Vol. 168: 259-271,1998
                      MARINE ECOLOGY PROGRESS SERIES
                          Mar Ecol Prog Ser
                                                  1
                                                  I
                                                      Published July 9




Changes in food web structure under the influence
  of increased anthropogenic nitrogen inputs
          to estuaries
                    James W. McClelland*, I. Valiela
     Boston University Marine Program, Marine Biological Laboratory. Woods Hole. Massachusetts 02543, USA




       ABSTRACT: Anthropogenic nitrogen loads to shallow coastal waters have been linked to shifts from
       seagrass- to algae-dominated communities in many regions of the world, yet the influence that these
       shifts have on the structure of nearshore food webs remains unclear. We used stable C and N isotope
       ratios to assess the relative importance of phytoplankton, macroalgae, and eelgrass Zostera marina as
       food sources to consumers in estuaries of Waquoit Bay, Massachusetts, USA, that receive low versus
       high nitrogen loads from their surrounding watersheds. We found that, in general, the diets of herbi-
       vores, suspension feeders, and detritlvores (collectively referred to as primary consumers) in the
       Waquoit Bay estuaries are influenced by the dominant forms of production that they are exposed to.
       Phytoplankton and macroalgae are the major food sources of primary consumers under both low and
       high N loading conditions, but eelgrass is also an important food so'urce where N loading is low Most
       of the primary consumers at the estuary with low N loading have between 10 and 16 % eelgrass in their
       diets. Some species, however, appear to have no eelgrass in their diets, while others have diets consist-
       ing of as much as 31 % eelgrass. Eelgrass C and N are passed on to benthic as well as pelagic secondary
       consumers. With losses of eelgrass as a consequence of nitrogen loading, an important pathway
       through which land-derived nitrogen enters food webs in the Waquoit Bay estuaries is eliminated. This
       fundamental change probably affects the rate at which land-derived nitrogen is cycled within estuar-
       ies, as well as its ultimate fate.

       KEY WORDS: Nitrogen loading . Estuanes . Food webs . Coastal watersheds . Urbanization




           INTRODUCTION                  In some shallow temperate waters, eelgrass Zostera
                                  marina dominates production under oligotrophic condi-
 Anthropogenic nutrient loads from coastal water-         tions, but is replaced by proliferations of phytoplankton
sheds to nearshore waters are dramatically altering        and macroalgae as eutrophication progresses. This shift
aquatic habitats (GESAMP 1990, National Research          occurs because eelgrass production is light-limited,
Council 1994). In estuarine and coastal marine waters,       whereas phytoplankton and macroalgal production are
increased nitrogen loads stimulate eutrophication         nutrient-limited (Duarte 1995). Increased nitrogen
(Kelly & Levin 1986, Nixon 1986, Galloway et al. 1995).      availability promotes algal growth, and as a result, eel-
As a consequence, the assemblage of primary produc-        grass is lost due to shading (Valiela et al. 1997a). A
ers found in these waters often changes (Orth & Moore       well-studied example of this phenomenon comes from
1983, Sand-Jensen & Borum 1991, Costa et al. 1992,         the Waquoit Bay system on Cape Cod, Massachusetts,
Duarte 1995, Short & Burdick 1996, Valiela et al.         USA (Fig. 1). In the 1950s eelgrass grew throughout
1997a).                              Waquoit Bay and the estuaries adjoining it (Valiela et
                                  al. 1992). Over the past 40 yr, however, nitrogen load-
                                  ing has increased phytoplankton and macroalgal pro-
                                  duction, causing eelgrass production to decrease as a
                                  secondary effect (Valiela et al. 1997a).

0 Inter-Research 1998
Resale of full article not permitted
  260                                                 mar Ecol Prog Ser 168: 259-271,        1998




              ' / i
              \
              L
              '
              ,                -
                                ;,
                                \            I
                                             I
                                             I
                                                                      studled recelve relatively low (Sage Lot Pond), high
                                                                      (Quashnet River), and higher (Childs Rlver) n ~ t r o g e n
                \
                                                                      loads (Table 1) In Sage Lot Pond, eelgrass is an irnpor-
                                                                      tant source of production, but at Quashnet Rlver and
              /FL                                                      Chllds River, production is dominated by macroalgae
           /I         r X                                               (Table 1) By maklng comparisons among estuaries
           II
           I I                                                         within the Waquoit Bay system, we galn a better
          I I               h
         I I             I                                              understanding of changes that take place in estuanes
         I I
        I I            l                                                as u ~ b a n ~ i a t i o n coastal watersheds Increases Thls
                                                                                  of
        I I             \                                                approach IS referred to as a space-for-time substttut~on
       I I              \
       I I                 \  *                                         (Pickett 1991) Of course, many factors are not con-
       I I                 \  '               h
        1          3         \\         4                                trolled In a quasi-expenment such as thls, addlng some
        l I                                                             uncertainty to data Interpretation Worklng at the scale
        I I
       I I                                                             of whole watersheds, however, reveals novel ~nforma-
       I I
     1    1          ,                                                   tion about ecosystem function that cannot be learned
     I    I                                                            by worklny at smaller scales Hence, the added uncer-
    I     \
                                                                      talnty Inherent to lnterpretatlon of data from cross-sys-
                                                                      t e n corr,peir:socs 1s oztwe:ghed by the scaic-spcc:fic
                                                   \
                                                                1)     information gained w ~ t h  this approach
                                                     \           N



                                                 "
    I                                                  \                Although it IS clear that pnmary producer assemblages
                                                         \
                                                           \   -        change in response to nutrient loadlng, the Influence of
                                                     --P         lkm
                                                                      such changes on the strl~ctlire ~ s t ~ ~ i i r i nW P ~ 1s
                                                                                       of    food ~    S

                                                         /
                                                                      poorly understood If all producers were equally edtble
                                                       /
   I         I                                                         (and accessible), then consumers would s ~ m p l y them
                                                                                               eat
  I
  J       I
           /                        4.
                                    "           0  'f '                   In proportion to their productlon, and the diets of con-
                                                 I
1-h1                   Waquo~t



                                     --
                                   /

                                                 ',
                                           A<--

I
                   /
                     Bay            /'
                                     /-
                                       -  7             /    /'          sumers would change as the mix of producers changed
                                                                      In fact, however, d~fferencesin chemlcal composition
I
I        --
          - .
          5
                -<             ,
                              '
                                 /.:
                                 <    L
                                     '
                                     -
                                     -/
                                      -   ,   ZCP ' /

                                                                      a n d morphological charactenstics among producers
                   \             ._-
                                                                      make them differ In edibility (Cebnan & Duarte 1994)
                            V~neyardSound
                                                                      For example, nutnent-poor structural tissues in eelgrass
                                                                      make it less desirable as a food source than phytoplank-
Fig. 1 T h e Waquolt Bay xvatershed-estuary system, C a p e
Cod, Massachusetts, USA. Dashed lines separate the W a q u o ~ t
                                                                      ton and macroalgae. Edibility also varies wldely among
Bav watershed into catchment basins associated xvith dlffer-                                        macroalgal species (NicOtn lg80, Heckscher et    lgg6).
e n t estuanes surrounding Waquoit Bay Each catchment                                           These differences in edtbility make it difficult to predict
basln is named for the estuary that it empties into: 1 = T ~ m m s                                     how a n d to what extent nutrient-loading-lnduced
Pond, 2 = Eel Pond, 3 = Childs River, 4 = Quashnet River,
                                                                      changes in       producer assemblages in coastal
   5 = Hamblin Pond, 6 = J e h u Pond, 7 = Sage Lot Pond
                                                                      waters affect food web structure.


                                                         Table 1 Nitrogen loads to the Sage Lot Pond, Quashnet Rlver, and Childs Rlver
 The W a q u o ~ t
         Bay system consists of
                                                         estuanes of Waquoit Bay, a n d percentages of primary product~oncontnbuted
7 shallow estuaries (maximum depth                                        by phytoplankton, macroalgae, a n d eelgrass wlthin these estuaries N - l o a d ~ n p
of 3 m , and average depth of 1 m at                                       a n d percentage primary product~onestimates are based on Waquoit Bdy Land
mean low water) that receive fresh-                                       b4argin Ecosystems Research (WBLMER) measurements (D'Avanzo et a1 1996,
water inputs from separate catchment                                               Peck01 & Rlvers 1996, WBLMER unpubl data)
basins within the Waquoit Bay water-
shed (Fig. 1 ) . Freshwater enters the                                                                  Estuary
estuaries almost exclusively via                                                            Sage Lot Pond  Quashnet River  Childs R ~ v e r
groundwater flow, a n d with it land-
derived nitrogen is conveyed to the
                                                         1    Nitrogen load ( k g N ha.' yr-l)
                                                             % of pnmary production by.
                                                                                64      520       624      1
estuaries (Val.iela et al. 1992). The                                           Phytoplankton            26
nitrogen loads delivered to each estu-                                          Macroalgae
ary differ according to land use on the                                          Cladophora vayabunda        19
surrounding watershed (Valiela et al.                                           Gracrlana t ~ k v a h i a e     19
                                                             Eelgrass
3997b). The 3 estuanes of Waquoit                                             Zostera marina           37
Bay that have been most extensively
                 McClelland & Valiela: Chalnges in food w e b structure              26 1




  An effective means to elucidate food web structure is     In this paper, we use stable C and N isotope ratios
to examine the stable carbon and nitrogen isotope       together to investigate how changes in primary pro-
ratios (expressed as 613C and 6I5N) of producers and      ducer assemblages brought about by increased waste-
consumers within a community (for reviews see Fry &      water N loads influence the diets of consumers in
Sherr 1984, Peterson & Fry 1987, and Lajtha & Mich-      Waquoit Bay. In particular, we assess the relative
ener 1994). This is because different producers often     importance of phytoplankton, macroalgae, and eel-
have unique stable isotope ratios that make them iden-     grass as food sources to the consumers. We focus our
tifiable in mixes of particulate organic matter (Haines    work on the Sage Lot Pond, Quashnet River, and
1977, Gearing 1988, Canuel et al. 1995) and in the tis-    Childs River estuaries, addressing 3 main questions:
sues of consumers (Zieman et al. 1984, Peterson et al.     (1) What are major primary producers contributing to
1985, Fry 1991).                        particulate organic matter (POM) in each estuary?
  Use of 613C and F15N in food web studies relies on     (2) What are the major food sources of consumers in
the assumption that the isotopic composition of a con-     estuaries with low versus high N loads? (3) How do
sumer (or mixture of particulate organic matter) is      pathways of C and N flow through estuarine food webs
equal to the weighted average of the isotopic compo-      change as eutrophication intensifies?
sitions of its food sources (Gannes et al. 1997). For
example, if an amphipod eats 2 food sources in equal
proportions, it is expected to have a 613C value                    METHODS
(whole body) that is intermediate to those of the 2
food sources. Where there are multiple food sources,       Sample collection. Biota and POM in Sage Lot Pond,
plots of 6I3C versus 615N provide a 2-dimensional       Quashnet River, and Childs River were sampled for
view of the diets of consumers. As carbon and nitro-      stable isotope analysis in November 1993, July 1994,
gen from producers are passed from one troph~c   level   May 1995, July 1995, and June 1996. These dates were
to the next, their 6I3C and F15N values tend to        chosen to cover spring, summer, and fall conditions
increase due to differential use of heavy and light iso-    within the estuaries, and to allow an interannual com-
topes during metabolism (Fry & Sherr 1984, Peterson      parison within the summer season. Samples were not
& Fry 1987, Lajtha & Michener 1994). There is also a      collected during winter months because of low animal
---fiw;n- 5&y of ex.ri.l,ence the! steb!e isotope rltics
3."  day                           sbur?d.-nce. Samples were col!ected from regions
can change during decomposition (Benner et al. 1987,      within each estuary characterized by a vertically
Currin et al. 1995). These factors must be considered     mixed water column with salinity between 25 and
in order to clearly interpret stable isotope data in food   30 PSU. Producers (except phytoplankton, see below)
web studies.                          and POM were sampled from all 3 estuaries, and con-
  Previously published nitrogen stable isotope data      sumers were sampled from Sage Lot Pond and Childs
from the Waquoit Bay system (McClelland et al. 1997)      River.
shows that differences in 615N values of biota among       We could not use suspended POM as a proxy for
the estuaries of Waquoit Bay are strongly correlated      phytoplankton isotope values-as    is often done in
with differences in nitrogen loading from wastewater.     open ocean .waters-because suspended POM in the
As the percentage of wastewater contributing to total     estuaries of Waquoit Bay has a large non-phytoplank-
N loading increases among estuaries, the 6'" values      ton (mainly detritus and amorphous organic aggre-
of estuarine biota increase. A second paper (McClel-      gates) conlponent. However, dense populations of
land & Valiela in press) defines the relationship       phytoplankton with minimal detrital and aggregate
between wastewater inputs to watersheds and the 6I5N      particles were available from the 'controls' of an ongo-
of dissolved inorganic nitrogen (DIN) in groundwater,     ing experiment on nutrient limitation of phytoplankton
and explicitly links groundwater 615N with producer      in Childs River (Tomasky & Valiela 1995). We mea-
6I5N in the different estuaries of Waquoit Bay. The 615N    sured the C and N stable isotope values of these 'con-
values of groundwater nitrate within the Waquoit Bay      trols'. Samples were collected for isotope analysis after
watershed increase from -0.9 to + 1 4 % as wastetvater     a July 1996 run of the experiment. Based on the fact
contributions increase from 4 % to 86% of the total      that the 613C values of other primary producers within
nitrogen pool. As a result, the average 615N of DIN      the Waquoit Bay system differed little among estuaries
(nitrate + ammonium) received by different estuaries      (data presented in this paper), we estimated the    of
around Waquoit Bay increases from +0.5%0to +9.5%o.       phytoplankton in Quashnet River and Sage Lot Pond to
This Increase is paralleled by the FI5N of eelgrass,      be the same as phytoplankton in Childs River. Simi-
while macroalgae, salt-marsh cordgrass, and sus-        larly, based on consistent differences in macroalgae
pended particulate organic matter increase to a lesser     615N among estuaries (McClelland et al. 1997, McClel-
degree among estuaries.                    land & Valiela in press, and data presented in this
262                     Mar Ecol Prog S e r 168: 259-271, 1998




paper) we estimated the 615N of phytoplankton in         The standard used for stable nitrogen isotope analysis
Quashnet River by subtracting l      from the Childs    was N, in air, and the standard used for stable carbon
River value, and we estimated the 615N of phytoplank-      isotope analysis was Vienna Peedee belemnite. 613C
ton at Sage Lot Pond by subtracting 2% from the         and 615N values, expressed as %, were calculated as
Childs River value.                       [(Rs,,ple/R,t,,d,,d)-l] X 103, with R equal to 15N/14Nor
  Other producers and consumers were sampled as         I3C/l2C. Precision of replicate analyses was *0.2% for
follows: Suspended POM was sampled by collecting 2 1       nitrogen and + 0 . 3 X for carbon.
bottles of sea water from a depth of 0.5 m below the
surface at 3 locations within each estuary. Benthic
POM was sampled by collecting bare sediment (ben-               RESULTS AND DISCUSSION
thic microbial and microalgal mats are not a prominent
feature in the estuaries of Waquoit Bay) to a depth of       In the sections below, we first examine the stable
1 cm at 10 locations within each estuary. Sediment        carbon and nitrogen isotope values of primary produc-
type in all 3 estuaries is a mixture of sand, silt, and clay.  ers in Sage Lot Pond, Quashnet River, and Childs
Benthic macrophytes were sampled from 15 locations        River. We then discuss the relationship between iso-
within each estuary. Benthic invertebrates were sam-       tope values of POM, and isotope values of dominant
pled with an Ekman dredge from 15 locations within        producers in each estuary. Finally, we use stable car-
Sage Lot Pond and Childs River Dredge contents were       bon and nitrogen isotope values to compare the food,
rinsed through a 1 mm sieve and invertebrate species       webs in Sage Lot Pond and Childs River, and discuss
retained on the sieve were sorted by species. Fish were     changes in the flow of land-derived nitrogen into food
col.lected from Sage Lot Pond and Childs River by        webs as eutrophication intensifies. Although we focus
seine.                              our analysis on the producers that contribute most to
  Sample preparation. To prepare samples for stable       production within the estuaries of Waquoit Bay
isotope analysis, phytoplankton and suspended POM        (Table l ) ,we acknowledge that a variety of less promi-
were filtered from water samples onto precombusted        nent producers must also contribute to POM and to the
(490°C)Gelman A/E glass fiber filters with a low pres-      diets of consumers.
sure vacuum pump, macrophytes were gently cleaned
of epiphytic material, and animals were held in filtered
sea water for 24 h to allow their guts to clear. Both          Stable isotope values of primary producers
macrophytes and animals were then rinsed with deion-
ized water, and all samples were dried at 60°C. When        The 613C values of primary producers (except for the
dry, samples were ground (filters with POM were kept       macroalga Gracilaria tikvahiae, see below) change
whole) into a homogenous powder, and combined to         little among estuaries, while the 6"'N values of pri-
make single composite samples of each species per        mary producers consistently differ among estuaries
estuary per sampling date. Whole organisms were          (Table 2). The similarity of producer 613C is reasonable
used in all cases except for Mya arenaria and Geuken-      because the 613C of dissolved inorganic carbon (DIC)
sia demissa, the shells of which were removed. Sepa-       largely depends on salinity (Fogel et al. 1992),and pro-
rate composites consisting of 15 to 250 individuals       ducers were sampled from waters of the same salinity
(depending on the size of the species collected) were      range at all estuaries. As discussed in the introduction,
prepared for stable isotope analysis from each estuary      differences in 615N values of producers among estuar-
for each season. Samples for stable carbon isotope        ies are linked to differences in N loading from waste-
analysis were acidified to remove inorganic carbon,       water (McClelland et al. 1997, McClelland & Valiela in
then dried and ground a second time.               press). Both 615N and 6I3C of producers change little
  Stable isotope analysis. Samples were analyzed in       among seasons (Table 2), suggesting that the isotope
the Boston University Stable Isotope Laboratory using      values of DIN and DIC, and the fractionation associ-
a Finnigan Delta-S isotope ratio mass spectrometer. All     ated with their use, do not vary much throughout the
samples (except filters, see below) were weighed and       year within the Waqu.oit Bay system.
loaded into tin capsules and combusted in a Heraeus         Unlike the other producers, the 6'" of Gracilaria
element analyzer. The resulting combustion gases         tikvahiae in Sage Lot Pond is substantially lower
were cryogen.ically separated and purified in a Finni-      than at the other 2 estuaries (Table 2). The reason for
gan CT-CN trapping box before introduction into the       this anomaly is unclear. Perhaps increased nitrogen
mass spectrometer. Filters were acidified with PtC12,      availability in Childs River and Quashnet River
dried, then combusted using the Dumas combustion         allows G. tikvahiae to incorporate carbon m.ore effi-
technique. Combustion gases were again separated         ciently than in Sage Lot Pond, so that fractionation of
and purified before analysis (Lajtha & Michener 1994).      DIC during uptake by this species becomes sma.ller
                   McClelland P1 Valiela: Changes in food web structure                263




Table 2. Primary producer and POM isotope values ("L)of composite samples collected spring, summer, and fall between 1993
and 1996 from Sage Lot Pond. Quashnet River, and Childs River (average t 3 SE). Values in parentheses are number of seasons
   sampled. Values in italics are estimates of phytoplankton isotope values based on measured values at Childs River


                    Sage Lot Pond           Quashnet River         Childs River
                  6I3c     615~        6' '(1     6I5N       613C      6%

POM (Suspended)       -18.4  * 1.4 (5)   3.7 + 0.3 (S)
POM (Benthic)        -15.2  t  0.5 (2)  2.4 + 0.3 (2)
Pnrnary producers
 Phytoplankton         -21.1         5.1
 Macroalgae
  Cladophora vagabunda   -15.3  * 0.3 (4)     *
                          3.4 0 1 (4)
  Cracilana tikvahiae    -19.5  * 0.6 ( 4 )  5.1 t 0 6 (4)
  Enteromorpha sp.     -19.1  * 2.1 (2)     *
                          4.9 0.3 (2)
 Eelgrass
  Zostera marina      -7.1  + 0.5 (4)   0.2 + 0.6 (4)




under higher N loading conditions (B. Peterson pers.        assumption that 20% of net aerial production (total
comm.).                              organic carbon) by S. alterniflora typically makes it
 Although the stable isotope values of phytoplankton       into the open waters of salt-marsh estuaiies (Taylor &
in Waquoit Bay were determined from populations          Allanson 1995). In fact, however, the hydrology and
grown up under artificial conditions (see 'Methods'),       geomorphology of the Waquoit Bay system are such
the values are typical of those measured for marine        that inputs from S. alterniflora may even be smaller.
phytoplankton in many locations around the world          The low tidal amplitude in Waquoit Bay is conducive to
(Gearing 1988). In fact, our estimate of phytoplankton       salt-marsh banks that drop off vertically from the
613C in Waquoit Bay (-21.1%0)is almost identical to the      marsh surface into the water. As a consequence, creek
                          *
sversge FI3C va!ue of phytop!ankton (-21.3 !.l":.-)ir!       ba~ks  with the tal! form of S. a!terr?iflor8 (that get
nearby Narragansett Bay, Rhode Island (Gearing et al.       flooded twice daily) are almost non-existent. The sur-
1984). Rased on this similarity, we conclude that any       face of the marsh is only flooded infrequently, and thus
non-phytoplankton particles in our 'phytoplankton'         remains uncoupled from the estuary much of the time.
samples did not measurably influence the isotope val-       This situation is consistent with the finding of Deegan
ues of these samples.                       & Garritt (1997) that salt-marsh production does not
                                  contribute significantly to subtidal food webs in estuar-
                                  ies with low tidal amplitudes.
         Composition of POM

 The 613C values of POM fall between those of phyto-



                                         -
plankton and macroalgae at all of the Waquoit Bay
estuaries (Fig. 2), indicating that these 2 producer         Phyto.    Macroalgae
                                       .......----.-...---.,
types are the dominant contributors to POM through-
out the system. It is apparent from the higher 6'" val-            :  SL
ues of POM in Sage Lot Pond compared to the other 2
estuaries (Fig. 2), however, that eelgrass makes an
important contribution to POM in Sage Lot Pond in
addition to phytoplankton and macroalgae.
  Salt-marsh cordgrass Spartina alterniflora is more
abundant at Sage Lot Pond that at the other 2 estuar-
ies. It is therefore possible that inputs from S. alterni-
flora (6I3c -12.5%) contribute to the higher 6'" val-
       =
ues of POM In Sage Lot Pond (Fig. 2 ) . We estimate the
influence of S. alterniflora to be small, however,         Fig. 2. 6I3C values of suspended (0)and benthic (m) POM in
                                  Sage Lot Pond (SL),Quashnet River (QR),and Chllds River
because the proportion of primary production in Sage        (CR) compared to those of phytoplankton (Phyto.), macroal-
Lot Pond coming from S. alterniflora is only about 10%       gae, and eelgrass in Waquoit Bay. Dashed lines indicate the
(WBLMER unpubl. data). This estimate is based on the            ranges of SI3C values for each producer type
264                       Mar Ecol Prog Ser 168: 259-271, 1998




Table 3 Percentage of POM and pnmary c o n w m e r d ~ e t s          eelgrass 6I3C and the 8°C of POM at
                               consisting of eel-
grass Zostera m a n n a w ~ t h l nthe Sage Lot Pond estuar) (average 2 1 SD) chllds River represents a 100% loss of eel-
Estimates a r e calculated from the difference In 6°C of consumers (and
                                        grass Thus, the slope term In the above
POM) llving In C h ~ l d s
             Rtver and Sage Lot Pond S p spring, Su summer,
                 .
                 F ~AII                    equatlon is 100 divided by the difference
                                        between eelgrass &':'C and the 6I3C of POM

I Suspended POM
                 Seasons In average Percent eelyrass

                  F., Su., S u , S p
                                        at Childs River There is no y-intercept
                                        term because the 6I3C values of POM at
                                        Sage Lot Pond and Childs River are
 Benth~c 0 k 1
      P            S u . Su                  expected to be the same if POM at Sage Lot
 Suspens~on  feeders                             Pond contains no eelgrass.
  Geukens~a em~ssa
        d          SP.
  ,klya arenarla          Su. Su                    Our estimates of the percentage of eel-
 Depos~t/detntus  feeders                          grass ln POM (Table 3) are probably con-
  Sclerodactyla bnareus      F , SLI., , Sp.
                       SLI               servative, because the larger contributlon
  Palaemonetes vulgans       Su.                    to production by C l a d o p h o r a v a g a b u n d a
  Le~toscoloplos  fraglhs     Su.
                                        in Childs River compared to Sage Lot
  Leptosynapta s p         Su
 Herbivores                                  Pond (Table 1) counters the decrease in
  Cypnnodon vanegatus       F . Su., Su                6I3C of POM caused by a loss of eelgrass.
  Cvmadusa compta         Su., Su.                  Nonetheless, o u r ~ s t i m i r t ~ s Sage 1.0t
                                                          for
  Gdlnfnarus rnucrona tub     Su.                    Pond are quite similar to estimates by
  Er~chsonella  fLOform~s     Su., Sp.
  fvl~crodeutopus  grj/llotalpa  S u . , SLI.                Thayer et al. (1978) for benthic (29%) and
                                        suspended (18%) POM in the Newport
                                        River Estuarv, North Carolina (also based
                                        on 6I3C).
 The consistently lower 6'"     values of suspended     Contributions by eelgrass (8'" = --7%0) to POM in
POM compared to benthlc POM at all 3 estuaries         Sage Lot Pond appear to change seasonally (Fig. 3, top
(Fig. 2) indicate that suspended POM contains a larger     panel). The &':'C of suspended POM is highest d u n n g
proportion of phytoplankton than does benthic POM.       summer months, when eelgrass biomass peaks within
This is not surprising, given that phytoplankton pro-     the estuary, and lowest in the fall and spring, when
duction takes place in the water column. What is more     eelgrass biomass is relatively low. In the other 2 estu-
interesting, however, is that suspended POM at all of     aries, where eelgrass is virtually absent, 613C values of
the estuaries contains organic matter from benthic pro-    POM change little among seasons (Table 2 ) . Using
ducers. Benthic macrophytes may be contributing to       E q . ( l ) ,we find that contributions of eelgrass to sus-
suspended POM (1) as sloughed particles, (2) as resus-     pended POM are undetectable during fall and spring,
pended detritus from the sediments or (3) as organic      but that in summer (when eelgrass biomass peaks in
aggregates formed by bacteria in association with       Sage Lot Pond), suspended POM consists of approxi-
macrophyte-derived dissolved organic matter (Alber &      mately 34 % eelgrass (Table 4 ) . The timing of the rela-
Valiela 1994). In any case, the 6°C data for POM 1.n the    tionship between eelgrass biomass and suspended
Waquoit Bay estuanes highlights the importance of       POM in Sage Lot Pond suggests that release of eel-
benthic-pelaglc coupling in shallow estuaries.         grass-derived organic matter into the water column is
 Based on the decrease in 613C between Sage Lot        closely coupled to active growth of eelgrass. Formation
Pond and Childs River (Fig. 21, w e estimate that (as an    of organic aggregates from DOM released by eelgrass
annual average) benthic POM is 23% eelgrass and        (Alber & Valiela 1994) is probably a n important mech-
suspended POM IS 17% eelgrass in Sage Lot Pond         anism behind this coupling.
(Table 3 ) . These percentages were calculated from the      As with 6I3C, differences in the 6"N values of POM
equation                            among estuanes are consistent with differences in eel-
                                grass contributions; the 615N values of POM are lower
% eelgrass =
                                at Sage Lot Pond than at the other 2 estuaries (Table 2).
                                Differences in the 6''N values of pnmary producers
                                among estuaries (as a result of dlfferences in waste-
where CR a n d SL represent Childs River and Sage Lot     water inputs), however, make it difficult to quantify the
Pond respectively, a n d POM is either benthic or sus-     relationship between eelgrass contributions and POM
pended particulate organic matter This equation        6I5N At most, we deduce from the different responses
expresses a simple linear relationship between % eel-     of 6'"C and 615N In POM to seasonal contributions of
grass and the decrease in 6I3C of POM between Sage       eelgrass (Fig. 3 ) that nitrogen from eelgrass makes a
Lot Pond a n d Childs River. The difference between      smaller contribution to POM than does carbon from
                           McClelland & Valiela. Changes in food web structure                                    265




                           +     15 Jul 94
                                                                                 SAGE LOT POND




                                                                                QUASHNET RIVER

    15 Apr 96

     "  ,   1  '    1    '  1  '   1  '  1




                                                                                    Z m

                                             I    "    "  "            "     "   '  "




                                           6


                                           4  I  Phyt.
                                                   -.__



                                                 '-'OM
                                                       -----
                                                     -------

                                                       B.POM
                                                         0
                                                          -  -  S  -
                                                                  4'
                                                                  _  _  _   ..
                                                                        '".C

                                                                         C v.
                                                                                 CHILDS RIVER
                                                                                     m
                                                                                     Z m




          Eelgrass b~ornass dry wt. m -2)
                  (g

Fig. 3. Biomass of eelgrass Zostera marina versus 613C (top)          Fig 4 6I3C versus &'"h of POM ( 0 )compared to 6IJC versus
and 615N (bottom) of suspended POM at various times of year           FI5N of the dominant pnmary producers ( 0 )at Sage Lot Pond,
in Sage Lot Pond Estimates of eelgrass biomass are from             Quashnet Rlver, and Childs Rlver Phyt = phytoplankton, G t
monthly samples (n = 10) collected with an Ekman dredge.            = Gracjlar~a trkvahlae, C v = Cladophora vagabunda, Z m =
Eelgrass blades were removed from rhizomes, cleaned of epi-           Zostera manna, S-POM = suspended POM, B-POM = benthlc
phytic material, rinsed wlth fresh water, dried at 60°C, and          POM Dashed llnes connect the donlinant producers w i t h ~ n
weighed. Averages i 1 SD were then calculated for each             each estuary (Table l ) Areas enclosed by the dashed lines
              month                        lndicate where points for POM are expected to fall when
                                           POM conslsts of a mix of the dominant producers



eelgrass. This is reasonable, g i w n the higher C:N of     Hanson 1984) may also uncouple the 615N signal from
eelgrass compared to phytoplankton and macroalgae        the 613C signal of eelgrass. Zieman et al. (1984),how-
(Duarte 1992). immobilization of inorganic nitrogen by      ever, found that the 615N values of 2 seagrass species in
heterotrophic bacteria associated with detritus (Rice &     Florida did not change during early decomposition.
                                       In dual isotope plots (613C VS 615N), points
                                      for POM consistentlv fall below the area
Table 4 Seasonal comparison of the percentage of eelgrass Zostera     where they are pred;cted to fall based on a
manna contributing to suspended POM and to the diets of sea cucum-
b e ~ sSclerodactyla brjareus, sheeps head minnow Cypnnodon van-          the dominant producers In each estu-
egatus, and isopods Er~chsonellahllformls at Sage Lot Pond Averages i   ary (Fig 4 ) This suggests that the 615N of
1 SE are reported where samples were collected durlng more than      producers decrease by a small but consistent
                one season                 amount during decomposition. Currin et al.
                                      (1995) found that the 6I5N of standing dead
 Season         Percentage eelgrass contributing to:       cordgrass decreased by about 2.4%0 over a
        Susp POM    S briareus C. vanegatus E. filiformis    10 mo period. The extent to which stable iso-
          -     -    -

                                      tope values of algae change during decom-
                       30 + 8
Summer       34  +7         25 r 6       39 + 5             position have received less study. Based on
                                                our data from Childs River (where over 80%
266                     Mar Ecol Prog Ser 168: 259-271, 1998




Table 5.Consumer isotope values (%D) o composite samples collected in spring, summer, and fall between 1993 and 1996 from
                   f
Sage Lot Pond, and Childs River (average * 1 SE). Values in parentheses are number o seasons sampled. B: b i v d l v ~ ;P: poly-
                                            f
                    chaete; C: crustacean; H: holothurian; F: fish

                            Sage Lot Pond                   Childs River
                        Si3C          6I5N           6I3c          6I5N

 Primary consumers
 Suspension feeders
  Geukensia demissa, B          -19.0 ( l )       7.3 (1)         -18.5 ( l )        9.1 (1)
  Mya arenaria, B               *
                    -16.0 0 5 (2)      5.3 k 0.4 (2)       -17.1 t 0 2 (2)     7 . 9 + 0.0 (2)
 DeposiUdetritus feeders
  Cirratulus grandis, P        -14.7 + 0.2 (2)     5.1 r 0.3 (2)          -
  Leitoscoloplos fragilis, P      -13.5 + 0.3 (2)     4.5 i 0.6 (2)        -14.4 ( l ]       7.5 (1)
  Palaemonetes vulgaris, C        -12.6 ( l )       5.8 (1)         -13.6 ( l ]       8.8 (1)
  Leptosynapta sp., H           -13.4 ( l )       6.1 (1)         -14.7 ( l )       9.4 (1)
  Sclerodactyla briareus, H      -13.4 + 0.2 (4)     6.7 r 0.1 (4)           *
                                              -16.0 0 2 (4)     10.3 + 0.4 ( 4 )
 Herbivores
  Cyprinodon variegatus, F       - 11.5 T 0.9 (3)        *
                                 5.2 0.5 (3)        -14.0  + 0.5 (3)    9.8 + 0.3 (3)
  Cymadusa cornpta, C         -14.4 + 0.3 (2)     4.1 2 0.2 (2)       -16.0  + 0.4 (2)    7.6 + 0.2 (2)
  Gammarus mucronatus. C         -14.2 ( l )        5.1 (1)        -14.8  + 0.4 (2)    7.8 * 0.3 (2)
  Erichsonella filiformis, C      -13.6 + 1.8 (2)     4 . 3 r 1.0(2)      -14.5  + 0.6 (2)    7.7 + 0.8 (2)
  Microdeutopus gryllotalpa. C     -13.4 + 0.1 (2)     3.8 r 0.1 (2)       -14.8  * 0.4 (2)    6.6 + 0.1 (2)
 Secondary consumers
  Menidia menjdia, F          -16.5 * 0.5 (4)        *
                                 9.8 0.4 (4)        -17.7 + 0 8 (4)    1 1 . 6 + 0 . 3(4)
  Gasterosteus aculeatus. F        -16.9 (1)        9.0 (1)        -15.4 * 0.4 (3)    12.3 * 0.4 (31
  Fundulus heteroclitus, F        -14 * 0.4 (3)        *
                                 8.3 0.2 (3)        -14.6 + 0.5 (3)    11.3 + 0.3 (3)
  Pseudopleuronectesamericanus, f.                  -           -13.1 (1)       10.7 (1)
  Podarke obscura, P          -12.7  + 0.2 (4)    7.1 r 0.1 (4)           *
                                              -14.9 0.2 (4)     10.7 + 0.3 (4)




of production comes from phytoplankton and Clado-            tracted from 6I3C, and 3.0%0 was subtracted from 6I5N.
phora vagabunda; Table l),it appears that decomposi-          These values reflect an average fractionation effect of
tion has decreased the F1'N of algae by about 1%0. This         a trophic step based on a broad range of consumer
decrease in 815N may be due to uptake of inorganic           types (Peterson & Fry 1987, Michener & Schell 1994).
nitrogen by bacteria associated with detritus (Rice            Primary consumers. The positions of primary con-
1982). Rice & Hanson (1984) demonstrated that bacte-          sumers compared to those of producers in 613C versus
rially mediated N accumulation in cordgrass detritus          6I5N plots for Sage Lot Pond (Fig. 5) and Childs River
was greater than N accumulation in macroalgal              (Fig. 6) indicate that the dominant producers found in
(Gracilaria foliifera) detritus. This would explain the         each estuary are important food sources. Primary con-
larger difference in 615N between llve and decom-            sumers in Sage Lot Pond (Fig. 5) eat phytoplankton,
posed cordgrass observed by Currin et al. (1995) than         macroalgae, and eelgrass, whereas primary consumers
between live and decomposed algae observed in this           in Childs River (Fig. 6) subsist on phytoplankton and
study (Fig. 4).                            macroalgae alone.
                                     The presence of eelgrass in the diets of primary con-
                                    sumers in Sage Lot Pond is confirmed by shifts in 6I3c
       Food sources of consumers                values of primary consumers between Sage Lot Pond
                                    and Childs River (Fig. 7). Shifts in the 615N values of
 Before comparing the stable isotope values of con-          consumers between Sage Lot Pond and Childs River
sumers with those of potential food sources within the         (Table 5) also reflect differences in eelgrass abundance
Waquoit Bay estuaries, we corrected for fractionation         between estuaries, but these shifts are confounded by
tha.t occurs as organic matter passes from one trophic         shifts in producer 6I5N (Table 2) as a result of differ-
level to the next. This was done by subtracting litera-        ences in N loading from wastewater (McClelland et al.
ture values for trophic fractionation from the measured        1997, McClelland & Valiela in press). We therefore
isotope values of consumers presented in Table 5.           focus on shifts in 6I3C between Sage Lot Pond and
1.0% was subtracted from the 613C and 615N values of          Childs River to estimate the proportions of primary
amphipods (and one isopod) to account for the troph~c         consumer diets derived from eelgrass in Sage Lot
fractionatlon measu.red by Macko et al. (1982)for these        Pond. We use Eq. ( l ) , but substitute 'consumers' for
consumers. For all other consumers, 1.0% was sub-           'POM' and '(6I3Ce1, + 1)'for '613CEe,, '. We add 1 to
                                              g          ,,
                                                        ,
                    McClelland & Valiela: Changes in food web structure                   267




                                       CHILDS RIVER
                      SAGE LOT POND                           I




                                            t  S-POM



                                                    +
                                                   S2.
                                                      B-POM
                                                           C v.




Fig. 5. 6'" versus 6'" of primary consumers (with adjust-      Fig. 6.SI3C versus 615N of primary consumers (with adjust-
ment for trophic fractionation, see below) compared to 6I3C     ment for trophic fractionation, see below) compared to 6l3C
versus SI5N of POM and the dominant producers in Sage Lot      versus ST5Nof POM and the dominant producers in Childs
?@.'d. Suspension feeders. S1 = Getlkprlia T I ~ m i c ~S2, = Mya
                            a      R~ver. Suspension feeders: S1 = Geukensia demissa. S2 = Mya
arenaria. Herbivores: H1 = Cypnnodon variegatus, H2 =        arenaria. Herbivores: H1 = 'Cyprinodon variegatus, H2 =
Cymadusa compta, H3 = Gammarus mucronatus, H4 = Erich-       Cymadusa compta, H3 = Gammarus mucronatus, H4 = Erlch-
sonella filiformis, H 5 = Microdeutopus gryllotalpa. Deposit/    sonella filiformis, H5 = Microdeutopus gryllotalpa. Deposit/
detritus feeder: D1 = Cirratulus grandis, D2 = Leitoscoloplos    detntus feeder: D2 = Leitoscoloplos fragilis, D 3 = Palae~non-
fragilis, D3 = Palaemonetes vulgaris. D4 = Leptosynapta sp.,    etes vulgaris, D4 = Leptosynapta sp., D5 = Sclerodactyla
D5 = Sclerodactyla briareus. Phyt. = phytoplankton, G. t. =     briareus. Phyt. = phytoplankton, C. t. = Gracilaria tikvahiae,
Gracilaria tikvahiae, C. v. = Cladophora vagabunda, Z. m. =     C. v = Cladophora vagabunda, 2.m. = Zostera marina,
Zostera mal-jna, S-POM = suspended POM, B-POM = benthic       S-POM = suspended POM, B-POM = benthic POM. Dashed
POM. Dashed llnes encompass the areas where polnts for pri-     lines encompass the areas where points for primary con-
mary consumers are expected to fall when their diets consist    sumers are expected to fall when thelr dlets consist of a mix of
of a mix of the dominant producers in Sage Lot Pond. 1 has     the dominant producers in Childs River. To correct for trophic
been subtracted from the stable C and N isotope values of      fractionation, 1% was subtracted from the stable C and N iso-
small crustacean grazers to correct for trophlc fractionation.   tope values of small crustacean grazers. l'?;;* 3%. were
                                                          and
1%" and 3%0  have been subtracted from the S1'C and 6"N val-    subtracted from the Fl3C and 6l5N values respectively of all
       ues respectively of all other consumers                  other primary consumers




the 6I3C value of eelgrass to account for trophic frac-       mussel Geukensia demissa has no detectable eelgrass
tionation.                             in its diet (Table 3).
  Sage Lot Pond: Using E q . ( l ) , estimate that eel-
                  we                The difference in diet between the ribbed mussel
grass makes a relatively small contribution to the diets      and the soft-shell clam in Sage Lot Pond (Table 3 ) ,
of most consumers in Sage Lot Pond (typically less than       even though they are both suspension feeders, is likely
16 %), and yet its presence in the food web of Sage Lot       a result of the different locations within the estuary
Pond is pervasive (Table 3). Furthermore, it is apparent      where each species lives. Ribbed mussels in Sage Lot
that eelgrass is more important in the diets of some        Pond live embedded in the salt-marsh bank at the
consumers than others; the sea cucumber Sclero-           estuary's edge, just below the marsh surface. In this
dactyla briareus and the sheepshead minnow Cypnn-          location, they can only feed near high tide, and thus
odon variegatus stand out as having particularly large       have diets dominated by phytoplankton (Fig. 5). On
proportions of eelgrass in their diets, while the ribbed      the other hand, soft shell clams in Sage Lot Pond live
268                       Mar Ecol Prog Ser 1




   Phyto.   Macroalgae
        .....~.~.............
                           Eelgrass
                            .....
                                     dition, S bnareus inay consume eelgrass detntus that
                             ,  .
                               ,      spans several (even many) seasons from day to day
m     :          a :     0      D5  I j
S A ;
                  :     +   H
                                ,
                                j ;
                                  .    Our findlng that eelgrass makes an important contri-
m
-
    m

3    :
               + : U          H2
                                ,  *
                                  j
                                     bution to the food web in Sage Lot Pond is conslstent
m   I

     :                               wlth other stable isotope studies of eelgrass communi-
m
C
m
      :
      :
                *      0      H5   i :
                                ,  .   ties In North Carolina (Thayer et a1 1978) and Alaska
                 :. 0            :
Z
- ;                          D4
                                ,  .
                                     (McConnaughey & McRoy 1979), suggesting that use
2     i        a+;             S2   :,  .
                  :   . O     D3        of eelgrass by consumers IS common Stephenson et a1
-   a
                                  ;
m                              . .
                               .
                                     (1986),in contrast, found that eelgrass does not make a
     :
                 -
S                                 .
                  : . G        D2  :   ;
m     ,                            ,
     1              c
                    .
                            H4
                                ,
                                j  j   s~gnif~cant contribution to consumers In an eelgrass
     :
                    V
V)
mJ                              ,  ,
m    .            - 0         H3   j  j   commun~ty Nova S c o t ~ aCanada Where eelgrass is
                                            in         ,
t
c
     i
     :    : 0.                 S1
                               .j ,j    a n important food source i t IS probably consumed
                               ,  ,
                               ,
                               ,
                               ,
                                  ,
                                  ,
                                  ,
                                     more as detntus than as live m a t e r ~ a l(Thayer et a1
                                     1975, Zieman 1975) Live eelgrass IS less nutritious
  -22    -20   -18  -16   -14   -12  -10  -8    -6
                                     than algae, but microbially mediated processes during
                                     decomposit~on make it (with assoclated bacteria) a
                                     more favorable food source with time (Tenore et a1
Fig 7 6I3C values of pnmarv consumers ilvlthout a d ~ u s t m ~ n t 1982)
for t r o p h ~ cfractionatlon) at Sage Lot Pond (0)n d Childs
                             a         Childs River: In C h ~ l d s River, where eelgrass IS vlr-
Rtver (a) compared to those of phytoplankton (Phyto )          tually absent, the stable isotope data show that herbi-
macroalgae, and eelgrass in Waquoit Bay Dashed hnes indi-
cate the ranges of 6I3C values for each producer type Sus-
                                    vores and detntus/deposit feeders eat a mix of Gracl-
penslon feeders S1 = Geukensla dernissa S2 = M v a arenarla       lana hkvahlae and Cladophora vaqabunda (Fiq 6 )
Herbivores H 1 = Cypnnodon vanegatus H2 = Cymadusa           Four herbivores (Cymadusa compta, Gammarus m u -
compta, H 3 = Garnmarus rnucronatus, H 4 = Er~chsonellaf111-      cronatus, Enchsonella filiform~sand Cyprlnodon var-
forms, H5 = ~V~crodeutopus       gryllotalpa Deposit/detntus  iegatus) fall Intermediate between G t ~ k v a h ~ a n d Cae
feeder D2 = Le~toscoloplosf r a g ~ l ~D3 = Palaemonetes vul-
                        s,
gans, D4 = Leptosynapta sp , D5 - Sclerodactyla bnareus Prl-      vagabunda In Fig 6, indicating that they eat similar
m a r y consumers a r e ordered from the bottom to the top of the    quantities of the 2 macroalgal species The herb~vore
flgure according to increasing d ~ f f e r e n c ein S13C between    M~crodeufopus    gryllotalpa falls close to C vagabunda
                 estuar~es               alone, indicating that it special~zes C vagabunda
                                                         on
                                    The w ~ d e scatter of points for deposlt/detrltus feeders
                                           r
                                    compared to herbivores in Fig 6 suggests that the diets
largely subtidally on the bottom of the estuary and           of depos~t/detr~tus   feeders vary more among species
their d ~ e t s  have a large benthic component (Fig 5),       than do the diets of herbivores It is more likely, how-
Including a contribution from eelgrass (Table 3) Peter-         ever, that v a n a t ~ o n sin C and N stable ~ s o t o p e
                                                                 values
son et a1 (1985) similarly suggest that the different          among deposit/detrltus feeders reflect taxon-speciflc
subhabitats of rlbbed mussels compared to quahogs            differences in fractionation of C and N during assimila-
~Clercenana     rnercenana are responsible for the larger      tion of organic matter (DeNlro & Epstein 1978, 1981
proport~on of phytoplankton found in the diets of            Mlnagawa & Wada 1984),since the species included In
ribbed mussels compared to hard-shell clams wlthin           the deposit/detrltus feeder group belong to 3 different
Great Slppewissett Marsh Cape Cod Massachusetts             phyla
  Where we had s u f f ~ c ~ e n t   data we also looked for      The positions of the rlbbed mussel Geukensla
seasonal differences in the percentage of eelgrass In          demjssa and the soft-shell clam Mya arenaria in Fig 6
the d ~ e t sof prlmary consumers As with suspended           indicate that these suspension feeders subsist on a mlx
POM the sheepshead m n n o w Cypnnodon vanegatus            of phytoplankton and macroalgae Cladophora vaga-
and the isopod Er~chsonella      fihformis are Influenced bv   bunda through the consumpt~on of POhZ in Childs
eelgrass more In summer than wlnter months (Table 4)          River As observed for Sage Lot Pond, the nbbed mus-
This seasonality suggests that these herbivores feed at         sel gets a larger proport~on ~ t d ~ efrom phytoplank-
                                                     of  s t
least partially on live eelgrass The sea cucumber Scle-         ton than does the soft-shell clam
rodactyla bnareus, on the other hand, shows no sea-            Secondary consumers. The pos~tions of mummi-
sonal shift in the percentage of eelgrass contnbuting to        chogs Fundulus heterochtus, winter flounder Pseudo-
~ t s ~et
   d    (Table 4 ) This is conslstent with the detntus       pleulonectes amencanus, and the polychaete Podarke
feeding habit of S bnareus Eelgrass detntus becomes           obscura In Fig 8 imply that these organisms feed on
more edible as ~t decays (Tenore et a1 1982), a n d thus        benthlc fauna In contrast, st~cklebacksGasterosteus
seasonal inputs of live eelgrass are temporally sepa-          aculeatus and silversldes Menldia rnenld~aappear to
rated from the detntus pool used by S bnareus In ad-          consume both b e n t h ~ c  fauna and zooplankton We do
                             McClelland & Valiela: Chalnges in food w e b structure                         269




not assign specific organisms to the diets of each sec-                  Shifts in stable isotope values of secondary con-
ondary consumer because, although the primary con-                   sumers from Sage Lot Pond to Childs River (Fig. 8) are
sumers we chose for this study are diverse, they rep-                  difficult to clearly attribute to a loss of eelgrass, since
resent only a fraction of the species found in Waquoit                 secondary consumers are 2 trophic steps away from
Bay. Other isotopic investigations of marine food webs                 eelgrass. Because most of the primary consumers in
also identify organisms that use planktonic and ben-                  Sage Lot Pond get a portion of their diet from eelgrass
thic food sources within the same community (Fry &                   (Table 3), however, it is reasonable to assume that eel-
Parker 1979, Fry et al. 1983, Simenstad & Wissmar                    grass carbon and nitrogen are making it into sec-
1985, Peterson et al. 1986, Thomas & Cahoon 1993,                    ondary consumers at this estuary.
France 1995, Deegan & Garritt 1997). Deegan & Gar-
ritt (1997) in particular, however, emphasize the mix-
ing of benthic and planktonic food sources in the diets                               Conclusions
of consumers in estuarine waters. Our stable isotope
data on secondary consumers (and also suspension                      Isotopic examination of subtidal food webs in differ-
feeders) in the Waquoit Bay estuaries support this                   ent estuaries of Waquoit Bay shows that diets of pri-
perspective.                                      mary consumers are influenced by the dominant forms
                                            of production that they are exposed to. Phytoplankton
                                 SAGE LOT POND     and macroalgae are the major food sources of primary
     I
                                            consumers under both low and high N loading condi-
                                            tions, but eelgrass is also an important food source
                                            where N loading is low. Seasonal changes in the
                                            percentages of eelgrass in the diets of some primary
                                            consumers suggest that they feed on live eelgrass or
                                            organic aggregates rapidly formed from eelgrass
                                            exudates. Other primary consumers have relatively
                                            constant contributions by eelgrass to their diets
                                            throughout the year, suggesting that they feed on eel-
--.t                                 l         g r a s s d e t r i t u s s t o r e d i n the sedimnntz. F ~ l r ~ r a Cs and
                                                                                s
&  3    ~  '  "  "  "  1  '  "    1  '  "  "  '           N are passed on to benthic as well as pelagic sec-
z
-
L0
                                     CHILDS RIVER  ondary consumers.
                                              With losses of eelgrass as a result of nitrogen load-
O  0  I                  I
                                            ing, an important pathway through which land-
                                            derived nitrogen enters food webs in the Waquoit Bay
                                            system is eliminated. The ecological implications of
                                            this fundamental change have not been defined, but
                                            the rate at which land-derived nitrogen is cycled
                                            within estuaries, as well as its ultimate fate, must be
                                            influenced by the path of nitrogen flow through estuar-
                                            ine food webs. When eelgrass is replaced by algae, it is
                                            llkely that detrital nitrogen becomes available to estu-
                                            arine consumers at a faster rate. This is because algal
                                            detritus starts out as a nutritious food source to con-
                                            sumers, whereas eelgrass detritus must be microbially
                                            altered before it can be used by many consumers
Flg 8 613C versus 6"N of secondary consumers (with adjust-               (Tenore et al. 1982). As a consequence of faster nutri-
ment for trophic fractionation, s e e below) compared to 6I3C              ent cycling. through the detrital pathway, detritivores
versus 6% of benthic fauna (B) and zooplankton (Z) in Sage               may become more closely coupled to annual produc-
Lot Pond (lop panel) and Childs River (bottom panel). The sta-             tion cycles. Levinton (1972)hypothesized that detritus-
ble isotope signature of the ribbed muscle is used as a proxy
for that of zooplankton, because ~ t s  diet is d o m ~ n a t e d by          based benthic communities are more stable than com-
phytoplankton. Common names for secondary consumers                   munities that feed on plankton directly because the
(open triangles) are: silversides. Menidia menidia; stickle-              food supply to detritus-based communities is relatively
back, Gasterosteus aculeatus; mummichog, Fundulus hetero-                constant. Shifts from eelgrass- to algae-dominated pro-
clitus; winter flounder, Pseudopleuronectes americanus; car-
nivorous polychaete, Podarke obscura. To correct for trophic
                                            duction brought about by anthropogenic nitrogen
fractionation, 1 ?Land 3 % o were subtracted from the 6'" and              loading may similarly result in less stable communities
   S 5 N values respectively of each secondary consumer                in shallow estuarine waters.
 270                       Mar Ecol Prog Ser 168. 259-271, 1998




Acknowledgements. This work is a result of research spon-         Freshwat Res 34:707-715
sored by NOAA National Sea Grant College Program Office,        Fry B. Sherr EB (1984) SI3C measurements as indicators of
Department of Commerce, under Grant No. NA46RG0470,            carbon flow in marine and freshwater ecosystems Contrib
Woods Hole Oceanographic Institution Sea Grant Project No.         Mar Sci 27 13-47
R/M-38, and a grant from the NOAA Coastal Oceans Pro-         Galloway JN, Schlesinger WH, Levy H 11, Michaels A,
gram. The views expressed herein a r e those of the authors        Schnoor JL (1995) Nitrogen fixation: anthropogenic en-
and do not necessarily reflect the views of NOAA or any of its       hancement-environmental response. Global Biogeochem
subagencies. We thank B. Peterson and D. Rhoads for their         Cycles 9:235-252
comments on earlier versions of this manuscript. We also        Gannes LZ,O'Brien DM, Martinez del Rio C (1997) Stable iso-
thank 3 anonymous reviewers for their kind comments and          topes in anlmal ecology: assumptions, caveats, and a call
helpful advice. Stable isotope analysis was done by R. H.         for more laboratory experiments. Ecology 78.1272-1276
Michener at the Boston University Stable Isotope Laboratory.      Gearing J N (1988) The use of stable isotope ratios for tracing
                                      the nearshore-offshore exchange of organic matter. In:
                                      Jannson B 0 (ed) Lecture notes on coastal and estuarine
           LITERATURE CITED                   studies, coastal offshore ecosystem interactions. Springer-
                                      Verlag, Berlln, p 69-101
Alber M, Valiela 1 (1994) Production of microbial aggregates     Gearing J N , Gearing PJ, Rudnick DT, Requejo AG, Hutchins
   from macrophyte-derived dissolved organlc material. Lim-       MJ (1984) Isotopic variability of organic carbon In a phyto-
   no1 Oceanogr 39:37-50                         plankton-based temperate estuary. Geochim Cosmochim
Benner R, Fogel ML, Sprague EK, Hodson RE (1987) Deple-          Acta 48:1089-1098
   tion of "C in lignin and its implications for stable carbon    GESAMP (1990) The state of the marine environment. Joint
   isotopc s:udics. Nature 329:708 -7lC                 G r c u ~ Expc:ts or, the Scientific Aspects of Marino Pol
                                          of
Canuel EA. Cloern JE, Ringelberg DB, Guckert JB, Rau GH          lution. Rep and Stud 39 United Nations Environmental
   (1995) Molecular and so topic tracers used to examinr.        Programme, Nairobi
  sources of organic matter and its incorporation into the      Haines EB (1977) The origins of detritus in Georgia salt marsh
  food web of San Francisco Bay. Lmnol Oceanogr 40:67-81         estuaries. Oikos 29:254-260
Cehrian _TV Ezarte CM (!%?/I.) The d e p e ~ d e n c ef herhivcrg nn
                           n         E!eckscher FIT, Flznxwe!! _T, Jimenoz EG, !?ie!cma C, Va!ie!a !
  growth rate in natural plant communities. Funct Ecol 8:        (1996) Selectivity by the herbivorous amphipod Microdeu-
  518-525                                topus gryllotalpa among five species of macroalgae. Biol
Costa J E , Howes BL, Giblin AE, Valiela l (1992) Monitoring        Bull (Woods Hole) 191:324-326
  nitrogen and indcators of nitrogen to support manage-       Kelly JR, Levin SA (1986) A comparison of aquatic and terres-
  ment action in Buzzards Bay. In: McKenzie DH, Hylact          trial nutrient cycling and production processes in natural
  DE, McDonald VJ (eds) Ecological indicators, Vol 1. Else-       ecosystems, with reference to ecological consequences to
  vier Press, London, p 499-530                     some waste disposal issues. In: Kullenberg G (ed)The role
Currin CA, Newel1 SY, Paerl HW (1995) The role of standing        of oceans as a waste disposal option. NATO Advanced
  dead Spartina alterniflora and benthic microalgae in salt       Research Workshop Series. D Reidel Pub1 CO, Dordrecht,
  marsh food webs: cons~derations  ba.sed on multiple stable     p 165-203
  isotope analysis. Mar Ecol Prog Ser l.21:99-116          Lajtha K, Mlchener RH (1994) Stable isotopes in ecology.
D'Avanzo C, Krerner JN. Wainright SC (1996) Ecosystem pro-        Blackwell Scientific Publications, Oxford
  duction and respiration in response! to eutrophication in     Levinton J (1972) Stability and trophic structure in deposit-
  shallow temperate estuaries. Mar Ecol Prog Ser 141:          feeding and suspension-feeding communities. Am Nat
  263-274                                106:472-486
Deegan LA, Garritt RH (1997) Evidence for spatial variability     Macko SA, Yuh Lee W, Parker PL (1982) Nitrogen and carbon
  in estuarine food webs. Mar Ecol Prog Ser 147:31-47          isotope fractionation by two species of manne amphipods:
DeNiro MJ, Epstein S (1978) Influence of d ~ e on the distribu-
                          t           laboratory and field experiments. J Exp Mar Biol Ecol 63:
  tion of carbon isotopes in animals. Geochim Cosmochim         145-149
  Acta 42:495-506                          McClelland JW, Valiela 1 (in press) Linking nitrogen in estu-
DeNiro MJ, Epstein S (1981) Influence of diet on the distribu-      arine producers to land-derived sources. Limnol Oceanogr
  tion of nitrogen isotopes in animals. Geochim Cosmochim      McClelland JUT, Valiela I, Michener RH (1997) Nitrogen sta-
  Acta 45:341-351                            ble isotope signatures in estuarine food webs: a record of
Duarte CM (1992) Nutnent concentration of aquatic plants         increasing urbanization in coastal watersheds. Limnol
  patterns across species. Limnol Oceanogr 37:882-889          Oceanogr 42:930-937
Duarte CM (1995) Submerged aquatic vegetation in relation       McConnaughey T, McRoy CP (1979) I3C label identifies eel-
  to different nutrient regimes. Ophelia 41:87-112            grass (Zostera marina) carbon in an Alaskan estuarine
Fogel ML. Cifuentes LA. Velinsky DJ. Sharp JH (1992) Rela-        food web. mar Biol53:263-269
  tionship of carbon availability in estu~trine   phytoplankton   Michener RH, Schell DM (1994) Stable isotope ratios as trac-
  to isotopic composition. Mar Ecol Prog Ser 82:291-300         ers in marine aquatic food webs. In: Lajtha K, Michen.er
France RL (1995)Carbon-13 enrichment I bcnthic compared
                       n              RH (eds)Stable isotopes in ecotogy and env~ronmental   sci-
  to planktonic algae: foodweb implications. Mar Ecol Prog        ence. Blackwell Scientific Publications. Oxford. p 138-157
  Ser 124:307-312                           Minagawa M. Wada E (1984) Stepwise enrichment of I5N
Fry B (1991) Stable isotope diagrams of freshwater food webs.       along food chains: further evidence and the relation
  Ecology 72:2293-2297                          between S15N and animal age. Geochim Cosmochlm Acta
Fry B. Parker PL (1979) Animal diet in Texas seagrass mead-        48:1135-1140
  ows. SI3C evidence for the importance of benthic plants       National Research Counc~l(1994) Prionties for coastal sci-
  Estuar Coast Shelf SCI8 499-509                    ence. Nationa.1.Academy Press, Washington, DC
Fry E, Scalan RS, Parker PL (1983) ' % / l Z ~ ratios in manne     Nicolri ME (1980) Factors ~nvolvedin herbivore food prefer-
  food webs of the Torres Strait, Queensland. Aust J Mar         ence. J Exp Mar Biol Ecol42:13-26
                      McClelland & Valiela: Changes in food web structure                    27 1




Nixon SW (1986) Nutrients and the productivity of estuarine         tween a high marsh and estuary, and the inapplicability
  and coastal marine ecosystems. J Limnol Soc S Afr 12:          of the Outwelling Hypothesis. Mar Ecol Prog Ser 120:
  43-71                                  263-270
Orth RJ, Moore KA (1983) Chesapeake Bay: an unprece-           Tenore KR, Cammen L, Findlay SEG, Phillips N (1982) Per-
  dented decllne in submerged aquatic vegetation. Science         spectives of research on detritus: do factors controlling the
  22251-53                                 availability of detritus to macroconsumers depend on its
Peckol P, Rivers JS (1996) Contribution by macroalgal mats to        source? J Mar Res 40:473-490
  primary production of a shallow embayment under high         Thayer GW, Adams SM. LaCroix M V (1975) Structural and
                                                        L
  and low nitl-ogen-loading rates. Estuar Coast Shelf Sci 43:       functional aspects of a recently established Zostera marina
  311-325                                 community. In: Cronin LE (ed) Estuarine research, Vol 1
Peterson BJ, Fry B (1987) Stable isotopes in ecosystem stud-         Academic Press, New York, p 518-540
  ies Annu Rev Ecol Syst 18:293-320                   Thayer GW, Parker PL, LaCroix MW, Fry B (1978)The stable
Peterson BJ, Howarth RW, Garritt RH (1985) Multiple stable          carbon isotope ratio of some components of an eelgrass,
  isotopes used to trace the flow of organic matter in estuar-       Zostera marina, bed. Oecologia 35:l-12
  ine food webs. Sclence 227:1361-1363                 Thornas CJ, Cahoon LE (1993) Stable isotope analyses differ-
Peterson BJ, Howarth RLV, Garntt RH (1986) Sulfur and car-          entlate between different trophic pathways supporting
  bon isotopes as tracers of salt-marsh organic matter flow.        rocky-reef fishes. Mar Ecol Prog Ser 9519-24
  Ecology 67.865-874                          Tomasky G, Valiela I (1995) Nutrient limitation of phyto-
Pickett STA (1991) Space for time substitution as an alterna-        plankton growth in Waquo~t   Bay, Massachusetts. B101 Bull
  tive to long-term studies. In: Likens GE (ed) Long-term         (Woods Hole) 189:257-258
  studies in ecology: approaches and alternatives. Springer-      Valiela I, Collins G , Kremer J . Lajtha K. Geist M, Seely B,
  Verlag, New York, p 110-135                       Brawley J , Sham CH (1997b) Nitrogen loading from
Rice DL (1982) The detritus nitrogen problem: new observa-          coastal watersheds to receiving estuaries: new method
  tions and perspectives from organic geochemistry. Mar           and application. Ecol App17:358-380
  Ecol Prog Ser 9:153-162                        Valiela 1, Foreman K, LaMontagne M, Hersh D , Costa J ,
Rice DL, Hanson RB (1984) A kinetic model for detritus nitro-        Peckol P, DeMeo-Anderson B, D'Avanzo C, Babione M,
  gen: role of the associated bacteria in nitrogen accumula-        Sham C, Brawley J. Lajtha K (1992) Couplings of water-
  tion. Bull Mar Sci 35:326-340                       sheds and coastal waters: sources and consequences of
Sand-Jensen K, Borum J (1991) Interactions among phyto-           nutnent ennchment in Waquoit Bay. Massachusetts. Estu-
  plankton, periphyton, and macrophytes in temperate            aries 15:443-457
  freshwaters and estuaries. Aquat Bot 41:137-175            Valiela I, McClelland J, Hauxwell J , Behr PJ, Hersh D, Fore-
Short FT, Burdick DM (1996) Quantifying eelgrass habitat           man K (1997a) Macroalgal blooms In shallow estuaries:
  loss in relation to housing development and nitrogen load-        Controls and ecophysiological and ecosystem conse-
  i?..; !E Waql~oit Bay. M a s s a c h ~ ~ s e t tEstuaries 19:730-739
                          s             quences. Limnol Oceanogr 42:1105-1118
Simenstad CA, Wissmar RC (1985) 613c evidence of the ori-         Zieman JC (1975) Quantitative and dynamic aspects of the
  gins and fates of organic carbon in estuarine and             ecology of turtle grass. Thalassia testudinum. In: Cronin
  nearshore food webs. Mar Ecol Prog Ser 22:141-152             LE (ed) Estuarine research, Vol 1. Academic Press, New
Stephenson RL, Tan FC, Mann KH (1986) Use of stable carbon          York, p 541-562
  isotope ratios to compare plant material and potential con-      Zieman JC, Macko SA, Mills L (1984) Role of seagrasses and
  sumers in a seagrass bed and a kelp bed in Nova Scotia,          mangroves in estuarine food webs: temporal and spatial
  Canada. Mar Ecol Prog Ser 30.1-7                     changes in stable isotope composition and amino acid con-
Taylor DI, Allanson BR (1995) Organic carbon fluxes be-           tent during decomposition. Bull Mar Sci 33:380-392

Editoi7a1responsibility: Otto Kinne (Editor),               Submitted: December 16, 1997; Accepted: April 3, 1998
Oldendorf/Luhe, Germany                          Proofs received from a uthor(s): June 3, 1998