Response and resilience of soil biocrust bacterial communities to chronic physical disturbance in arid shrublands

Abstract

The impact of 10 years of annual foot trampling on soil biocrusts was examined in replicated field experiments at three cold desert sites of the Colorado Plateau, USA. Trampling detrimentally impacted lichens and mosses, and the keystone cyanobacterium, Microcoleus vaginatus, resulting in increased soil erosion and reduced C and N concentrations in surface soils. Trampled biocrusts contained approximately half as much extractable DNA and 20-52% less chlorophyll a when compared with intact biocrusts at each site. Two of the three sites also showed a decline in scytonemin-containing, diazotrophic cyanobacteria in trampled biocrusts. 16S rRNA gene sequence and terminal restriction fragment length polymorphism (T-RFLP) analyses of soil bacteria from untrampled and trampled biocrusts demonstrated a reduced proportion (23-65% reduction) of M. vaginatus and other Cyanobacteria in trampled plots. In parallel, other soil bacterial species that are natural residents of biocrusts, specifically members of the Actinobacteria, Chloroflexi and Bacteroidetes, became more readily detected in trampled than in untrampled biocrusts. Replicate 16S rRNA T-RFLP profiles from trampled biocrusts at all three sites contained significantly more fragments (n = 17) than those of untrampled biocrusts (n≤6) and exhibited much higher variability among field replicates, indicating transition to an unstable disturbed state. Despite the dramatic negative impacts of trampling on biocrust physical structure and composition, M. vaginatus could still be detected in surface soils after 10 years of annual trampling, suggesting the potential for biocrust re-formation over time. Physical damage of biocrusts, in concert with changing temperature and precipitation patterns, has potential to alter performance of dryland ecosystems for decades.

Figure 1

Relative abundance of (a) DNA, (b) chlorophyll a, and (c) scytonemin extracted from trampled and untrampled plots sampled 1 year after trampling. Values are means (s.e.) for n=5 samples per treatment. Different letters denote a significant effect of trampling between pairs of untreated and trampled soils within each field site (two-tailed pair-wise t-test P<0.05 or smaller). The ^* denotes pairs that were significantly different by one-tailed pair-wise t-test at P<0.10.

Figure 2

Major bacterial phyla in 16S rRNA gene clone/sequence libraries from untrampled biocrusts (0–2 cm depth), trampled soils (0–2 cm depth) and sub-biocrust soils (5 cm depth). Sequences from clone libraries generated from all three field sites (Arches, ISKY-1 and ISKY-2) were pooled to generate the bars depicted in the figure. The untrampled biocrust bar depicts the pooled composition of sequences from six clone libraries (two field replicates for each site, total 385 sequences). Trampled biocrust and sub-biocrust soil bars each represent the pooled sequence composition of three clone libraries (one field replicate for each site, total 180 and 220 sequences, respectively). Individual clone library results are shown in Supplementary Figure 1.

Figure 3

T-RFLP profiles of biocrusts from the Arches site. (a) Untrampled biocrust, trampled biocrust collected (b) 1 year after trampling or (c) trampled biocrust collected 1 month after trampling. The dominant M. vaginatus peak is visible in the profiles at ∼420 bp in length.

Figure 4

Nonmetric multidimentional scaling plots derived from Manhattan distance matrices of 16S rRNA T-RFLP profiles show the uniformity of the bacterial community composition in replicate field samples from untrampled biocrusts (circles) at three different field sites (labeled within plot boxes). Bacterial community composition in trampled plots differed dramatically from the untrampled plots 1 month (triangles) or 1 year (diamonds) after a trampling event, and was highly variable across the field replicates.