Pihl et al 06
Estuarine, Coastal and Shelf Science 67 (2006) 123e132
www.elsevier.com/locate/ecss
Shift in fish assemblage structure due to loss of seagrass
Zostera marina habitats in Sweden
Leif Pihl a,*, Susanne Baden a, Nils Kautsky b, Patrik Ronnback b, Tore Soderqvist c,
¨ ¨ ¨
c a
˚
Max.Troell , Hakan Wennhage
a
Department of Marine Ecology, Goteborg University, Kristineberg Marine Research Station, 450 34 Fiskebackskil, Sweden
¨ ¨
b
Department of System Ecology, Stockholm University, 106 91 Stockholm, Sweden
c
Beijer International Institute of Ecological Economics, The Royal Swedish Academy of Science, 10405 Stockholm, Sweden
Received 6 June 2005; accepted 14 October 2005
Available online 19 January 2006
Abstract
The areal extent of Zostera marina in the archipelago of the Swedish Skagerrak has decreased by 60% over two decades. To investigate the
effects of Z. marina loss on the local fish assemblages, the fish fauna was compared between existing seagrass beds and sites where seagrass had
vanished. A field study was carried out at four shallow locations in the outer archipelago of the coast in June 2004. Within each location two sites
were sampled, one with an existing Z. marina bed and another where Z. marina had disappeared. Fish were sampled semi-quantitatively with
a beach seine. Samples were taken during both day and night and captured fish were examined to species, enumerated and measured in the field,
and released thereafter. The number of fish species was found to be significantly higher in Z. marina habitats compared to areas where seagrass
was missing, and density and biomass of fish were generally lower in areas dominated by bare sediment compared to those in the seagrass hab-
itats. Several species and groups of fishes (i.e., gadoids, labrids, syngnathids) were absent or occurred in low densities at sites where Z. marina
was missing. For example, juvenile 0-group cod density was reduced by 96% at sites where Z. marina had disappeared. Such a reduction in
recruitment of cod is in the same order of magnitude as the combined effect of seal predation and mortality due to by-catches in the eel
fyke-net fishery estimated for the archipelago of the Swedish Skagerrak. Hence, the results clearly indicate a shift in the fish assemblage, in-
cluding a loss of taxa at the family level as a result of degradation in habitat-forming vegetation.
Ó 2005 Elsevier Ltd. All rights reserved.
Keywords: habitat loss; eutrophication; cod; plaice; Skagerrak
1. Introduction are considered to be higher compared to alternative habitats
(Beck et al., 2001; Deegan et al., 2002).
Seagrass meadows provide habitat diversity in the coastal The global loss of seagrasses from the mid-1980s to the
seascape worldwide. This habitat-forming vegetation creates mid-1990s has been estimated to be 12,000 km2 (Short and
a three dimensional architecture over soft bottoms, which sta- Wyllie-Echeverria, 1996), which correspond to an overall re-
bilizes the sediment and reduces water movements. Seagrass duction by about 7% of the total areal extent (Spalding
meadows are known to harbour a diverse and abundant fauna et al., 2003). This fact has contributed to the listing of seagrass
of invertebrates and are generally considered as essential hab- beds as habitats worthy of protection in the Rio-declaration
itats for many fish species (Orth et al., 1984; Jenkins et al., (1992/93:13). The consensus from the report ‘‘World Atlas
1997). Fish may spawn in seagrass beds or use the area as of Seagrasses’’ is that the main reasons for the decline, apart
a nursery ground, where growth and survival of juvenile fish from natural threats such as storms and diseases, are anthropo-
genic (Green and Short, 2003). Most serious are the indirect
effects of human activity. Increased turbidity and overgrowth
* Corresponding author. by epiphytic algae reduce light penetration, resulting in subse-
E-mail address: leif.pihl@kmf.gu.se (L. Pihl). quent loss of seagrass. Turbidity and surplus algal growth may
0272-7714/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ecss.2005.10.016
124 L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132
derive from physical disturbances on land or in the water, and however, hard to evaluate and there may be alternative explan-
from eutrophication combined with trophic cascades (e.g. ef- ations for the changes observed (Orth et al., 1996; Browder
fects of over-fishing) in the coastal ecosystem (Howarth et al., 1999). To our knowledge there is only one study com-
et al., 2000; Jackson et al., 2001; Hughes et al., 2004). Direct paring areas with extant and lost seagrass habitats (Hughes
disturbances such as dredging and benthic trawling in seagrass et al., 2002). The study, performed in estuaries along the
will most likely cause drastic local effects, but are probably of NW Atlantic Coast, reported that the abundance, biomass
less importance on a larger scale compared to more diffuse, in- and species richness of fish were lower in areas where seagrass
direct anthropogenic affects (Hily et al., 2003). beds had disappeared, but the generality of these findings for
Zostera marina is the dominant seagrass species in Swedish other coastal areas are not known.
coastal waters. Along the SkagerrakeKattegat coast, it occurs The aim of this study was to investigate the effects of losses
in semi-exposed and protected areas within water depths of of Zostera marina on the local fish assemblages on the Swedish
0.5e6 m (Baden and Bostrom, 2001). These Zostera beds
¨ west coast. This was carried out by comparing the fish fauna in
have been shown to support a high production of benthic fauna existing seagrass beds with sites where seagrass has vanished
as well as epibenthic invertebrates and fish (Baden and Pihl, over the last two decades. The main purpose was to document
1984; Moller et al., 1985), and to serve as nursery and feeding
¨ shifts in fish assemblages through measurements of species
grounds for more than 40 fish species (Pihl and Wennhage, numbers, densities and biomass. By comparing the utilization
2002; Wennhage and Pihl, 2002). Over the last two decades, by fish of Z. marina beds and alternative habitats, changes in
a 60% reduction in distribution of the Z. marina has been ob- the ecological function of coastal areas with loss of seagrass
served in the Swedish Skagerrak archipelago (Baden et al., could be evaluated.
2003). Along most sections of the coast both the upper and
the lower depth distributions of seagrass have been reduced,
resulting in a narrowing of meadows, but in some areas sea- 2. Methods
grass meadows have disappeared completely. The lost Z. ma-
rina is commonly replaced by a bare sediment bottom, but This field study was carried out in the outer archipelago of
in some areas the sediment is partly covered by filamentous the Swedish Skagerrak coast (58 14e220 N; 11 23e320 E)
green algae or patches of Fucus spp., attached to shells and (Fig. 1) between 8 and 20 June 2004. The archipelago consists
stones. The reason for the degradation of the seagrass habitat of islands of varying size, and a shoreline characterized by
in the Skagerrak is not known, but coastal eutrophication and/ a mixture of rocky and soft-bottom substrata. On soft bottoms
or over-fishing have been suggested as plausible causes. In ad- Zostera marina is the dominating vegetation within the depth
dition, altered water exchange due to construction of road range of 1e5 m (Baden et al., 2003). This coastal region is
banks and leisure boat harbours could have reduced the distri- micro-tidal with a tidal amplitude of around 0.2 m. Mean sur-
bution of seagrass in the coastal Skagerrak. face water temperatures usually range from 5 to 15 C in
Historical distribution maps of seagrass from Scandinavia spring and autumn and from 15 to 20 C during the summer
are few, but extensive Danish investigations dating back to (Pihl and Rosenberg, 1982). Surface water salinity typically
1900 reveal that only about 25% of the former areal extension fluctuates between 20 and 25 psu in the summer.
remained in 1990 (Petersen, 1914). This large areal reduction June was selected for sampling in this investigation because
is partly attributed to losses of deep eelgrass stands as a conse- previous studies on seasonal dynamic of the fish community
quence of impoverished light conditions and partly to the slow had shown that the highest species richness, abundance and
recovery after the seagrass disease in the 1930s. Between 1900 biomass occurred at that time of the year (Pihl and Wennhage,
and 1990, maximum colonisation depths decreased from 2002). In June most of the fish are recruited to the coastal hab-
5e6 m in estuaries and 7e8 m in open waters, to 2e3 m itats, giving a full representation of age-classes in the fish
and 4e5 m, respectively (Bostrom et al., 2003).
¨ community. The investigated coastal region represents one
The change in habitat structure following the loss of of five areas where Zostera marina has been observed to de-
a Zostera marina bed is likely to shift the local system into crease significantly in its distribution over the last two decades
an alternative state. Primarily, the loss of the habitat-forming (Baden et al., 2003). Within this region four locations were
species Z. marina will alter habitat complexity, changing the chosen at random and two sites were selected in each: one
structure of associated fauna assemblages. In a review of the with an existing Z. marina bed and another where Z. marina
extensive literature testing the importance of seagrass mead- had disappeared (Fig. 1). Sites with existing seagrass were
ows as nursery areas for juvenile fish and invertebrates, chosen in close vicinity (about 500 m) to the sites without sea-
Heck et al. (2003) found that their abundance, growth and sur- grass. At the seagrass sites each Z. marina bed had a spatial
vival were generally higher in seagrass compared to unstruc- distribution of more than 10 ha, covering 60e100% of the
tured habitats, but similar to other structured habitat, bottom area within the depth range of 1e4 m. At sites where
indicating that habitat complexity may be more important to Z. marina had disappeared compared to the 1982 distribution,
fish diversity than the type of structures present. There are the bottom sediment was generally free of vegetation, except
also a few studies more specifically reporting changes in spe- from sparse occurrence of Z. marina shoots and small patches
cies composition following large-scale losses of seagrass beds. of Fucus spp. stands. Vegetation cover at these sites was
Historical comparisons without appropriate controls are, 1e15% of the bottom area.
L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132 125
Fig. 1. Map of the investigated area in the archipelago of the Swedish Skagerrak. Sampling locations indicated by open circles.
A sampling area of 5 ha was designated in the centre of haul during the night (24e01 hours) in each of the investigated
each selected site. Fish were sampled semi-quantitatively sites. Sampling order was randomly allocated between sites
with a beach seine according to Tveite (1984), a method and over time to avoid introduction of systematic errors.
most commonly used to study fish communities in the littoral Day and night samples were taken at random in the designated
zone. The method has the advantage of allowing for estimates area, and !50 m apart within a site. Captured fish were exam-
of the area being sampled; a feature not shared by some other ined to species, enumerated and measured (total length, mm)
methods in use (e.g. gillnets and fyke-nets). Methodological in the field, and released thereafter. Some fish from three large
studies have shown that benthic species and small-size individ- samples collected at night were brought to the laboratory for
uals may hide within substrata having a high complexity or es- further analysis. Estimates of the biomass (wet wt.) for all in-
cape underneath the foot-rope of the seine (Parsley et al., dividual fishes were derived from established lengtheweight
1989). Beach seines may also have selectivity towards relationships. In addition to fish, macro-crustaceans (mainly
small-sized species, at least in comparison to other methods shrimps and crabs) were also collected. Numbers and pooled
(Pierce et al., 1990; Weaver et al., 1993). However, the beach biomass (wet wt.) for each species of invertebrates were re-
seine maintains its performance better than visual sensing corded for each sample in the laboratory.
techniques when macrophyte cover increases (Brind’Amour Shoot density, mean and maximum length and biomass of
and Boisclair, 2004). The seine was 40 m long, 3 m high, Zostera marina were estimated in the four seagrass beds.
had a mesh opening of 10 mm in the arms and 5 mm in the Three quantitative samples were taken by a diver using
‘‘cod end’’ and was towed by 30 m long ropes. The gear a net-bag (mesh opening 1 mm) connected to a bottom metal
was deployed from a small boat in a rectangular shape with ring (diameter of 35 cm). Samples were allocated randomly in
its deepest part at around 3 m depth. It was pulled shoreward the centre of the Z. marina bed at approximately 2 m depth.
33 m by four people 15 m apart until reaching a water depth of Percentage cover of Z. marina was visually assessed by a diver,
1 m, giving an effective fishing area of approximately 500 m2. swimming over the investigated area. At sites where Z. marina
One haul was taken during daytime (12e13 hours) and one had disappeared, the vegetation cover was assessed visually
126 L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132
from the boat before sampling. To characterize the sediment Table 1
substrate three quantitative sediment cores were collected ran- Shoot density, blade length and biomass (wet wt.) of Zostera marina, and cov-
er of Z. marina and Fucus spp., as well as sediment organic content (%) in the
domly by a diver in each of the study sites (1e3 m water eight sample sites
depth) for analysis of organic content. Samples of the upper
Brofjorden Lindholmen Finnsbo ˚ ¨
Gaso
2 cm of the sediment core were dried at 60 C for 24 h and
combusted for 4 h at 450 C, and organic content was mea- Zostera marina sites
Cover of Zostera (%) 100 60 100 100
sured by weight loss.
Fish assemblage structure was compared between samples Number of shoots
Mean (g dw mÿ2) (n ¼ 3) 230 100 380 250
by using BrayeCurtis similarity indices, as described in the
SE 36 21 21 50
PRIMER package (Field et al., 1982). Fish abundance data
were log(X þ 1) transformed to weigh the relative numerical Blade length
Mean (cm) (n ¼ 3) 47.2 30.1 22.1 39.8
importance of common and rare species in the analysis. SE 4.6 2.3 1 3.9
BrayeCurtis similarity indices were computed and the result- Max (cm) 104 67 39 66
ing similarity matrix was used to perform non-metric Multi-
Blade biomass
Dimensional Scaling (MDS). An analysis of similarities Mean (g dw mÿ2) (n ¼ 3) 172.2 43.5 71.2 141.2
(ANOSIM) was used to test for differences in assemblage SE 13.8 21.5 11.3 31.4
structure between the two habitat types, and to compare day Sediment org. content
and night samplings (method in Clarke, 1993). The propor- Mean (%) (n ¼ 3) 16.6 14.5 1.7 13.6
tional contribution of different fish species to the dissimilarity SE 1.3 0.6 0.1 0.8
between groups was investigated using SIMPER (method in
Clarke and Warwick, 1994). Fish abundance, biomass and Non-Zostera marina sites
Cover of Zostera (%) 0 0 1 0
number of species in the two habitat types were compared us-
ing two-way ANOVAs, with habitat and time of day as factors. Cover of Fucus spp. (%) 5 10e15 0 5e10
Non-transformed data used as variances were shown to be ho- Sediment org. content
mogenous according to Cochran’s test. Mean (%) (n ¼ 3) 5.4 3.3 1 1.3
SE 0.5 0.4 0.06 0.06
3. Results
˚ ¨
two areas, one was considered as semi-exposed (Gaso) and
3.1. Vegetation one had a high physical exposure (Finnsbo). In the protected
and semi-exposed areas, content of organic matter in Zostera
In three out of four Zostera marina beds investigated the marina beds varied between 13.6 and 16.6%, whereas the ex-
vegetation had a homogenous density with full cover of the sed- posed Z. marina bed had a sandy sediment with only 1.7% or-
iment. Only at one site, Lindholmen (location 2), the Z. marina ganic content (Table 1). At the sampled sites mainly free of
had a patchy distribution covering approximately 60% of the vegetation content of organic matter in the sediment was esti-
sediment. Mean shoot density was estimated to be 100e mated to be between 1.0 and 5.4%, with the highest value in
380 shoots mÿ2 in the four Z. marina beds, the lowest densities the protected areas (Table 1). Thus, organic content in the sed-
occurring at the site where the bed had a patchy distribution iment was 2e10 times higher in Z. marina beds compared to
(Table 1). The mean blade length varied from 22 to 47 cm at unvegetated sites, but inter-location differences could be im-
the four sites, and maximum length was about double the portant due to variation in exposure.
mean length at all sites (Table 1). Blade biomass of Z. marina
was found to be between 44 and 172 g dw mÿ2 at the study sites,
3.3. Fish
with the lowest values in the beds having either low mean blade
length or a patchy distribution. At the four sites where Z. marina
Altogether, 33 fish species belonging to 15 families were
beds had disappeared over the last two decades the sediment
identified in this investigation (Table 2). Twenty-eight species
was mainly free of vegetation. At one site (the exposed area
were found in the Zostera marina beds, of which nine were ex-
Finsbo, location 3) a few remaining shoots of Z. marina were
clusive to this habitat. At the sites where Z. marina had disap-
found covering around 1% of the bottom (Table 1). The other
peared 19 fish species were found, and five of these species
three sites had patches of Fucus spp. growing on stones and
were only found here. When comparing Z. marina and non-
shells of blue mussels with an approximate cover of between
seagrass sites in pairs for each location, the number of fish spe-
5 and 15% of the bottom sediment.
cies was in all cases higher in the Zostera habitat (Fig. 2).
Overall the number of fish species was significantly higher
3.2. Sediment ( p < 0.005) in Z. marina beds than at non-seagrass sites, but
no difference was observed between day and night samplings
Two of the investigated locations (Brofjorden and Lindhol- (Table 3). The mean number of individuals sampleÿ1 exhibited
men) were situated in an enclosed part of the archipelago and a large variation between locations and sites (Fig. 2). Generally,
were protected from exposure to wind and waves. Of the other the numbers of individuals were higher in catches from
L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132 127
Table 2
Number of individuals and biomasses (wet wt.) of fishes recorded in beach-seine samples. Pooled data of day and night samples from four Zostera marina sites, and
from four sites where Z. marina has disappeared
Family Fish species Numbers Biomass
Zostera Non-Zostera Zostera Non-Zostera
marina marina marina marina
Anguillidae Anguilla anguilla 12 23 456.4 1436.7
Gobiidae Aphia minuta 1423 169 4705.5 261.0
Callionymidae Callionymus lyra 0 1 0.0 11.5
Clupeidae Clupea harengus 1 84 2.6 157.4
Labridae Ctenolabrus rupestris 272 2 2062.8 17.7
Syngnathidae Entelurus aequoreus 16 0 70.1 0.0
Gadidae Gadus morhua 146 12 947.1 1142.7
Gasterosteidae Gasterosteus aculeatus 1164 916 1781.7 1527.6
Gobiidae Gobius niger 738 304 4605.1 924.1
Gobiidae Gobiusculus flavescens 264 0 137.3 0.0
Pleuronectidae Limanda limanda 5 0 228.3 0.0
Gadidae Merlangius merlangus 16 0 482.1 0.0
Cottidae Myoxocephalus scorpius 22 2 868.9 81.6
Syngnathidae Nerophis lumbriciformis 1 1 2.6 5.5
Syngnathidae Nerophis ophidion 32 4 25.1 2.3
Pholidae Pholis gunnellus 3 0 42.5 0.0
Pleuronectidae Platichthys flesus 11 37 2206.6 1835.4
Pleuronectidae Pleuronectes platessa 21 143 74.8 310.8
Gadidae Pollachius virens 4 0 11.0 0.0
Gobiidae Pomatoschistus microps 0 90 0.0 85.1
Gobiidae Pomatoschistus minutus 30 72 75.6 178.6
Gobiidae Pomatoschistus pictus 16 18 23.6 23.5
Salmonidae Salmo trutta 9 4 785.4 777.8
Bothidae Scophthalmus rhombus 0 1 0.0 0.2
Soleidae Solea solea 0 1 0.0 48.8
Gasterosteidae Spinachia spinachia 2 0 8.4 0.0
Labridae Symphodus melops 4 0 62.8 0.0
Syngnathidae Syngnathus acus 35 0 179.2 0.0
Syngnathidae Syngnathus rostellatus 33 23 38.0 27.3
Syngnathidae Syngnathus typhle 129 37 135.2 52.6
Cottidae Taurulus bubalis 7 7 204.8 120.6
Gadidae Trisopterus esmarkii 5 1 6.0 0.5
Zoarcidae Zoarces viviparus 82 57 1648.0 474.5
4503 2009 21,877.5 9503.7
Z. marina beds compared to catches from areas where seagrass between day and night samplings. Furthermore, of the two
had disappeared when sites were compared in pairs. However, habitat types, Z. marina sites were more closely clustered in
overall no significant difference in fish density could be de- the analysis than bare sediment sites, indicating a higher sim-
tected between habitats or between day and night samplings ilarity of the fish assemblages in Z. marina beds.
(Table 3). Except for one sample in the Z. marina bed at loca- An SIMPER-analysis revealed that the distribution of 10
˚ ¨
tion 4 (Gaso), the fish biomass was generally low during day- species explained about 70% of dissimilarity between the two
time sampling, especially at non-seagrass sites (Fig. 2). Night habitat types (Table 4). Of these species, eight had higher den-
fish biomass was generally higher than daytime biomass, and sities in Zostera marina beds and two were more abundant in
similar total fish weights were recorded in the two habitat non-seagrass habitats. The high affinity of several fish species
types. Overall, a trend toward higher fish biomass was re- to the Zostera habitat is further emphasised by the fact that
corded in Z. marina beds, although the difference from non- 10 out of the 20 most abundant fish species were almost exclu-
seagrass sites was not significant ( p ¼ 0.10; Table 3). sively caught in the seagrass beds. Gadoids (Gadus morhua,
MDS ordination based on a BrayeCurtis similarity matrix Merlangius merlangus and Pollachius virens) labrids (Cteno-
showed that the fish assemblages were mainly structured ac- labrus rupestris and Symphodus melops), syngnathids (Syngna-
cording to habitat type, whereas time of the day was of less thus acus and Entelurus aequoreus) and Gobiusculus flavescens
importance for the structure of the fish assemblages (Fig. 3). were predominantly found in Z. marina beds, whereas flatfishes
An ANOSIM-test revealed a significant difference (Global such as Pleuronectes platessa, Platichthys flesus and Solea
R ¼ 0.37; p ¼ 0.01) between the fish assemblage structure in solea mainly occurred on bottoms dominated by bare sediment.
Zostera marina beds and non-seagrass habitat, but the test Cod (Gadus morhua) and plaice (Pleuronectes platessa) are
failed to show any difference (Global R ¼ ÿ0.07; p ¼ 0.74) the most important species of commercial interest that
128 L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132
25 Day
20
15
Species per sample
10
5
0
25 Night
20
15
10
5
0
1000 Day
600
Numbers per sample
200
0
1400 Night
1000
600
200
0
8000 Day Zostera
6000 Non-Zostera
Biomass per sample
4000
2000
0
8000 Night
6000
4000
2000
0
1 2 3 4
Sites
Fig. 2. Number of species, individuals and biomass of fish captured during day and night samplings at four Zostera marina and four non-Z. marina sites in the
archipelago of the Swedish Skagerrak.
occurred in high densities in this investigation. They both uti- 1993). From analysis of length distribution it was obvious
lize the coastal zone as a nursery and juveniles may stay in that 0-group juvenile cod mainly utilized the Zostera marina
shallow (<10 m) waters for about two years after settlement beds as a nursery, whereas 1-group cod were equally repre-
to the benthic habitat (Pihl, 1989; Pihl and Ulmestrand, sented in both habitats (Fig. 4). In the Z. marina beds cod
Table 3
Two-fixed-factor ANOVA-modes. Number of species, density and biomass of
fish as a function of habitat (Zostera, no Zostera) and time (day, night) Stress: 0.09
2
Source of variation SS df MS F p
2 1
Number of fish species
4 4
Habitat 203.1 1 203.1 21.8 0.005 2 2
3
Time 39.1 1 39.1 4.2 0.063
Habitat  Time 1.6 1 1.6 0.2 0.692 4
1
Residual 111.8 12 9.3 1
4
1
Density of fish
Habitat 388,752 1 388,752 2.58 0.134
Time 133,225 1 133,225 0.88 0.365 3
Habitat  Time 49,952 1 49,952 0.33 0.575
Residual 1,805,974 12 150,497 Day Night
Non-Zostera
Biomass of fish Zostera 3 3
Habitat 9,571,289 1 9,571,289 3.91 0.099
Time 2,051,340 1 2,051,340 0.68 0.424
Fig. 3. Similarities in fish assemblage structure between day and night
Habitat  Time 2,133,790 1 2,133,790 0.71 0.415
samplings in four Zostera marina and four non-Z. marina sites, based on
Residual 3.6Eþ07 12 2,996,293
Multi-Dimensional Scaling (MDS) ordination.
L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132 129
Table 4 and six species were found in Zostera marina habitats
Fish species most responsible for the difference in assemblage structure be- (Table 6). Carcinus maenas, Crangon crangon, Palaemon
tween Zostera marina and non-Zostera marina habitats, listed in the order
of their contribution to the average BrayeCurtis dissimilarity. Abundance is
adspersus and Palaemon elegans were the dominant species
mean individuals sampleÿ1 in both habitats. There was a trend towards higher densities
Rank Fish species Abundance % Contribution
and biomasses in non-seagrass compared to Z. marina sites,
however, the differences were not significant ( p > 0.05).
Zostera Non-Zostera
Catches were higher during the night, with significant larger
1 Aphia minuta 177.8 21.1 11.1 biomass ( p ¼ 0.005) at night compared to day samples in
2 Gasterosteus aculeatus 145.5 114.5 8.4
3 Gobiusculus flavescens 33.0 0 8.2
both habitats.
4 Gadus morhua 18.3 1.5 5.9
5 Gobius niger 92.3 38.0 5.4 4. Discussion
6 Syngnathus typhle 16.1 4.6 5.4
7 Zoarces viviparus 10.3 7.1 5.3 The main purpose of this study was to document shifts in
8 Ctenolabrus rupestris 34.0 0.3 5.0
9 Pleuronectes platessa 2.6 17.9 4.5
the assemblages of fish as a direct consequence of loss of
10 Pomatoschistus minutus 3.8 9.0 4.4 the habitat-forming vegetation, Zostera marina, in shallow
11 Nerophis ophidion 4.0 0.5 3.3 soft-bottom areas. It would be expected that species richness
12 Syngnathus rostellatus 4.1 2.9 3.2 and composition of fish species would change when vegetation
13 Pomatoschistus microps 0 11.3 3.1 disappears as a result of lower habitat complexity (Jackson
14 Syngnathus acus 4.4 0 2.7
15 Pomatoschistus pictus 2.0 2.3 2.6
et al., 2001; Hughes et al., 2002; Lazzari, 2002), but the den-
16 Anguilla anguilla 1.5 2.9 2.5 sity of fish does not necessarily decrease, since shallow soft
17 Platichthys flesus 1.4 4.6 2.5 bottoms are known to host large abundances of small fishes
18 Myoxocephalus scorpius 2.8 0.3 2.5 (Edgar and Shaw, 1995). The archipelago of the area investi-
19 Merlangius merlangus 2.0 0 2.1 gated consists of a mosaic of rocky- and soft-bottom habitats
20 Entelurus aequoreus 2.0 0 2.0
21 Taurulus bubalis 0.9 0.9 1.7
that are to a varying degree covered by vegetation, thereby of-
22 Salmo trutta 1.1 0.5 1.7 fering a suite of alternative habitats with different complexity
23 Clupea harengus 0.1 10.5 1.3 that could be utilized by littoral fish. When seagrass disappears
24 Trisopterus esmarkii 0.6 0.1 0.9 from an area, fish could either concentrate in alternative veg-
25 Pollachius virens 0.5 0 0.7 etated habitats or stay in the altered habitat dominated by
26 Limanda limanda 0.6 0 0.7
27 Symphodus melops 0.5 0 0.7
bare sediment. Therefore, as pointed out in previous studies
28 Nerophis lumbriciformis 0.1 0.1 0.6 comparing fish in seagrass and bare sediment (Ferrell and
29 Pholis gunnellus 0.4 0 0.6 Bell, 1991), it is important to consider the size of the area
30 Spinachia spinachia 0.3 0 0.5 where seagrasses have been lost and the distance to other alter-
31 Solea solea 0 0.1 0.3 native habitats. In our study, the areas where seagrass had dis-
32 Callionymus lyra 0 0.1 0.3
33 Scophthalmus rhombus 0 0.1 0.3
appeared had a size of several hectares and the distance to
vegetated habitats, in this case other seagrass beds or belts
of macroalgae, was between 200 and 500 m. Despite the close
were caught both during day and night, but in the non-seagrass proximity to complex habitats, a significant reduction in fish
habitats cod only appeared in night samples. In contrast to cod, species and change in species structure were observed in areas
juvenile 0-group plaice were almost exclusively caught at sites where seagrass had vanished. Thus, there is a clear indication
dominated by bare sediment (Fig. 5). They occurred on bare of a shift in the fish assemblage, including a loss of taxa at the
sediment both during day and night. Only a few individuals family level as a result of degradation in habitat-forming veg-
of 1-group plaice were captured in this study. They appeared etation at the observed scale.
in both habitats, but were only caught during night sampling. In our study, fish biomass was generally higher during night
In an attempt to analyze possible relationships between the in habitats both with and without Zostera marina. The high
structure of Zostera marina and the fish assemblages (although biomass was partly due to the larger size of individual fish cap-
based on a small sample size) density, length and biomass of tured during the night. This indicated that fish migrate into
vegetation were compared to species, density and biomass of shallow water at night, and these areas may function as
fish (Table 5). Highest number of fish species and biomass a night-time feeding ground for both types of habitat. Such
was observed in the two Z. marina beds having the greatest nocturnal shoreward migration has previously been described
blade length and the largest biomass of vegetation. These for the two dominating commercial species in the area, cod
two beds also had the largest geographical extension of the and plaice (Pihl, 1982; Gibson et al., 1998). In other investiga-
four investigated Z. marina sites. tions it has been shown that fish migrate from seagrass into
open sediment habitats at night for foraging (Gotceitas et al.,
3.4. Macro-crustaceans 1997).
Different fish species may show different degrees of depen-
In addition to fish, nine species of macro-crustaceans were dency on vegetation. Syngnathids is a group of fish species
found in samples from the sites dominated by bare sediment that are adapted to seagrass by their body shape. The habitat
130 L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132
8
Zostera 0-group 1-group
7
6
5
4
Individuals per length group
3
2
1
0
8
Non-Zostera
7
6
5
4
3
2
1
0
30 40 50 60 70 80 150 200 250 300
Length, mm Length, mm
Fig. 4. Number of individuals per length-class of 0- and 1-group juvenile cod (Gadus morhua) at Zostera marina and non-Z. marina sites.
dependence may vary for this group of fishes, but most species fish (Arntz, 1973). In this way, gobies provide an important
are found among vegetation. Six species of syngnathids were energy link from the highly productive shallow coastal system
found in this study, of which four species occurred in both to fish living in deeper less productive water (Moller et al.,
¨
habitat types and two species were caught exclusively in Zos- 1985). In our study, five species of gobies together contributed
tera marina beds. Altogether, over 80% of the individuals and 30% of the numbers and 15% of the biomass of the total fish
biomass of this group of fish were found in the Z. marina. The assemblage at sites dominated by bare sediment. In Zostera
reason that cryptic species like the syngnathids occupy the marina beds, the corresponding figures for gobies were 55
bare sediment habitat is probably because the missing Zostera and 45%, respectively. The total density of gobies was four
beds had partly been replaced by Fucus spp. that could give times higher in Z. marina beds compared to non-seagrass sites,
sufficient camouflage for these fishes. and biomass was more than six times higher. Thus, Z. marina
Gobies are important components in the food web of littoral beds seem to have a considerably higher capacity for produc-
fish assemblages, occurring in vegetated as well as unvege- tion of gobies that provides an essential energy transfer link to
tated habitats. They are typically small-sized fish with a short fish during the winter season of low productivity.
life span and high production (Fonds, 1973). In temperate Some fish species may vary in their utilization of habitats
waters, gobies utilize shallow water for growth during the over different time scales. The affinity to vegetation may
summer, but usually migrate to deeper water in wintertime change during ontogeny, as for example, early stages of juve-
where they comprise an important food resource for demersal nile cod are more habitat-specific and remain stationary in
vegetation compared to older juveniles (Borg et al., 1997).
Cod uses Z. marina beds as a nursery ground and immigrate
35
30 to these coastal habitats by larval transport. Juveniles settle
25 Non-Zostera in the seagrass by active selection usually avoiding open
20 bare sediment, and consequently the availability of seagrass
Individuals per length-group
15
10
beds may be considered as a bottleneck in the recruitment pro-
5 cess due to the specific habitat requirement of the early benthic
0
35 Table 5
30
Zostera
Density (individual mÿ2), length (cm) and biomass (g wet wt. mÿ2) of Zostera
25
marina blades (n ¼ 3), and number of species, density (individuals sampleÿ1)
20
15
and biomass (g wet wt. sampleÿ1) of fish (n ¼ 2) at four Z. marina sites
10 Zostera marina Zostera Fish
5 sites
0 Density Length Biomass Species Density Biomass
20 30 40 50 60 70 80 90 100 110 120 130
Brofjorden 230 47 172 21 463 1937
Length, mm Lindholmen 100 30 44 20 785 1086
Finnsbo 380 22 71 20 207 1133
Fig. 5. Number of individuals per length-class of juvenile plaice (Pleuronectes
˚ ¨
Gaso 250 40 141 24 797 2260
platessa) at Zostera marina and non-Z. marina sites.
L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132 131
Table 6 This investigation included areas with different characteris-
Density (individuals sampleÿ1) and biomass (g wet wt. sampleÿ1) of macro- tics in terms of exposure, sediment type and vegetation struc-
crustaceans at Zostera marina and non-Zostera marina sites during day and
night samplings
ture. However, organic content of the sediment, as well as
Zostera shoot density, blade length and biomass, of the studied
Sites Zostera marina Non-Zostera marina
Zostera marina beds in 2004 were within the range of what has
Day Night Day Night previously (1982e1990) been reported for the Swedish coastal
Density region (Baden and Pihl, 1984; Baden and Bostrom, 2001).
¨
Athanas nitescens 0.8 1.3 Thus, there are no indications that the characteristics of the ex-
Carcinus maenas 11.3 24.0 29.8 26.8
Crangon crangon 11.3 33.3 53.3 171.8
tant Z. marina meadows have changed over the last two dec-
Gammaridae 0.0 0.0 1.5 1.3 ades. The number of species and density of fish recorded in
Macropodia rostrata 20.0 12.8 0.3 2.5 this study are also in accordance with what have previously
Mysidae 0.3 0.0 9.3 1.5 been found in surveys of Z. marina beds on the Swedish Ska-
Palaemon adspersus 24.0 134.3 155.8 236.0 gerrak coast. Investigations carried out during June between
Palaemon elegans 17.0 75.8 113.3 86.3
Pagurus bernhardus 0.0 0.0 0.5 0.0
2000 and 2004, including 18 Z. marina beds from three costal
regions of the Swedish west coast, estimated number of spe-
Total density 83.8 280.0 364.3 527.3
cies and densities of fish per standard haul to 14.3 and 643, re-
Biomass 0.0 0.0 0.0 spectively (Anders Svenson, personal communication). The
Athanas nitescens 0.0 0.0 0.1 0.3 corresponding figures for number of species and density of
Carcinus maenas 62.0 339.5 120.3 254.0 fish in the present study was 16.2 and 389, respectively. There-
Crangon crangon 6.3 17.8 22.8 76.5 fore, the result from this study concerning both vegetation and
Gammaridae 0.0 0.0 0.1 0.1
fish could be considered representative for Z. marina beds in
Macropodia rostrata 14.8 7.8 0.1 2.8
Mysidae 0.0 0.0 0.7 0.3 the archipelago, and the findings would be expected to be gen-
Palaemon adspersus 12.7 64.5 57.4 153.6 erally applicable for the Swedish west coast.
Palaemon elegans 6.0 45.1 45.3 49.5
Pagurus bernhardus 0.0 0.0 1.0 0.0
Total biomass 101.8 474.6 247.6 536.9 Acknowledgments
This project is a part of the research program MARBIPP
stages. Other gadoids, such as Merlangius merlangus and Pol- (Marine Biodiversity: Patterns and Processes) funded by the
lachius virens, may also use vegetated coastal habitats as nurs- Swedish Environmental Protection Agency, which are kindly
ery grounds, and in our study juvenile of these two species acknowledged. We also thank Andreas Wikstrom for valuable
¨
were exclusively found in Z. marina beds. Thus, Z. marina assistance during field sampling.
beds are essential habitats during the recruitment process for
gadoids, and losses of seagrass will most likely reduce the
nursery function of the coastal zone for these commercial im- References
portant species.
In an attempt to make a rough estimate of the potential Arntz, W.E., 1973. Periodicity of diel food intake of cod Gadus morhua in the
effect of loss of seagrass on cod recruitment the following ex- Kiel bay. Oikos 15, 138e145.
ample is given: the present (2003e2004) spatial extent of Zos- Baden, S., Pihl, L., 1984. Abundance, biomass and production of mobile epi-
benthic fauna in Zostera marina (L.) meadows, western Sweden. Ophelia
tera marina in the Swedish Skagerrak archipelago has been 23, 65e90.
assessed to be around 130 kmÿ2 (Johan Stal, personal commu-
˚ Baden, S., Bostrom, C., 2001. The leaf canopy of seagrass beds: fauna com-
¨
nication). According to Baden et al. (2003) the distribution of munity structure and functional in a salinity gradient along the Swedish
Z. marina decreased by 60% over the last two decades, which coast. In: Reise, K. (Ed.), Ecological Comparisons of Sedimentary Shores.
would equal a total loss of 190 kmÿ2 (from 320 to 130 kmÿ2) Ecological Studies 151. Springer Verlag, pp. 213e236.
´
Baden, S., Gullstrom, M., Lunden, B., Pihl, L., Rosenberg, R., 2003.
¨
of the seagrass habitat. In our study, mean density of juvenile Vanishing seagrass (Zostera marina) in Swedish coastal waters. Ambio
(0-group) cod was estimated to be 17 individuals per 500 mÿ2 32, 374e377.
haul in Z. marina meadows. At non-Zostera sites, densities Beck, M.W., Heck Jr., K.L., Able, K., Childers, D.L., Eggelston, D.B.,
were estimated to be 0.6 individual per haul. Hence, using Gillanders, B.M., Halpern, B., Hays, C.G., Hoshino, K., Minello, T.J.,
the difference in density of juvenile cod (16.4 individuals Orth, R.J., Sheridan, P.F., Weinstein, M.P., 2001. The identification, con-
servation and management of estuarine and marine nurseries for fish and
per haul) between Zostera and non-Zostera habitats, the loss invertebrates. BioScience 51, 633e641.
in Z. marina meadows over the last two decades could have ˚
Borg, A., Pihl, L., Wennhage, H., 1997. Habitat choice of juvenile cod (Gadus
resulted in a reduction in recruitment of 6.3 million juvenile morhua L.) on sandy soft bottoms with different vegetation types. Helgo-
cod each year. This is of the same order of magnitude as the lander Meeresunersuchungen 51, 197e212.
¨
estimated combined effect of seal predation and mortality Bostrom, C., Baden, S.P., Krause-Jensen, D., 2003. The seagrasses of Scandi-
¨
navia and the Baltic Sea Chap. 1. In: Green, E.P., Short, F.T. 2003. World
due to by-catches in the commercial eel fyke-net fishery atlas of Seagrasses: present status and future conservation. Prepared by
(Mats Ulmestrand, personal communication) in the archipela- UNEP World Conservation Monitoring Centre. University of California
go of the Swedish Skagerrak. Press. Berkeley, USA. ISBN 0-520-24047-2. 298 pp.
132 L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132
Brind’Amour, A., Boisclair, D., 2004. Comparison between two sampling Jenkins, G.P., May, H.M.A., Wheatley, M.J., Halloway, M.G., 1997.
methods to evaluate the structure of fish communities in the littoral zone Comparison of fish assemblages associated with seagrass and adjacent un-
of a Laurentian lake. Journal of Fish Biology 65, 1372e1384. vegetated habitats of Port Philip Bay and Corner inlet, Victoria, Australia,
Browder, J.A., Restrepo, V.R., Rice, J.K., Robblee, M.B., Zein-Eldin, Z.Z., with emphasis on commercial species. Estuarine, Coastal and Shelf Sci-
1999. Environmental influences on potential recruitment of pink shrimp, ence 44, 569e588.
Farfantepanaeus duorarum, from Florida Bay nursery grounds. Estuaries Lazzari, M.A., 2002. Epibenthic fishes and decapod crustaceans in northern
22, 484e499. estuaries: a comparison between vegetated and unvegetated habitats in
Clarke, K.R., 1993. Non-parametric multivariate analyses of changes in com- Maine. Estuaries 25, 1210e1218.
munity structure. Australian Journal of Ecology 18, 117e143. Moller, P., Pihl, L., Rosenberg, R., 1985. Benthic faunal energy flow and bi-
¨
Clarke, K.R., Warwick, R.M., 1994. Change in Marine Communities: An Ap- ological interaction in some shallow bottom habitats. Marine Ecology
proach to Statistical Analysis and Interpretation. Natural Environment Re- Progress Series 27, 109e121.
search Council, UK, 144 pp. Orth, R.J., Heck Jr., K., van Montfrans, J., 1984. Faunal communities in sea-
Deegan, L.A., Wright, A., Ayvazian, S.G., Finn, J.T., Golden, H., grass beds: a review of the influence of plant structure and prey character-
Merson, R.R., Harrison, J., 2002. Nitrogen loading alters seagrass ecosys- istics on predatoreprey relationships. Estuaries 7, 339e350.
tem structure and support of higher trophic levels. Aquatic Conservation: Orth, R.J., van Montfrans, J., Lipcius, R.N., Metcalf, K.S., 1996. Utilization of
Marine and Freshwater Ecosystems 12, 193e212. seagrass habitat by the blue crab, Callinectes sapidus Rathbun, in Chesa-
Edgar, G.J., Shaw, C., 1995. The production and trophic ecology of shallow- peake Bay: a review. In: Kuo, J., Phillips, R.C., Walker, D.I., Kirkman, H.
water fish assemblages in southern Australia: 1. Species richness, size- (Eds.), Seagrass Biology. Proceedings of an International Workshop,
structure and production of fishes in Western Port, Victoria. Journal of Ex- Rottnest Island, Western Australia. Western Australian Museum, Perth,
perimental Marine Ecology and Biology 194, 53e81. pp. 213e224.
Ferrell, D.J., Bell, J.D., 1991. Differences among assemblages of fish associ- Parsley, M.J., Palmer, D.E., Burkhardt, R.W., 1989. Variation in capture effi-
ated with Zostera caricorni and bare sand over a large spatial scale. Marine ciency of a beach seine for small fishes. North American Journal of Fish-
Ecology Progress Series 72, 15e24. eries and Management 9, 239e244.
Field, J.G., Clarke, K.R., Warwick, R.M., 1982. A practical strategy for ana- Petersen, J.C.G., 1914. Om bændeltangens (Zostera marina) aars-produktion i
lyzing multispecies distribution patterns. Marine Ecology Progress Series de danske farvande. Kap. X i Jungersen, H.F.E. og Warming, Eug. Mind-
8, 37e52. ˚
eskrift i anledning af hundredaret for Japetus Steenstrups fødsel. I Kom-
Fonds, M., 1973. Sand gobies in the Dutch Wadden Sea (Pomatoschistus, Go- mission hos GEC Gad, København. Bianco Lunos Bogtrykkeri (in Danish).
biidae, Pisces). Netherlands Journal of Sea Research 6, 417e478. Pierce, C.L., Rasmussen, J.B., Leggett, W.C., 1990. Sampling littoral fish with
Gibson, R.N., Pihl, L., Burrows, M.T., Modin, J., Wennhage, H., Nickell, L.A., a seine: corrections for variable capture efficiency. Canadian Journal of
1998. Diel movements of juvenile plaice (Pleuronectes platessa) in Fisheries and Aquatic Sciences 47, 1004e1010.
relation to predators, competitors, food availability and abiotic factors on Pihl, L., 1982. Food intake of young cod and flounder in a shallow bay on the
a microtidal nursery ground. Marine Ecology Progress Series 165, 145e Swedish west coast. Netherlands Journal of Sea Research 15, 419e432.
159. Pihl, L., 1989. Abundance, biomass and production of juvenile flatfish in
Gotceitas, V., Fraser, S., Brown, J.A., 1997. Use of eelgrass beds (Zostera ma- southeastern Kattegat. Netherlands Journal of Sea Research 24, 69e81.
rina) by juvenile Atlantic cod (Gadus morhua). Canadian Journal of Fish- Pihl, L., Rosenberg, R., 1982. Production, abundance and biomass of mobile
eries and Aquatic Sciences 54, 1306e1319. epibenthic marine fauna in shallow waters, western Sweden. Journal of
Green, E.P., Short, F.T., 2003. World Atlas of Seagrasses: Present Status and Experimental Marine Biology and Ecology 57, 273e301.
Future Conservation (prepared by UNEP World Conservation Monitoring Pihl, L., Ulmestrand, M., 1993. Migration pattern of juvenile cod, Gadus
Centre). University of California Press, Berkeley, USA, ISBN 0-520- morhua, on the Swedish west coast. ICES Journal of Marine Science 50,
24047-2, 298 pp. 63e70.
Heck, K.L., Hays, G., Orth, R.J., 2003. Critical evaluation of the nursery role Pihl, L., Wennhage, H., 2002. Structure and diversity of fish assemblages on
hypothesis for seagrass meadows. Marine Ecology Progress Series 253, rocky and soft bottom shores on the Swedish west coast. Journal of Fish
123e136. Biology 61, 148e166.
Hily, C., van Katjwijk, M.M., den Hartog, C., 2003. Seagrasses of Western Short, F.T., Wyllie-Echeverria, S., 1996. Natural and human-induced distur-
Europe (Chapter 2). In: Green, E.P., Short, F.T. (Eds.), World Atlas of bances of seagrasses. Environmental Conservation 23, 17e27.
Seagrasses: Present Status and Future Conservation. University of Califor- Spalding, M., Taylor, M., Ravilious, C., Short, F., Green, E., 2003. Global
nia Press, Berkeley, USA, ISBN 0-520-24047-2, 298 pp. (prepared by overview: the distribution and status of seagrasses. In: Green, E.P.,
UNEP World Conservation Monitoring Centre). Short, F.T. (Eds.), World Atlas of Seagrasses: Present Status and Future
Howarth, R.W., Anderson, D., Cloern, J., Elfring, J., Hopkinson, C., Conservation. University of California Press, Berkeley, USA, ISBN
Lapointe, B., Malone, T., Marcus, N., McGlathery, K., Sharpley, A., 0-520-24047-2, 298 pp. (prepared by UNEP World Conservation Monitor-
Walker, D., 2000. Nutrient pollution of coastal rivers, bays, and seas. Is- ing Centre).
sues in Ecology 7, 1e15. Tveite, S., 1984. 0-Group cod investigations on the Norwegian Skagerrak
Hughes, J.E., Deegan, L.A., Wyda, J.C., Weaver, M.J., Wright, A., 2002. The coast. In: Dahl, E., Danielssen, D.S., Moksness, E., Solemdal, P. (Eds.),
effects of eelgrass habitat loss on estuarine fish communities of southern The Propagation of Cod Gadus morhua L.. Flødevigen. Rapport 1,
New England. Estuaries 25, 235e249. 581e590 (ISSN 033-2594).
Hughes, A.R., Bando, K.J., Rodriguez, L.F., Williams, S.L., 2004. Relative ef- Weaver, M.J., Magnuson, J.J., Clayton, M.K., 1993. Analyses for differentiat-
fects of grazers and nutrients on seagrasses: a meta-analysis approach. Ma- ing littoral fish assemblages with catch data from multiple sampling gears.
rine Ecology Progress Series 282, 87e99. Transactions of the American Fisheries Society 122, 1111e1119.
Jackson, E.L., Rowden, A.A., Attrill, M.J., Bossey, S.J., Jones, M.B., 2001. Wennhage, H., Pihl, L., 2002. Fish feeding guilds in shallow rocky and soft
The importance of seagrass beds as habitat for fishery species. Oceanogra- bottom areas on the Swedish west coast. Journal of Fish Biology 61,
phy and Marine Biology: An Annual Review 39, 269e303. 207e228.
www.elsevier.com/locate/ecss
Shift in fish assemblage structure due to loss of seagrass
Zostera marina habitats in Sweden
Leif Pihl a,*, Susanne Baden a, Nils Kautsky b, Patrik Ronnback b, Tore Soderqvist c,
¨ ¨ ¨
c a
˚
Max.Troell , Hakan Wennhage
a
Department of Marine Ecology, Goteborg University, Kristineberg Marine Research Station, 450 34 Fiskebackskil, Sweden
¨ ¨
b
Department of System Ecology, Stockholm University, 106 91 Stockholm, Sweden
c
Beijer International Institute of Ecological Economics, The Royal Swedish Academy of Science, 10405 Stockholm, Sweden
Received 6 June 2005; accepted 14 October 2005
Available online 19 January 2006
Abstract
The areal extent of Zostera marina in the archipelago of the Swedish Skagerrak has decreased by 60% over two decades. To investigate the
effects of Z. marina loss on the local fish assemblages, the fish fauna was compared between existing seagrass beds and sites where seagrass had
vanished. A field study was carried out at four shallow locations in the outer archipelago of the coast in June 2004. Within each location two sites
were sampled, one with an existing Z. marina bed and another where Z. marina had disappeared. Fish were sampled semi-quantitatively with
a beach seine. Samples were taken during both day and night and captured fish were examined to species, enumerated and measured in the field,
and released thereafter. The number of fish species was found to be significantly higher in Z. marina habitats compared to areas where seagrass
was missing, and density and biomass of fish were generally lower in areas dominated by bare sediment compared to those in the seagrass hab-
itats. Several species and groups of fishes (i.e., gadoids, labrids, syngnathids) were absent or occurred in low densities at sites where Z. marina
was missing. For example, juvenile 0-group cod density was reduced by 96% at sites where Z. marina had disappeared. Such a reduction in
recruitment of cod is in the same order of magnitude as the combined effect of seal predation and mortality due to by-catches in the eel
fyke-net fishery estimated for the archipelago of the Swedish Skagerrak. Hence, the results clearly indicate a shift in the fish assemblage, in-
cluding a loss of taxa at the family level as a result of degradation in habitat-forming vegetation.
Ó 2005 Elsevier Ltd. All rights reserved.
Keywords: habitat loss; eutrophication; cod; plaice; Skagerrak
1. Introduction are considered to be higher compared to alternative habitats
(Beck et al., 2001; Deegan et al., 2002).
Seagrass meadows provide habitat diversity in the coastal The global loss of seagrasses from the mid-1980s to the
seascape worldwide. This habitat-forming vegetation creates mid-1990s has been estimated to be 12,000 km2 (Short and
a three dimensional architecture over soft bottoms, which sta- Wyllie-Echeverria, 1996), which correspond to an overall re-
bilizes the sediment and reduces water movements. Seagrass duction by about 7% of the total areal extent (Spalding
meadows are known to harbour a diverse and abundant fauna et al., 2003). This fact has contributed to the listing of seagrass
of invertebrates and are generally considered as essential hab- beds as habitats worthy of protection in the Rio-declaration
itats for many fish species (Orth et al., 1984; Jenkins et al., (1992/93:13). The consensus from the report ‘‘World Atlas
1997). Fish may spawn in seagrass beds or use the area as of Seagrasses’’ is that the main reasons for the decline, apart
a nursery ground, where growth and survival of juvenile fish from natural threats such as storms and diseases, are anthropo-
genic (Green and Short, 2003). Most serious are the indirect
effects of human activity. Increased turbidity and overgrowth
* Corresponding author. by epiphytic algae reduce light penetration, resulting in subse-
E-mail address: leif.pihl@kmf.gu.se (L. Pihl). quent loss of seagrass. Turbidity and surplus algal growth may
0272-7714/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ecss.2005.10.016
124 L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132
derive from physical disturbances on land or in the water, and however, hard to evaluate and there may be alternative explan-
from eutrophication combined with trophic cascades (e.g. ef- ations for the changes observed (Orth et al., 1996; Browder
fects of over-fishing) in the coastal ecosystem (Howarth et al., 1999). To our knowledge there is only one study com-
et al., 2000; Jackson et al., 2001; Hughes et al., 2004). Direct paring areas with extant and lost seagrass habitats (Hughes
disturbances such as dredging and benthic trawling in seagrass et al., 2002). The study, performed in estuaries along the
will most likely cause drastic local effects, but are probably of NW Atlantic Coast, reported that the abundance, biomass
less importance on a larger scale compared to more diffuse, in- and species richness of fish were lower in areas where seagrass
direct anthropogenic affects (Hily et al., 2003). beds had disappeared, but the generality of these findings for
Zostera marina is the dominant seagrass species in Swedish other coastal areas are not known.
coastal waters. Along the SkagerrakeKattegat coast, it occurs The aim of this study was to investigate the effects of losses
in semi-exposed and protected areas within water depths of of Zostera marina on the local fish assemblages on the Swedish
0.5e6 m (Baden and Bostrom, 2001). These Zostera beds
¨ west coast. This was carried out by comparing the fish fauna in
have been shown to support a high production of benthic fauna existing seagrass beds with sites where seagrass has vanished
as well as epibenthic invertebrates and fish (Baden and Pihl, over the last two decades. The main purpose was to document
1984; Moller et al., 1985), and to serve as nursery and feeding
¨ shifts in fish assemblages through measurements of species
grounds for more than 40 fish species (Pihl and Wennhage, numbers, densities and biomass. By comparing the utilization
2002; Wennhage and Pihl, 2002). Over the last two decades, by fish of Z. marina beds and alternative habitats, changes in
a 60% reduction in distribution of the Z. marina has been ob- the ecological function of coastal areas with loss of seagrass
served in the Swedish Skagerrak archipelago (Baden et al., could be evaluated.
2003). Along most sections of the coast both the upper and
the lower depth distributions of seagrass have been reduced,
resulting in a narrowing of meadows, but in some areas sea- 2. Methods
grass meadows have disappeared completely. The lost Z. ma-
rina is commonly replaced by a bare sediment bottom, but This field study was carried out in the outer archipelago of
in some areas the sediment is partly covered by filamentous the Swedish Skagerrak coast (58 14e220 N; 11 23e320 E)
green algae or patches of Fucus spp., attached to shells and (Fig. 1) between 8 and 20 June 2004. The archipelago consists
stones. The reason for the degradation of the seagrass habitat of islands of varying size, and a shoreline characterized by
in the Skagerrak is not known, but coastal eutrophication and/ a mixture of rocky and soft-bottom substrata. On soft bottoms
or over-fishing have been suggested as plausible causes. In ad- Zostera marina is the dominating vegetation within the depth
dition, altered water exchange due to construction of road range of 1e5 m (Baden et al., 2003). This coastal region is
banks and leisure boat harbours could have reduced the distri- micro-tidal with a tidal amplitude of around 0.2 m. Mean sur-
bution of seagrass in the coastal Skagerrak. face water temperatures usually range from 5 to 15 C in
Historical distribution maps of seagrass from Scandinavia spring and autumn and from 15 to 20 C during the summer
are few, but extensive Danish investigations dating back to (Pihl and Rosenberg, 1982). Surface water salinity typically
1900 reveal that only about 25% of the former areal extension fluctuates between 20 and 25 psu in the summer.
remained in 1990 (Petersen, 1914). This large areal reduction June was selected for sampling in this investigation because
is partly attributed to losses of deep eelgrass stands as a conse- previous studies on seasonal dynamic of the fish community
quence of impoverished light conditions and partly to the slow had shown that the highest species richness, abundance and
recovery after the seagrass disease in the 1930s. Between 1900 biomass occurred at that time of the year (Pihl and Wennhage,
and 1990, maximum colonisation depths decreased from 2002). In June most of the fish are recruited to the coastal hab-
5e6 m in estuaries and 7e8 m in open waters, to 2e3 m itats, giving a full representation of age-classes in the fish
and 4e5 m, respectively (Bostrom et al., 2003).
¨ community. The investigated coastal region represents one
The change in habitat structure following the loss of of five areas where Zostera marina has been observed to de-
a Zostera marina bed is likely to shift the local system into crease significantly in its distribution over the last two decades
an alternative state. Primarily, the loss of the habitat-forming (Baden et al., 2003). Within this region four locations were
species Z. marina will alter habitat complexity, changing the chosen at random and two sites were selected in each: one
structure of associated fauna assemblages. In a review of the with an existing Z. marina bed and another where Z. marina
extensive literature testing the importance of seagrass mead- had disappeared (Fig. 1). Sites with existing seagrass were
ows as nursery areas for juvenile fish and invertebrates, chosen in close vicinity (about 500 m) to the sites without sea-
Heck et al. (2003) found that their abundance, growth and sur- grass. At the seagrass sites each Z. marina bed had a spatial
vival were generally higher in seagrass compared to unstruc- distribution of more than 10 ha, covering 60e100% of the
tured habitats, but similar to other structured habitat, bottom area within the depth range of 1e4 m. At sites where
indicating that habitat complexity may be more important to Z. marina had disappeared compared to the 1982 distribution,
fish diversity than the type of structures present. There are the bottom sediment was generally free of vegetation, except
also a few studies more specifically reporting changes in spe- from sparse occurrence of Z. marina shoots and small patches
cies composition following large-scale losses of seagrass beds. of Fucus spp. stands. Vegetation cover at these sites was
Historical comparisons without appropriate controls are, 1e15% of the bottom area.
L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132 125
Fig. 1. Map of the investigated area in the archipelago of the Swedish Skagerrak. Sampling locations indicated by open circles.
A sampling area of 5 ha was designated in the centre of haul during the night (24e01 hours) in each of the investigated
each selected site. Fish were sampled semi-quantitatively sites. Sampling order was randomly allocated between sites
with a beach seine according to Tveite (1984), a method and over time to avoid introduction of systematic errors.
most commonly used to study fish communities in the littoral Day and night samples were taken at random in the designated
zone. The method has the advantage of allowing for estimates area, and !50 m apart within a site. Captured fish were exam-
of the area being sampled; a feature not shared by some other ined to species, enumerated and measured (total length, mm)
methods in use (e.g. gillnets and fyke-nets). Methodological in the field, and released thereafter. Some fish from three large
studies have shown that benthic species and small-size individ- samples collected at night were brought to the laboratory for
uals may hide within substrata having a high complexity or es- further analysis. Estimates of the biomass (wet wt.) for all in-
cape underneath the foot-rope of the seine (Parsley et al., dividual fishes were derived from established lengtheweight
1989). Beach seines may also have selectivity towards relationships. In addition to fish, macro-crustaceans (mainly
small-sized species, at least in comparison to other methods shrimps and crabs) were also collected. Numbers and pooled
(Pierce et al., 1990; Weaver et al., 1993). However, the beach biomass (wet wt.) for each species of invertebrates were re-
seine maintains its performance better than visual sensing corded for each sample in the laboratory.
techniques when macrophyte cover increases (Brind’Amour Shoot density, mean and maximum length and biomass of
and Boisclair, 2004). The seine was 40 m long, 3 m high, Zostera marina were estimated in the four seagrass beds.
had a mesh opening of 10 mm in the arms and 5 mm in the Three quantitative samples were taken by a diver using
‘‘cod end’’ and was towed by 30 m long ropes. The gear a net-bag (mesh opening 1 mm) connected to a bottom metal
was deployed from a small boat in a rectangular shape with ring (diameter of 35 cm). Samples were allocated randomly in
its deepest part at around 3 m depth. It was pulled shoreward the centre of the Z. marina bed at approximately 2 m depth.
33 m by four people 15 m apart until reaching a water depth of Percentage cover of Z. marina was visually assessed by a diver,
1 m, giving an effective fishing area of approximately 500 m2. swimming over the investigated area. At sites where Z. marina
One haul was taken during daytime (12e13 hours) and one had disappeared, the vegetation cover was assessed visually
126 L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132
from the boat before sampling. To characterize the sediment Table 1
substrate three quantitative sediment cores were collected ran- Shoot density, blade length and biomass (wet wt.) of Zostera marina, and cov-
er of Z. marina and Fucus spp., as well as sediment organic content (%) in the
domly by a diver in each of the study sites (1e3 m water eight sample sites
depth) for analysis of organic content. Samples of the upper
Brofjorden Lindholmen Finnsbo ˚ ¨
Gaso
2 cm of the sediment core were dried at 60 C for 24 h and
combusted for 4 h at 450 C, and organic content was mea- Zostera marina sites
Cover of Zostera (%) 100 60 100 100
sured by weight loss.
Fish assemblage structure was compared between samples Number of shoots
Mean (g dw mÿ2) (n ¼ 3) 230 100 380 250
by using BrayeCurtis similarity indices, as described in the
SE 36 21 21 50
PRIMER package (Field et al., 1982). Fish abundance data
were log(X þ 1) transformed to weigh the relative numerical Blade length
Mean (cm) (n ¼ 3) 47.2 30.1 22.1 39.8
importance of common and rare species in the analysis. SE 4.6 2.3 1 3.9
BrayeCurtis similarity indices were computed and the result- Max (cm) 104 67 39 66
ing similarity matrix was used to perform non-metric Multi-
Blade biomass
Dimensional Scaling (MDS). An analysis of similarities Mean (g dw mÿ2) (n ¼ 3) 172.2 43.5 71.2 141.2
(ANOSIM) was used to test for differences in assemblage SE 13.8 21.5 11.3 31.4
structure between the two habitat types, and to compare day Sediment org. content
and night samplings (method in Clarke, 1993). The propor- Mean (%) (n ¼ 3) 16.6 14.5 1.7 13.6
tional contribution of different fish species to the dissimilarity SE 1.3 0.6 0.1 0.8
between groups was investigated using SIMPER (method in
Clarke and Warwick, 1994). Fish abundance, biomass and Non-Zostera marina sites
Cover of Zostera (%) 0 0 1 0
number of species in the two habitat types were compared us-
ing two-way ANOVAs, with habitat and time of day as factors. Cover of Fucus spp. (%) 5 10e15 0 5e10
Non-transformed data used as variances were shown to be ho- Sediment org. content
mogenous according to Cochran’s test. Mean (%) (n ¼ 3) 5.4 3.3 1 1.3
SE 0.5 0.4 0.06 0.06
3. Results
˚ ¨
two areas, one was considered as semi-exposed (Gaso) and
3.1. Vegetation one had a high physical exposure (Finnsbo). In the protected
and semi-exposed areas, content of organic matter in Zostera
In three out of four Zostera marina beds investigated the marina beds varied between 13.6 and 16.6%, whereas the ex-
vegetation had a homogenous density with full cover of the sed- posed Z. marina bed had a sandy sediment with only 1.7% or-
iment. Only at one site, Lindholmen (location 2), the Z. marina ganic content (Table 1). At the sampled sites mainly free of
had a patchy distribution covering approximately 60% of the vegetation content of organic matter in the sediment was esti-
sediment. Mean shoot density was estimated to be 100e mated to be between 1.0 and 5.4%, with the highest value in
380 shoots mÿ2 in the four Z. marina beds, the lowest densities the protected areas (Table 1). Thus, organic content in the sed-
occurring at the site where the bed had a patchy distribution iment was 2e10 times higher in Z. marina beds compared to
(Table 1). The mean blade length varied from 22 to 47 cm at unvegetated sites, but inter-location differences could be im-
the four sites, and maximum length was about double the portant due to variation in exposure.
mean length at all sites (Table 1). Blade biomass of Z. marina
was found to be between 44 and 172 g dw mÿ2 at the study sites,
3.3. Fish
with the lowest values in the beds having either low mean blade
length or a patchy distribution. At the four sites where Z. marina
Altogether, 33 fish species belonging to 15 families were
beds had disappeared over the last two decades the sediment
identified in this investigation (Table 2). Twenty-eight species
was mainly free of vegetation. At one site (the exposed area
were found in the Zostera marina beds, of which nine were ex-
Finsbo, location 3) a few remaining shoots of Z. marina were
clusive to this habitat. At the sites where Z. marina had disap-
found covering around 1% of the bottom (Table 1). The other
peared 19 fish species were found, and five of these species
three sites had patches of Fucus spp. growing on stones and
were only found here. When comparing Z. marina and non-
shells of blue mussels with an approximate cover of between
seagrass sites in pairs for each location, the number of fish spe-
5 and 15% of the bottom sediment.
cies was in all cases higher in the Zostera habitat (Fig. 2).
Overall the number of fish species was significantly higher
3.2. Sediment ( p < 0.005) in Z. marina beds than at non-seagrass sites, but
no difference was observed between day and night samplings
Two of the investigated locations (Brofjorden and Lindhol- (Table 3). The mean number of individuals sampleÿ1 exhibited
men) were situated in an enclosed part of the archipelago and a large variation between locations and sites (Fig. 2). Generally,
were protected from exposure to wind and waves. Of the other the numbers of individuals were higher in catches from
L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132 127
Table 2
Number of individuals and biomasses (wet wt.) of fishes recorded in beach-seine samples. Pooled data of day and night samples from four Zostera marina sites, and
from four sites where Z. marina has disappeared
Family Fish species Numbers Biomass
Zostera Non-Zostera Zostera Non-Zostera
marina marina marina marina
Anguillidae Anguilla anguilla 12 23 456.4 1436.7
Gobiidae Aphia minuta 1423 169 4705.5 261.0
Callionymidae Callionymus lyra 0 1 0.0 11.5
Clupeidae Clupea harengus 1 84 2.6 157.4
Labridae Ctenolabrus rupestris 272 2 2062.8 17.7
Syngnathidae Entelurus aequoreus 16 0 70.1 0.0
Gadidae Gadus morhua 146 12 947.1 1142.7
Gasterosteidae Gasterosteus aculeatus 1164 916 1781.7 1527.6
Gobiidae Gobius niger 738 304 4605.1 924.1
Gobiidae Gobiusculus flavescens 264 0 137.3 0.0
Pleuronectidae Limanda limanda 5 0 228.3 0.0
Gadidae Merlangius merlangus 16 0 482.1 0.0
Cottidae Myoxocephalus scorpius 22 2 868.9 81.6
Syngnathidae Nerophis lumbriciformis 1 1 2.6 5.5
Syngnathidae Nerophis ophidion 32 4 25.1 2.3
Pholidae Pholis gunnellus 3 0 42.5 0.0
Pleuronectidae Platichthys flesus 11 37 2206.6 1835.4
Pleuronectidae Pleuronectes platessa 21 143 74.8 310.8
Gadidae Pollachius virens 4 0 11.0 0.0
Gobiidae Pomatoschistus microps 0 90 0.0 85.1
Gobiidae Pomatoschistus minutus 30 72 75.6 178.6
Gobiidae Pomatoschistus pictus 16 18 23.6 23.5
Salmonidae Salmo trutta 9 4 785.4 777.8
Bothidae Scophthalmus rhombus 0 1 0.0 0.2
Soleidae Solea solea 0 1 0.0 48.8
Gasterosteidae Spinachia spinachia 2 0 8.4 0.0
Labridae Symphodus melops 4 0 62.8 0.0
Syngnathidae Syngnathus acus 35 0 179.2 0.0
Syngnathidae Syngnathus rostellatus 33 23 38.0 27.3
Syngnathidae Syngnathus typhle 129 37 135.2 52.6
Cottidae Taurulus bubalis 7 7 204.8 120.6
Gadidae Trisopterus esmarkii 5 1 6.0 0.5
Zoarcidae Zoarces viviparus 82 57 1648.0 474.5
4503 2009 21,877.5 9503.7
Z. marina beds compared to catches from areas where seagrass between day and night samplings. Furthermore, of the two
had disappeared when sites were compared in pairs. However, habitat types, Z. marina sites were more closely clustered in
overall no significant difference in fish density could be de- the analysis than bare sediment sites, indicating a higher sim-
tected between habitats or between day and night samplings ilarity of the fish assemblages in Z. marina beds.
(Table 3). Except for one sample in the Z. marina bed at loca- An SIMPER-analysis revealed that the distribution of 10
˚ ¨
tion 4 (Gaso), the fish biomass was generally low during day- species explained about 70% of dissimilarity between the two
time sampling, especially at non-seagrass sites (Fig. 2). Night habitat types (Table 4). Of these species, eight had higher den-
fish biomass was generally higher than daytime biomass, and sities in Zostera marina beds and two were more abundant in
similar total fish weights were recorded in the two habitat non-seagrass habitats. The high affinity of several fish species
types. Overall, a trend toward higher fish biomass was re- to the Zostera habitat is further emphasised by the fact that
corded in Z. marina beds, although the difference from non- 10 out of the 20 most abundant fish species were almost exclu-
seagrass sites was not significant ( p ¼ 0.10; Table 3). sively caught in the seagrass beds. Gadoids (Gadus morhua,
MDS ordination based on a BrayeCurtis similarity matrix Merlangius merlangus and Pollachius virens) labrids (Cteno-
showed that the fish assemblages were mainly structured ac- labrus rupestris and Symphodus melops), syngnathids (Syngna-
cording to habitat type, whereas time of the day was of less thus acus and Entelurus aequoreus) and Gobiusculus flavescens
importance for the structure of the fish assemblages (Fig. 3). were predominantly found in Z. marina beds, whereas flatfishes
An ANOSIM-test revealed a significant difference (Global such as Pleuronectes platessa, Platichthys flesus and Solea
R ¼ 0.37; p ¼ 0.01) between the fish assemblage structure in solea mainly occurred on bottoms dominated by bare sediment.
Zostera marina beds and non-seagrass habitat, but the test Cod (Gadus morhua) and plaice (Pleuronectes platessa) are
failed to show any difference (Global R ¼ ÿ0.07; p ¼ 0.74) the most important species of commercial interest that
128 L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132
25 Day
20
15
Species per sample
10
5
0
25 Night
20
15
10
5
0
1000 Day
600
Numbers per sample
200
0
1400 Night
1000
600
200
0
8000 Day Zostera
6000 Non-Zostera
Biomass per sample
4000
2000
0
8000 Night
6000
4000
2000
0
1 2 3 4
Sites
Fig. 2. Number of species, individuals and biomass of fish captured during day and night samplings at four Zostera marina and four non-Z. marina sites in the
archipelago of the Swedish Skagerrak.
occurred in high densities in this investigation. They both uti- 1993). From analysis of length distribution it was obvious
lize the coastal zone as a nursery and juveniles may stay in that 0-group juvenile cod mainly utilized the Zostera marina
shallow (<10 m) waters for about two years after settlement beds as a nursery, whereas 1-group cod were equally repre-
to the benthic habitat (Pihl, 1989; Pihl and Ulmestrand, sented in both habitats (Fig. 4). In the Z. marina beds cod
Table 3
Two-fixed-factor ANOVA-modes. Number of species, density and biomass of
fish as a function of habitat (Zostera, no Zostera) and time (day, night) Stress: 0.09
2
Source of variation SS df MS F p
2 1
Number of fish species
4 4
Habitat 203.1 1 203.1 21.8 0.005 2 2
3
Time 39.1 1 39.1 4.2 0.063
Habitat  Time 1.6 1 1.6 0.2 0.692 4
1
Residual 111.8 12 9.3 1
4
1
Density of fish
Habitat 388,752 1 388,752 2.58 0.134
Time 133,225 1 133,225 0.88 0.365 3
Habitat  Time 49,952 1 49,952 0.33 0.575
Residual 1,805,974 12 150,497 Day Night
Non-Zostera
Biomass of fish Zostera 3 3
Habitat 9,571,289 1 9,571,289 3.91 0.099
Time 2,051,340 1 2,051,340 0.68 0.424
Fig. 3. Similarities in fish assemblage structure between day and night
Habitat  Time 2,133,790 1 2,133,790 0.71 0.415
samplings in four Zostera marina and four non-Z. marina sites, based on
Residual 3.6Eþ07 12 2,996,293
Multi-Dimensional Scaling (MDS) ordination.
L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132 129
Table 4 and six species were found in Zostera marina habitats
Fish species most responsible for the difference in assemblage structure be- (Table 6). Carcinus maenas, Crangon crangon, Palaemon
tween Zostera marina and non-Zostera marina habitats, listed in the order
of their contribution to the average BrayeCurtis dissimilarity. Abundance is
adspersus and Palaemon elegans were the dominant species
mean individuals sampleÿ1 in both habitats. There was a trend towards higher densities
Rank Fish species Abundance % Contribution
and biomasses in non-seagrass compared to Z. marina sites,
however, the differences were not significant ( p > 0.05).
Zostera Non-Zostera
Catches were higher during the night, with significant larger
1 Aphia minuta 177.8 21.1 11.1 biomass ( p ¼ 0.005) at night compared to day samples in
2 Gasterosteus aculeatus 145.5 114.5 8.4
3 Gobiusculus flavescens 33.0 0 8.2
both habitats.
4 Gadus morhua 18.3 1.5 5.9
5 Gobius niger 92.3 38.0 5.4 4. Discussion
6 Syngnathus typhle 16.1 4.6 5.4
7 Zoarces viviparus 10.3 7.1 5.3 The main purpose of this study was to document shifts in
8 Ctenolabrus rupestris 34.0 0.3 5.0
9 Pleuronectes platessa 2.6 17.9 4.5
the assemblages of fish as a direct consequence of loss of
10 Pomatoschistus minutus 3.8 9.0 4.4 the habitat-forming vegetation, Zostera marina, in shallow
11 Nerophis ophidion 4.0 0.5 3.3 soft-bottom areas. It would be expected that species richness
12 Syngnathus rostellatus 4.1 2.9 3.2 and composition of fish species would change when vegetation
13 Pomatoschistus microps 0 11.3 3.1 disappears as a result of lower habitat complexity (Jackson
14 Syngnathus acus 4.4 0 2.7
15 Pomatoschistus pictus 2.0 2.3 2.6
et al., 2001; Hughes et al., 2002; Lazzari, 2002), but the den-
16 Anguilla anguilla 1.5 2.9 2.5 sity of fish does not necessarily decrease, since shallow soft
17 Platichthys flesus 1.4 4.6 2.5 bottoms are known to host large abundances of small fishes
18 Myoxocephalus scorpius 2.8 0.3 2.5 (Edgar and Shaw, 1995). The archipelago of the area investi-
19 Merlangius merlangus 2.0 0 2.1 gated consists of a mosaic of rocky- and soft-bottom habitats
20 Entelurus aequoreus 2.0 0 2.0
21 Taurulus bubalis 0.9 0.9 1.7
that are to a varying degree covered by vegetation, thereby of-
22 Salmo trutta 1.1 0.5 1.7 fering a suite of alternative habitats with different complexity
23 Clupea harengus 0.1 10.5 1.3 that could be utilized by littoral fish. When seagrass disappears
24 Trisopterus esmarkii 0.6 0.1 0.9 from an area, fish could either concentrate in alternative veg-
25 Pollachius virens 0.5 0 0.7 etated habitats or stay in the altered habitat dominated by
26 Limanda limanda 0.6 0 0.7
27 Symphodus melops 0.5 0 0.7
bare sediment. Therefore, as pointed out in previous studies
28 Nerophis lumbriciformis 0.1 0.1 0.6 comparing fish in seagrass and bare sediment (Ferrell and
29 Pholis gunnellus 0.4 0 0.6 Bell, 1991), it is important to consider the size of the area
30 Spinachia spinachia 0.3 0 0.5 where seagrasses have been lost and the distance to other alter-
31 Solea solea 0 0.1 0.3 native habitats. In our study, the areas where seagrass had dis-
32 Callionymus lyra 0 0.1 0.3
33 Scophthalmus rhombus 0 0.1 0.3
appeared had a size of several hectares and the distance to
vegetated habitats, in this case other seagrass beds or belts
of macroalgae, was between 200 and 500 m. Despite the close
were caught both during day and night, but in the non-seagrass proximity to complex habitats, a significant reduction in fish
habitats cod only appeared in night samples. In contrast to cod, species and change in species structure were observed in areas
juvenile 0-group plaice were almost exclusively caught at sites where seagrass had vanished. Thus, there is a clear indication
dominated by bare sediment (Fig. 5). They occurred on bare of a shift in the fish assemblage, including a loss of taxa at the
sediment both during day and night. Only a few individuals family level as a result of degradation in habitat-forming veg-
of 1-group plaice were captured in this study. They appeared etation at the observed scale.
in both habitats, but were only caught during night sampling. In our study, fish biomass was generally higher during night
In an attempt to analyze possible relationships between the in habitats both with and without Zostera marina. The high
structure of Zostera marina and the fish assemblages (although biomass was partly due to the larger size of individual fish cap-
based on a small sample size) density, length and biomass of tured during the night. This indicated that fish migrate into
vegetation were compared to species, density and biomass of shallow water at night, and these areas may function as
fish (Table 5). Highest number of fish species and biomass a night-time feeding ground for both types of habitat. Such
was observed in the two Z. marina beds having the greatest nocturnal shoreward migration has previously been described
blade length and the largest biomass of vegetation. These for the two dominating commercial species in the area, cod
two beds also had the largest geographical extension of the and plaice (Pihl, 1982; Gibson et al., 1998). In other investiga-
four investigated Z. marina sites. tions it has been shown that fish migrate from seagrass into
open sediment habitats at night for foraging (Gotceitas et al.,
3.4. Macro-crustaceans 1997).
Different fish species may show different degrees of depen-
In addition to fish, nine species of macro-crustaceans were dency on vegetation. Syngnathids is a group of fish species
found in samples from the sites dominated by bare sediment that are adapted to seagrass by their body shape. The habitat
130 L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132
8
Zostera 0-group 1-group
7
6
5
4
Individuals per length group
3
2
1
0
8
Non-Zostera
7
6
5
4
3
2
1
0
30 40 50 60 70 80 150 200 250 300
Length, mm Length, mm
Fig. 4. Number of individuals per length-class of 0- and 1-group juvenile cod (Gadus morhua) at Zostera marina and non-Z. marina sites.
dependence may vary for this group of fishes, but most species fish (Arntz, 1973). In this way, gobies provide an important
are found among vegetation. Six species of syngnathids were energy link from the highly productive shallow coastal system
found in this study, of which four species occurred in both to fish living in deeper less productive water (Moller et al.,
¨
habitat types and two species were caught exclusively in Zos- 1985). In our study, five species of gobies together contributed
tera marina beds. Altogether, over 80% of the individuals and 30% of the numbers and 15% of the biomass of the total fish
biomass of this group of fish were found in the Z. marina. The assemblage at sites dominated by bare sediment. In Zostera
reason that cryptic species like the syngnathids occupy the marina beds, the corresponding figures for gobies were 55
bare sediment habitat is probably because the missing Zostera and 45%, respectively. The total density of gobies was four
beds had partly been replaced by Fucus spp. that could give times higher in Z. marina beds compared to non-seagrass sites,
sufficient camouflage for these fishes. and biomass was more than six times higher. Thus, Z. marina
Gobies are important components in the food web of littoral beds seem to have a considerably higher capacity for produc-
fish assemblages, occurring in vegetated as well as unvege- tion of gobies that provides an essential energy transfer link to
tated habitats. They are typically small-sized fish with a short fish during the winter season of low productivity.
life span and high production (Fonds, 1973). In temperate Some fish species may vary in their utilization of habitats
waters, gobies utilize shallow water for growth during the over different time scales. The affinity to vegetation may
summer, but usually migrate to deeper water in wintertime change during ontogeny, as for example, early stages of juve-
where they comprise an important food resource for demersal nile cod are more habitat-specific and remain stationary in
vegetation compared to older juveniles (Borg et al., 1997).
Cod uses Z. marina beds as a nursery ground and immigrate
35
30 to these coastal habitats by larval transport. Juveniles settle
25 Non-Zostera in the seagrass by active selection usually avoiding open
20 bare sediment, and consequently the availability of seagrass
Individuals per length-group
15
10
beds may be considered as a bottleneck in the recruitment pro-
5 cess due to the specific habitat requirement of the early benthic
0
35 Table 5
30
Zostera
Density (individual mÿ2), length (cm) and biomass (g wet wt. mÿ2) of Zostera
25
marina blades (n ¼ 3), and number of species, density (individuals sampleÿ1)
20
15
and biomass (g wet wt. sampleÿ1) of fish (n ¼ 2) at four Z. marina sites
10 Zostera marina Zostera Fish
5 sites
0 Density Length Biomass Species Density Biomass
20 30 40 50 60 70 80 90 100 110 120 130
Brofjorden 230 47 172 21 463 1937
Length, mm Lindholmen 100 30 44 20 785 1086
Finnsbo 380 22 71 20 207 1133
Fig. 5. Number of individuals per length-class of juvenile plaice (Pleuronectes
˚ ¨
Gaso 250 40 141 24 797 2260
platessa) at Zostera marina and non-Z. marina sites.
L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132 131
Table 6 This investigation included areas with different characteris-
Density (individuals sampleÿ1) and biomass (g wet wt. sampleÿ1) of macro- tics in terms of exposure, sediment type and vegetation struc-
crustaceans at Zostera marina and non-Zostera marina sites during day and
night samplings
ture. However, organic content of the sediment, as well as
Zostera shoot density, blade length and biomass, of the studied
Sites Zostera marina Non-Zostera marina
Zostera marina beds in 2004 were within the range of what has
Day Night Day Night previously (1982e1990) been reported for the Swedish coastal
Density region (Baden and Pihl, 1984; Baden and Bostrom, 2001).
¨
Athanas nitescens 0.8 1.3 Thus, there are no indications that the characteristics of the ex-
Carcinus maenas 11.3 24.0 29.8 26.8
Crangon crangon 11.3 33.3 53.3 171.8
tant Z. marina meadows have changed over the last two dec-
Gammaridae 0.0 0.0 1.5 1.3 ades. The number of species and density of fish recorded in
Macropodia rostrata 20.0 12.8 0.3 2.5 this study are also in accordance with what have previously
Mysidae 0.3 0.0 9.3 1.5 been found in surveys of Z. marina beds on the Swedish Ska-
Palaemon adspersus 24.0 134.3 155.8 236.0 gerrak coast. Investigations carried out during June between
Palaemon elegans 17.0 75.8 113.3 86.3
Pagurus bernhardus 0.0 0.0 0.5 0.0
2000 and 2004, including 18 Z. marina beds from three costal
regions of the Swedish west coast, estimated number of spe-
Total density 83.8 280.0 364.3 527.3
cies and densities of fish per standard haul to 14.3 and 643, re-
Biomass 0.0 0.0 0.0 spectively (Anders Svenson, personal communication). The
Athanas nitescens 0.0 0.0 0.1 0.3 corresponding figures for number of species and density of
Carcinus maenas 62.0 339.5 120.3 254.0 fish in the present study was 16.2 and 389, respectively. There-
Crangon crangon 6.3 17.8 22.8 76.5 fore, the result from this study concerning both vegetation and
Gammaridae 0.0 0.0 0.1 0.1
fish could be considered representative for Z. marina beds in
Macropodia rostrata 14.8 7.8 0.1 2.8
Mysidae 0.0 0.0 0.7 0.3 the archipelago, and the findings would be expected to be gen-
Palaemon adspersus 12.7 64.5 57.4 153.6 erally applicable for the Swedish west coast.
Palaemon elegans 6.0 45.1 45.3 49.5
Pagurus bernhardus 0.0 0.0 1.0 0.0
Total biomass 101.8 474.6 247.6 536.9 Acknowledgments
This project is a part of the research program MARBIPP
stages. Other gadoids, such as Merlangius merlangus and Pol- (Marine Biodiversity: Patterns and Processes) funded by the
lachius virens, may also use vegetated coastal habitats as nurs- Swedish Environmental Protection Agency, which are kindly
ery grounds, and in our study juvenile of these two species acknowledged. We also thank Andreas Wikstrom for valuable
¨
were exclusively found in Z. marina beds. Thus, Z. marina assistance during field sampling.
beds are essential habitats during the recruitment process for
gadoids, and losses of seagrass will most likely reduce the
nursery function of the coastal zone for these commercial im- References
portant species.
In an attempt to make a rough estimate of the potential Arntz, W.E., 1973. Periodicity of diel food intake of cod Gadus morhua in the
effect of loss of seagrass on cod recruitment the following ex- Kiel bay. Oikos 15, 138e145.
ample is given: the present (2003e2004) spatial extent of Zos- Baden, S., Pihl, L., 1984. Abundance, biomass and production of mobile epi-
benthic fauna in Zostera marina (L.) meadows, western Sweden. Ophelia
tera marina in the Swedish Skagerrak archipelago has been 23, 65e90.
assessed to be around 130 kmÿ2 (Johan Stal, personal commu-
˚ Baden, S., Bostrom, C., 2001. The leaf canopy of seagrass beds: fauna com-
¨
nication). According to Baden et al. (2003) the distribution of munity structure and functional in a salinity gradient along the Swedish
Z. marina decreased by 60% over the last two decades, which coast. In: Reise, K. (Ed.), Ecological Comparisons of Sedimentary Shores.
would equal a total loss of 190 kmÿ2 (from 320 to 130 kmÿ2) Ecological Studies 151. Springer Verlag, pp. 213e236.
´
Baden, S., Gullstrom, M., Lunden, B., Pihl, L., Rosenberg, R., 2003.
¨
of the seagrass habitat. In our study, mean density of juvenile Vanishing seagrass (Zostera marina) in Swedish coastal waters. Ambio
(0-group) cod was estimated to be 17 individuals per 500 mÿ2 32, 374e377.
haul in Z. marina meadows. At non-Zostera sites, densities Beck, M.W., Heck Jr., K.L., Able, K., Childers, D.L., Eggelston, D.B.,
were estimated to be 0.6 individual per haul. Hence, using Gillanders, B.M., Halpern, B., Hays, C.G., Hoshino, K., Minello, T.J.,
the difference in density of juvenile cod (16.4 individuals Orth, R.J., Sheridan, P.F., Weinstein, M.P., 2001. The identification, con-
servation and management of estuarine and marine nurseries for fish and
per haul) between Zostera and non-Zostera habitats, the loss invertebrates. BioScience 51, 633e641.
in Z. marina meadows over the last two decades could have ˚
Borg, A., Pihl, L., Wennhage, H., 1997. Habitat choice of juvenile cod (Gadus
resulted in a reduction in recruitment of 6.3 million juvenile morhua L.) on sandy soft bottoms with different vegetation types. Helgo-
cod each year. This is of the same order of magnitude as the lander Meeresunersuchungen 51, 197e212.
¨
estimated combined effect of seal predation and mortality Bostrom, C., Baden, S.P., Krause-Jensen, D., 2003. The seagrasses of Scandi-
¨
navia and the Baltic Sea Chap. 1. In: Green, E.P., Short, F.T. 2003. World
due to by-catches in the commercial eel fyke-net fishery atlas of Seagrasses: present status and future conservation. Prepared by
(Mats Ulmestrand, personal communication) in the archipela- UNEP World Conservation Monitoring Centre. University of California
go of the Swedish Skagerrak. Press. Berkeley, USA. ISBN 0-520-24047-2. 298 pp.
132 L. Pihl et al. / Estuarine, Coastal and Shelf Science 67 (2006) 123e132
Brind’Amour, A., Boisclair, D., 2004. Comparison between two sampling Jenkins, G.P., May, H.M.A., Wheatley, M.J., Halloway, M.G., 1997.
methods to evaluate the structure of fish communities in the littoral zone Comparison of fish assemblages associated with seagrass and adjacent un-
of a Laurentian lake. Journal of Fish Biology 65, 1372e1384. vegetated habitats of Port Philip Bay and Corner inlet, Victoria, Australia,
Browder, J.A., Restrepo, V.R., Rice, J.K., Robblee, M.B., Zein-Eldin, Z.Z., with emphasis on commercial species. Estuarine, Coastal and Shelf Sci-
1999. Environmental influences on potential recruitment of pink shrimp, ence 44, 569e588.
Farfantepanaeus duorarum, from Florida Bay nursery grounds. Estuaries Lazzari, M.A., 2002. Epibenthic fishes and decapod crustaceans in northern
22, 484e499. estuaries: a comparison between vegetated and unvegetated habitats in
Clarke, K.R., 1993. Non-parametric multivariate analyses of changes in com- Maine. Estuaries 25, 1210e1218.
munity structure. Australian Journal of Ecology 18, 117e143. Moller, P., Pihl, L., Rosenberg, R., 1985. Benthic faunal energy flow and bi-
¨
Clarke, K.R., Warwick, R.M., 1994. Change in Marine Communities: An Ap- ological interaction in some shallow bottom habitats. Marine Ecology
proach to Statistical Analysis and Interpretation. Natural Environment Re- Progress Series 27, 109e121.
search Council, UK, 144 pp. Orth, R.J., Heck Jr., K., van Montfrans, J., 1984. Faunal communities in sea-
Deegan, L.A., Wright, A., Ayvazian, S.G., Finn, J.T., Golden, H., grass beds: a review of the influence of plant structure and prey character-
Merson, R.R., Harrison, J., 2002. Nitrogen loading alters seagrass ecosys- istics on predatoreprey relationships. Estuaries 7, 339e350.
tem structure and support of higher trophic levels. Aquatic Conservation: Orth, R.J., van Montfrans, J., Lipcius, R.N., Metcalf, K.S., 1996. Utilization of
Marine and Freshwater Ecosystems 12, 193e212. seagrass habitat by the blue crab, Callinectes sapidus Rathbun, in Chesa-
Edgar, G.J., Shaw, C., 1995. The production and trophic ecology of shallow- peake Bay: a review. In: Kuo, J., Phillips, R.C., Walker, D.I., Kirkman, H.
water fish assemblages in southern Australia: 1. Species richness, size- (Eds.), Seagrass Biology. Proceedings of an International Workshop,
structure and production of fishes in Western Port, Victoria. Journal of Ex- Rottnest Island, Western Australia. Western Australian Museum, Perth,
perimental Marine Ecology and Biology 194, 53e81. pp. 213e224.
Ferrell, D.J., Bell, J.D., 1991. Differences among assemblages of fish associ- Parsley, M.J., Palmer, D.E., Burkhardt, R.W., 1989. Variation in capture effi-
ated with Zostera caricorni and bare sand over a large spatial scale. Marine ciency of a beach seine for small fishes. North American Journal of Fish-
Ecology Progress Series 72, 15e24. eries and Management 9, 239e244.
Field, J.G., Clarke, K.R., Warwick, R.M., 1982. A practical strategy for ana- Petersen, J.C.G., 1914. Om bændeltangens (Zostera marina) aars-produktion i
lyzing multispecies distribution patterns. Marine Ecology Progress Series de danske farvande. Kap. X i Jungersen, H.F.E. og Warming, Eug. Mind-
8, 37e52. ˚
eskrift i anledning af hundredaret for Japetus Steenstrups fødsel. I Kom-
Fonds, M., 1973. Sand gobies in the Dutch Wadden Sea (Pomatoschistus, Go- mission hos GEC Gad, København. Bianco Lunos Bogtrykkeri (in Danish).
biidae, Pisces). Netherlands Journal of Sea Research 6, 417e478. Pierce, C.L., Rasmussen, J.B., Leggett, W.C., 1990. Sampling littoral fish with
Gibson, R.N., Pihl, L., Burrows, M.T., Modin, J., Wennhage, H., Nickell, L.A., a seine: corrections for variable capture efficiency. Canadian Journal of
1998. Diel movements of juvenile plaice (Pleuronectes platessa) in Fisheries and Aquatic Sciences 47, 1004e1010.
relation to predators, competitors, food availability and abiotic factors on Pihl, L., 1982. Food intake of young cod and flounder in a shallow bay on the
a microtidal nursery ground. Marine Ecology Progress Series 165, 145e Swedish west coast. Netherlands Journal of Sea Research 15, 419e432.
159. Pihl, L., 1989. Abundance, biomass and production of juvenile flatfish in
Gotceitas, V., Fraser, S., Brown, J.A., 1997. Use of eelgrass beds (Zostera ma- southeastern Kattegat. Netherlands Journal of Sea Research 24, 69e81.
rina) by juvenile Atlantic cod (Gadus morhua). Canadian Journal of Fish- Pihl, L., Rosenberg, R., 1982. Production, abundance and biomass of mobile
eries and Aquatic Sciences 54, 1306e1319. epibenthic marine fauna in shallow waters, western Sweden. Journal of
Green, E.P., Short, F.T., 2003. World Atlas of Seagrasses: Present Status and Experimental Marine Biology and Ecology 57, 273e301.
Future Conservation (prepared by UNEP World Conservation Monitoring Pihl, L., Ulmestrand, M., 1993. Migration pattern of juvenile cod, Gadus
Centre). University of California Press, Berkeley, USA, ISBN 0-520- morhua, on the Swedish west coast. ICES Journal of Marine Science 50,
24047-2, 298 pp. 63e70.
Heck, K.L., Hays, G., Orth, R.J., 2003. Critical evaluation of the nursery role Pihl, L., Wennhage, H., 2002. Structure and diversity of fish assemblages on
hypothesis for seagrass meadows. Marine Ecology Progress Series 253, rocky and soft bottom shores on the Swedish west coast. Journal of Fish
123e136. Biology 61, 148e166.
Hily, C., van Katjwijk, M.M., den Hartog, C., 2003. Seagrasses of Western Short, F.T., Wyllie-Echeverria, S., 1996. Natural and human-induced distur-
Europe (Chapter 2). In: Green, E.P., Short, F.T. (Eds.), World Atlas of bances of seagrasses. Environmental Conservation 23, 17e27.
Seagrasses: Present Status and Future Conservation. University of Califor- Spalding, M., Taylor, M., Ravilious, C., Short, F., Green, E., 2003. Global
nia Press, Berkeley, USA, ISBN 0-520-24047-2, 298 pp. (prepared by overview: the distribution and status of seagrasses. In: Green, E.P.,
UNEP World Conservation Monitoring Centre). Short, F.T. (Eds.), World Atlas of Seagrasses: Present Status and Future
Howarth, R.W., Anderson, D., Cloern, J., Elfring, J., Hopkinson, C., Conservation. University of California Press, Berkeley, USA, ISBN
Lapointe, B., Malone, T., Marcus, N., McGlathery, K., Sharpley, A., 0-520-24047-2, 298 pp. (prepared by UNEP World Conservation Monitor-
Walker, D., 2000. Nutrient pollution of coastal rivers, bays, and seas. Is- ing Centre).
sues in Ecology 7, 1e15. Tveite, S., 1984. 0-Group cod investigations on the Norwegian Skagerrak
Hughes, J.E., Deegan, L.A., Wyda, J.C., Weaver, M.J., Wright, A., 2002. The coast. In: Dahl, E., Danielssen, D.S., Moksness, E., Solemdal, P. (Eds.),
effects of eelgrass habitat loss on estuarine fish communities of southern The Propagation of Cod Gadus morhua L.. Flødevigen. Rapport 1,
New England. Estuaries 25, 235e249. 581e590 (ISSN 033-2594).
Hughes, A.R., Bando, K.J., Rodriguez, L.F., Williams, S.L., 2004. Relative ef- Weaver, M.J., Magnuson, J.J., Clayton, M.K., 1993. Analyses for differentiat-
fects of grazers and nutrients on seagrasses: a meta-analysis approach. Ma- ing littoral fish assemblages with catch data from multiple sampling gears.
rine Ecology Progress Series 282, 87e99. Transactions of the American Fisheries Society 122, 1111e1119.
Jackson, E.L., Rowden, A.A., Attrill, M.J., Bossey, S.J., Jones, M.B., 2001. Wennhage, H., Pihl, L., 2002. Fish feeding guilds in shallow rocky and soft
The importance of seagrass beds as habitat for fishery species. Oceanogra- bottom areas on the Swedish west coast. Journal of Fish Biology 61,
phy and Marine Biology: An Annual Review 39, 269e303. 207e228.