Ponderosa Pine Understory Response to Short-Term Grazing Exclusion (Arizona)

Livestock grazing is a pervasive land use practice on public lands in the American Southwest, making it an important variable in land management practices. While most grazing allotments are stocked each year, short rest periods (1–2 years) are used to minimize potential negative impacts of grazing and to sustain range resources (Curtin 2002). Moore and others (1999) specifically recommend deferment of livestock grazing on ponderosa pine (Pinus ponderosa) forests after thinning and burning as ecological restoration treatments. Few data are available, however, on the influence of short-term grazing exclusion on the understory plant community of pine forest restoration projects. In this study, we attempt to quantify the impacts of cattle grazing on plant cover, species richness, and non-native species on a ponderosa pine restoration project in northern Arizona.

In the summer of 2005, the Arizona Game and Fish Department erected several 35 m × 35 m cattle exclosures at the Mt. Trumbull, Arizona, ecological restoration research site on the Grand Canyon–Parashant National Monument. We used these exclosures to evaluate the short-term effects of cattle exclusion on the herbaceous understory in a ponderosa pine forest. We established ten paired plots consisting of a grazing treatment exclosure and a similarly sized grazed control. Of the ten pairs, five were located in a thinned and burned forest and five in unrestored areas. The plots were dispersed across the 1,200 ha Little Spring pasture of the Mt. Logan allotment, where elevations range from 2,000 to 2,250 m, and the mean annual precipitation from 2002 to 2007 was 413 mm (WRCC 2007). Cattle grazing occurs on the Mt. Logan allotment only from July through October at a maximum stocking rate of 13.6 ha/AUM (Whit Bunting, BLM, pers. comm.). In 2006, however, grazing pressure was reduced by a break in a fence that allowed cattle to stray into the adjacent Tuweap allotment.

Figure 1.

Median (25th, 75th percentiles) cheatgrass cover per plot, separated by both grazing and restoration treatments. Results of Kruskal-Wallis ranked signs analysis were significant for year × grazing × restoration (treated vs. control) interaction (χ² = 32.01, p = 0.001).

The unrestored sites within the project are predominantly dense stands of ponderosa pine with thick litter and duff layers and a sparse understory. The restored areas were thinned and burned to emulate pre-1870 tree density and distribution. Restoration treatment areas have a more diverse and abundant understory community and substantially fewer trees per hectare (399.2) than the prerestoration conditions (784.6) and the untreated controls (873.5). All restoration treatments on our study plots were completed by 2001 (see Roccaforte et al. 2009 for details), and cattle were reintroduced in the summer of 2002 (Whit Bunting, BLM, pers. comm.).

The dominant tree species is ponderosa pine, though there is a considerable amount of Gambel oak (Quercus gambelii) on site. The main native species are silver lupine (Lupinus argenteus) and the important perennial forage grasses bottlebrush squirreltail (Elymus elymoides) and western wheatgrass (Pascopyrum smithii). In 2003, cheatgrass (Bromus tectorum) invaded the site. Cheatgrass is the most abundant non-native species present, and this annual grass dominates much of the understory so that cheatgrass cover is typically greater than the cover of all native herbaceous species combined (McGlone et al. 2009).

Table 1.

Mean (± SE) or median (25th, 75th percentiles) values for aerial cover, species richness, and diversity in grazed versus ungrazed plots. All data are pooled across the restoration treatments. None of the results were significant except the year × grazing × restoration treatment interaction in cheatgrass cover (see Figure 1).

Vegetation was sampled annually in early July. Within each plot, we established a 25 m × 25 m study area that contained four 25 m modified Daubenmire transects along which we established ten permanent 1 m² quadrats. In each quadrat, we visually estimated aerial cover for each species. Additionally, we used the area between transects as a 5 m × 25 m belt transect to generate a list of all...