This item was digitized as part of the Million Books Project led by Carnegie Mellon University and supported by grants from the National Science Foundation (NSF). Cornell University coordinated the participation of land-grant and agricultural libraries in providing historical agricultural information for the digitization project; the University of Arizona Libraries, the College of Agriculture and Life Sciences, and the Office of Arid Lands Studies collaborated in the selection and provision of material for the digitization project.

Spatial decision support systems (SDSS) integrate the state of the art technology, such as GIS, database and distributed models into decision support systems to support geospatial analysis that is particularly useful for watershed management, such as TMDL development on watersheds required by the Clean Water Act. This dissertation focuses on the development of a SDSS to assess the economic and environmental impacts from various best management practices (BMPs) in reducing sediment yield on rangeland watersheds.The SDSS included three major parts: the models, database and web-based interfaces. The model part is the core of the SDSS that provides the functionality of watershed economic analysis. The model maximized the profit of a representative ranch assumed to cover the whole watershed with the constraints of production technology, resource, sediment control objectives and sustainable utilization. A watershed was spatially segmented into basic units, each unit with similar plant growth and forage utilization. There are two major types of models, static and dynamic. Each model type supported variations in plant growth, grazing and ranch operations. Upland erosion was estimated through RUSLE2 and the sediment yield of a watershed was estimated from upland erosion and sediment delivery ratios for each basic unit. GAMS programs were used to solve the optimization models. The SDSS provides a platform to automatically implement the models. The database was the major tool in managing spatial and non-spatial data. A series of customized web pages were developed to support users' inputs, watershed analysis and result visualization. The embedded procedures were integrated into the SDSS to support analytical functionality, including geospatial analysis, model parameterization and web page generation.The SDSS was used to assess sediment control on the Walnut Gulch Experimental Watershed. The SDSS was parameterized primarily using publicly available data and a preliminary validation was made. The SDSS functionality was illustrated through eight applications. The results showed that given recent prices, new infrastructure practices would cause a financial burden to ranches. Better grazing management may provide an economic alternative to meet the sediment control objective and cost sharing could provide ranchers the incentives to participate in conservation plans.

Sprouting response of Quercus emoryi and Quercus arizonica was studied following a May, 1987 wildfire on the Audubon Society Research Ranch Preserve in Santa Cruz, Co., AZ. Sprouting response of oaks was measured as total number and density of sprouts by species and diameter class. Thirty-five of the 109 trees examined (32%) sprouted in response to the fire, including 30.5% of all Q. emoryi and 36.4% of all Q. arizonica. In both species, small diameter (10 cm) trees are more likely to sprout than those larger than 10 cm. Quercus emoryi trends toward producing more sprouts/tree than Q. arizonica (avg. of 14.03 sprouts to 6.87 sprouts). Among all sprouters, however, there is no difference in the number of sprouts produced by trees in the two size classes. Sprout density is greater in Q. emoryi at 3.58 sprouts/cm. Q. arizonica averages 0.89 sprouts/cm. In all sprouters, density is greater in smaller stems (4.17 sprouts/cm) compared with 0.71 sprouts/cm in the larger size class.

Germination responses of P.I. 237110 and 'Ganada' yellow bluestem (Bothriochloa ischaemum (L.) Keng.) were compared in four experiments. P.I. 237110 had the highest total germination for all seed conditioning treatments. At the lowest temperature (15.6°C) P.I. 237110 did not germinate and Ganada had a low total germination. P.I. 237110 and Ganada require high (>20°C) incubation temperatures for optimum germination. Seed conditioning treatments did not significantly increase total germination for P.I. 237110 or Ganada. The caryopsis treatment gave the highest percent germination for all germplasm except Ganada. P.I. 237110 had the fastest time to 50% germination for all seed conditioning treatments. Seed conditioning did not significantly affect time to 50% germination for P.I. 237110 or Ganada. Ganada had significantly higher respiration rates than P.I. 237110 at 0 and -0.4 MPa but not at -0.8 and -1.2 MPa. Respiration rates for both germplasm decreased as osmotic concentrations became more negative.

I determined if Sonoran pronghorn (Antilocapra americana sonoriensis) could meet water and mineral requirements through consumption of forage. I sampled vegetation using dry weight rank and comparative yield methods in southwestern Arizona from June 1995 to May 1996 to determine preformed water, nutrient content, mineral content, and vegetation abundance. Water and nutrient content of plants varied between 2 sites and among 5 seasons (P ≤ 0.05). Plants consumed by pronghorn were higher in moisture and nutrients (P ≤ 0.05) than non-forage species. I constructed a model for Sonoran pronghorn diet based on field data. I used the model to describe dietary intake. Water intake predicted by diet models ranged from 1.1 to 6.1 L/animal/day; often exceeding predicted total water requirements (1.8-3.4 L/animal/day). Models predicted that forage provided adequate water for pronghorn during 5 seasons at 1 of 2 study sites and deficiencies for 5 of 11 minerals at both study sites.

I compared vegetation features and rodent communities between manipulated and non-manipulated ranges in southeastern Arizona during summers 1996 and 1997. I also examined the effect of burying traps to determine if this procedure altered trap sensitivity. I used two-way analysis of variance or paired t-tests for all comparisons and identified relationships between rodents and vegetation features with linear regression. There was no difference (P ) in rodent species richness or population size between mechanically treated and reference areas. Rodent species richness and population size were greater on ungrazed compared to grazed areas (P ), but there was no difference in rodent diversity (P = 0.13). Prescribed fire did not have an obvious impact on rodent species richness or population size, though there seemed to be an increase in kangaroo rats following the burn. Buried traps may have demonstrated a reduction in sensitivity because I caught fewer animals in those traps compared to non-buried traps (P = 0.087).

Historically, wildfires in mixed conifer forests of Southwestern sky islands were frequent events. Dendrochronological methods were used to reconstruct fire regimes and stand age structures in the Huachuca Mountains of Southeastern Arizona. Pre-settlement (i.e., before ca. 1870) fire intervals ranged from 4 to 10 years, with many fires spreading over the entire sample area. Stand age distributions show an increase in more shade-tolerant tree species. Although ponderosa pine is still the dominant overstory tree species, recent recruitment is predominantly southwestern white pine and Douglas-fir. Establishment of Ft. Huachuca in 1877 was a precursor to extensive use of timber, mineral, range and water resources in the Huachuca Mountains. The fire regime was altered at this time, with only one subsequent widespread surface fire recorded in 1899. Settlement era land-use practices may be responsible for changes in stand structure and composition.

Masked bobwhites used sites with more structural diversity than what was available on the Buenos Aires National Wildlife Refuge during 1994-96. Selected habitat variables that adequately predicted habitat use by masked bobwhites included percent herbaceous canopy cover, structure at 15 cm and 2 m, forb richness, and season. Masked bobwhite used more structural diversity than Gambel's and scaled quail. Gambel's quail had broader habitat tolerances than either masked bobwhite or scaled quail. Selected habitat variables that revealed differences among masked bobwhites, Gambel's quail, and scaled quail included percent woody canopy cover, structure at 15 cm, forb richness, and season. A historic perspective revealed that masked bobwhites used sacaton grasses that grew along the floodplains as important escape cover. I believe that we need to restore the integrity of the grassland ecosystems including the floodplain if we are to recover masked bobwhite quail.

I examined trends in relative abundance and species richness of breeding and wintering grassland birds before (1996) and after (1997, 1998) a spring prescribed burn in a mesquite-dominated desert grassland at Buenos Aires National Wildlife Refuge, Arizona. The burn was moderate in intensity, patchy in extent, and affected ground cover more strongly than shrub cover, smaller shrubs more strongly than larger shrubs, and killed 1% of velvet mesquite (Prosopis velutina). Species richness of breeding birds decreased in the first year post-burn. Of breeding species, black-throated sparrows (Amphispiza bilineata) and mourning doves (Zenaida macroura) increased; whereas Botteri's sparrows (Aimophila botterii), Cassin's sparrows (Aimophila cassinii), and pyrrhuloxias (Cardinalus sinuatus) decreased in relative abundance. Breeding species characterized as not shrub-dependent exhibited changes that were more pronounced than those for shrub-dependent species. Of wintering birds, ladder-backed woodpeckers (Picoides scalaris) and vesper sparrows (Pooecetes gramineus) increased, and cactus wrens (Campylorhynchus brunneicapillus) decreased in relative abundance.

I determined density and habitat use of desert mule deer (Odocoileus hemionus crooki) on the Buenos Aires National Wildlife Refuge, a semidesert grassland in southeastern Arizona, in 1996. I observed 219 groups of deer; densities varied from 0.9 ± 0.3 (SE) deer/km² in summer to 2.5 ± 1.3 in winter. Herd size varied from 1.5 ± 0.1 deer/group in summer to 9.7 ± 2.0 in winter. Density of females was greater than males (0.9 ± 0.3 and 0.03 ± 0.04, respectively). Mule deer used subshrub-grass more in summer (χ² = 54.8, 6 df, P < 0.0001) and Russian thistle (Sallsola kali) less and cactus more in autumn (χ² = 60.2, 6 df, P < 0.0001) than expected. Deer used mesquite (Prosopis velutina) less than expected, which may have been due to the difficulty of observing deer in dense cover. In general, the mule deer population was widespread, had a high proportion of females, and used vegetation associations as available.