This study took place at seven rangeland locations in the Foothills, Parkland, Dry Mixedwood, Foothills Fescue, Dry Mixedgrass and Mixedgrass Prairies. Research sites were set up at Kinsella, Mattheis, Onefour, Oyen, Sangudo, Stavely, and Twin River (Figure 1).
Field methods.
Figure 2. Diagram of experimental site layout and plot layouts. All experimental sites have a grid layout of 40 plots (except Kinsella and Mattheis, which have a grid layout of 50 plots). Sites are fenced to exclude grazing. Plot design is nested, consisting of a 2.5 x 2.5 m outer plot, and a 0.5 x 0.5 m inner plot. Biocrust measurements are completed on the inner plot, and soil and vegetation measurements are completed on the outer plot.
At all sites, biocrust composition and cover were measured at 40 plots (50 cm x 50 cm), except at the Kinsella and Mattheis sites, where 50 plots were established (Figure 2). Experimental plots are subjected to drought and defoliation treatments (Figure 3) in a fully factorial replicated design (ambient moisture versus 40% reduction; 5 grazing treatments) with 4 duplicates (5 duplicates at Kinsella and Mattheis). Rainout shelters were constructed over drought-simulated plots to reach a 40% reduction in moisture. The following defoliation treatments were applied to simulate grazing management practices:
Continuous grazing, simulated by clipping vegetation to a 3 cm height in June and September
Adaptive grazing, simulated by clipping to 7 cm height in June, and clipping to 3 cm height in September
Early spring grazing, simulated by clipping to a 3 cm height in June
Deferred grazing, simulated by clipping to a 3 cm height in September
Control, simulation of ungrazed pasture, no clipping treatment applied
Figure 3. Experimental plots were subjected to drought and defoliation treatments in a fully factorial replicated design: ambient moisture versus 40% reduction; 5 grazing treatments, with 4-5 duplicates. Rainout shelters were constructed over drought-simulated plots to reach a 40% reduction in moisture. Defoliation treatments consisted of an ungrazed control, spring grazed, fall grazed, adaptive grazing, and continuous grazing.
Lichen cover was measured with a ruler and recorded as cm2 area measurements to avoid subjectivity with percent cover visual assessments. Percent cover of vascular plant vegetation was assessed in the 2.5 x 2.5 m plot surrounding the lichen cover plots. In addition, environmental factors including soil moisture and soil temperature were measured. Soil samples were collected from the plots and analyzed for soil texture, soil pH, soil total nitrogen, soil total carbon, and soil bulk density.
Data analysis.
Nonmetric Multidimensional Scaling (NMDS) Nonmetric multidimensional scaling (NMDS) was used to evaluate the effect of drought and grazing treatments on biocrust community composition. NMDS is an ordination method that works well with data that is normally or non-normally distributed, and with data with multiple zeros. NMDS is a rank-based approach that computes a dissimilarity matrix (Bray-Curtis, in this case), and ordinates plots in space using pairwise dissimilarity betweeen ranked points (ie. the plots). Plots similar in species composition are ordinated closer together and plots that are dissimilar are ordinated further apart. Stress values produced by NMDS indicate how well distances in the ordination correspond to the dissimilarity values. NMDS was completed using the Vegan package in R (Oksanen 2015).
perMANOVA and pairwise perMANOVA PerMANOVA in the Vegan package (Adonis) was used to analyze differences between treatment groups (Oksanen 2015). In the perMANOVA analysis, the centroids for each group are found and then the squared deviations of each of site to its group centroid are calculated. Significance tests are performed using F-tests based on sequential sums of squares from raw data permutations. PerMANOVA tests in Adonis were followed by pairwise perMANOVAs to look at pairwise comparisons adjusted for multiple inference. Pairwise perMANOVAs were completed using the RVAideMemoire package (Hervé 2019).
Indirect Gradient Analysis Indirect gradient analysis was used to explore the relationships of treatment groups with environmental variables and univariate responses (biocrust cover, biocrust species richness, lichen cover, lichen species richness, and Shannon’s diversity). Indirect gradient analysis uses an NMDS ordination of the community data, and then variables are correlated to ordination scores. The direction of the arrow indicates the direction of the most rapid change in the variable (ie. the gradient), and the length of the arrow is indicative of the strength of the correlation between the ordination and environmental variable (Oksanen 2015).
Species Preferences Species preferences and sensitivities to treatment groups were analyzed using correlation indexes in the Indicspecies package (De C´aceres 2013). Species associations between treatment groups are calculated using point biserial correlation coefficients. Positive values close to 1 indicate an affinity for an environmental condition, whereas negative values of the index express that a species tends to ’avoid’ particular environmental conditions.
Disclaimer: This website was created for a class project for RENR 690 at the University of Alberta. Results presented are based on a randomly generated dataset.