Red-tailed hawks (Buteo jamaicensis), Swainson's hawks (Buteo swainsoni), great horned owls (Bubo virginiaus), and ravens (Covus corax and C. cryptoleucus) occur sympatrically in desert grasslands in Arizona. Desert grasslands have been invaded by trees since Anglo-American settlement. They may also be subject to greater human disturbance than historically, because of recent increases in illegal immigration along the U.S.-Mexico border. These changes may affect the abundance, diversity, competitive interactions, and reproductive success of raptors and corvids in desert grasslands. I studied raptors and corvids on the Santa Rita Experimental Range (SRER) and Buenos Aires National Wildlife Refuge (BANWR) in southern Arizona. My objectives were to (1) assess how spatial relationships among nesting raptors and corvids affect reproductive success, (2) identify characteristics of the environment selected by members of the assemblage for nest sites and home ranges, (3) identify similarities and differences in nest site and home-range characteristics among species, and (4) investigate how anthropogenic changes in desert grasslands have affected raptors and corvids. Productivity (x̄ [95% CI]) was high for red-tailed hawks on the SRER (1.82 [1.41, 2.23] young per breeding pair [YPB]), but low for all species on the BANWR (great horned owls: 0.96 [0.54, 1.38] YPB; ravens: 1.75 [1.39, 2.10]; red-tailed hawks: 1.04[0.83, 1.24]; Swainson's hawks: 0.91 [0.67, 1.16]) relative to range-wide averages (great horned owls: 1.42 [1.27, 1.56]; ravens: 2.39 [2.70, 3.07]; red-tailed hawks: 1.35 [1.20, 1.50]; Swainson's hawks: 1.34 [1.23, 1.45]). All species on both sites selected nest sites with taller nest trees and greater tree volume than available at random. Swainson's hawks selected home ranges with greater grass volume than available on BANWR. Overlap in habitat use was high among all species, exceeding 54% for nest trees, 80% for nest sites, and 90% for home ranges. Mesquite-dominated desert grasslands seemed to provide high quality habitat for red-tailed hawks on the SRER, but reproductive success for assemblage members on BANWR was poor. We suggest that competition, resulting from high habitat overlap, and human disturbance by illegal immigrants, has depressed raptor and corvid reproductive success on the BANWR.

I used archives of wildfire perimeters (fire atlases) within a geographic information system (GIS) to describe and evaluate fire frequency patterns for the 20th century in the 486,673-ha Gila/Aldo Leopold Wilderness Complex (GALWC), New Mexico and the 785,090-ha Selway-Bitterroot Wilderness Complex (SBWC) in Idaho and Montana. I addressed questions about changing 20th century fire frequencies and landscape-scale controls of fire frequency by analyzing fire atlases along with data for topography, vegetation, and climate. Similarities and differences in comparisons between study areas highlight important aspects of fire regimes and strengthen my interpretation and inference. In the GALWC, fire rotations were shortest during the recent wildfire use period (1975-1993) and longest during the pre-modern suppression period (1909-1946). In the SBWC, fire rotations were shortest during the pre-modern suppression period (1880-1934) and longest during the modern suppression period (1935-1975). Elevations with the highest fire frequencies differed between study areas. However, forest types found at these elevations are characterized by similar overstory tree species. Steeper northeastern slopes in the GALWC and southwestern slopes in the SBWC burned most frequently. I assert that, in the GALWC, horizontal fuel continuity is a main factor determining fire frequency. In the SBWC, fuel moisture status limits fire frequency. Fires are most frequent in areas where ignitions occur and neither fuel continuity nor fuel moisture are likely to limit fire spread. Three statistical modeling approaches were used to produce maps of reburn probabilities. Log-likelihood modeling provided the most satisfactory results, while logistic regression and classification and regression trees yielded statistically insignificant models. Empirical models contributed to the assertion that fuel continuity limits fire frequency in the GALWC while fuel moisture limits fire frequency in the SBWC. Mapped fire perimeters provide a valuable source of spatial historical information for describing the role of large fires over broad areas. This dissertation enhances scientific knowledge about broad scale changes in fire regimes. Comparisons between areas facilitate identification of unique versus general patterns. Results provide a contemporary baseline for comparison with estimates of Pre-EuroAmerican fire frequencies, and a historical, spatial context for modeling and managing future fire regimes.

I studied Gray Hawks (Asturina nitida) nesting along the upper San Pedro River in southeastern Arizona from 1995 to 1997. I identified 27 territories within the study area with a mean of 24.3 territories occupied per year. Productivity averaged 1.32 young per occupied nest during the study. Years did not differ in number of successful nests or in number of young produced. Mean size of home ranges (n = 10 males), based on the 90% adaptive kernel method, was 59.2 ha (range 21.4-91.2). Gray Hawk diet was composed of 68.6% reptiles, the majority being whiptail lizards ( Cnemidophorus spp). Daily foraging patterns were bimodal with peaks during mid-morning and late afternoon. Almost all Gray Hawk nests were located in cottonwoods (Populus fremontii). Nest trees tended to be dominant trees in the area. Gray Hawks did not use vegetation randomly within their home ranges. Mesquite (Prosopis juliflora) and cottonwood were used more than expected based on availability. Nearly 80% of all foraging locations were in mesquite. Within mesquite, Gray Hawks used areas of taller trees and lower horizontal cover than found at random sites. Trees used as foraging perches were taller than the surrounding canopy. Foraging and random sites did not differ in prey abundance indicating that prey availability is more important than prey abundance in determining where Gray Hawks forage. I developed a method of predicting home ranges for all 27 territories utilizing telemetry information from a sample of male hawks, a geographic information system, and logistic regression. This model produced estimates of vegetative composition within home ranges that were correlated strongly with vegetation composition in home ranges based on telemetry. The models I developed performed better than circular buffers around nest sites. Using the predicted home ranges, I developed an ordinal logistic model to identify what vegetative components influence Gray Hawk productivity. Area of mesquite was the primary determinant of habitat quality in Gray Hawks. Cottonwood-willow (Salix gooddingii) vegetation also can increase quality of home ranges when mesquite is present in low amounts.

Coyotes (Canis latrans) are common in many urban areas in North America, but little is known about how they adapt to urbanization. I studied the way they use the urban landscape by determining the patch types chosen for foraging, travelling, and resting. Home ranges encompassed a smaller proportion of natural areas and a greater proportion of parks and residential areas than were available in Tucson. Rates of nocturnal movement by coyotes peaked at midnight and at dawn. Natural areas, parks, and residential areas were used most by coyotes throughout the night, but rates of movement did not differ among these patch types. I created a GIS (geographic information system) model to predict the presence or absence of coyotes in all patch types throughout Tucson. The model indicates that most areas in Tucson are likely to be used by coyotes; it may be refined by adding new variables and by examining these variables at a finer spatial scale. Studies have not documented the health of coyotes in urban areas, where they may be reservoirs of disease for domestic canids. I determined the prevalence of pathogens, estimated survival rates, and identified sources of mortality. The prevalence of canine distemper virus was 27%, the prevalence of infectious canine hepatitis was 50%, the prevalence of canine parvovirus was 100%, and the prevalence of leptospirosis was 27%. The annual survival rate of coyotes was 0.72. Most necropsied coyotes were killed by humans. Coyotes in urban areas come into conflict with humans. Few humans have been attacked by coyotes, but small pets are susceptible to attack. The primary means of deterring coyotes from areas where their presence is unwanted is by eliminating all sources of food and water. The urban-dwelling public must become better educated about coyotes to minimize human-coyote conflicts.

In the states of Tabasco and Chiapas, Mexico there is a lack of long-term harvest scheduling models that consider the effects of the harvest activities on the surrounding areas. Additionally, these problems are combinatorial in nature, which makes them hard to solve. Consequently, only harvest scheduling for small areas can be solved to optimality using traditional approaches such as integer programming (IP). In this study, a genetic algorithms (GA) model was used to generate multiple viable solutions for long-term spatially constrained problems on large areas with a great number of management units. This model enables consideration of regeneration and reharvest in forest planning. The flexibility of the model allows it to handle a different set of time periods, database sizes, different species and diverse tree growth models. The data set employed corresponds to a eucalyptus plantation with a cutting cycle seven years and a planning horizon of 10 rotation periods. Total plantation area is 300,000 ha, divided in 5,388 harvest units. IP was used as a standard to validate the efficiency and accuracy of the GA method. The GA performance with different combinations of genetic operators was tested. Scheduled volume flow for simulated communities was computed. Additionally, three different volume assignment scenarios (low, medium and high) were compared to estimate the effect of volume assignment on the spatial optimization output. The significant findings of this research are: (1) a long-term spatially constrained robust solution was found through the use of genetic algorithms for a large area with more harvest units than those reported elsewhere, (2) the solution allowed re-harvest in the same planning horizon, (3) most of the genetic algorithms runs performed better than the integer programming, and (4) on average the volume scheduled for every simulated community was comparable for the two methods used in the work. In both cases, the percentages of the potential volume ranged between 7 and 29%.

The management of bighorn sheep in Mexico is not based on empirical data. We conducted phone interviews and a literature review to determine published information on desert bighorn sheep in Mexico. Most information on bighorn sheep was published prior to 1983 and focused on population size and structure. We concluded that desert bighorn sheep in Mexico are in need of further research. To provide better management to bighorn sheep, research needs to focus on basic life history strategies (e.g., minimum viable population size, diet, habitat relationships, movements, predation, diseases, and metapopulation dynamics). Bighorn sheep in Mexico have been scarcely studied. We conducted a habitat use study in Rancho el Plomito, Sonora, Mexico from April 1997 to December 1998 and the data were used to determine causes of sexual segregation. Vegetation associations were not used according to their availability. Segregated females formed larger groups, were closer to escape terrain and occupied more rugged areas than males. Sexual segregation exhibited by desert bighorn sheep in Rancho el Plomito may result from different energetic and reproductive strategies for males and females. Females tend to optimize conditions suitable for rearing offspring and males maximize body condition. Bighorn sheep in Mexico are important game animals and are confined to scattered populations (Villa 1.959). Little research has been conducted on desert bighorn sheep and consequently the species is listed in appendix II of CITES (i.e., vulnerable) (Valdez 1997). Tarangb and Krausman (1997) emphasized food selection as a fundamental research topic for the species in Mexico. We conducted a study of diet composition in Sonora, Mexico from April 1997 to December 1998. We identified 41 plant species consumed by desert bighorn sheep. The mean annual diet was composed of browse, forbs, succulents, and grasses. Composition of forage categories in diets of males and females was similar. The coefficient of overlap (Morisita 1959) of diets of male and female bighorns for seasons indicated significant overlap. Sexual segregation of desert bighorn sheep in Sonora, Mexico was not related to forage selection (i.e., botanical composition, forage quality), and suggested that other factors (i.e., physiographic characteristics) were responsible for this segregation.

From April 1988 to August 1988 and from March 1989 to August 1989 I studied habitat utilization of the endangered Sonoran pronghorn (Antilocapra americana sonoriensis) on the Cabeza Prieta National Wildlife Refuge and the Organ Pipe Cactus National Monument in southwestern Arizona. Pronghorn groups were small (x = 2.5) and dispersed widely throughout the study area. I never saw Sonoran pronghorns drinking water. Cacti were the major diet components in the dry seasons. Forbs were the major diet component in the wet season. Cover, plants, and vertical obstruction was generally greater in occupied areas than in unoccupied areas. Plants in occupied areas were higher in protein than plants in unoccupied areas. The percentage of indigestible material (e.g., cellulose) in plants from occupied areas was generally less than in unoccupied areas. My data represent the characteristics of vegetation in areas used by Sonoran pronghorn. They could be applied in an intensive habitat management program to improve the quality of pronghorn habitat.

I analyzed 1,182 gray wolf (Canis lupus) scats, representing summer diet in 1988-1990, collected from 9 packs in northwest Alaska. Using a linear regression model, I determined the biomass of prey species consumed using relative estimated bulk and frequency of occurrence of prey types in scats. There were no significant differences (P < 0.05) between methods in percent biomass of prey consumed and ranking of prey types. Caribou (Rangifer tarandus), moose (Alces alces), microtines, and birds composed 47.0, 21.9, 10.6, and 13.8%, of prey items, respectively. Ungulates composed 90.1%, birds 6.7%, and microtines 0.5% of biomass consumed. Ungulates composed 3.4%, birds 50.4%, and microtines 38.5% of numbers of individuals consumed. Significant (P < 0.05) observer bias was found in the detection and frequency of prey types in the analysis of 1,052 scats by multiple observers.

I evaluated different approaches for monitoring of amphibians, reptiles, and mammals in national parks based on data from Tonto National Monument, Arizona, and 3 criteria: statistical power, cost, and ability to meet monitoring goals. Detecting a declining trend in abundance (18% over 10 years) with 80% power would require approximately 65 person-days of annual sampling for the 2 most common rodents, and 110 person-days for the 2 most common lizards. Detecting this trend in species richness of terrestrial vertebrates (reptiles, amphibians, and mammals) would require only 30 person-days annually. Monitoring abundances of vertebrates is often recommended for parks, but my study suggests that this approach is too expensive for most parks, and results may be inadequate for achieving agency goals of monitoring biological diversity. I advocate use of species richness in monitoring, and provide field and analytical methods for estimating this parameter.

I evaluated water use by female Coues white-tailed deer (Odocoileus virginianus couesi) and calculated their diurnal ranges relative to available water in southeastern Arizona during hot-dry and hot-wet summer seasons. Developed water sources were scarce in 1988 and abundant during 1989. Freestanding water was not the only determinant of diurnal seasonal ranges. Not all diurnal seasonal ranges contained permanent water, but a water source density of 1/km² was sufficient during the 1989 hot-dry season and ephemeral water was common in hot-wet seasons. Deer visited developed water sources regularly during hot-dry seasons and infrequently 14 days after the summer rains began. Large amounts of preformed water were available in hot-wet season diets, but permanent or ephemeral water was necessary to meet total water requirements. Visits of cattle to troughs did not preclude deer use because of different activity patterns between the 2 species. Wildlife managers should provide free-standing water in Coues white-tailed deer ranges during hot-dry seasons and in the early part of hot-wet seasons. If winter precipitation is substantially below normal, available free-standing water in summer is especially critical.