Tortoise Habitat Research Articles

Enhancing and Restoring Habitat for the Desert Tortoise

Abstract Habitat has changed unfavorably during the past 150 y for the desert tortoise Gopherus agassizii, a federally threatened species with declining populations in the Mojave Desert and western Sonoran Desert. To support recovery efforts, we synthesized published information on relationships of desert tortoises with three habitat features (cover sites, forage, and soil) and candidate management practices for improving these features for tortoises. In addition to their role in soil health and facilitating recruitment of annual forage plants, shrubs are used by desert tortoises for cover and as sites for burrows. Outplanting greenhouse-grown seedlings, protected from herbivory, has successfully restored (>50% survival) a variety of shrubs on disturbed desert soils. Additionally, salvaging and reapplying topsoil using effective techniques is among the more ecologically beneficial ways to initiate plant recovery after severe disturbance. Through differences in biochemical composition and digestibility, some plant species provide better-quality forage than others. Desert tortoises selectively forage on particular annual and herbaceous perennial species (e.g., legumes), and forage selection shifts during the year as different plants grow or mature. Nonnative grasses provide low-quality forage and contribute fuel to spreading wildfires, which damage or kill shrubs that tortoises use for cover. Maintaining a diverse “menu” of native annual forbs and decreasing nonnative grasses are priorities for restoring most desert tortoise habitats. Reducing herbivory by nonnative animals, carefully timing herbicide applications, and strategically augmenting annual forage plants via seeding show promise for improving tortoise forage quality. Roads, another disturbance, negatively affect habitat in numerous ways (e.g., compacting soil, altering hydrology). Techniques such as recontouring road berms to reestablish drainage patterns, vertical mulching (“planting” dead plant material), and creating barriers to prevent trespasses can assist natural recovery on decommissioned backcountry roads. Most habitat enhancement efforts to date have focused on only one factor at a time (e.g., providing fencing) and have not included proactive restoration activities (e.g., planting native species on disturbed soils). A research and management priority in recovering desert tortoise habitats is implementing an integrated set of restorative habitat enhancements (e.g., reducing nonnative plants, improving forage quality, augmenting native perennial plants, and ameliorating altered hydrology) and monitoring short- and long-term indicators of habitat condition and the responses of desert tortoises to habitat restoration.

Tick (Amblyomma chabaudi) infestation of endemic tortoises in southwest Madagascar and investigation of tick-borne pathogens

Little is known about the role of endemic ticks as vectors for bacterial and protozoan pathogens for animals and humans in Madagascar and their interaction in anthropogenic habitats where humans, their livestock and native Malagasy species (vectors and hosts) come into more frequent contact than in natural forest ecosystems. The aims of the study were (1) to test whether habitat degradation is associated with increased infestation of tortoises by ticks and (2) to investigate whether ticks carried Babesia, Borrelia or Rickettsia species that might be pathogenic for humans and livestock. We studied hard ticks of two endemic Malagasy tortoises, Astrochelys radiata and Pyxis arachnoides in March and April 2013 in southwest Madagascar. Two tortoise habitats were compared, the National Park of Tsimanampetsotsa and the adjacent degraded pasture and agricultural land at the end of the wet season. Ticks were screened for protozoan and bacterial pathogens via PCR on DNA isolated from ticks using genus-specific primers. Only one out of 42 A. radiata collected from both habitats had ticks. The low prevalence did not allow further analyses of the effect of habitat degradation. Forty-two P. arachnoides were found in the anthropogenic habitat and 36 individuals in the national park. Tick infestation rates of P. arachnoides differed significantly between the two study sites. Tortoises inside the park had lower tick prevalence than outside (8 of 36 (22%) versus 32 of 42 individuals (76%)) and infected animals tended to have fewer ticks inside than outside the park. All ticks collected in both habitats were adults of the ixodid tick Amblyomma chabaudi, which is supposed to be a host-specific tick of P. arachnoides. Screening for Borrelia sp. and Babesia sp. was negative in all ticks. But all A. chabaudi ticks were infected with Rickettsia africae, known to cause spotted fever in humans. Thus, habitat degradation seems to be linked to higher infestation of tortoises with ticks with possible consequences for humans and their livestock.

Enhancing Quality of Desert Tortoise Habitat: Augmenting Native Forage and Cover Plants

Abstract Vegetation in habitat of the federally listed desert tortoise Gopherus agassizii in the Mojave and western Sonoran Desert is now partly or mostly dominated by nonnative annual plants. To improve forage quality and augment availability of perennial cover plants, we tested seeding (pelletized or bare seeding), watering, and fencing for increasing a native annual forage species (desert plantain Plantago ovata), a perennial forage species (desert globemallow Sphaeralcea ambigua), and two shrub species (cheesebush Hymenoclea salsola and winterfat Krascheninnikovia lanata) that provide cover in desert tortoise habitat of southern Nevada. Treatments were ineffective at establishing the perennial species, even though greenhouse assays confirmed that some bare and pelletized seeds were germinable. In contrast, pelletized seeding quadrupled the density of desert plantain compared with not seeding or seeding untreated seed by the end of the first year (autumn 2013). Fencing tripled density of desert plantain to 17 plants/m2. Pelletized seeding plus fencing produced a desert plantain density of 39 plants/m2, the highest average density among all treatment combinations. The positive effect of fencing persisted until at least the second year after treatment (autumn 2014). Augmenting native annual forage plants favored by desert tortoises is feasible.

Desert tortoise use of burned habitat in the Eastern Mojave desert

ABSTRACTWildfires burned 24,254 ha of critical habitat designated for the recovery of the threatened Mojave desert tortoise (Gopherus agassizii) in southern Nevada during 2005. The proliferation of non‐native annual grasses has increased wildfire frequency and extent in recent decades and continues to accelerate the conversion of tortoise habitat across the Mojave Desert. Immediate changes to vegetation are expected to reduce quality of critical habitat, yet whether tortoises will use burned and recovering habitat differently from intact unburned habitat is unknown. We compared movement patterns, home‐range size, behavior, microhabitat use, reproduction, and survival for adult desert tortoises located in, and adjacent to, burned habitat to understand how tortoises respond to recovering burned habitat. Approximately 45% of home ranges in the post‐fire environment contained burned habitat, and numerous observations (n = 12,223) corroborated tortoise use of both habitat types (52% unburned, 48% burned). Tortoises moved progressively deeper into burned habitat during the first 5 years following the fire, frequently foraging in burned habitats that had abundant annual plants, and returning to adjacent unburned habitat for cover provided by intact perennial vegetation. However, by years 6 and 7, the live cover of the short‐lived herbaceous perennial desert globemallow (Sphaeralcea ambigua) that typically re‐colonizes burned areas declined, resulting in a contraction of tortoise movements from the burned areas. Health and egg production were similar between burned and unburned areas indicating that tortoises were able to acquire necessary resources using both areas. This study documents that adult Mojave desert tortoises continue to use habitat burned once by wildfire. Thus, continued management of this burned habitat may contribute toward the recovery of the species in the face of many sources of habitat loss. Published 2015. This article is a U.S. Government work and is in the public domain in the USA.

Not putting all their eggs in one basket: bet-hedging despite extraordinary annual reproductive output of desert tortoises

Bet-hedging theory makes the counter-intuitive prediction that, if juvenile survival is low and unpredictable, organisms should consistently reduce short-term reproductive output to minimize the risk of reproductive failure in the long-term. We investigated the long-term reproductive output of an Agassiz's desert tortoise (Gopherus agassizii) population and conformance to a bet-hedging strategy of reproduction in an unpredictable but comparatively productive environment. Most females reproduced every year, even during periods of low precipitation and poor germination of food plants, and the mean percentage of reproducing females did not differ significantly on an annual basis. Although mean annual egg production (clutch size × clutch frequency) differed significantly among years, mean clutch size and mean clutch frequency remained relatively constant. During an El Nino year, mean annual egg production and mean annual clutch frequency were the highest ever reported for this species. Annual egg production was positively influenced by maternal body size but clutch size and clutch frequency were not. Our long-term results confirm earlier conclusions based on short-term research that desert tortoises have a bet-hedging strategy of producing small clutches almost every year. The risk of long-term reproductive failure is minimized in unpredictable environments, both through time by annually producing multiple small clutches over a long reproductive lifespan, even in years of low resource availability, and through space by depositing multiple annual clutches in different locations. The extraordinary annual reproductive output of this population appears to be the result of a typically high but unpredictable biomass of annual food plants at the site relative to tortoise habitat in dryer regions. Under the comparatively productive but unpredictable conditions, tortoises conform to predictions of a bet-hedging strategy of reproduction with relatively small but consistent clutch sizes. Published 2015. This article is a U.S. Government work and is in the public domain in the USA, Biological Journal of the Linnean Society, 2015, 115, 399–410.

Sample Grain Influences the Functional Relationship Between Canopy Cover and Gopher Tortoise (Gopherus polyphemus) Burrow Abandonment

Abstract Change in vegetation structure alters habitat suitability for the threatened gopher tortoise (Gopherus polyphemus). An understanding of this dynamic is crucial to inform habitat and tortoise management strategies. However, it is not known how the choice of the sample grain (i.e., cell size) at which vegetation structure is measured impacts estimates of tortoise–habitat relationships. We used lidar remote sensing to estimate canopy cover around 1573 gopher tortoise burrows at incrementally larger sample grains (1–707 m2) in 450 ha of longleaf pine (Pinus palustris) savanna. Using an information theoretic approach, we demonstrate that the choice of grain size profoundly influences modeled relationships between canopy cover and burrow abandonment. At the most supported grain size (314 m2), the probability of burrow abandonment increased by 1.7% with each percent increase in canopy cover. Ultimately, detecting the appropriate sample grain can lead to more effective development of functional relations...

Climatic variation and tortoise survival: Has a desert species met its match?

While demographic changes in short-lived species may be observed relatively quickly in response to climate changes, measuring population responses of long-lived species requires long-term studies that are not always available. We analyzed data from a population of threatened Agassiz’s desert tortoises (Gopherus agassizii) at a 2.59km2 study plot in the Sonoran Desert ecosystem of Joshua Tree National Park, California, USA from 1978 to 2012 to examine variation in apparent survival and demography in this long-lived species. Transect-based, mark-recapture surveys were conducted in 10 of those years to locate living and dead tortoises. Previous modeling suggested that this area would become unsuitable as tortoise habitat under a warming and drying climate scenario. Estimated adult population size declined greatly from 1996 to 2012. The population appeared to have high apparent survival from 1978 to 1996 but apparent survival decreased from 1997 to 2002, concurrent with persistent drought. The best model relating apparent survivorship of tortoises ⩾18cm over time was based on a three year moving average of estimated winter precipitation. The postures and positions of a majority of dead tortoises found in 2012 were consistent with death by dehydration and starvation. Some live and many dead tortoises found in 2012 showed signs of predation or scavenging by mammalian carnivores. Coyote (Canis latrans) scats and other evidence from the site confirmed their role as tortoise predators and scavengers. Predation rates may be exacerbated by drought if carnivores switch from preferred mammalian prey to tortoises during dry years. Climate modeling suggests that the region will be subjected to even longer duration droughts in the future and that the plot may become unsuitable for continued tortoise survival. Our results showing wide fluctuations in apparent survival and decreasing tortoise density over time may be early signals of that possible outcome.

Resource Selection Probability Functions for Gopher Tortoise: Providing a Management Tool Applicable Across the Species’ Range

The gopher tortoise (Gopherus polyphemus) is protected by conservation policy throughout its range. Efforts to protect the species from further decline demand detailed understanding of its habitat requirements, which have not yet been rigorously defined. Current methods of identifying gopher tortoise habitat typically rely on coarse soil and vegetation classifications, and are prone to over-prediction of suitable habitat. We used a logistic resource selection probability function in an information-theoretic framework to understand the relative importance of various environmental factors to gopher tortoise habitat selection, drawing on nationwide environmental datasets, and an existing tortoise survey of the Ft. Benning military base. We applied the normalized difference vegetation index (NDVI) as an index of vegetation density, and found that NDVI was strongly negatively associated with active burrow locations. Our results showed that the most parsimonious model included variables from all candidate model types (landscape features, topography, soil, vegetation), and the model groups describing soil or vegetation alone performed poorly. These results demonstrate with a rigorous quantitative approach that although soil and vegetation are important to the gopher tortoise, they are not sufficient to describe suitable habitat. More widely, our results highlight the feasibility of constructing highly accurate habitat suitability models from data that are widely available throughout the species' range. Our study shows that the widespread availability of national environmental datasets describing important components of gopher tortoise habitat, combined with existing tortoise surveys on public lands, can be leveraged to inform knowledge of habitat suitability and target recovery efforts range-wide.

Integration of Pipeline Operations Sourced with CO2 Captured at a Coal-fired Power Plant and Injected for Geologic Storage: SECARB Phase III CCS Demonstration

This paper presents a case study of the design and operation of a fit-for-purpose pipeline sourced with anthropogenic carbon dioxide (CO2) associated with a large-scale carbon capture and storage (CCS) Research & Demonstration Program located in Alabama, USA. A 10.2 centimeter diameter pipeline stretches approximately 19 kilometers from the outlet of the CO2 capture facility, located at Alabama Power Company's James M. Barry 2,657 - megawatt coal-fired electric generating plant, to the point of injection into a saline reservoir within Citronelle Dome.The CO2 pipeline has a 6.5 meter wide easement that primarily parallels an existing high-voltage electric transmission line in undulating terrain with upland timber, stream crossings, and approximately 61,000 square meters of various wetland types. In addition to wetlands, the route transects protected habitat of the Gopher Tortoise. Construction methods included horizontal drilling under utilities, wetlands, and tortoise habitat and ‘open cutting’ trenching where vegetation is removed and silt/storm-water management structures are employed to limit impacts to water quality and ecosystems. A total of 18 horizontal directional borings, approximately 8 kilometers, were used to avoid sensitive ecosystems, roads, and utilities.The project represents one of the first and the largest fully-integrated pulverized coal-fired CCS demonstration projects in the USA and provides a test bed of the operational reliability and risk management for future pipelines sourced with utility CO2 capture and compression operations sole-sourced to injection operations. An update on status of the project is presented, covering the permitting of the pipeline, risk analysis, design, construction, commissioning, and integration with compression at the capture plant and underground injection at the storage site.

Translocation as a conservation tool for Agassiz's desert tortoises: Survivorship, reproduction, and movements

AbstractWe translocated 120 Agassiz's desert tortoises to 5 sites in Nevada and Utah to evaluate the effects of translocation on tortoise survivorship, reproduction, and habitat use. Translocation sites included several elevations, and extended to sites with vegetation assemblages not typically associated with desert tortoises in order to explore the possibility of moving animals to upper elevation areas. We measured survivorship, reproduction, and movements of translocated and resident animals at each site. Survivorship was not significantly different between translocated and resident animals within and among sites, and survivorship was greater overall during non‐drought years. The number of eggs produced by tortoises was similar for translocated and resident females, but differed among sites. Animals translocated to atypical habitat generally moved until they reached vegetation communities more typical of desert tortoise habitat. Even within typical tortoise habitat, tortoises tended to move greater distances in the first year after translocation than did residents, but their movements in the second or third year after translocation were indistinguishable from those of resident tortoises. Our data show that tortoises translocated into typical Mojave desert scrub habitats perform well; however, the large first‐year movements of translocated tortoises have important management implications. Projects that employ translocations must consider how much area will be needed to contain translocated tortoises and whether roads need fencing to prevent the loss of animals. © 2012 The Wildlife Society.

Estimating the probability of illegal desert tortoise collection in the Sonoran Desert

AbstractThe expansion of road networks in desert tortoise (Gopherus agassizii) habitat in the Sonoran Desert has raised questions concerning appropriate mitigation to reduce impacts at the population level. Although some effects, namely road‐kill and habitat loss, have been well documented, illegal tortoise collection has been insufficiently addressed. It has become increasingly important for wildlife and land‐use managers to understand the cumulative impacts of roads on tortoises and the effect that those impacts have on population persistence. We estimated the probability of desert tortoise detection and collection along 2‐lane paved, maintained gravel, and non‐maintained gravel roads to evaluate whether collection probabilities were related to road type. Although collection probability did not vary by road type, the probability of desert tortoise detection by passing motorists was greatest on maintained gravel roads and fewest on non‐maintained gravel and paved roads. These results have implications for effectively mitigating the impacts of roads on desert tortoises. Published 2011. This article is a U.S. Government work and is in the public domain in the USA.

The turtle is in the details: microhabitat choice by Testudo hermanni is based on microscale plant distribution

Abstract Although research on habitat use and habitat selection is essential for understanding population ecology and behavior, most such zoological studies have used only general habitat categories describing main habitat features instead of using modern plant ecological approaches. Here, we analyze Testudo hermanni microhabitat use at a coastal Mediterranean site in central Italy by modeling tortoise presence/absence at three spatial scales, using a logistic regression design and quantitative vegetation and plant community analysis to reveal correlates of tortoise habitat use on a fine scale. Our analyses showed that only a few plant species among the many present, and these on a very small spatial scale, are important determinants of tortoise presence and site selection. We also find that tortoises chose a paradoxical combination of high levels of bare soil and high total vegetation cover. This suggests that these tortoises are selecting small patches of habitat in a matrix of less desirable habitat. Our findings also have important implications for habitat management, in that increasing the number of habitat patches containing the few significant plants is likely to be desirable, whereas increasing the size of such patches is probably less relevant.

Desert Tortoise Microhabitat Selection on the Florence Military Reservation, South-Central Arizona

We examined Desert Tortoise microhabitat selection on the Florence Military Reservation (FMR) in south-central Arizona where National Guard training and other activities, such as motorized recreation and grazing, are permitted. Previous research on the FMR indicated that Desert Tortoise home ranges overlapped with firing boxes, areas where most of the National Guard's training activity was concentrated. Our objective was to examine Desert Tortoise microhabitat selection within, and adjacent to, these firing boxes in an effort to guide future management for restoring degraded tortoise habitat on the installation. Desert Tortoises selected habitat that was characterized by a higher percentage of canopy cover, absence of cattle activity, and closer proximity to roads and washes than was available within their home range. Canopy cover had the highest calculated parameter importance and was included in each of the supported microhabitat selection models. Areas with sufficient canopy cover most likely provided shade for escaping the desert heat during periods of high tortoise activity. Our results suggest that management prescriptions that maintain, or increase, the amount of vegetative cover and that protect desert washes will have positive impacts on Desert Tortoise populations.

Seroprevalence of Mycoplasma agassizii in Wild Caught and Rescued Texas Tortoises (Gopherus berlanderi) in South Texas

The Texas tortoise (Gopherus berlanderi) is one of four native tortoise species found in North America. All four Gopherus species receive some form of government protection because of recent population declines. The Texas tortoise is considered threatened within the state of Texas and is of major conservation interest. As a result of human encroachment into Texas tortoise habitats, there has been an increase in the dissemination of infectious disease into native populations. One of the most important diseases studied in North American tortoises is Mycoplasma agassizii. Seroprevalence studies have been performed in the California desert tortoise (Gopherus agassizii) and Gopher tortoise (Gopherus polyphemus) but have yet to be performed in the Texas tortoise. The purpose of this study was to measure the seroprevalence of Mycoplasma agassizii using an ELISA in two populations of Texas tortoises, a wild population (N = 39) and a population in a wildlife rehabilitation facility (N = 15). Tortoises in the wild population were all seronegative (95% confidence intervals [CI]: 0–7.6%), whereas 80% (12/15; 95% CI: 60–100) of the tortoises from the rehabilitation facility were seropositive. The results of this study suggest that Texas tortoises are exposed to and can mount an immunoglobulin response to Mycoplasma agassizii, which may influence the status of this species in North America. This pilot study suggests that further research is needed to determine the epidemiology of mycoplasmosis in Texas tortoises from the wild and in captivity.

An Annual Plant Growth Proxy in the Mojave Desert Using MODIS-EVI Data.

In the arid Mojave Desert, the phenological response of vegetation is largely dependent upon the timing and amount of rainfall, and maps of annual plant cover at any one point in time can vary widely. Our study developed relative annual plant growth models as proxies for annual plant cover using metrics that captured phenological variability in Moderate-Resolution Imaging Spectroradiometer (MODIS) Enhanced Vegetation Index (EVI) satellite images. We used landscape phenologies revealed in MODIS data together with ecological knowledge of annual plant seasonality to develop a suite of metrics to describe annual growth on a yearly basis. Each of these metrics was applied to temporally-composited MODIS-EVI images to develop a relative model of annual growth. Each model was evaluated by testing how well it predicted field estimates of annual cover collected during 2003 and 2005 at the Mojave National Preserve. The best performing metric was the spring difference metric, which compared the average of three spring MODIS-EVI composites of a given year to that of 2002, a year of record drought. The spring difference metric showed correlations with annual plant cover of R2 = 0.61 for 2005 and R2 = 0.47 for 2003. Although the correlation is moderate, we consider it supportive given the characteristics of the field data, which were collected for a different study in a localized area and are not ideal for calibration to MODIS pixels. A proxy for annual growth potential was developed from the spring difference metric of 2005 for use as an environmental data layer in desert tortoise habitat modeling. The application of the spring difference metric to other imagery years presents potential for other applications such as fuels, invasive species, and dust-emission monitoring in the Mojave Desert.

Phenology of Attempted Matings in Gopher Tortoises

Observations of behaviors associated with reproduction in Gopher Tortoises (Gopherus polyphemus) have been reported in several studies, yet no clear temporal pattern of such activities has emerged from these observations. We used remote cameras to monitor a total of 37 focal female Gopher Tortoises, from which we documented seasonal and diel patterns of mountings by males. This was done at two sites in southern Georgia that represent the best remaining tortoise habitat of the northern portion of the geographic distribution of the species. Mounting behavior occurred as early as March and as late as October and had a unimodal peak during July, August, and September, a pattern that mimics testosterone cycles in males and estradiol cycles in females. For all months during which mountings were observed, the majority of mounts occurred from 1500 hr to dusk, a pattern that was more pronounced during peak months of attempted matings. We estimated that individual adult females experience 26 mount attempts from an average of 6 to 14 males each year.

Gopher Tortoise Response to Habitat Management by Prescribed Burning

ABSTRACT As quality of forested habitat declines from altered fire regimes, gopher tortoises (Gopherus polyphemus) often move into ruderal areas to the detriment of the animal and land manager. We evaluated effects of a dormant‐and‐growing‐season prescribed fire on habitat and gopher tortoise use of degraded longleaf pine (Pinus palustris) forests surrounding military training areas. We burned 4 of 8 sites in winter 2001–2002 and again in April 2003. Changes in vegetation measured during 2001–2004 indicated that burn treatments did not increase herbaceous vegetation. Similarly, movement patterns, burrow usage, and home range of tortoises radiotracked from 2002–2004 did not differ between treatments. Woody cover initially was reduced in the forests postburn, and we found more new burrows in burned forest sites. Once shrub cover was reduced, tortoises started using forested habitat that had become overgrown. However, shrub reduction may be temporary, as woody stem densities increased postburn. Thus, the one‐time use of fire to manage tortoise habitat may not rapidly restore the open canopy, sparse woody midstory, and abundant herbaceous vegetation that this species requires. Repeated prescribed fires or additional management techniques may be needed for complete restoration.

Habitat Modeling Within a Regional Context: An Example Using Gopher Tortoise

Changes in habitat are often a major influence on species distribution and even survival. Yet predicting habitat often requires detailed field data that are difficult to acquire, especially on private lands. Therefore, we have developed a model that builds on extensive data that are available from public lands and extends them to surrounding private lands. This model is applied for a five-county region in Georgia to predict habitats for the gopher tortoise (Gopherus polyphemus), based on analysis of documented locations of gopher tortoise burrows at the Fort Benning military installation in west central Georgia. Burrow associations with land cover, soil, topography and water observed within the military installation were analyzed with binary logistic regression. This analysis helped generate a probability map for the occurrence of gopher tortoise burrows in the five-county region surrounding Fort Benning. Ground visits were made to test the accuracy of the model in predicting gopher tortoise habitat. The results showed that information on land cover, soils, and distances to streams and roads can be used to predict gopher tortoise burrows. This approach can be used to better understand and effectively carry out gopher tortoise habitat restoration and preservation activities.

Abundance and distribution of selected elements in soils, stream sediments, and selected forage plants from desert tortoise habitats in the Mojave and Colorado deserts, USA

A baseline and background chemical survey was conducted in southeastern California, USA, to identify potential sources of toxicants in natural and anthropogenically-altered habitats of the threatened desert tortoise ( Gopherus agassizii). Soil, stream sediment, and plant samples were collected from six tortoise habitat study areas in the Mojave and Colorado deserts and analysed for up to 66 different elements. The chemical analyses provided new information on the abundances and distributions of selected elements in this region. Soil, stream-sediment, and plant analyses showed distinct variations in bulk chemistries from locality to locality. Variations were, in general, consistent with the many types of exposed rock units in the region, their highly variable bulk mineralogies, and chemical contents. Of elements in soils that might have been toxic to tortoises, only As seemed to be anomalous region-wide. Some soil and plant anomalies were clearly anthropogenic. In the Rand and Atolia mining districts, soil anomalies for As, Au, Cd, Hg, Sb, and(or) W and plant anomalies for As, Sb, and(or) W extend as far as ∼15 km outward from the present area of mining; soils containing anomalous Hg were found at least 6 km away from old piles of tailings. The anomalous concentrations of As and Hg may have been the source of elevated levels of these elements found in ill tortoises from the region. In the Goldstone mining district, soil anomalies extended several km from the mining area. These areas probably represented anthropogenic surface contamination of dust redistributed by wind, vehicles, and rainfall. One of two study areas transected by a paved road (Chemehuevi Valley) showed weakly elevated levels of Pb, which extended as far as ∼22 m from the pavement edge and were probably related to vehicle exhaust. No soil or plant samples from historically used military areas (Goldstone, Goffs, Chemehuevi Valley, Chuckwalla Bench) contained anomalous concentrations of the elements As, Cu, Cr, Fe, Pb, or Zn that could be ascribed to military maneuvers, vehicles, or ordnance. For future studies, the distribution and abundance of elements in the tortoise forage plants need to be evaluated for the respective roles of dust and systemic uptake. Additional chemical data from tortoise necropsies and nutritional studies are needed to determine the effects of potentially toxic elements in tortoise habitats on their health.

REGRESSION-TREE MODELING OF DESERT TORTOISE HABITAT IN THE CENTRAL MOJAVE DESERT

This paper describes an interdisciplinary study of the habitat requirements of threatened desert tortoises (Gopherus agassizii) on eight 225-ha study plots in a 14 000 ha study area near the southern boundary of the U.S. Army's National Training Center at Fort Irwin in the central Mojave Desert of southern California. The objective of the study was to produce an empirical, statistical, GIS-based model of desert tortoise habitat use based on a combination of field data and data derived from various spatial databases, including satellite imagery. A total of 11 primary and secondary data layers constitute the spatial database used for this project. Vegetation and tortoise relative density data were obtained from field surveys. Regression-tree methods were used to develop the statistical model. The tree has 11 terminal nodes and a residual mean deviance of 1.985. Out of 73 potential predictors in the model specification, only eight were selected by the algorithm to be used in construction of the tree. The mod...