Research Article |
Corresponding author: Fabricius Maia Chaves Bicalho Domingos ( fabriciusmaia@gmail.com ) Academic editor: Günter Gollmann
© 2021 Gabriel Preuss, Anna Victoria Silverio Righetto Mauad, Rafael Shinji Akiyama Kitamura, Thara Santiago de Assis, Marina Corrêa Scalon, Fabricius Maia Chaves Bicalho Domingos.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Preuss G, Silverio Righetto Mauad AV, Shinji Akiyama Kitamura R, Santiago de Assis T, Corrêa Scalon M, Domingos FMCB (2021) Lurking in the depth: Pond depth predicts microhabitat selection by Rhinella icterica (Anura: Bufonidae) tadpoles at two different sampling scales. Herpetozoa 34: 149-158. https://doi.org/10.3897/herpetozoa.34.e56326
|
Habitat selection has long been a central theme in ecology and has historically considered both physiological responses and ecological factors affecting species establishment. Investigating habitat selection patterns at different scales can provide important information on the relative roles of the environmental factors influencing the organisms’ abilities to use their surrounding habitat. This work aimed at investigating which environmental factors determine habitat selection by Rhinella icterica tadpoles, and also took the opportunity to investigate how the scale in which tadpoles and environmental data are sampled might influence the habitat selection results. A total of 2.240 tadpoles were counted in the whole sampling area, and while substrate cover and depth were the variables that better explained the abundance of tadpoles at the larger scale (plot level), depth and water turbidity better explained tadpoles’ abundance at the smaller scale (subplot level). The results suggest that avoiding predation by matching the background color is a likely process explaining tadpoles’ occupancy at both scales. Depth is known to influence tadpole habitat use in the tropics, and although its combination with turbidity and substrate cover varied between scales, our study suggests that sampling at different scales might not affect the inferred ecological processes driving habitat selection. This information might also be useful to predict tadpoles’ responses to micro-environmental perturbations and help in guiding the choice of parameters that should be taken into account when analyzing the effects of habitat degradation in Atlantic Forest amphibian populations.
Atlantic Forest, conservation, Cururu Toad, depth, distribution, habitat selection, substrate cover, water turbidity
There is currently little doubt that habitat loss and degradation are among the major threats to biodiversity worldwide (
In ectotherms, the biotic and environmental factors affecting habitat selection, such as predation pressure and temperature, will certainly influence population dynamics over time (
The Atlantic Forest is one of the most diverse and understudied ecosystems in the world (
Researchers working on tadpole ecology and habitat selection must always take practical decisions on how to effectively sample tadpoles in the field. In most cases, individuals are sampled near the stream or pond edges using dip-nets (
The toad Rhinella icterica (Spix, 1824) is distributed throughout southern South America, from eastern Paraguay to northern Argentina, in southern and southeastern Brazil, and northwards to the state of Bahia (Frost, 2019). This species differs from other Rhinella marina group species mainly by the presence of a subtriangular parotoid and tibial glands, as well as less developed foot webs, and tadpoles with isometric growth (
In this paper, we investigated which environmental factors determine microhabitat selection of R. icterica tadpoles. In addition, we examined how spatial scale might influence habitat selection results. We hypothesized that the environmental variables explaining tadpoles’ abundance will differ according to the sampling scale: variables related to the physical characteristics of the pond will be more important at the larger scale, whereas variables that provide better camouflage opportunities will be more important at the smaller scale. Thus, at the larger scale, we predict tadpole abundance will be greater in shallow and larger habitats, with higher substrate cover and slower water flow. These expectations are based on the assumption that in these conditions there are more opportunities for shelter and less risk of dragging. At the smaller scale, near to the pond edge, higher levels of vegetation cover, water turbidity, as well as substrate leaf content, will possibly provide better camouflage opportunities and a stronger role in the abundance of tadpoles.
The study area is located in the protected area ‘Reserva Natural Salto Morato’ (25°10'54"S; 48°17'52"W), municipality of Guaraqueçaba, Paraná State, Brazil (Fig.
The sampling site was a 30 m long temporary pond with width varying from 60 cm to approximately 6 m and maximum depth of approximately 30 cm. This pond was surrounded by vegetation, predominantly secondary forest, and emergent macrophytes in the pond margins. Since we sampled during the rainy season (August), the pond had a constant influx of water from a nearby creek, and the excess of water would escape from the opposite part of the pond through a small water channel excavated by the rain. Ultimately, this means that the pond had constant running water throughout its whole extension.
With the help of labelled stakes we delimited 20 plots along the pond extension where it was possible to visually count the tadpoles from the margin and without entering the pond. This condition was necessary to avoid biases due to the possibility of disturbing the tadpoles if an observer entered the pond. Each plot was 1 m long, consisting of the whole rectangular area across the pond, and this delimitation comprised our larger scale data collection. For the smaller scale data collection, each plot was subdivided in two equal parts considering a central imaginary line parallel to the pond edge (Fig.
Relationship between tadpole abundance and pond depth at both sampling scales. In the plot level (A), the substrate cover percentage was also selected as an important variable determining tadpole abundance and is represented by the gradient color scale from light to dark brown. In the subplot level (B), water turbidity was also selected as an important variable determining tadpole abundance and is represented by the gradient color scale from light to dark blue.
Within each plot and subplot we counted the total number of tadpole individuals (abundance), measured pond width, depth, water flow velocity, and visually estimated water turbidity, surrounding vegetation cover percentage (including emergent macrophytes present in pond margins), and percentage of the sand-soil covered by leaves (hereafter substrate cover). Tadpoles were counted by a single observer, and always considering one subplot and then the other (i.e., the plot count was the sum of both subplots). Tadpole counting was performed from the outer edge of the pound before all other measurements to avoid disturbing the tadpoles. Hence, after counting the tadpoles, water flow velocity was measured in meters per second (m/s) by the formula D/T, where D is the distance (meters) and T is time (seconds). Distance was measured using a polystyrene foam ball and a 30 cm scale, where the ball was gently placed on the water surface, parallel to the streamflow direction, and next to the scale for visualization. Pond width and depth were measured (in cm) with a measuring tape. Since the pond was very shallow, it was not possible to use the standard Secchi disk protocol to estimate water turbidity. Thus, we used a Secchi disk as a proxy to estimate turbidity by visually estimating how well it could be seen (similar to a ‘percentage’ of how well the disc could be visualized). Turbidity, surrounding vegetation cover, and substrate cover were estimated by the same observer.
The width was measured only once (for the plot), and the subplot width was half of the plot width. To replicate the usual data collection in habitat selection studies, all other measurements were taken twice (depth, velocity, turbidity, vegetation cover, and substrate cover), one in the plot center and another one in the subplot center. Since Rhinella tadpoles, if undisturbed, tend to remain still for long periods (
We separately analyzed the data of the plots (n = 20) and the subplots (n = 20) to compare the habitat selection results inferred in these two different ecological scales represented by the different sampling strategies. Although each plot consists of two subplots, we only analyzed the data of one randomly chosen subplot per plot, to try to reflect the most commonly adopted tadpole sampling strategy used in ecological studies.
We checked for multicollinearity of our data by calculating Variance Inflation Factors (VIF) and using a threshold of VIF > 5 to exclude variables (
After building the full WRM model, we performed an automated backward stepwise model selection using the Akaike Information Criterion (AIC) to select the environmental variables that better explain tadpole abundance. In short, the analysis starts with a previously created full WRM model where abundance is explained by all environmental variables, and variables are removed in a stepwise fashion while the AIC value of the models are compared until the best model fit with the lowest AIC value is reached. Thus, for each scale, the best model explaining the abundance of tadpoles was selected by AIC, and then the significance of the variables of the best model was calculated. All statistical procedures were performed in R v3.6.0 (R Core Team, 2019).
In total, we counted 2.240 tadpoles in the whole sampling area. Tadpole abundance at the plot scale was almost twice that of the subplot scale (Table
Predictor variables of tadpoles’ abundance at the plot and subplot spatial sampling scales.
Analyzed variables | Mean ± standard deviation | Minimum – Maximum (range) |
---|---|---|
PLOT (n = 20) | ||
Tadpoles abundance | 112.0 ± 109.4 | 0–392 (392) |
Depth (cm) | 13.4 ± 7.1 | 4–23 (19) |
Velocity (m/s) | 11.1 ± 16.9 | 0–59 (59) |
Width (cm) | 172.7 ± 86.6 | 39–305 (266) |
Surrounding vegetation cover (%) | 46.5 ± 21.8 | 20–90 (70) |
Substrate cover (%) | 38.7 ± 22.6 | 10–80 (70) |
Turbidity (%) | 16.2 ± 8.7 | 5–35 (30) |
SUBPLOT (n = 20) | ||
Tadpoles abundance | 59.1 ± 52.7 | 0–143 (143) |
Depth (cm) | 10.3 ± 5.7 | 2–20 (18) |
Velocity (m/s) | 3.7 ± 7.4 | 0–24 (24) |
Width (cm) | 86.38 ± 43.32 | 19.5–152.5 (133) |
Surrounding vegetation cover (%) | 46.0 ± 24.9 | 5–95 (90) |
Substrate cover (%) | 39.7 ± 23.6 | 10–80 (70) |
Turbidity (%) | 16.2 ± 8.7 | 5–35 (30) |
At the plot scale, the stepwise model selection procedure kept substrate cover and depth as the variables that better explained abundance, while depth and turbidity were the ones selected at the subplot level (Table
Summary statistics of the best Wavelet-Revised Models (WRM) explaining tadpole abundance for each sampling scale and kept after a stepwise model selection.
Explaining variables | Estimate | Standard Error (SE) | t statistic | P value |
---|---|---|---|---|
PLOT (n=20) | ||||
Intercept | -67.485 | 33.768 | -1.999 | = 0.06 |
Substrate cover | 2.579 | 0.610 | 4.227 | < 0.0001 |
Depth | 5.935 | 2.097 | 2.830 | = 0.01 |
SUBPLOT (n=20) | ||||
Intercept | -10.097 | 11.534 | -0.875 | = 0.4 |
Depth | 10.124 | 1.350 | 7.499 | < 0.0001 |
Turbidity | -2.187 | 0.791 | -2.763 | = 0.01 |
Habitat selection theory predicts that individual choices will influence foraging efficiency and survivorship, which will be directly related to individual fitness (
Many environmental variables have been found to influence tadpole habitat selection such as water flow and substrate composition (
There is no information on the predation of R. icterica tadpoles in the wild, but it can be inferred from results of other species of the genus Rhinella that they should be mainly consumed by invertebrate predators such as dragonfly larvae and waterbugs (
Although the selection of both substrate cover and turbidity by our analyses can be attributed to protection against predators (
According to our results, water flow was not important for tadpoles’ abundance. However, this result needs to be interpreted with caution because of the low variation of this data observed among the replicates (Table
Even though we were not able to test it, another factor that could influence tadpole microhabitat use is their toxicity. Previous studies detected toxins in the eggs and tadpoles of Rhinella marina, although in lower concentrations compared to adults (
Although our two different sampling scales might reflect common sampling practices in tadpole ecology studies, they were probably not divergent enough to capture different ecological properties of tadpole habitat selection. Contrary to our hypotheses, all three variables selected at the plot and subplot scales are putatively related to predation avoidance, even though at least one environmental variable differed. In this regard, we suggest that sampling at a smaller scale should be enough to capture the ecological processes behind habitat selection, and future studies in tadpole habitat selection might benefit from this information during the research planning stage. Nonetheless, given that different variables were selected in each scale, our results also indicate that tadpoles are sensitive to micro-environmental changes, which suggests that small scale habitat perturbations might influence the behavior of these animals. Tadpoles might be highly sensitive to environmental disturbances (
In conclusion, although variables that predict R. icterica tadpoles’ habitat selection in the Atlantic Forest might be explained by the different ecological processes discussed above, avoiding predation seems to be a reasonable explanation for our results, considering that water depth was selected at both scales and correlated with substrate cover and turbidity. Our study also indicates that sampling at different scales might not affect the inference of the ecological processes behind habitat selection. In face of the concerning conservation status of the Atlantic Forest, and considering the lack of data on the ecology of many amphibian species, the conservation of anurans reproductive sites might be one of the most important conservation measures to maintain viable populations in the Atlantic Forest (
We would like to thank all the staff at ‘Reserva Natural Salto Morato’ for their help and support during the time fieldwork was conducted. GP, AVSRM, RSAK, and TSA acknowledge graduate scholarships from CAPES. MCS acknowledges a PNPD postdoc fellowship from CAPES. This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - Finance Code 001.
R Markdown file
Data type: Adobe PDF file
Explanation note: This R Markdown file can be opened using any PDF reader. It contains all R codes (script) used for data exploration and analyses, as well as all results obtained at each analytical step.