Phenotypic plasticity in developmental rate is insufficient to offset high tadpole mortality in rapidly drying ponds S. M. Amburgey,1,3,† M. Murphy,1,4 and W. Chris Funk2 2Department
1Department of Biology, Colorado State University, Fort Collins, Colorado 80523 USA of Biology, Graduate Degree Program in Ecology, Colorado State University, Fort Collins, Colorado 80523 USA
Citation: Amburgey, S. M., M. Murphy, and W. C. Funk. 2016. Phenotypic plasticity in developmental rate is insufficient to offset high tadpole mortality in rapidly drying ponds. Ecosphere 7(7):e01386. 10.1002/ecs2.1386
Abstract. Habitat suitability is strongly regulated by seasonal conditions and stochastic processes, and this is
especially important in temporary aquatic systems that contain organisms with complex life cycles. We investigated the potential for phenotypic plasticity in timing of and size at metamorphosis to mitigate effects of altered habitat conditions, specifically shortened hydroperiod (duration of water in ponds) and altered predator-prey dynamics, in the pond-breeding boreal chorus frog (Pseudacris maculata). We simulated reduced hydroperiod and concentrated predator cue in the laboratory to understand potential benefits and costs of plasticity. Tadpoles developed faster in response to the combined effects of reduced hydroperiod and increased concentration of predator cue, potentially due to reduced conspecific density. In contrast, there was no effect of reduced hydroperiod or predator cue on size at metamorphosis. Alone, this result suggests that phenotypic plasticity may allow P. maculata to escape the negative effects of rapidly drying ponds. However, tadpole survival was significantly lower in reduced hydroperiod treatments relative to all other treatments, suggesting that even if plasticity acts as a buffer against reduced hydroperiod by facilitating metamorphosis, heightened mortality may offset benefits of this rapid response. Our results add to previous studies of plastic responses in amphibians by disentangling the costs and benefits of plasticity in habitats with multiple, simultaneous stressors. We show that while plasticity may accelerate metamorphosis, similar, heightened levels of mortality are experienced regardless of plasticity. This implies that plasticity may not completely buffer populations against the effects of altered habitat conditions, such as those that occur with climate change or urbanization.
Key words: amphibian decline; climate change; hydroperiod; predator cue; Pseudacris maculata. Received 31 December 2015; revised 2 March 2016; accepted 23 March 2016. Corresponding Editor: J. Benstead. Copyright: © 2016 Amburgey et al. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. 3 Present address: Department of Ecosystem Science and Management, Graduate Degree Program in Ecology, The Pennsylvania State University, University Park, Pennsylvania 16802 USA. 4 Present address: Department of Ecosystem Science and Management, Program in Ecology, University of Wyoming, Laramie, Wyoming 82071 USA. † E-mail:
[email protected]
Introduction
times of highly altered habitat conditions (Rudolf and Rödel 2007, Gienapp et al. 2008, Chevin et al. 2010, Anderson et al. 2012), but the roles and costs of plasticity are not entirely understood (Ghalambor et al. 2007). Plasticity may not be selected for in relatively stable environments (Merilä et al. 2000, Kulkarni et al. 2011) and may not be present in all species, as maintaining plasticity can be costly (DeWitt et al. 1998, Jannot 2009, Reed et al.
Natural environments are subject to inherent, stochastic processes that can drastically alter habitat suitability. Organisms whose habitats have changed beyond their physiological tolerance or ecological niche may disperse or die off, resulting in local extinction (Walther et al. 2002). Phenotypic plasticity can prevent local extinction during v www.esajournals.org
1
July 2016 v Volume 7(7) v Article e01386
AMBURGEY ET AL.
2010). Plasticity can also be inadequate when changes in habitat characteristics are extreme (Charmantier et al. 2008, Gienapp et al. 2008, Kelly et al. 2012, Lane et al. 2012). Understanding the potential costs and benefits of plasticity will allow us to better predict biological responses to a wide variety of habitat disturbances (e.g. climate change, urbanization) and their effects on biodiversity and species distributions. Organisms with complex life cycles tied to variable seasonal conditions are particularly sensitive to altered environments. Most amphibians are physiologically tied to water availability for reproduction and larval development (Buckley and Jetz 2007), making hydroperiod, the time water is present, a key variable in determining amphibian breeding success. For pond-breeding amphibians, developmental plasticity may affect their ability to respond to altered water availability and allow larvae to metamorphose when breeding pond hydroperiods are truncated (Donnelly and Crump 1998). Desiccation risk impacts survival of amphibian larvae directly (Pechmann et al. 1989, Richter-Boix et al. 2011), and larval survival is a particularly important factor in persistence of amphibian populations in some species (e.g., Rana temporaria in Biek et al. 2002, Matthews et al. 2013). Some amphibians can plastically adjust larval developmental rates to escape drying ponds and increase survival (Denver et al. 1998, Loman and Claesson 2003, Gervasi and Foufopoulos 2008). While this plasticity may provide a buffer against seasonal variability, accelerated developmental rate has been associated with tradeoffs in size at metamorphosis (Denver et al. 1998, Merilä et al. 2000), which may compromise survival and fitness in juveniles and adults (Terentyev 1960, Rudolf and Rödel 2007, Márquez-García et al. 2009). Moreover, stress from accelerated development or suboptimal habitat conditions may increase larval mortality (Newman 1992, Relyea and Mills 2001), making any buffer created by plasticity insufficient to ameliorate the detrimental effects of altered habitat conditions. In natural habitats, the effects of altered hydroperiod on amphibian development are complicated additionally by changes in community interactions (Gilman et al. 2010). Predator cue, either from the predator or dying conspecifics (Petranka et al. 1987), can trigger plastic responses in tadpole development (Werner 1986). Skelly and v www.esajournals.org
Werner (1990) showed accelerated development and smaller size at metamorphosis with predators, while other studies have shown equivalent or slower developmental rates with predator cue and an equal or larger resulting metamorphic size (Relyea 2007). Slower development might indicate dampened foraging to avoid detection (Altwegg 2002) but may provide more time for growth (Laurila and Kujasalo 1999). Hydroperiod and predator cue may also interact and alter development. As a pond dries, one prediction is that predator cues may increase in concentration in the remaining habitat, altering prey behavior (Mirza et al. 2006). Reducing hydroperiod and increasing predator cue may interact to stress amphibian larvae, resulting in faster development and metamorphosis (Werner 1986, Rowe and Ludwig 1991). However, simultaneous exposure to predators and reduced hydroperiod may also slow metamorphosis by behavioral suppression of foraging (Altwegg 2002). The combined effect of multiple stressors, from either scenario, could result in increased mortality (Altwegg 2002). Despite the plausible benefit of developmental plasticity, it is likely limited and may be unable to overcome multiple stressors and the overall effects may still be negative. Studies have examined plastic responses to reduced water levels, but the costs and benefits of phenotypic plasticity when simultaneously considering both altered hydroperiod and predation risk are less clear (see Skelly 1995, Lane and Mahony 2002). Although plasticity in larval developmental rate has been documented in some amphibian species, the potential impact of rapid development and multiple stressors on survival remains poorly understood (Newman 1992, Altwegg 2002). Here, we examine the role of plasticity as a buffer in the pond breeding boreal chorus frog (Pseudacris maculata). We assessed survival, time to metamorphosis, and size at metamorphosis in P. maculata tadpoles experiencing (1) truncated hydroperiod, (2) predator presence represented by a non-lethal predator cue, and (3) combined truncated hydroperiod and predator cue.
Methods and Materials Study species and field sampling
We used P. maculata because they primarily inhabit ephemeral wetlands that experience
2
July 2016 v Volume 7(7) v Article e01386
AMBURGEY ET AL.
Fig. 1. Experimental design. Wild-caught tadpoles from nine ponds were randomly assigned to five experimental treatments. Three replicates of each treatment per pond were used (3 replicates per treatment × 9 ponds = 27 replicates of each treatment). Water is gray while predator cue is represented by dots. Concentrations of predator cue were equivalent in the Interaction and Pred-Ratio treatments, though water volume decreased in the Interaction treatment.
seasonal variation in hydroperiod (Hammerson 1999). Breeding pond availability is dependent on snowpack and timing of snowmelt in these habitats (Corn 2005), such that hydroperiod, breeding, and larval success are likely to vary annually and phenotypic plasticity may play an important role in population persistence. Dispersal ability is limited in this species (Spencer 1964) and unlikely to be a viable response to altered habitat conditions. Early stage (Gosner stage 23–25; n = 540 tadpoles; Gosner 1960) P. maculata tadpoles were collected from ponds across mid (n = 4; 1923– 2432 m) and high elevations (n = 5; 2513–3014 m) in Larimer County, Colorado, USA from mid to late June 2010 due to asynchronous breeding across elevation. Ponds were on public lands occupied by P. maculata adults and early Gosner stage tadpoles. Odonate larvae are prevalent in ponds across Larimer County, and all collection sites had odonate larvae. The hydroperiod at each site (“natural hydroperiod”) was estimated during previous visits (2009) and biweekly visits tracking depth from the date of collection through August 2010. Of the four mid elevation sites, two were categorized as ephemeral (drying or nearly dry by the end of August) and two as permanent (retaining water through the end of August). One of the high elevation sites was classified as ephemeral while all others were permanent (Appendix S1). All sites retained odonate larvae of varying stages the e ntire season. v www.esajournals.org
Experimental design and animal care
Tadpoles from each pond were assigned randomly to five experimental treatments with three replicates each (9 ponds × 5 treatments × 3 replicates per treatment per pond × 4 animals/container = 540 tadpoles; Fig. 1). The Control treatment had a constant 1.5 L of water (approximately 6 cm deep) and no predator cue added for the entire experiment. The Hydroperiod-Reduction treatment was a constant hydroperiod reduction, decreasing by 350 mL every water change. The Constant-Pred treatment had a constant 1.45 L of water and a constant 50 mL of predator cue added (1.5 L volume). The Interaction treatment combined predator cue and hydroperiod reduction, involving a reduction of 350 mL of water and the addition of a constant 50 mL of predator cue every change (thus an increased ratio of predator cue to water as the weeks passed). The Pred-Ratio treatment had a constant 1.5 L volume but received the same proportion of predator cue as the Interaction treatment (thus the concentration of predator cue was similar to the Interaction treatment, but the volume of water remained constant). This was used to disentangle the effect of increasing concentration of predator cue without the combined effect of water reduction. Tadpoles were maintained in the laboratory with food and fresh water until treatments began, within 1–2 d after collection. Animals were fed daily ad libitum (to limit food competition) rabbit pellets (0.2 g) and chopped, raw organic spinach (0.1 g) (Lemmon and Lemmon 2010). In water 3
July 2016 v Volume 7(7) v Article e01386
AMBURGEY ET AL.
reduction treatments, water levels were reduced until they reached a final volume of 100 mL (approximately 1 cm deep), enough water to keep tadpoles partially submerged. Treatments were maintained until animals had metamorphosed or died, and dates of metamorphosis and mortality were recorded for all animals. Tadpoles (n = 4 per container) were housed in containers measuring 22 cm × 22 cm × 9.5 cm with ventilated lids (Lemmon and Lemmon 2010). De-chlorinated (AmQuel plus-treated), pH- neutral tap water was used (Lemmon and Lemmon 2010). Containers were randomized on 21 shelves, maintained at room temperature (21–25°C) and seasonal light cycle (14 h light, 10 h dark). Due to constraints on making cues (see below) and to limit disturbance of tadpoles, all replicates were given new containers with fresh water and applicable water reduction and/ or fresh predator cue every 5 d (to prevent predator cue degradation; Peacor 2006). Predator cue.—A pilot study revealed that tadpoles (n = 4 per container, three replicates) exposed to unfed odonate predator cue suppressed activity immediately with gradual (i.e., over 10 h) resumption of normal activity (Amburgey, unpublished data) as compared to tadpoles where only treated water was added to containers. P. maculata were present at all odonate larvae (OL) collection sites, so OL likely fed on tadpoles prior to collection, but OL were not fed in the laboratory. OL were collected from three sites, two at low elevation (1496–1519 m) and one at high elevation in Larimer County (Appendix S1), 48 h before water changes. Genera of Odonata differ by elevation in Larimer County and to remove the possibility that tadpoles only respond to local OL, we used a mixture from different elevations (all from the suborder Anisoptera; Appendix S2). We used OL from low elevation to make the predator cue in even weeks and a combination of half low and half high elevation OL in odd weeks. All OL were placed in individual containers with mesh lids to prevent intraspecific predation and were soaked in water for 48 h to make predator cue water (Eklöv 2000, Van Buskirk and McCollum 2000). A ratio of 1 OL to 100 mL water was used to produce predator cue water. After predator cue was produced, OL were preserved in 70% ethanol. Specimens were identified to family and genus (Appendix S2; Needham et al. 2000, Tennessen 2008). v www.esajournals.org
Data analysis
Correcting for starting stage.—Starting Gosner stage was assessed via individual tadpole photographs (stage 23–25; Gosner 1960) prior to randomization into treatments. Because initial developmental stage may influence time to and size at metamorphosis, a linear regression was used to establish the amount of variation exp lained by initial Gosner stage for each variable (SAS Institute 2007). Time to metamorphosis was correlated positively with initial Gosner stage (adjusted r2 = 0.288, df = 129, P = 0.030). Therefore, average starting stage was included as a cov ariate in all further models assessing time to metamorphosis but was not treated as a response variable. Size at metamorphosis was not cor related significantly with initial Gosner stage (adjusted r2 = −0.004, df = 129, P = 0.510), thus the raw size data were used. Significance was measured at the 0.05 level for all analyses. Time to metamorphosis.—Containers were checked daily for metamorphosed individuals, defined as any animal with both hind limbs and at least one front limb emerged (Lemmon and Lemmon 2010). Time to metamorphosis (days) was averaged by container (n ≤ 4 individuals/container; 135 containers). Data were checked for normality and log transformed prior to analysis. We ran a mixed effects ANOVA including initial starting stage, treatment, natural hydroperiod, and a treatment by natural hydroperiod interaction as fixed effects and source pond as a random effect (Eq. 1; Zar 1998, SAS Institute 2007).
log(Time) = Stage + Trt + NatHydro (1) + NatHydro × Trt + |Pond| In the mixed effects ANOVA, means were compared to the Pred-Ratio treatment and Natural Hydroperiod 2 (permanent hydroperiod). This allowed for us to investigate the effects of constant water, drying, and predator cue when treatment means were compared. Size at metamorphosis.—Animals were photo graphed with a size standard (ruler) at metamorphosis. We measured snout-vent length (SVL) of all individuals as an index of size at metamorphosis using ImageJ (Rasband 2011) and averaged measurements by container (n ≤ 4 individuals/container; 135 containers). Data were 4
July 2016 v Volume 7(7) v Article e01386
AMBURGEY ET AL.
checked for normality prior to analysis. We ran a mixed effects ANOVA including treatment, natural hydroperiod, and a treatment by natural hydroperiod interaction as fixed effects and source pond as a random effect (Eq. 2; Zar 1998, SAS Institute 2007).
Size =Trt + NatHydro + NatHydro × Trt + |Pond|
(2)
Means were compared to the Pred-Ratio treatment and Natural Hydroperiod 2 (permanent hydroperiod). Survival.—Individual mortalities were recorded daily for each container (n = 540 individuals). Because mortality occurred equally throughout replicates, the effect of container was excluded in the analysis. We used a general linear mixed model with a source pond random effect to examine the response of survival to experimental treatment, natural hydroperiod, and any interaction between the two (Eq. 3; Zar 1998, R Development Core Team, 2009).
Survival = Trt + NatHydro + NatHydro × Trt + |Pond|
(3)
Means were compared to the Pred-Ratio treatment and Natural Hydroperiod 2 (permanent hydroperiod). Effect of altered density.—To better understand the role of the experimental treatments vs. the potential effects of altered tadpole density, we analyzed time to and size at metamorphosis using a mixed effects ANOVA (analogous to the above) on each response variable except each model also included the effect of average survival by container (Zar 1998, SAS Institute 2007). This allowed for us to further test the respective importance of experimental treatments and the role altered density played in observed responses.
Fig. 2. Violin plots of (a) average time to metamorphosis and (b) average size at metamorphosis (SVL) by treatment. White markers indicate median values while black bars indicate interquartile range. Probability densities of the data are in grey and represent the probability distribution of data values for each treatment. The asterisk indicates a significant difference between time to metamorphosis and the Interaction treatment from the mixed effects ANOVA including average survival.
Results treatment tadpoles metamorphosing fastest (P = 0.002; Table 2). Source pond explained 29.6% of the variation in time to metamorphosis (estimate = 0.021, SE = 0.014 d). Time to metamorphosis was not correlated with natural hydroperiod or most of the interactions between treatment and natural hydroperiod
Time to metamorphosis
Tadpoles in the Interaction treatment and the Hydroperiod-Reduction treatment developed fastest (Fig. 2a). In the mixed effects ANOVA utilizing Gosner-corrected measures, treatment affected time to metamorphosis (F = 3.99, df = 4, P = 0.005; Table 1), with the Interaction v www.esajournals.org
5
July 2016 v Volume 7(7) v Article e01386
AMBURGEY ET AL. Table 1. Mixed-model ANOVA table of fixed-effects tests on log-transformed time to metamorphosis and Experimental Treatment and Natural Hydroperiod, controlling for the random effect of source pond of origin. Parameter Experimental Treatment Natural Hydroperiod Natural Hydroperiod × Experimental Treatment Stage
df
F
P
4 1 4
3.99 0.032 1.18
0.005* 0.864 0.325
4
1.22
0.308
Survival
Only 73% of the tadpoles survived and metam orphosed. Survival was 50% in the Hydro period-Reduction treatment and 55% in the Interaction treatment (Fig. 3), compared to 88.9% in the Control treatment. These two treatments accounted for approximately 74% of total experimental mortality, with other treatment mortalities between 9% and 20%. In the generalized linear mixed model, the Hydroperiod-Reduction treatment had a significant negative effect on survival (P 0.1; Tables 1 and 2). However, the interaction between ephemeral natural hydroperiod and the Constant-Pred treatment was significant (P = 0.045) with these tadpoles taking less time to metamorphose.
Size at metamorphosis
Effect of altered density
Average size at metamorphosis varied little among treatments (1.10 cm ± 0.05 cm; Fig. 2b; Table 3) although Interaction treatment individuals were smallest on average. In the mixed effects ANOVA, few of the variables and interactions had an effect on size at metamorphosis (P > 0.1; Tables 3 and 4). The interaction between ephemeral natural hydroperiod and Constant-Pred treatment was significant (P = 0.045; Table 4) with tadpoles metamorphosing at a larger size. The random effect of source pond accounted for only 8.63% of the variation (estimate = 2.53e−4, SE = 2.4e−4 cm).
In the mixed model ANOVA including the effect of average survival, overall treatment was no longer significantly related to time to metamorphosis, although the P-value was just slightly higher than the a priori significance level of α = 0.05 (F = 2.42, df = 4, P = 0.053; Table 6). However, when comparing the least means square output of experimental treatments, the Interaction treatment tadpoles still took a significantly shorter amount of time to metamorphose than other treatments (P = 0.006; Table 7). Source pond explained 36.1% of the variation in time to metamorphosis (estimate = 0.027, SE = 0.017 d). Time
Table 2. Individual parameter estimates from the mixed-model ANOVA of log-transformed time to metamorphosis and Experimental Treatment and Natural Hydroperiod (1 = ephemeral hydroperiod), controlling for the random effect of source pond of origin. In the analysis, all treatments are compared to the Pred-Ratio treatment and Natural Hydroperiod 2 (permanent hydroperiod). Interaction represents the Interaction treatment. Parameter
Estimate
SE
t ratio
P
(Intercept) Control Hydroperiod-Reduction Constant-Pred Interaction Natural Hydroperiod 1 Natural Hydroperiod 1 × Control Natural Hydroperiod 1 × HydroRed Natural Hydroperiod 1 × ConstPred Natural Hydroperiod 1 × Interaction
2.67 0.026 −0.055 0.059 −0.126 −0.010 0.011 0.042 −0.081 −0.008
0.176 0.040 0.042 0.040 0.040 0.053 0.041 0.043 0.040 0.042
15.2 0.66 −1.29 1.48 −3.12 −0.18 0.27 0.98 −2.03 −0.20