What is the magnitude and long-term pattern of freshwater runoff and organic carbon and nitrogen and inorganic N loading from the watershed to the estuary?
Human activities now dominate the N cycle in our watersheds and significantly impact the hydrologic cycle. Human diversion of water from the watershed now equals ~20% of annual runoff but the impact of this diversion has been largely offset by increased precipitation due to climate change. As a result, annual runoff to the estuary over the last century has not detectably changed (Claessens et al. 2006). Human N input via food/waste dominate the contemporary N budget of the watershed (Williams et al. 2004). Most of the basin-wide N inputs are retained on land (>80%), though retention declines with increasing imperviousness (Williams et al. 2004, Wollheim et al. 2005, Filoso et al. 2004) because flow paths bypass soils during storm events (Pellerin et al. 2008). The river network is highly effective at denitrifying DIN inputs under low flow (> 90% of inputs from land), and shows greater than predicted retention at high flows, potentially indicating that floodplain ecosystem are important regulators of nutrients at higher flows (Wollheim et al. 2008a).
Whole ecosystem tracer additions in PIE streams revealed that the denitrification becomes lessefficient as N concentrations increase, which is consistent with other stream studies (Mulholland et al. 2008). In other experiments we found the relative size of transient storage zones increases with stream size, while exchange parameters remain constant (Briggs et al. in press). Experimental in-situ manipulations of O2 and DOC indicate that O2 is the primary regulator of N forms and N cycling, while DOC levels fuel process intensity. This suggests that a better understanding of O2 dynamics may be needed to understand DOC vs. nitrate relationships across different streams (Thouin et al. 2009).
Our work with spatially distributed models has focused on understanding these processes at the network scale. Large rivers are potentially important N removal sites at network scales because total benthic surface areas typically remain similar across stream size (Wollheim et al. 2006, 2008). Loss of denitrification efficiency results in a declining importance of small relative to large rivers in networks as N loading increases (Mulholland et al. 2008, Wollheim et al. 2008). An enhanced network model that incorporates transient storage dynamics suggests that hyporheic zones dominate N removal at network scales, despite relatively slow exchange between surface water and the hyporheos (Stewart 2009).
Land use change models shows that if trends of deforestation since 1970 continue, 40% of the existing forest will be lost by 2050 (Pontius and Neeti 2009). About 14% of the study area consists of manicured grass, and about half of this is lawns of private residences. We also found that town-level residential water-use restrictions are increasing in duration and frequency, (controlling for climate variation), and that some state-level water- and land-use policies appear to be driving this trend, whereas other policies may be mitigating the trend (Polsky et al. 2009; Hill and Polsky 2005, 2007).
How are tidal marsh processes and their connections to estuarine waters regulated by sea level, storms, and water and material inputs from land and sea?
Research at the PIE LTER has confirmed that sea-level anomalies affect salt marsh primary productivity. This led to the development of a model that explains how marsh landscapes maintain equilibrium with sea level (Morris et al. 2002, Mudd et al. 2004, Morris et al. 2005, Morris 2005, Morris 2006, Morris 2007). Marsh primary productivity is strongly related to variations in flooding and salinity during the growing season (Morris 2000, Morris et al. 2002) and to temperature (Kirwan et al. 2008). The estuarine salinity gradient migrates in response to changes in freshwater discharge and sea level, and this affects ammonium export and import by marshes (Koop-Jakobsen and Giblin 2002; Weston et al. accepted). Increased sea level rise is likely to promote the migration of salt marshes upriver and to reduce the extent of tidal fresh and brackish marshes (Buchsbaum et al. 2006). PIE salt marshes are disintegrating due to a combination of lateral erosion and marsh ponding (Cavatorta et al. 2003), which is likely due to sea-level rise and reduced sediment loads as a consequence of reforestation of the watershed following abandonment of agriculture in New England. Disintegration initially results in an increase in marsh edge (Johnston et al. 2003), where there is considerable drainage of marsh porewater. Thus,
increased sea-level rise is likely to increase porewater drainage which is a significant source of inorganic and organic nutrients for the planktonic sub-system (Raymond and Hopkinson 2003, Gardner & Gaines 2008, Wilson and Morris, in prep.). The remaining high marsh at PIE appears to be maintaining its elevation as a consequence of cannibalization (Wang et al. 2009), i.e. eroded peat from the margins is deposited on the remaining surface. To examine how changes in marsh elevation alters sediment N cycling we investigated the relative importance of denitrification vs. dissimilatory nitrate reduction to ammonium. To facilitate this, we developed a new approach that allows denitrification and DNRA to be measured in the marsh rhizosphere (Koop-Jakobsen and Giblin 2009b).
ESTUARINE PLANKTONIC SYSTEM
How do planktonic community structure and production respond to changes in organic matter, nutrients and water fluxes?
The estuary is characterized by strong spatial gradients along its length including water residence time, community structure, the relative importance of pelagic and benthic primary production (Tobias et al. 2003a), bacterial community structure (Crump et al. 2004, Crump and Hobbie 2005) and patterns of metabolism (Vallino et al. 2005). Phytoplankton blooms occur in the oligohaline part of the estuary during midsummer when water residence time is high. When residence time is short, phytoplankton blooms never occur. Instead the base of the foodweb shifts to benthic microalgae (Tobias et al. 2003a). The phytoplankton community has a greater freshwater component in the upper estuary but is dominated by marine forms seaward. Shifts in bacterioplankton community composition along the salinity gradient are related to residence time and bacterial community doubling time in spring, summer, and fall. Freshwater and marine populations represent a large fraction of the bacterioplankton community in all seasons (Crump and Hobbie 2005), however, a unique estuarine community forms at intermediate salinities when bacterial doubling time is shorter than water residence time (Crump et al. 2004).
The estuary is heterotrophic along its entire length, but especially during times of high river discharge (Vallino et al. 2005). Seasonal patterns of GPP are driven primarily by variations in residence time, temperature and radiation, while R is driven primarily by variation in marsh flooding, temperature, GPP, and residence time. Bioassay and biomarker analyses suggest that allochthonous watershed inputs of OC meet NEP demands only in the upper 5 km of the estuary. Biomarkers (Raymond and Hopkinson 2003) show a large internal input of DOC from phytoplankton and marsh grass and that the estuary exports both watershed and internally-derived organic carbon to the continental shelf. Mass balance suggests that 1 to 45% of marsh NPP is required to sustain estuarine heterotrophy. While analysis of 14C shows an internal input of old particulate C, presumably from eroding marsh peat, it appears to pass unaltered to the ocean (or perhaps onto the marsh platform –see Marsh research section). One mechanism we think is particularly important in transferring marsh C to the water column is creekbank porewater drainage. Gardner and Gaines (2008) determined creekbank drainage to average about 160 m3 yr-1 per m of edge. Scaled to creekbank edge length throughout the estuary and porewater DOC concentrations, drainage volume and C input exceeds that from the entire Parker River. However, we also find compelling evidence that much of the estuarine heterotrophy arises from respiration on the marsh platform, as the sum of water column and benthic respiration is only half of free water measures of respiration. That much of the "estuarine respiration in water" must occur on the marsh proper, as opposed to in tidal creeks, agrees with similar observations for Georgia estuaries (Wang and Cai 2004).
How do benthic recycling of nutrients and processing of organic matter respond to changes in freshwater runoff and the quality and quantity of organic matter inputs?
Seasonal and inter-annual patterns of respiration in the upper Parker Estuary are well correlated with temperature, while, ammonium release is better correlated with salinity (Giblin et al. accepted). Salinity controls ammonium release through several mechanisms. First, salinity directly affects ammonium release through ion exchange (Weston et al. accepted). Second, salinity has a large indirect effect on rates of microbial processes including, nitrification, denitrification, and dissimilatory nitrate reduction to ammonium (DNRA) (Giblin et al. accepted). During summer and early fall, when salinity is high, nitrification rates decrease by nearly an order of magnitude compared to values measured in April. The numbers of ammonium oxidizing bacteria (AOB) also decrease by nearly ten fold from spring to fall, although the overall structure of the AOB community does not change (Bernhard et al. 2005, 2007). Lower nitrification rates decrease rates of coupled nitrification/denitrification so that N losses decrease over the summer. At the same time the preferred pathway for nitrate reduction shifts from denitrification (N2) to DNRA where the end product is ammonium. These changes all results in enhanced N release from sediments and may contribute to the large mid- to late-summer phytoplankton blooms we observe. At the mid-estuary site, which has a large bivalve population, we see a very different pattern. Here salinity changes are more modest and do not exert a strong control on N cycling. Ammonium release is well correlated with benthic respiration throughout the year. In contrast to the upper estuary where salinity variations control N cycling, variability at the mid-estuarine sites is largely driven by animal abundance. We also expanded our studies of N cycling pathways onto the marsh platform and found that DNRA and denitrification are of equal importance as a nitrate sink (Koop-Jakobsen and Giblin 2010), while anaerobic ammonium oxidation is a minor pathway (Koop-Jakobsen and Giblin 2009a).
HIGHER TROPHIC LEVELS
How do the structure and function of higher trophic levels respond to changes in land, atmospheric and oceanic forcing as well as fisheries harvest?
Our work has shown that higher trophic levels play an important role in controlling transfers of energy in PIE food webs and in determining estuarine response to environmental change (see Integrating Experiments below). Understanding the movements of top predators is key to predicting both their impacts on estuarine processes and their response to climate change. A recent advance, acoustic telemetry (remotely sensed tags), allows repeated observations of individuals and has provided these new insights into striped bass (a top predator) within-estuary foraging and long-distance migration. Striped bass traveled thousands of kilometers passing by other estuaries to become repeat summer residents in PIE (Mather et al. 2009, 2010) and individuals partitioned into two distinct feeding groups – those that foraged in tidal rivers and those that foraged in the open bay (Pautzke et al. 2010). These findings contradict the common assumptions that striped bass move opportunistically and randomly in the estuary, (Mather et al 2003) and suggest that bass may be more important top-down controls on prey populations than previously thought. Additionally, we have shown an unexpectedly tight connection between specific summer feeding and overwintering estuaries. Almost 2/3rds of the PIE summer resident striped bass wintered in Delaware Bay and over 60% of these fish returned to be PIE residents the next year (Mather et al. 2010). This confirmed our result obtained from using anchor tags that PIE tagged bass returned 2 to 7 years later (Mather et al. 2009). Our results contradict the assumption that bass show no fidelity to foraging estuaries within a year and randomly use different estuaries in subsequent years.
Landscape level differences in ecosystem characteristics may be important drivers of production transfer to higher trophic levels. In salt marshes, when mummichogs were caged in fresh, brackish, and areas where they had more access to the marsh platform and to terrestrial prey with higher protein content (Haas et al. 2009). Isotope characteristics of wild fish suggest some fish may be making much longer intra-estuarine movements (>1 km) to maximize growth than previously thought (Haas et al. 2009; Logan et al. 2006). In rivers, anadromous river herring are an important component of both freshwater and estuarine systems, and can serve as important conservation education tools (Frank et al. 2009a). Our acoustic tagging work in the Ipswich River suggests that restoration of decimated herring populations by transplanting spawning stock from other rivers may be problematic. Transplanted herring tended to return to estuarine locations rather than proceed upstream towards spawning areas (Frank et al, 2009b, c).
We have two on-going large-scale experiments that integrate among our program areas. These experiments manipulate plant resources and examine resulting changes in biogeochemical cycles and food webs. The first experiment is a long-term removal of marsh vegetation by haying (Buchsbaum et al. 2009). Marsh haying, despite altering plant species composition, resulted in only minor changes in the food web and had no effect on the abundances of breeding birds. The second experiment is a chronic N additions to salt marsh creeks designed to quantify the effects and interactions of increased nutrients and reduced abundance of a key fish on the function and sustainability of salt marsh ecosystems (TIDE experiment; Deegan et al. 2007). Contrary to theory and prior caging experiments, the mummichog is a weak interactor in the algae-based trophic cascade in creeks (Deegan et al. 2007), but exerts strong trophic control in the detrital food web on the marsh platform (Johnson et al. 2009; Galvan et al. 2008). This suggests that the ecological role of mummichogs varies across the landscape and that species behavior plays a significant role in structuring the salt marsh food web (Fleeger et al. 2008, Johnson and Fleeger 2009, Johnson et al. 2009). Although chronic nitrogen enrichment to creeks resulted in no large-scale plant community changes, we measured ecologically significant changes in ecosystem nitrogen cycling (Drake et al. 2009; Koop-Jakobsen & Giblin 2010).
SYNTHESIS AND MODELING
The synthesis and modeling component of PIE-LTER focused on two main areas during the last funding cycle: 1) coupled estuarine circulation and biogeochemistry modeling and 2) advancement of our distributed metabolic network model optimized by maximum entropy production (MEP). A very high resolution (sub meter) version of our 2D finite element model for PIE was developed for Sweeny Creek to examine how our long-term marsh fertilization experiment creates shifts in primary producer dominance. The transport model has been completed and coupled to a biogeochemistry model that focuses on capturing N flow through the four functional primary producers: phytoplankton, benthic microalgae, macroalgae, Spartina. As the long-term fertilization experiment continues, we will use this high-resolution model to test ideas on how primary producer abundances change as a function of N-loading, shading and local transport characteristics. At the whole estuary scale, we have been developing FVCOM (Finite-Volume Coastal Ocean Model: Chen et al. 2003, Chen et al. 2004) for PIE using a high resolution bathymetry model based on our LIDAR data for the estuary. We have been collaborating with Chen’s group at UMass Dartmouth to accurately simulate 3D circulation in PIE estuary. Because of the connection between Plum Island Sound and the Merrimack River, we have extended the domain of the model to incorporate both the local offshore current as well as the Merrimack River, where boundary conditions are obtained from FVCOM running in the Gulf of Maine. We have found, from drifter simulations and model guided observations, that a significant residual clockwise circulation develops between the Merrimack R. and PI Sound, where water in the Sound flows northward into the Merrimack R., then tracks southerly along the shore, and re-enters the Sound via its southern mouth (Zhao et al. 2009). Our results show that PI Sound and the Merrimack R. have a much tighter coupling than previously appreciated.
Although we have developed several classic compartment-type biogeochemistry models for the estuary (Hopkinson and Vallino 1995, Vallino 2000, Wan and Vallino 2005, Vallino et al. 2005) and watershed (Filoso et al. 2004), we have also been developing a metabolic approach (Vallino et al. 1996, Vallino 2003) that relies on the principle of maximum entropy production (Dewar 2003, 2005, 2009) to explain ecosystem biogeochemistry. During this cycle our major accomplishments include, a direct means of modeling entropy production rates from biogeochemical reactions, a spatiotemporal optimal control algorithm to determine how biological structure is allocated to distributed metabolic networks expressed by microbial communities, a mathematical definition that distinguishes living from nonliving systems, and how MEP arises from Darwinian competition and the establishment of a long term (> 4 years now) microcosm experiment for testing MEP hypotheses (Vallino 2009).