Phenological Differences of Two Sympatric Ranid Frogs in the Southeastern United States
The Southeastern Coastal Plain of the continental United States is a hotspot of amphibian diversification. Whilst most species have since dispersed and extended their geographic distributions beyond the coastal plain, the Florida Bog Frog, Rana okaloosae, remains restricted to three counties of northern Florida. Across its range, R. okaloosae co-occurs with its sister species, the Bronze Frog (R. clamitans clamitans). Hybridization between R. okaloosae and R. c. clamitans has been documented, raising concerns about the microendemic’s future. To date, however, the phenology of R. okaloosae, as well as the mechanisms by which reproductive isolation from R. c. clamitans is achieved, is poorly understood. Using 13 years of survey data from Eglin Air Force Base, Florida, we evaluated the environmental correlates of R. okaloosae and R. c. clamitans across 80 sites where they occur in sympatry. We found that although the two species had similar breeding seasons and were active at similar times of night, calling occurred under different environmental conditions. Specifically, R. c. clamitans were more likely to call on calm, dark, humid nights, whereas bog frogs were more likely to call across a range of conditions. Our results suggest that R. okaloosae may maintain reproductive isolation in part by timing breeding activity to nights when R. c. clamitans is less likely to be active. However, there remains a large degree of overlap in the phenology of the two species, highlighting the precarious nature of the Florida Bog Frog’s existence.
THE Southeastern Coastal Plain of the continental United States is a hotspot of amphibian diversification (Wolfe et al., 1988; Austin et al., 2003; Noss et al., 2015; Fedler et al., 2023). Climate-induced sea-level rises during global thermal maxima fragment the low-lying landscape, creating insularity that drives high rates of allopatric speciation (Austin et al., 2003; Noss et al., 2015). Moreover, the region has served as a glacial refuge during past ice ages, and thus species richness has been allowed to accumulate uninterrupted (Wolfe et al., 1988; Austin et al., 2003).
Whilst most species have since dispersed and extended their geographic distributions beyond the Coastal Plain, the Florida Bog Frog (Rana okaloosae) remains an endemic and is restricted to three counties in Florida (Gorman et al., 2009; Florida Fish and Wildlife Conservation Commission, 2011, 2013). Rana okaloosae is associated with slow-moving shallow seeps, non-stagnant boggy overflows, and floating mats of vegetation along open-canopy riparian areas (Moler, 1985; Gorman and Haas, 2011). Complex ground cover, emergent vegetation, and algae mats collectively are required to support breeding of R. okaloosae (Florida Fish and Wildlife Conservation Commission, 2013). Species with small geographic distributions (i.e., microendemics) or with specific habitat requirements are disproportionately threatened with extinction due to the potential for stochastic events to encompass the entire species’ range (Levin et al., 1996; Austin et al., 2011); R. okaloosae is currently listed as threatened by the state of Florida (Florida Fish and Wildlife Conservation Commission, 2013).
The sister taxon to R. okaloosae is the Bronze Frog (a subspecies of the Green Frog, R. clamitans clamitans). In contrast to R. okaloosae, R. c. clamitans is a widespread anuran with a geographic range that spans the southeastern United States. Owing to patterns of sea-level rise during the Pleistocene, the ancestors of R. okaloosae would have been isolated from populations of R. c. clamitans, allowing allopatric speciation to take place (Austin et al., 2003). Presently, however, R. okaloosae and R. c. clamitans co-occur (Gorman and Haas, 2012). Microendemics are vulnerable to shifts in the ranges of common species (Simovich et al., 2013; Todesco et al., 2016; Rudolf, 2019; Blackford et al., 2020). When environmental or temporal barriers between closely related species are broken down, microendemics can be lost as a result of competition and/or hybridization (Chunco, 2014; Todesco et al., 2016; Perez et al., 2021; Irwin and Schluter, 2022). Indeed, hybridization between R. okaloosae and R. c. clamitans has been documented (Austin et al., 2011). Over time, hybridization has the potential to erode reproductive barriers and drive the extinction of species (Rieseberg et al., 1989; Levin et al., 1996; Rhymer and Simberloff, 1996; Allendorf et al., 2001; Vuillaume et al., 2015; Todesco et al., 2016). To date, however, the phenology of R. okaloosae, as well as the mechanisms by which reproductive isolation from R. c. clamitans is achieved, is poorly understood. Developing a better understanding of these factors is essential to help predict future vulnerability and to tailor conservation measures for this at-risk species (Fukuyama and Kusano, 1992; Steelman and Dorcas, 2010; Perez et al., 2021).
Here, we use 13 years of call survey data from Eglin Air Force Base, Florida to analyze the nightly and seasonal patterns of calling behavior of R. okaloosae and sympatric R. c. clamitans. Long-term anuran monitoring programs use call surveys to describe phenological patterns and ascertain environmental correlates of breeding activities (Bishop et al., 1997; Weir and Mossman, 2005; Dorcas et al., 2009). Rainfall, temperature, light intensity, wind speed, and relative humidity are all documented triggers for anuran calling activity and breeding phenology (Cree, 1989; Fukuyama and Kusano, 1992; Duellman and Trueb, 1994; Hatano et al., 2002; Oseen and Wassersug, 2002; Van Sluys et al., 2012). There is a growing appreciation, however, for the importance of endogenous drivers (e.g., circannual rhythms) of breeding patterns that are decoupled from environmental conditions (Oseen and Wassersug, 2002; Gottsberger and Gruber, 2004; Van Sluys et al., 2006; Brooks et al., 2019). The relative importance of environmental factors and endogenous drivers in the timing of anuran calling will influence the degree of phenological overlap between co-occurring species. To understand the threat posed by R. c. clamitans to the future persistence of R. okaloosae, our objectives were to 1) describe the phenology of R. okaloosae to provide basic natural history information currently lacking for the species, 2) identify the environmental correlates of calling activity of R. okaloosae, and 3) contrast the calling behavior of R. okaloosae and R. c. clamitans where they co-occur.
MATERIALS AND METHODS
Data collection.—
From 2007 to 2019, aural surveys were conducted at 82 known R. okaloosae breeding sites on Eglin Air Force Base, Florida. We attempted to survey all sites at least three times per year between 1 May and 25 September, with at least one month separating consecutive surveys, but staffing shortages or extreme weather events led to some sites only being surveyed twice in some years (Table 1). No site was ever surveyed less than twice in a year. During each survey, observers listened for frog calls for a duration of five minutes at each location. All observers received training in frog call identification prior to surveying, and surveys were conducted >10 m from sites so as not to disturb calling individuals. The earliest survey was conducted at 1915 h and the latest at 0218 h. Air temperature, relative humidity, sky condition (0 = no clouds, 1 = partly cloudy, 2 = overcast), wind (Beaufort wind scale), and light scale (0 = new moon, 1 = quarter moon, 2 = intermediate phase moon, 3 = full moon) were recorded during each survey (i.e., multiple times per sampling night). For each survey, the frequency of calling for each species was indexed on a scale from 0 to 3, indicating whether calls were (0) absent, (1) audibly separated, (2) overlapping but non-continuous, or (3) continuous chorus.
Data analysis.—
To investigate the calling behavior of R. okaloosae and R. c. clamitans, we adopted a generalized mixed modeling framework (GLMM) with date, time of night, temperature, relative humidity, wind speed, light conditions, and sky code as fixed effects. Data exploration showed that none of the predictor variables were highly correlated (all pairwise comparisons had Kendall’s τ values less than 0.3), and thus we included all environmental variables as predictors in subsequent models. Quadratic relationships for time of night and date were included to test for seasonal and nightly peaks in calling activity. We included year as a random effect. Due to issues with model identifiability, we could not include site as a random effect. We modeled calling first as a binary trait (detection/non-detection) and then as an index of intensity. In this way, we could separate the factors that influence detectability from the factors that drive true breeding aggregations. We employed a logistic regression framework for the detection/non-detection data and an ordinal regression framework for the calling index data. The years 2007 to 2019 were used in models where detection and non-detection served as the response variable, and the years 2009 to 2019 were used in calling index models (2009 being the first year calling index was recorded).
To explicitly test for differences between calling activity of R. okaloosae and R. c. clamitans, we included interactive terms between each environmental covariate and species, such that significant interactions provide evidence that the two species have different associations between environmental conditions and calling. The significance of each covariate in explaining variability in calling behavior was assessed using likelihood ratio tests to compare models with and without focal predictors. Likewise, the significance of species-specific responses to environmental conditions was determined using likelihood ratio tests that compared models with and without the interaction term. Models were ranked using Akaike’s information criterion (AIC), with the lowest AIC score being used to determine the best performing model (Akaike, 1974). All analysis was performed in R using the lme4 and ordinal packages (Bates et al., 2015; Christensen, 2022; R Core Team, 2024).
RESULTS
From 2007 to 2019, we conducted 2,936 surveys across 82 sites. Rana okaloosae was heard during 1,027 of the 2,936 surveys (Table 1). Rana c. clamitans was heard during 2,121 surveys and documented at all 82 sites (Table 1). Across all sites, we documented 18 anuran species in total. In addition to the two focal species, we identified calls of Southern Cricket Frog (Acris gryllus), Oak Toad (Bufo quercicus), Southern Toad (Bufo terrestris), Fowler’s Toad (Bufo fowleri), Greenhouse Frog (Eleutherodactylus planirostris), Eastern Narrowmouth Toad (Gastrophryne carolinensis), Pine Barrens Treefrog (Hyla andersonii), Bird Voiced Treefrog (Hyla avivoca), Gray Treefrog (Hyla chrysoscelis), Green Treefrog (Hyla cinerea), Pine Woods Treefrog (Hyla femoralis), Barking Treefrog (Hyla gratiosa), Squirrel Treefrog (Hyla squirella), Pig Frog (Rana grylio), River Frog (Rana heckscheri), and Southern Leopard Frog (Rana utricularia).
Detection of R. okaloosae was predicted by time and date (Table 2, Fig. 1). Specifically, R. okaloosae was more likely to be detected later at night (χ2 = 34.12, P < 0.001) and during the middle of the breeding season (χ2 = 17.01, P < 0.001). We found no evidence that lunar illumination, cloud cover, wind speed, humidity, or temperature influenced detectability of R. okaloosae (Table 2). Rana c. clamitans was also more likely to be detected later at night (χ2 = 13.31, P < 0.001) and during the middle of summer (χ2 = 34.14, P < 0.001). In contrast to R. okaloosae, the detection of R. c. clamitans was found to be related to several environmental conditions (Fig. 2). Specifically, R. c. clamitans was more likely to be detected on cloudless nights (χ2= 22.82, P < 0.001), with low levels of moonlight (χ2 = 13.80, P = 0.008), when humidity was high (χ2 = 3.77, P = 0.05), and temperatures were low (χ2 = 5.69, P = 0.02). As a result, we found evidence for different responses of R. okaloosae and R. c. clamitans to lunar illumination (χ2= 11.50, P = 0.02), cloud cover (χ2 = 15.35, P < 0.001), and humidity (χ2 = 4.86, P = 0.02). Although detection of R. c. clamitans was predicted by temperature and detection of R. okaloosae was not, this difference was only partially supported (P = 0.08; Table 2).
Citation: Ichthyology & Herpetology 113, 2; 10.1643/h2023035
Citation: Ichthyology & Herpetology 113, 2; 10.1643/h2023035
Calling index of R. okaloosae was predicted by time, date, lunar illumination, wind speed, humidity, and temperature (Table 3, Figs. 3, 4). Specifically, calling index of R. okaloosae was predicted to be higher later at night (χ2 = 35.58, P < 0.001) and during the middle of the breeding season (χ2 = 51.12, P < 0.001; Fig. 3). Further, calling index of R. okaloosae was predicted to be highest on nights with no wind (χ2 = 14.37, P < 0.001), full moons (χ2 = 11.74, P = 0.02), high humidity (χ2 = 11.03, P < 0.001), and high temperatures (χ2 = 14.55, P < 0.001; Fig. 4). We found no evidence that calling index of R. okaloosae was related to cloud cover. Calling index of R. c. clamitans was predicted by all covariates examined (Table 3). Specifically, calling index of R. c. clamitans was highest late at night (χ2 = 50.13, P < 0.001) and during the middle of summer (χ2 = 190.63, P < 0.001; Fig. 3). Calling index of R. c. clamitans was also predicted to be highest on cloudless (χ2 = 32.86, P < 0.001), windless nights (χ2 = 83.33, P < 0.001), with low levels of moonlight (χ2 = 15.53, P = 0.004), when humidity was high (χ2 = 81.95, P < 0.001), and temperatures were high (χ2 = 8.11, P = 0.004; Fig. 4). We found no evidence for different responses between calling indexes of R. okaloosae and R. c. clamitans to temperature or time of night, but we did find evidence for species-specific responses to lunar illumination (χ2 = 13.20, P = 0.01), cloud cover (χ2 = 18.36, P < 0.001), wind speed (χ2 = 11.92, P = 0.003), and humidity (χ2 = 8.98, P = 0.003; Table 3). Lastly, we found evidence for a difference in the peak date of highest calling index between the two species (χ2 = 6.07, P = 0.01; Table 3).
Citation: Ichthyology & Herpetology 113, 2; 10.1643/h2023035
Citation: Ichthyology & Herpetology 113, 2; 10.1643/h2023035
DISCUSSION
Here we describe the breeding phenology of R. okaloosae and R. c. clamitans over 13 years at sites where they co-occur. Our results provide basic natural history information for R. okaloosae that will help with their continued monitoring and management. Further, our study highlights the large degree of phenological overlap between R. okaloosae and R. c. clamitans, both in terms of the dates of calling activities and in the relationships between environmental conditions and calling behavior. We discuss our findings in the context of conservation efforts of R. okaloosae, the risk hybridization poses to the species’ persistence, and more broadly with regard to mechanisms of coexistence and the vulnerability of microendemics to global change.
Coexistence theory posits that differences in phenology can facilitate the coexistence of closely related species by reducing niche overlap (Albrecht and Gotelli, 2001; Blackford et al., 2020; Irwin and Schluter, 2022). Temporal niche partitioning results in intraspecific competition exceeding interspecific competition, negating any inherent differences in fitness, and thus preventing a single species from dominating a community (Inger and Greenberg, 1966; Godoy and Levine, 2014; Blackford et al., 2020). In the present study, however, both species exhibited breeding activity throughout the summer with peak calling dates in mid-June. Additionally, both species were more likely to be calling after midnight, ruling out temporal niche partitioning as a primary mechanism of reproductive isolation. Time of day has been shown to strongly influence calling behavior across a range of amphibian taxa that exhibit a suite of reproductive strategies (Mohr and Dorcas, 1999; Bridges and Dorcas, 2000; Oseen and Wassersug, 2002; Todd et al., 2003; Kirlin et al., 2006). In this instance, the shared proclivity to delay calling activity until later at night is likely necessary for both species to minimize heat stress and desiccation risk (Bridges and Dorcas, 2000).
Amphibians time their breeding activity to minimize mortality risk (Yoo and Jang, 2012; Brooks et al., 2019). For instance, some anurans have been shown to avoid calling on windy nights because strong winds increase cutaneous evaporative water loss and desiccation risk (Weir et al., 2005; Liu and Hou, 2012). Similarly, amphibians will generally favor humid nights when the risk of desiccation is lower (Yoo and Jang, 2012). Several anuran species will reduce calling activity on cloudless, moonlit nights to avoid visual-based predators (Tuttle and Ryan, 1982; Tuttle et al., 1982; Baker and Richardson, 2006). Alternatively, frogs may use visual cues to avoid predators or find mates and thus prefer nights with more lunar illumination (Bishop, 2005; Granda et al., 2008; Grant et al., 2013). Regardless, there are strong incentives for anurans to time their calling activity to coincide with certain environmental conditions. The fact that detection of R. okaloosae appears unrelated to environmental conditions may indicate that R. okaloosae are calling during sub-optimal conditions to avoid competition with R. c. clamitans; this phenomenon of subordinate species being pushed to marginal environments has been documented for a variety of taxa (Jaeger, 1971; Svoboda and Henry, 1987). Alternatively, the relative lack of relationships between activity of R. okaloosae and ambient conditions may simply reflect an artifact of their calling locations. Calling perches of males are typically partially submerged, where changes in humidity or wind speed pose less of a threat (Gorman et al., 2009; Gorman and Haas, 2011). Given that R. c. clamitans occur at all sites of R. okaloosae in our study region (and hence across their entire known range), disentangling competitive exclusion from innate preference will likely require a more experimental approach (e.g., a removal experiment; Inger and Greenberg, 1966; Colwell and Fuentes, 1975; Díaz et al., 2003).
Anuran phenological patterns are more nuanced than simply the presence or absence of calling males. Documenting the presence of one male calling and attributing this to ‘favorable’ conditions fails to take into account all other males in the population that did not call in those conditions (Weir and Mossman, 2005; Steelman and Dorcas, 2010). When looking at the calling index of R. okaloosae, i.e., the intensity and size of the breeding chorus, we did find intuitive links to environmental conditions. For instance, calling index of R. okaloosae was highest on humid, wind-free nights. Thus, calling index might be seen as a more reliable gauge of what environmental cues frogs (as a whole) respond to (Steelman and Dorcas, 2010). Our results show that differences in phenology can be subtle and complicated, requiring multiple years of data collected under a variety of conditions to parse out. Although surveys are often timed to coincide with peak calling times, data collected either intermittently or over relatively short time scales will hamper efforts to accurately describe a species’ phenology and discern population trends (Oseen and Wassersug, 2002; Genet and Sargent, 2003; Weir et al., 2005; Saenz et al., 2006; Steelman and Dorcas, 2010). Moreover, chorus tenure (the average time an individual male spends calling) is typically a fraction of the length of the breeding season (Murphy, 1992; Bevier, 1997), such that estimating abundance or fitness metrics from call surveys alone remains problematic (Gaston et al., 2001; MacKenzie and Nichols, 2004; Corn et al., 2011; Grant et al., 2018). To fully understand interactions in anuran communities, therefore, will require additional sources of data (such as mark–recapture to quantify individual calling tenures, or genetic analyses to reveal rates of hybridization) or surveys that document breeding activities rather than just male calls.
Although the data presented here indicate that populations of R. okaloosae appear to be effectively mitigating competition with sympatric R. c. clamitans through a variety of mechanisms, concerns remain. Species with small range sizes are disproportionately threatened with extinction (Stuart et al., 2004; Wake and Vredenburg, 2008; González-del-Pliego et al., 2019). Further, the threats posed to microendemics are expected to be exacerbated by climate change (Levin et al., 1996; Chunco, 2014; Todesco et al., 2016). Climate change is expected to both alter species’ phenology and induce range shifts for a variety of taxa (Thomas et al., 2004; Renner and Zohner, 2018). If species respond differently to climate change, competitive asymmetries could emerge that disrupt coexistence mechanisms and drive local extinctions (Edwards and Richardson, 2004; Todesco et al., 2016; Scranton and Amarasekare, 2017). Whilst much work has been done on the loss of suitable habitat driving amphibian extinctions under future climate scenarios, the threat posed by a breakdown in reproductive barriers with closely related species has seldom been explored (but see Pfennig and Simovich, 2002; Pfennig, 2007). We hope this work brings attention to the endemic fauna of the Southeastern Coastal Plain and the precarious existence of the Florida Bog Frog.
DATA ACCESSIBILITY
All data and code needed to perform the analyses are available on GitHub: https://github.com/geobro1992/bog-frog-phenology. Supplemental material is available at https://www.ichthyologyandherpetology.org/h2023035. Unless an alternative copyright or statement noting that a figure is reprinted from a previous source is noted in a figure caption, the published images and illustrations in this article are licensed by the American Society of Ichthyologists and Herpetologists for use if the use includes a citation to the original source (American Society of Ichthyologists and Herpetologists, the DOI of the Ichthyology & Herpetology article, and any individual image credits listed in the figure caption) in accordance with the Creative Commons Attribution CC BY License.
AI STATEMENT
The authors declare that no AI-assisted technologies were used in the design and generation of this article and its figures.
Detectability of Bronze Frog (R. c. clamitans) and Florida Bog Frog (R. okaloosae) as a function of (A) date and (B) time of night. Shaded regions show the 95% confidence intervals.
Detectability of Bronze Frog (R. c. clamitans) and Florida Bog Frog (R. okaloosae) as a function of (A) relative humidity, (B) lunar illumination, and (C) cloud cover. Lunar illumination (0 = new moon, 1 = quarter moon, 2 = intermediate phase moon, 3 = full moon), cloud cover (0 = no clouds, 1 = partly cloudy, 2 = overcast), and wind speed (Beaufort wind scale) are categorical variables (light condition, sky code, and wind scale, respectively) that have been converted to continuous predictors for analysis. Shaded regions show the 95% confidence intervals.
Calling index of Bronze Frog (R. c. clamitans) and Florida Bog Frog (R. okaloosae) as a function of (A) date and (B) time of night. Calling index is measured on a scale from 0 to 3, indicating whether calls were (0) absent, (1) audibly separated, (2) overlapping but non-continuous, or (3) continuous chorus.
Calling index of Bronze Frog (R. c. clamitans) and Florida Bog Frog (R. okaloosae) as a function of (A) relative humidity, (B) lunar illumination, (C) cloud cover, (D) temperature, and (E) wind speed. Lunar illumination (0 = new moon, 1 = quarter moon, 2 = intermediate phase moon, 3 = full moon), cloud cover (0 = no clouds, 1 = partly cloudy, 2 = overcast), and wind speed (Beaufort wind scale) are categorical variables (light condition, sky code, and wind scale, respectively) that have been converted to continuous predictors for analysis. Calling index is measured on a scale from 0 to 3, indicating whether calls were (0) absent, (1) audibly separated, (2) overlapping but non-continuous, or (3) continuous chorus.
Contributor Notes
These authors contributed equally.
Present address: Center for Limnology, University of Wisconsin-Madison, 680 N Park St., Madison, Wisconsin 53706.
Present address: Aquatic Resources Division, Washington State Department of Natural Resources, 1111 Washington St. SE, Olympia, Washington 98504.
Associate Editor: J. M. Davenport.