Ecology Letters, (2003) 6: 613–622 REPORT Bacterial diversity patterns along a gradient of primary productivity Abstract M. Claire Horner-Devine1*, Primary productivity is a key determinant of biodiversity patterns in plants and animals Mathew A. Leibold2, but has not previously been shown to affect bacterial diversity. We examined the Val H. Smith3 and relationship between productivity and bacterial richness in aquatic mesocosms designed 1 Brendan J. M. Bohannan to mimic small ponds. We observed that productivity could influence the composition 1 Department of Biological and richness of bacterial communities. We showed that, even within the same system, Sciences, Stanford University, different bacterial taxonomic groups could exhibit different responses to changes in Stanford, CA 94305, USA productivity. The richness of members of the Cytophaga-Flavobacteria-Bacteroides 2Department of Ecology and group exhibited a significant hump-shaped relationship with productivity, as is often Evolution, University of Chicago, Chicago, IL 6063, USA observed for plant and animal richness in aquatic systems. In contrast, we observed a 3Department of Ecology and significant U-shaped relationship between richness and productivity for a-proteobacteria Evolutionary Biology, University and no discernable relationship for b-proteobacteria. We show, for the first time, that of Kansas, Lawrence, KS 6604, bacterial diversity varies along a gradient of primary productivity and thus make an USA important step towards understanding processes responsible for the maintenance of *Correspondence: E-mail: bacterial biodiversity. [email protected] Keywords Bacterial diversity, biodiversity, community composition, eutrophication, freshwater ecology, microbial ecology, primary productivity, ribosomal DNA, taxonomic diversity. Ecology Letters (2003) 6: 613–622 Dodson et al. 2000; Mittelbach et al. 2001). Other studies INTRODUCTION have reported a monotonic increase of diversity with A primary goal of ecology is to understand the distribution productivity, a decrease, a U-shaped relationship or no of organisms. Investigating patterns of biodiversity is a discernible pattern (Abrams 1995; Mittelbach et al. 2001). crucial step towards achieving this goal (Lubchenco et al. Variation in observed productivity–diversity patterns may 1991). While many factors likely affect the biodiversity of a result both from differences in study design and differential region, primary productivity (the rate of energy capture and responses of organisms at various spatial and temporal carbon fixation by primary producers) is emerging as a key scales (Mittelbach et al. 2001). For example, the geographical determinant of plant and animal biodiversity, especially scale (e.g. local vs. regional) and ecological scale (e.g. within species richness (i.e. the number of species present; vs. among communities) of studies often influences patterns Rosenzweig 1995; Mittelbach et al. 2001). However, it is even within the same taxonomic group (Waide et al. 1999; unknown how, or even if, bacterial diversity varies with Gross et al. 2000). It is also possible that some studies lump primary productivity. Understanding patterns of bacterial taxa or guilds that differ in their responses to productivity diversity is of particular importance because bacteria may and thus mask patterns at different taxonomic scales (e.g. well comprise the majority of the earth’s biodiversity and Haddad et al. 2000; Torsvik et al. 2002). In addition, our they mediate critical ecosystem processes (Cavigelli & understanding of productivity–diversity relationships is Robertson 2000; Torsvik et al. 2002). influenced strongly by particular well-studied taxonomic The relationship between productivity and diversity has groups such as terrestrial plants and, therefore, may be been of long-standing interest to ecologists. Many studies of biased by these taxa (Mittelbach et al. 2001). Other taxa plants and animals have reported a hump-shaped relationship (especially microorganisms) remain relatively unstudied. between productivity and diversity, where diversity peaks at Our ignorance regarding patterns of bacterial diversity is intermediate productivity (Rosenzweig 1995; Leibold 1999; primarily due to significant theoretical and practical problems Ó2003 Blackwell Publishing Ltd/CNRS 614 M. C. Horner-Devine et al. that have, until recently, hindered the quantification of assessed by using length heterogeneities in the intergenic bacterial diversity. These problems include the very small region between 16S and 23S ribosomal genes, changed in proportion of microbial species that can be cultured (Amann response to additions of nitrogen and phosphorous. et al. 1995), the very large number of individuals that may be We know of only three field studies that have attempted present per sample, the high diversity that may be present at a to document the relationship between primary productivity small scale and the difficulty of defining a microbial species and bacterial diversity (Benlloch et al. 1995; Torsvik et al. (Goodfellow & O’Donnell 1993). However, solutions to 1998; Schafer et al. 2001). Over a period of 13 days, Schafer many of these problems have recently been developed et al. (2001) found that nutrient addition first decreased (O’Donnell et al. 1994; Ovreas 2000; Hughes et al. 2001). bacterial diversity (measured as the number of DNA derived For example, a number of techniques have been developed DGGE bands). A subsequent increase in the abundance of that assess bacterial diversity without requiring growth. The protists, and thus possible increased grazing, was accom- most promising of these techniques use ribosomal gene panied by increased bacterial diversity. In the post-grazing sequences (obtained from DNA isolated from the environ- stage, bacterial diversity once again decreased. These results ment) as the indicators of bacterial phylogenetic richness suggest that both bottom–up and top–down processes (O’Donnell et al. 1994; Stackebrandt & Rainey 1995). might control bacterial diversity. They also observed that Evidence from laboratory studies suggests that produc- community composition varied with nutrient addition. tivity may influence microbial diversity, and such studies Benlloch et al. (1995) used the richness of 16S rDNA offer insight into possible mechanisms responsible for sequences as an estimate of bacterial diversity in two coastal observed productivity–diversity relationships (Kaunzinger & lagoons that differed in primary productivity. They observed Morin 1998; Bohannan & Lenski 2000; Kassen et al. 2000). a greater number of unique ribosomal gene sequences in the For example, Kassen et al. (2000) found that bacterial more productive lagoon than in the less productive lagoon, diversity peaked at intermediate productivity in heterogene- suggesting that bacterial richness may increase with primary ous lab microcosms. They suggested that this was likely due productivity. In addition, the sample from the more to niche specialization in a heterogeneous environment. productive lagoon contained individuals related to 10 major This work, however, considered only strains of one bacterial phylogenetic groups while the less productive lagoon species and, like most terrestrial plant studies, contained contained representatives from only five. However, due to only one trophic level. In contrast, Kaunzinger & Morin the time and effort necessary to clone and sequence, only 50 (1998) used multi-trophic level laboratory microcosms clones per lagoon were assessed, and 70% of these clones comprised of aquatic bacteria and protists, to demonstrate were unique. This suggests that the lagoons were likely very that increasing productivity resulted in increased food chain undersampled (Benlloch et al. 1995; Hughes et al. 2001). length. Furthermore, they observed that food chain length Torsvik et al. (1998) observed that pristine aquatic sediments determined the numerical response of a given trophic level had much higher prokaryotic diversity than sediments to productivity; for example, bacteria increased in abun- below fish farms (which receive a substantial input of dance in response to productivity only in food chains with nutrients via fish feed), suggesting that increased nutrients an odd number of trophic levels. Finally, Bohannan & may decrease diversity. In all three studies discussed above, Lenski (2000) observed that increasing productivity resulted only two productivity levels were sampled, and thus the in an increase in the relative importance of competition authors could not differentiate between a linear and (ÔÔbottom–upÕÕ effects) and predation (ÔÔtop–downÕÕ effects) quadratic trend. as determinants of bacterial community composition in To determine the relationship between both bacterial and laboratory microcosms. At low productivity levels, superior algal richness and primary productivity, we estimated competitors were favoured, while at high levels of produc- bacterial taxonomic richness along a gradient of primary tivity more predator-resistant types were favoured. These productivity in freshwater mesocosms. Small ponds and, observations are consistent with the keystone-predation similarly, mesocosms that mimic small ponds have proven model (Leibold 1996), which predicts a unimodal relation- to be excellent model systems with which to study ship between productivity and diversity. productivity–diversity relationships (Wilbur 1997; Leibold Observations from several field studies suggest that 1999; Chase & Leibold 2002; Downing