Mass Mortality in a Community of Headstarted (Emydoidea blandingii) and Naturally Occurring (Chrysemys picta marginata) Freshwater Turtles in Protected Urban Wetlands

Study sites.—

Three sites within the RNUP, totaling 33.8 ha of wetland habitat, were monitored as part of ongoing turtle headstarting and monitoring initiatives. Actual site names, locations, and associated details are omitted to protect the SAR involved, including Blanding’s, Midland Painted (hereafter Painted Turtle; Chrysemys picta marginata), and Common Snapping (hereafter Snapping Turtle; Chelydra serpentina) turtles.

Site A is composed of a complex of shallow ponds, meadow marsh, thicket swamp, and successional woodland. Its prominent aquatic vegetation species include Canadian Pondweed (Elodea canadensis), cattails (Typha spp.), Common Bladderwort (Utricularia vulgaris), Common Reed (Phragmites australis), Coontail (Ceratophyllum demersum), muskgrasses (Chara spp.), and water lilies (Nymphaea spp.). Site B is composed of shallow ponds and marsh surrounded by mineral deciduous swamp and successional woodland. Its prominent aquatic vegetation includes cattails, Common Duckweed (Lemna minor), Greater Duckweed (Spirodela polyrhiza), Purple Loosestrife (Lythrum salicari), and Sago Pondweed (Stuckenia pectinata). Site C is a cedar-grove wetland complex composed of a central shallow pond with nearby smaller waterbodies. Its prominent aquatic vegetation includes cattails, Common Duckweed, Common Reed, Coontail, and water lilies.

Blanding’s Turtle headstarting project.—

The Blanding’s Turtle headstarting project began with egg collection in 2012, and individuals were first released in 2014. This headstarting project aims to augment a historic population of Blanding’s Turtles that had decreased to fewer than ten individuals and was deemed functionally extinct following extensive surveys conducted in the 1990s and early 2000s (Wijewardena et al., 2023). The causes of the original population decline are unknown, but the effects of urbanization (e.g., habitat loss and fragmentation, road mortality) likely played a role. Populations at sites A (2014–2020) and B (2021–present) have been annually augmented with ∼22-month-old headstarted and 2–14-day-old hatchling Blanding’s Turtles. From 2014 to 2020, a total of 270 headstarted turtles and 184 hatchlings were released into site A, and from 2021 to 2022, 104 headstarted turtles and 34 hatchlings were released into site B. Headstarted individuals were marked with notches on their marginal scutes (Cagle, 1939) upon hatching, then prior to release, passive integrated transponder (PIT) tags were injected into their left hind-leg pocket and photos were taken of their plastron. Additionally, a randomly selected portion of headstarted turtles were fitted with an external glue-on radio transmitter. Notches, PIT tags, plastron photos, and transmitter frequencies (where possible) were used to verify the identity of dead Blanding’s Turtles. See Wijewardena et al. (2023) for additional details about the headstarting project.

Survey protocol.—

The Toronto Zoo conducted surveys from 2014 to 2022 using radio-telemetry of headstarted Blanding’s Turtles, live-trapping, opportunistic captures, nest site surveys, and road surveys for all freshwater turtle species at sites A–C (Table S1; see Data Accessibility). Due to the COVID-19 pandemic, sites were minimally surveyed in 2020. Radio-telemetry surveys were done more frequently during the active season of turtles (4–17 days/month; May–September) than during the non-active season (1–6 days/month; October–April) each year at site A (2014–2022) and sites B and C (2021–2022). A subset of released headstarted turtles and several adult Blanding’s Turtles were tracked using R4000 or R410 Receivers (Advanced Telemetry Systems, Inc., MN). Turtles were captured, processed, and released at the site of capture approximately monthly from May to October. Turtles of any species opportunistically captured by hand were also processed.

Live-trapping was conducted from 2017–2021, during the active season only, at site A (2017–2021), site B (2017–2021), and site C (2018–2019, 2021). We deployed 5–10 hoop traps, baited with cat food or canned sardines, and three basking traps (2018 only). Traps were checked daily, and captured turtles were photographed, measured (straight-line carapace length, carapace width, plastron length, plastron width, and shell height), unmarked individuals were notched and PIT tagged, and all turtles were released at the site of capture.

Nesting surveys were conducted during the nesting season (late May–early July 2017–2022) in the afternoon and evening to identify nesting turtles and depredated nests. Data were collected by walking 700–2,500 m transects, scanning for presence of nesting female turtles, other turtle nesting activity, and depredated nests and turtles. Road surveys were conducted at sites B and C during the active season (2017, 2018, 2021); in 2018 and 2021, road surveys were extended until October. Data were collected by walking each side of the road scanning for live and dead on- and beside-road wildlife from the center median line toward the shoulder. Mortalities of all animal species were recorded, but here we report only the mortalities of turtles.

In summary, survey effort generally increased from 2014 until 2019, and Toronto Zoo staff conducted rigorous surveys during summer across sites A–C. However, in 2020, only three survey types (radio-telemetry, live-trapping, and water chemistry sampling) were conducted in two sites (A and B) with reduced effort across all survey types, as a result of the COVID-19 pandemic (Table S1; see Data Accessibility).

Upon discovery of a carcass, the location, species, identifiers (notch, PIT tag number, plastron pattern, transmitter frequency), sex (male, female, juvenile, unknown), injuries, shell measurements, and probable cause of death (predation, road mortality, winterkill, unknown) of the turtle were recorded. Probable cause(s) of mortality were determined based on the condition of the carcass (Fig. 1). Predation was cued by obvious signs of injuries and damage to, or removal of, the head and limbs, which is characteristic of species known to prey on freshwater turtles (Lanszki et al., 2006). In contrast, road mortality was the documented cause of death when there was excessive bodily damage (e.g., cracked carapace and/or plastron) to carcasses located on or near a road. Winterkill was the documented cause of death when carcasses with no obvious signs of injuries or damage were observed in the winter or spring. Unknown cause of death was documented when there were no obvious signs of injury or damage on carcasses found outside of winter or spring, when destruction and/or decomposition of the carcass was too advanced to determine any signs of injury or damage, or as a result of missing data. Carcasses collected/identified by Parks Canada staff, who have regular presence in the RNUP, were also included in our summaries and analyses.

View Figure

• Download as Powerpoint
• Download Figure

Demography of carcasses.—

Sex was determined using secondary sex characteristics when present (Ernst and Lovich, 2009). Turtles were categorized into three age/sex classes based on relevant size for each species: female (adult), male (adult), and juvenile (unknown sex; Table 1). Shell measurements were recorded to the nearest 0.1 mm using ZEAST Vernier calipers (Table 2). Measurements were taken either in situ or at the Toronto Zoo. Carcass demography was recorded as “unknown” when characteristics required to determine sex or age class were not discernible and other identification methods (i.e., notch, PIT tag number, or transmitter frequency) could not be verified. The carcasses were brought to the Wildlife Health Centre at the Toronto Zoo to be temporarily stored in a freezer. Carcasses that had evidence of predation, or for which the cause of death was unknown and had enough soft tissue remaining, were sent to the Canadian Wildlife Health Cooperative (CWHC; Guelph, ON) for necropsy. Samples with tissue that were severely autolyzed only received gross examination. Where able, histopathology was used to rule out underlying disease. All other carcasses were sent to Parks Canada to be shared with partners where feasible, stored for future use, or respectfully buried.

Table 1.Number of mortalities by survey location, turtle species, and age class at the Rouge National Urban Park in Ontario, Canada from 2014 to 2022. Blanding’s Turtles include naturally occurring and headstarted individuals while all other turtles are naturally occurring individuals.

View Fullscreen

Table 2.Mean straight-line carapace length (CL) and standard deviation of turtle carcasses* based on species and age class at the Rouge National Urban Park in Ontario, Canada.

View Fullscreen

Ethics approval.—

The use of animals in this study was approved by the Toronto Zoo (Ref. No. 2010-01-01, 2014-03-01, 2017-03-01, 2020-02-01), Laurentian University (protocols AUP 2017-02-01 and 6020983), Parks Canada, and the Ministry of Natural Resources (MNR) Animal Care Committees, based on guidelines of the Canadian Council on Animal Care. All work was authorized by the provincial Endangered Species Act permits (GU-B-014-13, AU-B-010-14, AU-B-008-15, AU-B-011-16, AU-B-008-17, AU-B-009-18, and PS-B-004-18, CN-B-006-20), Fish and Wildlife Conservation Act Authorization to Keep Wildlife (1094480/Local Ref. No. AU20191892, 1095663/Local Ref. No. AU2020-2321, 1095690/Local Ref. No. AU2020-2325, 1097716/Local Ref. No. AU2021-00077, and 1100047/Local Ref. No. AU2022-00100), Wildlife Scientific Collector’s Authorizations (1077386, 1080550, 1083631, 1086727, 1090044, 1089107/Local Ref. No. AU2018-0533, 1092254/Local Ref. No. 053, 1092095/Local Ref. No. AU2019-1299, 1095664/Local Ref. No. AU2020-2308, 1095693/Local Ref. No. PS2020-0839, 1095690/Local Ref. No. AU2020-2325, 1096136/Local Ref. No. AU2020-2501, 1096490/Local Ref. No. AU2020-2550, 1097514/Local Ref. No. PS2021-1039, 1097470/Local Ref. No. AU2021-00104, 1100042/Local Ref. No. PS2022FW0014, and 1100043/Local Ref. No. AU2022-00099), Parks Canada Research and Collection (RNUP-2020-35017), and federal Species At Risk Act permits (RNUP-2020-35017).

Observed mortalities prior to mass-mortality event.—

From 2014 to 2019, while conducting radio-telemetry, nesting, live-trapping, and road surveys (typically from mid-April to early September; Table S1; see Data Accessibility), we observed 42 Blanding’s (41 were headstarted juveniles) and six Painted Turtle carcasses at site A, 26 Painted and 14 Snapping Turtle carcasses at site B, and two naturally occurring Blanding’s, 18 Painted, five Snapping Turtle carcasses, and one turtle carcass of an unknown species at site C. An additional 11 Painted Turtle carcasses were observed in areas outside sites A–C. During this period, roadkill resulted in 76 mortalities, predation resulted in 34 mortalities, winterkill resulted in six mortalities, and unknown causes resulted in nine mortalities, for a total of 125 mortalities (Fig. 2).

View Figure

• Download as Powerpoint
• Download Figure

Mass-mortality event.—

While conducting only live-trapping and radio-telemetry surveys in 2020 (Table S1; see Data Accessibility), 48 juvenile headstarted Blanding’s and 57 Painted Turtle carcasses were observed at site A, none of which were associated with roads or showed evidence of vehicle strikes. Carcasses of juvenile headstarted turtles were found mostly on land in Common Reed or grass patches. Of the 57 Painted Turtle mortalities documented in 2020, 43 were adult female Painted Turtles suspected to be depredated while nesting on land, while 14 were of unknown sex. The number of turtle mortalities observed for Blanding’s and Painted Turtles in 2020 was 105, while the number of observed mortalities in the previous six years combined was 125 (Fig. 2). Various degrees of damage to hard and soft tissues were observed in Painted and Blanding’s Turtles. Of the 83 turtles that were suspected to be depredated, field notes indicated 36 were decapitated, seven were missing head and at least one limb, four were decomposed, and 36 turtles lacked specific details of carcass condition. We could not determine the cause of death for some carcasses (n = 22) because of advanced decomposition, resulting in recovery of only shell fragments, intact shells without flesh, or carcasses in the later stages of autolysis. We acknowledge that disease is a potential cause of mortality and cannot be ruled out, especially where carcasses showed no obvious signs of predation; however, no evidence of disease was detected during necropsies (n = 25 carcasses).

Post-mass-mortality event.—

While conducting a combination of three to five survey types in 2021 and 2022 (radio-telemetry, nesting, road, live-trapping, water-chemistry sampling; Table S1; see Data Accessibility), we observed four headstarted Blanding’s, 31 Painted, and two Snapping Turtle carcasses at site A; 28 headstarted Blanding’s, seven Painted, and four Snapping Turtle carcasses at site B; and three Painted and two Snapping Turtle carcasses and one carcass of an unknown turtle species at site C. One Painted Turtle carcass was also found outside of these three study sites.

Mortality trends and demography by site.—

Overall, across all years, the majority of the turtle mortalities occurred at site A (n = 190), followed by site B (n = 79), and finally site C (n = 32). In sites A and B, headstarted Blanding’s and Painted Turtle mortalities were commonly documented (n = 94 of each), whereas in site C, where no headstart Blanding’s Turtles have previously been released, only Painted Turtle mortalities (n = 21) were documented (Table 1). In sites A and B, documented Blanding’s Turtle mortalities consisted of headstarted juveniles (except for one male). Among Painted Turtle carcasses, adult females were common in site A (n = 48), and juveniles were common in site B (n = 14). Snapping Turtle carcasses were found in relatively lower abundance across all three sites (n = 27), and most Snapping Turtle carcasses were juveniles (n = 23). Twelve additional carcasses were documented in areas not routinely surveyed outside of sites A–C (e.g., opportunistically found when surveyors traveled between sites and/or when conducting other monitoring activities). Of these carcasses, an adult male and a juvenile Painted Turtle were identified, while the age classes of the rest were unknown. At sites A–C, the sex or age class was determined for only 225 of 301 carcasses due to the condition of the carcass or because of missing data (Table 1).

Observed mortalities prior to mass-mortality event.—

Turtle mortalities have been observed at three sites in the RNUP since monitoring began for the headstarting program in 2014. In earlier years (2014–2016), observed mortalities were relatively low, and the carcasses were primarily of headstarted Blanding’s Turtles. Survey efforts increased in subsequent years (2017 onward), with a concomitant increase in observed mortalities. However, increased survey effort is likely just one factor that contributed to the increase in observed mortalities (see next section). A greater number of Painted Turtle carcasses were found than those of Blanding’s Turtle. This observation is not surprising given the higher abundance of Painted Turtles compared to Blanding’s Turtles across all surveyed sites. Painted Turtle mortalities were commonly associated with either predation or vehicular collisions, whereas the majority of headstarted Blanding’s Turtle mortalities were due only to predation. As the headstarted Blanding’s Turtles reach maturity, we expect individuals to exhibit larger home-range sizes (Edge et al., 2010), as observed in adults at our study sites (Toronto Zoo, unpubl. data). Given the highly fragmented nature of the RNUP and the extensive road network (Frost-Wicks, 2019), such movements will likely increase risk of road mortality for turtles. The major causes for pre-MME deaths (e.g., road mortality and predation) are typical of turtle populations inhabiting urban habitats (Marchand et al., 2002; Steen and Gibbs, 2004; Urbanek et al., 2016; Piczak et al., 2019). To a lesser degree, winterkill has also been documented in both urban and non-urban habitats (Christiansen and Bickham, 1989; Seburn et al., 2021).

Mass-mortality event.—

In 2020, a large number of headstarted Blanding’s and female Painted Turtles were depredated at site A. Despite relatively lower survey duration and survey area (i.e., site A only had radio-telemetry and trapping surveys for two months, site B was only surveyed using trapping for a week, and site C was not surveyed; Table S1; see Data Accessibility), mortalities were markedly higher in 2020 compared to previous years (2014–2019). Thus, we conclude the observed trends are not simply a result of changes in survey efforts.

A preliminary study based on a four-year mark–recapture survey at the three sites indicated a population size estimate of 306 (CI = 170–600) female and 142 (CI = 76–291) male Painted Turtles at site A in 2020 (Wijewardena et al., in press). As such, the observed mortalities resulted in ∼13% population loss at site A. Similarly, in 2020, we estimated the population size of headstarted Blanding’s Turtles to be 183 individuals (Wijewardena et al., 2023); thus, a loss of 48 individuals equates to a ∼26% population decrease in Blanding’s Turtles at site A. Different age classes of turtles are preyed upon by a variety mammals (Congdon et al., 1987; Cypher et al., 2018), such as American Mink (Neogale vison), Raccoon (Procyon lotor), Coyote (Canis latrans), and birds (Baxter-Gilbert et al., 2013), such as American Crow (Corvus brachyrhynchos). There are unique cases where one or a few individual predators caused MMEs of freshwater turtles (Brooks et al., 1991; Karson et al., 2018; Keevil et al., 2018; Jones et al., 2021; Bulté et al., 2024), and we suspect this to be the case in the RNUP as well. We suspect that Raccoons (Seigel, 1980) or mustelids (Stacy et al., 2014) were the likely cause of the mortality increase observed in 2020 due to similar damage to carcasses, which could have resulted from the dexterity of these predators. Coyote, Raccoon, River Otter (Lontra canadensis), and Mink have been observed at our study sites (Parks Canada, unpubl. data), and historical surveys (2005–2006) in the study areas indicated that Raccoons and other wildlife were fed by park visitors and residents in the area (Toronto Zoo, unpubl. data). We were unable to determine exactly when the mortalities at site A occurred because the decomposition rates of depredated turtles have not been examined at our study site. A study in Misery Bay Provincial Park in Ontario, Canada, reported that turtle soft tissue takes about three weeks to decompose under normal weather conditions, and decomposition rates can vary depending on whether the carcass was on land or submerged in water (Gasbarrini et al., 2021). We are unaware whether similar mortalities occurred in other sites, but a week-long survey at site B in 2020 and surveys at site C in subsequent years did not produce similar numbers of depredated turtles as we observed at site A. Drought conditions observed during 2020 and 2021 may have played a role in observed high mortalities (Table S2; see Data Accessibility). Droughts and intentional water drawdowns are associated with higher predation of turtles elsewhere, in part because turtles become concentrated into high densities in the limited remaining aquatic habitats, increasing their vulnerability (Gibbons et al., 1983; Lindeman and Rabe, 1990; Hall and Cuthbert, 2000; Anthonysamy et al., 2013; Gasbarrini et al., 2021; Beard and Powell, 2023). Freshwater turtles may also die due to diseases (Fey et al., 2015; Agha et al., 2017; Allender et al., 2018). In the subset of carcasses that were necropsied in our study, disease was not identified as a potential cause of mortality by CWHC. However, not all carcasses were fresh and, as a result, we cannot rule out disease alone or in combination with other factors.

Post-mass-mortality event trends.—

Blanding’s Turtle mortality rates returned to pre-MME (2014–2019) levels post-2020, whereas Painted Turtle mortalities remained elevated in 2021, returning to pre-MME levels in 2022. Suspected predation was the most common cause of mortality for Blanding’s and Painted Turtles; however, it is plausible that some deaths occurred in 2020 but were discovered in subsequent years based on carcass deterioration. Road surveys in the area of these sites also typically resulted in more observations of Painted Turtle mortalities than other species (Leermakers, 2020). These observations are typical for our study area given the relative abundance of each species, i.e., Painted Turtles have the highest abundance, followed by Blanding’s Turtles (due to headstarting), and Snapping Turtles.

Mortality trends and demography by site.—

Blanding’s Turtle mortalities were observed at sites A and B because they are release sites for the headstarted population that are routinely surveyed. Given that a greater number of headstarted turtles have been released at site A compared to site B, it is not surprising that the majority of Blanding’s Turtle mortalities were observed at site A. In contrast, at site C, Painted Turtle mortalities were common because of their higher abundance (Toronto Zoo, unpubl. data). Despite the proximity of the surveyed sites to each other (within <2 km), we have not observed juvenile headstarted Blanding’s Turtles at site C, indicating that headstarted turtles remain in the release sites. We have occasionally observed Snapping or naturally occurring Blanding’s Turtles at site C and believe both species are likely transients that occupy permanent wetlands elsewhere. Roadkill is a major threat to turtles at sites B and C because the wetlands are close to roads. Since female turtles use road shoulders to nest (Aresco, 2005), high female mortality is often associated with roads (Steen and Gibbs, 2004). However, juveniles are also affected by roads and may be killed more frequently than adults (Keevil et al., 2023). We observed several (n = 36) juvenile deaths during road surveys between 2017 and 2022. Ultimately, roads pose a threat to both sexes and all life stages of turtles (Carstairs et al., 2018; Keevil et al., 2023).

Conservation implications.—

Achieving and maintaining high survival is a challenge for turtle populations in fragmented habitats, such as the case in the RNUP. This is true for Blanding’s Turtles, which have been the target of conservation management for over ten years. Based on our findings from 2014 to 2022, at least 124 of the 374 (33%) headstarted Blanding’s Turtles released have been lost to predation, winter-kill, wildlife–vehicular collisions, and unknown causes. A population viability analysis of headstarted Blanding’s Turtles that incorporates the effect of catastrophes is currently being undertaken by the Toronto Zoo to determine management implications. We have yet to determine the extent of the long-term consequences that will follow the acute mortality of female Painted Turtles in 2020. Demographic models highlight the need for high adult (Enneson and Litzgus, 2008; Shoemaker et al., 2013) and juvenile survival for population stability in turtles (Congdon et al., 1993). If observed mortality trends continue, the Painted and Blanding’s Turtle community at the three sites is likely at risk of extirpation; as a result, they will require longer population recovery times and more intense human-mediated conservation interventions. Conservation initiatives focusing on endangered species already face multiple challenges in terms of limited funding, lack of long-term monitoring, and controversy regarding the most effective method to recover SAR (Dodd and Seigel, 1991; Heppell et al., 1996; Martin et al., 2018). MMEs can result in unanticipated strain on limited resources, which will delay recovery of populations with slow life histories. Additionally, due to stochasticity in natural ecosystems, catastrophes such as MMEs are perhaps the least understood in terms of frequency of occurrence and magnitude of impact. Our study is a cautionary tale for headstarting projects and highlights how MMEs could limit population recovery despite ongoing conservation efforts. Similarly, MMEs can impact naturally occurring populations (Keevil et al., 2018) that are perceived as robust given their relative abundance, as observed in our study for Painted Turtles. Continued monitoring of an endangered species provided us with a unique opportunity to monitor a Painted Turtle population that typically receives less conservation focus. Our study further highlights the importance of monitoring common species that may undergo catastrophes and face extirpation before we recognize the urgency of their conservation needs.

We conclude that long-term monitoring is necessary to understand population trends, especially in species with delayed sexual maturity and long generation times. MMEs, paired with ongoing challenges to survival (e.g., nest predation, road mortality), pose serious risks to the persistence of species at risk populations, especially in urban areas where multiple threats are simultaneously exacerbated (e.g., habitat fragmentation, degradation). Freshwater turtle populations, even in protected areas, can suffer from sudden, chronic, or synergistic sources of mortality due to natural or anthropogenic threats (Browne and Hecnar, 2007; Jones et al., 2021). Incorporation of catastrophes into projection tools for naturally occurring and augmented populations, such as population viability analyses, is needed to better understand the combined factors that affect individual- and population-level mortality, and inversely, survival. Ultimately, we strongly recommend that adaptive management (e.g., additional release sites) and multi-species approaches (e.g., subsidized predator control, nest protection, road mortality mitigation) be taken when pursuing conservation action specific to turtles.