Rivers and Short 07
MARINE ECOLOGY PROGRESS SERIES
Vol. 333: 271–279, 2007 Published March 12
Mar Ecol Prog Ser
Effect of grazing by Canada geese
Branta canadensis on an intertidal eelgrass
Zostera marina meadow
David O. Rivers, Frederick T. Short*
Department of Natural Resources, University of New Hampshire, Jackson Estuarine Laboratory, 85 Adams Point Road,
Durham, New Hampshire 03824, USA
ABSTRACT: Fishing Island, in Portsmouth Harbor on the Maine–New Hampshire border (USA), is
the site of an intertidal eelgrass (Zostera marina L.) bed that is part of SeagrassNet, an international
program for long-term seagrass monitoring. Eelgrass bed parameters of canopy height, percent
cover, and aboveground biomass have been monitored quarterly since October 2001 using the Sea-
grassNet protocol. A flock of nearly 100 Canada geese Branta canadensis L. over-wintered at Fishing
Island and grazed on eelgrass from January to April 2003, an event that had not been seen at this
meadow in 2 decades of observation. Before Canada geese were present, eelgrass parameters
demonstrated seasonal fluctuations typical of the region. During the grazing event, eelgrass parame-
ters declined drastically, and biomass losses reached 680 g m–2 in parts of the meadow. SeagrassNet
data demonstrated that eelgrass did not recover after the geese departed. Additional fieldwork con-
ducted from February to July 2003 showed that eelgrass recruitment via sexual reproduction at Fish-
ing Island was minimal, and vegetative recovery was impeded by Canada goose consumption of the
plant meristems. Unlike studies in other locations, which show seagrass quickly rebounding from
annual grazing events, eelgrass at Fishing Island showed little recovery from Canada goose grazing
through July 2003.
KEY WORDS: Eelgras · Zostera marina · Grazing · Canada goose · SeagrassNet · Climate change
Resale or republication not permitted without written consent of the publisher
INTRODUCTION goose grazing activity during this time can have a sub-
stantial effect on seagrass abundance (Portig et al.
Seagrass has long been recognized as an important 1984, Baldwin & Lovvorn 1994, Ganter 2000). At Euro-
food resource for migratory waterfowl, whose migra- pean seagrass meadows, waterfowl feeding activity
tion routes often coincide with seagrass meadow loca- was shown to reduce plant biomass by more than 50%
tions (Ganter 2000). In North America, eelgrass during the course of the winter grazing period (Jacobs
Zostera marina L. meadows are considered important et al. 1981, Nacken & Reise 2000).
staging areas for black brant geese Branta nigricans L. The rate of depletion of vegetation by grazing water-
on the west coast (Wilson & Atkinson 1995, Ward et al. fowl is primarily influenced by the number of birds
2003, Moore et al. 2004), and for brant B. bernicla L. present at a site and accessibility of the plants (Baldwin
and Canada geese B. canadensis L. on the east coast & Lovvorn 1994, Percival et al. 1996, Clausen 2000).
(Seymour et al. 2002, Hanson 2004). The length of stay Seagrass is accessible to birds in shallow water sys-
of waterfowl at staging areas is often closely correlated tems or in intertidal areas (Ganter 2000); changes in
to the available seagrass resources at that site (Wilson water levels due to the tidal cycle can limit the amount
& Atkinson 1995). Geese may remain at seagrass of time that food resources are obtainable (Fox 1996,
meadows for several months during the winter, and Clausen 2000). Many waterfowl species feed by up-
*Corresponding author. Email: fred.short@unh.edu © Inter-Research 2007 · www.int-res.com
272 Mar Ecol Prog Ser 333: 271–279, 2007
ending in the water (Buchsbaum 1987, Vogel 1995), geese are the most common over-wintering waterfowl
and the length of their neck-plus-head determines the species (Vogel 1995). Canada geese that over-winter
maximum depth at which they can feed (Clausen in this region rely on a few primary food sources,
2000). In intertidal areas, birds may not be able to including upland agricultural fields, golf course
reach plants at high tide; feeding is often restricted to grasses, coastal and estuarine salt marshes, and eel-
low tide when receding water makes the plants acces- grass meadows (Vogel 1995). Within the Great Bay
sible (Fox 1996, Percival & Evans 1997). Observations Estuary, the most extensive eelgrass meadows occur
of brant that congregate in intertidal areas have shown in Great Bay itself, with smaller meadows throughout
that the birds feed whenever they have access to sea- the rest of the estuary and along the coast. This study
grass, day or night, and rest when plants are inaccessi- was conducted at the Fishing Island eelgrass meadow,
ble due to high water levels (Percival & Evans 1997). located near the mouth of the estuary. The Fishing
Over time, the total seagrass biomass available at a Island meadow is a SeagrassNet site, which is part of
given site declines as the plants are consumed, causing an international long-term seagrass monitoring pro-
birds to spend a greater amount of time feeding in gram (Short et al. 2002). Documented cases of eel-
order to achieve an adequate food intake (Percival et grass decline due to Canada goose grazing are rare
al. 1996, 1998). (Hanson 2004), and studies focusing on other water-
Although grazing activity can significantly reduce fowl species often discuss how seagrass availability
plant biomass, at many meadows the biomass loss is a affects bird populations while ignoring the process of
short-term effect, and the affected meadows recover eelgrass recovery (Ganter 2002, Moore et al. 2004).
during the subsequent growing season (Vermaat & Our study presents comparative data of eelgrass para-
Verhagen 1996, Nacken & Reise 2000, Hughes & Sta- meters before, during, and after a Canada goose graz-
chowicz 2004). Winter waterfowl grazing coincides ing event and examines eelgrass recovery post-
with the biomass losses from natural seasonal declines impact. The objectives of this study were to compare
of temperate seagrasses (Short 1992, Ganter 2000). bed characteristics of Zostera marina between a year
Exclosure studies have demonstrated that patches of when grazing occurred and a year when no grazing
seagrass protected from grazing experience biomass occurred at a long-term monitoring site, and to quan-
declines of up to 65% during the winter months (Tubbs tify the extent of eelgrass vegetative and sexual
& Tubbs 1982, Madsen 1988), indicating a marked sea- reproduction in the growing season after the grazing
sonality in the plants regardless of grazing. The recov- event to assess plant re-establishment.
ery of seagrass during the summer months re-estab-
lishes the food resources such that waterfowl return on
a seasonal basis. The dependence of many waterfowl MATERIALS AND METHODS
species on recurring seagrass populations has been
well documented (Ganter 2000, Moore et al. 2004), and The 10 ha Fishing Island eelgrass meadow is located
waterfowl have been known to alter their migration at the mouth of the Piscataqua River in the Great Bay
routes when seagrass resources become unavailable Estuary, on the border of New Hampshire and Maine,
(Seymour et al. 2002). USA (43° 04.57’ N, 70° 41.89’ W; Fig. 1). The water
In the Great Bay Estuary, on the border of New depth at this site ranges from 0 to 0.5 m at low tide and
Hampshire and Maine, USA (see Fig. 1), Canada 3 to 4 m at high tide. Water temperature ranges from
Fig. 1 Great Bay Estuary, on
the border of New Hampshire
and Maine, USA, and the
Fishing Island eelgrass mea-
dow with SeagrassNet trans-
ects (Transects A, B and C).
Eelgrass coverage is based on
near-vertical aerial photogra-
phy taken in August 2002
Rivers & Short: Eelgrass affected by goose grazing 273
1.0 to 19.0°C, and salinity ranges from 25 to 34 PSU between 5 and 10 m apart, and the exact locations
(Short 1992). Eelgrass at this site shows seasonal fluc- were recorded using a Global Positioning System.
tuations in growth (Gaeckle & Short 2002) and biomass Quadrats were numbered according to their location,
(Burdick et al. 1993), with peak biomass levels occur- and measurements of shoot density, canopy height,
ring in September and the lowest levels occurring in and number of seedlings present within the quadrats
February. Change analysis of the eelgrass meadows in were recorded. Finally, each quadrat was excavated to
the Great Bay Estuary was made based on near-verti- a depth of 10 cm, and the collected sediment and all
cal aerial photography (Short & Burdick 1996). associated plant material were placed in plastic bags.
SeagrassNet is a seagrass monitoring program with Excavated samples were transported to the Jackson
sites throughout the world (www.SeagrassNet.org; Estuarine Laboratory for further analysis. In both April
Short et al. 2002). At each site, field sampling occurs and July 2003, each of the 6 previous excavation sites
along 3 permanent transects, situated inshore (Tran- was located, and new 0.0625 m2 quadrats were hap-
sect A), in the middle of the meadow (Transect B), and hazardly placed near the previous sites and sampled as
at the outer edge of the meadow (Transect C; Fig. 1). above.
Located along each transect are 12 permanent but ran- In the laboratory, each sediment sample was placed
domly selected 0.25 m2 quadrats from which data are in a bag with 2.0 mm2 mesh, and the sediment was
collected following the SeagrassNet protocol (Short et rinsed away using seawater. The rinsed samples were
al. 2002). Percent cover of eelgrass within each quadrat stored at 5°C and processed within 10 d of collection.
is recorded, and canopy height at each quadrat is The plant material was sorted into shoots, roots, rhi-
determined from the average height of 3 shoots. Bio- zomes, and seeds. Eelgrass detritus and any non-eel-
mass samples are collected from a 0.0035 m2 core grass material were discarded. The number of terminal
taken from an area of similar cover more than 0.5 m shoots, which grow directly from the end of the main
shoreward outside of each quadrat; plants are dried at rhizome segment, and lateral shoots, which grow from
60°C for 48 h to obtain dry weight. Monitoring occurs the end of branching rhizome segments, was recorded,
quarterly (January, April, July, and October). Sea- as well as the number of each type of shoot (terminal,
grassNet data collected at Fishing Island were used to lateral) that showed evidence of grazing. Grazed
compare bed characteristics of eelgrass between a shoots were defined, based on field observations, as
year when no waterfowl grazing occurred and a year shoots with leaves torn off near the sheath, or as shoots
when Canada geese over-wintered. Specifically, eel- with the leaves, sheath, and meristem missing from the
grass canopy height, percent cover, and aboveground end of the rhizome. The total length of the rhizome was
biomass from October 2001 through July 2003 were measured. The terminal shoots, lateral shoots, rhi-
measured for this study. zomes, and roots for each site were dried at 60°C for
Temperature, light, and salinity at the Fishing Island 48 h to obtain biomass.
meadow were also monitored as part of the Seagrass- Change analysis was used to compare eelgrass
Net protocol. Temperature data were collected using meadow percent cover between the Fishing Island
Onset TidbiT temperature sensors, which were placed meadow and meadows in Great Bay from 2002 to 2003.
on the meadow surface at Transect C and set to record The Fishing Island SeagrassNet data were analyzed
at hourly intervals. Temperature sensors were col- using a repeated-measures analysis of variance
lected after 6 to 31 d of recording. Light data were (ANOVA), with quadrats as subjects, transect as a
measured using 2 Onset Hobo light sensors placed on within-subject factor, and sampling date as an among-
the meadow surface at Transects A and C. The Hobo subject factor (Zar 1999). Multiple comparisons were
sensors recorded light (lumens ft–2) at 12:00 h for 10 d performed among transects and sampling dates using
prior to SeagrassNet sampling. Salinity was measured a Tukey’s honestly significant difference (HSD) test.
at each transect on an incoming tide immediately after Canopy height and aboveground biomass data were
SeagrassNet sampling. Average values of tempera- log transformed before analysis, and percent cover
ture, light, and salinity were calculated for each sam- data were logit transformed to produce homogeneous
pling period. variance. SeagrassNet environmental data (salinity,
In February, April, and July 2003, an additional field temperature, light) and eelgrass parameter data mea-
assessment was conducted to further capture the sured during February, April, and July 2003 field sam-
effects of Canada goose grazing activity at Fishing pling were analyzed using a 1-way ANOVA. A resid-
Island. The grazing study was conducted at an area of ual analysis was performed for each parameter, and
the eelgrass meadow adjacent to Transect B of Sea- where necessary, raw data were log transformed prior
grassNet. In February 2003, six 0.0625 m2 quadrats to analysis (Zar 1999). For all 1-way ANOVAs, multiple
were haphazardly placed such that they all contained comparisons among sampling dates were performed
at least 1 eelgrass shoot. Quadrats were positioned using a Tukey’s HSD test.
274 Mar Ecol Prog Ser 333: 271–279, 2007
RESULTS
Table 1. Zostera marina. Eelgrass parameter means (± SE) of SeagrassNet monitoring data (n = 12 for all parameters). Data from October 2001 through July 2002 represent
a year when no grazing of eelgrass by Canada geese Branta canadensis occurred at Fishing Island. Data from October 2002 through July 2003 span a period when
14.9 ± 1.1b,c
0.6 ± 0.3d
0.6 ± 0.3d
9.9 ± 1.7d
5.6 ± 2.1e
36 ± 12c
10.0 ± 2d
2 ± 2d
14 ± 7c
Jul 2003
The repeated-measures ANOVA showed significant
interactions between transect and sampling date for
approximately 100 Canada geese over-wintered at the site. For each transect, letters indicate significant differences between sampling periods (p < 0.05)
all eelgrass parameters monitored by SeagrassNet
(canopy height F14,231 = 13.08, p < 0.001; percent cover
F14,231 = 8.54, p < 0.001; aboveground biomass F14,231 =
6.42, p < 0.001). Although significant interactions were
6.0 ± 1.0d
0.4 ± 0.1d
2.2 ± 1.1d
6.5 ± 0.3d
5.5 ± 0.5d
6.3 ± 0.5d
Apr 2003
9 ± 2d
5 ± 1d
6 ± 1c
detected, comparisons made between transects did not
show a consistent pattern, and only comparisons
within each transect over time are shown (Table 1).
Although the SeagrassNet data were analyzed as a
complete dataset, for presentation, the results are sep-
9.4 ± 0.7c,d
12.0 ± 1.1c,d
15.5 ± 0.7c
arated into 1 yr blocks. During Year 1 (October 2001
Jan 2003
8 ± 2d
27 ± 5b
32 ± 4d
30 ± 4c
34 ± 7c
42 ± 7c
through July 2002), no Canada geese were observed at
the Fishing Island eelgrass meadow, and eelgrass
shoots examined during this time showed no evidence
of having been grazed. Year 2 of SeagrassNet monitor-
ing (October 2002 through July 2003) includes a winter
21.3 ± 1.8a,b
33.6 ± 1.4a,b
35.8 ± 1.7a,b
143 ± 22a,b
186 ± 23a
168 ± 21a
Oct 2002
when Canada geese were actively grazing on eelgrass
61 ± 4b
61 ± 7a
30 ± 5c
at the Fishing Island meadow. Canada geese were first
sighted during the January 2003 monitoring effort,
when a flock of nearly 100 geese was observed. Simi-
lar numbers of geese were seen in February and
March 2003. The geese left the meadow in early April
60.0 ± 2.7b
41.8 ± 2.5b
27.4 ± 2.3a
556 ± 91b
248 ± 32a
685 ± 75c
Jul 2002
98 ± 1b
85 ± 4a
2003 and no further sightings occurred during the
83 ± 6c
course of the study.
During Year 1, eelgrass canopy height and above-
ground biomass showed similar seasonal fluctuations
at all 3 transects, with low values occurring in January
126 ± 13a,b,c
18.0 ± 0.8a,b
24.4 ± 1.6a,b
and peak values occurring in July (Fig. 2). The same
30.6 ± 1.1a
67 ± 4a,b
381 ± 37b
203 ± 30a
Apr 2002
43 ± 7a,c
74 ± 3a,c
seasonal fluctuations appeared in the percent cover
data, with the exception of Transect B, where low per-
cent cover occurred in April. Throughout Year 1, eel-
grass canopy height and aboveground biomass were
consistently lowest at Transect A and highest at Tran-
14.3 ± 1.3b,c
28.1 ± 1.3a,c
21.0 ± 1.5a,c
109 ± 21b,c
sect B (Table 1).
193 ± 73a
150 ± 23a
47 ± 5a,c
Jan 2002
58 ± 6b
33 ± 5b
In Year 2, all eelgrass parameters declined from
October to January, and parameters continued to
decline into April 2003 (Table 1). Eelgrass shoots show-
ing evidence of grazing were abundant at the meadow
during SeagrassNet sampling in January and April
37.1 ± 2.0a,b
36.5 ± 4.1a,b
159 ± 23a,b
301 ± 54a,b
28.6 ± 2.1a
195 ± 35a
2003. Between April and July of Year 2, eelgrass per-
Oct 2001
81 ± 5a
60 ± 4a
62 ± 7a
cent cover did not change at any of the transects, and
only Transect A showed a slight increase in above-
Aboveground biomass (g m–2)
ground biomass.
By the end of January in Year 2, eelgrass above-
ground biomass at Transects B and C was significantly
Canopy height (cm)
lower than the January levels of Year 1 (Table 1). In
April and July of Year 2, all 3 transects had signifi-
Percent cover
cantly lower values for eelgrass percent cover, canopy
height, and aboveground biomass compared to the
Transect
same sampling periods in Year 1. Transect B showed
the largest overall declines in eelgrass parameters,
A
C
A
C
A
C
B
B
B
declining from 98% cover and 685 g m–2 aboveground
Rivers & Short: Eelgrass affected by goose grazing 275
cover (F. T. Short unpubl. data). The aerial pho-
60 Transect A tographs revealed that the first clear decline in eel-
Canopy height (cm)
Transect B grass percent cover occurred between Years 1 and 2 of
Transect C our study (Fig. 3).
40 Environmental parameters at Fishing Island did not
fall outside expected ranges in either Year 1 or Year 2
(Fig. 4) for the Great Bay Estuary (Short 1992). A 1-way
ANOVA showed a significant effect of sampling date
20 for all parameters (salinity F6,13 = 9.76, p < 0.001;
temperature F7,4695 = 4778.00, p < 0.001; light F7,72 =
11.54, p < 0.001). Temperature at the meadow surface
0 showed a seasonal fluctuation in both years, with aver-
age winter temperatures lower in Year 2 than in Year 1
100 (p < 0.05). Average salinities were not significantly dif-
ferent between the same months in Years 1 and 2 (p >
Percent cover
80 0.05), and salinities never fell below 24 PSU. Average
light reaching the meadow surface was not different
60 (p > 0.05) among sampling dates, with the exception of
July 2003, which had significantly higher average light
40
than the other dates (p < 0.05).
In February 2003, a month after Canada geese were
first sighted at the meadow, there was a higher shoot
20
density of lateral shoots (68 shoots m–2) than terminal
shoots (17 shoots m–2); a greater proportion of terminal
0
shoots showed evidence of having been grazed
(Table 2). Eelgrass shoot density decreased from Feb-
Aboveground biomass (g m–2)
800
ruary to April 2003 while geese were active at the Fish-
ing Island site, and by April 2003, the proportion of
600 grazed lateral shoots had increased to 60%. By July
2003, after Canada geese had left the meadow, no ter-
minal shoots remained, and the only evidence of graz-
400
ing appeared on a few lateral shoots (Table 2). For the
majority of grazed shoots examined during this study,
200 the meristem was completely removed by the geese. In
only a few cases did the plants have missing leaves but
an intact meristem.
0 Eelgrass rhizome weight declined from 15.4 g m–2 in
1 2 2 2 2 3 3 3
t0 n0 r 0 ul 0 t 0 an 0 pr 0 Jul 0 February 2003 to 9.0 g m–2 in April 2003, and did not
Oc Ja Ap J Oc J A
change from April to July (Table 3). Root biomass
Fig. 2. Zostera marina. Seasonal changes in eelgrass parame- demonstrated a similar pattern of decline, from 8.6 g
ters (mean ± SE, n = 12) measured by SeagrassNet monitoring m–2 in February to 7.3 g m–2 in April, and showed no
for Transects A, B, and C. Year 1 (October 2001 through July
change from April to July 2003. Eelgrass rhizome
2002) represents a year when no grazing of eelgrass by
Canada geese occurred at Fishing Island. Year 2 (October length did not change from February to April 2003, but
2002 through July 2003) spans the season when approxi- declined to a low of 854 cm m–2 by July 2003.
mately 100 Canada geese over-wintered at Fishing Island.
Significant differences between sampling periods are listed
in Table 1
DISCUSSION
biomass in July of Year 1 to <1% cover and 2 g m–2 Historical monitoring of both eelgrass and waterfowl
aboveground biomass in July of Year 2. at Fishing Island showed no evidence of the eelgrass
Further evidence for the decline in eelgrass percent meadow being grazed prior to January 2003. Aerial
cover was obtained through the analysis of aerial waterfowl surveys of the Great Bay Estuary conducted
photographs of Fishing Island. Two decades of annual annually since 1985 show that Canada geese typically
aerial photography of the meadow showed that the over-winter on the extensive Great Bay eelgrass mead-
eelgrass at Fishing Island had historically high percent ows, and no geese have been sighted over-wintering at
276 Mar Ecol Prog Ser 333: 271–279, 2007
made usual food resources inaccessible
and may have forced Canada geese
toward the coast to feed. During the
winter and spring of 2003, Canada
geese fed at Fishing Island from Janu-
ary through April. Dramatic evidence of
the effects of goose grazing is seen by
comparing the aerial photographs from
2002 and 2003 (Fig. 2).
The Fishing Island site is the only
coastal intertidal eelgrass meadow in
the Great Bay Estuary, but many other
intertidal eelgrass flats occur at the
upper end of the estuary in Great Bay
itself. However, in Year 2, which was
colder (Fig. 4) and had more ice cover
than Year 1, geese apparently fed more
at Fishing Island, as it was not frozen
over and remained accessible through-
out the winter. Aerial photography
(F. T. Short unpubl. data) shows that the
large intertidal eelgrass meadows in
Great Bay (Fig. 1) did not change sub-
stantially from Year 1 to Year 2.
Eelgrass in the Great Bay Estuary has
a cyclical growth pattern, with peak
growth occurring in late summer and
the lowest growth occurring in mid-
winter (Short et al. 1989), a pattern that
Fig. 3. Zostera marina. Aerial photographs of the Fishing Island eelgrass has been confirmed for the Fishing
meadow, from (A) August 2001 and (B) August 2002. The dark area on the mud- Island meadow (Gaeckle & Short 2002).
flat is the eelgrass Eelgrass parameters from Year 1 of this
study displayed the expected seasonal
Fishing Island (Vogel 1995, B. Smith pers. comm.). growth pattern and provide a baseline for comparison
Canada geese were first sighted at the Fishing Island with Year 2, when Canada goose grazing activity was
meadow during the January 2003 SeagrassNet moni- prevalent throughout the winter and spring. Eelgrass
toring effort. Annual aerial photography of the Great data from Year 2 did not conform to the typical season-
Bay Estuary revealed no evidence of eelgrass decline ality seen in Year 1, and instead showed severe
at Fishing Island until 2003. In the Great Bay Estuary declines. Changes in environmental factors are often
region, inland agricultural fields provide the most responsible for seagrass loss, but no dramatic changes
abundant food resources for over-wintering Canada in temperature, salinity, or light were observed at the
geese (Vogel 1995). The winter of 2003 was unusually Fishing Island meadow between Years 1 and 2 (Fig. 4).
cold, and inland agricultural fields were covered by The lower average temperatures in Year 2 are not
almost 1 m of snow. Great Bay itself was frozen over for atypical for the region, and the lower average salinity
part of the winter. The continuous snow cover and cold did not fall beneath typical eelgrass bed salinity levels
Table 2. Zostera marina. Eelgrass shoot density, seedling density, seed density, and percentage of grazed shoots at Fishing Island
from February to July 2003 (means ± SE, n = 6). Letters indicate significant differences between sampling periods (p < 0.05)
Terminal shoot Grazed terminal Lateral Grazed lateral Seeds m–2 Seedlings
density (shoots m–2) shoots (%) shoot density shoots (shoots m–2) (%) m–2
Feb 17 ± 3 a 56 68 ± 7 a 26 7±4a 0
Apr 1±1b 50 7±1b 60 5±1a 0
Jul 5±2b 0 10 ± 4 b 7 1±1a 0
Rivers & Short: Eelgrass affected by goose grazing 277
rate of eelgrass depletion at these areas to be greater
Salinity (PSU) and temperature (°C)
Salinity Temperature Light
than at the inshore portion of the meadow (Transect A).
35 1200
By consuming eelgrass aboveground biomass during
Light (lumens ft–2)
30 1000 the grazing event, Canada geese had an indirect effect
25 on eelgrass belowground biomass. Separation of
800 aboveground biomass quickly leads to senescence of
20
600 the belowground plant parts (Kenworthy & Thayer
15 1984). At Fishing Island, rhizome weight and root
400 weight declined quickly with the loss of aboveground
10
200 biomass (Table 3), suggesting a loss of stored carbon
5
typical for decaying plant material (Kenworthy &
0 0 Thayer 1984). Although rhizome weight dropped
1 02 02 02 02 03 03 03
t0 r l t r l
Oc Jan Ap Ju Oc Jan Ap Ju rapidly, rhizome length did not significantly decline
until 3 mo after geese had left the meadow (Table 3).
Fig. 4. Change in environmental parameters measured dur- The delay between the decline in rhizome weight and
ing SeagrassNet monitoring (mean ± SD). Data points show
complete decay of the rhizome structure has also been
salinity (n = 3), temperature (n ranges from 144 to 745) and
light (n = 10) observed in seagrass decomposition studies (Kenwor-
thy & Thayer 1984).
The patterns of eelgrass decline at the Fishing Island
Table 3. Zostera marina. Eelgrass belowground parameters at meadow indicate that Canada geese may have been
Fishing Island from February to July 2003 (means ± SE, n = 6).
Letters indicate significant differences between sampling selecting eelgrass for consumption based on shoot size.
periods (p < 0.05) Eelgrass terminal shoots are larger than lateral shoots
(Bak 1980), and if geese select shoots based on size,
Rhizome Rhizome Root terminal shoots would be consumed preferentially.
length weight weight Before Canada geese arrived at Fishing Island, the eel-
(cm m–2) (g m–2) (g m–2) grass at the site exhibited growth characteristics typi-
cal of a healthy meadow, with a higher proportion of
Feb 1757 ± 165a 15.4 ± 1.6a 8.6 ± 2.3a
Apr 1510 ± 272a 9.0 ± 1.1b 7.3 ± 1.1b lateral shoots than terminal shoots (Bak 1980). In Feb-
Jul 854 ± 175b 5.2 ± 1.3b 3.5 ± 1.1b ruary 2003, eelgrass terminal shoots were preferen-
tially grazed even though lateral shoot density was
higher. Canada geese selected the larger terminal
(Pinnerup 1980). The only detectable difference in shoots while they were still available, but by April
light reaching the meadow was an increase in light 2003 there were almost no terminal shoots left in the
that occurred after Canada geese had left the site. The meadow, and goose grazing activity on lateral shoots
goose grazing activity at Fishing Island from January increased accordingly. The overall changes in eelgrass
2003 to April 2003 caused a major decline in eelgrass parameters during the course of the grazing event
plant parameters, and eelgrass indicators remained (Table 2) suggest that Canada geese preferred to
low through July 2003. graze in areas where eelgrass was larger and more
Although the entire Fishing Island meadow was abundant.
greatly affected by the grazing event, the most heavily Several studies concerning waterfowl grazing activ-
grazed area of the meadow was Transect B, which lost ity on seagrass have reported ‘giving-up thresholds,’
over 680 g m–2 of eelgrass aboveground biomass by i.e. the seagrass biomass level (g m–2) at which water-
July 2003, followed by Transect C, which lost over fowl leave a meadow after depleting food resources
540 g m–2. These transects had the greatest initial eel- (Percival & Evans 1997, Clausen 2000). Most of the
grass biomass, and geese likely spent the greatest studies reporting giving-up thresholds concern brant
amount of time feeding at these areas. At a European grazing at European Zostera noltii meadows. For our
Zostera noltii meadow, brant spent the majority of their study, we used the April 2003 aboveground biomass as
time within 100 m of the low tide line, as shown in a an estimated giving-up threshold, since April was the
spatial depletion model developed by Percival et al. last month that Canada geese were observed at Fish-
(1996, 1998). At the Fishing Island meadow, Transects ing Island. The giving-up threshold for our study falls
B and C are closest to the low tide line, and Canada within the range of values listed for brant (Table 4).
geese may have similarly concentrated their feeding After Canada geese left the site, we continued
efforts at these locations. The greater initial biomass at monitoring eelgrass parameters through July 2003 to
Transects B and C, in combination with the increased determine the extent of eelgrass recovery after the
grazing time by Canada geese, would have caused the grazing event. Although aboveground biomass
278 Mar Ecol Prog Ser 333: 271–279, 2007
Table 4. Branta canadensis. Estimated giving-up thresholds (see ‘Discussion’) Island meadow during the winter–
for waterfowl grazing on seagrass meadows. Studies conducted at European spring of both 2004 and 2005 (Short et
sites concern brant Branta bernicla L. grazing on dwarf eelgrass Zostera noltii H.
The giving-up threshold for our study is the April 2003 aboveground biomass, as
al. 2006).
April was the last month in which Canada geese were sighted at Fishing Island
Acknowledgements. We thank B. Smith and
Location Giving-up threshold Source E. Robinson at the New Hampshire State
(g m–2) Fish and Game Department. Thanks to D.
Walker, T. Davis, J. Gaeckle, C. Ochieng,
Wadden Sea, Denmark 7.5 Madsen (1988) and N. Petit for assistance in the field and
Lindisfarne, UK 5.0 Percival & Evans (1997) laboratory. We thank C. Short for editing
Limfjorden, Denmark 1.7–5.9a Clausen (2000) and valuable suggestions on the manuscript.
Fishing Island, USA 6.7 Present study Thanks to 3 anonymous reviewers for help-
a
Biomass estimate based on a modeling approach ful comments. This work is Jackson Estuar-
ine Laboratory Contribution No. 438.
increased slightly from April to July 2003 at parts of LITERATURE CITED
the meadow, eelgrass at Fishing Island clearly did not
Bak HP (1980) Age populations and biometrics in eelgrass,
attain the biomass or percent cover levels seen in a Zostera marina L. Ophelia 19:155–162
typical summer, indicating that recovery was mini- Baldwin JR, Lovvorn, JR (1994) Expansion of seagrass habitat
mal. Few seeds or seedlings were found, and eelgrass by the exotic Zostera japonica, and its use by dabbling
shoot density did not change significantly from April ducks and brant in Boundary Bay, British Columbia. Mar
Ecol Prog Ser 103:119–127
to July 2003, suggesting that neither sexual propaga-
Buchsbaum R (1987) The geese and the grass. Sanctuary 27:
tion nor vegetative expansion of shoots contributed to 7–9
eelgrass recovery. At European Zostera noltii mead- Burdick DM, Short FT, Wolf J (1993) An index to assess and
ows where grazing by brant occurs every winter, Z. monitor the progression of wasting disease in eelgrass
noltii recovers to typical biomass levels in the sum- Zostera marina. Mar Ecol Prog Ser 94:83–90
Clausen P (2000) Modelling water level influence on habitat
mer (Vermaat & Verhagen 1996, Nacken & Reise choice and food availability for Zostera feeding brent
2000). The difference in recovery between that sys- geese Branta bernicla in non-tidal areas. Wildl Biol 6:
tem and Fishing Island appears to be a result of the 75–87
feeding behavior of the geese. Brant consume the Fox AD (1996) Zostera exploitation by brent geese and
wigeon on the Exe Estuary, southern England. Bird Study
leaves of Z. noltii, leaving the meristem intact (Ver-
43:257–268
maat & Verhagen 1996) to re-grow new leaves, Gaeckle JL, Short FT (2002) A plastochrone method for mea-
whereas Canada geese at Fishing Island typically suring leaf growth in eelgrass, Zostera marina L. Bull Mar
consumed the lower part of the shoot, including the Sci 71:1237–1246
sheath and the meristem. Once the meristem is Ganter B (2000) Seagrass (Zostera spp.) as food for brant
geese (Branta bernicla): an overview. Helgol Mar Res 54:
removed, eelgrass is unable to produce new leaves, 63–70
preventing vegetative growth of that shoot. Hanson AR (ed) (2004) Status and conservation of eelgrass
The Fishing Island case study represents one of the (Zostera marina) in Eastern Canada. Technical Report
few accounts of eelgrass decline due to Canada Series No. 412. Canadian Wildlife Service, Atlantic Region
Hughes AR, Stachowicz JJ (2004) Genetic diversity enhances
goose grazing in the Gulf of Maine region. Other
the resistance of a seagrass ecosystem to disturbance. Proc
studies in this region have reported how eelgrass Natl Acad Sci USA 101:8998–9002
availability can affect goose populations (Seymour et Jacobs RPWM, den Hartog C, Braster BF, Carriere FC (1981)
al. 2002, Hanson 2004). In contrast, our results show Grazing of the seagrass Zostera noltii by birds at Ter-
that Canada geese can have a substantial effect on schelling (Dutch Wadden Sea). Aquat Bot 10:241–259
Kenworthy WJ, Thayer GW (1984) Production and decompo-
eelgrass populations. In just 3 mo, a relatively small sition of the roots and rhizomes of seagrasses, Zostera
number of geese almost completely eliminated the marina and Thalassia testudinum, in temperate and sub-
long-standing Fishing Island eelgrass meadow. tropical marine ecosystems. Bull Mar Sci 35:364–379
Unlike many locations where seagrass recovers after Madsen J (1988) Autumn feeding ecology of herbivorous
wildfowl in the Danish Wadden Sea and the impact of food
a waterfowl grazing event, eelgrass at Fishing Island
supplies and shooting movements. Dan Rev Game Biol 13:
did not re-establish after Canada geese left the site. 2–33
The lack of eelgrass recovery will affect the ecology Moore JE, Colwell MA, Mathis RL, Black JM (2004) Staging
of Fishing Island. The small amount of eelgrass of Pacific flyway brant in relation to eelgrass abundance
biomass remaining at Fishing Island cannot provide and site isolation, with special consideration of Humboldt
Bay, California. Biol Conserv 115:475–486
the same functionality as the historical meadow. Fur- Nacken M, Reise K (2000) Effects of herbivorous birds on
ther SeagrassNet monitoring has shown a lack of eel- intertidal seagrass beds in the northern Wadden Sea.
grass recovery and the return of geese to the Fishing Helgol Mar Res 54:87–94
Rivers & Short: Eelgrass affected by goose grazing 279
Percival SM, Evans PR (1997) Brent Geese Branta bernicla Ocean Management. ASCE. July 11–14 Charleston, SC,
and Zostera: factors affecting the exploitation of a season- p 3689–3706
ally declining food resource. Ibis 139:121–128 Short FT, McKenzie LJ, Coles RG, Vidler KP (2002) Seagrass-
Percival SM, Sutherland WJ, Evans PR (1996) A spatial deple- Net Manual for Scientific Monitoring of Seagrass Habitat.
tion model of the responses of grazing wildfowl to the Queensland Department of Primary Industries, QFS,
availability of intertidal vegetation. J Appl Ecol 33:979–992 Cairns
Percival SM, Sutherland WJ, Evans PR (1998) Intertidal habi- Short FT, Koch E, Creed JC, Magalhaes KM, Fernandez E,
tat loss and wildfowl numbers: applications of a spatial Gaeckle J, Rivers DO (2006) SeagrassNet monitoring
depletion model. J Appl Ecol 35:57–63 across the Americas: case studies of seagrass change.
Pinnerup SP 1980. Leaf production of Zostera marina L. at PSZN I: Mar Ecol 27:277–289
different salinities. Ophelia 1:219–224 Tubbs CR, Tubbs JM (1982) Brent geese (Branta bernicla) and
Portig AA, Mathers RG, Montgomery WI, Govier RN (1994) The their food in the Solent, southern England. Biol Conserv
distribution and utilization of Zostera species in Strangford 23:33–54
Lough, Northern Ireland. Aquat Bot 47:317–328 Vermaat JE, Verhagen FCA (1996) Seasonal variation in the
Seymour NR, Miller AG, Garbary DJ (2002) Decline of intertidal seagrass Zostera noltii Hornem: coupling demo-
Canada geese (Branta canadensis) and common golden- graphic and physiological patterns. Aquat Bot 52:259–281
eye (Bucephala clangula) associated with a collapse of eel- Vogel L (1995) Wintering waterfowl on Great Bay. Technical
grass (Zostera marina) in a Nova Scotia estuary. Helgol Report No. 2 prepared for NOAA and the New Hampshire
Mar Res 56:198–202 Fish and Game Department, Concord. NH
Short FT (ed) (1992) The ecology of the Great Bay Estuary, Ward DH, Morton A, Tibbitts TL, Douglas DC, Carrera-
New Hampshire and Maine: an estuarine profile and González E (2003) Long-term change in eelgrass distribu-
bibliography. NOAA — Coastal Ocean Program Publ, tion at Bahía San Quintín, Baja California, Mexico, using
Durham, NH satellite imagery. Estuaries 26:1529–1539
Short FT, Burdick DM (1996) Quantifying eelgrass habitat Wilson UW, Atkinson JB (1995) Black brant winter and
loss in relation to housing development and nitrogen load- spring-staging use at two Washington coastal areas in
ing in Waquoit Bay, Massachusetts. Estuaries 19:730–739 relation to eelgrass abundance. Condor 97:91–98
Short FT, Wolf J, Jones GE (1989) Sustaining eelgrass to man- Zar JH (1999) Biostatistical analysis, 4th edn. Prentice Hall,
age a healthy ecosystem. Proc 6th Symp Coastal and Upper Saddle River, NJ
Editorial responsibility: Kenneth Heck (Contributing Editor), Submitted: November 7, 2005; Accepted: August 15, 2006
Dauphin Island, Alabama, USA Proofs received from author(s): March 1, 2006
Vol. 333: 271–279, 2007 Published March 12
Mar Ecol Prog Ser
Effect of grazing by Canada geese
Branta canadensis on an intertidal eelgrass
Zostera marina meadow
David O. Rivers, Frederick T. Short*
Department of Natural Resources, University of New Hampshire, Jackson Estuarine Laboratory, 85 Adams Point Road,
Durham, New Hampshire 03824, USA
ABSTRACT: Fishing Island, in Portsmouth Harbor on the Maine–New Hampshire border (USA), is
the site of an intertidal eelgrass (Zostera marina L.) bed that is part of SeagrassNet, an international
program for long-term seagrass monitoring. Eelgrass bed parameters of canopy height, percent
cover, and aboveground biomass have been monitored quarterly since October 2001 using the Sea-
grassNet protocol. A flock of nearly 100 Canada geese Branta canadensis L. over-wintered at Fishing
Island and grazed on eelgrass from January to April 2003, an event that had not been seen at this
meadow in 2 decades of observation. Before Canada geese were present, eelgrass parameters
demonstrated seasonal fluctuations typical of the region. During the grazing event, eelgrass parame-
ters declined drastically, and biomass losses reached 680 g m–2 in parts of the meadow. SeagrassNet
data demonstrated that eelgrass did not recover after the geese departed. Additional fieldwork con-
ducted from February to July 2003 showed that eelgrass recruitment via sexual reproduction at Fish-
ing Island was minimal, and vegetative recovery was impeded by Canada goose consumption of the
plant meristems. Unlike studies in other locations, which show seagrass quickly rebounding from
annual grazing events, eelgrass at Fishing Island showed little recovery from Canada goose grazing
through July 2003.
KEY WORDS: Eelgras · Zostera marina · Grazing · Canada goose · SeagrassNet · Climate change
Resale or republication not permitted without written consent of the publisher
INTRODUCTION goose grazing activity during this time can have a sub-
stantial effect on seagrass abundance (Portig et al.
Seagrass has long been recognized as an important 1984, Baldwin & Lovvorn 1994, Ganter 2000). At Euro-
food resource for migratory waterfowl, whose migra- pean seagrass meadows, waterfowl feeding activity
tion routes often coincide with seagrass meadow loca- was shown to reduce plant biomass by more than 50%
tions (Ganter 2000). In North America, eelgrass during the course of the winter grazing period (Jacobs
Zostera marina L. meadows are considered important et al. 1981, Nacken & Reise 2000).
staging areas for black brant geese Branta nigricans L. The rate of depletion of vegetation by grazing water-
on the west coast (Wilson & Atkinson 1995, Ward et al. fowl is primarily influenced by the number of birds
2003, Moore et al. 2004), and for brant B. bernicla L. present at a site and accessibility of the plants (Baldwin
and Canada geese B. canadensis L. on the east coast & Lovvorn 1994, Percival et al. 1996, Clausen 2000).
(Seymour et al. 2002, Hanson 2004). The length of stay Seagrass is accessible to birds in shallow water sys-
of waterfowl at staging areas is often closely correlated tems or in intertidal areas (Ganter 2000); changes in
to the available seagrass resources at that site (Wilson water levels due to the tidal cycle can limit the amount
& Atkinson 1995). Geese may remain at seagrass of time that food resources are obtainable (Fox 1996,
meadows for several months during the winter, and Clausen 2000). Many waterfowl species feed by up-
*Corresponding author. Email: fred.short@unh.edu © Inter-Research 2007 · www.int-res.com
272 Mar Ecol Prog Ser 333: 271–279, 2007
ending in the water (Buchsbaum 1987, Vogel 1995), geese are the most common over-wintering waterfowl
and the length of their neck-plus-head determines the species (Vogel 1995). Canada geese that over-winter
maximum depth at which they can feed (Clausen in this region rely on a few primary food sources,
2000). In intertidal areas, birds may not be able to including upland agricultural fields, golf course
reach plants at high tide; feeding is often restricted to grasses, coastal and estuarine salt marshes, and eel-
low tide when receding water makes the plants acces- grass meadows (Vogel 1995). Within the Great Bay
sible (Fox 1996, Percival & Evans 1997). Observations Estuary, the most extensive eelgrass meadows occur
of brant that congregate in intertidal areas have shown in Great Bay itself, with smaller meadows throughout
that the birds feed whenever they have access to sea- the rest of the estuary and along the coast. This study
grass, day or night, and rest when plants are inaccessi- was conducted at the Fishing Island eelgrass meadow,
ble due to high water levels (Percival & Evans 1997). located near the mouth of the estuary. The Fishing
Over time, the total seagrass biomass available at a Island meadow is a SeagrassNet site, which is part of
given site declines as the plants are consumed, causing an international long-term seagrass monitoring pro-
birds to spend a greater amount of time feeding in gram (Short et al. 2002). Documented cases of eel-
order to achieve an adequate food intake (Percival et grass decline due to Canada goose grazing are rare
al. 1996, 1998). (Hanson 2004), and studies focusing on other water-
Although grazing activity can significantly reduce fowl species often discuss how seagrass availability
plant biomass, at many meadows the biomass loss is a affects bird populations while ignoring the process of
short-term effect, and the affected meadows recover eelgrass recovery (Ganter 2002, Moore et al. 2004).
during the subsequent growing season (Vermaat & Our study presents comparative data of eelgrass para-
Verhagen 1996, Nacken & Reise 2000, Hughes & Sta- meters before, during, and after a Canada goose graz-
chowicz 2004). Winter waterfowl grazing coincides ing event and examines eelgrass recovery post-
with the biomass losses from natural seasonal declines impact. The objectives of this study were to compare
of temperate seagrasses (Short 1992, Ganter 2000). bed characteristics of Zostera marina between a year
Exclosure studies have demonstrated that patches of when grazing occurred and a year when no grazing
seagrass protected from grazing experience biomass occurred at a long-term monitoring site, and to quan-
declines of up to 65% during the winter months (Tubbs tify the extent of eelgrass vegetative and sexual
& Tubbs 1982, Madsen 1988), indicating a marked sea- reproduction in the growing season after the grazing
sonality in the plants regardless of grazing. The recov- event to assess plant re-establishment.
ery of seagrass during the summer months re-estab-
lishes the food resources such that waterfowl return on
a seasonal basis. The dependence of many waterfowl MATERIALS AND METHODS
species on recurring seagrass populations has been
well documented (Ganter 2000, Moore et al. 2004), and The 10 ha Fishing Island eelgrass meadow is located
waterfowl have been known to alter their migration at the mouth of the Piscataqua River in the Great Bay
routes when seagrass resources become unavailable Estuary, on the border of New Hampshire and Maine,
(Seymour et al. 2002). USA (43° 04.57’ N, 70° 41.89’ W; Fig. 1). The water
In the Great Bay Estuary, on the border of New depth at this site ranges from 0 to 0.5 m at low tide and
Hampshire and Maine, USA (see Fig. 1), Canada 3 to 4 m at high tide. Water temperature ranges from
Fig. 1 Great Bay Estuary, on
the border of New Hampshire
and Maine, USA, and the
Fishing Island eelgrass mea-
dow with SeagrassNet trans-
ects (Transects A, B and C).
Eelgrass coverage is based on
near-vertical aerial photogra-
phy taken in August 2002
Rivers & Short: Eelgrass affected by goose grazing 273
1.0 to 19.0°C, and salinity ranges from 25 to 34 PSU between 5 and 10 m apart, and the exact locations
(Short 1992). Eelgrass at this site shows seasonal fluc- were recorded using a Global Positioning System.
tuations in growth (Gaeckle & Short 2002) and biomass Quadrats were numbered according to their location,
(Burdick et al. 1993), with peak biomass levels occur- and measurements of shoot density, canopy height,
ring in September and the lowest levels occurring in and number of seedlings present within the quadrats
February. Change analysis of the eelgrass meadows in were recorded. Finally, each quadrat was excavated to
the Great Bay Estuary was made based on near-verti- a depth of 10 cm, and the collected sediment and all
cal aerial photography (Short & Burdick 1996). associated plant material were placed in plastic bags.
SeagrassNet is a seagrass monitoring program with Excavated samples were transported to the Jackson
sites throughout the world (www.SeagrassNet.org; Estuarine Laboratory for further analysis. In both April
Short et al. 2002). At each site, field sampling occurs and July 2003, each of the 6 previous excavation sites
along 3 permanent transects, situated inshore (Tran- was located, and new 0.0625 m2 quadrats were hap-
sect A), in the middle of the meadow (Transect B), and hazardly placed near the previous sites and sampled as
at the outer edge of the meadow (Transect C; Fig. 1). above.
Located along each transect are 12 permanent but ran- In the laboratory, each sediment sample was placed
domly selected 0.25 m2 quadrats from which data are in a bag with 2.0 mm2 mesh, and the sediment was
collected following the SeagrassNet protocol (Short et rinsed away using seawater. The rinsed samples were
al. 2002). Percent cover of eelgrass within each quadrat stored at 5°C and processed within 10 d of collection.
is recorded, and canopy height at each quadrat is The plant material was sorted into shoots, roots, rhi-
determined from the average height of 3 shoots. Bio- zomes, and seeds. Eelgrass detritus and any non-eel-
mass samples are collected from a 0.0035 m2 core grass material were discarded. The number of terminal
taken from an area of similar cover more than 0.5 m shoots, which grow directly from the end of the main
shoreward outside of each quadrat; plants are dried at rhizome segment, and lateral shoots, which grow from
60°C for 48 h to obtain dry weight. Monitoring occurs the end of branching rhizome segments, was recorded,
quarterly (January, April, July, and October). Sea- as well as the number of each type of shoot (terminal,
grassNet data collected at Fishing Island were used to lateral) that showed evidence of grazing. Grazed
compare bed characteristics of eelgrass between a shoots were defined, based on field observations, as
year when no waterfowl grazing occurred and a year shoots with leaves torn off near the sheath, or as shoots
when Canada geese over-wintered. Specifically, eel- with the leaves, sheath, and meristem missing from the
grass canopy height, percent cover, and aboveground end of the rhizome. The total length of the rhizome was
biomass from October 2001 through July 2003 were measured. The terminal shoots, lateral shoots, rhi-
measured for this study. zomes, and roots for each site were dried at 60°C for
Temperature, light, and salinity at the Fishing Island 48 h to obtain biomass.
meadow were also monitored as part of the Seagrass- Change analysis was used to compare eelgrass
Net protocol. Temperature data were collected using meadow percent cover between the Fishing Island
Onset TidbiT temperature sensors, which were placed meadow and meadows in Great Bay from 2002 to 2003.
on the meadow surface at Transect C and set to record The Fishing Island SeagrassNet data were analyzed
at hourly intervals. Temperature sensors were col- using a repeated-measures analysis of variance
lected after 6 to 31 d of recording. Light data were (ANOVA), with quadrats as subjects, transect as a
measured using 2 Onset Hobo light sensors placed on within-subject factor, and sampling date as an among-
the meadow surface at Transects A and C. The Hobo subject factor (Zar 1999). Multiple comparisons were
sensors recorded light (lumens ft–2) at 12:00 h for 10 d performed among transects and sampling dates using
prior to SeagrassNet sampling. Salinity was measured a Tukey’s honestly significant difference (HSD) test.
at each transect on an incoming tide immediately after Canopy height and aboveground biomass data were
SeagrassNet sampling. Average values of tempera- log transformed before analysis, and percent cover
ture, light, and salinity were calculated for each sam- data were logit transformed to produce homogeneous
pling period. variance. SeagrassNet environmental data (salinity,
In February, April, and July 2003, an additional field temperature, light) and eelgrass parameter data mea-
assessment was conducted to further capture the sured during February, April, and July 2003 field sam-
effects of Canada goose grazing activity at Fishing pling were analyzed using a 1-way ANOVA. A resid-
Island. The grazing study was conducted at an area of ual analysis was performed for each parameter, and
the eelgrass meadow adjacent to Transect B of Sea- where necessary, raw data were log transformed prior
grassNet. In February 2003, six 0.0625 m2 quadrats to analysis (Zar 1999). For all 1-way ANOVAs, multiple
were haphazardly placed such that they all contained comparisons among sampling dates were performed
at least 1 eelgrass shoot. Quadrats were positioned using a Tukey’s HSD test.
274 Mar Ecol Prog Ser 333: 271–279, 2007
RESULTS
Table 1. Zostera marina. Eelgrass parameter means (± SE) of SeagrassNet monitoring data (n = 12 for all parameters). Data from October 2001 through July 2002 represent
a year when no grazing of eelgrass by Canada geese Branta canadensis occurred at Fishing Island. Data from October 2002 through July 2003 span a period when
14.9 ± 1.1b,c
0.6 ± 0.3d
0.6 ± 0.3d
9.9 ± 1.7d
5.6 ± 2.1e
36 ± 12c
10.0 ± 2d
2 ± 2d
14 ± 7c
Jul 2003
The repeated-measures ANOVA showed significant
interactions between transect and sampling date for
approximately 100 Canada geese over-wintered at the site. For each transect, letters indicate significant differences between sampling periods (p < 0.05)
all eelgrass parameters monitored by SeagrassNet
(canopy height F14,231 = 13.08, p < 0.001; percent cover
F14,231 = 8.54, p < 0.001; aboveground biomass F14,231 =
6.42, p < 0.001). Although significant interactions were
6.0 ± 1.0d
0.4 ± 0.1d
2.2 ± 1.1d
6.5 ± 0.3d
5.5 ± 0.5d
6.3 ± 0.5d
Apr 2003
9 ± 2d
5 ± 1d
6 ± 1c
detected, comparisons made between transects did not
show a consistent pattern, and only comparisons
within each transect over time are shown (Table 1).
Although the SeagrassNet data were analyzed as a
complete dataset, for presentation, the results are sep-
9.4 ± 0.7c,d
12.0 ± 1.1c,d
15.5 ± 0.7c
arated into 1 yr blocks. During Year 1 (October 2001
Jan 2003
8 ± 2d
27 ± 5b
32 ± 4d
30 ± 4c
34 ± 7c
42 ± 7c
through July 2002), no Canada geese were observed at
the Fishing Island eelgrass meadow, and eelgrass
shoots examined during this time showed no evidence
of having been grazed. Year 2 of SeagrassNet monitor-
ing (October 2002 through July 2003) includes a winter
21.3 ± 1.8a,b
33.6 ± 1.4a,b
35.8 ± 1.7a,b
143 ± 22a,b
186 ± 23a
168 ± 21a
Oct 2002
when Canada geese were actively grazing on eelgrass
61 ± 4b
61 ± 7a
30 ± 5c
at the Fishing Island meadow. Canada geese were first
sighted during the January 2003 monitoring effort,
when a flock of nearly 100 geese was observed. Simi-
lar numbers of geese were seen in February and
March 2003. The geese left the meadow in early April
60.0 ± 2.7b
41.8 ± 2.5b
27.4 ± 2.3a
556 ± 91b
248 ± 32a
685 ± 75c
Jul 2002
98 ± 1b
85 ± 4a
2003 and no further sightings occurred during the
83 ± 6c
course of the study.
During Year 1, eelgrass canopy height and above-
ground biomass showed similar seasonal fluctuations
at all 3 transects, with low values occurring in January
126 ± 13a,b,c
18.0 ± 0.8a,b
24.4 ± 1.6a,b
and peak values occurring in July (Fig. 2). The same
30.6 ± 1.1a
67 ± 4a,b
381 ± 37b
203 ± 30a
Apr 2002
43 ± 7a,c
74 ± 3a,c
seasonal fluctuations appeared in the percent cover
data, with the exception of Transect B, where low per-
cent cover occurred in April. Throughout Year 1, eel-
grass canopy height and aboveground biomass were
consistently lowest at Transect A and highest at Tran-
14.3 ± 1.3b,c
28.1 ± 1.3a,c
21.0 ± 1.5a,c
109 ± 21b,c
sect B (Table 1).
193 ± 73a
150 ± 23a
47 ± 5a,c
Jan 2002
58 ± 6b
33 ± 5b
In Year 2, all eelgrass parameters declined from
October to January, and parameters continued to
decline into April 2003 (Table 1). Eelgrass shoots show-
ing evidence of grazing were abundant at the meadow
during SeagrassNet sampling in January and April
37.1 ± 2.0a,b
36.5 ± 4.1a,b
159 ± 23a,b
301 ± 54a,b
28.6 ± 2.1a
195 ± 35a
2003. Between April and July of Year 2, eelgrass per-
Oct 2001
81 ± 5a
60 ± 4a
62 ± 7a
cent cover did not change at any of the transects, and
only Transect A showed a slight increase in above-
Aboveground biomass (g m–2)
ground biomass.
By the end of January in Year 2, eelgrass above-
ground biomass at Transects B and C was significantly
Canopy height (cm)
lower than the January levels of Year 1 (Table 1). In
April and July of Year 2, all 3 transects had signifi-
Percent cover
cantly lower values for eelgrass percent cover, canopy
height, and aboveground biomass compared to the
Transect
same sampling periods in Year 1. Transect B showed
the largest overall declines in eelgrass parameters,
A
C
A
C
A
C
B
B
B
declining from 98% cover and 685 g m–2 aboveground
Rivers & Short: Eelgrass affected by goose grazing 275
cover (F. T. Short unpubl. data). The aerial pho-
60 Transect A tographs revealed that the first clear decline in eel-
Canopy height (cm)
Transect B grass percent cover occurred between Years 1 and 2 of
Transect C our study (Fig. 3).
40 Environmental parameters at Fishing Island did not
fall outside expected ranges in either Year 1 or Year 2
(Fig. 4) for the Great Bay Estuary (Short 1992). A 1-way
ANOVA showed a significant effect of sampling date
20 for all parameters (salinity F6,13 = 9.76, p < 0.001;
temperature F7,4695 = 4778.00, p < 0.001; light F7,72 =
11.54, p < 0.001). Temperature at the meadow surface
0 showed a seasonal fluctuation in both years, with aver-
age winter temperatures lower in Year 2 than in Year 1
100 (p < 0.05). Average salinities were not significantly dif-
ferent between the same months in Years 1 and 2 (p >
Percent cover
80 0.05), and salinities never fell below 24 PSU. Average
light reaching the meadow surface was not different
60 (p > 0.05) among sampling dates, with the exception of
July 2003, which had significantly higher average light
40
than the other dates (p < 0.05).
In February 2003, a month after Canada geese were
first sighted at the meadow, there was a higher shoot
20
density of lateral shoots (68 shoots m–2) than terminal
shoots (17 shoots m–2); a greater proportion of terminal
0
shoots showed evidence of having been grazed
(Table 2). Eelgrass shoot density decreased from Feb-
Aboveground biomass (g m–2)
800
ruary to April 2003 while geese were active at the Fish-
ing Island site, and by April 2003, the proportion of
600 grazed lateral shoots had increased to 60%. By July
2003, after Canada geese had left the meadow, no ter-
minal shoots remained, and the only evidence of graz-
400
ing appeared on a few lateral shoots (Table 2). For the
majority of grazed shoots examined during this study,
200 the meristem was completely removed by the geese. In
only a few cases did the plants have missing leaves but
an intact meristem.
0 Eelgrass rhizome weight declined from 15.4 g m–2 in
1 2 2 2 2 3 3 3
t0 n0 r 0 ul 0 t 0 an 0 pr 0 Jul 0 February 2003 to 9.0 g m–2 in April 2003, and did not
Oc Ja Ap J Oc J A
change from April to July (Table 3). Root biomass
Fig. 2. Zostera marina. Seasonal changes in eelgrass parame- demonstrated a similar pattern of decline, from 8.6 g
ters (mean ± SE, n = 12) measured by SeagrassNet monitoring m–2 in February to 7.3 g m–2 in April, and showed no
for Transects A, B, and C. Year 1 (October 2001 through July
change from April to July 2003. Eelgrass rhizome
2002) represents a year when no grazing of eelgrass by
Canada geese occurred at Fishing Island. Year 2 (October length did not change from February to April 2003, but
2002 through July 2003) spans the season when approxi- declined to a low of 854 cm m–2 by July 2003.
mately 100 Canada geese over-wintered at Fishing Island.
Significant differences between sampling periods are listed
in Table 1
DISCUSSION
biomass in July of Year 1 to <1% cover and 2 g m–2 Historical monitoring of both eelgrass and waterfowl
aboveground biomass in July of Year 2. at Fishing Island showed no evidence of the eelgrass
Further evidence for the decline in eelgrass percent meadow being grazed prior to January 2003. Aerial
cover was obtained through the analysis of aerial waterfowl surveys of the Great Bay Estuary conducted
photographs of Fishing Island. Two decades of annual annually since 1985 show that Canada geese typically
aerial photography of the meadow showed that the over-winter on the extensive Great Bay eelgrass mead-
eelgrass at Fishing Island had historically high percent ows, and no geese have been sighted over-wintering at
276 Mar Ecol Prog Ser 333: 271–279, 2007
made usual food resources inaccessible
and may have forced Canada geese
toward the coast to feed. During the
winter and spring of 2003, Canada
geese fed at Fishing Island from Janu-
ary through April. Dramatic evidence of
the effects of goose grazing is seen by
comparing the aerial photographs from
2002 and 2003 (Fig. 2).
The Fishing Island site is the only
coastal intertidal eelgrass meadow in
the Great Bay Estuary, but many other
intertidal eelgrass flats occur at the
upper end of the estuary in Great Bay
itself. However, in Year 2, which was
colder (Fig. 4) and had more ice cover
than Year 1, geese apparently fed more
at Fishing Island, as it was not frozen
over and remained accessible through-
out the winter. Aerial photography
(F. T. Short unpubl. data) shows that the
large intertidal eelgrass meadows in
Great Bay (Fig. 1) did not change sub-
stantially from Year 1 to Year 2.
Eelgrass in the Great Bay Estuary has
a cyclical growth pattern, with peak
growth occurring in late summer and
the lowest growth occurring in mid-
winter (Short et al. 1989), a pattern that
Fig. 3. Zostera marina. Aerial photographs of the Fishing Island eelgrass has been confirmed for the Fishing
meadow, from (A) August 2001 and (B) August 2002. The dark area on the mud- Island meadow (Gaeckle & Short 2002).
flat is the eelgrass Eelgrass parameters from Year 1 of this
study displayed the expected seasonal
Fishing Island (Vogel 1995, B. Smith pers. comm.). growth pattern and provide a baseline for comparison
Canada geese were first sighted at the Fishing Island with Year 2, when Canada goose grazing activity was
meadow during the January 2003 SeagrassNet moni- prevalent throughout the winter and spring. Eelgrass
toring effort. Annual aerial photography of the Great data from Year 2 did not conform to the typical season-
Bay Estuary revealed no evidence of eelgrass decline ality seen in Year 1, and instead showed severe
at Fishing Island until 2003. In the Great Bay Estuary declines. Changes in environmental factors are often
region, inland agricultural fields provide the most responsible for seagrass loss, but no dramatic changes
abundant food resources for over-wintering Canada in temperature, salinity, or light were observed at the
geese (Vogel 1995). The winter of 2003 was unusually Fishing Island meadow between Years 1 and 2 (Fig. 4).
cold, and inland agricultural fields were covered by The lower average temperatures in Year 2 are not
almost 1 m of snow. Great Bay itself was frozen over for atypical for the region, and the lower average salinity
part of the winter. The continuous snow cover and cold did not fall beneath typical eelgrass bed salinity levels
Table 2. Zostera marina. Eelgrass shoot density, seedling density, seed density, and percentage of grazed shoots at Fishing Island
from February to July 2003 (means ± SE, n = 6). Letters indicate significant differences between sampling periods (p < 0.05)
Terminal shoot Grazed terminal Lateral Grazed lateral Seeds m–2 Seedlings
density (shoots m–2) shoots (%) shoot density shoots (shoots m–2) (%) m–2
Feb 17 ± 3 a 56 68 ± 7 a 26 7±4a 0
Apr 1±1b 50 7±1b 60 5±1a 0
Jul 5±2b 0 10 ± 4 b 7 1±1a 0
Rivers & Short: Eelgrass affected by goose grazing 277
rate of eelgrass depletion at these areas to be greater
Salinity (PSU) and temperature (°C)
Salinity Temperature Light
than at the inshore portion of the meadow (Transect A).
35 1200
By consuming eelgrass aboveground biomass during
Light (lumens ft–2)
30 1000 the grazing event, Canada geese had an indirect effect
25 on eelgrass belowground biomass. Separation of
800 aboveground biomass quickly leads to senescence of
20
600 the belowground plant parts (Kenworthy & Thayer
15 1984). At Fishing Island, rhizome weight and root
400 weight declined quickly with the loss of aboveground
10
200 biomass (Table 3), suggesting a loss of stored carbon
5
typical for decaying plant material (Kenworthy &
0 0 Thayer 1984). Although rhizome weight dropped
1 02 02 02 02 03 03 03
t0 r l t r l
Oc Jan Ap Ju Oc Jan Ap Ju rapidly, rhizome length did not significantly decline
until 3 mo after geese had left the meadow (Table 3).
Fig. 4. Change in environmental parameters measured dur- The delay between the decline in rhizome weight and
ing SeagrassNet monitoring (mean ± SD). Data points show
complete decay of the rhizome structure has also been
salinity (n = 3), temperature (n ranges from 144 to 745) and
light (n = 10) observed in seagrass decomposition studies (Kenwor-
thy & Thayer 1984).
The patterns of eelgrass decline at the Fishing Island
Table 3. Zostera marina. Eelgrass belowground parameters at meadow indicate that Canada geese may have been
Fishing Island from February to July 2003 (means ± SE, n = 6).
Letters indicate significant differences between sampling selecting eelgrass for consumption based on shoot size.
periods (p < 0.05) Eelgrass terminal shoots are larger than lateral shoots
(Bak 1980), and if geese select shoots based on size,
Rhizome Rhizome Root terminal shoots would be consumed preferentially.
length weight weight Before Canada geese arrived at Fishing Island, the eel-
(cm m–2) (g m–2) (g m–2) grass at the site exhibited growth characteristics typi-
cal of a healthy meadow, with a higher proportion of
Feb 1757 ± 165a 15.4 ± 1.6a 8.6 ± 2.3a
Apr 1510 ± 272a 9.0 ± 1.1b 7.3 ± 1.1b lateral shoots than terminal shoots (Bak 1980). In Feb-
Jul 854 ± 175b 5.2 ± 1.3b 3.5 ± 1.1b ruary 2003, eelgrass terminal shoots were preferen-
tially grazed even though lateral shoot density was
higher. Canada geese selected the larger terminal
(Pinnerup 1980). The only detectable difference in shoots while they were still available, but by April
light reaching the meadow was an increase in light 2003 there were almost no terminal shoots left in the
that occurred after Canada geese had left the site. The meadow, and goose grazing activity on lateral shoots
goose grazing activity at Fishing Island from January increased accordingly. The overall changes in eelgrass
2003 to April 2003 caused a major decline in eelgrass parameters during the course of the grazing event
plant parameters, and eelgrass indicators remained (Table 2) suggest that Canada geese preferred to
low through July 2003. graze in areas where eelgrass was larger and more
Although the entire Fishing Island meadow was abundant.
greatly affected by the grazing event, the most heavily Several studies concerning waterfowl grazing activ-
grazed area of the meadow was Transect B, which lost ity on seagrass have reported ‘giving-up thresholds,’
over 680 g m–2 of eelgrass aboveground biomass by i.e. the seagrass biomass level (g m–2) at which water-
July 2003, followed by Transect C, which lost over fowl leave a meadow after depleting food resources
540 g m–2. These transects had the greatest initial eel- (Percival & Evans 1997, Clausen 2000). Most of the
grass biomass, and geese likely spent the greatest studies reporting giving-up thresholds concern brant
amount of time feeding at these areas. At a European grazing at European Zostera noltii meadows. For our
Zostera noltii meadow, brant spent the majority of their study, we used the April 2003 aboveground biomass as
time within 100 m of the low tide line, as shown in a an estimated giving-up threshold, since April was the
spatial depletion model developed by Percival et al. last month that Canada geese were observed at Fish-
(1996, 1998). At the Fishing Island meadow, Transects ing Island. The giving-up threshold for our study falls
B and C are closest to the low tide line, and Canada within the range of values listed for brant (Table 4).
geese may have similarly concentrated their feeding After Canada geese left the site, we continued
efforts at these locations. The greater initial biomass at monitoring eelgrass parameters through July 2003 to
Transects B and C, in combination with the increased determine the extent of eelgrass recovery after the
grazing time by Canada geese, would have caused the grazing event. Although aboveground biomass
278 Mar Ecol Prog Ser 333: 271–279, 2007
Table 4. Branta canadensis. Estimated giving-up thresholds (see ‘Discussion’) Island meadow during the winter–
for waterfowl grazing on seagrass meadows. Studies conducted at European spring of both 2004 and 2005 (Short et
sites concern brant Branta bernicla L. grazing on dwarf eelgrass Zostera noltii H.
The giving-up threshold for our study is the April 2003 aboveground biomass, as
al. 2006).
April was the last month in which Canada geese were sighted at Fishing Island
Acknowledgements. We thank B. Smith and
Location Giving-up threshold Source E. Robinson at the New Hampshire State
(g m–2) Fish and Game Department. Thanks to D.
Walker, T. Davis, J. Gaeckle, C. Ochieng,
Wadden Sea, Denmark 7.5 Madsen (1988) and N. Petit for assistance in the field and
Lindisfarne, UK 5.0 Percival & Evans (1997) laboratory. We thank C. Short for editing
Limfjorden, Denmark 1.7–5.9a Clausen (2000) and valuable suggestions on the manuscript.
Fishing Island, USA 6.7 Present study Thanks to 3 anonymous reviewers for help-
a
Biomass estimate based on a modeling approach ful comments. This work is Jackson Estuar-
ine Laboratory Contribution No. 438.
increased slightly from April to July 2003 at parts of LITERATURE CITED
the meadow, eelgrass at Fishing Island clearly did not
Bak HP (1980) Age populations and biometrics in eelgrass,
attain the biomass or percent cover levels seen in a Zostera marina L. Ophelia 19:155–162
typical summer, indicating that recovery was mini- Baldwin JR, Lovvorn, JR (1994) Expansion of seagrass habitat
mal. Few seeds or seedlings were found, and eelgrass by the exotic Zostera japonica, and its use by dabbling
shoot density did not change significantly from April ducks and brant in Boundary Bay, British Columbia. Mar
Ecol Prog Ser 103:119–127
to July 2003, suggesting that neither sexual propaga-
Buchsbaum R (1987) The geese and the grass. Sanctuary 27:
tion nor vegetative expansion of shoots contributed to 7–9
eelgrass recovery. At European Zostera noltii mead- Burdick DM, Short FT, Wolf J (1993) An index to assess and
ows where grazing by brant occurs every winter, Z. monitor the progression of wasting disease in eelgrass
noltii recovers to typical biomass levels in the sum- Zostera marina. Mar Ecol Prog Ser 94:83–90
Clausen P (2000) Modelling water level influence on habitat
mer (Vermaat & Verhagen 1996, Nacken & Reise choice and food availability for Zostera feeding brent
2000). The difference in recovery between that sys- geese Branta bernicla in non-tidal areas. Wildl Biol 6:
tem and Fishing Island appears to be a result of the 75–87
feeding behavior of the geese. Brant consume the Fox AD (1996) Zostera exploitation by brent geese and
wigeon on the Exe Estuary, southern England. Bird Study
leaves of Z. noltii, leaving the meristem intact (Ver-
43:257–268
maat & Verhagen 1996) to re-grow new leaves, Gaeckle JL, Short FT (2002) A plastochrone method for mea-
whereas Canada geese at Fishing Island typically suring leaf growth in eelgrass, Zostera marina L. Bull Mar
consumed the lower part of the shoot, including the Sci 71:1237–1246
sheath and the meristem. Once the meristem is Ganter B (2000) Seagrass (Zostera spp.) as food for brant
geese (Branta bernicla): an overview. Helgol Mar Res 54:
removed, eelgrass is unable to produce new leaves, 63–70
preventing vegetative growth of that shoot. Hanson AR (ed) (2004) Status and conservation of eelgrass
The Fishing Island case study represents one of the (Zostera marina) in Eastern Canada. Technical Report
few accounts of eelgrass decline due to Canada Series No. 412. Canadian Wildlife Service, Atlantic Region
Hughes AR, Stachowicz JJ (2004) Genetic diversity enhances
goose grazing in the Gulf of Maine region. Other
the resistance of a seagrass ecosystem to disturbance. Proc
studies in this region have reported how eelgrass Natl Acad Sci USA 101:8998–9002
availability can affect goose populations (Seymour et Jacobs RPWM, den Hartog C, Braster BF, Carriere FC (1981)
al. 2002, Hanson 2004). In contrast, our results show Grazing of the seagrass Zostera noltii by birds at Ter-
that Canada geese can have a substantial effect on schelling (Dutch Wadden Sea). Aquat Bot 10:241–259
Kenworthy WJ, Thayer GW (1984) Production and decompo-
eelgrass populations. In just 3 mo, a relatively small sition of the roots and rhizomes of seagrasses, Zostera
number of geese almost completely eliminated the marina and Thalassia testudinum, in temperate and sub-
long-standing Fishing Island eelgrass meadow. tropical marine ecosystems. Bull Mar Sci 35:364–379
Unlike many locations where seagrass recovers after Madsen J (1988) Autumn feeding ecology of herbivorous
wildfowl in the Danish Wadden Sea and the impact of food
a waterfowl grazing event, eelgrass at Fishing Island
supplies and shooting movements. Dan Rev Game Biol 13:
did not re-establish after Canada geese left the site. 2–33
The lack of eelgrass recovery will affect the ecology Moore JE, Colwell MA, Mathis RL, Black JM (2004) Staging
of Fishing Island. The small amount of eelgrass of Pacific flyway brant in relation to eelgrass abundance
biomass remaining at Fishing Island cannot provide and site isolation, with special consideration of Humboldt
Bay, California. Biol Conserv 115:475–486
the same functionality as the historical meadow. Fur- Nacken M, Reise K (2000) Effects of herbivorous birds on
ther SeagrassNet monitoring has shown a lack of eel- intertidal seagrass beds in the northern Wadden Sea.
grass recovery and the return of geese to the Fishing Helgol Mar Res 54:87–94
Rivers & Short: Eelgrass affected by goose grazing 279
Percival SM, Evans PR (1997) Brent Geese Branta bernicla Ocean Management. ASCE. July 11–14 Charleston, SC,
and Zostera: factors affecting the exploitation of a season- p 3689–3706
ally declining food resource. Ibis 139:121–128 Short FT, McKenzie LJ, Coles RG, Vidler KP (2002) Seagrass-
Percival SM, Sutherland WJ, Evans PR (1996) A spatial deple- Net Manual for Scientific Monitoring of Seagrass Habitat.
tion model of the responses of grazing wildfowl to the Queensland Department of Primary Industries, QFS,
availability of intertidal vegetation. J Appl Ecol 33:979–992 Cairns
Percival SM, Sutherland WJ, Evans PR (1998) Intertidal habi- Short FT, Koch E, Creed JC, Magalhaes KM, Fernandez E,
tat loss and wildfowl numbers: applications of a spatial Gaeckle J, Rivers DO (2006) SeagrassNet monitoring
depletion model. J Appl Ecol 35:57–63 across the Americas: case studies of seagrass change.
Pinnerup SP 1980. Leaf production of Zostera marina L. at PSZN I: Mar Ecol 27:277–289
different salinities. Ophelia 1:219–224 Tubbs CR, Tubbs JM (1982) Brent geese (Branta bernicla) and
Portig AA, Mathers RG, Montgomery WI, Govier RN (1994) The their food in the Solent, southern England. Biol Conserv
distribution and utilization of Zostera species in Strangford 23:33–54
Lough, Northern Ireland. Aquat Bot 47:317–328 Vermaat JE, Verhagen FCA (1996) Seasonal variation in the
Seymour NR, Miller AG, Garbary DJ (2002) Decline of intertidal seagrass Zostera noltii Hornem: coupling demo-
Canada geese (Branta canadensis) and common golden- graphic and physiological patterns. Aquat Bot 52:259–281
eye (Bucephala clangula) associated with a collapse of eel- Vogel L (1995) Wintering waterfowl on Great Bay. Technical
grass (Zostera marina) in a Nova Scotia estuary. Helgol Report No. 2 prepared for NOAA and the New Hampshire
Mar Res 56:198–202 Fish and Game Department, Concord. NH
Short FT (ed) (1992) The ecology of the Great Bay Estuary, Ward DH, Morton A, Tibbitts TL, Douglas DC, Carrera-
New Hampshire and Maine: an estuarine profile and González E (2003) Long-term change in eelgrass distribu-
bibliography. NOAA — Coastal Ocean Program Publ, tion at Bahía San Quintín, Baja California, Mexico, using
Durham, NH satellite imagery. Estuaries 26:1529–1539
Short FT, Burdick DM (1996) Quantifying eelgrass habitat Wilson UW, Atkinson JB (1995) Black brant winter and
loss in relation to housing development and nitrogen load- spring-staging use at two Washington coastal areas in
ing in Waquoit Bay, Massachusetts. Estuaries 19:730–739 relation to eelgrass abundance. Condor 97:91–98
Short FT, Wolf J, Jones GE (1989) Sustaining eelgrass to man- Zar JH (1999) Biostatistical analysis, 4th edn. Prentice Hall,
age a healthy ecosystem. Proc 6th Symp Coastal and Upper Saddle River, NJ
Editorial responsibility: Kenneth Heck (Contributing Editor), Submitted: November 7, 2005; Accepted: August 15, 2006
Dauphin Island, Alabama, USA Proofs received from author(s): March 1, 2006