Wetlands

, Volume 33, Issue 4, pp 707–715

Effects of Road Salts on Seasonal Wetlands: Poor Prey Performance May Compromise Growth of Predatory Salamanders

Authors

    • Department of BiologyUniversity of North Carolina at Asheville
  • Robert A. Francis
    • Department of BiologyUniversity of North Carolina at Asheville
Article

DOI: 10.1007/s13157-013-0428-7

Cite this article as:
Petranka, J.W. & Francis, R.A. Wetlands (2013) 33: 707. doi:10.1007/s13157-013-0428-7

Abstract

Road deicing agents that enter wetlands can affect amphibians both directly via their toxic effects and indirectly by altering food web interactions. We conducted experiments to determine whether larvae of the spotted salamander (Ambystoma maculatum) are more strongly influenced by direct versus indirect effects of salt concentration. Using outdoor mesocosms, we exposed salamanders and their prey to experimental salinities that were representative of values reported from salt contaminated breeding sites in North America. Increasing salinity depressed salamander growth but did not affect survival. Cladocerans were numerical dominants in samples taken 2 weeks after the experiment began, and markedly declined with increasing salinity. The number of cladocerans and total number of all invertebrates on this date were positively correlated with the mean mass of salamanders at the termination of the experiment. In a laboratory experiment where food was supplied in excess, increasing salinity did not affect the growth or survival of spotted salamanders that were chronically exposed to salinities that paralleled those in the mesocosm experiment. Our results suggest that spotted salamander larvae are more salt-tolerant than their prey, and that salamander growth may be compromised via indirect effects when breeding sites experience moderate salt contamination (i.e., < 1,000 mg L−1 of salts).

Keywords

Road saltsAmphibiansIndirect effectsFood websAmbystoma maculatumZooplankton

Introduction

Salinization of fresh water is a substantial threat to aquatic life, particularly in arid or semi-arid regions of the world (Williams 2002; Nielsen et al. 2003), and at northern latitudes where chemical deicers are applied to roadways to prevent ice formation (Findlay and Kelly 2011; Jin et al. 2011). Winter deicing agents consist mainly of sodium chloride, and are readily transported from roadways to nearby wetlands and streams where they can reduce the biomass and diversity of aquatic organisms and alter community composition (Hart et al. 1991; Environment Canada 2001; Kaushal et al. 2005; Karraker et al. 2008). Changes in community composition likely reflect the direct loss or decline of salt-intolerant taxa (Hart et al. 1991; Nielsen et al. 2003; Waterkeyn et al. 2008), along with altered ecological and behavioral interactions among salt-tolerant species (Petranka and Doyle 2010; Van Meter et al. 2011).

In central and eastern North America, seasonal freshwater wetlands such as fens, vernal pools, and marshes are the primary breeding habitats for many species. These habitats support unique faunal elements that are susceptible to salt contamination from road deicers (Brock et al. 2005; Karraker et al. 2008; Waterkeyn et al. 2008). It was previously thought that spring rains rapidly flushed out a majority of the NaCl from wetlands that adjoin roads, but this is not the case. Although NaCl enters wetlands in solution, some gets incorporated into the soil as Na+ evicts other cations from clay particles (Jackson and Jobbágy 2005; Findlay and Kelly 2011; Jin et al. 2011). The Na+ and Cl ions that are not retained within the soil collect in fresh water, and in some cases may concentrate as seasonal wetlands undergo annual drying (Turtle 2000; Kaushal et al. 2005).

Amphibians are vulnerable to salt-contamination because they have permeable skin. In addition, many species breed during the late winter/spring warm-up when the run-off of salt-laden water from roadways is highest (Sanzo and Hecnar 2006; Corsi et al. 2010; Findlay and Kelly 2011). Salt contamination can cause developmental anomalies (Karraker 2007; Karraker and Ruthig 2009), increase egg and larval mortality (Turtle 2000; Sanzo and Hecnar 2006; Karraker et al. 2008), reduce larval growth rates, and alter larval behavior (Christy and Dickman 2002; Winkler and Forte 2011). Although these direct (toxic) effects are well documented, the indirect effects of salinization on amphibians are poorly understood.

Freshwater organisms vary markedly in their salt tolerance, with larval insects generally appearing to be more salt-tolerant than either zooplankton or amphibians (Benbow and Merritt 2004; Sarma and Nandini 2006; Sarma et al. 2006; Petranka and Doyle 2010; Karraker and Gibbs 2011). The selective loss of salt-intolerant species from a community can produce a variety of indirect effects related to altered food webs. Because of the complexity of food web interactions, salinization might even enhance the growth of amphibian larvae (e.g., Petranka and Doyle 2010; Van Meter et al. 2011).

Indirect effects may be particularly important for predators because salinization can alter the size of prey populations and the composition of prey communities. Salamander larvae function as top predators in seasonal freshwater wetlands at northern latitudes and rely heavily on zooplankton as a dietary staple (Petranka 1998). Cladocerans and copepods are important prey in these systems and appear to be more salt-sensitive than other taxa (e.g., Sarma and Nandini 2006; Sarma et al. 2006; Petranka and Doyle 2010; Van Meter et al. 2011). This raises the possibility that deicing compounds could indirectly affect salt-tolerant predators by reducing their overall food supply. In addition, non-lethal effects in salt-intolerant predators (e.g., mildly reduced growth due to physiological stress) could be amplified by the loss of salt-intolerant prey.

The larvae of mole salamanders (Ambystoma spp.) are important community members in seasonal wetlands in eastern North America where they function as upper-level predators on zooplankton, insects, and other aquatic prey (Petranka 1998). Here we examine the effects of a commercial road deicer on the growth and survival of the spotted salamander (Ambystoma maculatum). This species is widespread in eastern North America and is vulnerable to salinization from road deicers (Turtle 2000; Karraker et al. 2008; Veysey et al. 2011). Our primary goal was to determine whether direct or indirect effects are more likely to adversely affect the growth and/or survival of spotted salamander larvae when exposed to a range of salinities that are typical of contaminated breeding sites in eastern North America.

Methods

We conducted an outdoor mesocosm experiment to determine the effects of salinity on the growth and survival of spotted salamander larvae and the relative densities of their invertebrate prey. We also conducted a laboratory experiment to help interpret the results of the mesocosm experiment with respect to the relative importance of direct versus indirect effects of salinity. We used a regression design (Cottingham et al. 2005) in which numerous experimental treatments were established without replication to examine performance across a salinity gradient. Salinities in contaminated seasonal wetlands near roads are generally < 1,000 mg L−1, but may exceed 4,000 mg L−1 (Environment Canada 2001; Sanzo and Hecnar 2006; Karraker et al. 2008). We used a range of salinities that encompassed the majority of values that have been reported for contaminated wetlands; most of our experimental salinities were < 600 mg L−1 (equivalent to 352 mg L−1 of Cl) and the highest values were < 2,500 mg L−1 (1,465 mg L−1 of Cl). All aspects of the experiments were randomized, including the experimental salinities that were assigned to containers and pools, and the assignment of salamander hatchlings to containers or pools.

Commercial deicers typically consist of either NaCl (rock salt) or mixtures of NaCl, CaCl2, KCl, or MgCl2. An anti-clumping agent is sometimes added (Karraker et al. 2008). We used Road Runner Ice Melt® (Scotwood Industries, Inc.) to facilitate comparisons with a previous study (Petranka and Doyle 2010). This commercial product lacks anti-clumping agents and has a base of NaCl and KCl, with relatively small amounts of CaCl2 (10 %) and MgCl2 (5 %) to enhance melting. Cl constituted ~ 58 % of the mass of the deicer (Petranka and Doyle 2010). Here, we use the terms ‘salts’ and ‘salt concentration’ to refer to the deicing agent and the concentration of the deicing agent. To facilitate comparisons with other studies, we measured the Cl content of the deicer based on titrations with silver nitrate (Mohr’s Method). We measured conductivity of solutions with a Corning Checkmate 90 Portable Meter and converted conductivity (μS/cm) to salt concentration (mg L−1) using a regression equation derived from measurements of solutions with known quantities of the deicing agent [P < 0.001; r2 = 0.999]. The regression line was fitted through the origin to avoid negative values for salt concentration.

Experiment 1: Effects of Salinity on Prey Abundance and Ambystoma Performance

Road salts that contaminate seasonal pools could compromise the growth and survival of Ambystoma larvae either directly via physiological stress (toxicity) or indirectly by reducing their food supply. The primary goal of this experiment was to document how spotted salamander larvae and their invertebrate prey perform across a salinity gradient. Local populations of the spotted salamander begin breeding from late December-early May in association with the winter/spring warm-up, and a wave of breeding generally progressing from the southern US (December-February) to the northern US and southern Canada (March-early May). Small larvae feed almost entirely on zooplankton such as cladocerans and copepods, while older larvae feed on zooplankton as well as larger prey such as chironomids, chaoborids and isopods (Petranka 1998).

Experiment 1 was conducted in 2009 using mesocosms that were established in 42 plastic wading pools (0.9 m in diameter, 20 cm deep; 0.64 m2). Pools were placed on 4 March in a partially shaded field in Buncombe Co., North Carolina and filled with ~ 84 L of tapwater. We then added 126 g of dried southern red oak leaves (Quercus falcata) and 3 g of powdered commercial rabbit chow to each pool to provide physical structure and nutrients to the system. On 7 March we collected zooplankton and other invertebrates from three local seasonal ponds (all within 30 m of each other) using a fine-mesh 300 μm net. These were placed in a bucket with pond water and mixed thoroughly before adding ~ 300-ml aliquots of pond water with concentrated zooplankton to each pool. We added three additional grams of rabbit chow on 17, 24, and 31 March to provide a continuing nutrient base for the mesocosm communities.

We left pools undisturbed for 3 weeks to allow phytoplankton and zooplankton communities to develop, and midges and other aquatic insects to oviposit in pools. Copepods and large numbers of cladocerans (primarily Daphnia and Ceriodaphnia) were visible in pools within 2 weeks after adding zooplankton. After 3 weeks (29 March), we randomly assigned salt treatments to pools. These included seven control pools that did not receive supplemental salts (range of initial values = 30–43 mg L−1 of salts) and 35 pools that received different experimental quantities of the deicer. Initial salt concentrations in pools that received deicer based on 30 March readings varied from 100 to 2,468 mg L−1 of salt (equivalent to ~ 59–1,447 mg L−1 of Cl and conductivities of 169–4,166 μS cm−1) and were similar to values reported from salt-contaminated streams and seasonal wetlands near roadways (e.g., Environment Canada 2001; Karraker et al. 2008; Corsi et al. 2010). We included control pools in order to calculate a mean value that would serve as a reference point for judging declines in cladocerans, which were known to be sensitive to the deicer (Petranka and Doyle 2010).

After adding salts, we randomly selected and added 20 spotted salamander hatchlings to each pool (all hatchlings < 5 days post-hatch). These were selected from a large group of hatchlings that were obtained from 16 egg masses that were collected on 7 March from a seasonal pond in Buncombe Co., North Carolina and maintained in an outdoor wading pool through hatching. We established hatchlings at a density (31 m−2 of pool bottom or 0.24 L−1) that would likely result in competition for food resources (Burley et al. 2006), since one of our goals was to examine how salt concentration could indirectly influence density-dependent growth and survival through resource competition. Densities of hatchlings in natural ponds may sometimes exceed 100 larvae m−2 of pond bottom (Karraker et al. 2008).

We left pools uncovered for the duration of the experiment to allow colonization by insects. One concern in having uncovered pools was that dragonflies might colonize and feed on spotted salamander larvae. However, we have never observed ovipositing dragonflies at the site in past years and no dragonfly larvae were recovered from our invertebrate samples (see results).

Near the end of the experiment, we inspected pools daily to determine whether larvae were undergoing metamorphosis. We terminated the experiment 1 day after the first metamorphosing larva was observed, and 50 days after hatchlings were added to the pools. Because water levels and salt concentrations fluctuated in pools due to evaporation and rainfall, we could not precisely control salinity during the experiment. To provide more realistic estimates of the conditions that organisms experienced, we measured conductivities on five dates (30 March; 3 April; 12 April; 9 May; 18 May) with a Corning Checkmate 90 Portable Meter.

We sampled invertebrates 14 days after adding hatchlings (12 April) and at the termination of the experiment (18 May; Day 50). Because cladoceran populations tend to decline seasonally as water temperatures increase, we included the early sample to examine the effects of salts on this important prey group. To determine the relative abundance of invertebrates, we made eight standardized sweeps with a fine-mesh plankton net (300 μm; 15 × 13 cm) from the midpoint to the edge of each pool. Sweeps were taken at standardized locations (cardinal and subcardinal points) for all pools and the net was sufficiently deep to sample the entire water column.

Specimens were preserved, then identified as cladocerans (Cladocera), copepods (Copepoda), midges (Chironomidae), or miscellaneous taxa. Miscellaneous taxa constituted < 2 % of the invertebrates in samples and included mosquito larvae, beetle larvae, and hydras. Lower taxonomic identification was not feasible due to the shear number of specimens in samples (total for all samples > 140,000). We subsampled if a sample contained exceptionally large numbers of invertebrates (i.e., when we estimated there to be > 1,000 individuals in a sample). To subsample, we placed invertebrates from a sample in 500 ml of water, shook the solution vigorously to suspend the animals, and then poured 10 % of the solution off for analysis. All but 16 % of samples were subsampled in this manner.

We removed all spotted salamander larvae from pools on the last day of the experiment (18 May; Day 50) to determine the effects of salinity on growth and survival. After sampling invertebrates (see above), leaf litter was removed from each pool and carefully searched for larvae. The water in each pool then was poured through a net and all remaining larvae were counted. We expressed survival as the percentage of larvae out of the initial 20 that survived to Day 50. To determine the mean mass of larvae in each pool, we briefly blotted larvae on paper towels, then collectively weighed all larvae on an electronic balance.

Experiment 2: Direct Effect of Salts on Spotted Salamander Larvae

We conducted a laboratory experiment to help interpret the results of the mesocosm experiment. The primary purpose was to determine whether chronic exposure to salts would reduce the growth or survival of spotted salamander larvae that were supplied food in excess. If so, it would provide evidence that physiological stress (toxicity) contributed to reduced growth in the mesocosm experiment (see below).

On 5 March 2012, we filled each of 41 plastic containers (17 × 30 × 11 cm) with 3 L of pond water that was filtered through a fine-mesh net (300 μm) to remove any macroinvertebrates. The water was from one of the same seasonal ponds that were the source of zooplankton used in the mesocosm experiment (salt concentration = 51.6 mg L−1). We added 75–2,300 mg L−1 of Road Runner Ice Melt® in concentrations that paralleled those used in the mesocosm experiment (equivalent to 44–1,348 mg L−1 of Cl and conductivities of 127–3,882 μS cm−1). On 12 March, we randomly assigned and added five spotted salamander larvae (Harrison stages 41–45; in Duellman and Trueb 1986) to each container. The larvae were < 5 days post-hatch and were drawn from a pool of individuals that hatched from seven egg masses that were from the same breeding site as those used in the mesocosm experiment. The containers were maintained at room temperature that varied from 19.9 C (5 March) to 21.7 C (3 April). They were fitted with lids to minimize evaporative loss, and bubbled with air to oxygenate the water.

We added equal increments of zooplankton (consisting mainly of Daphnia and Ceriodaphnia) every 2 days to ensure that the larvae were at or near satiation. To minimize variation in the amount of food added to containers, we collected zooplankton with a fine-mesh net from the same pond that served as the water source, concentrated the entire catch in 9–10 L of pond water, then mixed the solution thoroughly before withdrawing two ~100 mL units of water for each container. To prevent dilution of the salt solutions, we poured the samples through a 300 μm net to remove most water before adding the zooplankton to the containers. Every 2 days, we removed salamander larvae from each container, filtered the water through a fine mesh net to remove fecal matter and zooplankton, then poured the water into a duplicate clean container and added freshly caught zooplankton. We monitored larvae and zooplankton daily to ensure that live zooplankton were present in each container and that larvae were at or near satiation as evidenced by their swollen guts.

To facilitate comparisons with the mesocosm study, we terminated the experiment on day 25 (5 April) when the mean mass of larvae approximated the mass of larvae at the termination of the mesocosm study (respective means ± SE for lab experiment and mesocosm experiment = 0.56 ± 0.018 g and 0.54 ± 0.14 g; t-test: t = 0.70; df = 80; P = 0.62). We estimated the mean mass of larvae on 26 March (Day 15) and 5 April (Day 25) after blotting larvae on paper toweling and weighing the aggregate.

We estimated the salt concentrations in the containers on 5 March, 15 March, 22 March, 30 March, and 3 April based on conductivity measurements using a Corning Checkmate 90 Portable Meter (see above). Mean concentration did not differ significantly between sample dates (ANOVA: F4,205 = 0.15; P = 0.95) and we used the grand mean for all dates in statistical analyses.

Statistical Methods

Salt concentrations in pools changed due to evaporation and rainfall (see results). To provide more realistic representations of the salinities that aquatic organisms experienced, we analyzed trends for invertebrate samples taken on 12 April using the mean values for the first three measurements of salt concentration (30 March; 3 April; 12 April). We analyzed trends for the final invertebrate sample of 18 May using the means for all five measurements. We utilized least-squares regression analysis to examine the effects of salinity on the growth and survival of larvae, and Pearson’s correlation analysis to examine relationships between prey abundance and larval growth. In Experiment 1, some of the relationships between salinity and prey abundance were non-linear, and we used second-order least-squares polynomial models to fit trend lines that illustrated general patterns in the data. Polynomial models were only used if they significantly increased the r2 value relative to a linear regression model. In Experiment 2, one or more larvae in six containers died or experienced cannibalism during the experiment. Because of selective mortality, the mean masses for these containers may have been biased and were not use in the growth rate analysis. Exclusion of these data points did not affect the statistical significance of any of the analyses. We used the mean mass of larvae for each mesocosm or container as a proxy for mean growth rate.

Results

Experiment 1: Effects of Salinity on Prey Abundance and Ambystoma Performance

Salinities in pools increase by 6.4 % between 30 March and 12 April due to evaporation, then decline sharply after mid-April due to increased pool volume from rain events. The respective means (± 1 SE) for 30 March, 3 April, 12 April, 9 May, and 18 May were 831 ± 112, 873 ± 120, 884 ± 128, 575 ± 81, and 487 ± 68 mg L−1 salts. Relative to the initial values on 30 March, salinities were significantly higher on 3 April (paired t-test; t = −3.90; df = 41; P = 0.0004) and 12 April (t = −2.24; df = 41; P = 0.03), and significantly lower on 9 May (t = 7.47; df = 41; P < 0.0001) and 18 May (t = 7.35; df = 41; P < 0.0001). The decline after mid-April was similar to declines that often occur in natural habitats following the late-winter/spring warm-up (e.g., Corsi et al. 2010).

Invertebrate samples on 12 April and 18 May yielded an estimated 74,595 and 65,973 organisms. Cladocerans, midges, and copepods comprised 66 %, 29 %, and 5 % of all invertebrates on 12 April, respectively. Midges were the numerical dominants on 18 May (77 % of all invertebrates), followed by cladocerans (16 %) and copepods (7 %).

The relative abundance of all taxa varied significantly across the salinity gradient (P < 0.05); however, patterns differed between taxa and sampling periods (Fig. 1). The relative abundance of cladocerans markedly declined with mean salinity. On 12 April (Day 14 after salt additions), cladocerans were present in all pools with mean salinities < 2,390 mg L−1 salts (~ 1,401 mg L−1 of Cl and 4,034 μS cm−1), but numbers were reduced to 50 % of the mean for control pools at approximately 800 mg L−1 (469 mg L−1 of Cl and 1,350 μS cm−1). On 18 May (Day 50), cladocerans were either absent (71 % of pools) or uncommon (≤ 30 individuals in a sample) in samples from pools with mean salts concentrations > 420 mg L−1 (246 mg L−1 of Cl and 709 μS cm−1). In contrast, cladocerans were absent in only 17 % of pools with salinities < 420 mg L−1 and averaged 589 ± 232 individuals per sample (mean ± 1 SE).
https://static-content.springer.com/image/art%3A10.1007%2Fs13157-013-0428-7/MediaObjects/13157_2013_428_Fig1_HTML.gif
Fig. 1

Relationship between mean salinity (mg L−1) and estimated number of invertebrates in samples from 42 outdoor mesocosms. Invertebrate samples were collected 14 days (12 April; left column) and 50 days (18 May; right column) after road deicer was added to pools. Mean salinities are based on measurements made from 30 March – 12 April 2009 for the 12 April samples and from 30 March-18 May 2009 for the 18 May samples of invertebrates. Data were fit using either linear (df = 40) or second-order (df = 39) polynomial models

Copepods were less sensitive to salinity than cladocerans, but showed a weak tendency to decline with increasing salinity (Fig. 1; P = 0.03 for both 12 April and 18 May samples). Mean salinity explained only 16 % and 11 % of the variation in copepod numbers for the two sampling dates, suggesting that other factors such as biotic interactions with other community members were more important. The number of midges was independent of salinity for 12 April samples (Fig. 1; df = 40; P = 0.10; r2 = 0.07), but in 18 May samples the number of midges tended to increase with salinity in pools with mean salinities between 28 and 1,000 mg/L−1 (second-order polynomial regression; df = 39; P = 0.005; r2 = 0.24). A combined analysis of all invertebrates revealed that the total number of invertebrates declined significantly with salinity on 12 April (df = 39; P < 0.0001; r2 = 0.75), but was independent of salinity (df = 40; P = 0.74; r2 = 0.002) on 18 May (Fig. 1; lowermost panels).

Survival of spotted salamander larvae to the termination of the experiment (Day 50; 18 May) varied from 30 to 100 % among pools (Mean = 76 ± 2.7 %). We did not detect an influence of salinity on survival (Fig. 2; df = 40; P = 0.96; r2 = 0.001). In contrast, the average growth rates of larvae decreased significantly with salinity (df = 40; P < 0.0001; r2 = 0.33), with larvae at the highest salinities growing about 33 % slower than those at the lowest salinities. The mean mass of larvae at the termination of the experiment was positively correlated with the number of cladocerans (Pearson’s correlation; df = 40; P = 0.0006; r = 0.51) and total number of all invertebrates (df = 40; P = 0.0004; r = 0.52) in sweep-net samples on 12 April, and with the number of copepods (df = 40; P = 0.17; r = 0.37) on 18 May. Mean mass was not significantly correlated with the number of copepods on 12 April (df = 40; P = 0.43; r = 0.13), cladocerans on 18 May (df = 40; P = 0.53; r = −0.10), or chironomids on either 12 April (df = 40; P = 0.13; r = 0.24) or 18 May (df = 40; P = 0.31; r = −0.16).
https://static-content.springer.com/image/art%3A10.1007%2Fs13157-013-0428-7/MediaObjects/13157_2013_428_Fig2_HTML.gif
Fig. 2

Relationship between salinity (mg L−1) in mesocosms and survivorship and growth of spotted salamander (Ambystoma maculatum) larvae over 50 days

Experiment 2: Direct Effect of Salts on Spotted Salamander Larvae

In the chronic exposure experiment, spotted salamander larvae were salt tolerant. Survivorship to the termination of the experiment (day 25) was 100 % in all but six of the containers. Almost all mortality occurred early in the experiment (15–25 March). Of the six containers with mortality (range = 20–100 %), five had relatively low experimental salt concentrations (< 102 mg L−1 or 60 mg L−1 of Cl) and the cause of death did not appear to be due to salt stress. We did not statistically analyze the survival data since practically all values for survival on Day 15 and Day 25 were identical (100 %).

Larvae grew more rapidly in the lab than in the mesocosm experiment, presumably due to warmer temperatures and a greater supply of food. In contrast to the mesocosm experiment, larval growth rate was independent of salinity on both Day 15 (df = 33; P = 0.34, r2 = 0.03) and Day 25 (df = 33; P = 0.17, r2 = 0.06; Fig. 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs13157-013-0428-7/MediaObjects/13157_2013_428_Fig3_HTML.gif
Fig. 3

Relationship between salinity (mg L−1) and growth rates of spotted salamander (Ambystoma maculatum) larvae in laboratory containers. Larvae were weighed on Day 15 (open symbols) and Day 25 (filled symbols) of the experiment. Relationships were not significant for either date (P > 0.05)

Discussion

Although the toxic effects of salts on aquatic organisms are well known from laboratory studies of acute and chronic exposure, a far more challenging problem is to understand how salinization of freshwater systems can affect species through indirect effects that involve changes in community composition, species interactions, and trophic structure. Because of the complexity of food webs and trophic level interactions, the responses of salt-sensitive species may vary across a salinity gradient, and in some cases may be positive. For example, Petranka and Doyle (2010) found that the growth of wood frog (Lithobates sylvaticus) hatchlings in mesocosms was positively correlated with experimental salt concentrations between 0 and 3,000 mg L−1, but negatively correlated between 3,000-6,000 mg L−1. The positive trend may have been caused in part by a reduction of interspecific competition with zooplankton, which declined with increasing salinity. Similarly, Van Meter et al. (2011) reported that Gray Tree Frog (Hyla versicolor) larvae in a high-salinity treatment grew faster than controls, perhaps because a die-off of zooplankton triggered an increase in phytoplankton in mesocosms.

The trophic level that a species occupies may also affect the performance of a species across a salinity gradient. Unlike most anuran larvae, Ambystoma larvae are strict predators and could be adversely affected by decreases in the abundance of prey taxa (i.e., zooplankton). Numerical changes in prey could in turn be due to either taxon-specific salt-sensitivity or altered interactions with other community members. A primary finding of this study was that a top predator (spotted salamander) was affected far less than its primary prey (cladocerans). In the lab, spotted salamander larvae from our study population tolerated chronic exposure to salt concentrations as high as 2,300 mg L−1 (1,348 mg L−1 of Cl) without any detectable effects on survival or growth. These tolerances exceed those of Collins and Russell (2009), who reported an LD50 of ~ 1,938 mg L−1 NaCl for a 96-h acute toxicity study of spotted salamander hatchlings from Nova Scotia. Cladocerans showed much stronger responses to salt contamination. They declined by > 50 % in pools averaging > 800 mg L−1 of salts (469 mg L−1 of Cl) after 2 weeks, and were mostly absent in pools with > 420 mg L−1 of salts (246 mg L−1 of Cl) after 7 weeks. The rapid collapse of zooplankton communities with increasing salinity was the presumed primary cause of parallel declines in the growth rates of spotted salamander larvae.

The extreme sensitivity of cladocerans (primarily Daphnia and Ceriodaphnia) in our mesocosm experiment contrasts with the results from acute and chronic exposure studies in the lab, where LD50 values for NaCl often exceed 1,500–2,000 mg L−1 (e.g., Harmon et al. 2003; Gonçalves et al. 2007; Santos et al. 2007; Corsi et al. 2010; Latta et al. 2012) and rates of population increase are generally not affected below 1,000–1,500 mg L−1 (Cowgill and Milazzo 1991; Sarma et al. 2006). However, our results are consistent with those of other studies that were conducted in more natural settings. Petranka and Doyle (2010) found that cladocerans were rare or absent in outdoor mesocosms with salinities > 1,200 mg L−1, while Van Meter et al. (2011) reported a 93.5 % decline of cladocerans relative to controls in mesocosms with ~1,300 mg L−1 NaCl. In a survey of ponds in Australia, Hart et al. (1991) rarely found cladocerans in natural habitats with salinities > 1,000 mg L−1. These studies suggest that moderate contamination of breeding sites from road deicers could significantly reduce food resources for larval salamanders and compromise growth or survival. However, much remains to be learned about the seeming discrepancies between the results of laboratory versus mesocosm studies, and how salt contamination actually affects food resources in natural breeding sites.

Although the dietary staple of Ambystoma larvae is zooplankton, older larvae often consume chironomids and other larger benthic prey (Petranka 1998). Midges tended to increase in our mesocosms as the spring warm-up progressed, and replaced zooplankton as the numerical dominants in pools. At the termination of the experiment, we found a general pattern for chironomids to increase in abundance between 28 and 1,000 mg L−1 salts. This paralleled a decline of zooplankton over this range, and may reflect competitive interactions between members of these taxa. Petranka and Doyle (2010) reported a similar inverse relationship between the numeric abundance of zooplankton and midge larvae. Because of the compensatory replacement of zooplankton by midges, salinity had no effects on the numerical abundance of all invertebrates in samples from 18 May. Nonetheless, spotted salamander growth rates through the initiation of metamorphosis declined with salinity, which may reflect the strong depletion of food resources that occurred earlier in the season.

One challenge with regards to understanding the impact of road deicers on amphibians is that populations near roads may be smaller than those in nearby forests for reasons unrelated to salt contamination. These include edge effects, the loss of adult habitat from paving and deforestation, mortality of juveniles and adults from road kill, and exposure of eggs and larvae to heavy metals and other toxins from run-off (Veysey et al. 2011). Despite these confounding factors, collective evidence from a variety of studies provides a compelling argument that salinization of breeding sites can adversely impact amphibians through both direct and indirect mechanisms. These include laboratory assays (e.g., Sanzo and Hecnar 2006; Collins and Russell 2009; Karraker and Gibbs 2011), mesocosm studies (Petranka and Doyle 2010; Van Meter et al. 2011), field experiments (Turtle 2000; Karraker et al. 2008; Brady 2012), and inverse correlations between the salinity of breeding sites and amphibian abundance and richness (Sanzo and Hecnar 2006; Karraker et al. 2008; Hecnar and M’Closkey 1996; Collins and Russell 2009).

Ambystoma larvae are important predators in seasonal woodland ponds, and may be more sensitive to salinization than many anuran larvae because they rely heavily on zooplankton for food. Competition among Ambystoma larvae for food resources is often keen, and the depletion of zooplankton can slow growth, delay development, increase mortality, and reduce the size of larvae at metamorphosis (Petranka 1998; Wells 2007). Higher premetamorphic mortality and reduced size at metamorphosis may ultimately reduce the size of adult populations, since size at metamorphosis is positively correlated with overall lifetime fitness (Scott 1994; Wu et al. 2012).

Evidence to date suggests that even moderate salt contamination of breeding sites can result in population declines or losses of spotted salamanders and other amphibians (Karraker et al. 2008; Collins and Russell 2009). However, we have a poor understanding of the extent to which population declines reflect direct toxic effects on the eggs and larvae, or indirect effects on larvae that are associated with altered trophic level and food web interactions. A growing body of evidence suggests that salinity may have a greater indirect effect than a direct lethal effect of aquatic organisms when at low concentrations (Findlay and Kelly 2011). Karraker et al. (2008) found that spotted salamander larvae had lower survival than controls when grown at ~180 mg L−1 NaCl under semi-natural conditions in the field, but it is uncertain to what extent reductions in zooplankton contributed to larval mortality. Brady (2012) provided evidence that spotted salamander embryos in Connecticut were locally adapted to better tolerate environmental contaminants in roadside pools. This could act to ameliorate the impact of salinization to varying degrees, depending on the level of gene flow between local demes. Even if selection favors salt tolerance in embryos, depletion of zooplankton could act as a bottleneck that reduces the size of adult populations.

Amphibians are experiencing global population declines due to a variety of factors (Stuart et al. 2004), and increased stress from salinization of freshwater wetlands may exacerbate the problem. Researchers are actively seeking alternatives to conventional road deicers such as acetate, but these have so far proven to be expensive and lethal to amphibians (Harless et al. 2011). There is an obvious need for safe roads during times of inclement weather, but the long-term effects of increased salinity on the environment is a pressing problem in many areas of North America where deicers are applied (Kaushal et al. 2005; Corsi et al. 2010). We encourage researchers to further explore ways to find inexpensive deicers with minimal environmental impacts, and to better understand the indirect effects of deicers on biotic communities.

Acknowledgments

We thank Kayla Bott for her assistance in the study.

Copyright information

© Society of Wetland Scientists 2013