Demographic Assessment of a Freshwater Turtle Assemblage in an Urban Protected Area in the Context of Ongoing Threats and a Mass-Mortality Event
Abstract
Chronic threats and mass-mortality events in urban areas impact wildlife and can disproportionately affect turtle populations because their slow life history limits population recovery. Demographic studies of urban turtles are important, especially where ongoing conservation efforts (e.g., headstarting) occur and historical data on species status are absent. Our study focused on a community of headstarted Blanding’s (Emydoidea blandingii) and naturally occurring painted (Chrysemys picta) and snapping (Chelydra serpentina) turtles that inhabit a wetland complex in an urban area of Toronto, Ontario, Canada. We assessed demography and biomass at 3 sites using mark-recapture data collected from 2018 to 2021. Abundance estimates were highest for painted turtles, followed by Blanding’s and snapping turtles, while survival estimates varied depending on the species. Sex ratios of Blanding’s turtles in human care and postrelease were similar. Both painted and snapping turtles displayed unbiased sex ratios, except at 1 site, where painted turtles displayed a female-biased sex ratio. The Blanding’s turtle population was juvenile-biased because they are part of an ongoing headstarting program, whereas the painted and snapping turtle populations were adult-biased. Biomass of snapping turtles was the highest despite low abundance, highlighting their functional role in the ecosystem. Our findings indicate that turtle communities can persist in urban habitats; however, ongoing threats and catastrophes may pose a risk to population stability.
Native wildlife in urban areas experience numerous direct and indirect threats of varying magnitude (McKinney 2008; Murray et al. 2019). These threats include altered habitat quality and quantity, introduction of nonnative predators, collection of wildlife as pets, fatal behavioral changes in animals, and exposure to physiological stressors such as pollutants (Garber and Burger 1995; Salmon et al. 1995; Pitts et al. 2017; Fritsch et al. 2019; Baker and Potter 2020). Although urbanization is considered one of the major causes of biodiversity loss, conservation efforts in urban areas are sometimes met with skepticism (Sanderson and Huron 2011). Mitigating threats to wildlife in urban areas is challenging because of a lack of high-quality continuous habitats, increasing negative effects of human activities, and the perception that wildlife populations in urban areas inevitably face extirpation (Soanes et al. 2019). However, recent evidence suggests that some wildlife can thrive in urban areas by shifting their behaviors or modifying life-history traits (Ditchkoff et al. 2006), and urban areas may even support endemic species (Aronson et al. 2014).
In addition to chronic threats, wildlife populations also experience stochastic acute events such as catastrophes. A mass-mortality event is a catastrophe that results in the loss of a significant portion of a population over a short period, leading to detrimental shifts in population size and structure (Fey et al. 2015). Although catastrophes can occur in both urban and nonurban habitats (Brooks et al. 1991; Wijewardena et al. 2025), wildlife populations in urban habitats may have a harder time recovering from such catastrophes, given the multitude of chronic threats present in urban habitats. Both natural and human-induced causes of mass-mortality events have been documented for herpetofauna, including disease, native predators, invasive species, and extreme climatic conditions (Brooks et al. 1991; Letnic et al. 2008; Agha et al. 2017; Gasbarrini et al. 2021).
Freshwater turtles are impacted by numerous threats (Lovich et al. 2018; Stanford et al. 2020; Cox et al. 2022), including urbanization (French et al. 2018). Turtles in urban habitats experience a suite of threats, including habitat degradation and fragmentation, increased mortality from subsidized predators, mortality from collisions with vehicles, altered nesting habitats resulting from nonnative vegetation, and reduced hatching success due to pollutants (Gibbs and Shriver 2002; Kolbe and Janzen 2002; de Solla et al. 2008; Piczak et al. 2019; Wijewardena et al. 2023). Although harmful effects of urbanization on turtles have been documented, some populations benefit from the direct and indirect effects of urbanization. For example, turtles in highly eutrophic habitats can have greater population densities and somatic growth rates and beneficial shifts in size class structures compared to nonurban populations (Galbraith et al. 1988; Spinks et al. 2003; Roe et al. 2011). In some populations, the effects of urbanization are neutral, such as in a population of eastern long-necked turtles (Chelodina longicollis) where turtles in urbanized areas had similar abundance, growth rate, reproductive output, and sex ratios compared to turtles in nonurban sites (Serrano et al. 2020). Turtles are especially sensitive to the effects of mass-mortality events because of their slow life history (i.e., delayed sexual maturity, long generation times), and they have a weak ability to compensate for losses in abundance and shifts in age structures (Heppell 1998; Enneson and Litzgus 2008; Keevil et al. 2018). Thus, we have a limited understanding of how turtle populations respond to urbanization at a local scale.
To effectively manage species of conservation interest in the long term, a robust understanding of the effects of chronic threats and catastrophes at a local scale is needed. Such knowledge can inform conservation priorities in areas where long-term conservation efforts such as headstarting programs (i.e., captive rearing and release of turtles; Burke 2015) are already underway. For example, evaluating the demography of sympatric species can inform the status and trends of well-established populations and provide insights on the persistence of the headstarted population following human intervention.
Our study focused on an urban community of freshwater turtles at the Rouge National Urban Park (RNUP) in Toronto, Ontario, Canada, where headstarting is an ongoing conservation priority for 1 turtle species. The turtle community comprises a functionally extinct population of Blanding’s turtle (Emydoidea blandingii) and 2 other sympatric species, the painted (Chrysemys picta) and snapping (Chelydra serpentina) turtles. Fewer than 10 wild Blanding’s turtles have been observed since the early 2000s, and the population is annually supplemented with headstarted turtles. Given the highly urbanized nature of the area, the turtle community at the RNUP has been subjected to a multitude of chronic threats, including nest predation by subsidized predators (Toronto Zoo, unpubl. data), road mortality (Leermakers 2020), habitat fragmentation (Frost-Wicks 2019), and a recent mass-mortality event (Wijewardena et al. 2025). We previously assessed the demography of the population to evaluate the short-term success of the headstarting program (Wijewardena et al. 2023). Both painted and snapping turtles have been observed since the early 2000s; however, their demography has not been assessed. The objective of our current study was to provide an assessment of abundance (population density), survival, sex ratio, size distribution, and biomass for all 3 species in the community. Biomass is an informative metric in demographic studies in urban areas because urban pressures could disproportionately affect larger animals with slow life histories (Fisher and Owens 2004) and potentially alter population structures and energy flow within the ecosystem. These demographic and biomass metrics collectively provide insights into the ecological roles and contributions of the turtle community in the urban environment of the RNUP.
METHODS
Study Areas. —
We surveyed 3 wetland sites within the RNUP in Toronto, Ontario, Canada (Fig. 1). Site A is a constructed wetland complex with plant species including alders (Alnus spp.), cattails (Typha spp.), sedges (Carex spp.), and willows (Salix spp.). Site B is a natural wetland complex with vegetation such as cattails, common duckweed (Lemna minor), greater duckweed (Spirodela polyrhiza), purple loosestrife (Lythrum salicaria), and sago pondweed (Stuckenia pectinata). At sites A and B, 22-month-old juvenile Blanding’s turtles were released as part of the headstarting program at the Toronto Zoo. From 2014 to 2020, 270 headstarted turtles were released at site A, while 48 were released at site B in 2021. The third study site (site C) is a small pond with vegetation such as cattails, common duckweed, and waterlily (Nymphaea spp.). We excluded the exact location of the study sites to minimize the poaching risk for the endangered (COSEWIC 2016) Blanding’s turtle.
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 24, 1; 10.2744/CCB-1641.1
Mass-Mortality Event. —
We observed a mass mortality of Blanding’s (n = 48) and painted turtles (n = 57) in 2020 in site A (Wijewardena et al. 2025). The majority of painted turtle carcasses were adult females, whereas, for Blanding’s turtles, the carcasses were all headstarted juveniles. Based on carcass condition (e.g., loss of limbs and head), we suspected that the mortality was caused by a mammalian predator such as a racoon (Procyon lotor) or a mustelid species. Additional information about the mass-mortality event can be found in Wijewardena et al. (2025).
Survey Protocol. —
We conducted trapping and visual encounter surveys during the turtle active season (late April to early September) from 2018 to 2021 in all 3 sites, except in 2020 when our surveys were restricted due to COVID-19 regulations (Fig. 1). In 2020, we conducted trapping surveys from June to early August at site A and for 1 wk in late July–early August at site B. No surveys were conducted at site C in 2020 because we were unable to obtain permits. We captured turtles using hoop traps and basking traps (2018 only) and by hand. All hoop traps were baited with canned sardines or cat food. We checked traps once every 24 h from 2018 to 2021 and checked traps twice daily in 2021 from July to August. Number of traps deployed varied each year based on site; we deployed 6–10 traps at site A, 4–6 at site B, and 2 at site C per year. All captured turtles were processed (see below) in the field and then released at their capture locations.
For Blanding’s and painted turtles, we measured shell height, midline carapace length and width, and plastron length and width in mm using ZEAST Vernier calipers and body mass in g using a Pesola scale. For snapping turtles, we measured shell height, curved carapace length and width, curved plastron length and width in mm (measuring tape), and body mass in lb (hanging scale), which was converted to g prior to data entry. The majority of captured Blanding’s turtles were juveniles from the headstarting program that were notched and tagged (see below); thus, we determined sex using the incubation temperature. Blanding’s turtles have temperature-dependent sex determination (Gutzke and Packard 1987; Ewert and Nelson 1991); eggs incubated at 27.5°C were considered males, whereas eggs incubated at 29.5°C were considered females. The male:female sex ratio reared in human care was 1:1.5. We also captured an adult female Blanding’s turtle in the spring of 2021. The sex of adult Blanding’s and painted turtles was determined using secondary sexual characteristics (Ernst and Lovich 2009). We used the stimulated penis-eversion method described in Dustman (2013) to determine the sex of snapping turtles. Each headstarted Blanding’s turtle was uniquely marked by marginal scute notching (Cagle 1939) and a Passive Integrated Transponder (PIT) tag injected into in the left hind leg prior to release. Painted turtles were also notched and PIT-tagged in the field upon capture. In 2021, we took digital photographs of Blanding’s and painted turtle plastrons as another method to identify individuals. All snapping turtles were PIT-tagged in their left hind leg. Each snapping turtle was given a unique notch in 2018, but only 1 notch was given from 2019 onwards because fewer marginal scutes were safely accessible to the researcher. Female painted turtles were palpated to check for the presence of eggs. If an individual was confirmed to be gravid, we marked the turtle with a unique notch but avoided PIT tagging to minimize stress to the turtle. We did not catch any gravid snapping turtles.
STATISTICAL ANALYSIS
Abundance and Survival. —
We previously estimated the abundance and survival of headstarted Blanding’s turtles at site A (Wijewardena et al. 2023) using the POPAN formulation in Program MARK (White and Burnham 1999). Only 1 wild female Blanding’s turtle has been captured in site A, and at least 4 other transient individuals have been observed in the RNUP from the early 2000s to 2021 (Toronto Zoo, unpubl. data), but we could not estimate survival from these data. We were unable to estimate the abundance and survival of snapping turtles using population models because of low captures and recaptures (Supplemental Table S1; all supplemental material is available at 10.2744/CCB-1641.1.s1); thus, we computed minimum abundance as the unique number of individuals captured at each site.
For painted turtles, we used a robust design model (RDHuggins) in Program MARK via the RMARK package (Laake 2013) in R (R Core Team 2021). We used the robust design because road mortality and nesting surveys in the area indicated that painted turtles move between and away from sites in contrast to Blanding’s turtles that remained at their release site (Toronto Zoo, unpubl. data). We estimated abundance, apparent survival (Φ), temporary emigration (TE), and capture (p) and recapture probability (c). The robust design model combines both closed- and open-population structures to obtain robust estimates of parameters that are not estimable under either closed- or open-population models alone. For example, Cormack-Jolly-Seber models cannot account for individual capture heterogeneity within a sampling occasion or permanent trap response (trap-happy or trap-shy behaviors) among individuals. Robust design models include primary sampling occasions during which the population is considered open and secondary sampling occasions during which the population is assumed to be closed. They estimate abundance based on the secondary occasions (closed model) and survival based on the primary occasions (open model). These models also incorporate temporary emigration such that an animal marked in a primary occasion may be temporarily unavailable for capture during subsequent occasions but become available for capture again at a later sampling occasion.
We included 4 primary periods (sampling years 2018–2021) and 3 secondary periods (each 2 mo long; April–May, June–July, and August–September) per primary period. The secondary periods were delineated based on the assumption that the population is closed during each 2-mo period and based on turtle behavior (e.g., prenesting season: April–May, nesting season: June–July, postnesting season: August–September) that may influence capture rates. We were unable to begin surveys until June in 2020; thus, we fixed the capture probability in April–May (secondary period) in 2020 at zero and further fixed the capture and recapture probability for site C at zero to account for missed surveys (see “Methods”). We captured 367 adult painted turtles and grouped them based on sex and site, which resulted in six groups. We omitted 115 juveniles from our analysis because juveniles often violate model assumptions due to their capture heterogeneity. During the study period, 2 male and 1 female painted turtles moved between sites. The first turtle was found at site A in 2018 and later recaptured at site B in 2019, whereas the second turtle was found at site C in early 2018 and recaptured at site A in late 2018. Thus, we created a site-specific identity for each migrant turtle in the encounter history; doing so does not bias the apparent survival or abundance estimates.
We considered several factors that could affect the survival of painted turtles, including sampling year, sex, and site. We further considered factors such as transient individuals and the mass-mortality event in 2020. Thus, we parameterized the model such that survival in 2019–2020 would be different from survival in 2018–2019 and 2020–2021. We also considered additive and interactive models of a subset of variables (Supplemental Table S2). We included a model with constant survival because the estimates provided by the top-ranked models indicated overlapping confidence intervals for survival. For capture and recapture probability, we assumed no capture heterogeneity (i.e., p = c) based on estimates provided by U-CARE for the fully time-dependent model. We considered capture probability to vary depending on sex and site. We further considered biologically relevant interactive effects (e.g., survey protocol) and parameterized capture probability to vary depending on sampling year and season, or site and season. For temporary emigration, we included random, null, and Markovian emigration, but models with Markovian emigration yielded inestimable parameters, so we did not examine Markovian models further. We evaluated model fit using Fletcher’s c-hat calculated with Program MARK for the most parameterized model. This model included the effects of site on survival, sampling year and season on capture probability, and random emigration parameters. We ranked models using the Akaike Information Criterion (AICC) for small sample sizes (Burnham and Anderson 2002).
Sex Ratio. —
We examined whether the sex ratio of painted and snapping turtles deviated from parity (1:1 male:female) and whether the presumed sex ratio (based on incubation temperature) of headstarted Blanding’s turtles differed from 1:1.5 using χ2 tests. Sex ratio was assessed using trapping data collected from 2018 to 2021. We used the first capture of each turtle in our analysis. For Blanding’s turtles, given that sex was presumed based on incubation temperature, we excluded 18 headstarted turtles that lacked incubation temperature data.
Size Distribution. —
For size distribution, we used frequency histograms for each species. We categorized midline carapace length into 5 mm increments because Blanding’s and painted turtles were typically less than 200 mm in length. We used a similar method for snapping turtles to facilitate comparison of size distributions. We further grouped each species into male, female, and unknown sex categories. We selected size as opposed to age because age cannot be accurately determined in freshwater turtles, especially in adults (Brooks et al. 1997).
Biomass. —
We calculated biomass by multiplying the mean body mass (g) of each species by their abundance estimate, then dividing by pond surface area. For Blanding’s and painted turtles, we obtained abundance estimates using POPAN and robust design models, respectively. For snapping turtles, we used the minimum abundance by counting unique individuals at each site. For both painted and snapping turtles, we included the unique number of juveniles in the biomass estimation. To calculate the mean body mass, we only used data from the first capture of all individuals. In painted turtles, we excluded 3 females because of missing or inaccurate body mass data and 7 females that were gravid at the time of capture. In snapping turtles, we removed 2 juveniles due to missing body mass data. Given that painted and snapping turtles display sexual size dimorphism (Mosimann and Bider 1960; Litzgus and Smith 2010), we calculated biomass separately for each sex at each site. Pond area was obtained using information on the Ontario Hydro Network Waterbody shapefile (Ontario Geohub 2022).
RESULTS
Abundance and Survival. —
For painted turtles, 3 models in our candidate model set had ΔAICc < 2 with similar deviances (Supplemental Table S2). The top models included the null model and models with the effects of sex and the mass-mortality event. Thus, based on model-averaged estimates at the end of the study period in 2021, site A had the highest abundance with 110 (CI = 74–178) females and 126 (CI = 85–202) males; site B had 28 (CI = 17–57) females and 50 (CI = 32–90) males; and site C had the lowest abundance, with 9 (CI = 5–27) females and 28 (CI = 17–57) males. In 2020, female painted turtle abundance in site A was 3-fold higher than in other years (Fig. 2); however, we were unable to estimate abundance at sites B and C for 2020 because our sampling effort was hindered by restrictions associated with the COVID-19 pandemic. Based on model-averaged estimates, the annual survival of painted turtles was similar among sites and between sexes across years. For females at site A, survival ranged from 0.53 (CI = 0.34–0.72) to 0.57 (CI = 0.35–0.77), while for males, survival ranged from 0.58 (CI = 0.36–0.77) and 0.62 (CI = 0.41–0.79) depending on the year. At site B, female survival ranged from 0.54 (CI = 0.34–0.72) to 0.57 (CI = 0.35–0.77) and male survival ranged from 0.58 (CI = 0.36–0.77) to 0.62 (CI = 0.41–0.79). At site C, female survival ranged from 0.54 (CI = 0.34–0.73) to 0.58 (CI = 0.36–0.77) and male survival ranged from 0.58 (CI = 0.36–0.77) to 0.62 (CI = 0.41–0.79). Recapture probability varied from 0.002 (CI = 0.002–0.017) to 0.18 (0.12–0.28) depending on sampling year and season and was similar between sexes and sites.
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 24, 1; 10.2744/CCB-1641.1
For snapping turtles, the minimum abundance at site A was 9 males, 13 females, and 12 juveniles, while at site B it was 18 males, 13 females, and 4 juveniles, and at site C it was 2 males, 1 female, and 4 juveniles.
Sex Ratio. —
Based on data collected in 2018–2021, the male:female sex ratio of Blanding’s turtles at site A was 1:1.15 and was similar to the sex ratio produced in human care. The male:female sex ratio of painted turtle differed from 1:1 at site A, but not at sites B and C (Table 1). In snapping turtles, the adult sex ratio was equal at sites A and B (we captured only 3 snapping turtles at site C, thus we were unable to assess sex ratio at that site).
Size Distribution. —
The size distribution of Blanding’s turtles showed a juvenile-biased population structure (Fig. 3). The mean midline carapace length of headstarted turtles was 107.4 mm (min = 87.5; max = 139; n = 76). The size distribution of painted turtles showed an adult-biased population structure (Fig. 3). The mean midline carapace length of adult female turtles was 134.8 mm (min = 87; max = 180; n = 186), and 119.3 mm (min = 86; max = 160; n = 175) for adult males. The size distribution of snapping turtles was adult-biased (Fig. 3), as no individuals with a curved carapace length of 40–90 mm were captured during surveys. The mean curved carapace length of adult males was 317 mm (min = 200; max = 380; n = 29), while for females it was 298 mm (min = 233; max = 370; n = 27).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 24, 1; 10.2744/CCB-1641.1
Biomass. —
The biomass of snapping turtles was the greatest, followed by painted and Blanding’s turtles (Table 2). The biomass of each species was highest at site A and lowest at site C. The biomass of female painted turtles was higher than that of males. In contrast, the biomass of male snapping turtles was higher than that of females.
DISCUSSION
Abundance and Survival. —
Painted turtles were the most abundant species at our study sites, followed by headstarted Blanding’s turtles and then snapping turtles. This result is not surprising given that painted turtles are a common species in North America (Ernst and Lovich 2009) with density estimates typically ranging from 2–40 turtles/ha (COSEWIC 2018) and as high as 827 turtles/ha (Frazer et al. 1991). Painted turtles occupying coastal wetlands in Toronto and other urban areas in Canada have densities ranging from 0.08 to 1.8 turtles/ha (Marchand et al. 2018; Dupuis-Desormeaux et al. 2021). Blanding’s turtles were more abundant than snapping turtles as a result of the headstarting program, and we expect Blanding’s turtle abundance to increase with time as headstarting continues and as juveniles become sexually mature and contribute to population growth. Snapping turtles had the lowest abundance and density but the greatest biomass. The low capture and recapture rate of snapping turtles could be a result of survey methods. For example, traps with larger mesh sizes (e.g., 10 cm) are better at capturing snapping turtles compared to traps with smaller mesh sizes (Ennen et al. 2021). The majority of the traps used in our surveys had a mesh size of 2.5 cm and had a trap diameter of 50–55 cm. Additionally, we caught juvenile Blanding’s turtles and painted turtles of various life stages by hand; however, catching snapping turtles by muddling was not successful.
A caveat in density estimation is that density is sensitive to abundance and area calculations. Both metrics are difficult to obtain without long-term monitoring. Population size estimates can be biased depending on the sampling method and analyses (Koper and Brooks 1998). In addition, pond surface areas vary depending on seasonal and annual weather patterns. In urban landscapes, aquatic habitats also face additional challenges such as habitat degradation from nutrient loading, soil erosion, and invasive vegetation (Faulkner 2004; Ruas et al. 2022). As a result, urban landscapes may undergo drastic changes in habitat quality in a shorter amount of time compared to nonurban habitats. Thus, abundance and density estimates should be reviewed periodically to obtain an accurate assessment of the dynamics of demography and community structure.
Survival of headstarted Blanding’s turtles was high (∼89%), except during the mass-mortality event that occurred in 2019 and 2020 when survival was 43% (Wijewardena et al. 2023). Long-term population stability generally requires high adult survivorship (Congdon et al. 1993; King et al. 2021), and it remains to be seen whether juvenile headstarted turtles will maintain high survivorship as they reach adulthood. However, the adult survival estimate of other species inhabiting the same wetland complex is relatively low. For example, the survival of painted turtles was 53%–62% and is a cause for concern. Throughout its range, survival of adult painted turtles varies widely depending on sex and geography. In 2 well-studied populations in Ontario and Virginia, the survival of adults was over 94% and 96%, respectively (Mitchell 1988; Samson 2003), whereas, in a Michigan population, survival ranged from 64% to 83% in males and 29% to 50% in females (Frazer et al. 1991). There are several ongoing threats to survival of turtles at the RNUP, including subsidized predators such as coyote (Canis latrans), American mink (Neogale vison), raccoon, and red fox (Vulpes vulpes). We have observed nest, hatchling, and adult depredation by mammalian mesopredators at the RNUP (Toronto Zoo unpubl. data). In addition, the extensive road network that fragments the park (Frost-Wicks 2019; Leermakers 2020) could lead to turtle population extirpation by increasing mortality of adults that migrate in search of mates, nesting habitats, and overwintering habitats. For example, a road network in southern Ontario caused an ∼80% decline in a snapping turtle population within 17 yr (Piczak et al. 2019). Mitigating loss of adult turtles to predators and vehicular collisions could become a conservation priority in the future and add strain to the ongoing conservation efforts to recover more critically endangered species in the area.
Sex Ratio. —
The male-to-female sex ratio of the headstarted Blanding’s turtle population (1:1.15) was similar to the sex ratio produced in human care (1:1.5). Although statistically there was no difference, the observed ratios indicate lower number of female captures in the wild, suggesting that female mortality may be higher than that of males. Additionally, these estimates are based on presumed sex based on incubation temperature. Once the headstarted turtles reach sexual maturity and when the validity of presumed sex is confirmed, future studies should explore whether female turtles experience higher mortality in the wild. Sex ratio of snapping turtles indicated parity; however, our sample size was small, especially for site C. Sex ratio of painted turtles at site A was female-biased. This bias likely resulted from the relatively higher number of females captured in 2020 compared to other years and may have resulted from a drought that caused many ponds to dry and water levels to decrease. We suspect that female turtles from sites B and C may have moved to site A during this period. Similar movements have been reported in other populations during droughts when turtles emigrate in search of permanent waterbodies (Congdon and Gibbons 1996; Roe and Georges 2007; Krochmal et al. 2018). In addition, female turtles may have entered the baited hoop traps more frequently than males, given the higher energy needed for egg development and nesting. A study examining pond use by painted turtles indicated that they prefer shallow ponds with a large littoral zone, low visibility, and no fish (Hughes et al. 2016). These characteristics are more common at site A, compared to sites B and C. Thus, the drought may have motivated painted turtles to emigrate to search for better refugia. Unfortunately, we were unable to sample sites B and C in 2020, which makes it difficult to test this hypothesis. Sex ratios may also differ from unity because of unequal primary sex ratios (i.e., sex ratio at hatching) and differences in age at maturity (Lovich and Gibbons 1990).
Size Distribution. —
The size distribution of Blanding’s turtles showed a juvenile-biased distribution given the low abundance of adult turtles and the annual addition of headstarted juveniles into the RNUP. The size distribution of painted and snapping turtles was adult-biased, which is typical of turtle populations in Canada (Browne and Hecnar 2007) and the United States (Gulette et al. 2019; Vanek and Glowacki 2019), although a bimodal distribution has also been observed for snapping turtles (Hughes and Meshaka 2020). Interestingly, we observed several large males but few small males in the RNUP wetland complex, which may suggest that smaller males are limiting home-range overlap with larger individuals to avoid aggressive interactions (Galbraith et al. 1987; Keevil et al. 2017). Size distribution can be an artifact of sampling methods as opposed to true size distribution because trapping can generate biases in sex and age classes (Tesche and Hodges 2015). However, we sampled the turtle community at RNUP using multiple methods, which likely reduced such biases. In addition, juveniles are difficult to detect using conventional survey methods (Pike et al. 2008). Bias toward adults is generally consistent with the life-history strategy of turtles and contributes to stable population growth (Congdon et al. 1993, 1994).
Biomass. —
Compared to the other 2 turtle species at RNUP, biomass of snapping turtles was the highest despite their low abundance. In other populations, biomass of snapping turtles ranged from 5 to 25 kg/ha (Dreslik et al. 2005) and reached as high as 340 kg/ha in highly eutrophic ponds (Galbraith et al. 1988). The abundance of snapping turtles in the RNUP is an underestimation of their true population size, given that we counted only the number of unique individuals. Thus, their biomass contribution is likely greater than we estimated. In contrast, the painted turtle biomass at RNUP was lower than that of snapping turtles despite painted turtles being most abundant at each site. Biomass estimates for many painted turtle populations ranged from 4.6 to 154 kg/ha (COSEWIC 2018). In Blanding’s turtles, juveniles contributed most to the observed number, which is not surprising given the low number of adults in the park. Biomass estimates of Blanding’s turtles are lacking but can vary from 0.26 kg/ha to 3.5 kg/ha (Rohde 2023) in Michigan and 45 kg/ha in Wisconsin (Ross 1989). The mass-mortality event that occurred in 2019–2020 substantially decreased the abundance and biomass of both Blanding’s and painted turtles (Wijewardena et al. 2025).
At the RNUP, the total biomass of turtles was approximately 93 kg/ha. Total biomass of freshwater turtles in other restored sites ranges from 160 to 250 kg/ha in Missouri (Nickerson et al. 2019), whereas, for other vertebrates, such as salamanders and frogs, biomass ranges from 68 to 1490 kg/ha, respectively (Gibbons et al. 2006). Typically, the biomass of turtles is disproportionately higher compared to other vertebrates and, thus, turtles can disproportionately influence ecosystem processes (e.g., energy flow within and between ecosystems, mineral cycling, bioaccumulation, trophic status; Lovich et al. 2018). In urban environments where large mammalian predators (e.g., bears and wolves) are absent (Bateman and Fleming 2012), turtles may even become essential niche replacements of large mammals. We are unaware of any biomass estimates of other vertebrates (e.g., birds, mammals, reptiles) in the RNUP, but we suspect that it would be lower compared to that of turtles.
Management Implications. —
Based on our findings, the RNUP provides habitat to a substantial assemblage of turtles with several individuals of snapping turtles contributing to most of the biomass. As the wild population of Blanding’s turtles is annually supplemented by headstarted individuals, these biomass trends will likely change and may lead to shifts in turtle community metrics. We are uncertain whether the mass-mortality event occurred as a result of density-dependent regulation due to turtles reaching carrying capacity or whether it was a stochastic event that was exacerbated by climatic conditions. Evidence for density-dependent population regulation is limited in turtle populations (Rodríguez-Caro et al. 2016; King et al. 2021). From an ethical and conservation standpoint, we recommend that additional sites should be considered for the release of headstarted turtles to avoid the potential negative consequences of reaching carrying capacity.
Given the absence of historical data for the areas we surveyed, we are unable to assess whether the RNUP turtle populations are growing, stable, or declining. Road mortality, low habitat connectivity, and subsidized predators are major threats that affect the turtle community. Studies have indicated that chronic mortality, even at low levels, can have adverse effects on turtle population persistence (Congdon et al. 1993, 1994), and catastrophic events can increase the extinction risk of healthy populations of turtles (Keevil et al. 2018). In the short period that we surveyed the turtle community, we observed both mortality processes. Thus, it is highly likely that turtle populations in RNUP are declining, except for the Blanding’s turtle population, which is annually supplemented by headstarting. Additional conservation interventions should be considered to mitigate the ongoing threats to turtle populations in the park. Identifying road-mortality hot spots, improving habitat connectivity by installing eco-passages, creating more wetlands that are resilient to effects of extreme weather (e.g., deeper ponds), and creating and maintaining nesting habitats that are safe from human disturbances and mesopredators are the minimum requirements needed to address the conservation needs of the 3 species. A landscape-level conservation action plan will likely provide the maximum benefit for the turtle community of the RNUP. Our results suggest that turtle populations can reach high biomass in urban habitats and that conservation activities in urban areas should not be abandoned despite the numerous challenges that persist in such habitats.
Schematic diagram of the study sites (A–C) and the sampling periods (denoted by transparent hoop trap) for each site at the Rouge National Urban Park in Ontario, Canada. The Blanding’s turtle (Emydoidea blandingii) icon indicates the sites where headstarted turtles were released. (Stock images obtained from Canva 2024.)
Abundance estimates for female and male painted turtles (Chrysemys picta) at the Rouge National Urban Park in Toronto, Ontario, Canada, 2018–2021. The estimates were generated using a robust design model. Error bars indicate standard errors. In 2020, site B was sampled for only 1 wk, and site C was not sampled due to logistic challenges associated with the COVID-19 pandemic.
Size distributions of Blanding’s (Emydoidea blandingii), painted (Chrysemys picta), and snapping turtles (Chelydra serpentina) based on carapace length (mm) at first capture during a 4-yr (2018–2021) study in the Rouge National Urban Park in Toronto, Ontario, Canada. In Blanding’s turtles, “unknown” refers to headstarted juvenile turtles.
Contributor Notes
Handling Editor: Peter V. Lindeman
