Stochastic and deterministic processes interact in the assembly of desert microbial communities on a global scale

Abstract

Extreme arid regions in the worlds' major deserts are typified by quartz pavement terrain. Cryptic hypolithic communities colonize the ventral surface of quartz rocks and this habitat is characterized by a relative lack of environmental and trophic complexity. Combined with readily identifiable major environmental stressors this provides a tractable model system for determining the relative role of stochastic and deterministic drivers in community assembly. Through analyzing an original, worldwide data set of 16S rRNA-gene defined bacterial communities from the most extreme deserts on the Earth, we show that functional assemblages within the communities were subject to different assembly influences. Null models applied to the photosynthetic assemblage revealed that stochastic processes exerted most effect on the assemblage, although the level of community dissimilarity varied between continents in a manner not always consistent with neutral models. The heterotrophic assemblages displayed signatures of niche processes across four continents, whereas in other cases they conformed to neutral predictions. Importantly, for continents where neutrality was either rejected or accepted, assembly drivers differed between the two functional groups. This study demonstrates that multi-trophic microbial systems may not be fully described by a single set of niche or neutral assembly rules and that stochasticity is likely a major determinant of such systems, with significant variation in the influence of these determinants on a global scale.

Figure 1

World map showing locations of desert sampling sites and number of samples recovered per location. Two non-arid locations (Chiang Mai, Thailand (warm tropical) and Yellowstone National Park, USA (cold temperate)) were also included for reference, but were not included in the analysis. Climate classification follows an updated Köppen–Geiger scheme (Peel et al., 2007) as follows, where MAT=mean annual temperature (°C) and MAP=mean annual precipitation (xP_threshold): BWh, hot arid desert (MAT>18, MAP <5); BWk, cold arid desert (MAT <18, MAP <5); ET, polar tundra (MAT <10); EF, polar frost (MAT <0); Dfa, cold winter (MAT_hot>10, MAT_cold <0). Deserts experiencing mean annual temperature <18 °C but hot diurnal and/or seasonal temperatures (Atacama, Taklimakan, Turpan) are categorized as warm arid deserts (BWw; MAT>10, MAP <5; Thomas, 1997). Quartz substrate was identified using USGS (The United States Geological Survey) mineral identification criteria (http://pasadena.wr.usgs.gov/office/given/geo1/).

Figure 2

Summary of desert hypolithic community structure, based upon environmental 16S rRNA gene sequences. Bracketed numbers show the range for phyla among all samples (N=92). All values shown are percentage of overall community.

Figure 3

Ordination plots of the redundancy analysis on the bacterial community matrix as dependent on continental location of each desert. In this case, the analysis represents the unique contribution of continent identity to (a) the entire bacterial community; (b) phototrophic bacteria; and (c) heterotrophic bacteria. The effect of continent was significant for the entire community only. Instead of plotting each sampling point, ellipses were drawn to represent the standard error around the continental centroids: 1, Africa; 2, Antarctica; 3, Arctic; 4, Asia; 5, Australia; 6, North America; 7, South America.