Ecology Letters, (2006) 9: xxx–xxx doi: 10.1111/j.1461-0248.2005.00873.x LETTER Early Cambrian food webs on a trophic knife-edge? A hypothesis and preliminary data from a modern stromatolite-based ecosystem Abstract James J. Elser,1* James Watts,1 Here we use the theory of ecological stoichiometry to propose and provide a preliminary John H. Schampel1 and Jack test of a novel hypothesis that the Cambrian ÔexplosionÕ may have been triggered by Farmer2 changes in circulating P availability in the biosphere. We exposed living stromatolites 1 School of Life Sciences, Arizona from a spring-fed stream in Mexico to a gradient of P enrichment to examine subsequent State University, Tempe, AZ effects on stromatolite C : P ratio and on the primary grazer, an endemic snail. 85287, USA Consistent with a previously hypothesized stoichiometric Ôknife-edgeÕ, snail performance 2Department of Geology, was maximal at intermediate P-enrichment, indicating in situ stoichiometric constraints Arizona State University, because of high stromatolite C : P ratio along with high sensitivity to excessive P intake. Tempe, AZ 85287, USA *Correspondence: E-mail: These results are consistent with the idea that stoichiometric constraints may have [email protected] delayed the evolutionary proliferation of animals in ancient stromatolite-dominated ecosystems and also suggest that high food P content can significantly impair consumers. We propose that ecosystem P availability may have impacted both the expansion and decline of animal taxa in the history of life. Keywords Cambrian explosion, ecological stoichiometry, phosphorus, snails, stromatolites. Ecology Letters (2006) 9: xxx–xxx Shergold 1984; McMenamin 1988; Brasier 1990; but see INTRODUCTION Butterfield 1997). However, various findings emerging from One of the most dramatic events in the history of life on studies of ecological stoichiometry (the study of the balance Earth was the sudden expansion and diversification of of energy and multiple chemical elements in ecological metazoan animals c. 520 million years ago (Ma), the interactions, Sterner & Elser 2002) highlight the importance so-called ÔCambrian explosionÕ. These events have long- of mismatches in the elemental composition of food items fascinated scientists, leading to the production of a variety (i.e. biomass of autotrophic algae or higher plants) and of hypotheses regarding possible intrinsic and extrinsic consumers (i.e. herbivores) as major factors affecting (environmental) triggers responsible for this rapid diversi- consumer growth and production (Sterner & Elser 2002). fication. These include causes involving increasing atmo- In general, autotrophic biomass has variable but generally spheric oxygen (Towe 1970; Valentine 1973; Runnegar high C : nutrient ratio, while animals have low C : nutrient 1982; Thomas 1997), the appearance of key genomic ratios in their bodies that they homeostatically regulate innovations (Erwin 1993; Raff et al. 1994; Davidson et al. around taxon- or stage-specific values (Sterner & Elser 1995), and increases in total ecosystem productivity (Cook 2002). These physiological contrasts lead to frequent & Shergold 1984; McMenamin 1988; Brasier 1990). situations in which herbivores face strong stoichiometric With respect to the latter, considerations of the role of constraints on growth and reproduction (Elser et al. 2000). nutrients in potentially constraining the development In fact, stoichiometric models of autotroph–consumer of ancient food webs have focused entirely on overall rates interactions predict the existence of stable equilibria of total productivity. Low productivity has been proposed dominated by autotroph biomass into which consumers to somehow limit the success of large-bodied animals likely cannot invade (Ôdeterministic extinctionÕ of the consumer; to occupy trophic levels high in food webs (Cook & Andersen et al. 2004). Ó 2006 Blackwell Publishing Ltd/CNRS 2 J. J. Elser et al. The preceding discussion suggests a hypothetical scenario compared with values exhibited by aquatic and terrestrial (Elser 2003) for the Precambrian Eon, when, for billions of autotrophs in other ecosystems (Elser et al. 2000), the years, Earth’s ecosystems were dominated by microbial P-content of microbial biomass comprising the surficial biomass (heterotrophic Bacteria and Archaea, autotrophic layers (< 1 mm) of these stromatolites is extremely low (high Cyanobacteria and, eventually, eukaryotic algae), in the C : P ratios ranging from c. 500 to 5000; N : P ratios are also absence of large metazoan herbivores such as those that play correspondingly high). Since animal biomass is generally a key role in modern trophic transfer. Perhaps microbial much more P-rich (Elser et al. 2000), P-limitation of growth mats during the Precambrian, such as those generating for herbivores consuming this material seems likely. During a widespread stromatolites (bio-laminated sediments; see preliminary 21-d experiment performed in 2001, enrichment Walter 1994), suffered severe nutrient (PO4) limitation, of PO4 to stromatolites dramatically lowered its C : P ratio resulting in extremely high biomass C : P ratio that excluded from c. 3000 to c. 700. Consistent with P limitation of snail large metazoan herbivores because of stoichiometric con- growth (Elser et al. 2003), the P- and RNA-contents of straints. According to this hypothesis, this state was M. quadripaludium were significantly higher in the + P maintained until environmental changes altered microbial treatment (Elser et al. 2005a,b), consistent with higher nutritional status (lowered microbial C : P ratio) in such a growth rate in response to the P enrichment. However, in way as to improve its ability to support large, fast-growing, 2002 a similar but longer-term experiment yielded conflicting metazoans (Elser 2003). Paleoenvironmental evidence sug- results: P-enrichment lowered stromatolite C : P ratio from gesting the feasibility of such a scenario exists and will be c. 800 to < 100 but snails in the + P treatment exhibited discussed below. significantly lower RNA content, growth rate, and survivor- Thus, application of stoichiometric theory might help ship than in unenriched controls. These conflicting results, reduce existing uncertainty (Martin 2003) surrounding the coupled with similar laboratory observations recently ecological events that may have triggered the Cambrian reported for the crustacean Daphnia (Plath & Boersma ÔexplosionÕ. Testing causal hypotheses about events of the 2001), led to the hypothesis (Elser et al. 2005b) that snails at distant past is, of course, very difficult. Generally, there are Cuatro Cie´negas live on a kind of Ôstoichiometric knife edgeÕ, two ways to proceed: one can make predictions about things in which, in nature, their growth is limited by the high C : P in the fossil or geological record that would be observed if ratio of stromatolite biomass but their performance can be the hypothesis were correct or make predictions about the impaired by unnaturally P-rich (low C : P) food. outcomes of experiments that can be conducted with Thus, we propose that new developments in ecological modern systems similar to those of the past (i.e. ÔanaloguesÕ). stoichiometry, both the existence of stoichiometric con- In this paper, we take both of these strategies. First, the straints because of low food nutrient content as well as stoichiometric constraint hypothesis predicts that the negative effects of high food nutrient content (Boersma & geological record should show major changes in the global Elser in press), may help in understanding important events P cycle at the time of the Cambrian explosion. As discussed in the history of life. To bolster this suggestion we present below, this is, in fact, what is reported by Cook & Shergold the results of a field experiment in which stromatolites were (1984). Second, the stoichiometric constraint hypothesis exposed to a range of experimental P enrichment from 0 to predicts certain outcomes in studies performed in modern 10 lM P, after which various aspects of snail performance ecosystems analogous to the ancient situation: laminating were assayed. Given the Ôknife edge hypothesisÕ,we microbial mats (living stromatolites) should have high expected P fertilization to lower stromatolite C : P ratio biomass C : P ratios and P addition should stimulate animal and, in turn, to stimulate snail growth, survivorship, and growth and performance, at least in some part of the range. production but only over a limited range of P enrichment. A preliminary evaluation of these ideas comes from our At higher levels of P enrichment, the hypothesis predicts studies in the Chihuahuan desert near Cuatro Cie´negas, decreased snail performance. Our data are consistent with Coahuila, Mexico (see Methods for detailed site description). these predictions and lead us to suggest that consideration Here, a diverse array of active stromatolite-generating of stoichiometric food quality may improve our under- microbial communities (ÔstromatolitesÕ, hereafter) occupy standing of the evolution of early food webs. desert springs discharging warm and highly mineralized waters rich in Ca, Mg, CO3,SO4 and NO3 but relatively low in PO4 (Elser et al. 2005a). Furthermore, these stromatolites METHODS often co-occur with several metazoan consumers, including Study system and organisms an endemic hydrobiid snail Mexithauma quadripaludium, and thus offer a unique opportunity to examine the interactions Our experiment involved studies in the Rio Mesquites between