Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

A taxa–area relationship for bacteria

Abstract

A positive power-law relationship between the number of species in an area and the size of that area has been observed repeatedly in plant and animal communities1. This species–area relationship, thought to be one of the few laws in ecology2, is fundamental to our understanding of the distribution of global biodiversity. However, such a relationship has not been reported for bacteria, and little is known regarding the spatial distribution of bacteria, relative to what is known of plants and animals3. Here we describe a taxa–area relationship for bacteria over a scale of centimetres to hundreds of metres in salt marsh sediments. We found that bacterial communities located close together were more similar in composition than communities located farther apart, and we used the decay of community similarity with distance to show that bacteria can exhibit a taxa–area relationship. This relationship was driven primarily by environmental heterogeneity rather than geographic distance or plant composition.

Main

In the 1920s, the empirical relationship between the number of species and area was generalized4,5 as a power-law, S = cAz, where S is the number of species, A is the area sampled and c is the intercept in log–log space. The species–area exponent, z, is a measure of the rate of change of the slope with increasing area, that is, the rate of turnover of species across space. Variation in the values for c, and especially for z, is of interest because it may indicate that different processes underlie the species–area relationship at different spatial scales6,7. Although not as well studied as species–area relationships, other taxa–area relationships (for example, genera–area and family–area) have been identified for plants and animals; such relationships conform to the same power-law as species–area relationships, although they differ in their values of c and z8,9.

Bacteria are among the most abundant and diverse groups of organisms on earth10 and mediate important ecosystem processes, including trace gas emissions, decomposition and nitrogen cycling. Whereas taxa–area relationships have been observed repeatedly for numerous plant and animal taxa regardless of ecosystem type1, they have not been explicitly examined for bacteria. Unique aspects of bacterial biology may prevent bacteria from exhibiting taxa–area relationships. For example, if most bacteria are not dispersal limited (for example, owing to small size and environmental hardiness)11 and if they exhibit a high degree of ecological redundancy (for example, if bacteria are flexible in habitat requirements and physiological abilities, or if they can easily obtain traits through horizontal gene transfer that are necessary for survival in a given habitat), then one would not expect to observe a taxa–area relationship3.

Here we investigated whether bacteria exhibit a taxa–area relationship in a New England salt marsh. We conducted our work in a salt marsh because the spatial ecology of salt marshes is especially well understood12. There is an extensive literature regarding the main physical gradients in salt marshes, the spatial distribution of plant species and the ecological processes that underlie this distribution. This information provides an ideal reference point from which to investigate the spatial distribution of bacteria. We sampled 1-cm-diameter sediment cores in a nested manner over a scale of centimetres to hundreds of metres. With the possible exception of the most extreme and depauperate environments13, the diversity of bacterial communities is too high to be exhaustively sampled. Therefore we used a previously refined distance decay approach14, which uses data on the spatial turnover of taxa, to determine the taxa–area exponent, z. This approach uses comparisons of community composition rather than richness estimations to describe taxa–area relationships. For comparison, we also estimated the relationship between the number of plant species and area in this ecosystem, using the same distance decay approach.

Because a large proportion of microbes cannot be cultured with current laboratory techniques15, bacterial taxa are often identified from the sequences of indicator genes extracted from environmental samples16. We determined the bacterial community composition of our salt marsh samples by amplifying via the polymerase chain reaction (PCR), cloning and sequencing a region of 16S ribosomal DNA (rDNA), the most commonly used indicator gene for bacterial biodiversity. Because the bacterial diversity of salt marshes is often very high, we used PCR primers targeting a subset of the bacterial biota (the β-proteobacteria and relatives) to constrain the potential community we sampled. Proteobacteria are commonly found in many different environments, including salt marshes, and are often numerically dominant17. Because there is no single best definition of ‘species’ using this sequencing approach18, taxa are usually defined as operational taxonomic units (OTUs) based on sequence similarity groupings. We used the three most commonly used groupings—95%, 97% and 99% sequence similarity—to define OTUs in our study. Using multiple OTU definitions is analogous to comparing different taxonomic resolutions (for example, comparing genus, species and sub-species)16.

We sequenced a total of 945 partial 16S rDNA sequences. These sequences grouped into 88 OTUs, using a taxon resolution of 97% sequence similarity. Approximately 34% of the OTUs were singletons (n = 30), and 15 OTUs were represented by more than ten sequences. Five-hundred and twenty-three sequences were members of the β-proteobacteria; the remainder were members of the γ-proteobacteria and the δ-proteobacteria.

Bootstrapping of linear regressions between log-transformed values of bacterial similarity and geographic distance revealed significant distance decay curves for all taxonomic resolutions; that is, samples that were located closer in space were significantly more similar in bacterial composition than samples that were located farther apart (Table 1; see Supplementary Methods 2). We then computed the taxa–area z-value for bacteria from the slope of these distance decay curves (Table 1; Fig. 1).

Table 1 Taxa–area z-values
Figure 1: The taxa–area relationship for salt marsh organisms varied with taxonomic focus (a) and taxonomic resolution (b).
figure1

a, Plants and bacteria both exhibited significant taxa–area relationships, but the z-value for plants (z = 0.103) was significantly greater than the z-value for bacteria (z = 0.040; coeff. = 0.0198, t = 159.51, P = 0.001). b, Both taxonomic resolution and taxonomic focus influenced the taxa–area relationship for salt marsh bacteria. Lower resolution OTUs (that is, those defined using the criteria of 97% or 95% sequence similarity) had a lower slope (z = 0.020 and z = 0.019, respectively; coeff. = -0.041, t = -3.32, P = 0.0008) than the higher resolution OTUs (99% sequence similarity). The β-proteobacteria only exhibited a significant taxa–area relationship for the 99% resolution (grey solid line, z = 0.019).

We also observed a significant species–area relationship for plants in this same marsh (Fig. 1a). The plant z-value was significantly larger than those observed for bacteria but was similar to values estimated for other wetland plant communities19. To determine whether taxonomic focus (the particular group of taxa analysed) also affects the taxa–area relationship among different bacterial taxa, we repeated the analyses above for a subgroup of the sequences, restricting our analyses to the β-proteobacteria (Table 1). The z-value for the 99% sequence similarity grouping of the β-proteobacteria was significantly lower than that of the entire group of sequences at the 99% resolution (Fig. 1b). This suggests that the turnover in space of β-proteobacteria was lower than that of the other bacteria we sampled.

The z-value also varied by taxonomic resolution, increasing with increased taxonomic resolution for bacteria. The estimated z-value for all sequences at the 99% resolution was significantly larger than those estimated at the 97% and 95% resolutions. Similarly, the z-value at the 99% resolution for β-proteobacteria had a significantly positive z-value, whereas the z-values at the 97% and 95% resolutions were not significantly different from zero (Table 1; Fig. 1b). A similar increase in z-values with increasing taxonomic resolution has been observed for plants8 and animals9 (that is, from family to genera to species). We expect that the z-value will continue to change with taxonomic resolution at both higher and lower resolutions. We did not present the 100% sequence similarity OTUs because of the possibility that even minor PCR and/or sequencing errors could result in artefactual OTUs at this resolution. The taxonomic resolutions we used probably include a greater diversity of ecological types than contained within an animal or plant species. However, the presence of a taxa–area relationship for bacteria at this relatively coarse resolution suggests that a taxa–area relationship would probably also be present at finer resolutions, resolutions that may more closely correspond to the ecological breadth of animal and plant species.

Environmental heterogeneity often increases with increased area. The increase in heterogeneity with area together with the specificity of taxa for different habitats is the most common explanation of taxa–area patterns1, especially at scales where dispersal is not limiting to the distribution of taxa. Environmental heterogeneity may also underlie the bacterial taxa–area relationship observed here. When we removed the effect of geographic distance, partial Mantel tests showed that sites that were similar in environmental characteristics were also similar in bacterial composition (Table 2). In contrast, there was no effect of geographic distance on community similarity when we removed the effect of environmental similarity (Table 2). Although there was a significant relationship between environmental similarity and plant community composition (r = 0.36, P < 0.001), there was no relationship between plant similarity and bacterial similarity independent of environmental similarity (Table 2).

Table 2 The influence of geographic distance and habitat heterogeneity on bacterial community composition

The z-values for bacteria reported here are among the lowest z-values reported for any organisms (Fig. 2), suggesting that turnover of bacterial taxa at these spatial and taxonomic scales may be lower than that of most other organisms. Bacterial z-values may be low, in part, because it is unlikely that bacteria are dispersal limited within a salt marsh. Bacteria can probably disperse easily by air and tidal waters at this scale, and indeed we found no evidence for an effect of geographic distance on bacterial community composition when controlling for environmental heterogeneity and plant community composition. However, it is unlikely that the plants, which have a z-value at least two times higher than the bacteria, are dispersal limited at these spatial scales either, suggesting that dispersal alone cannot be responsible for the low bacterial z-values. We can think of at least three other potential explanations. First, the levels of taxonomic resolution we examined for bacteria probably reflect a broader ecological breadth than the plant species units, and thus the lower z-values for bacteria may be a reflection of taxonomic resolution. In other words, if we could define bacterial OTUs in terms that were equivalent to the ecological breadth of a typical plant species, then perhaps the bacterial species–area curve would have a z-value similar to that observed for many macro-organisms. Second, although environmental factors were related to community composition, the bacteria we sampled could have low habitat specificity (relative to the plants), reducing the spatial turnover of taxa relative to the plants in this system. Third, horizontal transfer of ecologically relevant genes could uncouple the relationship between phylotypes (phylogenetically distinct groups) and ecotypes (ecologically distinct groups), resulting in a lower z-value3.

Figure 2: A comparison of z-values for both microbial and macrobial taxa in different ecosystems.
figure2

The bacterial z-values observed in this salt marsh system represent some of the lowest values observed so far. Representative z-values were selected from a review of taxa–area relationships (see Supplementary Methods 3)

We saw no evidence that the bacterial z-value was scale-dependent (that is, the slope of the distance decay curve did not vary with distance). However, it is possible that the z-value may increase at larger spatial scales, as has been observed for plants6, or even at smaller scales, where microscale environmental heterogeneity can be very high3. Indeed, recent reports demonstrate that dispersal limitation is possible at large spatial scales for some microbes20 (although not for all21), and this would tend to increase the z-value.

For over two-hundred years, ecologists have repeatedly documented that there is a relationship between the number of eukaryotic species found in a given area and the size of the area. Our study demonstrates that both prokaryotes and eukaryotes can exhibit taxa–area relationships, although the quantitative form of the relationship may differ. The taxa–area relationship seems to be a universal law.

Methods

Sampling

We collected samples from a salt marsh on Prudence Island, Rhode Island in July 2002. Twenty-six sediment samples for bacterial community composition were collected from a nested grid that ranged from 300 × 300 m to 0.03 × 0.03 m (see Supplementary Methods 1).

We sampled plant community composition in two ways. First, we sampled along three transects at the 300-m and 30-m scales, with transects spaced at 100-m and 10-m intervals, respectively. We recorded the presence and per cent cover of each species in 1 × 1 m quadrants every 20 and 10 m, respectively, and used these data to determine the plant species–area relationship. Second, we recorded similar data in a 10 × 10 cm quadrant at each bacterial sampling point and used these data to examine the influence of plant composition on bacterial community composition.

We collected porewater samples adjacent to each sediment core to measure salinity, pH and nutrient concentrations (phosphate, ammonium and sulphate). Water was filtered with a 40-micron filter immediately after collection. Sulphate, ammonia and phosphate were measured in the field using Hach portable colorimeters (http://www.hach.com). Salinity was measured with a refractometer and pH with a calibrated probe.

DNA analyses

DNA was extracted from the top 0.5 g of each sediment core using a combination of phenol/chloroform extraction and a MoBio Ultraclean Soil DNA Kit (MoBio Laboratories). We chose primers (βAMOf and βAMOr) that amplify 16S rDNA from a subset of the domain Bacteria, as it is not tractable to sufficiently sample bacteria in our samples using domain level primers22. We reduced PCR biases by limiting the PCR cycles to twenty-five and included BSA in the reaction mix23. Each cycle consisted of 30 s at 94 °C, 30 s at 57 °C and 90 s at 72 °C. Gel-extracted and purified amplicons were cloned using the TOPO-TA cloning kit for sequencing (Invitrogen). We used an ABI 377 automated DNA sequencer to determine the sequence of the 5′ terminal 600 nucleotides of 945 of the cloned rDNA amplicons.

Phylogenetic analysis

We used the RDP database24 and ARB software25 to align the rDNA sequences from our 26 clone libraries. Ambiguously and incorrectly aligned positions were aligned manually on the basis of conserved primary and secondary structure. We identified and excluded potential chimaeras using the Chimera_Check program of the RDP24 and using ARB to compare trees generated from the 5′-end versus the 3′-end sequences separated at the break point suggested by Chimera_Check. Similarity matrices were generated using 510 unambiguously aligned positions. We grouped sequences into OTUs on the basis of rDNA sequence similarity, using DOTUR26. We used the three most commonly used groupings—95%, 97% and 99% sequence similarity—to define OTUs in our study and to examine how the taxa–area relationship varies with taxonomic resolution.

Community similarity

First, we calculated turnover using the Sorensen index, for use in estimating the z-values14. Second, we calculated the Bray–Curtis similarity coefficient (SBC) for each pair of samples, for use in the Mantel tests27. We calculated SBC on square-root-transformed data to decrease the influence of highly dominant sequences, because the most dominant sequences in a clone library may not be the most active or dominant types in the actual community, owing to primer bias28.

To control for unequal sampling (numbers of sequences) between sediment cores, we used a form of community rarefaction when calculating the pairwise community similarity indices. We randomly sampled the lowest number of sequences found at any point from each core and calculated the similarity values between all cores. We then repeated this randomization 1,000 times to get a rarefied community similarity matrix.

Taxa–area relationships

We used linear regression to examine the relationship between geographic distance between samples and similarity in bacterial composition. Because our data consisted of pairwise comparisons and thus were not independent, we used bootstrapping (10,000 replications) to test if the slope of the regression was significantly different from zero (see Supplementary Methods 2). We estimated the power-law exponent z with a distance decay approach14, using the equation log(SS) = constant - 2zlog(D), where SS is pairwise similarity in community composition and D is distance between two samples. Because some similarity values were equal to zero (that is, no OTUs in common), we coded the similarity data by adding 0.01 before log transforming each value29. The approach outlined in ref. 14 allowed us to use relative comparisons of bacterial community composition rather than richness to examine the taxa–area relationship; richness is very difficult to estimate accurately in hyperdiverse communities such as bacterial communities. There is no reason to believe that undersampling and/or PCR biases will co-vary with intersample distance or will result in preferential sampling of those taxa most likely to be shared among samples located close together in space; thus, these factors, although likely to be present, are unlikely to influence the z-values we observed. In addition, we can think of no model in which PCR biases and/or undersampling could generate a taxa–area relationship that was completely artefactual.

We compared the z-values for different taxon definitions by testing if the slopes of the regressions differed (see Supplementary Methods 2). We used the same distance decay approach to determine the z-value for plants, using data from the transect quadrants.

Influence of habitat heterogeneity

We used partial Mantel tests (9,999 permutations) to examine the influence of abiotic factors and aboveground plant composition on bacterial community composition, while holding geographic distance constant and vice versa.

We constructed a distance matrix for plant community composition from per cent cover estimates for the four dominant species (Spartina alterniflora, Spartina patens, Salicornia virginica and Limonium nashii). We constructed a matrix of environmental distance from the abiotic factors identified as most important to community composition (phosphate and ammonia concentrations), using BIO-ENV28. The BIO-ENV procedure selects a subset of available abiotic variables to maximize rank correlation between community similarity and abiotic dissimilarity matrices. We then used these matrices to test for additional distance and plant effects.

References

  1. 1

    Rosenzweig, M. L. Species Diversity in Space and Time 8–48, 190–284 (Cambridge Univ. Press, Cambridge, 1995)

    Google Scholar 

  2. 2

    Lawton, J. H. Are there general laws in ecology? Oikos 84, 177–192 (1999)

    Article  Google Scholar 

  3. 3

    Horner-Devine, M. C., Carney, K. M. & Bohannan, B. J. M. An ecological perspective on bacterial biodiversity. Proc. R. Soc. Biol. Sci. B 271, 113–122 (2004)

    Article  Google Scholar 

  4. 4

    Arrhenius, O. Species and area. J. Ecol. 9, 59–99 (1921)

    Article  Google Scholar 

  5. 5

    Gleason, H. A. On the relationship between species and area. Ecology 3, 158–162 (1922)

    Article  Google Scholar 

  6. 6

    Crawley, M. J. & Harral, J. E. Scale dependence in plant biodiversity. Science 291, 864–868 (2001)

    ADS  CAS  Article  Google Scholar 

  7. 7

    Losos, J. B. Analysis of an evolutionary species–area relationship. Nature 408, 847–850 (2000)

    ADS  CAS  Article  Google Scholar 

  8. 8

    Bennett, J. P. Nested taxa–area curves for eastern United States floras. Rhodora 99, 241–251 (1997)

    Google Scholar 

  9. 9

    Harcourt, A. H. Biogeographic relationships of primates on South-East Asian islands. Glob. Ecol. Biogeogr. 8, 55–61 (1999)

    Article  Google Scholar 

  10. 10

    Whitman, W. B., Coleman, D. C. & Wiebe, W. J. Prokaryotes: The unseen majority. Proc. Natl Acad. Sci. USA 95, 6578–6583 (1998)

    ADS  CAS  Article  Google Scholar 

  11. 11

    Baas-Becking, L. G. M. Geologie of Inleiding Tot de Milieukunde (W. P. Van Stockum & N. V. Zoon, The Hague, The Netherlands, 1934)

    Google Scholar 

  12. 12

    Bertness, M. D. The Ecology of Atlantic Shorelines 313–376 (Sinauer Associates, Sunderland, Massachusetts, 1999)

    Google Scholar 

  13. 13

    Tyson, G. W. et al. Community structure and metabolism through reconstruction of microbial genomes from the environment. Nature 428, 37–43 (2004)

    ADS  CAS  Article  Google Scholar 

  14. 14

    Harte, J., McCarthy, S., Taylor, K., Kinzig, A. & Fischer, M. L. Estimating species–area relationships from plot to landscape scale using species spatial-turnover data. Oikos 86, 45–54 (1999)

    Article  Google Scholar 

  15. 15

    Torsvik, V., Goksoyr, J. & Daae, F. L. High diversity in DNA of soil bacteria. Appl. Environ. Microbiol. 56, 782–787 (1990)

    CAS  PubMed  PubMed Central  Google Scholar 

  16. 16

    Stackebrandt, E. & Goebel, B. M. Taxonomic note: A place for DNA–DNA reassociation and 16S rRNA sequence analysis in the present species definition in bacteriology. Int. J. Syst. Bacteriol. 44, 846–849 (1994)

    CAS  Article  Google Scholar 

  17. 17

    Burke, D., Hamerlynck, E. & Hahn, D. Interactions among plant species and microorganisms in salt marsh sediments. Appl. Environ. Microbiol. 68, 1157–1164 (2002)

    CAS  Article  Google Scholar 

  18. 18

    Rossello-Mora, R. & Amann, R. The species concept for prokayotes. FEMS Microbiol. Rev. 25, 39–67 (2001)

    CAS  Article  Google Scholar 

  19. 19

    Peitinger, M., Bergamini, A. & Schmid, B. Species–area relationships and nestedness of four taxonomic groups in fragmented wetlands. Basic Appl. Ecol. 4, 385–394 (2003)

    Article  Google Scholar 

  20. 20

    Whitaker, R. J., Grogan, D. W. & Taylor, J. W. Geographic barriers isolate endemic populations of hyperthermophilic archea. Science 301, 976–978 (2003)

    ADS  CAS  Article  Google Scholar 

  21. 21

    Roberts, M. S. & Cohan, F. M. Recombination and migration rates in natural-populations of Bacillus subtilis and Bacillus mojavensis. Evolution 49, 1081–1094 (1995)

    Article  Google Scholar 

  22. 22

    McCaig, A. E., Embley, T. M. & Prosser, J. I. Molecular analysis of enrichment cultures of marine ammonia oxidizers. FEMS Microbiol. Lett. 120, 363–367 (1994)

    CAS  Article  Google Scholar 

  23. 23

    Qiu, X. et al. Evaluation of PCR-generated chimeras, mutations, and heteroduplexes with 16S rRNA gene-based cloning. Appl. Environ. Microbiol. 67, 880–887 (2001)

    CAS  Article  Google Scholar 

  24. 24

    Maidak, B. L. et al. The RDP-II (Ribosomal Database Project). Nucleic Acids Res. 29, 173–174 (2001)

    ADS  CAS  Article  Google Scholar 

  25. 25

    Ludwig, W. et al. ARB: a software environment for sequence data. Nucleic Acids Res. 32, 1363–1371 (2004)

    CAS  Article  Google Scholar 

  26. 26

    Schloss, P. D. & Handelsman, J. Introducing DOTUR, a computer program for defining operational taxonomic units and estimating species richness. 〈http://www.plantpath.wisc.edu/fac/joh/dotur.htmlAppl. Environ. Microbiol. (in the press)

  27. 27

    Magurran, A. E. Measuring Biological Diversity 175–176 (Blackwell, 2004)

    Google Scholar 

  28. 28

    Clarke, K. R. & Warwick, R. M. Change in Marine Communities: an Approach to Statistical Analysis and Interpretation 2.3, 11.8–11.11 (PRIMER-E, Plymouth Marine Laboratory, 2001)

    Google Scholar 

  29. 29

    Legendre, P. & Legendre, L. Numerical Ecology 33–47 (Elsevier Science B. V., Amsterdam, 1998)

    Google Scholar 

Download references

Acknowledgements

We are grateful to C. Anderson, H. P. Horz, A. Martiny, S. Reddy, members of M. Bertness' laboratory at Brown University and K. Nusslein's laboratory at the University of Massachusetts, Amherst for their technical assistance. We thank J. Green, D. Ackerly, P. Ehrlich, D. Relman and D. Petrov for comments on a previous draft of this manuscript. We also thank E. Bathgate, the American Association of University Women, and the National Science Foundation for their support.

Author information

Affiliations

Authors

Corresponding author

Correspondence to M. Claire Horner-Devine.

Ethics declarations

Competing interests

The authors declare that they have no competing financial interests.

Supplementary information

Supplementary Information Methods 1

Describes sampling design and includes Supplementary Information Figure S1. (PDF 45 kb)

Supplementary Information Methods 2

Describes the statistical analyses in more detail. This includes Supplementary Information Figure S2, which shows the relationship between pairwise geographic distance between samples and similarity in community composition for bacteria considered at 99%, 97% and 95% similarity. (PDF 79 kb)

Supplementary Information Methods 3

This file describes how data included in Figure 2 were selected. (PDF 126 kb)

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Horner-Devine, M., Lage, M., Hughes, J. et al. A taxa–area relationship for bacteria. Nature 432, 750–753 (2004). https://doi.org/10.1038/nature03073

Download citation

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing