The relationship between spindly leg syndrome incidence and water composition, overfeeding, and diet in newly metamorphosed harlequin frogs (Atelopus spp.)

Spindly Leg Syndrome (SLS) is a persistent animal welfare issue associated with the rearing of amphibians in captivity. We conducted two experiments to investigate the effects of diet, water composition and overfeeding on prevalence of SLS in newly metamorphosed harlequin frogs (Atelopus spp.). In our first experiment, we offered 400 full-sibling tadpoles of Atelopus certus isocaloric diets in treatments of 31%, 37%, 42% and 48% crude protein respectively. Tadpoles fed higher protein diets metamorphosed faster, but the incidence of SLS exceeded 80% in all treatments leading to the conclusion that variation in dietary protein was not responsible for causing SLS. We used 720 full-sibling Atelopus glyphus tadpoles in a second experiment to examine the effects of diet type, water composition and diet ration on SLS. We found that an overall incidence of 58% spindly leg in tadpoles reared in tap water, but reduced to about 10% in water treated by reverse osmosis and then reconstituted. It is possible that the reverse osmosis treatment removed some factor that caused the SLS, or that the reconstitution may have added a mineral lacking in the original tap water. Within tap water treatments, overfeeding tadpoles in tanks increased the incidence of SLS. We recommend further experimental research into this condition to identify the causative factors in the water. Additional research into the nutritional composition of food available to wild tadpoles would be useful in formulating captive diets, that have to date been solely based on surrogate species.


Introduction
Spindly Leg Syndrome (SLS) is a musculoskeletal abnormality commonly associated with captive-rearing of amphibian larvae resulting in under-developed front limbs [1][2][3][4]. The causes of SLS are poorly understood [2,3,5,6] some postulate a nutrient deficiency in larval diets such as vitamin B complex [3], husbandry factors such as water quality, or some complex multifactorial cause including genetics and maternal nutrition, but there has been a lack of controlled studies of the underlying etiology. The problem is a widespread and enigmatic amphibian health and welfare issue documented in several wildlife medicine text books [3,7,8].
Metabolic bone diseases lead to reduced bone density and skeletal deformities in postmetamorphic amphibians. They have been linked to improper regulation of calcium metabolism related to poor dietary Ca:P ratios, vitamin D deficiency or inadequate UV B exposure [9][10][11][12][13][14]. Calcium is a major developmental requirement during metamorphosis and tadpoles are able to selectively absorb and store calcium from their environment and is stored in endolymphatic sacs [12,15]. In amphibian tadpoles, calcium is primarily absorbed across the gills and skin, rather than the gut, and the ability to absorb calcium in this way is enhanced with dietary vitamin D supplementation [16]. Recently, others have suggested that MBD in some anurans may be a result of factors that interfere with calcium metabolism such as excessive fluoride [17], but these frogs may also have experienced conventional risk factors for MBD. Despite a significant focus on amphibian calcium metabolism in the 1970's [12], there hasn't been much recent attention on the subject. However, two independent studies suggest that a lack of calcium in tadpole-rearing water affected limb development [18][19][20]. One of these studies suggests that calcium deficiency at a key tadpole life-stage affects normal neurological development, that in turn induces limb malformations [19].
The Panama Amphibian Rescue and Conservation (PARC) Project was established in 2009 to create ex-situ assurance populations of amphibians in Panama at risk of extinction from the amphibian chytrid fungus [21]. Project staff have encountered spindly-leg syndrome in captive offspring of species with aquatic tadpoles from a variety of species including Atelopus limosus, A. certus, A. glyphus, Andinobates geminisae, Oophaga vicentei, Hyloscirtus colymba (Pers. obs. J.G. Fig 1). The prevalence of SLS in captive-reared amphibians is a factor that introduces uncertainty for population managers and is a research priority for animal husbandry and welfare of captive Atelopus [5].
We conducted two controlled captive-rearing experiments designed to improve our understanding of the occurrence of SLS in captive harlequin frogs of the genus Atelopus. In the first experiment, we tested the hypothesis that lack of dietary protein in Atelopus tadpoles caused SLS. The hypothesis is from anecdotal observations that increasing animal protein in Mantella aurantiaca tadpole diets reduced SLS incidents at the National Zoo (L. Augustine, pers. comm).
We tested three hypotheses in the second experiment. The first hypothesis is reducing quantity of food offered might reduce SLS based on anecdotal observations by JG that Atelopus tadpoles from the same clutch that were fed lower food quantities, took longer to metamorphose and had a lower SLS incidence. The second hypothesis is that some unknown nutritional factor in our experimental diet may cause SLS based on reports from the Maryland Zoo that they did not observe any SLS in Atelopus tadpoles fed exclusively on a commercial food recommended by the Species Survival Program (K. Barrett, pers. comm). The third hypothesis is that reverse-osmosis (RO) treatment of tap water will reduce SLS incidence based on observations that SLS disappeared from the collection following a transition from well water to reconstituted reverse osmosis-treated water at the PARC Project facility at the Nispero Zoo in El Valle de Anton (HR pers. obs).
provides core operational support for this project. RI, HR and JG were supported in part by a grant from First Quantum Minerals, JFCC was supported by a grant from the Woodtiger Fund. BG is grateful to Susan and Frank Mars for their continued financial support. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Donna Snellgrove is employed by Waltham Center for Pet Nutrition. Waltham Center for Pet Nutrition provided support in the form of salary for author DS, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific role of this author is articulated in the 'author contributions' section.
Competing interests: Donna Snellgrove is employed by Waltham Center for Pet Nutrition. This study was funded in part by First Quantum Minerals. There are no patents, products in development or marketed products to declare. This does not alter our adherence to all the PLOS ONE policies on sharing data and materials.

Experiment 1: Investigating the influence of dietary protein content on SLS
A clutch of Atelopus certus eggs laid on 9 December 2014 hatched on 15 December 2014 at the PARC Project facility in Gamboa, Panama. We took 400 tadpoles at Gosner stage 22, two days after hatching, and split them into 16 groups with 25 tadpoles per tank. We used 16 identical 20-liter tanks set up on a single rack with about 2.5 cm of gravel covering an undergravel filter and air stone. Tanks were filled to a 16l mark (80%) and water temperature was maintained at 22.3-24.1C and lighting was provided on a 12-hour cycle from a shop light on each shelf with 2x T8 10% UVB bulbs. Water was partially (20%) changed with carbon-filtered tap water twice weekly through siphoning, in order to reduce fouling. We randomly assigned four tanks to each of four experimental isocaloric diet treatments that contained 31%, 37%, 42% and 48% crude protein respectively (Table 1). These ranges are based on diets of other surrogate frog species, but are within the potential protein content of periphyton collected at sites where other Atelopus species have been encountered in Panama (S1 File). Atelopus are specialized grazers adapted to eating periphyton, so we weighed food and mixed it to a paste by adding water. We spread the paste > 1mm thickness on acrylic plates and dried it to a film. Each day we replaced two feeding plates in the front and rear of each tank. Initially we offered 0.38 g of food per tank but this led to excessive water fouling. Therefore, we reduced the food offering to 0.24 g per tank after the first 2 weeks. When tadpoles developed hind limbs, we place floating polystyrene slices in the tanks to facilitate froglet emergence. For each tank we recorded 1) proportion of 'spindly leg' syndrome by classifying each tadpole into 'spindly' or 'healthy' [2], 2) mortality or survival through metamorphosis (Gosner stage 47), 3) number of days to metamorphosis of each surviving tadpole. Commercial brand names for equipment and diets used throughout this paper are detailed in S2 File. The Smithsonian Tropical Research Institute IACUC approved this project under proposal #2014-0901.

Experiment 2: Investigating water composition, ration and diet type on SLS
A clutch of Atelopus glyphus eggs hatched on Nov 27, 2016 at the Panama Amphibian Rescue and Conservation Project in Gamboa Panama. Within 3 days of hatching (Gosner stage 22), we took 720 tadpoles and split them into 36 groups with 20 tadpoles per tank. We used a  completely randomized design to examine the effects of ration, diet type and water composition using tadpoles as our experimental units ( Table 2). We established four replicate tanks per treatment. We offered a custom-made experimental diet designed to meet recommended nutritional requirements for amphibians at full, half and quarter rations to each treatment group. We tested for differences between the experimental diet and the commercial diet type at the half-ration level ( Table 2). We provide full nutrient composition analyses conducted in the same laboratory using the same experimental batches for both diet types in Table 1. Our final treatment included the commercial diet and water supplemented with a vitamin b complex in the following ratios: Vitamin B1 10mg/ml, Vitamin B2 2mg/ml, Vitamin B6 2 mg/ml, Vitamin B12 50μg/, D-panthenol 2.6mg/, Nicotinamide 50mg/ml. To compare the effect of water composition we compared municipal tap water provided by the Gamboa Municipality with reconstituted RO water. We prepared RO water in 90L reservoirs using a non-commercial 100G gallon per day Reverse Osmosis System and reconstituted the RO water by dissolving 0.0395g calcium chloride, 0.0465g magnesium sulfate, 0.0358g potassium bicarbonate and 0.0298g sodium bicarbonate per liter of RO water. We measured various water composition parameters for tap water, RO water and reconstituted RO using portable colorimeter according to the manufacturer specifications (Table 3), a single sample of tap water and RO water was mailed to Triton Labs, Germany for an elemental water analysis (Table 4) We used 36 identical tanks filled to a 26-liter mark without any gravel or lighting other than ambient fluorescent overhead shop lights in the laboratory. We placed a 5cm PVC "T" fitting at the bottom of each tank as cover for the tadpoles. We filtered water in each tank using a small sponge filter with air stone in each tank. We added a supplemental air stone to each tank to improve aeration and circulation of water and maintained the temperature at 22.3-22.5˚C. We partially (30%) changed the water three times per week through siphoning to prevent water fouling. Weighed food was and mixed with water to a paste then spread to about 1mm thickness on acrylic plates scoured with sandpaper to allow better adhesion. Initially we offered 0.88g as the full ration but reduced this to 0.44g after two weeks due to excessive water fouling, adjusting the half and quarter rations similarly. Once per month, immediately prior to water changes we measured temperature, pH, dissolved oxygen, total dissolved solids, ammonium and ammonia in every tank using a commercial water quality probe. When tadpoles developed hind limbs, we placed floating polystyrene slices in the tanks to facilitate froglet emergence. For each tank we recorded: 1) proportion of SLS by classifying each tadpole into 'spindly' or 'healthy'; [2]; 2) mortality or survival through metamorphosis (Gosner stage 47); 3) number of days to metamorphosis of each surviving tadpole. Any animals displaying spindly leg syndrome were humanely euthanized by immersion in saturated, buffered MS222 as advocated in AMVA humane euthanasia guidelines. The Smithsonian Tropical Research Institute IACUC approved this project under proposal #2014-0901-2017-2-A1.

Statistical analysis
We visualized data by plotting mean values per treatment +/-standard error. In order to maximize the power to detect statistical differences we defined individual tadpoles as experimental units. Using individual binary response variables such as spindly/healthy required the use of a generalized linear mixed model (GLMM) to analyze data. A GLMM permitted binary data responses and allowed us to incorporate blocking factors such as tank as random covariates [22]. We conducted all analyses using the Package 'lme4' in R [23]. In each model, we examined the main effects such as protein, ration, water composition and diet type. Random covariates accounted for variation introduced from other factors including tank and differential tadpole survival in each tank as a proxy indicator for unintended density effects. All relevant data are within the paper and its supporting Information files. Raw experimental data are available in S3 File.

Experiment 1: Investigating the influence of dietary protein content on SLS
All protein treatments produced more than 80% spindly-leg metamorphs and experienced average mortality rates of 30-50% (Fig 2). A slight, but statistically significant trend indicated that higher protein diets let to metamorphosis a few days earlier than lower protein diets and was associated with marginally lower percentages of SLS (Fig 2, Table 5). However, the overwhelming proportion of SLS metamorphs produced by all treatment groups in this experiment leads us to conclude that dietary protein levels are not the primary causative factor associated

Tap water RO Reconstituted RO Units
Unwanted   with SLS. The high quantities of B vitamins in each experimental diet (Table 1), and the provision of UVB lights for all tanks leading us to conclude that neither UVB nor thiamine were likely limiting factors causing the observed SLS.

Experiment 2: Investigating water composition, ration and diet type on SLS
Diet type did not significantly affect the prevalence of SLS, but both water source and ration significantly influenced spindly leg incidence with probabilities > 0.001 ( Table 6). Regardless of ration, the provisioning of reconstituted RO water drastically reduced the incidence of SLS (Fig 3). Despite the severe reduction in SLS cases with reconstituted RO water, they were not completely eliminated. A closer examination of the RO water data (Table 3) indicates that the RO process removed about 84% of the total dissolved solids in the tap water so it is likely that a small proportion of all the dissolved solids remained even post RO treatment. The regular tap water treatments had consistently higher incidence of spindly leg than reconstituted RO treatments but the proportion of SLS in tap water treatments was correlated with ration. Tadpoles receiving a full ration had about 70% SLS and dropped to about 25% spindly in the quarter ration treatments (Fig 3). It is unclear which water composition parameters were responsible for the dramatic differences between RO and tap water. A single measurement of the starting ratio of Phosphate to total Calcium hardness in tap water was 1:189, while in the reconstituted RO it was 1:986. Starting concentrations of total dissolved solids, chlorine, phosphates, and iron were substantially higher in tap water compared to RO, while RO had higher sulfate concentrations (Table 3). Multiple measurements of a more limited set of parameters in the actual tanks revealed that dissolved solids, ammonia and ammonium were consistently higher in RO treatments than tap water (Table 7). This pattern was not evident in the starting concentrations of the water provided (Tables 3 & 4), indicating that some differential biological processes may be occurring in the different water treatment groups. Furthermore, post-metamorph survivorship rates in RO treatments were double those of regular tap water (Table 8). Diet type and ration significantly affected tadpole mortality but type of water did not ( Table 6). Both commercial diets and the full ration experimental diets were associated with reduced tadpole survival ( Table 6, Fig 3) and provisioning of supplementary vitamins to the water of tadpoles fed the commercial diet had no effect on survival (Fig 3). Ration was the only experimental factor that influenced the number of days to metamorphosis. We designed the experiment anticipating that limited food availability would be a treatment that would be associated with slower growth rates. Counter to our expectations, however, tadpoles receiving a full ration took on average 15-20 days longer to metamorphose than tadpoles receiving a quarter ration (Fig 3). This observation, combined with the higher mortality rates observed in full ration diets leads us to conclude food quantity was not a limiting factor for any of the diet treatments. In fact, it is likely that all ration treatments were overfeeding the tadpoles. Despite the three times per week water changes, it seems likely that water fouling associated with the overfeeding is likely to be responsible for the slower development and higher mortality rates observed in higher ration treatments (See Table 7 for water quality data immediately prior to water changes).

Experiment 1: Investigating the influence of dietary protein content on SLS
Similar to other tadpole experiments, we found that tadpoles with higher protein diets metamorphosed sooner. Other studies have noted that the optimum protein levels in bullfrog and natterjack toad tadpole diets are 44-46% [24,25], while extremely reduced dietary protein in tadpoles impairs growth rates and disease resistance [26]. Despite a relative abundance of studies of dietary protein on tadpole development, none report any relationship to diet and poor limb development [25][26][27][28]. The lack of any wild-type dietary baselines for Atelopus species is a limiting factor, forcing us to use surrogate species that may differ substantially in their biological requirements. The preliminary data on nutritional composition of food available to Atelopus tadpoles (S1 File), shed more light on the difficulties associated with this task, than the actual nutritional composition of the food. The high prevalence of SLS across all protein treatments in this experiment and the lack of SLS encountered in other dietary protein studies lead us to conclude that dietary protein is not a causal factor associated with SLS. Two commonly cited limiting factors involved in the endocrine regulation of calcium are the provisioning of UVB light, and vitamin D3 content of the diet [29], but these factors were not limiting in this experiment and UVB was provisioned at 2.6-3.5 UVI, and vitamin D3 levels were 6 times the recommended concentrations [1]. The Ca:P ratio of these tadpole diets was about 1.5:1, exceeding the recommended minimum ratio of 1:1 for adult amphibians [9]. Recommended Ca:P ratios for tadpole diets have not been determined but may be a less-relevant factor, given tadpoles' ability to absorb calcium from their environment [12].

Experiment 2: Investigating water composition, ration and diet type on SLS
The clearest pattern from this experiment was that provision of reconstituted RO water drastically reduced the occurrence of SLS, and we can say with some confidence that differences in  (Table 5).
https://doi.org/10.1371/journal.pone.0204314.g002 Table 5. Generalized Linear Mixed Model analysis for Experiment 1 examining the effects of dietary protein content on 1) SLS, 2) mortality and 3) days to metamorphosis in 400 Atelopus certus tadpoles that were fed isocaloric diets in four treatments of 28, 34, 40 and 46% protein, replicated in four tanks per treatment and 25 tadpoles in each tank, means shown in Fig 2. Block effects of tank and density-dependent effects resulting from differential mortality between tanks were accounted for in the model as covariates where applicable. Effects of water composition, overfeeding, and diet on spindly leg syndrome in harlequin frogs water composition were responsible for the observed differences. Given our knowledge of the importance of calcium metabolism in amphibian metamorphosis [12], a more detailed experimental examination of Ca:P ratios in Atelopus tadpole water seems warranted. According to the U.S. Geological Survey, soft waters are less than 60 mg/l hardness, while very hard water exceeds 251 mg/l on the water hardness scale [30]. Our Gamboa experimental water is on the lower end of the scale with Ca hardness values of 26 mg/l in tap water 56 mg/l in reconstituted water (Table 3). One other study noted scoliosis and crooked limb development in tadpoles reared in water with a hardness <4mg/l [18] but it is not clear if SLS was observed. A second study observed hind limb deformities connected with poor neuron development that were induced in deionized water that were ameliorated by the addition of CaCl 2 [19]. Based on these similar, independent observations, it seems likely that a lack of dissolved calcium in the water is a potential limiting factor in this system. A second clear pattern within tap water treatments is that the prevalence of SLS was aggravated by overfeeding. Traditional indicators of poor water quality such as ammonia were actually higher in the reconstituted RO treatments (Table 7, Kruskall Wallace Test 1df, Chi square 27, p <0.001). The fact that the tanks with the poorest water quality indicators of organic matter decomposition had the lowest spindly leg incidence indicates that ammonia release was not the likely factor causing SLS (Table 3). It is possible, however, that some other soluble factor associated with surplus food decomposition in the tap water could be cumulatively adding undesirable elements to the water in both the RO and the tap-water treatments. The more Table 6. Linear Mixed Models examining the effects of ration, water and diet on 1) SLS, 2) mortality and 3) days to metamorphois in 720 Atelopus glyphus tadpoles reared in 36 tanks with 20 tadpoles per tank and 4 tanks per treatment. Block effects of tank and density-dependent effects resulting from differential mortality between tanks were accounted for in the model as covariates where applicable. Effects of water composition, overfeeding, and diet on spindly leg syndrome in harlequin frogs advantageous starting water composition provided by the RO treatment mitigated the effects somewhat in that treatment. In light of these results, the combination of tap water and overfeeding associated with all treatments in experiment 1 may explain the high spindly leg ratios consistently observed in that experiment.
The dietary analyses showed that the experimental diets compared favorably to the recommended amphibian nutrition intake for post-metamorphic amphibians ( Table 3, [1]). The diet type did not significantly affect SLS prevalence but it did significantly affect mortality rates, with the highest mortality rates observed in the commercial diet treatments, and this may be connected to vitamin deficiencies noted in the nutritional analysis of that diet.
Overall, the results of these experiments lead us to conclude that spindly leg syndrome is directly related to the water composition in which the tadpoles are raised, rather than the diet. Further experiments should experimentally focus on the addition or removal of specific compounds to water that can reduce the incidence of spindly leg.  Effects of diet type, ration and water source on A) proportion of SLS cases, B) tadpole mortality and C) average number of days to metamorphosis in Atelopus glyphus tadpoles. Bars for each treatment represent mean values from 4 tanks +/-SEM. 20 tadpoles were placed in each tank at the beginning of the experiment. Proportion of SLS for commercial 0.5, RO water is not presented because 100% tadpole mortality in some tanks led to missing replicates.

Supporting information
https://doi.org/10.1371/journal.pone.0204314.g003 Table 7. Summary of water quality variables by treatment using a water quality probe. Means for each treatment were determined measuring all 36 tanks on 4 separate sampling dates immediately prior to water changes. Comm = commercial diet, Expt = experimental diet, CommVit = Commercial diet with vitamin B complex supplement added to the water. DO = Dissolved oxygen, TDS = Total Dissolved solids.