Human activities are increasing the rate of species extinction to more than 1000 times the natural level (Ceballos et al. 2010) and chelonians, composed of turtles, terrapins, and tortoises, are a group of species experiencing dramatic declines. Of the roughly 360 chelonian species worldwide, more than half are threatened with extinction (Stanford et al. 2020). Turtles are particularly vulnerable to extirpation because of their requirement for wetland habitats and life history characteristics, such as slow growth rates, delayed sexual maturity, and extreme longevity (Stanford et al. 2020).

Two chelonian species gaining attention because of recent declines are the northwestern pond turtle (NWPT; Actinemys marmorata) and southwestern pond turtle (SWPT; Actinemys pallida), California's only native freshwater turtles. Both species are cryptic, semiaquatic turtles found along the western region of the United States. These species were long classified as a single species. However, recent analysis using a combination of mitochondrial DNA, nuclear DNA, and single-nucleotide polymorphisms found the turtles to be 2 distinct species, with A. marmorata occurring north of the San Francisco Bay to Puget Sound, Washington, as well as into the Central Valley of California, while A. pallida extends from the San Francisco Bay south along the coast to Baja California (Spinks et al. 2014). Once considered widespread throughout its range, populations of both Actinemys species have dropped considerably. The NWPT is listed as Endangered in Washington State (Washington Department of Fish and Wildlife 2022). In Oregon, the NWPT is designated as “Sensitive-Critical” (Oregon Department of Fish and Wildlife 2021), while in California, both the NWPT and SWPT are listed as Species of Special Concern (California Department of Fish and Wildlife 2016).

Although highly adapted to living in a variety of aquatic environments—including permanent and seasonal water bodies, natural and human-altered landscape, human-created water bodies, and lentic or lotic systems—both species are threatened by habitat loss due to habitat alteration, fragmentation, and destruction (Cook and Martini-Lamb 2004; Bury and Germano 2008). Predation by largemouth bass (Micropterus salmoides) and bullfrogs (Lithobates catesbeianus), particularly of turtles at the juvenile stages, are also of concern (Manzo et al. 2021). The SWPT is believed to be more vulnerable to extirpation than the NWPT because of their smaller range, smaller population size, and potentially severe impacts from natural disasters such as droughts, wildfires, and floods (Manzo et al. 2021). These factors all make protecting the SWPT a high priority.

Human-created habitats, such as livestock ponds (Fig. 1), can have the potential to support declining native species such as the SWPT (Fig. 2). Livestock ponds are found throughout the western United States, often in dry regions with seasonal rainfall. They are built along low-order tributaries, which drain from nearby hill slopes to support grazing livestock and farming activities (Florsheim et al. 2013). These ponds may serve as temporary or permanent habitat for a variety of wildlife species. Livestock ponds are considered high-value conservation areas because they contribute significantly to regional biodiversity and help support endangered species (Declerck et al. 2006; Ruggiero et al. 2008; Usio et al. 2013). Livestock ponds provide alternative habitat when natural habitat has disappeared (Casas et al. 2012) and they can enhance preexisting habitat. Further, Knutson et al. (2004) concluded that if small livestock ponds were managed properly, they could help sustain amphibian populations in areas where natural wetlands are uncommon.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Both species of pond turtles have been found in abundance in livestock ponds (Bury and Germano 2008). Although livestock ponds may be important features in the conservation and management of SWPT populations, little is known regarding the extent to which these ponds may provide long-term habitat for SWPTs. Santa Clara County, California, is an ideal location to study this question. In this county, urbanization along the valley floor, known worldwide as “Silicon Valley,” has greatly altered the natural landscape, and has likely negatively impacted the population of SWPTs residing there. However, the landscape along the Diablo Range of eastern Santa Clara County has been ranch land grazed by cattle for many decades and this land use has helped to preserve many natural features. In addition, cattle ranching activities have led to the creation and long-term preservation of hundreds of livestock ponds in the Diablo Range that SWPTs can potentially inhabit.

The goal of this study was to assess the value of livestock ponds as conservation sites for the protection of SWPTs by 1) determining the relative occupancy of livestock ponds by SWPTs in the Diablo Range of eastern Santa Clara County, California, and 2) comparing how habitat features differ between occupied and nonoccupied ponds in order to characterize the features that attract SWPTs. Ultimately, we conducted this study to provide managers with information that can help them preserve and enhance SWPT populations.

Methods. — The Diablo Range in Santa Clara County, California, reaches an elevation of 1300 m on top of Mount Hamilton (lat 37°20′30.2″N, long 121°38′34.2″W). The area has a Mediterranean-type climate characterized by warm and dry summers followed by cool and wet winters. Annual precipitation averages 60 cm and snow may accumulate under certain conditions. Five properties (study sites) were included in this study. These sites and their managing entities were Rancho Canada de Pala (RCDP, managed by Santa Clara Valley Water District), Blue Oak Ranch Reserve (BORR, University of California Natural Reserve System), Joseph Grant County Park (JGCP, Santa Clara County Parks), San Felipe Ranch (SFR, Hewitt Packard Family), and Henry W. Coe State Park (HCSP, California State Parks) (Fig. 3). All properties experienced cattle grazing at one point, but at the time of this study grazing was active only in JGCP and SFR. The study sites supported a variety of different habitats including oak woodland, chaparral, riparian forests, and open grassland, and included many perennial and seasonal streams and water bodies.

View Figure

• Download as Powerpoint
• Download Figure

A search of previous studies, review of the California Natural Diversity Database (CNDDB 2017), and conversations with property managers indicated that SWPTs were found in ponds at all 5 study sites. At RCDP, 2 ponds and 1 lake had documented SWPT occurrences (N. Jassal, pers. comm., May 2017) while BORR had 5 ponds and 1 lake with SWPTs (Blue Oak Ranch Reserve 2018). Records of SWPT occurrence at JGCP were incomplete, but 3 ponds and 1 lake were documented in the CNDDB (2017). Only 2 ponds and 1 lake were known to have had SWPTs at SFR (H. Wittkopp, pers. comm., June 2017), plus 1 record from a different lake on the property (CNDDB 2017).

To determine the extent to which SWPTs occupied livestock ponds in these areas, we visited ponds that had been previously surveyed by other researchers and ponds that had not been visited. One turtle survey was conducted at each pond in order to visit as many ponds as possible. Surveys were conducted in teams of 2 from 12 March to 17 September 2017, when temperatures are rising, and turtles are more likely to be active and spotted. Each turtle survey consisted of a 15-min visual encounter survey between 0900 and 1700 hrs using 10 × 50 binoculars within 5 m of the pond edge. After each survey, 1 person conducted a perimeter walk to look for signs of turtles (i.e., eggshells or dead carcasses).

Data on habitat features were collected on a subset of all ponds surveyed for this study that were large enough (> 10 m in diameter) and deep enough (> 0.33 m in depth) to be likely to support SWPTs. In addition, this subset of ponds was visually surveyed 3 times within a 1-wk period, with the intention of confirming presence/absence of SWPTs more accurately. Water bodies larger than 100 m were considered lakes and were excluded from habitat data collection as they did not represent the type of water body that would be created for a typical livestock pond. Heavy rainfall and landslides limited access to some ponds; thus, only ponds that were safely accessible were surveyed.

Data on 6 habitat features were collected on ponds occupied by SWPTs and on nonoccupied ponds. Habitat features included pond surface area, maximum pond depth, elevation, number of basking structure(s), presence of bullfrogs, and amount of tree/shrub cover. Surface area was estimated using the World Imagery basemap layer in ArcMap (Version 10.6). Pond depth was estimated by averaging the results of 3 casts of a sonar fishing bobber (Deeper Pro) into the center of the pond. Tree and shrub cover were also estimated using the World Imagery Basemap layer. A 100-m buffer zone was created in ArcMap around each pond and the amount of cover was divided by the total buffer zone area to get percentage of tree/shrub cover, combined. Tree/shrub cover was then grouped into 3 categories: low (0%–10%), medium (10%–25%), and high (> 25%) tree/shrub cover.

Statistical analysis was conducted using SPSS 24. A ttest was used to compare habitat features (surface area, maximum water depth, elevation, and number of basking structures) between occupied and unoccupied ponds. Each variable was tested for normality and, when necessary, the data were transformed by ln(x + 1). Pearson correlations were used to test for associations among the variables. The t-test α was set at 0.10 due to the small sample size of ponds with SWPTs. Prior to running the t-test, a Levene's F-test was used to test for equality of variance. No variables were significant except for ln(surface area) (p < 0.05). A power curve analysis was conducted using SYSTAT 13 (Systat Software Inc. 2017) to determine what an appropriate sampling size would be to confidently detect a difference. The habitat variable with the lowest ttest p-value was used for the power curve analysis. Descriptive statistics were used to evaluate whether the presence of bullfrogs and tree/shrub cover appeared to affect SWPT occupancy of ponds.

Results. — We surveyed a total of 78 ponds for SWPTs at least once (Table 1). Of these, 55 had been previously surveyed by other researchers (CNDDB 2017) and/or property managers who found SWPTs at 16 of the ponds and no turtles at the remaining 39 ponds. Of the 78 ponds included in this study, 23 had not previously been surveyed by other researchers; these ponds provide a more random sampling compared with the previously visited ponds. We conducted 2 additional visual encounter surveys and collected habitat data at a subset of 31 of the 78 ponds. All 16 previously surveyed ponds with SWPT observations were included in the habitat data collection.

Table 1. Previously documented southwestern pond turtles (SWPT) occurrence in livestock ponds vs. the occurrence documented by this study.

View Fullscreen

Although visual encounter surveys were conducted 3 times at the 31 subset ponds to try to ensure adequate observation, when turtles were detected, they were always found on the first survey, as well as in subsequent 2 surveys. Thus, 1 survey seemed to have been adequate to ensure detections. During our surveys, we detected SWPTs at 12 of 16 previously surveyed ponds (75%) where turtles were recorded. Of the 39 previously surveyed ponds where turtles had not been found, we detected turtles at 1 pond—a pond in which the dam was repaired in 2012. There were no turtles detected at the 23 ponds included in this study that had no previous records of being surveyed (Table 1). In total, we found turtles at 13 of 78 (16.7%) ponds surveyed, including at 1 pond that was not previously searched or for which turtles had not been documented (1 of 62 or 2%).

There were moderate correlations between ln(surface area + 1) and ln(depth + 1) (r[29] = 0.63, p < 0.01) as well as ln(depth + 1) and basking structure (r[29] = 0.60, p < 0.01). We considered all these parameters to be important and conducted analyses on each of them. Surface area, water depth, and elevation did not differ between occupied and nonoccupied sites (Table 2). The average number of basking structures in occupied sites was twice that of the number in ponds without SWPTs (1.38 vs. 0.67 structures per pond) and this difference was significant at the 0.10 level (Table 2). The power analysis for basking structures indicated that a sample size of 39 for each dependent variable group was needed for the power to be 0.80 with an α of 0.05. Our actual smaller sample size for basking structures was much smaller and the observed power was low (0.41), indicating a low likelihood of detecting a difference; despite this, a difference was detected at the 0.10 level. There was little obvious difference in the presence or absence of bullfrogs between occupied and nonoccupied ponds (Fig. 4). However, the presence of tree/shrub cover did seem to be a factor in pond occupancy by SWPTs; specifically, only ponds with medium and high tree/shrub cover were occupied by pond turtles (Fig. 4).

Table 2. Comparison of habitat features between ponds occupied and nonoccupied by southwestern pond turtles (n = 31). Bold p-values are significant.

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

Discussion. — The results of this study showed that livestock ponds can provide long-term habitat for SWPTs, as we found that 12 of the 16 ponds documented to have SWPTs within the last 15 yrs still supported SWPTs. But, overall, the livestock pond occupancy rate was relatively low in the Diablo Range of Santa Clara County. While we found approximately 17% of the ponds we surveyed were occupied, this result was probably high since our sample was biased toward ponds previously surveyed and was not random. In fact, our random sample of 23 ponds that were not previously surveyed revealed no occupancy by SWPT. Similarly, a complete survey of ponds at the SFR site also showed low SWPT occupancy of livestock ponds, with only 4 records of SWPT occupied ponds and lakes out of 124 surveyed (4.0%) (H. Wittkopp, pers. comm., June 2017).

Other regions may potentially support higher occupancy rates in livestock ponds. For example, 26 of the 50 randomly surveyed ponds in the Umpqua, North Umpqua, and South Umpqua watersheds of western Oregon were occupied (Horn and Gervais 2018). However, these areas support the NWPT and Manzo et al. (2021) has found SWPT populations to be generally smaller than those of the NWPT. These smaller populations make the SWPT especially vulnerable to extirpation (Manzo et al. 2021). Given the vulnerability of SWPT populations, our results indicate that preserving ponds known to be occupied is especially important.

Most of the habitat features we assessed did not differ between occupied and nonoccupied ponds, except potentially for basking structures and tree/shrub cover, which appeared to be important habitat features. These findings are consistent with other studies, such as one conducted by Reese and Welsh (1998) in the Trinity River in Northern California in which NWPTs were associated with basking structures. Although this association may be less critical in areas where higher water temperature is available (Reese and Welsh 1998), such as livestock ponds which are stagnant and warmed by ambient air, we still found basking structures to be important features. SWPTs were found only at ponds with moderate to high tree/shrub cover, results that are consistent with radiotelemetry studies of NWPT and SWPT terrestrial movements in which turtles are often found hidden under various substrates and vegetation (Pilliod et al. 2013; Zaragoza et al. 2015). These findings all highlight the importance of having available tree/shrub cover nearby to support SWPTs at livestock ponds. A fuller investigation is needed of habitat features likely to be important to SWPTs, such as distance to source populations, seasonality of the ponds, nutrient levels, amount of aquatic vegetation growth, and level of grazing activity.

Although SWPT occupancy of livestock ponds in the Diablo Range of eastern Santa Clara County was relatively low, these ponds—especially occupied ponds—provide valuable habitat and can increase SWPT abundance in the area. Whereas this study occurred in the range of the SWPT, we believe our results also apply to NWPTs because of the similarity in threats and in habitat requirements of these 2 species (Manzo et al. 2021). We recommend monitoring and protecting existing turtle-occupied ponds, as well as undertaking habitat enhancement of nonoccupied and previously occupied ponds, especially by providing basking structures and tree/shrub cover, in an attempt to increase the value of livestock ponds for SWPTs in rangeland habitats.