ECOLOGY Friday Sep 18 11:35 AM ecol 78 817 Mp 2303 v File # 17sc Allen Press x DTPro 96-0380 Ecology, 78(8), 1997, pp. 2303±2315 q 1997 by the Ecological Society of America EFFECT OF RESOURCE ENRICHMENT ON A CHEMOSTAT COMMUNITY OF BACTERIA AND BACTERIOPHAGE BRENDAN J. M. BOHANNAN AND RICHARD E. LENSKI Center for Microbial Ecology, Michigan State University, East Lansing, Michigan 48824 USA Abstract. We determined the responses of a model laboratory community to resource enrichment and compared these responses to the predictions of prey-dependent and ratio- dependent food chain models. Our model community consisted of Escherichia coli B and bacteriophage T4 in chemostats supplied with different concentrations of glucose. We ob- served the following responses to enrichment: (1) a large and highly signi®cant increase in the equilibrium population density of the predator, bacteriophage T4, (2) a small but signi®cant increase in the equilibrium population density of the prey, E. coli, and (3) a large and highly signi®cant decrease in the stability of both the predator and prey popu- lations. These responses were better predicted by a prey-dependent model (altered to include a time delay between consumption and reproduction by predators) than by a ratio-dependent model. Enrichment had a large effect on evolutionary change in our system. Enrichment signi®cantly decreased the amount of time required for mutants of E. coli that were resistant to predation by bacteriophage to appear in the chemostats. Enrichment also signi®cantly increased the rate at which these bacteriophage-resistant mutants invaded the chemostats. These results were also better predicted by the prey-dependent model. Invasion by bacte- riophage-resistant mutants had a large effect on the subsequent population dynamics of both predator and prey. Both the equilibrium density and stability of the E. coli population increased following invasion, and the population shifted from being primarily limited by predators to being primarily limited by resources. After invasion by the mutants, the T4 population decreased in equilibrium density, and the population cycled with an increased period. These results were compared to the predictions of a ratio-dependent model and a prey-dependent model altered to include T4-resistant mutants. The dynamics of this com- munity were better predicted by the modi®ed prey-dependent model; however, this model was more complex mathematically than the simpler ratio-dependent model. Key words: bacteriophage T4; Escherichia coli; population dynamics; predation; prey-dependent models; ratio-dependent models; resource enrichment. INTRODUCTION plitude and period of population oscillations (Rosen- zweig 1971). The dynamics of predator±prey and other exploit- ative interactions have long been recognized as fun- These classical models are considered ``prey-depen- damentally important to the structure of ecological dent'' because they assume that the attack rate of pred- communities (Hairston et al. 1960, Paine 1966, Lub- ators depends only on the instantaneous density of prey. chenco 1978). Nonetheless, there remains considerable Some theorists have argued that the attack rate is often debate over such basic issues as the effects of resource better modeled as a function of the ratio of prey to enrichment on these interactions and how best to model predator density (Arditi and Ginzburg 1989). Such these effects (Arditi et al. 1991a, Ginzburg and Ak- ``ratio-dependent'' models make very different predic- cËakaya 1992, Diehl et al. 1993, Abrams 1994, Berry- tions concerning the effect of enrichment on prey and man et al. 1995). Classical predator±prey models (i.e., their predators. Enrichment is not predicted to be de- Lotka-Volterra models and modern variations thereof) stabilizing, and the equilibrium population sizes of both make two controversial predictions concerning the ef- predators and prey are predicted to increase in response fect of resource enrichment on prey and their predators. to enrichment. First, these models predict that enrichment will result Proponents of ratio-dependent models have sug- in an increase in the equilibrium population density of gested that this approach is superior because it captures the predator but have no effect on the equilibrium pop- the effects of heterogeneity on predator±prey dynam- ulation density of the prey (Rosenzweig 1977). Second, ics. Such heterogeneity could include differences in the classical predator±prey models predict that enrichment time scales of feeding by predators and reproduction can destabilize a predator±prey pair, increasing the am- by predators, discontinuous prey reproduction, spatial heterogeneity, and heterogeneity in prey edibility (Ar- Manuscript received 3 June 1996; revised and accepted 10 diti and Ginzburg 1989). The superiority of ratio-de- May 1997. pendent models in these situations has been hotly de- 2303 ECOLOGY Friday Sep 18 11:35 AM ecol 78 817 Mp 2304 Allen Press x DTPro File # 17sc 2304 BRENDAN J. M. BOHANNAN AND RICHARD E. LENSKI Ecology, Vol. 78, No. 8 bated (Oksanen et al. 1992, Diehl et al. 1993, Abrams bacteria increased in abundance in response to in- 1994, Gleeson 1994, AkcËakaya et al. 1995, Berryman creased nutrient input. However, they sampled the bac- et al. 1995). This debate has centered on whether ratio- teria population only twice during their 52-d experi- dependent models do indeed capture the effects of het- ment (they were primarily interested in protozoan pop- erogeneity, whether it is better to model heterogeneity ulation dynamics); if the bacteria population cycled in by using a ratio-dependent model or by explicitly in- response to predation, these estimates of population corporating heterogeneity into a prey-dependent mod- density could be inaccurate. In addition, Balciunas and el, and what the trade-offs are in using these two ap- Lawler used a heterogeneous population of bacteria in proaches. their experiments, and the increase in bacteria abun- There have been a number of attempts to answer dance could be due to an increase in the abundance of these questions using ®eld systems. Most of these at- less edible members of the mixed population. Balciunas tempts have involved comparing trophic structure and/ and Lawler found some evidence for predator mutual or trophic level biomass across natural gradients of interference and could not rule out ratio-dependent pre- productivity (e.g., Arditi et al. 1991a, Ginzburg and dation in their system. AkcËakaya 1992, Hansson 1992, Oksanen et al. 1992, Although most predator±prey theory assumes a Persson et al. 1992) or measuring the response of a ``chemostat-like'' environment (i.e., continuous input natural community to enrichment (e.g., O'Brien et al. of resources, constant mortality, etc.), both studies 1992, Wootton and Power 1993, Stow et al. 1995). The above used batch culture systems rather than chemo- results of these attempts have been inconclusive. In stats. In batch culture, an aliquot of the culture is trans- some studies prey-dependent models appeared to better ferred at regular intervals to fresh culture medium. The predict the responses (Hansson 1992, Oksanen et al. effect of such serial transfer is potentially confounding; 1992, Persson et al. 1992, Wootton and Power 1993), it was considered by Harrison (1995) to be the major while in other studies the responses appeared to be reason that he was unable to get a close ®t between better predicted by ratio-dependent models (Arditi et some of Luckinbill's data and the predictions of math- al. 1991a, Ginzburg and AkcËakaya 1992, O'Brien et ematical models. al. 1992, Schmitz 1993). The limitations inherent in We have built on these previous attempts by using using ®eld systems to test these models have been well chemostat communities of bacteria and bacteriophage discussed in the literature (e.g., Power 1992). These (viruses that feed on bacteria) to test prey-dependent limitations include dif®culty determining whether pop- and ratio-dependent models. We observed the response ulations are at or near equilibrium, problems with quan- of these communities to resource enrichment and com- tifying trophic level biomass, and dif®culty de®ning pared this response to quantitative predictions of prey- the physical boundaries of food chains. dependent and ratio-dependent models. Both predator Some of these limitations can be circumvented by and prey persisted in all replicates and we were able using laboratory model systems. Ecological experi- to estimate equilibrium densities and quantify stability ments with model laboratory systems can bridge the for all populations. In addition, bacteria and bacterio- gap between mathematical models and natural com- phage have suf®ciently short generation times that we munities, by allowing the predictions of mathematical were able to observe the effect of enrichment on the models to be rigorously examined in a biological sys- evolution of predator±prey interactions during the tem that is easily manipulated, replicated, and con- course of our experiment. trolled before such models are applied directly to nat- ural systems (Lawton 1995). Two attempts have been METHODS made to test prey-dependent and ratio-dependent mod- Experimental system els using laboratory model communities. In the ®rst attempt, Harrison (1995) reanalyzed the classic exper- Bacteria and bacteriophages have been proposed as iments of Luckinbill (1973). Luckinbill observed that ideal experimental systems for studying predator±prey decreasing the concentration