We analyzed 15 cases from 3 sources in the NCPN database, primarily in and around Capitol Reef National Park. Shane Green of the NRCS kindly provided some vegetative composition data from an additional 14 sites that were on file, the location of these was not documented in the data files.
Caveats and alternative explanations
In general, our observed clusters seem to provide a reasonable validation of the NRCS ecological site description. We see a gradient of communities dominated by putative decreaser species identified in the description, grade into ones dominated by increaser species, which in turn grades into states dominated by exotic invasive grasses. Tp our knowledge, all of our sites are currently in active grazing allotments and have likely been impacted by livestock in recent years; from these results we would infer that the more strongly impacted cluster had experienced more grazing activity.
We cannot rule out observer bias in generating these differences. The annualized cluster is exclusively from data provided by the NRCS. The less impacted cluster are primarily sites sampled by the NPS Inventory and Monitoring Program, whereas the other cluster consists of a mixture of sites from the Capitol Reef vegetation mapping project, and from the Grand Staircase-Escalante National Monument Rangeland Health Assessment. Biases could be present resulting from different personnel collecting data, or the more southerly distribution of the samples in the more impacted cluster.
Cluster analysis
We subjected all available data on plant community composition to hierarchical cluster analysis to determine if clusters could validate the existence of a priori states and phases. All species with fewer than 3 occurrences were excluded from the analysis. We applied to relativization by row maximum transformation to focus the analysis on relative species abundance. A double relativization to further standardize data was not appropriate, because the NRCS data did not provide absolute abundance of species, only relative abundance. Based on a rather simple apriori concept of a state with 2 phases (one dominated by grazing decreasers, and one dominated by grazing increasers), and an additional devegetated state subject to erosion, we examined 2,3 and 4 cluster solutions. All of these were "clean" clusters with little overlap and all could be potentially be explained by a hypothesized unquantified grazing gradient (from lower right to upper left in Fig. 1).
Fig. 1. NMDS ordination diagram depicting results of cluster analysis. Clusters do not overlap, and seem to correspond to a plausible degradation sequence (Click image to enlarge).
Based upon the four cluster solution, we determined the degree of correlation of various species with the clusters. We found that there were clusters indicated by: 1. Atriplex canescens, 2. Atriplex confertifolia, 3. Pleuraphis jamesii, and Bromus tectorum (Fig 2).
Fig. 2. Image of the same ordination as above, with symbols resized to reflect the relative abundance of key species (click image to enlarge).
The clusters depicted in panels a - c seem to correspond well to NRCS concepts of vegetation changes under grazing pressure (loss of decreasers, expansion of increasers). All three of these could be interpreted as different phases of the same state; because of the shift in growth form dominance we interpreted the perrenial grass rich cluster (open circles) as its own state. We were surprised to also detect a Bromus tectorum-dominated cluster, which we deemed functionally distinct enough to be its own state, and due to floristic similarities and closeness in ordination space hypothesized that it can arise from the P. jamesii-dominated cluster. The resultant state-and-transition model can be viewed here.
No comments:
Post a Comment