, Volume 721, Issue 1, pp 1–8

Insensitivity to road salt: an advantage for the American bullfrog?


    • Illinois Natural History Survey, Prairie Research InstituteUniversity of Illinois
    • Department of BiologySusquehanna University
  • Chris A. Phillips
    • Illinois Natural History Survey, Prairie Research InstituteUniversity of Illinois
  • Dave J. Soucek
    • Illinois Natural History Survey, Prairie Research InstituteUniversity of Illinois
Primary Research Paper

DOI: 10.1007/s10750-013-1626-2

Cite this article as:
Matlaga, T.H., Phillips, C.A. & Soucek, D.J. Hydrobiologia (2014) 721: 1. doi:10.1007/s10750-013-1626-2


The health of freshwater ecosystems is negatively affected by a multitude of pollutants. In northern latitudes, road deicing agents enter nearby ponds and waterways elevating chloride concentrations in winter and spring. Few studies have examined how amphibians respond to road salt contamination and no study has focused on the response of an invasive amphibian. We examined the effects of NaCl, the most commonly used deicing agent, on the embryos and tadpoles of the American bullfrog, Lithobates catesbeianus, a species that is invasive in many regions around the world. In the first experiment, we exposed L. catesbeianus embryos to ecologically relevant levels of chloride for 60 days. The second experiment examined the indirect consequences of chloride contamination by exposing L. catesbeianus tadpoles to dragonfly larvae. Lithobates catesbeianus did not experience reduced survival, growth, or ability to evade predation in elevated chloride concentrations compared to controls. The lack of a response by L. catesbeianus suggests that its population growth will not be negatively impacted by road salt contamination. This result may be good news for L. catesbeianus, but raises concern for sympatric amphibians that have to contend with negative impacts of both chloride contamination and non-native L. catesbeianus.


AmphibianDeicing agentsDragonfly larvaeLithobates catesbeianusSublethal effectsSurvival


Deicing agents, primarily sodium chloride (NaCl), are used to reduce the accumulation of ice on roads during the winter in the northern latitudes. Benefits of using deicing agents include reduced accident rates, reduced delays, and improved accessibility, yet these benefits are difficult to demonstrate quantitatively (National Research Council, 1991a, b). After application to roads, deicing agents travel through the environment as an aerosol spray, surface runoff, or by infiltrating groundwater (Marsalek, 2003; Karraker, 2008). Studies have demonstrated the array of negative effects that road salts have on the diversity and health of aquatic plant and animal communities (Environment Canada, 2001; Karraker, 2008). Amphibians may be especially vulnerable to road salts because the timing of reproduction for many species coincides with melting of snow and influx of salts into ponds in the spring.

Research investigating the response of amphibians to deicing agents has focused on acute and long-term effects of chloride concentrations (1–6100 mg/l) and has included six North American species (Spotted salamander, Ambystoma maculatum; Wood frog, Lithobates sylvaticus; Green frog, L. clamitans; American toad, Anaxyrus americanus; Spring peeper, Pseudacris crucifer and Eastern gray Treefrog, Hyla versicolor). These studies demonstrate that there is much interspecific variation in response to elevated chloride concentrations (Karraker, 2008). Responses include direct lethal and sublethal consequences (Sanzo & Hecnar, 2006; Denoël et al., 2010) as well as indirect consequences via changes in food-web structure and species interactions (Van Meter et al., 2011). At ecologically relevant chloride concentrations, some species have high mortality during egg and tadpole stages such as L. sylvaticus (Sanzo & Hecnar, 2006; Karraker et al., 2008) and A. maculatum (Turtle, 2000; Karraker et al., 2008); however, others such as L. clamitans (Karraker, 2007) and A. americanus (Dougherty & Smith, 2006; Collins & Russell, 2009) are relatively tolerant. Chloride contamination can also alter the growth of developing tadpoles, with consequences that are sublethal in the short-term yet still likely to affect populations in the long-term (Karraker, 2008).

Additional sublethal effects include the development of malformations such as tail curvature, causing tadpoles to swim erratically (Sanzo & Hecnar, 2006). Such physical and behavioral changes may influence the ability of larvae to evade predation, ultimately reducing their survival. Few studies have addressed the consequences of sublethal effects of elevated chloride levels on amphibians [but see Denoël et al. (2010) and Chambers (2011)].

The American Bullfrog, Lithobates catesbeianus, is native east of the Rocky Mountains (Stebbins, 1985), yet it now occupies much of western North America (Casper & Hendricks, 2005). It has also been introduced to many countries worldwide and is highly invasive (Casper & Hendricks, 2005). Besides competing with native species (Kupferberg, 1997; Kats & Ferrer, 2003), it is also known to be an efficient predator (Kiesecker & Blaustein, 1997) and to carry pathogens, such as the amphibian chytrid fungus, Batrachochytrium dendrobatidis (Garner et al., 2006). For these reasons, L. catesbeianus are implicated as the cause of declining populations of many native amphibian species (Blaustein & Wake, 1990).

Deicing agents are used on a regular basis during winter months in each midwest and northeastern state where the bullfrog is native (National Research Council, 1991a, b), including Minnesota and Iowa where bullfrog range expansions have been reported (Christiansen, 1998; Moriarty, 1998, respectively). In addition, deicing agents are used in some western states, including California and Nevada (National Research Council, 1991a, b), Colorado (Lewis, 1999), and Washington (WSDOT, 2012) where bullfrogs are non-native. It is important to note, however, that deicing agents are used at much lower levels in western states (2% of total salts used within the United States) than in midwest and northeastern states (National Research Council, 1991a, b). If L. catesbeianus shows tolerance to elevated chloride concentrations, we can deduce that road salt contamination of pools will not be a deterrent of population growth and stability for this species. Insensitivity to elevated chloride levels may offer advantages to L. catesbeianus compared to sympatric species that are affected by elevated chloride levels, especially in regions where L. catesbeianus is non-native.

Here, we assessed whether L. catesbeianus is vulnerable to road salt. First, we examined how chloride concentration affects the performance (survival and growth) of L. catesbeianus embryos and tadpoles. We expected tadpoles to have poorer survival and growth at higher chloride concentrations. Then, we exposed dragonfly larvae to tadpoles from the first experiment to examine whether dragonflies are equally capable of predating tadpoles reared at different chloride concentrations. We expected tadpoles reared at high chloride concentrations to have lower survival than those reared without chloride.


Experimental conditions and egg mass collection

We conducted both experiments in an environmental chamber that maintained air temperature at 24 ± 1°C, relative humidity at 50%, and a regular photoperiod (16:8, L:D) cycle. We reared egg masses, embryos, and tadpoles in a water mixture (60% declorinated tap water and 40% deionized water) similar to water quality conditions measured in the pond where the egg masses originated [pond conductivity (250 μS/cm), chloride (7.29 mg/l), pH (8.4), alkalinity (110 mg/l), and hardness (80 mg/l)]. The experimental units were clear plastic basins (27 × 49 × 15 cm), covered loosely with plastic wrap to reduce water evaporation. Tadpoles were fed rabbit chow ad libitum, in an amount sufficient to satisfy their hunger yet minimize the accumulation of waste in the basins.

We collected three L. catesbeianus egg masses from one pond at the Fisheries Ponds Site, University of Illinois, Urbana, IL, on 15 July 2010. Pond water temperature was 28.3°C when the eggs were collected at 09:00, higher than the water temperature of basins in the experimental chamber (24 ± 1°C). The egg masses were transported to the environmental chamber on campus and transferred into basins of water (3 l) to acclimate to experimental conditions for 24 h before experimentation began.

Experimental treatments included a control (standard water mix; 7 mg/l), and low (100 mg/l), medium (500 mg/l), and high (1,000 mg/l) chloride treatments. These levels were selected because they were similar to levels used in other studies (Sanzo & Hecnar, 2006; Karraker, 2007; Karraker et al., 2008), facilitating comparisons among species, and because they covered the normal range of levels measured in natural ponds in IL (Lake County Health Department, 2012). Samples of control and treatment water (with known quantities of NaCl (mg/l) and conductivity) were tested for chloride concentration using ion chromatography based on US EPA Method 300.0 (Pfaff, 1993). Then a standard curve relating chloride concentration (mg/l) to conductivity (μS/cm) was established. This relationship (conductivity (μS/cm) = 3.3167 * chloride (mg/l) + 196.02; R2 = 0.999) was used to determine the appropriate conductivity levels for the desired chloride treatment levels. To maintain chloride levels throughout the experiments, conductivity was measured as a proxy for chloride concentration. The chloride concentrations used in the experiments corresponded to the following conductivity levels: control = 7 mg/l chloride or 220 μS/cm conductivity, low = 100 mg/l or 500 μS/cm, medium = 500 mg/l or 1,800 μS/cm, and high = 1,000 mg/l or 3,500 μS/cm. Experimental treatment water was mixed using standard rock road salt (active ingredient, sodium chloride; obtained from the local highway department) dissolved in the water mixture. Although NaCl is the main ingredient, standard rock road salts also include anti-caking agents, such as sodium ferrocyanide, and additional elements such as phosphorus, zinc, and lead (Oberts, 1986).

Experiment 1: Do L. catesbeianus have reduced survival or growth when exposed to road salt?

On July 16, we placed ten randomly selected embryos at Gosner stages 14–18 (Gosner, 1960) into each prepared control and experimental basin. There were 12 replicates for each treatment, with four replicates for each of three egg masses, for a total of 48 experimental units. Placement of the basins on the lab benches was randomized. We maintained water volume at 1 l until July 28 when we increased it to 3 l to provide more space for the growing tadpoles for the remainder of the experiment.

We measured conductivity and temperature daily with a YSI model 55 meter for a subset of basins (four of each treatment). When conductivity differed from initial values by 10% or more, half of the water volume was changed for all basins in a treatment. To change the water, we siphoned off excess food and wastes, removed additional water and then added fresh water at a given chloride concentration. Typically water changes occurred biweekly. We measured conductivity of all basins immediately before and after water changes took place.

Tadpole survival was determined for all basins daily and dead individuals were removed from the basins. We estimated embryo and hatchling survival (non-feeding to Gosner stage 24) on day 6 and overall tadpole survival on day 60, when the experiment ended. We measured growth in the form of total length (to the nearest 0.01 mm using digital calipers) and wet mass (to the nearest 0.0001 g using an Ohaus Scout Pro SPE 123 balance) of all tadpoles on day 60. We examined tadpoles for gross morphological and behavioral abnormalities, such as tail curvature and erratic swimming or listlessness, throughout the experiment.

Survival data were not normal and transformations did not correct normality, so we used separate non-parametric Kruskal–Wallis tests to compare the proportion of embryos and tadpoles surviving in different treatments. We used a multivariate analysis of variance (MANOVA) to examine whether treatment or parental source had an effect on tadpole growth. Survival on day 60 was included as a covariate in the analysis because tadpole survival and size are not independent. Tadpole mass and length passed tests of normality and equality of variance. If a significant multivariate effect was found, separate univariate analyses were conducted to differentiate between effects on tadpole mass and length.

Experiment 2: Are L. catesbeianus reared in road salt more vulnerable to predation?

We collected dragonfly larvae (Anax junius, Odonata: Libellulidae) on September 27 from the same pond at which we collected the L. catesbeianus egg masses. We allowed the larvae to acclimate to the lab conditions and water mixture for 24 h. Then, we sorted the larvae into two size classes [small (mean ± SE; 25.21 ± 0.40 mm) and large (31.29 ± 0.65 mm) measured from photographs using ImageJ software (Abramoff et al., 2004)]. Then we randomly selected one individual from each size class to add to each 3 l basin (total of two larvae per basin). There were six replicates of each chloride treatment (low, medium, and high as in Experiment 1) and control. We allowed dragonfly larvae to acclimate to experimental conditions for 24 h to ensure that all would survive in the treated water. An aquatic plant (10 cm piece of curly pondweed; Potamogeton crispus) provided cover for both the dragonfly and tadpole larvae in each basin. Dragonfly larvae were not given food from the time we collected them from the pond until exposure to the tadpoles.

Surviving tadpoles from the first experiment were pooled by treatment into larger basins. We randomly selected ten individuals from each treatment basin to add to basins containing the dragonfly larvae of the same treatment water on 29 September at 14:00. Photos of each tadpole group were taken to determine mean tadpole length per basin (24.81 ± 0.19 mm) using ImageJ software (Abramoff et al., 2004). Water temperature at the start of the experiment was 23.7°C. Conductivity was measured at the start of the experiment to confirm that levels were the same as used in Experiment 1. We assessed tadpole survival every 8 h, for a total of 48 h.

Dragonfly larvae and tadpole lengths were compared with separate univariate analyses to determine whether there was variation in size among chloride treatments, which could affect the results of the experiment. We used a Kaplan–Meier survival analysis (Levesque, 2007) to compare tadpole survivorship over time in control and chloride-treated water. The analysis examined the proportion of units in each treatment with at least 50% of tadpoles surviving over time. Survivorship curves generated in the analysis were compared among treatments using a log-rank test. We used ANOVA to compare the survival of tadpoles after 48 h among treatments.


Experiment 1: Do L. catesbeianus have reduced survival or growth when exposed to road salt?

Embryo and hatchling survival was high, with an average of over 96% of individuals surviving in each treatment; survival did not differ among treatments (Kruskal–Wallis; χ2 = 2.19, df = 3, P = 0.53) (Table 1). Most mortality occurred in the first 3 to 4 days of the experiment and appeared to be due to a lack of egg development. Overall tadpole survival (cumulative throughout the experiment) was at least 90% in each chloride treatment and did not differ by treatment (χ2 = 2.25, df = 3, P = 0.52) (Table 1). Observations of malformed tails and erratic swimming behavior followed by death were noted for a small number of individuals (n = 5) in the low and medium chloride treatments.
Table 1

Survival and size measurements for tadpoles exposed to different levels of chloride; all values represent the mean ± SE


Embryo and hatchling survival (%)

Tadpole survival (%)

Mass (g)

Length (mm)


97.5 ± 1.3

90.8 ± 2.9

0.19 ± 0.01

24.7 ± 0.6


96.7 ± 1.4

95.0 ± 2.0

0.18 ± 0.01

24.7 ± 0.4


99.2 ± 0.8

95.8 ± 1.5

0.19 ± 0.01

24.1 ± 0.4


97.5 ± 1.3

95.8 ± 1.5

0.18 ± 0.01

24.2 ± 0.4

Chloride treatment did not have a significant multivariate effect on tadpole growth (Wilks lambda = 0.890; F6,68 = 0.678, P = 0.668) (Table 1); tadpole length varied little, from 24.12 mm ± 0.37 (mean ± SE) in the medium chloride treatment to 24.70 mm ± 0.61 in the control. Tadpole mass also varied little among treatments, ranging from 0.176 g ± 0.009 in the high chloride treatment to 0.185 g in both the control (±0.012) and medium chloride treatments (±0.010) (Table 1). Parental source contributed to tadpole growth (Wilks lambda = 0.642; F4,68 = 4.22, P = 0.004), affecting tadpole mass (F2,47 = 4.329, P = 0.021) by as much as 12% depending on clutch, but not tadpole length (F2,47 = 0.421, P = 0.659). However, the significant effect was lost when the survival covariate was removed from the analysis. The survival covariate had a significant multivariate effect on tadpole growth (Wilks lambda = 0.785; F2,34 = 4.663, P = 0.016). Tadpoles in basins with lower survival (70%) grew 11% larger in mass (F1,47 = 6.243, P = 0.017) and 5% longer in length (F1,47 = 9.601, P = 0.004) than tadpoles in basins with 100% survival.

Experiment 2: Are L. catesbeianus reared in road salt more vulnerable to predation?

In several basins in each treatment, dragonfly larvae started to attack and consume tadpoles within minutes of exposure. Tadpole mortality followed similar trends in all treatments, with consistent declines in survival over time. There was little variation across treatments in the number of tadpoles surviving after 48 h of exposure to dragonfly larvae, ranging from 2.33 ± 0.80 tadpoles in medium chloride to 2.83 tadpoles in both control (±0.83) and high (±0.79) chloride treatments (F3,23 = 0.095, P = 0.962) (Fig. 1). Tadpole survivorship curves did not differ by chloride treatment (c2 = 0.79, df = 3, P = 0.851, n = 24) (Fig. 1).
Fig. 1

Tadpole survival (raw data) decreased with amount of time exposed to dragonfly larvae in all chloride treatments


Road salt had no effect on the survival or growth of embryos or tadpoles of L. catesbeianus. In addition, chloride concentration did not affect the rate at which dragonflies consumed tadpoles. The chloride concentrations we examined (7, 100, 500, 1000 mg/l chloride) fall within the normal ranges reported for roadside pools in Illinois (Lake County Health Department, 2012), New York (Karraker, 2007), Ontario (Sanzo & Hecnar, 2006), and Nova Scotia (Collins & Russell, 2009), although reports of chloride values exceeding 1,000 mg/l in roadside pools do exist (see Environment Canada, 2001; Karraker, 2008). Our study suggests that contamination of roadside ponds by road salts at levels up to 1,000 mg/l chloride will have little to no effect on L. catesbeianus embryos or tadpoles.

Tolerance of high chloride concentrations by L. catesbeianus embryos and tadpoles is not unheard of for an amphibian. In acute toxicity tests, A. americanus tadpoles had a median lethal chloride concentration of 3,926 mg/l (95% CI: 3729–4132), which was about 2–3.5 times that of A. maculatum or L. sylvaticus (Collins & Russell, 2009). However, the tolerance of A. americanus to high chloride conditions has not been studied in a long-term test. Embryos and tadpoles of L. clamitans had high survival at 945 mg/l chloride during a 58-day study; yet, the incidence of malformations increased significantly at 945 mg/l chloride compared to lower chloride treatments (Karraker, 2007). In contrast, L. sylvaticus and A. maculatum had poor survival during long-term tests at chloride concentrations at or above 145 mg/l (Karraker et al., 2008). The ultimate factors that determine the ability of amphibians to tolerate chloride contamination are not well understood.

There were a few observations of individuals with apparent malformed tails and erratic swimming behavior in low and medium treatments. However, statistically there were no differences in survival across chloride treatments, suggesting that the observations reflect background levels of random malformations in bullfrog larvae. Other studies qualitatively report that malformations increase at higher chloride concentrations (Sanzo & Hecnar, 2006; Collins & Russell, 2009), yet only one study (Karraker, 2007) quantified different malformations by conductivity level. Additional studies that quantify malformations and differentiate between those that result in mortality and do not will be helpful in understanding how sublethal effects of chloride contamination affect amphibian populations.

Tadpoles of L. catesbeianus were equally vulnerable to predation by dragonfly larvae at all chloride concentrations. Sublethal effects of chloride may be observable only at higher concentrations or with different methods. The Common Frog (Rana temporaria) had reduced swimming speeds and shorter movement distances at high chloride concentrations (600 and 900 mg/l; Denoël et al., 2010), suggesting a reduced ability to evade predation compared to individuals at lower chloride concentrations. Similar to L. catesbeianus, R. temporaria showed no difference in survival or growth at chloride concentrations up to 900 mg/l over a 2-month period (Denoël et al., 2010). These findings potentially raise a different explanation for our results. L. catesbeianus may have had reduced swimming speeds at higher chloride concentrations, but if dragonfly larvae were efficient predators regardless of the swimming speed of tadpoles, this effect would have gone undetected in our study. An experimental design similar to that of Denoël et al. (2010) where swimming speed is explicitly measured, followed by exposure of tadpoles to predators, would be an effective way to understand the consequences of sublethal effects of chloride contamination on amphibians.

Although NaCl is the most commonly used deicing agent because of its low expense (National Research Council, 1991a, b), many alternatives are in limited use today (Environment Canada, 2001) and some are thought to be more environmentally friendly than NaCl (D’Itri, 1992). Only a single study (Harless et al., 2011) has compared the response of an amphibian to different deicers. The results showed that L. sylvaticus is more sensitive to acetates (calcium magnesium acetate, magnesium acetate, and calcium acetate) and CaCl2 than to urea (CH4N20), NaCl, and MgCl2 (Harless et al., 2011). These findings emphasize the need to conduct similar testing on additional species, especially ones that may be insensitive to NaCl such as L. catesbeianus, to determine whether they exhibit sensitivity to other deicing agents. Results should be of particular interest to communities that are considering alternatives to NaCl.

Road salt contamination is known to alter community composition and food web interactions (Petranka & Doyle, 2010; Van Meter et al., 2011). At medium and high salinities (≥4000 mg/l commercial deicer product made of several deicers), L. sylvaticus survival decreased and cladocerans and copepods were rare or absent, while shore fly (Ephydridae), mosquito (Culex restuans and Anopheles punctipennis) and midge larvae (Chironomidae) abundance increased (Petranka & Doyle, 2010). Reductions in zooplankton densities with increased chloride (645 mg/l NaCl) resulted in blooms of periphyton and phytoplankton, benefiting H. versicolor tadpoles by reducing time to metamorphosis and increasing size at metamorphosis (Van Meter et al., 2011). No studies have examined how road salt affects dynamics between co-occurring amphibians, yet such a study would be particularly interesting for understanding the impact of L. catesbeianus on other amphibians. Insensitivity to road salt contamination by L. catesbeianus may have important consequences for other amphibians, especially if the performance of sympatric species is impacted by exposure to deicing agents.

Introduced populations of L. catesbeianus in Iowa (Christiansen, 1998), Minnesota (Moriarty, 1998), and California (Bury & Luckenbach, 1976) are reported to be expanding their ranges. In Iowa and Minnesota, where road salts are used at up to 5.0 tons/lane-mile (National Research Council, 1991a, b), chloride contamination of ponds may exacerbate interactions between L. catesbeianus and other species, leading to further population declines for chloride-sensitive species and potential population growth for L. catesbeianus. In regions where L. catesbeianus is non-native, it is important to consider the direct and indirect implications of road salt use on other amphibians. Our study suggests that road salt use will not hinder the population growth of L. catesbeianus, which in turn, could contribute to its success as an invader. It would be informative to use a population modeling approach to examine different scenarios between L. catesbeianus and sympatric species exposed to chloride contamination.

Our understanding of the effects of road salt on amphibians has grown in recent years, yet much remains to be studied. With wide disparity in the concentrations at which amphibians are susceptible to chloride contamination it remains vital to study the tolerance of species, especially those of conservation concern. In addition, studies that examine the sublethal effects of deicing agents and consequences at population and community levels will be of continued interest. Finally, understanding how alternative deicing agents affect amphibians will enhance our ability to inform communities about how to reduce the negative impacts of deicing agents on aquatic life.


We thank David Matlaga, Joseph Frumkin, Kyle Livengood, and Natalie Marioni for assistance in the field and the lab. We are grateful for funding from the Illinois Toll Highway Authority to complete this project.

Copyright information

© Springer Science+Business Media Dordrecht 2013