Mojave desert tortoises (Gopherus agassizii) inhabit the southwestern United States in the Mojave and Sonoran deserts from southern California through southern Nevada and northwestern Arizona, and into the southwestern tip of Utah (Fig. 1). The species has been protected under the Endangered Species Act since 1989 (US Fish and Wildlife Service 1994), and populations are in decline throughout most of the range, including in the Red Cliffs Desert Reserve in St. George, Utah (Allison and McLuckie 2018). However, the biogeography and origin of the species in the northeastern extreme of their distribution, east of Red Cliffs, has long been contested (reviewed in Bury et al. 1994) because of the area's suboptimal habitat, distance to optimal habitat, and known movements of tortoises from elsewhere in the desert. Southwestern Utah is where the Mojave Desert, Great Basin, and Colorado Plateau meet, representing the northeastern edge of preferred tortoise habitat. As such, the extent of their natural limit into Utah is of interest to tortoise biologists.

View Figure

• Download as Powerpoint
• Download Figure

The majority of desert tortoise habitat is defined by creosote–white bursage (Larrea tridentata–Ambrosia dumosa) communities, including additional perennial plants such as Ephedra spp., Opuntia cacti, and occasional Joshua tree and Mojave yucca (Yucca spp.), among others (McLuckie et al. 2000; Nussear et al. 2009, 2012). Although most desert tortoise habitat is found below 1000 m above sea level (masl), they do occupy higher regions where plant communities shift from creosote–bursage to blackbrush-dominated (Coleogyne ramosissima) and sagebrush-dominated (Artemisia filifolia) areas (Nussear et al. 2012). Indeed, Mojave desert tortoises often occupy habitats along the edges of their distribution that differ from those in the core of their range; the unique habitat characteristics and accompanying changes in tortoise ecology in the northeastern portion of the distribution around St. George, Utah, justified the area to be designated as its own recovery unit in the species recovery plan (US Fish and Wildlife Service 2011). The St. George area, including Red Cliffs Desert Reserve, is on the edge of typical tortoise habitat, with higher elevations than much of the desert (approx. 950–1200 masl), and more blackbrush and less creosote than in the core range (Nussear et al. 2012). Contemporary genetic analyses demonstrate that tortoises from the Red Cliffs Desert Reserve cluster with those from neighboring Nevada (Hagerty and Tracy 2010). Approximately 25 km farther east, in and near Zion National Park (hereafter, Zion) near Springdale, Utah, tortoises inhabit even higher elevations (> 1300 masl), where they experience different temperature regimes (e.g., colder winters), and the area predominantly supports blackbrush and pinyon pine–juniper (Pinus monophyla, Juniperus osteosperma) plant communities. Whether or not tortoises entered this suboptimal habitat naturally is thus ecologically interesting and could help us understand how the tortoise distribution may change in the future.

Even though Red Cliffs is generally accepted as within the natural tortoise range, the origins, history, and genetic identity of tortoises farther east, in and around Zion, are still controversial. Historically, there were many opportunities and reports of prehistoric and modern humans moving various desert tortoises. People have often kept tortoises as pets, bringing them when they move around the desert Southwest. In recent studies, approximately 40% of captive tortoises sampled in Arizona were identified as Mojave desert tortoises and not as Sonoran desert tortoises native to that area, and a captive tortoise in California was identified as a Sonoran desert tortoise (Edwards et al. 2010; Edwards and Berry 2013). Pet owners may also release unwanted tortoises into the desert, and Woodbury and Hardy (1948) found nearly 20 free-ranging tortoises with signs of previous captivity, such as holes in marginal scutes and initials carved into scutes, during studies from the 1930s to 1940s in Beaver Dam Slope, Utah. Facilitating these tortoises being kept as pets and moved around the desert, at one point, desert tortoises were purchased and given away as souvenirs along old highway 91, which passed from Nevada into Utah, through St. George (Beck 2005). Although some have opined that tortoises in Zion were translocated from elsewhere within the range, others have suggested that individuals from different environments would not be able to withstand the harsher climate of this area of Utah (Bury et al. 1994). Although many tortoise movements were likely aided by people, tortoises naturally inhabit the area around St. George, and their natural occurrence even farther east toward Zion is plausible. Some tortoises move long distances, and long-distance movements are more likely when tortoises find themselves in unfamiliar or poor habitat (e.g., Edwards et al. 2004; Nussear et al. 2012).

Our objective in this study was to elucidate the genetic identity of Zion's desert tortoises using microsatellite data to better understand the natural limits of the Mojave desert tortoise distribution. We use 2 genetic analyses to determine the likely origin of 9 tortoises from Zion. First, we identify the genetic clustering of these individuals. Mojave desert tortoises cluster into 7 genetic groups (see fig. 1 in Hagerty and Tracy 2010), and if tortoises in Zion originated in the nearby areas, they would be assigned to the Virgin River cluster. Next, we use an analysis that assigns individuals to local population segments (e.g., Zion or Red Cliffs). In this analysis, we expect tortoises originating from the vicinity to assign to Zion or Red Cliffs local populations, versus populations farther away in the distribution.

METHODS

Blood was collected from the subcarapacial sinus (Hernandez-Divers et al. 2002) of 9 adult desert tortoises (240–320-mm midline carapace length) in Springdale, Utah, near Zion National Park (Fig. 1), with whole blood samples dried onto filter paper dots. We extracted total genomic DNA using the Qiagen DNeasy Blood and Tissue Kit (Qiagen Inc., Valencia, CA) protocol for dried blood. DNA was eluted in PCR-grade water, quantified using a Labsystems Fluoroskan Ascent fluorometer, and diluted to between 5 and 10 ng/µL for microsatellite amplification. We amplified 17 microsatellite loci (GOA1, GOA2, GOA3, GOA4, GOA6, GOA8, GOA9, GOA11, GOA12, GOA13, GOA14, GOA22, GOA23, GOAG3, GOAG7, GP15, GP30; Edwards et al. 2003; Schwartz et al. 2003; Hagerty et al. 2008) using polymerase chain reaction (PCR) protocols described by Hagerty and Tracy (2010). All amplified segments were analyzed for fragment size using an ABI 3730 DNA sequencer at the Nevada Genomics Center (http://www.ag.unr.edu/genomics/), and we scored alleles using GeneMapper 5.0 (Applied Biosystems, Inc, Foster City, CA).

Data Analyses. — To assign the 9 individual tortoises from Zion to genotype clusters, we added them to a reference data set that included 585 desert tortoises from a previous genetic analysis (Hagerty and Tracy 2010) for a total of 594 samples. Up to 30 sampled individuals were included from each of the 25 original sampling sites of Hagerty and Tracy (2010) to prevent overrepresentation from one region.

We employed 2 analyses to compare the microsatellite data from Zion tortoises with those from elsewhere in the species distribution, with a focus on the individual assignments of Zion tortoises. We used the Bayesian clustering technique in STRUCTURE (ver. 2.34, Pritchard et al. 2000; Falush et al. 2003) to infer the number of genotype clusters in the data set and to assign individuals to one of those groups. Although our data set includes tortoises from 26 local populations or sampling sites, STRUCTURE groups those local populations into fewer genetic clusters. In STRUCTURE, we used the admixture model and allowed for correlated gene frequencies. We performed 3 runs at each value of the fixed parameter K (number of clusters; K = 1 to K = 12). We specified a burn-in of 750,000 followed by 750,000 MCMC replicates. All other parameters were set to the default values (Pritchard et al. 2003).

Structure results were summarized using Structure Harvester (Earl and vonHoldt 2012). For each value of K, we evaluated the mean estimated natural log of the probability of the data, or ln P(D). Once the true K is reached, the phenomenon of experiencing a plateau in the likelihoods is common; therefore, we also evaluated the number of clusters using the Evanno method (ΔK), which can provide a more robust measure of K (Evanno et al. 2005). To assign individuals to a genotype cluster, we used values of q, the proportion of an individual's sampled genome characteristic of each cluster, with a 50% threshold for assignment.

For comparison, we also used a frequency-based assignment test (Paetkau et al. 1995) that assigned individuals to 1 of the 26 sampling locations using GENALEX (version 6.5; Peakall and Smouse 2006, 2012). To assess genetic divergence in the Zion tortoises compared with other sampling locations, we calculated pairwise F_ST and G′_ST(Nei) using GENALEX 6.5. Both F_ST and G′_ST(Nei) are metrics of genetic diversity based on differences in heterozygosity within and among populations, providing information on genetic distance among tortoise sampling sites and increasing in value as genetic drift increases. G′_ST(Nei) measures include corrections applied for multiple loci and small population size (Peakall and Smouse 2012).

We were only able to sample 9 tortoises from Zion, so we compared the number of GENALEX assignments found in our Zion samples with the number of assignment groups from a site represented by more tortoises. We compared results from Zion with those from the nearby Red Cliffs site (n = 30) by randomly subsampling GENALEX results from 9 Red Cliffs tortoises over 1000 iterations.

RESULTS

The most likely number of genotype clusters resulting from STRUCTURE was 7. These results closely match those of Hagerty and Tracy (2010); however, local population membership to the 7 clusters differed slightly based on the addition of a new sampling location and inclusion of fewer samples in the analysis to reduce overrepresentation of any one sampling site. The Zion site was most closely assigned to the Virgin River genetic cluster (43% assignment), which includes the nearby locations of Red Cliffs, Beaver Dam, and Mormon Mesa (Fig. 1).

Of the 9 Zion tortoises, 3 were assigned to the Virgin River cluster, 1 tortoise was assigned to the Muddy Mountains cluster (northeast of Las Vegas in Nevada and Arizona), and 2 tortoises were assigned to the Colorado Desert cluster (mostly California into the southern tip of Nevada; the Colorado Desert is a subdivision of the Sonoran Desert). The remaining 3 tortoises had split assignment (below 50%), although their most likely clusters were those already represented in the other Zion tortoises (Table 1). In GENALEX assignments, the 9 Zion tortoises assigned to 5 local populations, 3 of which are in the Virgin River genetic cluster (Zion, Red Cliffs, and Beaver Dam Slope), 1 in the Muddy Mountains genetic cluster (NE Las Vegas Valley), and 1 in the Colorado Desert cluster (Chuckwalla Valley). Three tortoises sampled outside Zion, one each from Red Cliffs, Northeast Las Vegas Valley, and Southwest Las Vegas Valley, were assigned to the Zion population in GENALEX analyses.

Table 1. STRUCTURE and GENALEX assignment of 9 tortoises (ID) sampled from Zion National Park. Assignment agreement is between the two analyses. Geographic agreement indicates whether the tortoise is likely native (y) or introduced (n) to the region, or of unknown origin (unk), based on genetic assignment. STRUCTURE assignments below 50% agreement are in parentheses. The STRUCTURE assignment of Virgin River includes locations in southwestern Utah, Muddy Mountains includes areas of Nevada and Arizona northeast of Las Vegas, and Colorado includes both Eastern and Northern Colorado Desert locations in California and Nevada.

View Fullscreen

Structure and GENALEX assignments for these 9 tortoises were in agreement for 5 individuals; 4 of the 9 tortoises sampled from Zion were self-assigned from GENALEX (Table 1). F_ST and G′_ST values were comparable, with G′_ST values being consistently higher. F_ST values comparing Zion with other sites included in the analyses ranged from 0.029 (nearest location—Red Cliffs, Utah) to 0.070 (Fremont–Kramer, California), and increased as geographic distance increased between paired locations (G′_ST values ranged from 0.021 to 0.094). These values indicate low to moderate differentiation among the sites.

The 30 tortoises sampled in Red Cliffs were assigned to 10 of the 26 local population groups by GENALEX, including sites farther west into California, while the 9 tortoises from Zion were assigned to 5 local population groups. When Red Cliffs results were subsampled (n = 9, 1000 iterations), 2 to 9 unique populations were represented, with 5 populations occurring most frequently. Therefore, our result of 5 local population groups represented by Zion tortoises is in accordance with what we would expect from nearby population groups.

DISCUSSION

Mojave desert tortoises (Gopherus agassizii) in the Mojave and Sonoran deserts are federally protected wherever found, although it is important to understand whether individuals found at the northeastern extreme of the distribution represent a natural extent of the range. Using microsatellite data from tortoises near Zion compared with previously published data (Hagerty and Tracy 2010), our results suggest that tortoises in and near Zion include a mixture of residents to the region and individuals brought in from elsewhere or their descendants.

Between the 2 analyses conducted, most of the 9 Zion tortoises sampled were best assigned to the Virgin River genetic cluster of southwestern Utah (STRUCTURE analysis) or its local populations (GENALEX), indicating that Zion tortoises may represent an extension of the local population in the nearby Red Cliffs Desert Reserve (Table 1). Our results suggest that most of the tortoises are native to the region, although whether some were moved by humans or naturally dispersed is unknown. Two individuals clustered with populations much farther away in the tortoise distribution, indicating that there has likely been some artificial admixture.

Tortoises found in and near Zion are separated from other conspecifics (in the Red Cliffs Desert Reserve) by areas dominated by atypical habitat, including blackbrush instead of creosote bush, a prominent shrub in most of the desert tortoise range (Nussear et al. 2009, 2012). The elevation difference between the Red Cliffs Desert Reserve and the locations where Zion tortoises occur is > 200 m, and the straight-line distance between the 2 areas is nearly 30 km. This distance is not beyond the extremes that have been reported for desert tortoise movements. In a translocation study in the Red Cliffs Desert Reserve, a tortoise moved > 25 km in 1 yr, 3 yrs after being released at the site (Nussear et al. 2012). In a study on this species' close relative, a Sonoran desert tortoise (Gopherus morafkai) moved 30 km in 1 yr, although it received assistance by humans to overcome anthropogenic barriers (Edwards et al. 2004). Mojave desert tortoises are able to move long distances occasionally, so their occurrence in the northeastern portion of their range in Utah is not unexpected.

Although tortoises have not been reported between Zion and the nearby St. George, Utah area, habitat modeling incorporating landscape, climatic, plant, and soil characteristics suggests that this area provides some suitable tortoise habitat and moderate connectivity (Nussear et al. 2009; Gray et al. 2019). Furthermore, projections of climate warming for the next 100 yrs in the southern portions of Zion National Park are expected to make the area more suitable to tortoises (Thoma and Shovic 2013).

Tortoises in Zion may represent a relic population. If the habitat between what are now St. George and Springdale was once more continuous and hospitable to tortoises, then they may have once roamed the area more freely. However, Pleistocene habitat reconstructions along the Colorado River from the Sonoran Desert to the Colorado Plateau suggest that creosote bush likely did not reach elevations > 500 m, and it has spread upward over time (Cole 1990). At least in the relatively recent past, around the Wisconsin glaciation, tortoises were not likely to reside in the higher elevation areas near Zion. Furthermore, Woodbury and Hardy (1948) claim that the natural Mojave desert tortoise distribution ends at the Beaver Dam Mountains, suggesting that even the well-established Red Cliffs Desert Reserve tortoises were introduced by humans. Indeed, there is extremely poor habitat connectivity between the Red Cliffs area and the Beaver Dam Slope to the west (Averill-Murray et al. 2013), but regardless of how tortoises became established in the Red Cliffs area, their spread toward Zion could have been natural.

What is still unknown about Zion tortoises, however, is if the population size, habitat, and genetic variation allow for it to be a stable local population. If tortoises occasionally wander to Zion but are unable to successfully reproduce and recruit, the area might result in a population sink. Previous studies indicate that while most of the tortoises encountered at Zion are older adults, there is evidence of some reproductive success, including sightings of hatchlings and juveniles (McLuckie et al. 2000; National Park Service 2016), and genetic analysis of juvenile tortoises would uncover whether tortoises from native or introduced populations are reproducing. Importantly, further studies should aim to determine the spatial extent of tortoises in and around Zion, and programs should be put in place to monitor the population structure over time. Sampling of more individuals could also determine the extent of representation of other population segments.

In summary, Mojave desert tortoises in and near Zion National Park represent both resident and relocated or introduced tortoises, fitting with both of the competing hypotheses regarding this population group. This local population is the northeastern-most extent of the species distribution, and its viability and stability should be further studied and monitored.