There is considerable public and political pressure to reduce woody plant cover on rangelands as a means of increasing water yield, despite the lack of studies documenting that such a strategy is effective. In the Texas Hill Country, runoff from the Edwards Plateau recharges the highly productive and regionally vital Edwards Aquifer. The dominant woody plant on the Plateau is Ashe juniper (Juniperus ashei Buchholz). To understand how woody plant cover may affect the amount and timing of runoff in this region, we monitored streamflow from nine small (3‐ to 6‐ha) watersheds over a 13‐year period. After the first two years (initial observations), 100% of the shrub cover was removed from three of the watersheds and ∼70% from another three. Following these treatments we continued to monitor runoff for four years, suspended monitoring for four and a half years, and then resumed monitoring for an additional three years. Runoff from these nine first‐order watersheds generally accounted for <5% of the total precipitation and occurred entirely as stormflow (there was no baseflow before or after treatment). Some runoff was generated as subsurface flow, as indicated by hydrographs showing prolonged runoff (typically lasting hours longer than the rainfall). We evaluated the influence of woody plant cover on streamflow by comparing streamflow during the four‐year treatment period with that during the posttreatment period (when considerable recovery of woody plants had taken place). Our findings indicate that changes in woody plant cover had little influence on the amount, timing, or magnitude of streamflow from these watersheds. On the basis of this work and other observations in the region, we hypothesize that, for small watersheds, changes in shrub cover will have little or no effect on streamflow except where springs are present.

Understanding hydrologic processes on rangelands is essential to determine if water yield will increase through shrub management. Nine nonweighable lysimeters were monitored for 3 years to determine the water balance as hrfhrenced by vegetation. Cover types studied were honey mesquite (Prosopis glandulosa) plus herbaceous vegetation (M+H), mesquite removed leaving only herbaceous vegetation (H), and mesquite and herbaceous vegetation removed (BG). Throughout the study, BG lyshneters had greater soil water content than the vegetated sites but, regardless of cover type, only O&1.4% of precipitation drained below 3 m. Runoff and interrill erosion were closely associated with rainfall amount, peak short-term storm intensity, and amount of bare ground. Evapotranspiration accounted for over 95% of water leaving the vegetated sites. Herbaceous vegetation on the I-l lysimeters increased following mesquite removal. This increase offset any water yield benefit that may have accrued through shrub management. Results indicate that there is essentially no net change in deep drainage, evapotranspiration, or runoff on sites where the herbaceous component increases in response to shrub removal.

Research was initiated in August 1982 at the Texas Experhnental Ranch to evaluate effect of selected grazing treatments on watershed condition. Two production scale grazing treatments were sampled on 4 dates over a period of 15 months. Treatments were yearlong continuous grazing stocked at a moderate rate (MC) and a 16-paddock rotational grazing treatment stocked at a heavy rate (RG). In addition, hydrologic conditions in an ungrazed exclosure (EX) and a moderately stocked epasture, 3-herd deferred rotation treatment (DR) were examined during the summer of 1982. Regression analyses indicated infiltration rates increased and sediment production declined as vegetation standing crop and cover increased, soil bulk density decreased, and soil organic matter and aggregate stability increased. Averaged across the 4 sample dates, sediment production was least (33 kg/ha) and infiltration rate greatest (89 mm/hr) in the MC treatment as compared to the RG treatment (63 kg/ha and 82 mm/hr). Infiltration rates and sediment production in the RG and DR treatments before grazing were not significantly different from those in the MC treatment; however, grazing caused a significant decline in infiltration rates and a significant increase in sediment production in both treatments. Sediment production was least in the exclosure (23 kg/ha) while infiltration rates were generally greater and sediment production less in the midgrass communities as compared to the shortgrass communities. All effects were closely related to the effect of the various treatments on vegetation standing crop and cover.

Table 1. Streambank erosion observation periods and treatment identification. Streambank erosional patterns have been studied for 3 years (1 year of calibration and 2 years of active grazing treatment) on the Starkey Experimental Forest and Range in the Blue Mountains of northeastern Oregon. Livestock grazing use at the rate of 3.2 ha/AUM (8 acres/AUM) has not accelerated streambank degradation on Meadow Creek. Most erosion occurred during wintering periods and this erosion has been independent of grazing season treatments. It appears that high runoff and occasional ice flows are the most significant factors in bank cutting on this stream. Literature Cited Hayes, F.A. 1978. Streambank stability and meadow condition in relation to livestock grazing in mountain meadows of central Idaho. M.S. Thesis. Univ. of Idaho. Moscow, Id_91 p.

Termhtal infiltration rates were simiiar in soils on which a heavy whitebrush (Abysia lycioides) cover had been aerirny treated with 2 kg/ha (active ingredient) of 20% tebuthiuron pellets 4 years previously or on tebuthiuron-treated plots which had been prescribed burned the winter about 9 months prior to infiltration measurements, compared to untreated sites. However, sediment production was greater from plots treated with the herbicide than from areas subjected to the herbicide-fire system or from untreated plots. Only minor variations in infiltration rates occurred among sites originaiiy dominated by running mesquite (Prosopis reptuns) which were aerially sprayed with 2,4,5-T + picioram (1:l) at 1.1 kg/ha 3.5 years previously, burned 10 months previously, subjected to the herbicide-fire system or left untreated. However, sediment production on the running mesquite areas which had been sprayed tended to be greater than on untreated plots. Sediment production on areas subjected to the herbicide-prescribed burning system tended to be less than from brush-covered plots. Differences in sediment production In both experiments were generally attributed to siightly reduced mulch loads and mulch covers where the brush was removed as a leaf mulch donor by herbicide treatment. Prescribed burning apparently compensated for 10s~ of brush leaf mulch by promoting grass cover on herbicide-treated areas.