| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Behavioral Ecology Vol. 14 No. 5: 626-633
© 2003 International Society for Behavioral Ecology
Context-dependent reproductive site choice in a Neotropical frog
Department of Biology, Duke University, Durham, NC 27708-0338, USA
Address correspondence to P.J. Murphy, who is now at the Department of Biological Sciences, Box 8007, Idaho State University, Pocatello, ID 83204-8007, USA. E-mail: pjmurphy{at}isu.edu.
Received 4 April 2002; revised 4 October 2002; accepted 25 October 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
In organisms whose offspring develop in discrete habitat patches such as pools, studies have frequently shown that adults avoid sites based on a single risk factor facing offspring. However, natural reproductive sites often vary in multiple risk factors in both space and time. In this study, I used choice tests among field mesocosms to determine whether adults of a Neotropical anuran, Edalorhina perezi, select the safest pools for offspring based on two biotic risks of different magnitude: insect predators and conspecific competitors. I also investigated whether adult site-choice was context dependent (i.e., whether it varied by sex or by season or whether it was based on local pool quality). I found that both sexes avoided pools containing predatory insects, but only females significantly avoided those containing conspecific tadpoles. When offered two risky options, both sexes favored pools with competitors over those with predators. Site-choice behavior also varied depending on the temporal and spatial context. Female sensitivity to insect predators decreased late in the season. In addition, both sexes exhibited dampened reproductive activity when only risky sites were available locally. This study emphasizes that social and environmental factors simultaneously impact reproductive site choice. Whether a site is accepted ultimately depends not only on the assessment of mortality risks to offspring but also on the sex and spatiotemporal context of the decision maker.
Key words: amphibians, calling sites, choice experiments, Edalorhina perezi, insect predators, oviposition sites, Peru, pool mesocosms, reproductive site choice, tadpole competitors, tadpole mortality risk.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
In organisms such as butterflies, mosquitoes, and frogs whose juvenile stages develop in single patches of suitable habitat (i.e., a single host plant or pool), the ability of adults to distinguish and avoid poor-quality patches for oviposition should strongly influence reproductive success and lifetime fitness through enhanced juvenile growth and survival. Previous studies such taxa have demonstrated that adults avoid sites where the likelihood of predation or competition facing larvae is high (Fincke, 1992
; Howard, 1978; Rausher, 1979; Stav et al., 1999). Yet egg-laying sites often vary in multiple risk factors simultaneously. For example, ephemeral pools may have higher densities of larval competitors, whereas more persistent pools have higher predator densities (Skelly, 1996). In iteroparous species, moreover, several sets of offspring, often from multiple reproductive sites, contribute to lifetime fitness. Hence, the most favorable behavioral response to larval risks at any single site may reflect various aspects of reproductive strategy including sex, body condition, and the alternatives currently available and attainable by adults. In this study, I used experiments to explore how reproductive site choice in a Neotropical frog, Edalorhina perezi (Anura: Leptodactylidae), varied given two biotic risks facing offspring. In particular, I considered how site-choice behavior varied depending on sex, season, and the reproductive sites locally available to reproducing adults.
Where variation in multiple risk factors is common, selection should favor the ability of adults to weigh relative risk and select pools where offspring have higher survival probabilities. Few studies of reproductive site choice, however, have either tested or confirmed that adults successfully weigh multiple risk factors facing larvae (but for mosquitoes, see Edgerly et al., 1998). Crump (1991) did not detect a differential response by adult frogs to two direct risks (drying and cannibalism), nor did Resetarits and Wilbur (1989) find that frogs distinguished a direct and an indirect risk (predation and competition). A drawback of these studies was that the relative importance of the risk factors at natural pools was not known or not reported. In both cases, experimental pools of lower risk were available nearby, and thus adults were not forced to choose among risky sites as may often occur at natural pools. In a 3-year observational study, I identified three primary risk factors facing E. perezi larvae: pool drying, insect predators, and conspecific tadpoles (Murphy, 2003). An estimated 48% of tadpole mortality resulted from the two biotic risks: 47% directly from insect predation and 1% indirectly from intraspecific competition, which decreased growth rate and increased exposure to pool drying. When calculating the risks of conspecific tadpoles, I ignored the possibility that conspecifics may increase the risk of disease transmission to adults or to larvae (Kiesecker et al., 1999). I used information on quantified risks from natural pools to make predictions about site preference in this study when choice was limited to pools with larval predators or competitors.
The tree-fall pools favored by E. perezi not only expose larvae to more than one risk, but sites differ spatially and temporally in their social context (i.e., the number and quality of mates available to males and females). When males display and mate at the sites where offspring develop, site choice and mate choice may not be independent. Hence, sex differences in the costs of courtship and reproduction may favor distinct site-choice criteria in males and females (e.g., Emlen and Oring, 1977; Johnstone et al., 1996; Trivers, 1972). Females may weigh site quality more heavily when their per-brood investment or their opportunity costs (measured as time until they can reproduce again) are higher than males (Emlen and Oring, 1977; Rosenheim, 1999). In species where reproductively mature males outnumber females at reproductive sites (male-biased operational sex ratio; Emlen and Oring, 1977), males may select sites that are high in quality for offspring only insofar as it increases their access to females (reviewed in Andersson and Iwasa, 1996). Therefore, depending on the social and environmental conditions, a male may sacrifice site quality to increase mating opportunities (Emlen and Oring, 1977). Given that mark–recapture data demonstrate that the operational sex ratio at pools is male biased and that male mating success is highly skewed in E. perezi (Murphy, unpublished), I tested whether sex differences in site choice were evident in this study.
In addition to differences between the sexes, reproductive site choice may be flexible, involving seasonal shifts in criteria, facultative adjustments by individuals, or both. Due to the impact of climate on both larval survival and development, site-choice criteria may shift seasonally. Yet most experimental studies of reproductive site choice, particularly in anurans, have not considered potential seasonal shifts in criteria (but see Spieler and Linsenmair, 1997). For example, when few days remain in the reproductive season, cues that formerly triggered site rejection may be ignored or actually enhance site acceptance. In mosquitoes, Edgerly et al. (1998) found that adults generally avoided pools with larval predators. Late in the reproductive season, however, adults favored sites with predators, presumably using their presence as a cue to pool persistence. I explored a similar possibility for seasonal flexibility in E. perezi by conducting an identical choice experiment both early and late in the rainy season.
Beyond any seasonal shift in site-choice criteria, the condition of an individual as well as their local context (social and spatial factors) may induce facultative changes in their site selectivity. For example, in fish and amphibians, a male's condition relative to competitors may influence either mating strategy (e.g., sneaking vs. courting; review in Gross, 1991) or choice of display sites (Murphy, unpublished). Any increase in mating opportunities that a male attains by sacrificing site quality will also depend on the options locally available. When reproductive sites are clustered, females can feasibly inspect both males and sites, potentially rendering ineffectual any male display that does not reflect site quality (i.e., dishonest displays; Candolin and Reynolds, 2001, Johnstone and Grafen, 1993). Third, temporal physiological factors also may interact with condition to influence site-choice decisions. For example, female Physalaemus pustulosus (i.e., túngara frogs) may delay oviposition until their complement of mature ova increases (Davidson and Hough, 1969), unless a particularly high-quality site becomes available. In this study, I focused on how the local spatial context (i.e., the quality of pools locally available) may impact E. perezi site choice, particularly calling frequency by males and time to oviposition by females.
This study addressed reproductive site choice by adult frogs in response to two biotic risks to offspring and looked for evidence of facultative behavior based on sex, season, or the quality of local options. I addressed four questions. First, do both sexes select pools to reduce insect predation and competition risks to offspring? Second, do adults avoid pools with greater direct mortality risks more readily than pools with lesser indirect risks? Third, does female site choice in response to insect predators show seasonal flexibility? And fourth, do male and female site-choice behaviors vary facultatively based on local pool quality for offspring? I addressed the first and second questions by testing male calling and female oviposition behavior in three paired experiments in field mesocosms: tadpole predators present or absent (the predator experiment), tadpole competitors present or absent (the competitor experiment), and tadpole predators or tadpole competitors present (the predator vs. competitor [PC] experiment). To address the third question, I repeated the predator experiment late in the rainy season and compared female site choice in both experiments. Finally, to address the fourth question, I compared frequency of male calling, female time to oviposition, and adult pool rejection across all three experiments.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
System
I conducted the experiments in the floodplain forest surrounding the Cocha Cashu Biological Station (elevation 384 m) in Manu National Park, Peru (Terborgh, 1983
). Edalorhina perezi is an iteroparous, continuous breeder and member of the foam-nesting Leptodactylines (Leptodactylidae; Downie, 1993). The natural history of this species is covered in detail elsewhere (Duellman and Morales, 1990, Murphy, 2003, Schlüter, 1990). I note here the following details relevant to site-choice behavior. First, mean development time for tadpoles (egg to metamorph) was 24.3 ± 1.3 (SE) days, with the minimum being 21 days. Therefore, larvae hatching within 3–4 weeks of the average end of the rainy season face an increasing likelihood of desiccation, unless found at more persistent pools (Murphy, 2003). Increasing risk of drying may favor a change in site-choice criteria by adults, a possibility I explored by repeating the predator experiment late in the season. Second, mated pairs may interrupt nest construction and resume at a different site, with the female carrying the male. To account for potential nest subdivision in choice experiments, I recorded both nests and total eggs placed at pools. Third, while in amplexus (mated), males contribute energetically to nest construction by whipping up, with their hind legs, the egg mixture secreted by females during a 1–2 h period. Despite this energetic contribution by males, evidence from mark–recapture studies suggests a greater temporal investment per clutch by females: the median time between matings was 43% longer for females than males (20 vs. 16 days), and some males mated on successive nights (Murphy, unpublished). These data suggest that if sex differences exist in site-choice criteria, females should be more sensitive to risks facing offspring than males (Emlen and Oring, 1977).
Experimental calendar
Edalorhina perezi breeds opportunistically over a 7-month rainy season in southeastern Peru. Rainfall sufficiently frequent to fill a majority of the active breeding sites at Cocha Cashu—a mean (±SE) of 47 ± 2 mm falling each 5-day period (11-year record)—usually begins by late October and ends by early April (Murphy, 2003). Beginning 1 April, 5-day rainfall totals decrease to a mean (±SE) of 29 ± 9 mm, insufficient to sustain water in many forest pools. I conducted all choice experiments during the 1996–97 rainy season. I ran the predator experiment for 60 days beginning 22 October 1996, followed by the competitor and PC experiments for 70 days beginning 26 January 1997 (Figure 1). These experiments were not run simultaneously due to an initial scarcity of E. perezi tadpoles. To test for a possible seasonal shift in site-choice criteria, I repeated the predator experiment for 36 days beginning on 7 March, a start date approximately 25 days from the end of the rainy season as defined by the decrease in rainfall regularity.
|
General procedures
I used artificial pools (mesocosms) as experimental reproductive sites because they allowed me to control for differences in soil characteristics and basin shape and to facilitate replicate arrays with equal spacing among pools. Similar advantages have made their use standard in site-choice experiments (e.g., Edgerly et al., 1998
, Marsh et al., 2000, Resetarits and Wilbur, 1989). Two tubs (25 l), buried to their rim, were separated by 8 m in fenced arrays in relatively uniform high-ground forest (Figure 1). Tub volume was well within the range of natural pools selected by E. perezi: 8 of 36 active pools from 1995–1997 had mean daily volumes of less than 25 l. Separation by 8 m, shorter than the mean nearest neighbor distance between active natural tree-fall pools (31.1 ± 5.4 m; range in 20-ha study area 1.0–522 m; Murphy, 2003), allowed me to feasibly fence arrays to control adult frog movement yet leave 6 m of undisturbed vegetation between pools. Arrays were separated by at least 100 m in the landscape. Each buried tub received a 1-cm layer of sifted soil and was then lined with "no-see-um" screening (Balson-Hercules Group, Ltd.) that hung from the tub edges and rested on the soil base. By simply lifting this screening, I could rapidly census insect predators and tadpole competitors and remove unwanted intruders. The screening also provided texture to tub walls so that frogs could easily enter and exit. To increase habitat heterogeneity, I added 20 g of dry leaf litter to each pool. Drift fences, constructed of sturdy plastic 0.6 m in height, surrounded arrays. Fences had eight one-way funnel traps (Figure 1), half with entrances directed outward and half inward, so that I could regulate adult movement, retain a group of calling males, and count attempts to enter and exit. Vegetation was cleared 0.5 m from fences and from buried tubs to prevent frogs from escaping and to facilitate visual census of calling males and amplectant pairs.
Do both sexes select pools to reduce insect predation and competition risks to offspring?
I conducted the predator and competitor experiments to test male calling and female oviposition site preference based on insect predation and conspecific competition risks to offspring (Figure 1). In the eight replicate arrays erected for both experiments, a control tub, maintained free of tadpole predators and competitors, was paired with a high-risk tub. In the predator experiment, high-risk tubs contained four Aeshna odonate naiads, 20–40 mm in total length, the most common predator in natural pools (Murphy, 2003). Each week, naiads were fed 15–20 E. perezi hatchling tadpoles (Gosner stage [GS] 25–26; Gosner, 1960) to ensure proper chemical cues (Kats and Dill, 1998; Spieler and Linsenmair, 1997). These tadpoles were usually consumed in 1 day, but always before 3 days. In the competitor experiment, high-risk tubs contained 30 E. perezi hatchling tadpoles. Experimental predator and tadpole densities (6.6/m2 and 49/m2 of pool floor surface area, respectively) approximated the mean densities observed at natural, persistent tree-fall pools during the 1995–1996 rainy seasons (7.5/m2 and 31/m2 of pool floor surface area, respectively). I checked tub contents every 3 days, adding or removing insects and tadpoles as necessary. I used laboratory populations of naiads and tadpoles to maintain treatment densities.
I captured adults for experiments at natural breeding sites 2–5 km from the mark–recapture study area and 0.2–3 km from arrays. Additional adults, attracted by calls, were captured in entrance funnel traps (Figure 1). Males, individually marked with toe-clips (Heyer et al., 1994), were introduced into arrays first, a randomly selected group of four to six per array. All included males were within 2.0 mm in snout-vent length (SVL); I excluded any particularly small or large male. Individually marked females were released into arrays near the low-risk pool one at a time, such that females were not able to copy other females (Pruett-Jones, 1992). I removed long-time resident males (>20 days), releasing them at least 100 m away. I left a female in an array until she mated and placed a clutch, entered exit traps at least twice, or showed no reproductive activity for at least 7 days. Arrays were visited twice daily, morning and evening, to remove frogs from funnel traps, count calling males, and remove recently active females and their foam nests from arrays. Calling site was noted by counting males in calling posture (or with an inflated vocal sac) near each pool. I did not recapture (i.e., identify) individual males during visits to minimize behavioral disturbance. I counted the number of eggs in each foam nest in the laboratory.
I calculated three responses in both experiments to test male and female site choice: (1) the mean number of males calling at each tub per day, (2) the total nests placed in each tub, and (3) the total eggs placed in each tub. Because the response at one tub was not independent of the other tub within an array, I treated arrays as replicates. I then used paired t tests (two tailed) to evaluate whether adults exhibited a non-zero, directional preference within the arrays. I log-transformed calling data and square root-transformed nest and egg data before statistical analysis.
Do adults avoid pools with greater direct mortality risks more readily than pools with lesser indirect risks?
In the PC experiment conducted spring 1997, I restricted adult choice to two risky sites for offspring, a pool with high predator density and a pool with high conspecific tadpole density (Figure 1). I used four replicate arrays and set treatment insect and tadpole densities as above. Arrays were visited twice daily to vacate traps, note behaviors, and remove females and foam nests as necessary. I measured male and female site choice for each replicate array using the same response variables as in the previous two experiments. Because no low-risk option was available, I predicted that frogs would prefer high competition over high predation tubs because of the direct and higher mortality costs of insect predation quantified at natural pools (Murphy, 2003). I tested this prediction using a paired t test (two tailed) as in the predator and competitor experiments. As previously, I treated arrays as replicates and log-transformed calling data and square root-transformed nest and egg data before statistical analysis.
Does female site choice in response to insect predators show seasonal flexibility?
I repeated the predator experiment (Figure 1) late in the rainy season to see whether female site choice in response to insect predators changed temporally. I set up four replicate arrays, regulating treatments and monitoring female behavior as above. I measured two responses for each replicate array: the number of nests placed in each tub and the total eggs placed in each tub. Due to insufficient time to census during calling activity, I did not collect data on male behavior late in the season. The two response variables were analyzed as in the earlier predator experiment, using paired t tests (two tailed) to evaluate whether a consistent directional preference was evident within arrays. In addition, to see if site-choice criteria changed from early to late in the season, I used a t test (two tailed, not paired) to compare the percentage of nests in each array placed in the predator-free tubs in both trials. For each array, I weighted the response variable (arcsine square-root transformed) by the total nests placed in that array because the early and late season trials were of unequal duration.
Do male and female site-choice behaviors vary facultatively based on local pool quality for offspring?
I used comparisons across the predator, competitor, and PC experiments to test how E. perezi reproductive behavior changed when safe pools for offspring were not locally available. First, did adults attempt to leave arrays more frequently? Second, did males call less frequently? Finally, did the time between female release and nest placement increase? To quantify adult tendency to leave arrays, I counted the number of frogs found in exit traps and divided this by the number of days the array was in use. I proceeded similarly for calling activity, calculating the mean number of calls per day for each array. For female delay, I calculated the mean time from female release to oviposition in each array (excluding frogs that never oviposited). I then compared exit rates, call frequency, and female delay by experiment (and, for exit rates, by sex) using ANOVA. For the predator experiment, I included only the data from the eight arrays run early in the rainy season. I pooled data by sex and by experiment (i.e., data from the predator and competitor experiments) when justified by the initial analyses. When comparing the PC experiment with pooled data from the other two experiments, I used Brown and Forsythe's F test (Zar, 1996) due to unequal sample sizes (n = 20 vs. n = 4). To normalize data, exit rates and call rates were arcsine-square root transformed and female delay was log transformed.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Experimental activity summary
Overall, 105 females and 93 males were active in the three choice experiments: 51 (27 early and 24 late) and 55 (35 early and 20 late) in the predator experiment, 43 and 42 in the competitor experiment, and 25 and 37 in the PC experiment, respectively. More males (38) than females (14) were active in multiple experiments. Similarly, more males (40) than females (34) were active in multiple arrays within an experiment. Differences in total activity by sex were due in part to the recovery time required by females before subsequent oviposition (see Methods). In addition, males that were removed from one array after 20 days frequently moved to distant arrays. A total of 212 nests were found at tubs within arrays: 105 in the predator experiment (67 early, 38 late), 78 in the competitor experiment, and 29 in the PC experiment.
Do both sexes select pools to reduce insect predation and competition risks to offspring?
In the predator experiment, when given a choice between pools that contained insect predators and those that did not, E. perezi males preferred predator-free pools as calling sites (t2,7 = 4.5, p =.003; Figure 2A). Males called more often at predator-free than at control pools (mean ± 1 SE): 30% ± 7 vs. 3% ± 1 of the experimental days, respectively. However, in the competitor experiment in which one of two pools contained conspecific tadpoles, male behavior was not definitive. Males called more often at tadpole-free pools (16% ± 6 vs. 8% ± 3 of the days), but this trend was not significant (t2,7 = 1.2, p = 0.279) due to lack of consistency of calling activity across arrays (Figure 2A).
|
Unlike males, E. perezi females exhibited definitive risk-sensitive behavior in both experiments. Females were more likely to choose predator-free than control pools: 50 of 67 nests were placed in pools without predators (t2,7 = 3.2, p =.015; Figure 2B). Females also avoided oviposition at pools with competitors: 61 of 78 nests were placed at pools without conspecific tadpoles (t2,7 = 4.0, p =.005; Figure 2B). Statistical analyses on egg placement data yielded similar results to those for nest placement for both experiments (t2,7 = 4.9, p =.002 and t2,7 = 4.2, p =.004, respectively), indicating that subdivision of nests by females did not confound the results.
Do adults avoid pools with direct mortality risks more readily than pools with indirect risks?
When forced to choose between pools containing either tadpole predators or conspecific tadpoles, adults, on average, preferred pools with tadpoles, yet the trends were not as definitive as when a safe site for offspring was available. Males called more often (mean % of days ± 1 SE) at pools with tadpole competitors (8 ± 3) than at those with insect predators (1 ± 1; Figure 3A), but the trend was only marginally significant (t2,3 = 1.9, p =.147). Females exhibited a significant preference for pools with tadpoles in terms of nest placement (19 vs. 10 nests; t2,3 = 4.0, p =.027; Figure 3B). When data on total eggs were analyzed, the trend favoring predator-free pools was not significant (total 1710 vs. 804; t2,3 = 2.0, p =.143). This increase in variability resulted from unusual nest sizes (mean ± 1 SE) in one PC array: two large nests were found in the predator pool (165 ± 27), while four smaller nests were found in the competitor pool (78 ± 8).
|
Does female site choice in response to insect predators show seasonal flexibility?
Unlike early in the rainy season, females showed no preference for pools without insect predators late in the season in terms of either nests (t2,3 = 0.1, p = 0.983) or total eggs (t2,3 = 0.5, p =.652). Late in the season, mean (±1 SE) reproductive activity was roughly equal in predator and control pools for both nests placed (4.5 ± 1.2 vs. 5.0 ± 1.4) and eggs placed (270 ± 65 vs. 255 ± 76, respectively). Thus, the preference for predator-free pools within arrays (mean % of nests ± 1 SE) decreased from 75% ± 6 early to 52% ± 10 late in the year, a change that was marginally significant (t2,10 = 2.1; p =.063). This change was driven by an increase in variability of female choice late in the year: the coefficient of variation (CV) in the proportion of nests placed in control pools late in the year (CV = 0.378) was nearly twice that early in the year (CV = 0.223).
Do male and female site-choice behaviors vary facultatively based on local pool quality for offspring?
When no low-risk pool was available within arrays, adults entered exit traps more often, males called less frequently, and females appeared to delay oviposition. However, only the latter two results were clearly independent of any seasonal effect. Mean exit rates (Figure 4A) differed by experiment (F2,36 = 3.7, p =.034) but not by sex (F1,36 = 0.1, p =.735). Pooled by sex, exit rates (mean/day ± SE) were 0.10 ± 0.02 in the predator experiment, 0.23 ± 0.05 in the competitor experiment, and 0.32 ± 0.14 in the PC experiment. Post-hoc contrasts (
= 0.017) revealed that exit rates were significantly higher in the PC experiment (F1,36 = 6.6, p =.015) and near-significantly higher in the competitor experiment (F1,36 = 4.2, p =.048) than in the predator experiment. Rates in the PC and competitor experiments did not differ (F1,36 = 0.8, p =.380).
|
Male calling frequency did not differ in the predator and competitor experiments in which a low-risk pool was available (F1,17 = 0.7, p =.408). However, males called less frequently in the PC experiment than in the former two experiments pooled (FB&F:1,16 = 11.4, p =.004). For male calling frequency (Figure 4B), the means per day (±1 SE) were 0.33 ± 0.07 in the predator experiment, 0.24 ± 0.07 in the competitor experiment, and 0.08 ± 0.03 in the PC experiment.
As indicated by the experimental means (days ± 1 SE), oviposition delay (release to nest placement) in the PC experiment (3.4 ± 0.7) was greater than in the predator (1.6 ± 0.3) and competitor experiments (1.9 ± 0.2). Statistically, delay did not differ in the predator and competitor experiments, run in fall and spring, respectively (F1,17 = 1.0, p =.339). Similarly, the trend for greater delay in the PC experiment was also not significant when compared to the other experiments pooled (FB&F:1,4 = 2.5, p =.190). The lack of significance was driven by one array in the PC experiment in which six of nine females were excluded from analysis due to inactivity (i.e., each was removed after a delay of 7 days). Hence the calculated mean did not reflect average delay in that array (Figure 4C). Upon exclusion of this array, delay in the PC experiment was higher than in the other experiments (FB&F:1,4 = 9.9, p =.051). In sum, I noted two behavioral trends when adults were restricted to two poor sites: males called less frequently and females appeared to delay nest placement.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
In species that exploit discrete reproductive sites, numerous studies have shown that adults select sites to reduce a single mortality risk facing their offspring. Yet several risk factors, including predation, competition, and impermanence, often simultaneously face offspring. Moreover, adults must weigh their decision to reject a site based on their current social context, the time remaining to reproduce, and the nearby environment. This study demonstrated that E. perezi adults select reproductive sites with regard to two biotic risks facing their offspring and that their decisions are to some degree context dependent. Below, I discuss the risk sensitivity I observed experimentally and how it varied depending on the sex, season, and nearby environment of the decision maker. I also consider why context-dependent site-choice behavior may be favored in E. perezi and similar species.
Site choice in response to two biotic risks facing offspring
E. perezi adults avoided pools with direct larval mortality risks (i.e., insect predators) more strongly than pools with indirect mortality risks (i.e., conspecific tadpoles; Figures 2 and 3). Sex differences were apparent in the response to tadpoles, and I discuss this topic in the next section. The adult response to these larval mortality risks is consistent with their impact at natural pools (see above). Nevertheless, objections could be raised as to how risks to larvae were quantified from field data (i.e., I considered the impact of conspecific tadpole density on exposure to drying and insect predation [Murphy, 2003], not any potential link to disease transmission [Kiesecker et al., 1999]). At this time, however, I have no data indicating that tadpole disease is an important mortality risk in E. perezi.
Recent work has reemphasized the importance of confirming mesocosm predictions in the field (Marsh and Borrell, 2001), where, in addition to mortality risks, pools vary in size, topography, cover, degree of isolation, and other factors. Field data suggest that E. perezi adults weigh both biotic risks but that predation risks do indeed receive greater emphasis. For example, oviposition rates were lower at pools with recently hatched larvae than at unoccupied pools (Murphy, 2003). Yet at new pools or at pools that had recently dried (both situations with near-zero insect predator densities), E. perezi females oviposited in great numbers, ignoring high conspecific tadpole densities (Murphy, 2003, unpublished data). Spieler and Linsenmair (1997) similarly noted a graded response by adult Hoplobatrachus occipitalis to pools with varying degrees of drying and cannibalism risk to tadpoles.
The degree of pool isolation in natural habitats may also affect the adult response to offspring risks. In primary lowland forests, a single tree often triggers a series of tree falls, producing pools that are clustered on the forest floor. Hence, E. perezi adults can frequently compare nearby pools (see nearest neighbor data in Methods). Yet in patches of forest with low pool density, the rejection threshold will likely be higher, which could be tested by varying pool quality and spacing within experimental arrays. Future work is also necessary to identify risk detection mechanisms in E. perezi. Chemosensory detection, via byproducts of predator excretory metabolism (reviewed in Kats and Dill, 1998), currently appears more plausible than visual detection because of the turbidity of experimental and natural pools.
Site choice that is sensitive to invertebrate predators, as opposed to vertebrate predators or competitors, has not been previously observed in amphibians (e.g., Resetarits and Wilbur, 1989, 1991). In tree-fall pools at Cocha Cashu, odonate naiads were the dominant biotic mortality risk facing tadpoles (Murphy, 2003), consistent with observations from terra firme pools in the Brazilean Amazon (Gascon, 1989, 1992). Because fish and other vertebrates are uncommon and insect predators exact high costs, it is not surprising that E. perezi has evolved the ability to detect them. Such sensitivity probably exists in many tropical amphibians that reproduce in temporary forest pools.
E. perezi adults may be able to obtain more information on future mortality risks to offspring in a single pool visit by assessing predator density (via the concentration of a waterborne or volatile chemical cue) than by attempting to measure drying risk. Pool drying was the most important abiotic mortality risk at pools, matching the estimated mortality due to insect predators (Murphy, 2003). Based on regression analyses, sampled insect predator density explained 32% of the probability of E. perezi tadpole survival. Conversely, pool depth and volume explained, on average, only 25% of a pool's probability of persistence through tadpole development (Murphy, 2003). Obtaining more precise information on pool persistence would require remaining at a pool or visiting it repeatedly (Spieler and Linsenmair, 1997). Yet making a single assessment of future risk may be critical, particularly at pools formed by recent disturbances (tree falls or peccaries; Gascon and Zimmerman, 1998). Such pools, while attractive to E. perezi, have no drying history, leading adults to often mistakenly select them even when highly ephemeral (Murphy, 2003).
Context dependence: sex differences in site choice
Unlike the response to insect predators, only females strongly avoided pools with conspecific tadpoles (Figure 2). Because both experiments had equal replication, the weaker response by males in the competitor experiment is probably not an artifact. Decreased site selectivity in male E. perezi is consistent with other anurans (Crump, 1991; Resetarits and Wilbur, 1991), suggesting that similar forces may produce sex differences.
At least three nonexclusive hypotheses may explain sex differences in site selectivity. First, selection may favor more stringent site-choice criteria in females because of their higher temporal and energetic investment per clutch. In E. perezi, the median recovery time between matings was 43% longer in females than in males, and the average (±SE) clutch made up 11% ± 1 of a female's mass (Murphy, unpublished). Thus males, although they contribute to foam nest construction, clearly invest less in each clutch. Such male "emancipation" (Emlen and Oring, 1977) may favor male strategies that maximize matings, even given higher mean mortality risks to sired clutches. At first glance, the best strategy to maximize matings would appear to be anticipating female site choice; however, factors other than mere arrival (i.e., social status, see below) may influence access to females.
A second explanation for differences in site selectivity is that sex-specific behavior may expose one sex either to greater risks or to more information. For example, male display may incur additional predation risks (Tuttle and Ryan, 1981). Or females, by entering a pool to hydrate before oviposition (e.g., Pyburn, 1970), may be exposed to waterborne cues (Kats and Dill, 1998) more often than males. Yet site-choice experiments with a randomized design should control for any site-specific predatory risks to males that could impact their site choice. Moreover, if males are less likely to detect waterborne cues, their response should decrease to both predators and competitors unless only one of the two cues is volatilized. Unfortunately, at present, the detection mechanism for either cue is unknown.
A third hypothesis is that male social status, by changing potential access to mates, influences site choice. In mating systems with a male-biased operational sex ratio (OSR; Emlen and Oring, 1977), low-status males may sacrifice site quality to increase mating opportunities. This hypothesis may best explain the variation in male behavior observed across arrays in the competitor experiment. In all the experiments, the OSR was five males to every female, higher than the mean (±SE) at natural sites (5.6 ± 0.6 males to 3.2 ± 0.4 females; Murphy, unpublished). Because the traits that determine male status are not known in E. perezi, certain arrays may have contained a broader spectrum of males than others. In these arrays, males of low status may have moved to pools with conspecific larvae, whereas in arrays in which males were more equally matched, fewer abandoned control pools. In the predator experiment, the greater mortality risk posed by insect predators may have outweighed any tendency of low-status males to abandon the crowded control pool.
Context dependence: seasonality in site choice
When selecting among experimental pools late in the rainy season, female E. perezi showed no sensitivity to insect predators as they had earlier in the rainy season. This pattern is consistent with observations at natural pools where the majority of activity during the last third of the rainy season occurred at riskier pools (Murphy, 2003). I provide three hypotheses that could explain this seasonal context dependence in risk sensitivity. First, a shift in the response to a predator cue may reflect its decreasing importance seasonally in light of mortality risks from other sources. For example, mosquitoes lay more eggs at predator-free pools at the start of the wet season when pools are unlikely to dry but favor pools with predators late in the year, perhaps using predators as indicators of pool persistence (Edgerly et al., 1998). Second, risk-prone or risk-neutral (neither avoidance nor preference for risk) behavior may increase when adults become time limited, as they could late in the reproductive season. For example, parasitic wasps with many eggs remaining in their ovaries were less discriminating when selecting hosts, particularly late in the year (Heimpel et al., 1996; Roitberg et al., 1992). Finally, it has been suggested that the response to site-specific cues may depend on adult experience (Stamps, 1988). For example, naive túngara frogs, abundant early in the year, may favor pools with conspecifics; conversely, late in the season few adults are naive, and thus a majority may avoid occupied pools (Marsh and Borrell, 2001).
Based on the available evidence, the seasonal shift in the response of E. perezi to insect predators appears to be most consistent with a seasonal change in the importance of predators and an increase in risk-neutral behavior by adults. Predator densities decrease over the rainy season (Murphy, 2003), and hence adults late in the year may become unresponsive to predator cues. Neither the predator experiment nor evidence from natural pools, however, suggest that E. perezi females begin to favor pools with predators late in the year. Likewise, field data do not suggest that adults become more selective with respect to predators with more experience (Murphy, 2003). In sum, late in the season, risks facing offspring appear to weigh less heavily on E. perezi site choice; thus, as with insect parasitoids, limited time may decrease site selectivity. A complete explanation of this apparently maladaptive behavior requires further study.
Context dependence: impact of local pool quality
When, in the PC experiment, only risky pools for offspring were locally available, both sexes exhibited a decrease in reproductive activity: males called less frequently (Figure 4B), and females delayed oviposition (Figure 4C). A decrease in courtship based on site quality for offspring has been infrequently documented (in reef fish: Sikkel, 1995; Warner and Dill, 2000; in anurans: Mitchell, 2001). Reducing courtship may permit males to allocate more energy to growth, thereby improving their future social status at better sites. Such a strategy may be especially beneficial if females reject poor sites independent of male behavior (Candolin and Reynolds, 2001, Murphy, unpublished).
An increase in time to oviposition by E. perezi females in the PC experiment similarly indicates a facultative response to local site quality (Figure 4C). Females may have delayed oviposition in order to carefully sample pool quality in PC arrays. For example, telemetry data on Hoplobatrachus occipitalis showed that females visited up to six closely-spaced pools in a single night (Spieler and Linsenmair, 1997). An in increase in the exit rate during the PC experiment may likewise suggest a desire to sample other pools (Figure 4A). Yet this increase cannot be disentangled from a seasonal increase in exit rate because it did not differ from the competitor experiment conducted concurrently. Females may also delay oviposition to await improvement in pool quality. Such delay has been noted for various taxa in response to food supply and temperature (Farfan et al., 1998; Vannoordwijk et al., 1995). In pools used by E. perezi, drying events are common and eliminate most tadpoles and insect predators, making oviposition more favorable after the next rainfall. Thus, delaying egg-laying may benefit E. perezi females due to both spatial and temporal variability in pool quality in the lowland rain forest.
Conclusions
Many tropical amphibians such as E. perezi breed continuously over long seasons in which favored reproductive sites change temporally in several mortality risks to offspring. Moreover, the forest floor is dynamic, particularly in undisturbed forests, successionally producing both clusters and spatially isolated reproductive sites. Under these conditions, selection should favor adults that wisely allocate their reproductive effort in a variety of situations. In response to two biotic risks, I found that E. perezi favored sites with competitors over predators, consistent with the strength of these risk factors at natural pools. Males on average were less selective than females in response to tadpole competitors, suggesting that either courtship behavior or the mating system influence site selectivity. Context-dependent site choice was evident in two additional results: site selectivity in response to predators decreased seasonally, and overall activity decreased when only risky pools were locally available. In sum, reproductive site choice in E. perezi involves social, seasonal, and local adjustment to the environment. Further experimental work is necessary to select among competing hypotheses for context dependence and to quantify their adaptive benefit versus fixed reproductive strategies.
|
ACKNOWLEDGEMENTS |
|---|
I am grateful to Bill Morris, Peter Tiffin, Kurt Regester, the Pfennig lab, and two anonymous reviewers who greatly improved the manuscript. Dianne Purdy, Wilfredo Arizabal, and Ariadne Angulo provided assistance in the field. John Terborgh and Mercedes Foster provided help with field logistics. The Instituto de Recursos Naturales, Lima, and Parque Nacional Manu provided field permits. Special thanks go to Carmela Landeo, who was a supportive throughout the project. Support for the research came from the Organization for American States, the Duke Chapter of Sigma Xi, the National Science Foundation (Dissertation Improvement Grant 9700515 to W.F. Morris), and the U.S. Department of Education.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Andersson M, Iwasa Y, 1996. Sexual selection. Trends Ecol Evol 11: A53-A58. [CrossRef]
Candolin U, Reynolds JD,
2001. Sexual signaling in the European bitterling: females learn the truth by direct inspection of the resource.
Behav Ecol
12:
407-411.
Crump ML, 1991. Choice of oviposition site and egg load assessment by a treefrog. Herpetologica 47: 308-315.
Davidson EH, Hough BR, 1969. Synchronous oogenesis in Engystomops pustulosus, a Neotropical anuran suitable for laboratory studies: localization in the embryo of RNA synthesized at the lampbrush stage. J Exp Zool 172: 25-48. [CrossRef][Web of Science][Medline]
Downie JR, 1993. Functions of the foam in foam-nesting leptodactylids: the nest as a post-hatching refuge in Physalaemus pustulosus. Herpetol J 3: 35-42.
Duellman WE, Morales VR, 1990. Variation, distribution, and life history of Edalorhina perezi (Amphibia, Anura, Leptodactylidae). Stud Neotrop Fauna Environ 25: 9-30.
Edgerly JS, McFarland M, Morgan P, Livdahl T, 1998. A seasonal shift in egg-laying behaviour in response to cues of future competition in a treehole mosquito. J Anim Ecol 67: 805-818. [CrossRef]
Emlen ST, Oring LW,
1977. Ecology, sexual selection, and the evolution of mating systems.
Science
197:
215-223.
Farfan C, Peralta AME, Vazquez OT, 1998. Egg laying delay in Modiolus capax (Bivalvia: Mytilidae). Rev Biol Trop 46: 633-641.
Fincke OM, 1992. Consequences of larval ecology for territoriality and reproductive success of a neotropical damselfly. Ecology 73: 449-462. [CrossRef]
Gascon C, 1989. Predator-prey size interaction in tropical ponds. Rev Bras Zool 6: 701-706.
Gascon C, 1992. Aquatic predators and tadpole prey in central Amazonia: field data and experimental manipulations. Ecology 73: 971-980. [CrossRef]
Gascon C, Zimmerman B, 1998. Of frogs and ponds and peccaries. Nat Hist 107: 43-45.
Gosner KL, 1960. A simplified table for staging anuran embryos and larvae with notes on identification. Herpetologica 16: 183-190.
Gross MR, 1991. Evolution of alternative reproductive strategies: frequency-dependent sexual selection in male bluegill sunfish. Phil Trans R Soc Lond B 332: 59-66.
Heimpel GE, Rosenheim JA, Mangel M, 1996. Egg limitation, host quality, and dynamic behavior by a parasitoid in the field. Ecology 77: 2410-2410. [CrossRef]
Heyer WR, Donnelly MA, McDiarmid RW, Hayek L-AC, Foster MS, (eds), 1994. Measuring and monitoring biological diversity. Standard methods for amphibians. Washington, DC: Smithsonian Institution Press.
Howard RD, 1978. The influence of male-defended oviposition sites on early embryo mortality in bullfrogs. Ecology 59: 789-798. [CrossRef]
Johnstone RA, Grafen A, 1993. Dishonesty and the handicap principle. Anim Behav 46: 759-764. [CrossRef]
Johnstone RA, Reynolds JD, Deutsch JC, 1996. Mutual mate choice and sex differences in choosiness. Evolution 50: 1382-1391. [CrossRef][Web of Science]
Kats LB, Dill LM, 1998. The scent of death: chemosensory assessment of predation risk by prey animals. Ecoscience 5: 361-394.
Kiesecker JM, Skelly DK, Beard KH, Preisser E,
1999. Behavioral reduction of infection risk.
Proc Nat Acad Sci USA
96:
9165-9168.
Marsh DM, Borrell BJ, 2001. Flexible oviposition strategies in túngara frogs and their implications for tadpole spatial distributions. Oikos 93: 101-109. [CrossRef]
Marsh DM, Rand AS, Ryan MJ, 2000. Effects of inter-pond distance on the breeding ecology of túngara frogs. Oecologia 122: 505-513. [CrossRef]
Mitchell NJ, 2001. Males call more from wetter nests: effects of substrate water potential on reproductive behaviours of terrestrial toadlets. Proc R Soc Lond B 268: 87-93. [Medline]
Murphy PJ, 2003. How closely does reproductive site choice in a neotropical frog mirror variable mortality risks facing offspring? Ecol Monogr 73: 45-67.
Pruett-Jones S, 1992. Independent versus nonindependent mate choice: do females copy each other? Am Nat 140: 1000-1009. [CrossRef][Web of Science]
Pyburn WF, 1970. Breeding behavior of the leaf-frogs Phyllomedusa callidryas and Phyllomedusa dacnicolor in Mexico. Copeia 1970:209–218.
Rausher MD, 1979. Larval suitability and oviposition preference in three related butterflies. Ecology 60: 503-511. [CrossRef]
Resetarits WJ, Wilbur HM, 1989. Choice of oviposition site by Hyla chrysoscelis: role of predators and competitors. Ecology 70: 220-228. [CrossRef]
Resetarits WJ, Wilbur HM, 1991. Calling site choice by Hyla chrysoscelis: effect of predators, competitors, and oviposition sites. Ecology 72: 778-786. [CrossRef]
Roitberg BD, Mangel M, Lalonde RG, Roitberg CA, Vanalphen JJM, Vet L,
1992. Seasonal dynamic shifts in patch exploitation by parasitic wasps.
Behav Ecol
3:
156-165.
Rosenheim JA, 1999. Characterizing the cost of oviposition in insects: a dynamic model. Evol Ecol 13: 141-165.
Schlüter A, 1990. Reproduction and tadpole of Edalorhina perezi (Amphibia, Leptodactylidae). Stud Neotrop Fauna Environ 25: 49-56.
Sikkel PC, 1995. Effects of nest quality on male courtship and female spawning: site choice in an algal-nesting damselfish. Bull Mar Sci 57: 682-689.
Skelly DK, 1996. Pond drying, predators, and the distribution of Pseudacris tadpoles. Copeia 1996:599–605.
Spieler M, Linsenmair KE, 1997. Choice of optimal oviposition sites by Hoplobatrachus occipitalis (Anura: Ranidae) in an unpredictable and patchy environment. Oecologia 109: 184-199. [CrossRef]
Stamps JA, 1988. Conspecific attraction and aggregation in territorial species. Am Nat 131: 329-347. [CrossRef][Web of Science]
Stav G, Blaustein L, Margalith J, 1999. experimental evidence for predation risk sensitive oviposition by a mosquito, Culiseta longiareolata. Ecol Entomol 24: 202-207. [CrossRef]
Terborgh J, 1983. Five New World primates. Princeton, New Jersey: Princeton University Press.
Trivers RL, 1972. Parental investment and sexual selection. In: Sexual selection and the descent of man (Campell BG, ed). Chicago: Aldine; 139–179.
Tuttle MD, Ryan MJ,
1981. Bat (Trachops cirrhosus) predation and the evolution of frog vocalizations in the Neotropics.
Science
214:
677-678.
Vannoordwijk AJ, McCleery RH, Perrins CM, 1995. Selection for the timing of great tit breeding in relation to caterpillar growth and temperature. J Anim Ecol 64: 451-458. [CrossRef]
Warner RR, Dill LM,
2000. Courtship displays and coloration as indicators of safety rather than of male quality: the safety assurance hypothesis.
Behav Ecol
11:
444-451.
Zar JH, 1996. Biostatistical analysis. Upper Saddle River, New Jersey: Prentice Hall.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  W. J Palen, C. E Williamson, A. A Clauser, and D. E Schindler Impact of UV-B exposure on amphibian embryos: linking species physiology and oviposition behaviour Proc R Soc B, June 22, 2005; 272(1569): 1227 - 1234. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||