Salted roads lead to edema and reduced locomotor function in wood frogs

Human activities have caused massive losses of natural populations across the globe. Like many groups, amphibians have experienced substantial declines worldwide, driven by environmental changes such as habitat conversion, pollution, and disease emergence. Each of these drivers is often found in close association with the presence of roads. Here we report a novel consequence of roads affecting an amphibian native to much of North America, the wood frog (Rana sylvatica). Across 38 populations distributed from southern to central New England, we found that adult wood frogs living adjacent to roads had higher incidence and severity of edema (bloating caused by fluid accumulation) during the breeding season than frogs living away from the influence of roads. This effect was best explained by increased conductivity of breeding ponds, caused by runoff pollution from road salt used for de-icing. Edema severity was negatively correlated with locomotor performance in more northerly populations. Interestingly, northern populations experience more intense winters, which tends to result in more de-icing salt runoff and increased energetic demands associated with overwintering cryoprotection needs. Thus, this emerging consequence of roads appears to impose potential fitness costs associated with locomotion, and these effects might be most impactful on populations living in regions where de-icing is most intense.


18
Human activities are a leading driver of wild population declines, extirpations, and extinctions, culminating

23
Among the many groups of organisms declining across the globe, amphibians have captured a 24 special place in the attention of researchers, fueling a surge in scientific inquiry over the past three 25 decades (Green et al. 2020). Causes of declines in amphibian populations are complex and dependent 26 on both regional context and species-specific adaptations, but unsurprisingly share a common theme of 27 human activity (Grant et al. 2020). Drivers include habitat loss and alteration, pollution, climate change, 28 3 and disease. These and many other human-mediated drivers can act synergistically, confronting 29 populations with conditions far exceeding the capacity of existing trait variation or the potential for 30 adaptive evolutionary or plastic responses (reviewed in Green et al. 2020).

31
Lurking among the numerous drivers of amphibian declines lies a common but often overlooked  (Lalo 1987) -a phenomenon that disproportionately affects amphibians (Glista et al. 2008).

38
Roads can facilitate the spread of both disease (Urban 2006) and invasive species (Mortensen et al.

39
2009); they fragment habitats (Reed et al. 1996), reduce connectivity (Shepard et al. 2008), and affect 40 dispersal and gene flow (Riley et al. 2006). Roads are also substantial sources of pollution. Leaching and 41 stormwater runoff have deposited countless contaminants into the environment for decades (Huber et al.    beyond the road (Forman and Deblinger 2000). Some contaminants are fleeting, broken down by 45 chemical or biological processes, but others are long-lived or inert and tend to increase over time 46 (Kaushal et al. 2005). In short, this singular and physically narrow change to the landscape caused by 47 roads spurs numerous and wide-reaching drivers of population decline. Identifying and understanding 48 these drivers and their effects is essential for conservation efforts.

49
In areas with cold winters, a chief pollutant threatening amphibian populations is road salt. Due to 50 de-icing practices and the pervasiveness of roads, salt pollution of surface and groundwaters has become 51 common and widespread (Mullaney et al. 2009, Corsi et al. 2010. Even in large waterbodies (e.g., 52 permanent ponds, lakes), runoff pollution can result in substantial salt concentrations exceeding national 53 water quality criteria for aquatic life (Dugan et al. 2017, Kaushal et al. 2018). In smaller waterbodies 54 however (e.g., vernal pools, wetlands), levels can be many times higher, approaching brackish conditions 55 typical of coastal waters, a stark contrast to the generally salt-free conditions of small, inland surface 56 4 waters. For many pool-breeding amphibians, these smaller waters are critical sites for reproduction, 57 where embryos and larvae develop before metamorphosing and dispersing into terrestrial habitats.

58
Across the complex life-history cycle common to many amphibians, sensitivity to salt tends to be highest 59 for embryos and larvae (Gordon and Tucker 1965, Uchiyama et al. 1990, Karraker et al. 2008

69
In the present study, we investigated an additional and novel physiological consequence of roads:

75
During spring thaw, the reversal of the process whereby water is sequestered out of cells to prevent intra-76 cellular ice formation, is thought to cause temporary edema (Kling et al. 1994, Irwin et al. 1999). However, 77 our preliminary observations suggest that spring edema is more severe in populations from polluted, 78 roadside ponds compared to those from unpolluted, woodland ponds.

79
Here, we quantified the severity of edema in wood frogs from 38 ponds in New England, where 80 road salt pollution is common. We posited that road salt might be causing severe edema, and therefore 81 asked whether variation in edema prevalence and severity correlates with road adjacency and water 82 conductivity, and if edema prevalence and severity increases with age, a proxy for lifetime exposure. We 83 also investigated directly the effect of roadside pond water on edema by estimating body mass gain for 84 5 animals exposed to water from roadside versus woodland ponds and compared to spring water. Next, we 85 used dissections and histological preparations to characterize the relation between edematous outward 86 appearance and potential underlying signs and causes. Since edema could be associated with winter 87 physiology and spring thaw, we also asked if blood glucose levels correlated with edema severity. Finally,

88
we investigated a potential fitness cost of edema. Specifically, we asked how jumping performance, a 89 commonly used fitness proxy for frogs, is affected by edema severity.

117
frogs were captured on their inbound breeding migration using drift fences with pitfall traps placed 118 adjacent to breeding ponds. Frogs from these populations were therefore not exposed to pond water 119 immediately prior to capture. In the other two study areas, drift fences with pitfalls were supplemented 120 with minnow traps. Thus, frogs from these populations had mixed histories of exposure to pond water 121 immediately before capture. Both types of traps were checked daily for new captures. Captured frogs 122 were returned to the lab and dorsal photographs were taken within two days. Edema severity was scored 123 from these photographs on a 1-3 integer scale, corresponding to 1) non-edematous, 2) moderate edema,

128
We used separate mixed models to evaluate the effect of population type (roadside vs. woodland) 129 on 1) edema severity score and 2) edema prevalence. Edema severity was fit using a linear model with a 130 Gaussian distribution while edema prevalence was fit using a binomial model with a logit link. In each 131 case, we first used AIC model selection to compare three candidate models fit with maximum likelihood

139
Because our edema severity data could potentially be considered ordinal, we also composed a mixed 7 model using the R package 'ordinal' (Christensen 2019). Inference was similar between these two 141 modeling approaches (P < 0.001 in each case) and we therefore interpreted results from the linear model.

142
For the binomial responses of prevalence, inference was made using a likelihood ratio test between the 143 selected model and a reduced model in which the response mean was used as the sole fixed effect in 144 place of population type (i.e. intercept-only model). In a separate suite of standard linear models, we 145 analyzed edema in response to conductivity, dissolved oxygen, and pH. Pond-level averages of edema 146 score were used as the response variable in each model. Conductivity was log-transformed because of 147 its wide range (26-984 µS/cm) and bimodal distribution between woodland and roadside ponds.

148
We analyzed the relation between adult age and both edema severity and prevalence using data  14) breeding pairs were stocked into each treatment (i.e. breeding-pond water or spring water). Exposure 169 period was determined by the time required to breed. Following breeding, adults were re-weighed.

170
We analyzed change in body mass relative to SVL for males prior to and after breeding. Females 171 were excluded from this analysis because change in mass caused by water is confounded with change in 172 mass due to oviposition. Specifically, we calculated a body condition index (BCI) as mass per unit SVL 173 (i.e. mass/SVL), both before and after exposure to pond or spring water that occurred while breeding. We 174 then calculated delta BCI as the ratio of post-breeding BCI to pre-breeding BCI. Thus, delta BCI values > 175 1 indicate body mass gain whereas those < 1 indicate body mass loss. We used separate linear mixed 176 models with population as the random effect to analyze both delta BCI and exposure duration across the 177 interaction of population type and exposure water type. We used the R package 'emmeans' (Lenth et al. 178 2018) to estimate marginal means for each of these models and to compute 95% confidence intervals 179 with respect to a null value of 1 (i.e. no change in body mass). We applied Tukey contrasts to make

208
We used a mixed model with population as the random effect to analyze the relation between jumping 209 performance and edema. For each frog, we used maximum jump distance from among all jumps in a 210 given trial as the response variable. We analyzed jump performance collectively and separately for the 211 two regions. We also used a mixed model with population as the random effect to analyze whether jump 212 performance varied between the two regions.

215
Edema severity/prevalence and environmental variation 216 Across 38 ponds, edema was more severe and more prevalent in roadside than woodland ponds (Fig. 1).

232
Exposure-induced change in mass 233 Delta BCI, or the change in body mass relative to SVL, varied across the interaction of water type X 234 population type ( Fig. 2; F1,339.3 = 18.02, P < 0.001). Specifically, wood frogs exposed to roadside water 235 had the highest relative change in mass compared to those in all other treatments. Model estimates 236 indicated that roadside wood frogs exposed to roadside water maintained their mass, whereas wood 237 frogs in all other combinations of water and population type lost mass (Supplemental Table 1  On gross examination, marked subcutaneous edema was evident in the frog with severe edema (Fig. 2).

247
Upon dissection, edema fluid rapidly dissipated from the subcutis, and was evident to a much lesser 248 degree in the coelomic cavity. In all frogs, the gastrointestinal tract was devoid of ingesta or digesta, no 249 fat stores were present, and testes were well developed. On histology, the grossly edematous animal  salt runoff being a causative agent of edema, we found that edema severity was correlated with 279 conductivity in breeding ponds, and that frogs experimentally exposed to roadside pond water during 12 breeding increased fluid retention. Edema also correlated with increased blood glucose, as well as with 281 decreased jumping performance in northerly (but not southerly) populations. Together, these results 282 reveal that pollution from roads appears to be increasing the severity of edema in amphibians, and that 283 this relatively unexplored consequence of roads imposes a potential fitness cost through its effect on 284 locomotion.

285
Our gross morphological and histological findings indicate that fluid accumulation in the subcutis

293
Although we cannot completely rule out ranavirus as a factor, infected individuals typically present 294 ulcerative and hemorrhagic syndromes in multiple tissues (Cunningham et al. 1996, Docherty et al. 2003), 295 which we did not observe in the edematous animal dissected or among the animals scored for edema.

296
In our study, exposure to low solute water can definitively be ruled out as a possible factor, since 297 animals presenting edema were more frequently those exposed to high solute water. Frogs have highly 298 permeable skin and must osmoregulate in aquatic environments. For frogs in freshwater, the tendency is 299 for water to move from the environment into the frog. Curiously, from an osmotic perspective, adding salt 300 to freshwater should decrease the osmotic potential of the external environment relative to that of the 301 frog. That is, by adding ions to the pond, water pressure exerted on frog skin decreases. In contrast, in 302 unpolluted woodland ponds, external water pressure is higher than that found in polluted roadside ponds.

303
Thus, based purely on osmotic pressure, there should be a greater tendency for woodland, not roadside, 304 water to move into frogs. And yet we found edema was more severe in salty, roadside environments, 305 pointing to dysfunction in osmoregulation. Therefore, it is likely that the culprit of edema is kidney 306 disorder, potentially incurred from osmoregulatory stress from prior exposure to high salinity.

13
Increased edema severity in roadside ponds might also be linked to overwintering physiology.

308
Beginning in fall as temperatures decline, wood frogs accumulate urea (Costanzo and Lee 2005)

332
The edema we describe occurred in breeding frogs captured either on route to or soon after 333 arriving in their breeding ponds. This means that for at least a subset of frogs, the onset of edema 334 occurred before they were exposed to pond water. This finding indicates that edema sets in either during 335 14 the short migration from upland hibernacula to breeding ponds, or that frogs are edematous prior to 336 migration. In late fall, frogs tend to choose overwintering sites close to their breeding ponds (O'Connor 337 and Rittenhouse 2016), such that salt runoff exposure in the terrestrial environment might be 338 exacerbating edema that forms naturally following the reversal of winter cryoprotection. Interestingly, we 339 found no relation between age and edema severity or prevalence, suggesting that a single exposure to 340 the roadside environment might be just as likely to generate edema as a history of adult breeding in 341 roadside ponds.

342
Our study additionally suggests a significant impact of salt runoff exposure in the aquatic 343 environment. We found higher edema severity with increased conductivity of the aquatic environment, a

351
Moreover, we found that frogs experimentally exposed to roadside water during breeding maintained 352 mass on average while those exposed to woodland or spring water lost mass. We interpret this change in 353 mass as being driven primarily by water gain or loss, thus representing a relative increase in water 354 retention or uptake in roadside water, because frogs did not have access to food during this period.

355
Saltier aquatic environments should result in frogs taking up more ions for osmoregulation (Greenwald 356 1972), which could lead to higher water retention (Park and Do 2020) consistent with our findings (Fig. 2).

357
For example, Park and Do (2020) found that exposure of black-spotted frogs (Pelophylax nigromaculatus) 358 to about 9,400 µS/cm resulted in increased blood electrolytes and that frogs became edematous. The  that it results in skeletal muscle atrophy. The authors suggest that this mechanism is essential to increase 371 nitrogen availability for urea production given the more extreme winters in Alaska. Conceivably, harsher 372 winters in Vermont and New Hampshire induce this same mechanism and cause muscle atrophy that 373 diminishes jumping performance. Also, the serous fat atrophy we found with severe edema suggests that 374 edematous frogs consume more energy for cryoprotection. Given the importance of energy and 375 locomotion during spring breeding, particularly in the context of road crossings and swimming 376 performance during mating, these consequences could bear impacts on fitness components including 377 survival and reproductive success and, therefore, on the viability of higher latitude populations.

378
In this study, we investigated possible causes and fitness consequences of increased edema 379 severity, representing an amphibian population threat from roads and runoff pollution that has remained 380 largely unexplored. Increased edema severity in wood frogs appears to be a wide-ranging phenomenon 381 in road-adjacent habitats, driven by road salt pollution. Given the regional distribution of edema reported 382 here, we expect that this phenomenon likely affects wood frogs across much of their range, although the 383 consequences of edema might be most severe for populations experiencing more intense winters.

384
Interestingly, northern areas where winters are more intense tend to use more road salt. Thus, the 385 presence of these multiple, correlated stressors suggests that wood frogs in more northerly populations   contrast, on average, frogs from woodland ponds exposed to (left center) woodland water lost mass.

589
Similarly, individuals exposed to spring water -whether from (right center) roadside ponds or (far right)