Microbial Ecology

, Volume 63, Issue 3, pp 694–700

The Effect of Resource Islands on Abundance and Diversity of Bacteria in Arid Soils


  • Ami Bachar
    • Zuckerberg Institute for Water Research, The Jacob Blaustein Institutes for Desert ResearchBen-Gurion University of the Negev
  • M. Ines M. Soares
    • Zuckerberg Institute for Water Research, The Jacob Blaustein Institutes for Desert ResearchBen-Gurion University of the Negev
    • Zuckerberg Institute for Water Research, The Jacob Blaustein Institutes for Desert ResearchBen-Gurion University of the Negev
Soil Microbiology

DOI: 10.1007/s00248-011-9957-x

Cite this article as:
Bachar, A., Soares, M.I.M. & Gillor, O. Microb Ecol (2012) 63: 694. doi:10.1007/s00248-011-9957-x


Bacteria and nutrients were determined in upper soil samples collected underneath and between canopies of the dominant perennial in each of three sites along a steep precipitation gradient ranging from the Negev desert in the south of Israel to a Mediterranean forest in the north. Bacterial abundance, monitored by phospholipid fatty acid analysis, was significantly higher under the shrub canopy (compared to barren soils) in the arid and semi-arid sites but not in the Mediterranean soils. Bacterial community composition, determined using terminal restriction fragment length polymorphism and clone libraries, differed according to the sample’s origin. Closer examination revealed that in the arid and semi-arid sites, α-Proteobacteria are more abundant under the shrub canopy, while barren soils are characterized by a higher abundance of Actinobacteria. The bacterial communities in the Mediterranean soils were similar in both patch types. These results correspond to the hypothesis of “resource islands”, suggesting that shrub canopies provide a resource haven in low-resource landscapes. Yet, a survey of the physicochemical parameters of inter- and under-shrub soils could not attribute the changes in bacterial diversity to soil moisture, organic matter, or essential macronutrients. We suggest that in the nutrient-poor soils of the arid and semi-arid sites, bacteria occupying the soil under the shrub canopy may have longer growth periods under favorable conditions, resulting in their increased biomass and altered community composition.


Plant communities in resource-limited environments, such as desert soil, are shaped by harsh and unstable environmental conditions. Underneath the canopy of perennials in these regions, soil temperature and evaporation are both moderated and limiting nutrients might be more abundant, creating patches of higher fertility in the low-productivity landscape [1, 2]. These patches are referred to as “resource islands” [3] and have been found in diverse ecosystems, including sagebrush steppe [4], arid woodlands [5], coastal dunes [6], chaparral [7], cold and warm deserts [3, 8], and pine forests [9]. In deserts, these patchy habitats are created by shrub vegetation and may determine the distribution of annuals [5], invertebrates [10, 11], and vertebrates [12, 13].

In arid and semi-arid soils, the microbial community and its associated biogeochemical transformations may reflect the sparse distribution of the shrub vegetation. In arid lands in Australia, resource accumulation and milder climatic conditions in plant-covered patches increase the abundance and richness of protozoan species [14]. Bacteria have been found to be more abundant beneath shrubs in the desert soils of the Chilean coast [15], the Sonora [16], and the Negev [17]. In semi-arid Mediterranean [18] and heather moorland [19] soils, plant patches have been found to enhance microbe-driven processes, showing bursts of microbial activity under the plant canopy. Indeed, the distribution of microorganisms in a patchy landscape was found to be closely related to the concentrations of available resources, resulting in higher microbial biomass and nutrient mobilization [3, 18].

Bacterial diversity has been found to decrease in non-vegetated inter-shrub patches relative to under the shrub canopy in the Great-Basin [20] and Negev [21] deserts. In semi-arid regions, vegetation was shown to produce environmental gradients that exhibit a positive correlation to microbial abundance and diversity. In more moderate environments, where water is not a limiting factor, the physical gradient from the vegetation is reduced [22]. However, no study has yet compared the diversity and community composition of soil bacteria in patches of plant-covered and barren matrices along a low-precipitation gradient. One might hypothesize that bacterial diversity in shrubland ecosystems will be patchy, mirroring the plant matrix, while in forest ecosystems, bacterial diversity will be more evenly distributed. Soil bacterial abundance and community composition will vary around plant clusters compared to inter-plant patches and, accordingly, an increase in vegetation cover will result in a decrease in bacterial patchiness. To test this hypothesis, we explored how bacterial abundance and community composition reflect vegetation patchiness in distinct climatic regions. We examined local-scale microbial diversity in patch types consisting of bare soil and soil under the canopy of the predominant shrub at geographical sites ranging from the Negev desert in the south to the Mount Carmel Mediterranean forests in the north of Israel.

Materials and Methods

Site Description and Sampling

The study was conducted at three long-term ecological research (LTER) sites in Israel (http://lter.bgu.ac.il/sites/rhd.aspx) representing three climatic regions: arid, semi-arid, and Mediterranean (average annual precipitation of 100, 300, and 600 mm, respectively). Sampling took place in May 2007 as previously described [21, 23]. Briefly, sampling was performed in triplicate plots (40 × 25 m) at each site (Supplementary Table 1), all fenced and thus protected from grazing livestock and undisturbed by human activity. At each plot, eight randomly selected subsamples were taken from the under the canopy and inter-shrub patches, i.e., between the perennial plants and under the dominant perennial canopy (Supplementary Table 1). The eight subsamples of each patch type at each plot were composited to represent an average for that site, resulting in a total of six composite soil samples per site. Following crust and litter removal, the top 5 cm of the soil was collected into sterile Whirl-Pak sample bags (Nasco, Fort Atkinson, WI, USA) and placed in a cooler. The samples were transported to the laboratory and homogenized within 24 h of sampling. A 50-g subsample of each soil sample was stored at −80°C for molecular analysis while the rest was used for physicochemical analysis.

Soil Physicochemical Analyses

Soil analyses were conducted according to standard protocols [24]. Water content was determined by gravimetric method; organic matter content by dichromate oxidation method; pH and electrical conductivity (EC) in saturated soil extract (SSE), sodium, calcium, potassium, and magnesium in SSE by inductively coupled plasma spectroscopy and flame photometer; phosphate by the “Olsen method” (in sodium bicarbonate extract); and nitrogen as nitrate in aqueous extract, nitrogen as ammonium in KCl solution extract (including adsorbed nitrogen), and calcium carbonate content by HCl digestion.

Analysis of Phospholipid Fatty Acids (PLFAs)

Analyses of bacterial PLFAs were conducted on 4 g of soil homogenate sampled from each patch in three replicate plots for each site. The soil PLFAs were extracted and analyzed as previously described [23].

DNA Extraction

Each soil homogenate was extracted three times for DNA using the PowerSoil™ DNA Isolation Kit (MoBio, West Carlsbad, CA, USA) according to the manufacturer’s recommendations. The three DNA extracts were then pooled and measured spectrophotometrically, verified on an ethidium bromide-stained agarose gel and used for molecular analysis, as described below. All products were stored at −20°C until use.

Terminal Restriction Fragment Length Polymorphism (T-RFLP) Analyses

The DNA extracted from soils sampled from each patch in three replicate plots of each site was subjected to bacterial fingerprint analysis using T-RFLP as previously described [21].

Clone Library Construction and Analysis

DNA was extracted from one soil homogenate of each patch type at each of the LTER sites, as described above, and further subjected to clone library construction and analysis as previously described [23]. The clone libraries of the 16S rRNA genes amplified from DNA extracted from each homogenate yielded about 290 true clones, except for the clone library amplified from the soil sample taken from under the canopy in the Mediterranean site, which amounted to 213 clones. Rarefaction analysis suggested that over 400 clones are needed to yield the minimal bacterial estimation and that the number of clones per sample used in this study was not enough to reach saturation (Supplementary Fig. 1). We thus regard our cloning efforts as minimal values.

Statistical Analysis

The TRFLP patterns of each sample were standardized and analyzed as previously described [21]. Briefly, the profiles were aligned and a consensus profile was computed for each sample by eliminating non-reproducible peaks and averaging shared peaks. All standardization and normalization procedures were performed using MATLAB cluster and distribution analyses. Hypothesis testing was performed using block-design redundancy analysis (RDA) and tested using Monte Carlo permutation tests. Differences in PLFA and clones between soils sampled from under the canopy and inter-shrub patches were analyzed using a paired t test.

Accession Numbers

Sequence data generated in this study were deposited in GenBank under accession numbers JF295130 to JF295976.


Soil Physicochemical Characteristics

Table 1 summarizes the physicochemical characteristics of the soils sampled in this study. In the arid site, the soil water content showed similar values for the two patch types, probably because the scanty shrub canopy could not prevent evaporation. Calcium carbonate and phosphate concentrations, as well as pH and organic matter values, were also unchanged between barren soils and under-shrub patches. Yet, nitrate, sodium, magnesium, and potassium concentrations were almost three times higher in the soils collected from under the canopy than in the inter-shrub patches; the salinity measurements (EC) as well as the calcium and ammonium concentrations were almost doubled.
Table 1

Soil chemical characteristics in patches under shrub canopy (US) and in inter-shrub (IS) patches in arid, semi-arid, and Mediterranean sites


Patch type


Na (mg/kg)

Ca (mg/kg)

Mg (mg/kg)

NO3–N (mg/kg)

NH4–N (mg/kg)

P (mg/kg)

K (mg/kg)

ECa (dS/m)

CaCO3 (%)

Organic matter (%)

Water content (%)



8.09 ± 0.01

15.49 ± 6.34

20.14 ± 2.06

2.31 ± 0.65

3.20 ± 0.86

3.86 ± 0.11

0.03 ± 0.00

6.98 ± 6.34

0.65 ± 0.07

35.50 ± 2.12

0.75 ± 0.26

1.85 ± 0.28


7.99 ± 0.06

40.22 ± 8.31

35.36 ± 11.61

8.39 ± 4.60

9.37 ± 2.27

6.25 ± 3.59

0.04 ± 0.01

19.11 ± 5.91

1.24 ± 0.21

33.67 ± 4.51

0.79 ± 0.35

1.72 ± 0.31



7.18 ± 0.07

18.79 ± 6.26

128.98 ± 13.21

19.48 ± 2.28

2.82 ± 0.33

49.02 ± 7.48

0.07 ± 0.01

7.00 ± 0.85

0.65 ± 0.07

17.00 ± 2.65

1.99 ± 0.29

2.54 ± 0.44


6.97 ± 0.24

19.32 ± 3.45

263 ± 73.02

44.46 ± 10.44

3.47 ± 0.23

62.52 ± 8.70

0.09 ± 0.03

14.08 ± 1.55

2.23 ± 0.43

16.3 ± 2.89

3.39 ± 0.44

3.68 ± 0.41



6.95 ± 0.20

13.73 ± 3.60

142.9 ± 64.84

29.49 ± 8.44

6.72 ± 1.47

40.82 ± 10.05

0.09 ± 0.02

9.92 ± 3.26

1.02 ± 0.27

1.67 ± 1.15

4.93 ± 1.40

8.71 ± 3.43


7.05 ± 0.09

45.82 ± 16.38

325.0 ± 186.00

70.00 ± 25.81

6.52 ± 0.22

51.0 ± 10.88

0.02 ± 0.01

34.52 ± 15.10

1.83 ± 0.58

1.67 ± 1.15

10.53 ± 3.43

12.78 ± 1.84

aEC electrical conductivity

In the semi-arid site, soil water content and organic matter were higher in the under the shrub canopy patches compared to barren soils. The concentrations of calcium carbonate, phosphate, and sodium, and the pH values, did not change between inter- and under-shrub patches. The EC and the concentration of magnesium were always higher under the shrub canopy, while ammonium, nitrate and potassium concentrations, although higher in the inter-shrub patch, were not significantly so.

In the Mediterranean site, the concentration of organic matter under the shrub canopy was almost double of that in barren soils. The concentrations of calcium carbonate and nitrate as well as the pH and salinity values did not change between inter- and under-shrub patches. The concentrations of potassium, sodium, calcium, and magnesium were always higher under the shrub canopy; in contrast, phosphate concentrations were significantly higher in barren soils. Average values for water and ammonium content were higher under the shrub canopy.

Soil Bacterial Abundance

The concentration of bacterial PLFAs in the soil was taken as an indicator of bacterial abundance. Figure 1 describes the concentrations of PLFAs in inter- and under-shrub patches, demonstrating significant differences between the two patch types in the arid and semi-arid sites but not in the Mediterranean site. Moreover, the concentration of biomass in the semi-arid soil samples collected under the shrub canopy was similar to that in the samples taken from both patch types at the Mediterranean site.
Figure 1

Content of bacterial phospholipid fatty acids in soils sampled from inter-shrub patches and under the canopy at the three sites, distinct in their annual precipitation (Table 1). ‘S’ marks statistically significant differences (t test, P < 0.05) between the two patch types

Bacterial Diversity

T-RFLP Analysis

To verify the validity of our sampling approach, we analyzed the bacterial fingerprints in all samples taken from triplicate plots within each site and tested their effect on community composition [21]. The bacterial T-RFLP analysis showed no statistically significant differences in the distribution of diversity among each patch type of the three replicate plots (P = 0.33), suggesting that the soil samples taken from the three plots at each site were true replicates. The bacterial communities were distributed according to patch type (P = 0.0091), indicating that the dominant perennial canopy influences the soil’s bacterial diversity. The analysis was block designed so that permutations were only allowed within each site (reflecting the six composite soil samples) resulting in two distinct groups clustered according to patch type.

Clone Library

Since each replicate plot reflected the soil bacterial diversity within the study site [21], we focused our attention on one patch at each site to explore bacterial community composition using clone library analysis.

Both the T-RFLP [21] and the clone library analyses indicated high bacterial diversity within patch types at each of the sites. However, estimations of the number of phyla and richness in each sample were higher in the arid site and decreased with the increase in precipitation, with smaller differences between the semi-arid patches and no difference observed for samples obtained from the Mediterranean site (Fig. 2). Our results indicate higher bacterial diversity in barren soils compared to the communities detected in patches sampled under the canopy of the dominant perennial.
Figure 2

Diversity and richness of phyla in soils from inter-shrub patches and under the canopy along a precipitation gradient. Bars indicate the number of clusters at a level of 15% similarity and their correspondence to phylum level. Circles indicate the calculated richness (Chao 1) index and vertical lines mark upper and lower intervals at 95% confidence.

The most pronounced differences between inter- and under-shrub soils were observed in the composition of the bacterial community in the arid soils (Fig. 3). The differences detected between the semi-arid patches were small, whereas samples obtained from the Mediterranean site were very much alike. In the arid site, we observed significant differences (t test, P < 0.001) in the abundance of Actinobacteria and Proteobacteria (Fig. 3a), which were further established when higher phylogenetic levels within these phyla were analyzed (data not shown). Trends toward differences between patches were detected for the phyla Acidobacteria, Firmicutes, Chloroflexi, Verrucomicrobia, Gemmatimonadetes, and Cyanobacteria (Fig. 3a). In the semi-arid soils, significant differences (t test, P < 0.01) between bacterial communities were only observed for the Proteobacteria (Fig. 3b). However, trends toward differences were detected for the phyla Actinobacteria and Bacteroidetes, and further significant differences were detected in higher phylogenetic levels (data not shown). The analysis of soils collected from the Mediterranean site yielded no significant differences in any of the phyla or higher taxonomic levels (Fig. 3c).
Figure 3

Community composition along a precipitation gradient. Relative abundance of detected phyla in soils obtained from inter-shrub (IS) patches and under shrub (US) canopy, in arid (A), semi-arid (B) and Mediterranean (C) sites. ‘S’ marks statistically significant differences between the two patch types (t test, P < 0.001); ‘T’ marks trends toward differences between the patches


In resource-limited environments, plants provide important nutritional flux to the soil by excreting organics through root exudates and microbe-decomposed plant material. This cycle is beneficial to both the plant and the microorganisms under its canopy, especially in environments that support low biomass. Here, we show that soil bacterial abundance and diversity in the arid and semi-arid sites differ significantly between inter- and under-shrub patches. In contrast, soils retrieved from Mediterranean patches shared similar bacterial communities (Figs. 1 and 3). This observation is in agreement with previous studies showing that in water-limited areas (<300 mm/year), the perennial canopy creates a “resource island” and thus spatial heterogeneity in annual plant species [5], and in microbial diversity, abundance, and productivity [14, 15, 18]. However, in our study, the changes in bacterial diversity and abundance cannot be attributed to access to resources under the canopy as the measured soil moisture and organic pool were found to be similar in both patch types at the arid and semi-arid sites (Table 1). We further noted that the differences in soil physicochemical parameters increased with precipitation while the bacterial community abundance and composition in the patch types became more similar (Figs. 1 and 3). Similarly, fungi isolated from soils sampled at inter- and under-shrub patches in a cold desert formed spatially distinctive communities, which were not significantly correlated to soil factors [8].

The prominent taxonomic groups detected in the arid and semi-arid soils, Actinobacteria and Proteobacteria (Fig. 3), were affected by patch type. In contrast, the same bacterial groups dominate inter- and under-plant patches in the Sonoran desert soil crusts, but showed little difference between patch types [25]. Here, the crust was removed prior to sampling and we detected patch-mediated differences between these phyla in the upper soils. Upon closer examination of the actiobacterial clones, higher abundance of the radiotolerant bacteria Rubrobacterales [26] and Geodermatophilaceae, a group isolated from deserts around the world [27], was detected in the barren soils, while the α-Proteobacteria group dominated in the soils collected under the canopy in the arid and semi-arid sites. The latter includes bacteria that can form associations with plants [28], which might explain their higher abundance under the shrub canopy. Indeed, α-Proteobacteria clones were also found to dominate agricultural soils in the Sonoran desert [29].

Bacterial diversity might be governed not only by a patchy landscape but also by the type of shrub under which the community resides and by seasonality. In a study that compared bacterial diversity under the canopy of the shrub Reaumuria negevensis and in barren patches of arid soils [30], the detected communities differed from the ones described in our study, although both samplings took place at close proximity. This might correspond to the plant type under which canopy the soils were sampled: the soil under the canopy of R. negevensis contains high concentrations of salt excreted by the plant [31], unlike the soil under the shrub examined here, Hammada scoparia (Table 1).

Physicochemical measures of the soil samples, including soil moisture, carbon availability, salinity, and pH, did not correlate to the changes observed in bacterial abundance and diversity. Moreover, although some nutrient concentrations differed among patch types, no differences were detected in the nutrients that might limit bacterial growth [32, 33]. A possible explanation for the observed differences in diversity could lay in the microclimatic characteristics in inter- and under-shrub canopy patches. Although we did not measure these parameters, it has been shown that the meager canopy of the Negev desert dwarf shrubs reduces soil temperature and radiation [2], which might in turn contribute to changes in bacterial diversity: bacterial exposure to high temperatures [34] and radiation [35] is known to alter community composition. We could thus hypothesize that under favorable conditions, bacteria residing under the canopy experience longer growth-favorable periods, leading to the higher biomass we detected (Fig. 1). In addition, the bacterial groups found in barren soils have to tolerate higher desiccation, radiation, and temperatures, compared to the groups associate with the shrub canopy, which might be more susceptible to these harsh conditions. This might explain the patchiness observed in the bacterial community composition. Similarly, a study of bacterial abundance and community composition in the Chihuahuan desert suggested that soil moisture and heat load, rather than carbon input, regulate the spatial structure of the bacterial community among vegetation patches [36]. Likewise, a study exploring fungi diversity in arid and semi-arid environments indicated that the occurrence of favorable conditions for fungi in arid systems vary considerably in time and space and could not be predicted based solely on abiotic considerations [37].

In the arid and semi-arid soils, shrub-mediated distinctions were more pronounced than in the Mediterranean site (Fig. 3 and Table 1). Shrub canopy has been shown to decrease evaporation following rain events in the desert [2], and it has been suggested that plants sustain microbial banks in the soil that can undergo bursts of activity in response to rain events [18]. We suggest that in the Mediterranean site, characterized as a dense shrubland, the distinction between inter- and under-shrub soils is less pronounced. The physicochemical differences between the soil patches during favorable conditions are minute leading to similar abundance and diversity of the soil bacteria. In order to test our hypothesis, we will need to perform further studies addressing the correlation between plant density and bacterial patchiness surveying a larger number of patches in diverse climatic regions.


We would like to acknowledge the contribution of Michigan State University (Ribosomal Database Project) and the technical assistance of Rina Myaskovsky and Reuma Arusi. We thank the Israeli LTER program (http://lter.bgu.ac.il) for allowing sampling on their sites.

Supplementary material

248_2011_9957_MOESM1_ESM.doc (31 kb)
Supplementary Table 1Main characteristics of the three LTER sites sampled in this study (DOC 31 kb)
248_2011_9957_MOESM2_ESM.doc (84 kb)
Supplementary Figure 1Rarefaction curves calculated for observed OTUs with ≥85% sequence similarity. The clone libraries were amplified from inter-shrub (IS) patches and under shrub canopy (US), in arid, semi-arid and Mediterranean sites (DOC 83.5 kb)

Copyright information

© Springer Science+Business Media, LLC 2011