MARINE ECOLOGY PROGRESS SERIES
   Vol. 247: 59–73, 2003                                           Published February 4
                           Mar Ecol Prog Ser




Eelgrass Zostera marina loss in temperate estuaries:
 relationship to land-derived nitrogen loads and
   effect of light limitation imposed by algae
            Jennifer Hauxwell1, 2, 4,*, Just Cebrián1, 3, 5, Ivan Valiela1
     1
      Boston University Marine Program, Marine Biological Laboratory, Woods Hole, Massachusetts 02543, USA
2
   Wisconsin Department of Natural Resources, DNR Research Center, 1350 Femrite Drive, Monona, Wisconsin 53716, USA
      3
       Dauphin Island Sea Lab, 101 Bienville Boulevard, PO Box 369-370, Dauphin Island, Alabama 36528, USA
    4
   Present address: Wisconsin Department of Natural Resources, DNR Research Center, 1350 Femrite Drive, Monona,
                         Wisconsin 53716, USA
5
 Present address: Dauphin Island Sea Lab, 101 Bienville Boulevard, PO Box 369-370, Dauphin Island, Alabama 36528, USA




      ABSTRACT: In this paper, we explicitly link changes in community structure of estuarine primary pro-
      ducers to measured nitrogen loading rates from watersheds to estuaries, and quantify the relationship
      between nitrogen load, annual dynamics of algal growth and Zostera marina L. productivity, and over-
      all eelgrass decline at the watershed-estuarine scale in estuaries of Waquoit Bay, Massachusetts, USA.
      Substantial eelgrass loss (80 to 96% of bed area lost in the last decade) was found at loads of ~30 kg N
      ha–1 yr–1, and total disappearance at loads ≥ 60 kg N ha–1 yr–1. Rather than decreased eelgrass growth
      rates, we observed an exponential decrease in shoot densities and bed area (and subsequently, areal pro-
      duction) as nitrogen loads increased, suggesting that eelgrass decline in higher-nitrogen estuaries of the
      Waquoit system occurred largely via lack of recruitment or enhanced mortality of established shoots.
      Similar to the patterns observed in many other systems and the experimental results obtained in labo-
      ratories or mesocosms, the relationship we observed between nitrogen loads and eelgrass health within
      the Waquoit system was indirect: increased nitrogen stimulated growth and standing stocks of algal pro-
      ducers, that may have caused severe light limitation of eelgrass. From light budgets that considered
      water column, epiphyte, and macroalgal shading, we estimated chronic, severe light limitation to newly
      recruiting shoots in higher-nitrogen estuaries, due mainly to shading by a coexisting ≤15 cm macro-
      algal canopy. Two management recommendations aimed at eelgrass preservation emerge from this
      work. First, development and management of watersheds must be conducted such that land-derived
      nitrogen loading to estuaries is restricted. In the Waquoit Bay estuaries, for example, eelgrass is absent
      or rapidly disappearing from all but those receiving the lowest (≤15th percentile) loads. Second, shoot
      density and meadow area, rather than growth rates per shoot, seem to be adequate variables for routine
      monitoring of eelgrass health. We also show that the shift from eelgrass- to algae-dominated commu-
      nities has important consequences for total system primary production and carbon and nitrogen cycling.
      Estimated total primary production by coastal assemblages in the Waquoit Bay system was 135% higher
      in estuaries receiving relatively high versus low loads of land-derived nitrogen, suggesting important
      trophic and biogeochemical alterations to temperate estuarine ecosystems as a result of eutrophication.

      KEY WORDS: Seagrass · Macroalgae · Epiphytes · Phytoplankton · Irradiance · Waquoit Bay ·
      Eutrophication · Estuary
                Resale or republication not permitted without written consent of the publisher




            INTRODUCTION                   induced. Natural disturbances, such as hurricanes,
                                     earthquakes, ice scour, bioturbation, and herbivory,
 Eelgrass Zostera marina L. habitat has been lost from         may account for a small percentage of worldwide loss
temperate estuaries worldwide. This loss has occurred          (Short & Wyllie-Echeverria 1996). Wasting disease
via several mechanisms, both natural and human-             caused an extensive die-off in the 1930s along the

*Email: jennifer.hauxwell@dnr.state.wi.us                © Inter-Research 2003 · www.int-res.com
60                     Mar Ecol Prog Ser 247: 59–73, 2003




Atlantic coastlines of North America (Cottam 1933,      the available literature indicates that most often, light
Cotton 1933, Renn 1935) and Denmark (Petersen 1934,      limitation is a primary mechanism of eelgrass decline
Rasmussen 1973), but is now only locally important      under enhanced eutrophication (see reviews by Duarte
(Short et al. 1986). Human-induced disturbances, such     1995 and Cloern 2001).
as dredging, addition of docks, mooring of boats, har-      The associations between increased nitrogen load-
vesting of shellfish using rakes or trawls, and use of    ing, light interception by algal producers, and seagrass
motorboats in shallow waters have created ‘scarred’      decline have been largely established with experi-
areas within eelgrass meadows. Sediment transport       mental manipulations in laboratory microcosms and
and herbicide runoff as a result of development and      mesocosms (Twilley et al. 1985, Burkholder et al. 1994,
agricultural activities in adjoining land parcels may     Neckles et al. 1994, Short et al. 1995, Taylor et al.
also have affected eelgrass habitat (Kemp et al. 1983).    1995a,b, Moore & Wetzel 2000). Large-scale compar-
  Anthropogenic nitrogen loading from watersheds to     isons, integrating changes in nitrogen delivery, algal
estuaries, or increased delivery of nutrients into re-    growth, and eelgrass loss at the watershed –estuarine
ceiving waters generated by human activities, may       level are lacking. Development of numerical relation-
also be an important cause of eelgrass loss (Short et al.   ships between nitrogen loading and response of estu-
1995, Valiela et al. 2000b, Cloern 2001, Hauxwell et al.   arine producers at the watershed –estuary scale is
2001). Increased nutrient loads into estuarine waters     important for 3 reasons. First, it is necessary for under-
result from the disproportionate increase in human      standing how interactions among nutrients, algae, and
population near the coast (as compared to inland pop-     seagrasses observed under laboratory conditions trans-
ulations) and associated transformation of natural land    late into natural, large-scale scenarios. Second, it can
into urban development, agricultural land, and recre-     assist in developing efficient management practices for
ational facilities (i.e. golf courses: Nixon 1995, Cloern   eelgrass preservation. For instance, knowledge of the
2001). Recent estimates reveal that 40% of the world’s    threshold land-derived nitrogen loading rate at which
population live within 100 km of the coastline (Cohen     eelgrass declines, and understanding the proximate
et al. 1997), and it is predicted that this imbalance will  causes for such decline, would help in formulating
become greater, because coastal populations have       policies to manage nitrogen loads to estuaries. Because
faster growth rates than inland populations. In fact,     eelgrass restoration may be difficult (Harrison 1990,
anthropogenic nitrogen loading is now viewed as one      Davis & Short 1997, Davis et al. 1998), development of
of the most pervasive, world-wide, human impacts on      indicators of incipient eelgrass decline in response to
estuaries (NRC 1994, Jackson et al. 2000, Tilman et al.    eutrophication would be useful for directing manage-
2001).                            ment efforts towards the prevention of loss. Finally,
  Increasing evidence shows that an important proxi-     eelgrass loss and algal overgrowth in estuaries may
mate cause by which increased nitrogen supply leads      have important implications for ecosystem production,
to eelgrass decline is via intense light limitation due to  carbon and nutrient cycling, and trophic linkages to
the overgrowth of fast-growing, nitrogen-limited algal    adjacent systems (carbon and nutrient export: Nixon
producers (Kemp et al. 1983, Short et al. 1993, 1995,     et al. 1986, Cebrián et al. 1998, Cebrián 1999). Under-
Short & Wyllie-Echeverria 1996, Valiela et al. 1997b,     standing the nature and dynamics of the transition
Hauxwell et al. 2001). Increased nitrogen delivery into    from eelgrass to algae under increasing eutrophication
estuarine waters stimulates the growth of opportunistic    is a first step toward understanding the ecological
algae, including phytoplankton, epiphytes, and fila-     implications associated with that change.
mentous macroalgae (Sand-Jensen & Borum 1991,          The estuaries of Waquoit Bay, Massachusetts, USA,
Duarte 1995, Short et al. 1995, Taylor et al. 1995b,     offer the opportunity to examine the response of estu-
Hauxwell et al. 1998), which may attenuate a large      arine primary producers to increased nitrogen loads at
percentage of light that was available to eelgrass      the watershed –estuarine scale. We used the estuaries
under low-nitrogen loads. Other processes associated     of Waquoit Bay in a space-for-time substitution (Pickett
with large accumulations of algal producers, such as     1989) to infer the time course of increased eutrophica-
anoxia (Pregnall et al. 1984, Koch et al. 1990), redox    tion created by increasing urbanization of watersheds.
changes resulting from low-oxygen concentration (i.e.     In the Waquoit Bay estuarine system, different land
high-sulfide concentrations: Goodman et al. 1995, Ter-    use patterns within the watersheds of 7 estuaries
rados et al. 1999), and high and toxic ammonium        (Fig. 1), similar in depth and water residence times,
concentrations (Van Katwijk et al. 1997), may also con-    have resulted in different annual loads of nitrogen
tribute to eelgrass decline. Direct effects of nitrogen    delivered to those estuaries (Table 1). Increased urban-
loading, such as high and toxic concentrations of       ization within certain watersheds is accompanied by
nitrate have also been found to cause eelgrass decline    increases in delivery of land-derived nitrogen (Table 1).
(Burkholder et al. 1992, 1994). In aggregate, however,    The range of nitrogen loads delivered to Waquoit Bay
                       Hauxwell et al.: Eutrophication and eelgrass loss                  61




                                     mass, and total annual growth and areal production
                                     of leaves, rhizomes, and roots. The effect of nitrogen
                                     supply on various aspects of eelgrass productivity is not
                                     direct, but seems most likely to be mediated through
                                     the stimulation of biomass, and increased light inter-
                                     ception by algal producers. To assess this potential
                                     indirect effect of nitrogen, we (3) determined annual
                                     patterns of phytoplankton, epiphyte, and macroalgal
                                     biomass, and (4) estimated how standing stocks of
                                     these producers may have modified available irradi-
                                     ance for eelgrass in estuaries subject to different rates
                                     of nitrogen loading. Presence of a relationship be-
                                     tween these measurements and nitrogen load would
                                     yield insight into the mechanisms by which eelgrass
                                     decline occurs and reveal potentially useful indicators
                                     of incipient decline. We conclude by discussing the
                                     implications of our results toward understanding
                                     broad-scale ecosystem alterations accompanied by
                                     increased nitrogen loading and in the development of
                                     management recommendations aimed at preventing
                                     eelgrass loss.


                                           MATERIALS AND METHODS

                                      Measurement of eelgrass bed area and recent loss.
                                     We determined Zostera marina bed area in estuaries
                                     subject to different rates of nitrogen loading in Septem-
                                     ber 1997, using a viewbox held over the side of a boat
Fig. 1. Map of Waquoit Bay estuarine system (inset: location on
Cape Cod, Massachusetts). Watershed delineations for each        traveling along transects approximately 30 m apart,
      estu-ary are indicated with dashed lines           and running E to W and N to S. Maps in which we de-
                                     lineated eelgrass area in each estuary were scanned
                                     and percentage cover was calculated after digitizing
  estuaries (5 to 407 kg N ha–1 yr–1, Valiela et al. 1997a,       the total area of each basin and the coverage of eelgrass
  2000a) encompasses ~75% of the range of reported           (Adobe Photoshop 4.0, Adobe Systems 1996). Loss of
  loads to different estuaries around the world (Nixon         eelgrass in the past decade was determined by compar-
  1992).                                ing 1997 maps to those obtained in 1987 by L. Deegan
   In this paper, we make use of inter-estuary com-          & I. Valiela (pers. obs.) for Sage Lot Pond, and Short &
  parisons to evaluate the effect of nitrogen supply on         Burdick (1996) for the remainder of the system.
  (1) the extent of eelgrass bed area and losses of eel-          Measurements of eelgrass variables. To evaluate
  grass habitat over the past decade, and (2) eelgrass         how the seasonal patterns and magnitude of several
  mean annual shoot density, biomass and areal bio-           eelgrass variables may have responded to different


  Table 1. Land-derived nitrogen loading rate (normalized for estuarine + salt marsh surface area) (Valiela et al. 1997a, 2000a),
  number of houses within each watershed (Valiela et al. 1997a, 2000a), mean depth (mean low water + 0.5 m tidal range), and
              water residence time (Valiela unpubl. data) for the 7 estuaries of Waquoit Bay


   Variable                                     Estuary
                        Timms    Sage Lot   Hamblin    Jehu    Eel   Quashnet   Childs
                         Pond    Pond     Pond     Pond   Pond    River    River

   Nitrogen loading rate (kg N ha–1 yr–1)    5.3     7.6     28.4     30.1    62.7    298     407
   Houses (watershed–1)             0      0     340     529    718     767     1233
   Depth (m)                  1.3     1.3     1.5     1.7    1.4     0.8     1.4
   Water residence time (d)           1.5     1.5     2.3     2.7    2.0     1.7     2.3
62                    Mar Ecol Prog Ser 247: 59–73, 2003




nitrogen loads, we conducted a field study from       rates on a shoot basis (mg DW shoot–1 d–1). Weight-spe-
November 1997 to November 1998. We routinely mea-      cific leaf growth rates were determined by dividing
sured shoot densities, shoot biomass, areal above-     growth rates per shoot by aboveground shoot biomass.
ground biomass, plastochrone intervals, and leaf, rhi-     For each marking period, we calculated the plas-
zome, and root absolute and weight-specific growth     tochrone interval (number of days elapsed between
rates, and production rates of eelgrass. Measurements    the appearance of 2 consecutive leaves) by dividing
were taken every 2 to 8 wk in the 4 estuaries of      the total days elapsed by the mean number of new
Waquoit Bay that still supported eelgrass meadows      leaves (i.e. bearing no holes) emerged per shoot
(Timms, Sage Lot, Hamblin Ponds, and Jehu). Of the     (Brouns 1985). Rhizome and root growth rates per
7 total estuaries, these 4 received the lowest loads of   shoot (mg DW shoot–1 d–1) were calculated from the
nitrogen from their watershed. Timms and Sage Lot      rate of node formation and growth along the horizontal
Ponds have forested watersheds and receive very       rhizome, as outlined by Sand-Jensen (1975) and subse-
low loads of land-derived nitrogen (≤ 8 kg N ha–1 yr–1)   quently applied by Pedersen & Borum (1992, 1993) and
(Table 1). Jehu and Hamblin Ponds have watersheds      Duarte et al. (1994). This technique, however, only pro-
that are somewhat urbanized and receive higher loads    vides conservative estimates of root growth, since root
of nitrogen from their watersheds (~30 kg N ha–1 yr–1).   turnover is fast and the estimates are only based on
 To quantify shoot density, SCUBA divers counted      standing root biomass at the time of collection (Duarte
the total number of shoots (vegetative and flowering)    et al. 1998). Shoot-specific rhizome and root growth
within randomly tossed 0.25 m2 quadrats; 3 to 4 mea-    rates were also determined by dividing growth rates
surements were made in Timms, Sage Lot, and Jehu      per shoot by aboveground shoot biomass.
Ponds, where spatial distributions of shoots were rea-     Eelgrass aboveground areal biomass was estimated
sonably homogenous, and 6 to 12 measurements were      for each sampling interval by multiplying mean shoot
made in Hamblin Pond, where spatial distribution was    density and mean aboveground shoot biomass. We
relatively patchy. We used the marking technique      chose not to measure belowground biomass, due to
described by Zieman & Wetzel (1980) to measure       the destructive nature of the collection method, and
in situ leaf growth rates. To insure that growth rates   we only report data for Sage Lot Pond, where long-
of all shoot size classes were represented, SCUBA      term above- and belowground biomass data have
divers tagged all shoots within a given area (at least   been taken since 1994 using Eckman grabs (15 cm ×
25 shoots) and punched 2 holes at the blade–sheath     15 cm) (Hersh 1996, Hauxwell et al. 1998, Stieve
junction of the oldest leaf with a 23-gage hypodermic    2001). Areal leaf, rhizome, and root production were
needle. Shoots were retrieved 2 to 6 wk later (depend-   derived by multiplying the corresponding growth
ing on the season), with as much intact rhizome and     rates per shoot by shoot density for each sampling
root material as possible, and a new batch of shoots    interval, and annual total estimates were derived by
were tagged and marked. A total of 12 sets of mea-     summing the production of all compartments over
surements were made throughout the study period.      the annual cycle.
Collected shoots were brought to the laboratory and      Friedman’s method for randomized blocks (Sokal &
frozen until processing was possible.            Rohlf 1995) was used to compare shoot densities, shoot
 In the laboratory, we ranked the leaves on each      biomass, areal aboveground biomass, plastochrone
shoot by age, and on each leaf we measured total      intervals, leaf, rhizome, and root absolute and weight-
length, width and growth as the distance between the    specific growth rates of eelgrass over time among
sheath–blade junction and marked holes. We noted      Timms, Sage Lot, Hamblin, and Jehu Ponds. If signifi-
whether the tips of the leaves were intact or broken,    cant differences were observed among estuaries, a
and used only shoots for which we could unambigu-      Wilcoxon’s signed-ranks test was employed to deter-
ously determine leaf growth (for instance, we dis-     mine significant differences between pairs.
carded shoots where leaf tips were not intact and holes    Biomass of and light interception by algal pro-
not observed on the remaining leaf blade portion). To    ducers. To examine the response of algal producers
convert aboveground shoot characteristics and leaf     and to quantify their potential effect on light supply for
growth from units of area to biomass, we calculated an   eelgrass, we first determined the biomass of phyto-
average leaf specific density (mg DW cm–2 of leaf sur-   plankton, epiphytes, and macroalgae in each estuary.
face) for each estuary on each date. To determine leaf   These data were then used to evaluate potential light
specific density, we first removed epiphytic material    interception by each producer. Based on estimates of
using a glass slide from each leaf on each of 5 shoots,   saturating and compensating irradiances for eelgrass
then dried leaves in an oven at 70°C, and weighed      over an annual cycle (a function of temperature), and
them. These values were used to calculate above-      light availability after attenuation by phytoplankton,
ground shoot biomass (mg DW shoot–1) and leaf growth    epiphytes, and macroalgae, we could compare the re-
                    Hauxwell et al.: Eutrophication and eelgrass loss                 63




lative importance of potential light limitation of eel-      8 cm, light penetration was <1% of that reaching the
grass in estuaries subject to different rates of nitrogen     surface of the canopy.
loading, as detailed below.                     Because eelgrass shoot height has an effect on its
  Annual measurements of phytoplankton, epiphytes,        depth in the water column and, therefore, the intensity
and macroalgal biomass: During each visit for the eel-      of incoming light it receives, we considered scenarios
grass measurements described above, we also quanti-        for average established shoots (>15 cm in height) and
fied phytoplankton, epiphyte, and macroalgal stand-        also for smaller newly recruiting shoots. For estab-
ing stocks within the 4 estuaries. Measurements of        lished shoots, we calculated water-column light atten-
chlorophyll a concentrations in the water column were       uation as the mean of attenuation between (1) the air-
made by collecting replicate 1 l water samples within       water interface to tips of leaves as shoot height varies
each eelgrass meadow; in the laboratory, samples         throughout a year, and (2) the air-water interface to the
were filtered (Whatman GF/F filters), and chlorophyll       sediment-water interface. Epiphyte light attenuation,
a was determined spectrophotometrically using the         as described above, was assumed to occur. Macroalgal
method of Lorenzen (1966). Epiphyte biomass was          light attenuation was calculated by multiplying the
determined for leaves ranked 1 (youngest) to 5, for at      percentage of photosynthetic material buried by macro-
least 5 shoots from each estuary. We used a glass slide      algae (range: 0 to 32% depending on estuary and date,
to scrape epiphytes from each leaf into preweighed        based on mean shoot height and macroalgal canopy
aluminum foil envelopes, dried them overnight, and        height) by light attenuation through the macroalgae
reweighed them. Macroalgal canopy heights were de-        canopy. For smaller new shoots, water column light
termined using SCUBA and taking 5 to 15 measure-         attenuation was calculated from the air-water interface
ments at random within the eelgrass meadows. We used       to the sediment-water interface, interception of light
the same statistical approach described for eelgrass       by epiphytes was assumed to be negligible, and 100%
annual variables to determine whether phytoplankton,       of the photosynthetic material was assumed to be
epiphyte, and macroalgal standing stocks differed in       beneath the macroalgal canopy.
estuaries receiving different loads of nitrogen.           Calculations of epiphytic and macroalgal shading
  Calculations of light attenuation due to phytoplank-      were made assuming that both processes occur simul-
ton, epiphytes and macroalgae: Estimates of light         taneously, and were both based on the magnitude of
intercepted by each type of producer and resulting        incoming irradiance after total water-column attenua-
irradiance (I) reaching eelgrass surfaces were calcu-       tion. We did not include light interception due to eel-
lated over the annual cycle as described in Hauxwell et      grass canopies, so estimates of resulting light intensi-
al. (2001). Our general approach was to use measure-       ties are conservative and only represent the potential
ments of each biological parameter (i.e. phytoplank-       modifying effects of water-column, epiphytes, and
ton, epiphyte, and macroalgal biomass) in equations        macroalgae.
describing the relationship between biomass and light        Irradiance and prediction of saturating (Isat ) and
attenuation for each producer (Bannister 1974, Kirk        compensating (Icomp) irradiances for eelgrass: To con-
1994, Twilley et al. 1985, Peckol & Rivers 1996). Previ-     vert percentage reductions in irradiance by algal pro-
ous studies showed that background attenuation did        ducers to absolute values, we needed to apply irradi-
not vary among different sites in the Waquoit Bay sys-      ance at the air-water interface. Surface irradiance data
tem (Hauxwell et al. 2001), so we used the mean water-      for the annual cycle were provided by R. Payne of
column light-attenuation factor for background scatter      Woods Hole Oceanographic Institution. Based on previ-
(KB = 0.85 m–1) from Hauxwell et al. (2001). Light atten-     ous measurements in Waquoit Bay throughout a variety
uation due to phytoplankton (KP) was calculated by        of weather conditions, we assumed a surface re-
multiplying chlorophyll concentrations by a chlorophyll-     flectance of 35% (Peckol & Rivers 1996).
specific light-attenuation coefficient (kc = 0.016 m2 mg–1      To assess the effect of light interception by algal pro-
chlorophyll: Bannister 1974, Kirk 1994). Total water-       ducers on eelgrass growth, we compared our estimates
column light-attenuation from surface to depth, z, was      of irradiance reaching eelgrass in the Waquoit estu-
calculated according to Beer’s law: Iz /I 0 = e– (KB+KP)(z).   aries to estimates of saturating and compensating
Light attenuation due to mean epiphyte biomass on         irradiance necessary for eelgrass photosynthesis.
Leaves 1 to 5 was calculated according to Twilley et al.     We determined the annual cycle of Isat and Icomp for eel-
[1985: Fig. 8, top: ln Iz /I 0 = 0.32 – (0.42) (epiphyte bio-   grass in Waquoit Bay, using the relationships between
mass, in units mg DW cm–2 of leaf material)]. Light        temperature and I sat and I comp for a nearby Woods Hole
attenuation due to macroalgal canopies was calculated       eelgrass population (Marsh et al. 1986), applied to
according to the relationship provided by Peckol &        Waquoit Bay 1998 temperature data (provided by the
Rivers (1996), in which irradiance decreased exponen-       Waquoit Bay National Estuarine Research Reserve
tially as macroalgal canopy heights increased; by 6 to      Baywatchers Program).
64                       Mar Ecol Prog Ser 247: 59–73, 2003




             RESULTS                 Jehu Ponds, where nitrogen loading rates were ~30 kg
                                  N ha–1 yr–1, eelgrass nearly disappeared. Only ~6500 of
   Eelgrass bed area and recent loss in relation to       the ~180 000 m2 of eelgrass present in 1987 remained
          nitrogen loading                in Hamblin Pond, a loss of > 96%. In Jehu Pond, 85% of
                                  the area in 1987 was lost, with only ~20 000 m2 remain-
 For the estuaries that supported eelgrass in 1987 (i.e.     ing out of the ~130 000 m2 of eelgrass habitat present in
those exposed to loading rates between 5 and 63 kg N        1987. There was some loss of eelgrass habitat (13 to
ha–1 yr–1), we regressed eelgrass bed area in 1997 and       32%) even in Timms Pond and Sage Lot Pond, where
areal loss between 1987 and 1997 versus nitrogen          nitrogen loads were low (≤ 8 kg N ha–1 yr–1). In 1997,
loading rates (Fig. 2). Across those estuaries, the area      Timms Pond and Sage Lot supported 14 000 and 30 000
of eelgrass habitat decreased logarithmically as nitro-      m2 of eelgrass habitat, respectively.
gen loading rates increased (Fig. 2, top). Similarly, loss
of eelgrass habitat, expressed as a percentage of the
existing area in 1987, increased logarithmically with         Eelgrass annual variables in relation to nitrogen
higher loads (Fig. 2, bottom).                              loading
 Loss of eelgrass habitat from estuaries of Waquoit
Bay between 1987 and 1997 was extensive. Eelgrass          Annual means of our eelgrass measurements are
disappeared entirely from Eel Pond, the estuary with        summarized in Table 2. Of all the eelgrass measure-
the highest nitrogen loading rate (63 kg N ha–1 yr–1)       ments, shoot density, shoot and areal aboveground
that still supported eelgrass in 1987. In Hamblin and       biomass, leaf growth rate, and leaf, rhizome, and root
                                  production differed among the 4 estuaries (Table 3).
                                  For all of these measurements, the meadow in Sage Lot
                                  Pond (a low-nitrogen site) consistently represented the
                                  highest values recorded, while values for the Hamblin
                                  Pond meadow (a higher-nitrogen site) were the lowest
                                  (Table 2). Values for Timms Pond (a low-nitrogen site)
                                  and Jehu Pond (a higher-nitrogen site), however, were
                                  aligned with the other estuary of similar nitrogen load
                                  (Sage Lot Pond or Hamblin Pond, respectively) only in
                                  the cases of shoot density, aboveground areal biomass,
                                  and production by leaves, rhizomes, and roots (Table 2).
                                  Within-meadow differences in shoot density contri-
                                  buted to the differences observed in areal biomass and
                                  production rates within meadows, which were arith-
                                  metically derived by multiplying shoot density by
                                  shoot biomass or shoot growth rates, variables that
                                  were not aligned in relation to nitrogen load (Table 3).
                                   Regression analyses were conducted to ascertain
                                  which eelgrass variables might be indicators of decline
                                  associated with increased nitrogen loading before dis-
                                  appearance of habitat. These regressions were, there-
                                  fore, restricted to the narrow, relatively low range of
                                  loads represented by estuaries that supported eelgrass
                                  within the past decade (0 to 63 kg N ha–1 yr–1). Mean
                                  annual shoot density and aboveground areal biomass
                                  (within meadows) decreased exponentially as nitrogen
                                  loading rates increased (Fig. 3). Total annual eelgrass
                                  production within meadows decreased exponentially
                                  as nitrogen loading rates increased (Fig. 4, top). Total
Fig. 2. Zostera marina. Eelgrass areal cover (top) and areal    eelgrass production within the estuary, calculated as
loss (bottom) between 1987 and 1997 in estuaries of the Wa-    the sum of the products between within-meadow leaf,
quoit Bay system subject to different rates of nitrogen loading.  rhizome, and root production applied to the area of
Regression analyses were conducted within the nitrogen
loads represented by the estuaries that still had eelgrass in
                                  eelgrass bed in the estuary, also decreased expo-
1987 (Timms, Sage Lot, Hamblin, Jehu, and Eel Ponds: 5 to     nentially as nitrogen loading rates increased (Fig. 4,
  63 kg N ha–1 yr–1). Here and in later figures, * p < 0.05   bottom).
                     Hauxwell et al.: Eutrophication and eelgrass loss                  65




Table 2. Zostera marina. Summary of results of annual measurements of shoot densities (annual within-meadow mean ± SE),
aboveground shoot biomass (annual mean ± SE), aboveground areal biomass (annual within-meadow mean ± SE), number of
plastochrone units produced (annual total), and leaf, rhizome and root absolute and weight-specific growth rates (annual total
± SE) and production rates (annual within-meadow total ± SE) of eelgrass in 4 estuaries of Waquoit Bay subject to different rates
                            of nitrogen loading


Variable                                        Estuary
                             Timms Pond     Sage Lot Pond    Hamblin Pond    Jehu Pond
          –2
Density (shoots m )                    196 ± 25      266 ± 28       57 ± 5      141 ± 23
Aboveground shoot biomass (mg DW shoot–1)         130 ± 24      253 ± 47      160 ± 19     244 ± 41
Aboveground areal biomass (g DW m–2)            32 ± 8       080 ± 19       09 ± 1      36 ± 9
Plastochrone units (number yr–1)              30.4        27.2        31.1       29.9
Leaf growth rate (mg DW shoot–1 yr–1)           690 ± 30      1280 ± 800      870 ± 60     1169 ± 600
Rhizome growth rate (mg DW shoot–1 yr–1)         210 ± 10      240 ± 20      220 ± 10     270 ± 10
Root growth rate (mg DW shoot–1 yr–1)           110 ± 50      170 ± 10      180 ± 10     180 ± 20
Weight-specific leaf growth rate (% yr–1)         574 ± 13      523 ± 10      562 ± 17     493 ± 12
Weight-specific rhizome growth rate (% yr–1)       200 ± 15      122 ± 90      150 ± 23     117 ± 24
Weight-specific root growth rate (% yr–1)         142 ± 10      111 ± 70      129 ± 14     094 ± 16
Leaf production (mg DW m–2)                159 ± 9       395 ± 29       49 ± 6      203 ± 70
Rhizome production (mg DW m–2)              044 ± 3       74 ± 6       13 ± 2      54 ± 3
Root production (mg DW m–2)                020 ± 1       46 ± 3       12 ± 1      30 ± 3



Table 3. Zostera marina. Results of Friedman’s method for       Light interception by algal producers and relation-
randomized blocks (Statview®, SAS Institute 1999) used to               ship to nitrogen load
compare shoot densities, aboveground shoot biomass, above-
ground areal biomass, plastochrone intervals, and leaf,        Annual measurements of phytoplankton, epiphytes and
rhizome and root absolute and weight-specific growth rates              macroalgal biomass
and production rates of eelgrass over time in 4 estuaries
of Waquoit Bay subject to different rates of nitrogen loading      Phytoplankton biomass peaked at the end of August
(ns: not significant, p > 0.05; **p < 0.01; ***p < 0.001). When
                                   in all estuaries between 11 and 23 µg chlorophyll a l–1,
significant differences were observed among sites, a Wil-
coxon’s signed-ranks test (Statview®) was employed to com-      and was ≤ 4 mg chlorophyll a l–1 from November to
pare pairs. Results of the ranking were presented in order      May (Fig. 5, top). Epiphyte biomass was lowest in the
from highest to lowest; =: no significant difference between     spring when plastochrone intervals were at a mini-
adjacent pairs; >: significant differences (p < 0.05) between     mum, and accumulation of epiphytic material was
adjacent pairs (S: Sage Lot Pond, T: Timms Pond, J: Jehu
          Pond, H: Hamblin Pond)
                                   reduced due to the relatively fast appearance of new
                                   leaves and shedding of old leaves (Fig. 5, middle).
Variable              p   Statistical ranking
                                   Epiphyte biomass was highest during late summer,
                                   fall, and early winter when plastochrone intervals
Density (shoots m–2)        ***  S>T=J>H           were longer and hence, the time in which epiphytes
Aboveground shoot biomass     **  S=J>T=H           might colonize leaves was longer (Fig. 5, middle). The
 (mg DW shoot–1)
                                   highest peak in epiphyte biomass was in Hamblin
Aboveground areal biomass     ***  S>J=T>H
 (g DW m–2)
                                   Pond in late August at 13 mg DW cm–2 of leaf material.
Plastochrone interval (d leaf–1)  ns                Macroalgal biomass fluctuated over the year in all
Leaf growth rate          **  S = J = H = T, S > H,    estuaries and was generally highest in Hamblin Pond
 (mg DW shoot–1 d–1)           S > T, J > T        (4 to 10 cm) and lowest in Timms Pond (0 to 2 cm)
Rhizome growth rate        ns                (Fig. 5, bottom). Macroalgal blooms occurred in Jehu
 (mg DW shoot–1 d–1)                         Pond during December 1997 and in November 1998
Root growth rate          ns                when mean canopy heights reached 13 cm.
 (mg DW shoot–1 d–1)
                                    Of these measurements, macroalgal canopy height
Weight-specific leaf growth    ns
 rate (% d–1)                            was the only variable to differ significantly among
Weight-specific rhizome      ns                estuaries (Table 4). Maximum macroalgal canopy
 growth rate (% d–1)                         height increased linearly as nitrogen loading rate in-
Weight-specific root growth    ns                creased, within the range of loads in which eelgrass
 rate (% d–1)
                                   was still present, and also across the entire range of
Leaf production (mg DW m–2)    ***  S > J = T>H
                                   loads present to the Waquoit Bay estuaries (Fig. 6).
Rhizome production         ***  S > J = T>H
 (mg DW m–2)                             Canopies were primarily comprised of the fast-
Root production (mg DW m–2)    **  S = J = T > H, S > T    growing, nutrient-limited taxa Cladophora vagabunda
                                   and Gracilaria tikvahiae (Peckol et al. 1994).
66                        Mar Ecol Prog Ser 247: 59–73, 2003




Fig. 3. Zostera marina. Mean annual 1998 within-meadow        Fig. 4. Zostera marina. Total annual 1998 eelgrass areal pro-
shoot density (top) and areal biomass (bottom) of eelgrass in     duction within meadows and normalized to bed area within
estuaries of the Waquoit Bay system subject to different rates    estuaries of the Waquoit Bay system subject to different rates
of nitrogen loading. Regression analyses were conducted        of nitrogen loading. Regression analyses were conducted
within the nitrogen loads represented by the estuaries that      within the nitrogen loads represented by the estuaries that
still had eelgrass in 1987 (Timms, Sage Lot, Hamblin, Jehu,      still had eelgrass in 1987 (Timms, Sage Lot, Hamblin, Jehu,
       and Eel Ponds: 5 to 63 kg N ha–1 yr–1)                and Eel Ponds; 5 to 63 kg N ha–1 yr–1)



 Light attenuation due to phytoplankton, epiphytes          high-standing stocks of phytoplankton in all estuaries
          and macroalgae                   at that time (Fig. 5, top). Due to the increase in eel-
                                   grass shoot height from spring to summer (shoots
 Estimated light attenuation through the water            ranged from 15 to 95 cm in height over the year) and
column from the surface to the sediment-water inter-         hence, shallower depth (z), estimated water-column
face remained relatively constant throughout the           light attenuation from the surface to leaf tips was
year, ranging from 68 to 82% (Fig. 7, top right).           lowest during late spring and early summer (Fig. 7,
The small peak at the end of August was due to the          top left).

Table 4. Results of Friedman’s method for randomized blocks (Statview®, SAS Institute 1999) used to compare standing stocks of
phytoplankton, epiphytes, macroalgae, and estimated irradiance reaching established and new eelgrass Zostera marina shoots
over time in 4 estuaries of Waquoit Bay subject to different rates of nitrogen loading (ns: not significant, p > 0.05; ***p < 0.001). If
significant differences were observed among sites, a Wilcoxon’s signed-ranks test (Statview®) was employed to compare pairs.
Results of the ranking were presented in order from highest to lowest; =: no significant difference between adjacent pairs;
>: significant differences (p < 0.05) between adjacent pairs (S: Sage Lot Pond, T: Timms Pond, J: Jehu Pond, H: Hamblin Pond)


Variable                                        p        Statistical ranking

Phytoplankton (µg chlorophyll a l–1)                         ns
Epiphyte biomass (mg cm–2)                              ns
Macroalgal canopy height (cm)                             ***       H>J=S>T
Irradiance reaching established eelgrass shoots (µmol photons m–2 s–1)        ***       S > T = J = H, T > H
Irradiance reaching new eelgrass shoots (µmol photons m–2 s–1)            ***       T>S=J>H
                     Hauxwell et al.: Eutrophication and eelgrass loss                   67




                                  photosynthesis by established eelgrass shoots was
                                  highest for the Sage Lot Pond meadow and lowest
                                  for the Hamblin Pond meadow (Fig. 8, middle). Our
                                  conservative estimates yielded irradiance above satu-
                                  rating levels for established shoots in all meadows
                                  between January and August. During late summer
                                  and fall, established shoots may have received be-
                                  tween compensating and saturating levels of irradi-
                                  ance, except in Hamblin Pond where established
                                  shoots may have received less than compensating
                                  levels of irradiance.
                                    Estimated irradiance available for photosynthesis
                                  by new shoots was highest (above or near saturation)
                                  in Timms Pond, where macroalgal canopies were
                                  lowest (Fig. 8, bottom). Estimates for Sage Lot and
                                  Jehu Ponds ranged between compensating and
                                  above-saturating levels of irradiance. Estimates were
                                  lowest in Hamblin Pond, at or below compensating
                                  levels of irradiance for all but 1 date in the annual
                                  cycle.
                                    Over an annual cycle, the estimated irradiance
                                  which reached established and new shoots varied sig-
                                  nificantly among estuaries (Table 4). The mean annual
                                  irradiance which reached established and new eel-
                                  grass shoots was significantly lower in estuaries of
                                  higher-nitrogen loads (Fig. 9). Within estuaries of
                                  similar nitrogen load, established shoots were esti-
Fig. 5. Standing stocks of phytoplankton (top), epiphytes of eel-  mated to receive significantly higher irradiance than
grass Zostera marina (middle), and macroalgae (bottom) between   new shoots (paired t-tests for values throughout the
November 1997 and November 1998 in 4 estuaries of Waquoit      annual cycle, p < 0.05).
Bay subject to different rates of nitrogen loading (S: Sage Lot
   Pond, T: Timms Pond, J: Jehu Pond, H: Hamblin Pond)




  Estimated light attenuation by epiphytes for estab-
lished shoots ranged from 0 to 100% (Fig. 7, middle
left) after accounting for water-column attenuation.
During peaks in epiphyte biomass, light attenuation
by epiphytes may have been of greater importance
than water column attenuation for established shoots
(Fig. 7, top left). Since epiphyte biomass was negligible
for new shoots, light attenuation was 0% (Fig. 7,
middle right).
  Estimated light attenuation by macroalgal canopies
was relatively minor for established shoots, ranging
from 0 to 31% (Fig. 7, bottom left) after accounting for
water-column attenuation. For new shoots, however,
light attenuation by macroalgal canopies may have          Fig. 6. Maximum canopy height of macroalgae during 1998 in
been severe; the lowest peak occurred in Timms Pond         estuaries of the Waquoit Bay system subject to different rates
at 77% (Fig. 7, bottom right). Estimated light attenua-       of nitrogen loading. Regression analyses were conducted for
tion in Hamblin Pond was 91 to 100% throughout the         the 4 estuaries in which we took measurements (Timms, Sage
                                  Lot, Hamblin, and Jehu Ponds: 5 to 30 kg N ha–1 yr–1) and for
annual cycle (Fig. 7, bottom right).                the entire range of values present in the Waquoit Bay system
  After attenuation by the water column, epiphytes,        (dashed line) including Childs and Quashnet Rivers (5 to
and macroalgae, estimated irradiance available for            407 kg N ha–1 yr–1; data from Hauxwell et al. (2001)
68                       Mar Ecol Prog Ser 247: 59–73, 2003




                                 Fig. 8. Zostera marina. Monthly surface irradiance (mean for
                                 daylight hours) corrected for reflectance (top) and estimates
Fig. 7. Zostera marina. Percentage of surface irradiance atten-  of available irradiance (after attenuation by the water column,
uated through the water column (top), by epiphytes (middle),   epiphytes, and macroalgal canopies) to established (middle)
and by macroalgal canopies (bottom) for established (left) and  or new (bottom) eelgrass shoots between November 1997 and
newly recruiting (right) eelgrass shoots between November     November 1998 in 4 estuaries of Waquoit Bay subject to dif-
1997 and November 1998 in 4 estuaries of Waquoit Bay sub-     ferent rates of nitrogen loading (S: Sage Lot Pond, T: Timms
ject to different rates of nitrogen loading (S: Sage Lot Pond,  Pond, J: Jehu Pond, H: Hamblin Pond) (surface irradiance
T: Timms Pond, J: Jehu Pond, H: Hamblin Pond). Epiphyte      data from R. Payne, Woods Hole Oceanographic Institution).
and macroalgal light attenuation were based on incoming      Saturating (Sat.) and compensating (Comp.) levels of irradi-
      irradiance after water-column attenuation       ance for eelgrass, corresponding to actual temperatures over
                                         the annual cycle are also shown



            DISCUSSION                 Ponds; Fig. 2). In the Waquoit estuaries, nitrogen loads
                                 that allow eelgrass survival appear to be < 28 to 63 kg N
     Nitrogen loading and eelgrass loss           ha–1 yr–1. This load matches hindcast estimates of nitro-
                                 gen loading (based on historical land-use data and the
  Nitrogen loading to the Waquoit Bay system has con-      Waquoit Bay Nitrogen Loading Model, Valiela et al.
tributed to extensive loss of eelgrass Zostera marina       2000a) to now eutrophic subestuaries for the period of
(Figs. 2–4). Results of this large-scale comparison at      time when eelgrass loss was documented historically
the watershed –estuarine scale are in agreement with       (Bowen & Valiela 2001). Presence of seed coats in cores
previous laboratory and mesocosm experiments (Short        taken from estuaries with the highest loads of nitrogen
et al. 1993,1995), i.e. they both show that (1) eelgrass     revealed that Childs River and Quashnet River, estuar-
is very sensitive to eutrophication, with large losses      ies now receiving loads between 298 and 407 kg N ha–1
occurring rapidly even at relatively low-nitrogen load-      yr–1, once supported eelgrass prior to extensive resi-
ing rates, and (2) light limitation imposed by nutrient-     dential development of their watersheds (Safran et al.
limited algae is an important mechanism by which         1998). Seagrasses in general appear to be sensitive
such losses may occur. Substantial eelgrass loss was       indicators of nitrogen loading. In a compilation by
observed in all Waquoit Bay estuaries, except in those      Valiela & Cole (2002) of worldwide seagrass loss over a
with the lowest loading rates (Timms and Sage Lot         range of reported nitrogen loads, an identical general
                    Hauxwell et al.: Eutrophication and eelgrass loss                 69




pattern was observed, with substantial losses (> 50%)       tom & 9). From light budgets which considered water
occurring within 50 to 100 kg N ha–1 yr–1 and total dis-      column, epiphyte, and macroalgal shading, we esti-
appearance at loadings exceeding 100 kg N ha–1 yr–1.        mated chronic, severe light limitation to newly recruit-
  In Waquoit Bay, loss of eelgrass under increasing        ing shoots in Hamblin Pond, due mainly to shading by
nitrogen loads seems to occur mainly through a           a coexisting 4 to 10 cm macroalgal canopy (Figs. 5, bot-
decrease in shoot density (Fig. 3, top), rather than as a     tom & 8, bottom). Recruiting shoots were exposed to
result of reduced growth rates per shoot (Tables 2 & 3).      under-compensating irradiances for most of the year in
Under higher-nitrogen loads (Hamblin and Jehu Ponds),       Hamblin Pond due to a high, persistent macroalgal
we observed rapid loss of eelgrass bed area (Fig. 2) and      canopy, and during the 1998 fall in Jehu Pond due to a
relatively low shoot densities (Table 2, Fig. 3, top);       concurrent macroalgal bloom (Fig. 8). Severe light lim-
however, we found no significant relationship between       itation of recruiting shoots, however, was not observed
nitrogen loading rates and the above- or belowground        in the 2 low-nitrogen estuaries. Additional deleterious
growth rates of established shoots (Table 2). Hence,        effects associated with large macroalgal canopies
across the estuaries compared, more eutrophic estuar-       may occur via unfavorable biogeochemical conditions
ies have fewer shoots per unit of meadow area, but the       imposed on buried eelgrass shoots, such as anoxia
remaining shoots generally grow at the same rate as        (Pregnall et al. 1984, Koch et al. 1990), other redox
those in low-nitrogen estuaries. These differences sug-      changes resulting from low-oxygen concentration (i.e.
gest reduced shoot recruitment or promoted shoot          high sulfide concentrations: Goodman et al. 1995, Ter-
mortality as plausible mechanisms leading to eelgrass       rados et al. 1999), and toxic ammonium concentrations
decline in eutrophic estuaries. This discrepancy be-        (Van Katwijk et al. 1997). In fact, the capacity for large
tween density data and growth data implies that estab-       macroalgal canopies (>12 cm) to preclude eelgrass
lished shoots function similarly on a per shoot basis       shoot recruitment has been experimentally demon-
among estuaries, but that established shoots from         strated by Hauxwell et al. (2001). Our results also con-
higher-nitrogen estuaries lack sufficient resources to       tribute evidence for increased mortality of established
translocate energy to newly recruiting clonal branches.      shoots in higher-nitrogen estuaries due to intense
Hence, loss occurs at both the edges of a meadow and        shading by epiphytes; established shoots in Hamblin
within a meadow.                          Pond were estimated to receive less than compensat-
  The speculation that diminished recruitment accounts      ing irradiance during the fall of the two years of survey
for eelgrass decline is further supported by our esti-       (i.e. highest epiphyte biomass). Nitrate concentrations
mates of light availability to new shoots (Figs. 8, bot-      in these estuaries were below toxic levels (Burkholder
                                  et al. 1992,1994, Hauxwell et al. 2001)


                                           Coastal management

                                   Management of eelgrass habitats is often mandated
                                  by state and local governments. In Massachusetts, for
                                  instance, state regulations dictate that stakeholders in
                                  projects involving dredging, filling, or altering parcels
                                  of coastland must first demonstrate that they will mini-
                                  mize deleterious impacts, or have no adverse effects on
                                  eelgrass beds. In addition, there are other stringent
                                  local bylaws on land development, such as regulations
                                  that often make dock construction illegal. The very
                                  important link, however, between watershed develop-
                                  ment > 30 m from the water’s edge and adjoining estu-
                                  arine eelgrass health/water quality is largely unregu-
                                  lated. Watershed influences on nitrogen load arguably
Fig. 9. Zostera marina. Estimated mean (± SE) annual 1998
irradiance that reached established and new shoots of eel-     have more far-reaching negative impacts on eelgrass
grass in estuaries of Waquoit Bay receiving low (5 to 8 kg N    habitat and water quality (Figs. 2–6), and these issues
ha–1 yr–1, Timms and Sage Lot Ponds as replicates) versus     continue to be addressed on a regional/local level on
higher (28 to 30 kg N ha–1 yr–1, Hamblin and Jehu Ponds as     Cape Cod.
replicates) loads of land-derived nitrogen. Paired t-tests for
values throughout the annual cycle were used to determine
                                   Two general recommendations emerge from this work
whether there were significant differences between low and     for managers investing in eelgrass preservation. First,
      higher-nitrogen estuaries (***p < 0.001)        since this comparison and others (Valiela et al. 2000b,
70                     Mar Ecol Prog Ser 247: 59–73, 2003




Valiela & Cole 2002) show that eelgrass is lost within a   Sage Lot and Hamblin Ponds do not consistently reflect
relatively low and narrow range of nitrogen loading      the slight but crucial increase in wastewater nitrogen
rates, watersheds should be developed or managed such     inputs that apparently contribute to eelgrass decline
that land-derived loads are kept low. The threshold      (Table 5). A more thorough investigation of this ap-
value necessary for eelgrass preservation is difficult to   proach in detecting wastewater nitrogen inputs across
establish accurately, since many factors may influence    the relative small loading range relevant to eelgrass
land-derived nitrogen loading and fate in estuaries (i.e.   decline is highly recommended and may still reveal
retention by surrounding marsh, water residence time:     this technique to be useful for the management of eel-
Valiela et al. 2000a, 2001), but the present results and   grass preservation.
others (Valiela et al. 2000b, Valiela & Cole 2002) suggest
that eelgrass is likely to decline substantially at
values <30 to 100 kg N ha–1 yr–1. Several strategies, some   Relative contribution by eelgrass, macroalgae and
economically feasible and some not, within already-       phytoplankton to total primary productivity under
developed communities may reduce nitrogen loads to             low- or high-nitrogen loads
such levels (Valiela et al. 2000b: sewering towns, green
space/salt marsh preservation, minimizing fertilizer       Our comparison shows a shift from eelgrass-domi-
use, etc.).                          nated to macroalgal-dominated communities follow-
  Our second recommendation involves assessing eel-     ing increased eutrophication, similar to the results of
grass health. Because eelgrass restoration is difficult    past field comparisons and manipulative experiments
(Harrison 1990, Davis & Short 1997, Davis et al. 1998),    (Kemp et al. 1983, Twilley et al. 1985, Sand-Jensen &
simple but accurate indicators of incipient eelgrass de-   Borum 1991, Duarte 1995, Short et al. 1995). To exam-
cline due to nitrogen loading are needed. Since de-      ine the implications of such shift on total primary pro-
pressed shoot recruitment and increased mortality seem    duction in Waquoit Bay estuaries, we compared annual
to be important processes in eelgrass decline, we       eelgrass, macroalgae, and phytoplankton production,
recommend routine monitoring of shoot density within     standardized to estuarine area, in 2 estuaries repre-
meadows and, if possible, eelgrass bed area. Since algae,   senting extremes of the nitrogen loading gradient
like eelgrass, are very sensitive indicators of nitrogen   found in Waquoit Bay (Table 6). Under low-nitrogen
loading (Fig. 6), routine monitoring of macroalgal distri-  conditions, eelgrass production within the meadow
butions/canopy heights and Secchi depths (in estuaries    (i.e. scaled to m2 of meadow, 515 g m–2 yr–1) was similar
with longer residence times) may also prove useful.      in magnitude to that of macroalgae and approximately
  In contrast, shoot growth measurements require       twice that of phytoplankton. However, after extrapo-
SCUBA, are time-consuming, and did not yield a rela-     lating to m–2 of estuarine area, production by eelgrass
tionship with nitrogen loading rate in the meadows we     was lower than that by both macroalgae and phyto-
studied. Morphological features of eelgrass and shoot     plankton, even under low-nirogen conditions. Under
biomass vary widely among and within stable pop-       high-nitrogen conditions, eelgrass disappeared, macro-
ulations (van Lent & Verschuure 1994) and may not       algal production almost tripled, and phytoplankton
be useful indicators. Physiological measurements of      production more than doubled.
eelgrass tissues (C:N, [chlorophyll a]) require access      In the Waquoit system, total primary production in
to expensive scientific equipment and again, do not      the estuary exposed to the highest annual load of nitro-
necessarily yield consistent comparative information     gen was more than twice that in the estuary exposed to
(J. Hauxwell et al. unpubl. data), although C:N ratios    the lowest load of nitrogen (Table 6). Replacement of
have been hypothesized to be potential indicators of     eelgrass habitat by macroalgal- and phytoplankton-
nutrient availability (Fourqurean et al. 1997). Recent
findings by McClelland & Valiela (1998) suggested       Table 5. δ15N values in macroalgae from 2 estuaries of
that stable isotopic nitrogen signatures (δ15N) mea-     Waquoit Bay subject to different rates of nitrogen loading.
                               Macroalgal samples were composites of specimens collected
sured in estuarine primary producers may be useful in
                               from 5 sites within each estuary in November 1999 (means
detecting wastewater-derived nitrogen loading. Over              ± SE of replicate composites)
a broad range of nitrogen loading rates (5 to 407 kg N
ha–1 yr–1), this signature could even be used to estimate    Species              Estuary
wastewater-nitrogen loading rates. Preliminary evi-                   Sage Lot Pond Hamblin Pond
dence, however, suggests that this approach is not
sensitive enough to detect differences in wastewater      Cladophora vagabunda    3.5 ± 0.09   4.1 ± 0.09
                                Gracilaria tikvahiae    5.8 ± 0.05   5.6 ± 0.06
input within the low and narrow range of nitrogen
                                Codium fragile       6.1 ± 0.07   6.5 ± 0.07
loads in which eelgrass disappears. Stable isotopic       Fucus vesiculosis     4.8 ± 0.00   4.2 ± 0.07
nitrogen signatures measured in macroalgae from
                     Hauxwell et al.: Eutrophication and eelgrass loss                 71




Table 6. Comparison of annual net production estimates of eelgrass Zostera marina at reclaiming nitrogen from senescent
(normalized to estuarine surface area), macroalgae (Hauxwell et al. [1998], ex-  leaf material (Borum et al. 1989,
tended for the annual cycle), and phytoplankton (J. H. Foreman et al. unpubl. data)
                                         Pedersen & Borum 1992), our estimate
in 2 estuaries of Waquoit Bay subject to relatively low (Sage Lot Pond, 0.5 g N m–2
                                         of nitrogen incorporation for eelgrass
yr–1) or high (Childs River, 36 g N m–2 yr–1) rates of land-derived nitrogen loading
                                         may be high (Table 6). Pedersen & Bo-
(see Table 1). Production is expressed in terms of biomass, carbon fixation, and ni-
trogen incorporation (g DW m–2 yr–1). To convert from biomass units to carbon fix-rum (1993) estimated total nitrogen up-
ation or nitrogen incorporation, we assumed carbon and nitrogen content relative take to be 73% from external sources
to eelgrass biomass to be 40 and 3%, respectively (J. Hauxwell pers. obs.). Carbon
and nitrogen content relative to macroalgal biomass were 25 and 2–3%, respec-
                                         and 27% from internal recycling.
tively (Hauxwell et al. 1998). For phytoplankton, we assumed a carbon to biomass Hence, a more realistic estimate of eel-
        ratio of 0.4 and a C:N molar ratio of 7:1 (Redfield 1958)     grass nitrogen demand in the low-ni-
                                         trogen estuary might be approxi-
 Producer      Biomass       Carbon       Nitrogen       mately 3.7 g N m–2 yr–1. Overall, total
         low-   high-   low-   high-   low-    high-    nitrogen demand for primary produc-
        nitrogen nitrogen  nitrogen nitrogen  nitrogen nitrogen     tion shifted from ~42 to 101 g N m–2 yr–1
 Eelgrass    0167   0000    067    000    05     000
                                         across the range of nitrogen loads en-
 Macroalgae   0725   2071    180    518    22     062     countered in Waquoit Bay. Since land-
 Phytoplankton 0235    0583    094    233    15     039     derived loading rates of nitrogen were
 Total     1127   2654    341    751    42     101     only 0.5 g N m–2 yr–1 in the low-nitro-
                                         gen estuary and 36 g N m–2 yr–1 in the
                                         high-nitrogen estuary, regenerated ni-
dominated communities under high rates of nitrogen       trogen and/or nitrogen imported in seawater during
loading may have resulted in a 120% increase in total     tidal exchange must have supported 99% of nitrogen
carbon fixation. Annual measurements of net eco-        production in the low-nitrogen estuary and 64% in the
system production for these estuaries corroborate       high-nitrogen estuary. Of the 13 estuaries summarized
this difference; ecosystem net production was 4-fold      in Nixon & Pilson (1983), only the lower New York Bay
greater in the high-nitrogen versus the low-nitrogen      and a section of San Francisco Bay received more ‘new’
estuary (D’Avanzo et al. 1996). These results are con-     nitrogen from land than was required by primary pro-
trary to the non-relationship between total primary      ducers; for the remaining estuaries, nitrogen regen-
production and nitrogen loading rates recorded by       eration was estimated to support 50 to 91% of primary
Borum & Sand-Jensen (1996), who found that for many      production, similar to the range represented by the
systems phytoplankton productivity was stimulated,       Waquoit system.
but that benthic production declined as nitrogen load-
ing rates increased. This was probably due to phyto-
                                Acknowledgements. This research was supported by an Envi-
plankton shading of macroalgae, rendering them less
                                ronmental Protection Agency STAR Fellowship for Graduate
productive than the seagrass systems they replace.       Environmental Study (U-915335-01-0) and a National Estuar-
Two factors may explain the discrepancy between        ine Research Reserve Graduate Research Fellowship from
our results and the findings of Borum & Sand-Jensen:      the National Oceanic and Atmospheric Administration (award
(1) for the shallow Waquoit system, macroalgae have      number NA77OR0228) awarded to J.H. We thank the Que-
                                bec-Labrador Foundation Atlantic Center for the Environ-
higher areal rates of production in estuaries of higher-    ment’s Sounds Conservancy Program and the Boston Univer-
nitrogen loads despite stimulated phytoplankton        sity Ablon/Bay Committee for research funds. We are grate-
growth; (2) eelgrass production in the Waquoit system     ful to Richard Payne for providing surface irradiance data and
under low-nitrogen loads is relatively low (515 g DW      the Waquoit Bay Estuarine Research Reserve volunteer Bay-
 –2        –1                     watchers and David Giehtbrock for providing temperature
m of meadow yr ) compared to the worldwide aver-
                                data. Anne Giblin, Paulette Peckol, Philip Lobel, Rainer Voigt,
age (J. Hauxwell et al. unpubl. data: 1145 g DW        and Jennifer Bowen provided valuable comments on prelimi-
m–2 yr–1) due to low shoot densities relative to other     nary drafts of the manuscript. We also thank the Waquoit Bay
populations. Overall, these shifts in quantity and qual-    National Estuarine Research Reserve for the use of their facil-
ity of organic material imply ecologically significant     ities throughout the study.
consequences in terms of carbon fixation and the vari-
ous fates of production (changes in rates of herbivory,               LITERATURE CITED
decomposition, storage, and export).
  As in several other systems for which nitrogen bud-     Adobe Systems (1996) Adobe Photoshop 4.0. Adobe Systems
gets have been estimated (summarized in Nixon &          Incorporated, San Jose, CA
                                Bannister TT (1974) Production equations in terms of chloro-
Pilson 1983), estimated nitrogen incorporation by all       phyll concentration, quantum yield, and upper limit to
producers in the Waquoit estuaries exceeded land-         production. Limnol Oceanogr 19:1–12
derived inputs (Table 6). Since eelgrass may be efficient   Borum J, Sand-Jensen K (1996) Is total primary production
72                       Mar Ecol Prog Ser 247: 59–73, 2003




  in shallow coastal marine waters stimulated by nitrogen    Hauxwell J, McClelland J, Behr PJ, Valiela I (1998) Relative
  loading? Oikos 76:406–410                     importance of grazing and nutrient controls of macroalgal
Borum J, Murray L, Kemp WM (1989) Aspects of nitrogen        biomass in three temperate shallow estuaries. Estuaries
  acquisition and conservation in eelgrass plants. Aquat Bot    21:347–360
  35:289–300                           Hauxwell J, Cebrián J, Furlong C, Valiela I (2001) Macroalgal
Bowen JL, Valiela I (2001) The ecological effects of urbaniza-    canopies contribute to eelgrass (Zostera marina) decline
  tion of coastal watersheds: Historical increases in nitrogen   in temperate estuarine ecosystems. Ecology 82:1007–1022
  loads and eutrophication of Waquoit Bay estuaries. Can     Hersh DA (1996) Abundance and distribution of intertidal and
  J Fish Aquat Sci 58:1489–1500                   subtidal macrophytes in Cape Cod: The role of nutrient
Brouns JJWM (1985) The plastochrone interval method for       supply and other controls. PhD thesis, Boston University
  the study of the productivity of seagrasses: possibilities   Jackson LE, Kurtz JC, Fisher WS (eds) (2000) Evaluation
  and limitations. Aquat Bot 21:71–88                guidelines for ecological indicators. EPA/620/R-99/005.
Burkholder JM, Mason KM, Glashow HB Jr (1992) Water-         US Environmental Protection Agency, Office of Research
  column nitrate enrichment promotes decline of eelgrass      and Development, Research Triangle Park, NC
  Zostera marina: evidence from seasonal mesocosm exper-     Kemp WM, Twilley RR, Stevenson JC, Boynton WR, Means
  iments. Mar Ecol Prog Ser 81:163–178               JC (1983) The decline of submerged vascular plants in
Burkholder JM, Glasgow HB Jr, Cooke JE (1994) Compara-        upper Chesapeake Bay: Summary of results concerning
  tive effects of water-column nitrate enrichment on eel-      possible causes. Mar Technol Soc J 17:78–89
  grass Zostera marina, shoalgrass Halodule wrightii, and    Kirk JTO (1994) Light and photosynthesis in aquatic ecosys-
  widgeongrass Ruppia maritima. Mar Ecol Prog Ser 105:       tems. Cambridge University Press, Cambridge
  121–138                            Koch MS, Mendelssohn IA, McKee KL (1990) Mechanism for
Cebrián J (1999) Patterns and the fate of production in plant    the hydrogen sulfide-induced growth limitation in wet-
  communities. Am Nat 154:449–468                  land macrophytes. Limnol Oceanogr 35:396–407
Cebrián J, Williams M, McClelland J, Valiela I (1998) The     Lorenzen CJ (1966) Determination of chlorophyll and pheo-
  dependence of heterotrophic consumption and carbon        pigments: spectrophotometric equations. Limnol Oceanogr
  accumulation on autotrophic nutrient content in ecosys-      12:343–346
  tems. Ecol Lett 1:165–170                   Marsh JA Jr, Dennison WC, Alberte RS (1986) Effects of tem-
Cloern JE (2001) Our evolving conceptual model of the        perature on photosynthesis and respiration in eelgrass
  coastal eutrophication problem. Mar Ecol Prog Ser 210:      (Zostera marina L.). J Exp Mar Biol Ecol 101:257–267
  223–253                            McClelland JW, Valiela I (1998) Linking nitrogen in estuarine
Cohen JE, Small C, Mellinger A, Gallup J, Sachs JD (1997)      producers to land-derived sources. Limnol Oceanogr 43:
  Estimates of coastal populations. Science 278:1211        577–585
Cottam C (1933) Eelgrass dying of mysterious disease. Sci     Moore KA, Wetzel RL (2000) Seasonal variations in eelgrass
  News Lett 24:73                          (Zostera marina L.) response to nutrient enrichment and
Cotton AD (1933) Disappearance of Zostera marina. Nature       reduced light availability in experimental ecosystems.
  132:277                              J Exp Mar Biol Ecol 244:1–28
D’Avanzo C, Kremer JN, Wainright SC (1996) Ecosystem pro-     Neckles HA, Koepfler ET, Haas LW, Wetzel RL, Orth RJ
  duction and respiration in response to eutrophication in     (1994) Dynamics of epiphytic photoautotrophs and hetero-
  shallow temperate estuaries. Mar Ecol Prog Ser 141:        trophs in Zostera marina (eelgrass) microcosms: Responses
  263–274                              to nutrient enrichment and grazing. Estuaries 17:597–605
Davis RC, Short FT (1997) Restoring eelgrass, Zostera marina   Nixon SW (1992) Quantifying the relationship between nitro-
  L., habitat using a new transplanting technique: the hori-    gen input and the productivity of marine ecosystems. In:
  zontal rhizome method. Aquat Bot 59:1–15             Takahashi M, Nakata K, Parsons TR (eds) Proceedings
Davis RC, Short FT, Burdick DM (1998) Quantifying the        of the Advanced in Marine Technology Conference, No. 5
  effects of green crab damage to eelgrass transplants.       Marine Technology Society of Japan, Japan, p 57–83
  Restor Ecol 6:297–302                     Nixon SW, Pilson MEQ (1983) Nitrogen in estuarine and
Duarte C (1995) Submerged aquatic vegetation in relation to     coastal marine ecosystems. In: Carpenter EJ, Capone DG
  different nutrient regimes. Ophelia 41:87–112           (eds) Nitrogen in the marine environment. Academic
Duarte CM, Marbà N, Agawin NSR, Cebrián J and 8 others        Press, New York, p 565–648
  (1994) Reconstruction of seagrass dynamics: age determi-    Nixon SW, Oviatt CA, Frithsen J, Sullivan B (1986) Nutrients
  nation and associated tools for the seagrass ecologist. Mar    and the productivity of estuarine and coastal marine
  Ecol Prog Ser 107:195–209                     ecosystems. J Limnol Soc South Africa 12:43–71
Duarte CM, Merino M, Agawing NSR, Uri J, Fortes MD, Gal-     Nixon SW (1995) Coastal marine eutrophication: a definition,
  legos ME, Marbà N, Hemminga MA (1998) Root produc-        social causes and future concerns. Ophelia 41:199–219
  tion and belowground seagrass biomass. Mar Ecol Prog      NRC (National Research Council) (1994) Priorities for coastal
  Ser 171:97–108                          science. National Academy Press, Washington, DC
Fourqurean JW, Moore TO, Fry B, Hollibaugh JT (1997)       Peckol P, Rivers JS (1996) Contribution by macroalgal mats to
  Spatial and temporal variation in C:N:P ratios δ15N, and     primary production of a shallow embayment under high
  δ13C of eelgrass Zostera marina as indicators of ecosystem    and low nitrogen-loading rates. Estuar Coast Shelf Sci 43:
  processes, Tomales Bay, California, USA. Mar Ecol Prog      311–325
  Ser 157:147–157                        Peckol P, DeMeo-Anderson B, Rivers J (1994) Growth, nutri-
Goodman JL, Moore KA, Dennison WC (1995) Photosynthetic       ent uptake capacities, and tissue constituents of the macro-
  responses of eelgrass (Zostera marina L.) to light and sedi-   algae Gracilaria tikvahiae and Cladophora vagabunda,
  ment sulfide in a shallow barrier island lagoon. Aquat Bot    related to site-specific nitrogen loading rates. Mar Biol
  50:37–47                             121:175–185
Harrison PG (1990) Variations in success of eelgrass trans-    Pedersen MF, Borum J (1992) Nitrogen dynamics of eelgrass
  plants over a five-year period. Environ Conserv 17:157–163    Zostera marina during a late summer period of high
                     Hauxwell et al.: Eutrophication and eelgrass loss                   73




  growth and low nutrient availability. Mar Ecol Prog Ser 80:    Waquoit Bay, Massachusetts. Masters thesis, Boston Uni-
  65–73                               versity
Pedersen MF, Borum J (1993) An annual nitrogen budget for      Taylor D, Nixon S, Granger S, Buckley B (1995a) Nutrient
  a seagrass Zostera marina population. Mar Ecol Prog Ser      limitation and the eutrophication of coastal lagoons. Mar
  101:169–177                            Ecol Prog Ser 127:235–244
Petersen HE (1934) Wasting disease of eelgrass (Zostera       Taylor DI, Nixon SW, Granger SL, Buckley BA, McMahon JP,
  marina). Nature 134:143–144                    Lin HJ (1995b) Responses of coastal lagoon plant commu-
Pickett STA (1989) Space-for-time substitution as an alterna-     nities to different forms of nutrient enrichment — a meso-
  tive to long-term studies. In: Likens GE (ed) Long-term      cosm experiment. Aquat Bot 52:19–34
  studies in ecology: approaches and alternatives. Springer-   Terrados J, Duarte CM, Kamp-Nielsen L, Agawin NSR and 6
  Verlag, New York, p 110–135                    others (1999) Are seagrass growth and survival con-
Pregnall AM, Smith RD, Kursar TA, Alberte RS (1984) Meta-       strained by the reducing conditions of the sediment?
  bolical adaptations of Zostera marina (eelgrass) to diurnal    Aquat Bot 65:175–197
  periods of root anoxia. Mar Biol 83:141–147           Tilman D, Fargione J, Wolff B, D’Antonio C and 6 others
Rasmussen E (1973) Systematics and ecology of the Isefjord       (2001) Forecasting agriculturally driven global environ-
  marine fauna (Denmark) with a survey of the eelgrass        mental change. Science 292:281–284
  (Zostera) vegetation and its communities. Ophelia 11:      Twilley RR, Kemp WM, Staver KW, Stevenson JC, Boynton
  1–495                               WR (1985) Nutrient enrichment of estuarine communities.
Redfield AC (1958) The biological control of chemical factors     1. Algal growth and effects on production of plants and
  in the environment. Am Sci 46:205–221               associated communities. Mar Ecol Prog Ser 23:179–191
Renn CE (1935) A mycetozoan parasite of Zostera marina.       Valiela I, Cole ML (2002) Comparative evidence that salt
  Nature 135:544–545                         marshes and mangroves may protect seagrass meadows
Safran RE, Legra JC, Valiela I (1998) Effects of nitrogen       from land-derived nitrogen loads. Ecosystems 5:92–102
  loading on eelgrass seed coat abundance, C to N ratios,     Valiela I, Collins G, Kremer J, Lathja K, Geist M, Seely B,
  and δ15N in sediments of Waquoit Bay. Biol Bull 195:        Brawley J, Sham CH (1997a) Nitrogen loading from
  245–246                              coastal watersheds to receiving estuaries: new methods
Sand-Jensen K (1975) Biomass, net production and growth        and application. Ecol Appl 7:358–380
  dynamics in an eelgrass (Zostera marina L.) population in    Valiela I, McClelland J, Hauxwell J, Behr PJ, Hersh D, Fore-
  Vellerup Vig, Denmark. Ophelia 14:185–201             man K (1997b) Macroalgal blooms in shallow estuaries:
Sand-Jensen K, Borum J (1991) Interactions among phyto-        controls and ecophysiological and ecosystem conse-
  plankton, periphyton, and macrophytes in temperate         quences. Limnol Oceanogr 42:1105–1118
  freshwaters and estuaries. Aquat Bot 41:137–175         Valiela I, Geist M, McClelland J, Tomasky G (2000a) Nitrogen
SAS Institute (1999) Statview. SAS Institute, Cary, NC        loading from watersheds to estuaries: Verification of the
Short FT, Burdick DM (1996) Quantifying eelgrass habitat        Waquoit Bay nitrogen loading model. Biogeochemistry 49:
  loss in relation to housing development and nitrogen load-     277–293
  ing in Waquoit Bay, Massachusetts. Estuaries 19:730–739     Valiela I, Tomasky G, Hauxwell J, Cole ML, Cebrián J,
Short FT, Wyllie-Echeverria S (1996) Natural and human-        Kroeger KD (2000b) Operationalizing sustainability: man-
  induced disturbance of seagrasses. Environ Conserv 23:       agement and risk assessment of land-derived nitrogen
  17–27                               loads to estuaries. Ecol Appl 10:1006–1023
Short FT, Mathieson AC, Nelson JJ (1986) Recurrence of the     Valiela I, Cole ML, McClelland J, Hauxwell J, Cebrián J, Joye
  eelgrass wasting disease at the border of New Hampshire      S (2001) Role of salt marshes as part of coastal landscapes,
  and Maine, USA. Mar Ecol Prog Ser 29:89–92             In: Weinstein MP, Kreeger DE (eds) Concepts and contro-
Short FT, Burdick DM, Wolf JS, Jones GE (1993) Eelgrass in       versies in tidal marsh ecology. Kluwer Academic Publish-
  estuarine research reserves along the East Coast, USA,       ers, Dordrecht, p 23–28
  Part I: Declines from pollution and disease; Part II: Man-   Van Katwijk MM, Vergeer LHT, Schmitz GHW, Roelofs JGM
  agement of eelgrass meadows. National Oceanic and         (1997) Ammonium toxicity in eelgrass Zostera marina.
  Atmospheric Administration, Coastal Ocean Program         Mar Ecol Prog Ser 157:159–173
  Publication, NOAA, Rockville, MD                van Lent F, Verschuure JM (1994) Intraspecific variability of
Short FT, Burdick DM, Kaldy JE III (1995) Mesocosm experi-       Zostera marina L. (eelgrass) in the estuaries and lagoons of
  ments quantify the effects of eutrophication on eelgrass,     the southwestern Netherlands. I. Population dynamics.
  Zostera marina. Limnol Oceanogr 40:740–749             Aquat Bot 48:31–58
Sokal RR, Rohlf FJ (1995) Biometry. The principles and prac-    Zieman JC, Wetzel RG (1980) Productivity in seagrasses:
  tice of statistics in biological research, 3rd edn. WH Free-    methods and rates. In: Phillips RC, McRoy CP (eds) Hand-
  man, New York                           book of seagrass biology. Garland Press, New York,
Stieve E (2001) Controls of macrophyte abundance in          p 87–116

Editorial responsibility: Kenneth Heck (Contributing Editor),    Submitted: April 16, 2001; Accepted: October 4, 2002
Dauphin Island, Alabama, USA                    Proofs received from author(s): January 13, 2003