This experiment was conducted on the USDA Agricultural Research
Service Jornada Experimental Range near Las Cruces, New Mexico, USA
(http://jornada.nmsu.edu/). Climate is arid to semi-arid, with
long-term (90-year) mean annual temperature of 15C and mean annual
precipitation of 250 mm, over half of which occurs in summer
(July–September). The study was conducted in an upland grassland type
dominated by the perennial C4 grass black grama (Bouteloua eriopoda
Torr.). This grassland type has undergone transitions to C3 shrublands
dominated by honey mesquite shrubs (Prosopis glandulosa Torr.) and
eroding bare ground and is a classic case of feedback-driven
desertification.
The experiment was conducted in a level, uniform area of sandy loam
soils (Hueco series; coarse?loamy, mixed, superactive, thermic Argic
Petrocalcids) that is typical of sandy plains in the region. The
presence of a petrocalcic horizon imparts greater grassland resilience
compared to deeper soils, potentially contributing to the persistence
of grasses through historical periods of heavy grazing. Prior to the
experiment, cattle stocking rates were ca. 29 ha/animal unit month
(AUM) from 1986?1995, somewhat lighter than is typical for regional
grasslands (ca. 15 ha/AUM).
In 1995, eighteen 0.5-ha (70m x 70m) grazing paddocks were established
to include three blocks and six paddocks per block. Each paddock in
each block was assigned one of three grazing treatments (summer
grazed, winter grazed and non-grazed) and a shrub-removal treatment
(intact or removed), resulting in six treatment combinations per
block. Beginning in late spring 1996, all P. glandulosa shrubs in
‘removed’ treatments were clipped at the ground level, the vegetation
removed and the stumps painted with glyphosate (360 g/L) to prevent
resprouting. Removal of new plants or resprouts from underground was
maintained each year until 2000. Beginning in August 1996, grazing was
conducted in either August (summer) or February (winter) such that
65–80% removal of aboveground perennial grass biomass was achieved in
all parts of the paddock over a 24- to 36-h period, resulting in a
stubble height of < 5 cm. This level of biomass removal is over
twice that typically recommended, and is known to harm B. eriopoda
plants. Furthermore, this treatment was applied yearly from 1996 to
2000 such that each grazed paddock was defoliated four times in either
summer or winter, effectively compounding the damage. No livestock
grazing had occurred in the paddocks since spring 2000.
Pre-manipulation measurements of vegetation were gathered in spring
1996. Post-manipulation measurements were gathered in fall 2002
(measuring the effects of the grazing perturbation) and again in fall
2009 and 2016 (measuring the effects of the response). Line-point
intercept, following Herrick et al. 2005, was used to estimate basal
and foliar cover for each plant species in each paddock. In 1996,
2002, and 2009, measurements were gathered on seven, 70-m transects
spaced 10 m apart with 10-cm spacing between intercept points (4900
points per paddock). Analysis of these data revealed that less
sampling effort was required to detect treatment effects, so the 2016
sampling consisted of only 3 transects per paddock.
In the 1996 and 2002 samplings standing dead vegetation was recorded
as “D” and not identified to species. In the 2009 and 2016 samplings
standing dead vegetation was recorded to species.
