The sustainability of coastal ecosystems in the face of widespread environmental change is an issue of pressing concern throughout the world (Emeis et al. 2001). Coastal ecosystems form a dynamic interface between terrestrial and oceanic systems and are one of the most productive ecosystems in the world. Coastal systems probably serve more human uses than any other ecosystem and they have always been valued for their rich bounty of fish and shellfish. Coastal areas are also the sites of the nation’s and the world’s most intense commercial activity and population growth; worldwide, approximately 75% of the human population now lives in coastal regions (Emeis et al. 2001).
Over the past three decades nutrient enrichment of coastal and estuarine waters has become the premier issue for both scientists and managers (National Research Council 2000). Our understanding of coastal eutrophication has been developed principally through monitoring of estuaries, with a focus on pelagic or subtidal habitats (National Research Council 2000, Cloern 2001). Because estuarine systems are usually nitrogen limited, NO3- is the most common nutrient responsible for cultural nutrient enrichment (Cloern 2001). Increased nitrogen delivery to pelagic habitats of estuaries produces the classic response of ecosystems to stress (altered primary producers and nutrient cycles and loss of secondary producer species and production; Nixon 1995, Rapport and Whitford 1999, Deegan et al. 2002).
Salt marsh ecosystems have been thought of as not susceptible to nitrogen over-loading because early studies found added nitrogen increased marsh grass production (primarily Spartina spp., cordgrass) and concluded that salt marshes can adsorb excess nutrients in plants and salt marsh plant-derived organic matter as peat (Verhoeven et al. 2006). Detritus from Spartina is important in food webs (Deegan et al. 2000) and in creating peat that forms the physical structure of the marsh platform (Freidrichs and Perry 2001). However, the accumulation of peat and inputs of sediments and loss of peat through decomposition and sediment through erosion may be altered under high nutrient regimes and threaten the long-term stability of marsh systems.
Nitrogen addition may lead to either net gain or loss of the marsh depending on the balance between increased marsh plant production and increased decomposition. Absolute change in marsh surface elevation is determined by marsh plant species composition, production and allocation to above- and belowground biomass, microbial decomposition, sedimentation, erosion and compaction (Friedrichs and Perry 2001). Levine et al. (1998) suggested that competitive dynamics among plants might be affected by nutrient enrichment, potentially altering relative abundance patterns favoring species with less belowground storage and thus lowering rates of peat formation. When combined with the observation that nutrient additions may also stimulate microbial respiration and decomposition (Morris and Bradley 1999), the net effect on the salt marsh under conditions of chronic nitrogen loading is a critical unknown.
Although most research treats nutrient enrichment as a stand-alone stress, it never occurs in isolation from other perturbations. The effect of nutrient loading on species composition (both plants and animals) and the resultant structure and function of wetlands has been largely ignored when considering their ability to adsorb nutrients (Verhoeven et al. 2006). Recent studies suggest the response of estuaries to stress may depend on animal species composition (Silliman et al. 2005). Animal species composition may alter the balance between marsh gain and loss as animals may increase or decrease primary production, decomposition or N recycling (Pennings and Bertness 2001).
Failure to understand interactions between nutrient loading and change in species composition may lead to underestimating the impacts of these stresses. The 'bottom up or top down' theory originated from the observation that nutrient availability (bottom up)sets the quantity of primary productivity, while other studies have shown that species composition (top down), particularly of top consumers, has a marked and cascading effect on ecosystems, including controlling species composition and nutrient cycling (Matson and Price 1992, Pace et al. 1999). Most examples of trophic cascades are in aquatic ecosystems with fairly simple, algal grazing pelagic food webs (Strong 1992). The rarity of trophic cascades in terrestrial systems has been attributed to the importance of detrital food webs (Polis 1999). Detritus-based aquatic ecosystems, such as salt marshes, bogs, and swamps, have classically been considered bottom-up or physically controlled ecosystems.
Recent experiments, however, suggest that salt marshes may exhibit top-down control at several trophic levels (Silliman and Zeiman. 2001, Silliman and Bertness 2002, Quiñones-Rivera and Fleeger 2005). One abundant, ubiquitous predator, a small (<10 cm total length) killifish (Fundulus heteroclitus, mummichog) has been suggested to control benthic algal through a trophic cascade because they prey on the invertebrates that graze on the benthic algae (Kneib 1997, Sarda et al. 1998). In late summer, killifish are capable of consuming 3-10 times the creek meiofauna production and meiofauna in the absence of predators appear capable of grazing over 60% of the microalgal community per day (Carman et al. 1997). Strong top-down control by grazers is considered a moderating influence on the negative effects of elevated nutrients on algae (Worm et al. 2000). Small-scale nutrient additions and predator community exclusion experiments have demonstrated bottom-up and top-down control of macroinfauna in mudflats associated with salt marsh creeks (Posey et al. 1999, Posey et al. 2002). Together, these observations suggest mummichogs are at the top of a trophic cascade that controls benthic algae (Sarda et al. 1998).
Mummichogs are also omnivorous and ingest algae, bulk detritus and the attached microbial community (D’Avanzo and Valiela 1990). As a result, marsh decomposition rates may be limited by top-down controls through trophic pathways or by release from competition with algae for nutrients.
Whole-ecosystem experiments have shown that responses to stress are often not predictable from studies of the individual components (Schindler 1998). Developing the information needed to predict the interacting impacts of nutrient loading and species composition change requires experiments with realistic alterations carried out at scales of space and time that include the complexities of real ecosystems. Whole ecosystem manipulation experiments have been used effectively in other ecosystems (Bormann and Likens 1979, Carpenter et al. 1995), but they are rare in coastal research. Experiments in salt marshes have traditionally been less than a few m2. Our understanding of the response of salt marsh plants to nutrient enrichment is from small (<10 m2), plot-level additions where uniform levels of dry inorganic fertilizer (20 to > 1000 g N m-2 y-1) are sprinkled on the marsh surface at low tide. Dry fertilizer additions were usually made every two weeks or monthly and the duration of elevated nutrient levels after these additions was usually not determined. Tidal water is the primary vector for N delivery to coastal marshes, suggesting that dry fertilizer addition to the marsh surface may not be the best basis for determining if Spartina production responds to nutrient enrichment of tidal waters. Similarly, our understanding of top-down controls in salt marshes also relies on small (1 - 4 m2) exclusion experiments that use cages to isolate communities from top consumers. While the design of these cage experiments has improved, there are some remaining drawbacks. For example, it is impossible to selectively exclude single species using cages, and recruitment or size-selective movement into or out of the cages may obscure interpretations. In addition, while these small-scale experiments provide insight into controls on isolated ecosystem processes, they do not allow for interaction among different parts of the ecosystem which may buffer or alter the impacts and are not appropriate for determining the effects of populations of larger more motile animals on whole-ecosystems or the effects of ecosystem changes on populations. For example, interactions may be caused when a motile species alters its distribution among the habitats available to it because of an experimental treatment. Small-scale experiments generally do not allow such events to happen. Complex feedbacks among physical and biological processes can alter accumulation rates and affect marsh elevation relative to sea level rise making extrapolation of small plot level experiments to whole marsh ecosystems problematic.
We are conducting an ecosystem-scale, multi-year field experiment including both nutrient and biotic manipulations to coastal salt marsh ecosystems. We are testing, for the first time at the ecosystem level, the hypothesis that nutrient enrichment and species composition change have interactive effects across multiple levels of biological organization and a range of biogeochemical processes. We altered whole salt marsh creek watersheds (~60,000 m2 of saltmarsh) by addition of nutrients (15x ambient) in flooding waters and by a 60% reduction of a key fish species, the mummichog. Small marsh creek watersheds provide an ideal experimental setting because they have the spatial complexity, species composition and processes characteristic of the larger salt marsh ecosystem, which are often hundreds of thousands of m2. Manipulating entire salt marsh creeksheds allowed us to examine effects on large motile animals and the interactive effects of motile species changes on ecosystem processes without cage artifacts. Because our manipulations were done on whole-marsh ecosystems, we are able to evaluate the integrated and interactive effects on all habitats (e.g., water column, tidal creeks and marsh) and on populations. These experiments are similar in many respects to the small watershed experiments carried out in forested catchments.
Our nutrient enrichment is novel compared to past studies in two important ways. We added nutrients (N and P) directly to the flooding tidal creek waters to mimic the way in which anthropogenic nutrients reach marsh ecosystems. All previous experimental salt marsh nutrient enrichment studies used a dose-response design with spatially uniform dry fertilizer loading on small plots (<10 m2). Nutrients carried in water will interact and reach parts of the ecosystem differently than dry fertilizer. Our enrichment method also creates a spatial gradient of nutrient loading across the landscape that is proportional to the frequency and depth of inundation in the marsh. Spatial gradients in loading within an ecosystem are typical in real world situations in many terrestrial and aquatic ecosystems. Because of our enrichment method, at any location in the ecosystem, nutrient load will be a function of the nutrient concentration in the water, the frequency and depth of tidal flooding and the reduction of nutrients from the flooding waters by other parts of the ecosystem. Uniform loading misses important aspects of the spatial complexity of ecosystem exposure and response.
This work is organized around two questions that are central to understanding the long-term fate of coastal marshes:
1. Does chronic nutrient enrichment via flooding water increase primary production more than it stimulates microbial decomposition? 2. Do top-down controls change the response of the salt marsh ecosystem to nutrient enrichment?
Here we present findings on the first 2 years of these experiments including 1) water chemistry, 2) standing stocks and species composition of benthic microalgae, 3) microbial production, 4) species composition and ecophysiology of macrophytes, 5) invertebrates, and 6) nekton. Because even highly eutrophic waters result in nutrient loading that is an order of magnitude less than most plot level experiments, we expected little stimulation of salt marsh vascular plant growth. However, moderate levels of nutrient enrichment in the water column were expected to increase benthic algal biomass and to stimulate bacterial activity and detrital decomposition throughout the ecosystem because of direct uptake of nitrogen from the water column and availability of more high quality organic matter from increased algal production. We predicted nutrient enrichment would increase invertebrate production because of an increase of high quality microalgal and microbial production at the base of the food web. Finally, we predicted that fish reduction would reduce predation on benthic invertebrates resulting in increased abundance of benthic invertebrates that would graze down the benthic algae.