Desert Sand - - rough draft results

We identified: 1) a “reference grassland” group with the indicators (Indicator value >30, corresponding to P ≤ 0.10) Achnatherum hymenoides, Sporobolus spp., and Hesperostipa comata., 2) A “sand shrub community” with the indicators Artemisia filifolia, Bouteloua gracilis, and Poliomintha incana, 3) An “annual-invaded shrubland” with the indicators Coleogyne ramosissima, Psorothamnus fremontii, and Bromus tectorum, and 4) A “coppiced system” with the indicators Ephedra spp., Artemisia filifolia, and Opuntia spp (Fig 1, Fig 2). These four groups provide a reasonable degree of confirmation of our apriori expectation of a three state model (with four phases), and confirms the existence of our conception of the reference state and the coppiced state (Fig 2). A new state was identified however, the invaded shrubland. This state was defined by dominance by Coleogyne ramosissima or Psorothamnus montanum (in drier sites), and the presence of Bromus tectorum. Coleogyne may facilitate Bromus invasion by providing climatic buffering and litter resources (ref.). We did not find evidence in our dataset of an annualized state, dominated by Bromus or Salsola spp., but cannot rule out that it exists. One interpretational caveat is that there was a strong correlation between the data source and cluster membership. The most extreme example was that over 70% of the reference grassland sites were from the Miller et al. Canyonlands data, whereas the other groups were reasonably well spread across the various data sources. The Canyonlands data was primarily gathered within a national park which has been ungrazed for 40 years, a degree of rest not matched in any other datasource. Nevertheless, we used these clusters as a reasonable approximation of state and phase membership.





























[ maybe i'll change this silly color scheme for some error bars instead ]

The set of four community-structure based states and phases corresponded well to several functional attributes of the ecosystem (Fig 3). The variance explained in our ANOVA models corresponded well to the hypothesized causal sequence, with more proximate causes exhibiting a higher R², and ultimate causes lower R² values: BSC cover, Plant cover < style=""> The reference state (RG and SS) retained 2-4 × as much biological crust cover, and consequently much higher soil aggregate stability. Somewhat surprisingly, all of the shrub-dominated states or phases had greater total perennial vegetative cover than the reference grassland, although it should be noted that grassland cover varies from year to year. The sand shrub phase of the reference state retained a relatively low shrub:grass ratio, whereas the coppiced and invaded shrub communities were almost exclusively woody-dominated. Perhaps most instructive were differences in the gap size distribution (R² = 0.49 – 0.66). The reference grassland was characterized by 30 – 50% shorter gap length, and about twice as many gaps per meter. The difference in the shape parameter of the gamma distribution illustrated this contrast most clearly, indicating a much stronger positive skew in the gap size distribution. The invaded shrub state also differed from the coppice state in this functional attribute reinforcing that it is a distinct state; in general the invaded state was characterized by a larger number of smaller gap sizes. Data was not available to examine these properties adequately in the sand shrub phase.

Our logistic regressions provided the values of several of these predictors at which transition form reference state to the coppiced state are 25%, 50% and 100% probable (Table 2). The solutions for mean gap length and k failed to converge, likely because of the large magnitude of the difference between reference and coppiced samples. However they did produce reasonable values which should be interpreted cautiously.