Biogeochemical Cycling Constraints on Stream Ecosystem Recovery D. L. DEANGELIS* P. J. MULHOLLAND d. W. ELWOOD A. V. PALUMBO A. D. STEINMAN Environmental Sciences Division Oak Ridge National Laboratory Oak Ridge, Tennessee 37831-6038, USA ABSTRACT / In systems where production is limited by the availability of a nutrient, nutrient input to and recycling within the system is related to the resilience, or speed of recovery, of a system to its steady state following a disturbance. In particular, it is shown that the return time T s of the system to steady state, or the inverse of the resilience, is approximately equal to the mean turnover time of the limiting nutrient in the system. From this relationship, it is possible to understand and predict how various properties of food webs and their environments affect resilience, These properties include nu- trient input rate, loss rate, size of the detritus compartment, and trophic structure. The elfects of these properties on re- silience are described by using simple mathematical models. To test model predictions, experimental studies of the re- sponse of periphyton-dominated stream ecosystems to dis- turbance are being conducted on a set of laboratory streams in which nutrient inputs and grazing intensity are regulated at different levels. In streams without snail grazers (low-grazed streams), 90% recirculation of stream water to reduce nutrient inputs resulted in longer turnover times (T,) of phosphorus within the stream compared with once-through flow. How- ever, in streams with snail grazers (high-grazed streams), there were no differences in phosphorus turnover time be- tween once-through and partially recirculated treatments. Results on the rate of recovery of periphyton from a flood/ scour disturbance to each stream partially support the model prediction of a positive relationship between ecosystem re- turn time (T,) and nutrient turnover time (7-,) within the streams. Return Time, Resilience, and Nutrient- Limited Systems Ecosystem stability has long been a primary interest of ecologists (e.g., Odum 1969, Woodwell and Smith 1969, Holling 1973, May 1973, Goodman 1975, McNaughton 1977). Although most investigations of the stability of ecological systems have centered on how stability is related to either the number of inter- acting species populations in the community (e.g., di- versity-stability ideas) or on the types and functional forms of the biotic interactions between species (com- petition, predation), the relationship between nutrient cycling and stability also has been a recurrent theme (Hutchinson 1948, Pomeroy 1970, Jordan and Kline 1972, O'Neill and others 1975, Webster and others 1975, 1983, DeAngelis 1980). For example, Pomeroy (1970) noted that both rain forests and coral reefs have stable physical environments but in many cases lack major inputs of essential nutrients and have little nutrient reserve. Both have evolved "tightly closed, KEY WORDS: Nutrient limitation; Streams; Resilience; Resistance;Ar- tificial streams; Periphyton *Author to whomcorrespondenceshouldbe addressed. rapid cycles of nutrients," and, following a distur- bance, both "may be slow to recover, because they have little input of nutrients." Both systems have a high de- gree of nutrient recycling relative to nutrient input, that is, a high ratio of recycled to new nutrient atoms in the system. Using the results of computer model studies of mineral cycles in forest ecosystems, Jordan and others (1972) concluded that amelioration of low nutrient input rates by the efficient or tight cycling of limiting nutrients results in reduced system resilience. The term resilie~uze refers to the rate at which a system re- turns to a steady-state equilibrium tollowing a pertur- bation away from that equilibrium (e.g., Smith 1972, Webster and others 1975, Pimm and Lawton 1977, DeAngelis 1980, Harwell and others 1981, Pimm 1984). As an example, suppose that the nutrient standing stock of a system, including portions stored in the biomass, can be described by a simple linear, one compartment model, dX(t) - IN -- kX(t) (1) dt Here, X(t) is the standing stock of nutrient, IN is the external input of nutrient to the system, assumed con- stant with time, and k is the loss rate coefficient of nu- trient from the system. The solution of this equation is EnvironmentalManagement Vol. 14, No. 5, pp. 685-697 9 1990Springer-VerlagNew York Inc.