Appendix A. Detailed methods for primary production and soil sampling..
Methods for estimating production of forbs and annuals
At each site, we laid out three transects with five sampling stations distributed systematically on each transect. All three transects fit within the 10 × 10 m temporary exclosures. A 0.5 × 0.5 m square quadrat was placed over each sampling station on each transect and all current year’s aboveground growth of annuals and perennial forbs harvested. The five samples on each transect were pooled, yielding 3 replicates representing an area of 1.25 m2 each (only four sampling stations per transect were harvested at the subhumid study areas in each region). Biomass samples were dried for 48 hrs at 55°C and then weighed.
Methods for estimating perennial graminoid production
We developed a two-step procedure to estimate perennial grass production in order to dampen variability created by spatial heterogeneity. First, we measured the basal cover of all perennial grass species using 3 40-m line transects laid out in parallel around each 10 × 10 m temporary exclosure. Second, we harvested individual plants of all common perennial grass species (2 to 5 species per site) within the temporary ungrazed exclosure as follows: we harvested the two individuals of each species located nearest to each sampling station, recording the longest basal diameter and a second, perpendicular diameter of each plant in order to calculate basal cover using the formula for an ellipse. The ten plants on each harvesting transect were pooled, yielding 3 replicates representing 10 plants each. These samples were weighed, subsampled by 20%, sorted into live, recently senesced, and dead components, and then the sorted samples were dried and weighed. The proportion live plus recently senesced was used to estimate current year’s production for the original, dry-weight corrected sample (Lauenroth et al. 1986). Using the basal cover data from the harvested plants, we calculated the production per unit basal cover of each species. We then used the basal cover data from the 40-m line transects to scale production per unit basal area to production per m2 for each species. The main assumption of this method is that the form and vigor of plants on the longer 40-m transects is identical to plants located within the 10 × 10 m exclosure. This assumption should be valid given our effort to select sites that were homogeneous at this scale. Production estimates using these methods fall within historical ranges (Jobbágy and Sala 2000) and, based on a Monte Carlo simulation, reduce variability in estimates of the mean compared to the standard method (Adler 2003).
Methods for estimating shrub production
Fernandez et al. (1991) estimated
shrub production at the arid PAT site based on a relationship between the aboveground
production of a particular species, ANPPshrub, and two predictor
variables: (1) the surface area of the shrub assuming a perfect hemispheric
shape, SA, which is calculated from height and diameter measurements,
and (2) the production of leaves and twigs within a small quadrat centered on
the top of the shrub, the production per plant area or PPA. Then,
ANPPshrub = k* SA* PPA where k is the
regression coefficient, estimated by complete harvest of a number of individual
shrubs. In order to estimate total production per m2 of this
species, ANPPtot, we used belt transects to measure mean shrub
surface area, 
, and shrub density,
D. In addition, we harvested small quadrats at the center of a number
of randomly selected shrubs to determine mean production per plant area, 
.
Then, ANPPtot = k* D**,
where k is the regression coefficient determined above.
For the arid PAT site, we used
regression coefficients for the three dominant shrub species estimated by Fernández
et al. (1991). We estimated both D and using
3 40 × 2 m belt transects at each site, and was
estimated by harvesting 25 × 10 cm quadrats on 5 shrubs of each
species (if present) within the temporary exclosures. At the semiarid and
subhumid study areas, where the growth form of the dominant shrub M. spinosum
appeared to differ from the arid study area, we recalibrated the relationship
by complete harvesting of 25 randomly selected shrubs. In the arid and
semiarid SGBR study areas, we calibrated the relationship for the dominant shrub
A. tridentata, harvesting 30 individual shrubs. For subdominant
shrub species at the SGBR study areas, we calibrated a relationship based only
on shrub density and mean surface area: ANPPtot = k*
D*. At all SGBR sites,
we used 3 40 × 1 m belt transects to calculate D and
at each site, and 25 × 25 cm quadrats to determine. We
tested each relationship using different methods of calculating SA (such
as averaging the diameters and height measurements, or using only the diameter
mean) and chose the one that minimized least square errors. All analyses used
oven-dried weights. Note that production contributed by rare shrub species
went unrecorded, but based on the data from the subdominant species we measured,
this error was less than 1–2 g·m-2·yr-1.
Our recalibration of the relationship ANPPshrub = k* SA* PPA for M. spinosum at the PAT semiarid and subhumid study areas produced almost the same value of k (0.377 instead of 0.37) found by Fernández et al. (1991) for M. spinosum at the arid PAT site. The calculation of surface area (SA) was based only on shrub diameters, ignoring height. This relationship explained 93.5% of variation about the origin in individual shrub biomass (n = 25). The calibration of this relationship for A. tridentata at the SBGR study areas also produced a good fit (Table A1), with an r2 of 0.898. Relationships based only on shrub size for A. tripartita and Chrysothamnus viscidiflorus explained over 90% of variation about the origin, while the relationship for the smaller shrub Eriogonum microthecum explained 84% of variation (Table A1).
Table A1. Models used to estimate shrub leaf and fine twig production per m2 in sagebrush steppe.
Species
Model
k
r2
n
Artemisia tridentata v. tridentata
0.2480
0.898
30
Artemisia tripartita
0.0086
0.915
30
Chrysothamnus viscidiflorus
0.0052
0.956
25
Eriogonum microthecum
0.0027
0.844
25
Notes: Shrub surface area was calculating using the formula for a hemisphere (2
r2). “SAd” denotes that r was based on two shrub diameter measurements, while r used in “SAdh” was the mean of diameter and height measurements. Density “D” is shrubs/m2, and “G” is the production per plant unit surface area (g/cm2). All regressions were forced through the origin. “k” is the regression coefficient, and “r2” measures the proportion of variability in observed production about the origin explained by regression.
Methods for estimating total soil C and N
At each site, we took 5 randomly located soil cores from each of three microsites: bare soil, under grass, and under shrub. The depth of cores beneath grasses and shrubs was variable to accommodate the accumulation of material underneath plants (Burke et al. 1998). To measure the height of these plant “mounds” we used a carpenter’s level to run a string horizontally from the center of the plant to the nearest patch of bare ground, and recorded the height of the string above the ground. Total core depth thus equaled 5 cm (the below-mound component) plus the height of the plant mound. Mound heights were easily measured in PAT, where ground was level, but much harder to measure in SGBR, where the siltier soils preserved greater microtopography and slopes were steeper. Whenever mound height could not be measured with confidence, we cored to 5 cm. Soil cores were immediately air-dried.
In the laboratory, we used a 2-mm sieve to separate gravel from soil and weighed each of these components. We determined both total and gravel-free bulk density by averaging the soil mass divided by the core volume of the 5 bare ground samples from each site. Subsamples of all soil samples were oven-dried for 24 hours at 100°C to determine moisture content. We aggregated the 5 samples from each microsite (bare, grass, shrub) by combining 25% subsamples of each replicate’s moisture corrected dry mass. The aggregated bare ground samples were then used to determine soil texture by the hydrometer method (Gee and Bauder 1986). After grinding subsamples of the aggregated samples in a roller mill, we determined total C and N concentration by combustion using a LECO CHN-1000 analyzer (St. Joseph, Michigan, USA).
We scaled the C and N concentration data up to total C and N expressed per unit area (g/m2) by first multiplying total C and N concentrations by fine soil bulk density to estimate total C and N per core. We divided this value by the surface area of the core and then multiplied by the proportion of ground covered by the corresponding microsite, determined by canopy cover estimates (bare ground = 1 – grass – shrub). Total C and N (g/m2) is the sum of the bare, grass, and shrub fractions.
Adler, P. B. 2003. A comparison of livestock grazing effects of sagebrush steppe, USA and Patagonian steppe, Argentina. Dissertation. Colorado State University, Fort Collins, Colorado, USA.
Burke, I. C., W. K. Lauenroth, M. A.Vinton, P. B. Hook, R. H. Kelly, H. E. Epstein, M. R. Aguiar, M. D. Robles, M. O. Aguilera, K. L. Murphy, and R. A. Gill. 1998. Plant-soil interactions in temperate grasslands. Biogeochemistry 42:121–143.
Fernández, R. J., O. E. Sala, and R. A. Golluscio. 1991. Woody and herbaceous aboveground production of a Patagonian steppe. Journal of Range Management 44:434–436.
Gee, G. W., and J. W. Bauder. 1986. Particle size analysis. Pages 383–411 in A. Klute, editor. Methods of Soil Analysis Part I: Physical and Mineralogical Methods. Madison, Wisconsin: Agronomy Society of America and Soil Science Society of America.
Jobbagy, E. G., and O. E. Sala. 2000. Controls of grass and shrub aboveground production in the Patagonian steppe. Ecological Applications 10:541–549.
Lauenroth, W. K., H. W. Hunt, D. M. Swift, and J. S. Singh. 1986. Estimating aboveground net primary production in grasslands: a simulation approach. Ecological Modeling 33:297–314.
