Couplings of watersheds and coastal waters: Sources and ... › nrs › classes › NRS555 ›...

Estuaries Vol. 15, No. 4, p. 443-457 December 1992

Couplings of Watersheds and Coastal Waters:

Sources and Consequences of Nutrient

Enrichment in Waquoit Bay, Massachusetts

IVAN VALIELA

�9 KENNETH FOREMAN

MICHAEL LAMONTAGNE

DOUGLAS ]-IERSH

JOSEPH COSTA 5

Boston Universit'r Marine Program Marine Biological Laboratory Woods Hole, Massachuseits 02543

PAULETTE PECKOL

BARBARA DEMEO-ANDRESON

Department of Biological Sciences Smith College Northampton, Massachusetts 01063

CHARLENE D ' A V A N Z O

MICHELE BABIONE

Department of Natural Sciences Hampshire College Amherst, Massachusetts 01003

C H I - H o SHAM 4

JOHN BRAWLEY

Department of Geography Boston University Boston, Massachusetts 02215

KATE LAJTHA

Biology Department Boston University Boston, Massachusetts 02215

ABSTRACT: H u m a n activities on coastal watersheds provide the major sources of nut r ien ts en te r ing shallow coastal ecosystems. Nutr ient loadings f rom watersheds are the most widespread factor that alters s t ructure and funct ion of receiving aquatic ecosystems. To investigate this coupl ing of land to mar ine systems, we are s tudying a series of subwatersheds of Waquoit Bay that differ in degree of urbanizat ion and hence ;ire exposed to widely different nu t r i en t loading rates. The subwatersheds differ in the number of septic tanks and the relative acreage of forests. In the area of our study, groundwater is the major mechanism that t ranspor ts nut r ien ts to coastal waters. Al though there is some at tenuat ion of nu t r i en t concentra t ions within the aquifer or at the sediment-water interface, in urbanized areas there are significant increases in the nu t r ien t content of groundwater a r r iv ing at the shore 's edge. The groundwater seeps or flows through the sediment-water boundary, and sufficient groundwater-borne nut r ien ts (ni t rogen in part icular) traverse the sediment-water boundary to cause significant changes in the aquatic ecosystem. These loading-dependent al terat ions include increased nut r ien ts in water, greater pr imary product ion by phytoplankton, and increased macroalgal biomass and growth (mediated by a suite of physiological responses to abundance of nutr ients) . The increased macroalgai biomass dominates the bay ecosystem th rough second- or th i rd-order effects such as al terat ions of nu t r ien t status of water columns and increas ing frequency of anoxic events. The increases in seaweeds have decreased the areas covered by eelgrass habitats. The change in habi ta t type, plus the increased frequency of anoxic events, change the composi t ion of the benthic fauna. The data make evident the importance of bottom-up control in shallow coastal food webs. The coupl ing of land to sea by groundwater-borne nu t r i en t t ranspor t is mediated by a complex series of steps; the cascade of processes make it unlikely to find a one- to-one relat ion between land use and condi t ions in the aquatic ecosystem. Study of the process and synthesis by appropr ia te models may provide a way to deal with the complexities of the coupling.

' We dedicate this paper to our friend and colleague, the late William E. Odum of the University of Virginia.

Work reported here is part of the Waquoit Bay Land Margin Ecosystems Research project, funded by the National Science Foumlation, the United States Environmental Protection Agen- cy, and the National Oceanic and Atmospheric Administration. "File work was carried out in the Waquoit Bay National Estu- arine Research Reserve, and we thank its executive director, Christine Gauh, tbr her cooperation.

s Present address: Executiw~ Office of Environmental Affairs, Coastal Zone Management, Marion, Massachusetts 02738.

I n t r o d u c t i o n

Coastal and es tuar ine waters are the most nu- t r ien t -enr iched ecosystems on ear th (Nixon et al. 1986; Kelly and Levin 1986). Nu t r i en t loading has led to increased a b u n d a n c e of" nuisance algae, reduced oxygen content o f water, dec imated shell

" Present address: The Cadmus Group, 135 Beaver Street, Waltham, Massachusetts 02154.

�9 1992 Estuarine Research Federation 443 0160-8347/92/040443-15501.50/0

4 4 4 I. Valieta et al.

Fig. 1. Growth in housing units in Massachusetts counties, 1980-1990 in relation to distance from the sea. Housing data from U.S. Census. "Distance" is calculated as the average km from the center of the county to the shoreline.

and finfish, and many other changes in structure and function of aquatic communities. The increasing delivery of nutrients to coastal waters is caused largely by the many kinds of human activities on coastal watersheds. Urbanization increases release of wastewater, which in turn increases delivery of nutrients to groundwater, streams, and coastal waters. Deforestation attendant to urbanization re- duces interception and storage of precipitation- borne nutrients, and may favor nitrification; both these effects increase loading rates to coastal waters. Agricultural land uses involve fertilizer and presence of livestock. Significant amounts of nutrients from fertilizer and manure leach into wa- terways, groundwater, and coastal bays.

Some nutrient inputs to coastal waters may orig- inate from human activities far from shore. Nitric acid in acid rain, for instance, may contribute sub- stantially to nitrogen budgets of estuaries in the Northeast United States (Culliton et al. 1989). In most cases, however, nutrient sources within watersheds are larger than contributions from remote sources (Valiela and Costa 1988).

Nutrient enrichment is the most pervasive agent for change in coastal waters, yet we lack knowledge of basic relationships between nutrient supply rates and many key ecological processes involved in function of ecosystems (Nixon et al. 1986). We cannot, for example, reliably quantify how a given

rate of nitrogen loading will affect coastal macroalgae, phytoplankton, or fish.

We need experiments to quantify the effects of nutrient loading and identify causal mechanisms. The experiments should be done at sufficiently- large space and time scales to accommodate the regional nature of the land-sea coupling and the long-term features of the processes involved. Fur- ther, because waterflow, sediment characteristics, and bathymetry modify the effects of nutrient loading, design and execution of the experiments re- quires collaboration among diverse disciplines such as geology, hydrology, and hydrography, as well as the more usual suite of biological and chemical measurements.

The density of human populations within the coastal zones of the world is increasing. Nixon (1988) compiled data that showed that populations near 14 coastal bays in the United States had increased between 1860 and 1980. The increases ranged from 3-fold to 25-fold. Moreover, population densities in coastal areas have been increasing at rates faster than those of the general population. One recent survey, for instance, showed that the population of coastal counties in the Unit- ed States was growing at a rate 3 times that of the total United States population (Culliton et al. 1989).

The characteristic increases in human populations can also be seen on Cape Cod, Massachusetts. The rate of growth in housing units in Barnstable County, for example, was 35.3% between 1980 and 1990 (United States Census data). Barnstable, which covers all of Cape Cod, is the fastest growing of all the mainland Massachusetts counties. The rate of increase in housing units during 1980-1990 in Massachusetts was about 10% for inland rural counties (lower right of Fig. 1) and for coastal counties in and around Boston (lower left Fig. 1). The latter are either fully developed or economically incapable of supporting further urbanization. In contrast, relatively rural coastal areas (Nantucket, Martha's Vineyard, and Cape Cod) show increases of 30 to 45% in housing units (top left of Fig. 1). The high rate of urbanization on Cape Cod (and similar coastal areas elsewhere) must result in more nutrient delivery to groundwater, streams, and coastal waters.

Information on mechanisms involved in the coupling of watersheds and receiving waters is of basic and applied interest. In terms of basic principles, we need to know how processes within forests, aquifers, and the sediment-water interfaces, as well as hydrodynamic flow and mixing, affect cycling, sources, and fates of biogeochemically important substances. We also need to know more about basic controls of primary and secondary producers on coastal bays. From an applied perspective, we need

Couplings of Watersheds and Coastal Waters 445

Fig. 2. Inputs and fates of nitrogen (mol N x 108 yr ') entering the watershed and traveling toward Buttermilk Bay (from Valiela and Costa 1988).

to evaluate the consequences of a given house den- si ty--or other land use--on water quality in the receiving bay, and determine how far back from shore do we need to extend management practices if we intend to protect coastal environmental quality.

We have addressed these kinds of basic and applied questions in case studies of shallow coastal bays and their watersheds by evaluating contributions of different sources of nutrients, and as- sessing the consequences of increased nutrient loading to shallow bays along the coast of Massa- chusetts. We chose to study small, shallow water bodies because these are the coastal ecosystems that are affected first. Shallow nearshore waters can be thought of as bellweathers of future trends in eutrophication, and moreover, are easier to study because of their smaller size. Results from studies in shallow ecosystems could then be used to de- velop and test hypotheses relevant to larger, hard- er-to-study systems.

In this paper we first review some of our earlier work on how human activities on coastal watersheds lead to increased nutrients in groundwater and how this in turn increases nutrient loading to

adjoining receiving waters. Then we describe an interdisciplinary study on Waquoit Bay, where we have been conducting a regional-scale experiment in nutrient loading that permits examination of how loading rates from watersheds take place, and quantifies the effects of loading on components of the coastal aquatic ecosystem.

W a t e r s h e d s as N u t r i e n t Sources

The major sources of nitrogen to coastal watersheds in our area are precipitation, fertilizers, and domestic wastewater. In earlier studies in Butter- milk Bay, a small, shallow bay on Cape Cod, Mas- sachusetts, we found that precipitation brought almost 1.2 x 10 s moles of nitrogen per year to the surface of the watershed (Fig. 2). This value was obtained by multiplying the annually-averaged volume of precipitation by the average concentration of dissolved inorganic nitrogen (NH4 + + NOs- + NO2-) (Valiela et al. 1978; Valiela and Costa 1988; Valiela et al. 1990). Fertilizer added 8.1 x 105 moles of N yr -I to the watershed surface, an estimate obtained using information on agricultural use of fertilizers and area of cropland around the bay.

4 4 6 I. Valiela et al.

Nitrogen delivered to the watershed surface can be taken up by vegetation, be denitrified in the soil, or percolate through soil. Some of the N added in fertilizer was harvested as crops, so our cal- culation of fertilizer input is somewhat of an over- estimate. Incorporation of N into aggrading forests is quantitatively important in our study area (Fig. 2), because forests of Cape Cod have been growing since the mid-1800s, when most of Cape Cod was pastureland. The rates of accumulation of nitrogen in the pine-oak forest that cover Cape Cod are about 5-10 kg N ha -~ yr -I (unpublished data, J. Melillo and colleagues). This range of nitrogen ac- cretion could account for more than half to all of the nitrogen delivered to the watershed surface (Fig. 2).

Losses of nitrogen from the watershed surface caused by denitrification in soils are likely to be relatively low. Rates of denitrification (Groffman and Tiedje 1989) in sandy soils similar to those of our study area would remove only 10 to 20% of the nitrogen reaching the watershed surface (Va- liela and Costa 1988). The 10% estimate seems most likely for soils in our study area (Peter Groff- man, personal communication).

The rough estimates we calculated for Fig. 2 suggest that, at least in our study area, most of the nitrogen delivered by precipitation is intercepted at the watershed surface. Percolation of nutrients through to the subsoil and water table should be a small fraction of the inputs.

Domestic wastewater from septic tanks provides more nitrogen than precipitation or use of fertilizers (Fig. 2). Moreover, septic tanks release nitrogen deep in the subsoil, below the root systems of nearby trees. We calculated that about 1.5 x 106 moles N yr -~ leave septic tanks in the watershed of Buttermilk Bay, move toward the water table, and then are entrained down-gradient in groundwater. This estimate was made using results from engineering studies on septic tank performance, and data on house occupancy patterns in our area (Va- liela and Costa 1988).

The nitrogen (mainly ammonium) released by septic tanks is largely nitrified as it moves toward the water table. It is highly likely that the remain- ing ammonium is adsorbed to surfaces of particles along the way (Caezan et al. 1987; Kipp 1987). It is also possible that some of the nitrate will be denitrified while passing through the aquifer (Smith and Duff 1988). Denitrification is indeed likely to take place near septic systems, where there may be both NO3- and sufficient amounts of low molecular dissolved organic matter (DOM). Such DOM is needed as an energy source by denitrifier organisms. The engineering estimates of loading from septic tanks do include losses by denitrification in

the vicinity of septic tanks. Activity of denitrifying bacteria in aquifers may be limited at greater distances from septic tanks because of low concentrations of DOM. This is a supposition that needs confirmation by further studies.

We do not know the extent of nutrient attenuation by denitrification during transport of nitrogen in groundwater. However, a rough estimate of potential nitrogen reductions in the aquifer can be made by comparing the inputs from septic tanks with the amounts of groundwater-borne nitrogen near Buttermilk Bay (Fig. 2). Concentration of nitrogen in groundwater adjacent to Buttermilk Bay were measured by sampling and analyzing groundwater obtained from shallow well points (Valiela and Costa 1988). We also had estimates of groundwater flow into the bay. From these data we calculated that 1.1 x 106 moles N yr-' were carried into Buttermilk Bay by groundwater. This value is somewhat smaller than the 1.5 x 106 moles N yr -I delivered by septic tanks (Fig. 2). This 27% difference may be caused by adsorption and denitrification. On the other hand, because our calculations are based on rough approximations, we cannot be sure that the difference is significant. More ac- curate measurements need to be done, but to be conservative we assume for the moment that all of the nitrogen that enters the aquifer from septic tanks eventually makes its way to the shore of receiving waters.

We conclude, first, that in watersheds with accreting forests, nitrogen inputs to groundwater are largely uncoupled from precipitation-borne inputs to the watershed. Second, concentrations of nutrients in groundwater seem more closely related to the density of septic tanks in the watershed. Evidence for the second conclusion is provided by Persky (1986), who carried out a survey of solute concentrations in groundwater below areas of Cape Cod with different building densities. This study found that median ni t rate concent ra t ions increased as building density increased (Fig. 3), a result that implies that nutrient enrichment of groundwater - -and hence of receiving coastal wa- t e r s -w i l l increase along with urbanization. Of course, urbanization results in felling of forests, so that, in addition to direct input of nutrients via septic tanks, urbanization may also increase loading to groundwater by removing forests that intercept precipitation-borne nutrients.

A Regional Experiment in Nutrient Loading: The Waquoit Bay LMER Project

The Waquoit Bay Land-Margin Ecosystems Re- search project (WBLMER) is a regional-scale experiment in nutrient loading in which we can ex- amine how major components of coastal landscapes


(watersheds, sediment-water interface, coastal bay) are coupled by nutrient transport and transformations. These studies involve understanding nutrient dynamics in the terrestrial components (watershed vegetat ion, soils, and aquifers), and transport and transformations of nutrients at and through the second component, the sediment-water interface. Lastly, we follow the dynamics of nutrients, and the first-, second-, and third-order effects of the different nutrient loading rates on the receiving aquatic ecosystem.

The basis of the nutrient-loading experiment is that the degree of urbanization in different subwatersheds differs markedly. We delineated subwatersheds of Waquoit Bay on the basis of water table elevations (Fig. 4). Thus we can show that the watershed of Childs River, for example, is more densely urbanized than other subwatersheds such as Quashnet River and Sage Lot Pond (Fig. 5). We intend to show that differences in the land-use mo- saic among subwatersheds result in differences in the rate of nutrient loading to groundwater, and hence to receiving coastal waters. The choice of subwatersheds therefore provides a regional-scale experiment in nutrient loading.

A second purpose in choosing subwatersheds that are at different stages in the process of change from forest to urban is to use them as a space-for-time substitution (Pickett 1988), and thereby obtain a depiction of the effects that result as nutrient loading increases over time. Development of the present differences in land use can be reconstructed from analysis of aerial photographs taken at intervals from 1938 to the present. We will ultimately provide a detailed account of different types of land use in each subwatershed, but the history of urbanization is captured by a simple count of buildings within each subwatershed over the course of time (Fig. 5).

Quantification of nutrient loading to receiving waters begins with an assessment of freshwater flow. Fresh water flows into Waquoit Bay via groundwater seepage and spring-fed streams. The flow of groundwater via seepage into the three selected subestuaries of Waquoit Bay was estimated by means of hydrological stream-tube calculations (Freeze and Cherry 1979). Flow through spring-fed streams was done using stream-gauge data (Table 1). The sum of these two sources of the fresh water is re- markably similar to estimates of annual flows of groundwater calculated using values for annual precipitation, evapo-transpirational losses, and recharge to groundwater (Table 1). The close agree- ment of these two independent estimates suggests seepage and streams account for all flow. It seems unlikely that freshwater flows beneath Waquoit Bay to the sea. Thus, we have a system where we can

Fig. 3. Nitrate concentrations in groundwater below areas of Cape Cod having different densities of buildings. Data from Persky (1986).

almost completely account for entry and exit of flesh water, and can therefore carry out the calculations needed for nutrient loading estimates.

We estimated nutrient loadings to groundwater in the three selected subwatersheds, using land-use data and procedures by Frimpter et al. (1988) and other (Table 2). In these calculations only nitrogen provided by precipitation and septic tanks is con- sidered. For each subwatershed, surface N inputs were divided into those intercepted by forests and those passing into the subsoil. These calculations were made based on estimates of forest area. Pre- cipitation N entering the subsoil was added to septic system inputs to calculate loading to the water table.

The estimated nitrogen loading rates to the different subwatersheds span a range of up to 2 orders of magnitude. The range of nitrogen loading rates that affect subestuaries of Waquoit Bay span much of the range of loading rates seen in other estuaries (cf., the span across the x-axis in Fig. 7).

To see if differences in house density, and hence in presumed N loading, actually translated into differences in nutrients concentrations in groundwater near the shoreline, we used well points to obtain samples of groundwater at or above the high tide mark. Samples were obtained from well points placed at 1-m intervals on transects visited at different times of the year, and on one occasion at intervals of 20 m to 50 m around the periphery of the subestuaries receiving water from each subwatershed. Concentrations of dissolved inorganic nitrogen (DIN) in groundwater from Childs River were an order-of-magnitude greater than those


Fig. 4. The subwatersheds in Waquoit Bay. Boundaries drawn by M. Babione and Chi-ho Sham.

from Quashnet River or Sage Lot watersheds (Ta- ble 2). The differences were most marked for nitrate, so that the comparison of subwatersheds con- stitutes an experiment in nitrate loading.

The concentrations of nitrogen measured in the well-point samples can be used to check estimates derived from loading calculations. We calculated predicted concentrations of DIN expected to appear at the water's edge assuming that outflow of

fresh water over a year is roughly equivalent to annual recharge to the aquifer, and that no nutrient attenuation occurs during transport in the aquifer ('Fable 2). The expected concentrations of dissolved inorganic nitrogen match closely the values actually found in the well-point samples (Table 2). Even though these calculations are rough, they do suggest that we can have some confidence about the magnitude of transport of nitrogen from wa-


TABLE 2. Nitrogen contributions (kg N yr -~) by precipitation and septic systems, and N loading to water table and hence to estuaries.

Subwatersheds

Q~ashet Sage Lot Childs River ~ l v e r Pond

Precipitation Intercepted by forests 1,092 4,809 114 To soil in nonforest areas 1,209 2,634 11.5

Septic systems 13,000 11,900 320 Loading to water table 14,209 1 4 , 5 3 4 331.5 Calculated DIN conc (M)' 160 3.6 6.2 Measured DIN conc (M) in

well point samples (geometric mean) 132.7 4.2 15.1

Assuming that seepage plus stream flow equals recharge.

Fig. 5. ttistorical trend in the degree of urbanization of three different subwatersheds of Waquoit Bay. Data from aerial photographs. Cf. Fig. 4 for the location of the subwatersheds.

te rshed t h rough the aquifer , and to the water ' s edge.

Consequences of Nutrient Loading in Waquoit Bay

Above we reviewed evidence on delivery of" nutrients, par t icular ly o f ni t ra te addit ions, f rom watersheds to the water ' s edge of Waquoi t Bay, and concluded that dif ferences in urbaniza t ion led to di f ferent concen t ra t ions of ni t ra te in g roundwa te r moving f r o m the watershed to receiving waters. Below we focus on how the di f ferent rates o f ni t ra te loading change some key ecological componen t s of the aquatic c o m p o n e n t s o f the th ree selected subwatersheds of Waquoi t Bay (Childs River, Quashne t River, and Sage Lot Pond).

NUTRIENTS IN WATER INCREASE WITH URBANIZATION OF WATERSHEDS

Concen t ra t ions of" ni t ra te in the water co lumn o f Childs River were larger than those o f Quashne t River; concen t ra t ions of ni t ra te in Sage Lot Pond were the lowest (Fig. 6). T h e negat ive re la t ionship

to salinity suggests that the source o f ni t ra te must be the f reshwater e n d - m e m b e r o f the estuaries. Clearly, the ni t ra te must come f rom the watershed, e i ther f r o m seepage a long the shores of the subestuary or f rom the spr ing-fed streams.

T h e t ranslat ion be tween loading to a wate rshed and condit ions in the water is ar t icula ted t h rough a series of steps. T h e first step involves the en- t r a inment and possible a t tenua t ion o f n i t rogen in flowing g roundwate r . T h e total quant i ty of ni trogen furn ished by wastewater and precipi ta t ion to the aquifers o f the Childs River and Quashne t Riv- er subwatersheds is abou t the same (Table 2). How- ever, the Quashne t River wate rshed is la rger than the Childs River watershed (Table 2). Consequent - ly, the re is m o r e water involved in nu t r ien t transpor t within the aquifer o f the Quashne t River watershed. We calculate mean concent ra t ions o f ni t ra te in g r o u n d w a t e r abou t to en te r Quashne t River by dividing the loading ra te by the r echa rge volume, and assuming that n i t rogen mixes uni- formly within the aquifer . T h e calculated concen- t ra t ions o f n i t rogen in g r o u n d w a t e r abou t to dis- cha rge into the Quashne t es tuary are much lower than those calculated for Childs River (Table 2). Because the assumpt ion of mixing within the aquifer is too simple, we are present ly using g round-

TABLE 1. Annual flows of groundwater from watersheds into subestuaries (Childs River, Quashnet River, Sage Lot Pond) of Waquoit Bay. Data calculated byJ. Brawley, C.-H. Sham, and I. Valiela.

Childs River Quashnet River Sage Lot Pond

2,657 59 Surt'ace area (ha) 687 Total flow of groundwater calculated from stream gauges and aquifer data

Groundwater flow via stream (m s yr-0 (% of total)

Groundwater flow via seepage (m s yr -~) (% of total)

S u m

Total flow of groundwater calculated from annual rain and recharge

2.01 x 106 10.91 x 106 (56) (80.4) - -

1.56 • 106 2.66 • 106 (44) (19.6%) 3.76 x 10'

3.57 • 106 1.36 • 107 3.76 • 106

3.1 • 106 1.2 x 107 2.67 x 10 ~


Fig. 6. Concen t r a t ion o f n i t ra te vs. salinity in water co lumns o f Childs River, Q u a s h n e t River, and Sage Lot Pond. Data f rom samples taken over the course o f a year by K. Foreman .

water flow and transport models to incorporate a more realistic set of conditions for nutrient move- ment within the aquifer. In any case, the calculations do show that loading rates can be sharply modified by the volume of recharge water; there is no one-to-one relationship between loading from land use and nutrient concentrations discharged into coastal waters.

The second step in the translation occurs in the shallow sediments where groundwater enters the estuaries. Groundwater about to enter Childs Riv- er holds about 133/~M NO3- (Table 2), but these concentrations were not found in the receiving water column (Fig. 6). At most, extrapolation~ of the nitrate vs. salinity curve for Childs River in Fig. 6 yields about 60 ~M at salinity of 0~/~. This suggests a loss of about half the dissolved inorganic nitrogen during passage of groundwater through the thin sediment-water interface. Denitrification could be responsible for these losses (Seitziger 1988). Such losses are not seen in the Quashnet River data (Ta- ble 2 and Fig. 6). Perhaps, as argued by Seitzinger and Nixon (1985) and Seitzinger (1988), rates of nitrification and denitrification increase as nitrogen loading rates increase. If such were the case, we might expect that denitrification rates are higher in Childs River than in Quashnet River and Sage Lot Pond.

Differences in location of homes (and hence septic tanks) may be an alternative explanation for the discrepancy between measured concentrations in

Fig. 7. Distr ibut ion o f n u m b e r o f h o m e s (N) at d i f ferent d is tances f rom the shore l ine o f Waquoi t Bay. Data f rom aerial photos taken in 1938, 195.6, 1979, and 1989 are included in bars in indicate the history o f u rban iza t ion in the area. Data o f M. Babione and C.-H. Sham.

groundwater .near the shore and in regional estimates. Perhaps the nearshore concentrations are higher because of the greater density of houses very close to the water's edge (Fig. 7). Further, a location away from shore also means longer travel time for nitrate in the groundwater. Average ve- locity of groundwater in our area is about 0.3 m d -~, so a distance of only 300 m away from shore implies a year's travel time. If there is any attenuation along the travel path, the importance of septic tanks located farther away could be propor- tionally smaller. Thus, septic tanks near the water's edge are probably the most critical for nutrient loading. This idea needs testing with actual measurements of attenuation and with groundwater models.

PHYTOPLANKTON PRODUCTION AND BIOMASS INCREASE WITH ENRICHMENT

Production by phytoplankton varied seasonally, and reached 100-200 mg C m -3 h -I in Childs River during middle to late summer. These are consid-

T A B L E 3. Chlorophyl l concen t ra t ions (mean + SE) in water f rom stat ions located u p s t r e a m and at the m o u t h o f each o f t h ree subes tuar ies o f Waquoi t Bay. Each m e a n was calculated f rom eight m e a s u r e m e n t s ob ta ined at intervals over a one-year per iod. Data o f J. Costa.

Chlorophyll concentrations (mg m s)

Subestuary Upstream Sites Mouth Sites

Childs River 47.7 +- 21.1 25.5 + 7.6 Q u a s h n e t River 8.6 + 2.6 5.9 + 1.7 Sage Lot Pond - - 3.9 + 1.2


Fig. 8. Potential input of nitrogen entering the ecosystem related to mean phytoplankton chlorophyll in water in subestuaries of Waquoit Bay (open circles; Sage Lot Pond, Quashnet River, Childs River, in increasing order along the loading axis) and in other coastal waters [data compiled by Nixon and Pilson, and modified from Valiela (1991)].

erably higher rates than the 10-50 mg C m -~ h -~ measured in Sage Lot Pond during the peak of the season. At other times of year, phytoplankton production was very low, generally less than 1 mg C m -3 h -~ everywhere in Waquoit Bay.

Standing crops of phytoplankton, measured as chlorophyll, are higher in Childs River than that in Quashnet River, and both of these estuaries in turn have more phytoplankton chlorophyll than Sage Lot Pond (Table 3), the estuary that is exposed to the lowest potential nutrient loading (Ta- ble 2). The chlorophyll values are averages of mean, in tegrated water-column chlorophyll measurements taken over many sampling periods during 1988-1989. There are strong down-estuary gra- dients in chlorophyll concentrations in our subestuaries (Table 3), but the differences between Childs River, Quashnet River, and Sage Lot Pond are maintained down-estuary.

The responses of average chlorophyll in the water to potential nutrient loading rate in our estuaries fall within the ranges observed in other estuarine environments (Fig. 8). The values for Childs River, the most nutrient-loaded estuary, lie within the scatter around the line fitted to observations from other estuaries. Although all data from Wa-

quoit Bay lie within the scatter of Fig. 8, those of Quashnet and Sage Lot Pond lie below the fitted line. Perhaps nutrient uptake by macroalgae or denitrification in these estuaries are high enough to lower phytoplankton chlorophyll. We are cur- rently testing these hypotheses.

MACROALGAE INCREASE W I T H ENRICHMENT

Easily the most obvious response of the Waquoit Bay estuary ecosystems to nutrient loading is the proliferation of benthic macroalgae. Total macroalgal biomass, for example, increases 3- to 4-fold between Sage Lot Pond and Childs River (Table 4).

The increases in macroalgal biomass in nutrient- loaded estuaries are the result of alteration to physiological features of macroalgae. Cladophora vagabunda is one of the two major species of macroalgae in Waquoit Bay (Table 4). C. vagabunda from Childs River, the most nutrient loaded site (Table 4), have higher photosynthetic rates at any irradiance com- pared to C. vagabunda from Sage Lot Pond, the water body receiving the lowest nutrient load (Fig. 9, top).

C. vagabunda from Childs River also shows faster nitrogen uptake rates than C. vagabunda from Sage Lot Pond, regardless of the concentrations of ammonium in which they were incubated (Fig. 8, bottom). The asymptotic uptake rate of Childs River C. vagabunda was about twice that of the Sage Lot Pond population.

Differences in physiological performance such as shown in Fig. 9 result in differences in growth rate (Table 5). The Childs River C. vagabunda grow faster than those in Sage Lot Pond when exposed to saturating irradiences. Within the macroalgal canopy there is less light and slower growth rate. Since algae deep within the canopy appear to be healthy and active, there must be sufficient physical disturbances to turnover the canopy so that algae are exposed to sufficient light at frequent enough intervals.

Physiological responses to enhanced nutrient supply, on aggregate, result in large accumulations of macroalgae. Although macroalgae were present in the bay in the past (Curley et al. 1971), biomass and cover were less than at present. The cover of macroalgae over the bay is now very extensive (Va-

TABLE 4. Mean (+ SD) of biomass of macrophytes in three selected subestuaries of Waquoit Bay. Data averaged from multiple samples per site collected monthly over one year. Data of D. ltersh.

Number of Biomass (g m 2) Subestuary Samples Cladophora vagabunda Gracilaria tikoahiae Total Macroalgal Biomass Zostera marina

Childs River 131 220 + 29.4 109 _+ 10.5 335 + 39.8 0 Quashnet River 130 85 _+ 11.1 48 _+ 5 150 +_ 14.3 0 Sage Lot Pond 78 36 _+_ 6 36 -+ 5.9 90 + 12.1 117 + 12.6


T A B L E 5. Mean (_+_ SD) g rowth rates tbr Cladophora vagabunda f rom Childs River and Sage Lot Pond. M e a s u r e m e n t s were done in situ as increases in fresh weight in t e the red f ronds s i tuated above the canopy and within the canopy.

Growth Rates (Doublings d ~)

Above Canopy Within Canopy Location (High Light Intensity) (Low Light Intensity)

Childs River 0.09 -+ 0.02 0.04 -+- 0 . 0 2 Sage Lot Pond 0.04 _+ 0.02 0.03 + 0.02

Fig. 9. Net pho tosyn thes i s vs. light intensity (top) and ni- t rogen up take rate vs. concen t ra t ion in water (bot tom) for Cla- dophora vagabun<~a collected in Childs River and f rom Sage I.ot Pond. Data o f P. Peckol.

liela et al. 1990), and in some places within Waquoit Bay the algal canopy may be up to 75 cm thick (unpublished data). In addition to the enhanced biomass, there is a marked shift in species composition of the macrophyte canopy. Nutrient loading eliminates eelgrass and fosters growth of a green (Cladophora vagabunda) and a red (Gracilaria tikvahiae) algal species (Table 4). The current abundance of Gracilaria tikvahiae is of particular interest, since it was not mentioned in a survey done in 1967 (Curley et al. 1971).

Changes in biomass and composition of vegetation such as shown in Table 4 have significant repercussions throughout the bay ecosystem. For example, we calculate that at measured uptake rates (from Fig. 9) and an average ambient 20 gM DIN, the approximately 200 g m -~ of C. vagabunda in Childs River would take up about 4 • 101~ -< moles DIN in the Childs River estuary. There are about 7.5 x 109 #moles DIN in the Childs River estuary, on average, at any one instant. It is difficult to compare standing crops to uptake rates, but for

comparison's sake, we can see that the potential uptake over a day, by only one of the algal taxa present in Childs River, is an order of magnitude higher than the stock of DIN. The nitrogen loading rate (Table 2), when converted to appropriate units, is 2-8 • 10 l~ ~mol N d -I. Loading rate is therefore of the same magnitude as the potential uptake by C. vagabunda, which of course is only one of the macroalgal species present in Childs River. All these comparisons are rough and ignore features such as regeneration of DIN, but they do suggest the potential importance of macroalgae in nutrient dynamics in our estuaries. In fact, uptake by macroaigae, we believe, is responsible for low nutrients in the water of Waquoit Bay during warmer months (unpublished data).

A second example of the importance of macroalgae is their effect on oxygen content of the water. Photosynthesis and respiration by macroalgae determine the vertical profiles of oxygen in Waquoit Bay and its estuaries. This is unusual, since physical mixing by wind, currents, and tides commonly overwhelm biological processes and largely determine oxygen content in coastal waters. Evidence for the role of macroalgal control of oxygen content is presented in detail elsewhere (D'Avanzo et al. in preparation), but Fig. 10 is a clear example. At dawn, oxygen is depleted near the bottom (Fig. 10, black circles). As the sun comes up, photosynthesis by phytoplankton and macroalgae generates oxygen, and the profile of oxygen concentrations slides to the right (Fig. 10, open circles). The dom- inant effect of macroalgal photosynthesis is evident from the pattern of" the daily swing; the largest change in oxygen occurs just above the macroalgal canopy. After dusk, respiration consumes oxygen, and the daily fluctuation is completed. Of course, there are many kinds of organisms in the bay, all of which respire, but unpublished data of P. Peckol suggest that respiration by the macroalgae may account for most of the oxygen depletion that takes place during the night.

Phytoplankton of course are partially responsible for the excursion of oxygen concentrations seen in Fig. 10. The increased oxygen within the upper


Fig. 10. Vertical profiles of oxygen in Childs River during dawn and midafternoon of a sunny day (circles), and during afternoon after three cloudy days in a row. Data from C. O'Avanzo.

por t ion o f tile water column is probably due in some measure to phytop lankton product ion. K. Foreman ' s measurements of phytoplankt0n product ion suggest that dur ing most o f the year phytoplankton are not as p roduc t ive as macroalgae, and account for about 20% of total pr imary production. During a few weeks in mid to late summer phytoplankton peak, and cont r ibu te perhaps 70% of the oxygen.

I f there is a sequence of a few cloudy days, photosynthesis by the macroalgae may be low enough that the dayt ime recovery o f oxygen content seen in Fig. 10 does not take place. In as little as th ree cloudy days, the consumpt ion o f oxygen by respi- rat ion uncompensa ted by photosynthesis can pro- duce hypoxic or anoxic condit ions (Fig. 10, tri- angles) near the bo t tom, or even over the ent i re water column.

Anoxic events have far-reaching second- and th i rd -order effects t h roughou t the food web o f Wa- quoit Bay. During the summer o f 1988, for example, we r eco rded a p ro longed overcast per iod July 19-23 (Fig. 11, top graph). This occur red

Fig. 11. Time course of light, temperature, oxygen, ammonium, phosphate, and chlorophyll in water of Waquoit Bay, summer 1988. Modified from Costa et al. (in press).


Fig. 12. Changes in eelgrass distr ibution in Waquoit Bay, 1951-1987. Black areas are eelgrass beds with cover near 100%. T h e diagonal stripes in 1951 map show an area o f patchy eelgrass cover. T h e dot ted ones in 1951 map refer to parts o f pho tograph where it was impossible to in terpre t the information. From Costa et al. (in press).

during a time when the water was warm (Fig. 11, second graph) and the water column was strongly stratified. During this period, net production by macroalgae was nil, while macroalgal respiration exceeded 80 -< mol O~d-' h-' (Peckol et al. in preparation). The oxygen content of surface water decreased sharply after that cloudy period (Fig. 11, second graph). The lower oxygen was accompa- nied by increases in ammonium and phosphate concentrations (Fig. 11, third graph), presumably caused by increased regeneration. The increase in nutrients was followed almost immediately by a bloom of phytoplankton growth, as shown by the chlorophyll data (Fig. 11, bottom graph). We have unpublished data that show that phytoplankton are nitrogen-limited in the more saline areas of Wa- quoit Bay. Anoxic events thus have repercussions throughout key components of aquatic ecosystems. Anoxic events may occur on occasion in unen- riched shallow coastal water bodies, even if macroalgae are scarce, and these events have effects on their ecosystems also. The point here, however, is that abundant macroalgae may increase the frequency of such events and hence increase the consequent changes in nutrient availability and phytoplankton as well as throughout the food webs.

Enrichment leads to dominance of macroalgae,

and the effect of macroalgae on the vertical distribution of oxygen therefore poises the bay ecosystem in a rather unstable state, one in which there is an increase in the frequency of anoxic events that can severely alter the trophic web and geochemistry of the bay. Consumers, as well as nutrients and producers, are affected by increased fre- quencies of anoxia. Anoxic events can result in extensive kills offish and invertebrates. In Waquoit Bay we have recorded kills following anoxic events in each of the summers during our study (1988 to present). Nutrient loading thus leads to a cascade of fairly complicated second- and third-order effects.

Loading rates are also likely to prompt third- order effects on bacteria. One of the always puz- zling features of denitrification is the paradox of how anaerobic bacteria manage to find sufficient nitrate, a compound that is obviously in short supply in anaerobic environments. Perhaps one reason for the increases in denitrification we have already mentioned is that in enriched waters there are wide daily fluctuations in oxygen conditions, as implied in Fig. 10. These fluctuations could allow nitrate to reach the vicinity of anoxic microzones where denitrifiers can survive and somehow have access to the new nitrate.

EELGRASS ABUNDANCE IS REDUCED BY LOADING

Eelgrass is absent from nutrient-loaded subestuaries of Waquoit Bay (Table 4). This is a second- order phenomenon rather than a direct conse- quence of nutrient additions (Costa 1988). Growth of" eelgrass is light-limited rather than nutrient- limited in Cape Cod (Dennison and Alberte 1985). Nutrient enrichment encourages growth of epiphytes on eelgrass leaves, and epiphytes intercept light (Costa 1988). The potential result is a decrease in vigor of eelgrass, with eventual reduced cover, and loss of the eelgrass habitat.

There is a historical parallel to the differences in eelgrass cover we found in comparing the subwatersheds exposed to different loading rates. Eel- grass habitat in Waquoit Bay has decreased (Fig. 12), while the number of dwellings and the result- ing potential nutrient loading have increased during recent decades. The data of Fig. 12 were obtained from aerial photographs using interpretative methods developed by Costa (1988). We cannot be sure that the decrease in eelgrass was the direct result of nutrient loadings, but there is a negative correlation between nutrient loading calculated from the number of houses in the watershed and the loss of eelgrass cover in the bay (unpublished data). The disappearance of eelgrass over recent decades is documented by long-time residents, who recall that Childs River was an eelgrass-dominated estuary before the 1960s. Moreover, in the space- for-time substitution afforded by comparisons of the subestuaries (Table 4), it is clear that macroalgae replace eelgrass in the bottom of enriched estuaries.

ABUNDANCE OF ANIMALS DECREASES W I T H LOADING

The abundance and species richness of invertebrates found on the bottom and macroalgal canopy are lower in parts of Waquoit Bay in which more macroalgae are present (Fig. 13). These results are from replicated samples obtained using an Ekman dredge at several sites over Waquoit Bay.

We can show a historical parallel to the spatial comparison of Fig. 12, using data from shellfish catch reports. The catch of bay scallops (Argopecten irradians) from Waquoit Bay have decreased since the 1960s (Fig. 14). The decrease has been even more marked in Eel Pond, which is on the western section of the bay and receives loadings from the Eel Pond and Childs River subwatersheds (Fig. 4), the two most urbanized estuaries of Waquoit Bay. The scallop yield from the more urbanized water- body decreased more than that in the relatively less urbanized Waquoit Bay proper (Fig. 14).

On the other hand, the catch of quahogs (Mer- cenaria mercenaria) in Waquoit Bay and Eel Pond

Couplings of Watersheds and Coastal Waters 45S

Fig. 13. Abundance and species richness of invertebrates from various sites in Waquoit Bay (samples of macroalgae from which animals were obtained were collected with an Eckman dredge: data from J, Buckley, A. Forrester, and R. Siwko). The sample point with lowest algal biomass and highest richness and abundance was taken in the part of the bay where eelgrass still exists.

has remained about the same over the last several decades (Fig. 14), so not all benthic invertebrate species are affected similarly. There may be some differential tolerance for low oxygen between these two shellfish species. Perhaps more likely to account for the difference between the responses of scallops and clams may be differential effects of eutrophication on the specific habitats used by these


Fig. 14. Reported catch of two major species of shellfish, scallops (Argopecten irradians) and quahogs (Mercenaria mercenaria), from Waquoit Bay proper and from Eel Pond. Data from Town of Falmouth Annual Reports, compiled by the Shellfish Warden. One bushel equals 35.2 1.

two stocks of shellfish: clams use the very shallow sandy littoral strips, and scallops depend on eelgrass beds in deeper areas. Eutrophication has, for some reason, not greatly altered the vegetation cover of littoral sands in Waquoit Bay. The strip of littoral sands is still largely devoid of vegetation. In contrast, the favored habitat of scallops has been much reduced by nutrient enrichment (Fig. 12).

Macrolgal canopies are rather inhospitable habitats for scallops--scallops that move into macroalgal beds sink into the thick canopy of loose fronds. Suspension feeding within the canopy must be im- paired, and water within the canopy often becomes anoxic on a daily basis. Scallops that wander into the macroalgal beds certainly are unlikely to thrive there.

It may be argued that the decrease in scallop catch reported in Fig. 14 is a regional recruitment phenomenon, perhaps unrelated to local eutrophication. This may be so, but even if there were a sudden increase in larval recruitment of scallops into Waquoit Bay, we suspect that the great re-

duction in eelgrass habitat brought about by eutrophication would make it difficult for the current population or for new recruits to find sufficiently large areas of suitable habitat. In addition, the increased frequency of anoxia must deplete scallop populations.

General Conclus ions

Nutrient loading has had pervasive repercussions throughout the food web of the Waquoit Bay ecosystem. The first-, second-, and third-order effects associated with different degrees of nutrient loading result in very different configurations of the food web, differences in nutrient and carbon cycling, elemental budgets, and input-output budgets. Although we have enough knowledge to show that the differences exist, we need further study to establish cause and effect mechanisms and to quantify actual trends created by increased nutrient loading.

The suite of effects that seem related to nutrient enrichment in Waquoit Bay strongly imply the im-

portance of "bottom-up" controls in this aquatic ecosystem. We have as yet not explored the importance of "top-down" controls.

Our data so far suggest that land-use mosaics on coastal watersheds, even though separated by con- siderable distances from receiving waters, are strongly coupled to adjoining aquatic ecosystems, and this coupling has fundamental consequences for the structure and function of the receiving water ecosystem. Human activities on the watersheds work to increase "natural" nutrient sources and transport, and are ultimately responsible for increases in nutrient delivery to coastal waters.

Ideally, we would like to have a suitable index of land use, for example, "number of houses" or "area of accreting forests," or some combination of these variables. We could then search for a cor- respondence of the index to water quality in the receiving water body. As is obvious from the dis- cussion above, one-to-one translations of land use to nutrient loading, and to consequences for water quality and resources are unlikely. The many linked mechanisms, and their first-, second-, and third- order effects, make the finding of such correspon- dences a challenging task. Many ostensibly "obvious" relationships fail to materialize. There is no doubt that the translation exists. Comprehensive, interdisciplinary study and modelling of the inter- actions, and of the responses of processes involved, promise timely answers to the central question of how land use on watersheds is coupled to consequences in the receiving aquatic ecosystem.

LITERATURE CITED

CAEZAN, M. C., E. M. "I'HURMAN, AND R. L. SMI'I'tt. 1987. The role of cation exchange in the transport of ammonium and nitrate in a sewage--contaminated aquifer, p. B53-B58. In Toxic Waste and Groundwater Contamination Program, 3rd "Technical Meeting. United States Geological Survey, Boston, Massachusetts.

COSTA, J. E. 1988. Distribution, production, and historical changes in abundance of eelgrass (Zostera marina) in south- eastern Massachusetts. Ph.D. Thesis. Boston University, 352 p.

COSTA, J. E., B. L. IIowzs, A. E. GmLIN, AND I. VALIELA. In press. Monitoring nitrogen and indicators of nitrogen to sup- port management action in Buzzards Bay. In D. II. McKenzie, D. E. Hylact, and V.J. McDonald (eds.), Ecological Indicators. Volume 1. Elsevier Press, London.

CULLITON, T. J., C. M. BLACKWELL, D. G. REMER, T. R. GoonSPEEI), AND M. A. WARREN. 1989. Selected characteristics in coastal states, 1=980-2000. NOAA's Coastal Trends Series: Report 1. National Oceanic and Atmospheric Admin- istration, Strategic Assessment Branch, Rockville, Maryland. 15p.

CURLEY, J. R., R. P. LAWTON, J. M. HIGK~'I', AND J. D. FISKE. 1971. A study of the marine resources of the Waquoit Bay- Eel Pond Estuary. Department of Marine Fisheries, Sandwich Massachusetts. 40 p.

DENNISON, W. C. AND R. S. ALBERTE. 1985. Role of daily light


period in the depth distribution of photosynthetic response of Zostera marina (eelgrass). Marine Ecology Progress Series 25: 51-61.

FREEZE, R. A. ANDJ. A. CHERRY. 1979. Groundwater. Prentice- llall. Englewood Cliffs, New Jersey. 604 p.

FRIMPTER, M. H . , j . j . DONOHOE, AND M. V. RAPACZ. 1988. A mass balance nitrate model for predicting the effects of land use on groundwater quality in municipal wellhead protection areas. Cape Cod Aquifer Management Project. United States Geological Survey, Marlboro, Massachusetts. 54 p.

GROFFMAN, P. M. AND J. M. TIEDJE. 1989. Denitrification in North Temperate forest soils: Spatial and temporal patterns and the landscape level and seasonal sinks. Soil Biology Bio- chemistry 21:613-620.

KELLY, J. R. ANn S. A. LEVIN. 1986. A comparison of aquatic and terrestrial nutr ient cycling and production processes in natural ecosystems, with reference to ecological concepts of relevance to some waste disposal issues, p. 165-203. In G. Kullenberg (ed.), The Role of the Oceans as a Waste Disposal Option. NATO Advanced Research Workshop Series. D. Reidel Publishing Company, Dordrecht.

KIPP, K. !.. 1987. Preliminary one-dimensional simulation of ammonium and nitrate in the Cape Cod sewage plume, p. B43-B44. In Toxic Waste and Groundwater Contamination Program, 3rd Technical Meeting. United States Geological Survey, Boston, Massachusetts.

NIXON, S. W., C. A. OVIATT, J. FRVrHSEN, AND B. SULLIVAN. 1986. Nutrients and the productivity ofestuarine and coastal marine ecosystems. Journal of the Limnological Society of South Africa 12:43-71.

NIXON, S. W. 1988. Physical energy inputs and the comparative ecology of lake and marine ecosystems. Limnology and Ocean- ography 33(pt.2): 1005-1025.

PERSKV, J. H. 1986. The relation of ground-water quality to housing density, Cape Cod, Massachusetts. U.S.G.S. Water- Resources Investigations Report 86-4093, Boston, Massachu- setts. 29 p.

PICKE'r~', S. T. A. 1988. Space for time substitution as an alternative to long-term studies, p. 110-135. In G. E. Likens (ed.), Long-Term Studies in Ecology. Springer-Verlag, New York.

SEITZINGER, S. P. 1988. Denitrification in freshwater and coastal marine ecosystems: Ecological and geochemical significance. Limnology and Oceanography 33:702-724.

SEITZINGER, S. P. ANn S. W. NIXON. 1985. Eutrophication and rate of denitrification and N20 production in coastal marine sediments. Limnology and Oceanography 30:1332-1339.

SMITH, R. L. ANnJ. H. DUTY. 1988. Denitrification in a sand and gravel aquifer. Applied EnvironmentalMicrobiology 54:1071- 1078.

VALIV.LA, I. 1991. Ecology of water columns, p. 29-56. In R. S. K. Barnes and K. H. Mann (eds.), Fundamentals of Aquatic Ecology. Blackwell, Oxford.

VALIELA, I. AND J. COSTA. 1988. Eutrophication of Buttermilk Bay, a Cape Cod coastal embayment: Concentrations of nutrients and watershed nutr ient budgets. Environmental Man- agement 12:539-551.

VALIELA, 1., J. TEAL, S. VOLKMANN, D. SHAFER, AND E. CAR- PENTER. 1978. Nutrient and particulate fluxes in a salt marsh ecosystem: Tidal exchanges and inputs by precipitation and groundwater. Limnology and Oceanography 23:798-812.

VALIELA, I.,J. COSTA, K. FOREMAN,J. M. TEAL, B. HOWES, AND D. AUBREY. 1990. Transpor t of groundwater-borne nutrients from watersheds and their effects on coastal waters. Bin- geochemistry 10:177-197.

Received for consideration, September 1, 1990 Accepted for publication, March 13, 1992

Couplings of watersheds and coastal waters: Sources and ... › nrs › classes › NRS555 ›...

Documents

Transcript of Couplings of watersheds and coastal waters: Sources and ... › nrs › classes › NRS555 ›...