cornell University Press. next generation of prey to their site by indirectly restricting Dall SRX, Houston AI, McNamara JM. 2004. The behavioural ecology prey dispersal. of personality: consistent individual differences from an adaptive perspective. Ecol Lett. 7:734–739. My study contributes to a general growing appreciation of Dame EA, Petren K. 2006. Behavioural mechanisms of invasion and the existence and importance of behavioral syndromes (e.g., displacement in Pacific island geckos (Hemidactylus). Anim Behav. in aggressiveness, boldness, or general activity level) that carry 71:1165–1173. over across multiple contexts (e.g., foraging, predator avoid- De Block M, Stoks R. 2005. Fitness effects from egg to reproduction: ance, mating, dispersal, etc.). In addition, I here present novel bridging the life history transition. Ecology. 86:185–197. 36 Behavioral Ecology

Dingemanse NJ, Both C, Drent PJ, Tinbergen JM. 2004. Fitness con- McElreath R, Strimling P. 2006. How noisy information and individual sequences of avian personalities in a fluctuating environment. Proc asymmetries can make ‘personality’ an adaptation: a simple model. R Soc Lond B Biol Sci. 271:847–852. Anim Behav. 72:1135–1139. Dingemanse NJ, Both C, van Noordwijk AJ, Rutten AL, Drent PJ. 2003. McPeek MA. 1990a. Behavioral differences between Enallagma species Natal dispersal and personalities in great tits (Parus major). Proc R (Odonata) influencing differential vulnerability to predators. Ecol- Soc Lond B Biol Sci. 270:741–747. ogy. 71:1714–1726. Drent PJ, van Oers K, van Noordwijk AJ. 2003. Realized heritability of McPeek MA. 1990b. Determination of species composition in the personalities in the great tit (Parus major). Proc R Soc Lond B Biol Enallagma damselfly assemblages of permanent lakes. Ecology. Sci. 270:45–51. 71:83–98. Fincke O. 1982. Lifetime mating success in a natural population of the Mettke-Hofmann C, Ebert C, Schmidt T, Steiger S, Stieb S. 2005. damselfly Enallagma hageni (Walsh) (Odonata: Coenagrionidae). Personality traits in resident and migratory warbler species. Behav- Behav Ecol Sociobiol. 10:293–302. iour. 142:1357–1375. Fincke O, Hadrys H. 2001. Unpredictable offspring survivorship in the Moran NA. 1994. Adaptation and constraint in the complex life cycles damselfly, Megaloprepus coerulatus, shapes parental behavior, con- of animals. Annu Rev Ecol Syst. 25:573–600. strains sexual selection, and challenges traditional fitness estimates. O’Steen S, Cullum AJ, Bennett AF. 2002. Rapid evolution of escape Downloaded from https://academic.oup.com/beheco/article/20/1/30/212812 by guest on 01 October 2021 Evolution. 55:762–772. ability in the trinidian guppies (Poecilia reticulata). Evolution. 56: Fraser DF, Gilliam JF, Daley MJ, Le AN, Skalski GT. 2001. Explaining 776–784. leptokurtic movement distributions: intrapopulation variation in Overli O, Sorensen C, Korzan W, Summers C, Nilsson G. 2004. Track- boldness and exploration. Am Nat. 158:124–135. ing trout personality traits: subtle differences in male and female Godfrey JD, Bryant DM. 2000. State-dependent behaviour and energy stress coping style in juvenile fish. Horm Behav. 46:118. expenditure: an experimental study of European robins on winter Overli O, Sorensen C, Nilsson GE. 2006. Behavioural indicators of territories. J Anim Ecol. 69:301–313. stress-coping style in rainbow trout: do males and females react Godin J-G, Crossman SL. 1994. Hunger-dependent predator inspec- differently to novelty? Physiol Behav. 87:506–512. tion and foraging behaviours in the threespine stickleback (Gaster- Pierce CL, Crowley PH, Johnson DM. 1985. Behavior and ecological osteus aculeatus) under predation risk. Behav Ecol Sociobiol. 34: interactions of larval Odonata. Ecology. 66:1504–1512. 359–366. Plaistow S, Siva-Jothy MT. 1996. Energetic constraints and male mate- Godin J-G, Smith SA. 1988. A fitness cost of foraging in the guppy. securing tactics in the damselfly Calopteryx splendens xanthostoma Nature. 333:69–71. (Charpentier). Proc R Soc Lond B Biol Sci. 263:1233–1239. Groothuis TGG, Carere C. 2005. Avian personalities: characterization Plaistow S, Siva-Jothy MT. 1999. The ontogenetic switch between odo- and epigenesist. Neurosci Biobehav Rev. 29:137–150. nate life history stages: effects on fitness when time and food are Hedrick AV. 2000. Crickets with extravagant mating songs compensate limited. Anim Behav. 58:659–667. for predation risk with extra caution. Proc R Soc Lond B Biol Sci. Quinn JL, Cresswell W. 2005. Personality, anti-predation behaviour 267:671–675. and behavioural plasticity in the chaffinch Fringilla coelebs. Behav- Jeffries M. 1990. Interspecific differences in movement and hunt- iour. 142:1377–1402. ing success in damselfly larvae (Zygoptera: Insecta): responses Reale D, Gallant BY, Leblanc M, Festa-Bianchet M. 2000. Consistency to prey availability and predation threat. Freshw Biol. 23: of temperament in bighorn ewes and correlates with behaviour and 191–196. life history. Anim Behav. 60:589–597. Johansson F, Hjelm J, Giles BE. 2005. Life history and morphology of Reale D, Reader SM, Sol D, McDougall PT, Dingemanse NJ. 2007. Rana temporaria in response to pool permanence. Evol Ecol Res. Integrating animal temperament within ecology and evolution. Biol 7:1025–1038. Rev. 82:291–318. Johansson F, Rowe L. 1999. Life history and behavioral responses to Sawchyn WW, Gillott C. 1974. The life history of three species of time constraints in a damselfly. Ecology. 80:1242–1252. Lestes (Odonata: Zygoptera) in Saskatchewan. Can Entomol. 106: Johansson F, Stoks R, Rowe L, De Block M. 2001. Life history plasticity 1283–1293. in a damselfly: effects of combined time and biotic constraints. Schluter D. 1995. Adaptive radiation in sticklebacks: trade-offs in feed- Ecology. 82:1857–1869. ing performance and growth. Ecology. 76:82–90. Johnson JC, Sih A. 2005. Pre-copulatory sexual cannibalism in fishing Sih A, Bell AM, Johnson JC. 2004. Behavioural syndromes: an ecolog- spiders (Dolomedes triton): a role for behavioural syndromes. ical and evolutionary overview. Trends Ecol Evol. 19:372–378. Behav Ecol Sociobiol. 58:390–396. Sih A, Bell AM, Johnson JC, Ziemba RE. 2004. Behavioral syndromes: Jones AG, Arnold SJ, Burger R. 2003. Stability of the G-matrix in an integrative overview. Q Rev Biol. 79:341–377. a population experiencing pleiotropic mutation, stabilizing selec- Sih A, Kats LB, Maurer EF. 2003. Behavioural correlations across sit- tion and genetic drift. Evolution. 57:1747–1760. uations and the evolution of antipredator behaviour in a sunfish- Ketterson ED, Nolan V Jr. 1999. Adaptation, exaptation, and con- salamander system. Anim Behav. 65:29–44. straint: a hormonal perspective. Am Nat. 154:S4–S25. Sinervo B. 2001. Selection in local neighborhoods: the social environ- Koolhaas JM, Korte SM, de Boer ST, van der Vegt BJ, van Reenen CG, ment and ecology of alternative strategies. In: Dugatkin LA, editor. Hopster H, de Jong IC, Ruis MAW, Blokhuis HJ. 1999. Coping styles Model systems in behavioral ecology.Integrating conceptual, theo- in animals: current status in behavior and stress-physiology. Neuro- retical, and empirical approaches. Princeton (NJ): Princeton Uni- sci Biobehav Rev. 23:925–935. versity Press. p. 191–226. Laurila A, Kujasalo J. 1999. Habitat duration, predation risk and phe- Slos S, Stoks R. 2006. Behavioural correlations may cause partial sup- notypic plasticity in common frog (Rana temporaria) tadpoles. port for the risk allocation hypothesis in damselfly larvae. Ethology. J Anim Ecol. 68:1123–1132. 112:143–151. Lima SL. 1998. Nonlethal effects in the ecology of predator-prey in- Sokolovska N, Rowe L, Johansson F. 2000. Fitness and body size in teractions. Bioscience. 48:25–34. mature odonates. Ecol Entomol. 25:239–248. Lima SL, Dill LM. 1990. Behavior decisions made under the Sol D, Lefebvre L. 2000. Behavioural flexibility predicts invasion suc- risk of predation: a review and prospectus. Can J Zool. 68: cess in birds introduced to New Zealand. Oikos. 90:599–605. 619–640. Sol D, Timmermans S, Lefebvre L. 2002. Behavioural flexibility and Magnhagen C, Staffan F. 2005. Is boldness affected by group compo- invasion success in birds. Anim Behav. 63:495–502. sition in young-of-the-year perch (Perca fluviatilis)? Behav Ecol So- Stapley J, Keogh JS. 2005. Behavioural syndromes influence mating ciobiol. 57:295–303. systems: floater pairs of a lizard have heavier offspring. Behav Ecol. Magurran AE. 1990. The inheritance and development of minnow 16:514–520. anti-predator behaviour. Anim Behav. 39:834–842. Stoks R. 1998. Effect of lamellae autotomy on survival and foraging Maurer EF, Sih A. 1996. Ephemeral habitats and variation in behav- success of the damselfly Lestes sponsa (Odonata: Lestidae). Oecolo- iour and life history: comparisons of sibling salamander species. gia. 117:443–448. Oikos. 76:337–349. Stoks R, Johansson F. 2000. Trading off mortality risk against foraging McElreath R, Luttbeg B, Fogarty SP, Brodin T, Sih A. 2007. Evolution effort in damselflies that differ in life cycle length. Oikos. 91: of animal personalities. Nature. 450:E5. 559–567. Brodin • Behavioral syndrome over the boundaries 37

Stoks R, McPeek MA. 2003. Antipredator behavior and physiology van Oers K, Klunder M, Drent PJ. 2005. Context dependence of per- determine Lestes species turnover along the pond-permanence gra- sonalities: risk-taking in a social and a nonsocial situation. Behav dient. Ecology. 84:3327–3338. Ecol. 16:716–723. Stoks R, McPeek MA, Mitchell JL. 2003. Evolution of prey behavior in Walker EM. 1953. The Odonata of Canada and Alaska. Vol. 1. response to changes in predation regime: damselflies in fish and London, Ontario (Canada): University of Toronto Press. dragonfly lakes. Evolution. 57:574–585. Wellborn GA. 2002. Tradeoff between competitive ability and anti- SYSTAT 11. 2004. SYSTAT Software Inc. Suite F Richmond (CA). predator adaptation in a freshwater amphipod species complex. Tessier AJ, Leibold MA, Tsao J. 2000. A fundamental trade-off in re- Ecology. 83:129–136. source exploitation Daphnia and consequences to plankton commu- Werner EE, Anholt BR. 1993. Ecological consequences of the trade-off nities. Ecology. 81:826–841. between growth and mortality rates mediated by foraging activity. van Oers K, de Jong G, Drent PJ, van Noordwijk AJ. 2004. A genetic Am Nat. 142:242–272. analysis of avian personality traits: correlated response to artificial Wolf M, van Doorn S, Leimar O, Weissing FJ. 2007. Life-history trade-offs selection. Behav Genet. 34:611–619. favour the evolution of animal personalities. Nature. 447:581–585. Downloaded from https://academic.oup.com/beheco/article/20/1/30/212812 by guest on 01 October 2021