Temporal, spatial and tidal influences on benthic and demersal fish abundance in the Forth estuary

Temporal, spatial and tidal influences on benthic and demersal fish abundance in the Forth estuary

Estuarine, Coastal and Shelf Science 58 (2003) 211–225 Temporal, spatial and tidal influences on benthic and demersal fish abundance in the Forth estua...

327KB Sizes 0 Downloads 17 Views

Estuarine, Coastal and Shelf Science 58 (2003) 211–225

Temporal, spatial and tidal influences on benthic and demersal fish abundance in the Forth estuary M.F.D. Greenwooda,*, A.S. Hillb,1 a Department of Biological Sciences, University of Stirling, Stirling FK9 4LA, UK Tidal Waters Section, Scottish Environment Protection Agency East Region, Clearwater House, Heriot–Watt Research Park, Avenue North, Riccarton, Edinburgh EH14 4AP, UK

b

Received 27 August 2002; accepted 24 February 2003

Abstract The 10 most common demersal and benthic fish species collected during a 1982–2001 Agassiz trawling programme in the mid/ lower Forth estuary, east Scotland, were assessed for possible influences of trawl site, month of sampling and tide height on abundance. All species were seasonal in their occurrence in the study area, with trends generally similar to those shown in other temperate European estuaries. Six species (whiting (Merlangius merlangus), dab (Limanda limanda), pogge (Agonus cataphractus), fatherlasher (Myoxocephalus scorpius), ÔgobiesÕ (Pomatoschistus spp.) and sea snail (Liparis liparis)) exhibited high winter and low summer abundance while plaice (Pleuronectes platessa), flounder (Platichthys flesus) and eelpout (Zoarces viviparous) were least numerous in winter and mostly present in summer. Cod (Gadus morhua) were low in abundance in late spring, but did not show pronounced peaks of abundance. Trawl station significantly influenced fish abundance in all but one case (ÔgobiesÕ). There was relatively low abundance of most species at the middle trawl location, Tancred; only plaice, dab and whiting were numerous at this site. The majority of species were abundant at Port Edgar, adjacent to the estuary mouth, which may have reflected the relatively stable salinity conditions at this station. Flounder was most abundant at the mid-estuarine site of Longannet, and this was likely to have been in order to feed on the large mudflats nearby. Dispersal of fish onto inundated intertidal areas is suggested as a possible mechanism explaining the significant decrease in abundance of several species (flounder, pogge, eelpout and ÔgobiesÕ) at high water (HW). The significant increase in abundance of whiting in the study area at HW may have been attributable to immigration of individuals from inshore regions of the Firth of Forth by the action of the flood tide. Interactions of location, month and tide height significantly influenced abundance of most species captured during trawling, though explained less variation in the data than the individual main effects. Ó 2003 Elsevier Ltd. All rights reserved. Keywords: fish; trawl; abundance; generalised linear model; Forth estuary

1. Introduction Many marine fish use estuaries seasonally as adults and as nursery grounds when young. This is not necessarily dependence on the estuarine environment (Potter, Claridge, Hyndes, & Clarke, 1997), merely * Corresponding author. Present address: Florida Fish and Wildlife Commission, Florida Marine Research Institute, 100 8th Avenue SE, St Petersburg, FL 33701, USA. E-mail address: [email protected]fl.us (M.F.D. Greenwood). 1 Present address: Countryside Council for Wales, Maes-Y-Ffynon, Penrhosgarnedd, Bangor, Gwynedd LL57 2DW, UK. 0272-7714/03/$ - see front matter Ó 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0272-7714(03)00071-4

opportunistic utilisation of a near-shore environment offering refuge from predators and increased feeding possibilities (Blaber & Blaber, 1980). Relatively few species are adapted to thrive in conditions such as fluctuating salinity and their resultant physiological demands (Haedrich, 1983). The ichthyofauna present in estuaries varies temporally and spatially, based mainly on migrations of seasonal visitors (e.g. Elliott, O’Reilly, & Taylor, 1990; Maes, Taillieu, van Damme, Cottenie, & Ollevier, 1998), the distribution of fish within estuaries according to salinity tolerance (e.g. Henderson, 1989; Loneragan, Potter, Lenanton, & Caputi, 1986) and sediment or vegetation preferences

212

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225

(e.g. Elliott & Dewailly, 1995; Marshall & Elliott, 1998). Fish distribution in estuaries is mostly driven by fluctuations in environmental (physical) variables as opposed to biological interactions (e.g. Kupschus & Tremain, 2001). The Forth estuary, east Scotland, is a well-mixed inlet of the North Sea, with mean tidal range averaging 5.0 and 2.5 m on spring and neap tides, respectively (Webb & Metcalfe, 1987). Fish in the estuary are typical of northeastern Atlantic temperate estuaries. There is little change in the composition of the fish community in the lower section of the estuary, with sites of salinity >30 tending to consist of about 50% estuarine resident species, 30% marine juvenile (MJ—see Table 1 for definition) and 15% marine seasonal (MS) species (Elliott et al., 1990). Where salinity starts to fall considerably, the proportion of estuarine resident species increases substantially, while MS species become more numerous than those of the MJ category. Intertidal mud flats comprise approximately 22.6 km2 of the estuary; the largest of these is Kinneil, which at low water (LW) measures 6.42– 5.71 km2 during the spring–neap tidal cycle (McLusky, 1982). Use of the intertidal flats at Skinflats and Kinneil (Fig. 1) has been shown for a variety of species including flounder (Platichthys flesus (L.)), plaice (Pleuronectes platessa (L.)), whiting (Merlangius merlangus (L.)), gobies (Pomatoschistus spp.), eelpout (Zoarces viviparous (L.)), cod (Gadus morhua (L.)), pogge (Agonus cataphractus (L.)) and common dab (Limanda limanda (L.)) (see Table 1 for species characteristics) (Elliott & Taylor, 1989). Abundance of flounder was shown to vary considerably spatially and temporally: abundance was high in the lower estuarine subtidal and intertidal areas in summer, whereas the highest numbers in the upper estuary were observed in winter (Elliott & Taylor, 1989). Elliott et al. (1990) further examined seasonal trends in abundance of marine species from 1981 to 1989 and found that winter maxima were generally coupled with summer minima some 6 months later in dab, whiting and cod. Plaice peaked in July but did not possess clear minima, while the estuarine resident eelpout showed bimodal annual peaks of abundance in spring and summer. Comparisons of seasonal changes in abundance of fish in the Forth and Tyne estuaries showed similarities in the cases of cod, whiting and flounder, while plaice, dab and eelpout differed somewhat between the locations (Pomfret, Elliott, O’Reilly, & Phillips, 1991). Long term abundance trends in the mid-/lower Forth estuary were analysed over the period 1982–2001 and suggested that while there were significant differences in annual abundances of fish captured, most species appeared to have population densities that were at equilibrium in the estuary (Greenwood, Hill, & McLusky, 2002). The exceptions included fatherlasher, which increased marginally in abundance, and significant decreases of whiting and eelpout. In the case of the latter

Table 1 Species studied and classification according to Elliott and Dewailly (1995) Species Whiting, Merlangius merlangus (L.) Cod, Gadus morhua (L.) Plaice, Pleuronectes platessa (L.) Flounder, Platichthys flesus (L.) Common dab, Limanda limanda (L.) Pogge, Agonus cataphractus (L.) Fatherlasher (bull rout), Myoxocephalus scorpius (L.) Eelpout (viviparous blenny), Zoarces viviparus (L.) ÔgobiesÕ, Pomatoschistus spp. Common sea snail, Liparis liparis (L.)

Ecological Vertical Substratum Feeding guild distribution preference guild MJ

Demersal

F

I,F

MJ MJ

Demersal Benthic

F F

I,F I

ER

Benthic

F

I,F

MJ

Benthic

S

I,F

ER

Benthic

F

I

ER

Benthic

F,V

I,F

ER

Benthic

M,V

I

ER

Benthic

S

I

ER

Benthic

M

I,F

MJ, marine juvenile migrant species, which use the estuary primarily as a nursery ground, usually spawning and spending much of their adult life at sea but often returning seasonally to the estuary; ER, truly estuarine resident species, which spend their entire lives in the estuary. F, soft bottom (sand, mud, and/or fine gravel); S, solely sand; V, living above/among vegetation or seaweed on a certain bottom type; M, mixed/various bottom (indiscriminate). I, feed on invertebrates; F, feed on fish.

species this decline may be attributable to a general trend of increasing water temperatures negatively impacting upon reproductive success. The present study utilised the same data as those of Greenwood et al. (2002), which were a continuation of data analysed by Elliott et al. (1990) and Elliott and Taylor (1989). The aim of this study was to use the 20year dataset to examine closely the abundance of the 10 most commonly captured demersal and benthic fish species at fixed mid-/lower estuarine trawl stations in relation to environmental factors. These factors were temporal (month of sampling), tidal (high or low water—HW/LW) and spatial (three sites moving upstream from Port Edgar at the estuary mouth and ending at Longannet in the lower section of the midestuary (Fig. 1)). The seasonality of certain species was shown graphically by Elliott et al. (1990); the present study aimed to extend this analysis for additional lessstudied but common species and to provide statistical validation for seasonal trends by assessing the 20 years of collected data. Possible concentration of fish in the estuarine channel at LW and movement into intertidal areas at HW was investigated by assessment of the tidal factor. The trawl stations examined differ principally in their observed salinity ranges; almost fully marine

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225

213

Fig. 1. Location map of the study area. L, Longannet trawl station, T, Tancred trawl station, P, Port Edgar trawl station (only Longannet station data were used in the present study); ÔlwostÕ, ÔhwostÕ: low water/high water of spring tides, respectively. Dashed line (- - -) indicates area of dredging influence; d.c., dredged channel; s.g., spoil dumping ground.

conditions exist for most of the time at Port Edgar, while salinity at Longannet may be that of full strength sea water at HW and almost zero at LW (Pomfret et al., 1991). Thus the study aimed to assess if fish, particularly those of marine origin, exhibited a gradual reduction in abundance with movement upstream or otherwise.

giving a total of 30 trawls per year, i.e. 10 trawls at each site. Fish were sampled in January/February, March/ April, June/July, September/October and December, with minor exceptions. Only 24 trawls were undertaken in 1986 because samples were not taken in January, therefore the total number of trawls undertaken from 1982 to 2001 was 594.

2. Materials and methods

2.2. Data analysis

2.1. Study site and sampling techniques

The 10 most commonly captured benthic and demersal species (Table 1) were assessed for possible influences of tidal state, month of sampling and trawl station on abundance. Only benthic and demersal species were selected for analysis as the gear was unlikely to have captured pelagic fish consistently. Counts of fish captured in each trawl were highly skewed, with most values being low and relatively few trawls having large abundances of individuals. Each of the 10 species analysed possessed a negative binomial count distribution (negative binomial fitting function for S-PlusÒ 6 provided by D.J. Shaw, Veterinary Epidemiology Group, University of Edinburgh). Log transformation of data with this type of distribution to allow classical linear techniques (e.g.

Trawling took place at three lower Forth estuary stations, Port Edgar, Tancred and Longannet (Fig. 1; Table 2), from January 1982 to December 2001, using the SEPA survey vessel, the Forth Ranger. Each sample consisted of an approximately 0.8 km haul with a 2 m Agassiz trawl of 15 mm stretched mesh diameter. Samples were taken during spring tides. All fish collected were identified to species, counted and measured for total length (LT) and wet mass (g). Gobies were classed as Pomatoschistus spp. and were probably mostly Pomatoschistus minutus (Pallas). Trawling was undertaken at HW and LW of spring tides in 5 months of the year,

214

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225

Table 2 Details of trawl stations Station

Zone

Depth (m)

Salinity

Substratuma

Port Edgar Tancred Longannet

Lower estuary Lower estuary Middle estuary

5–15 7–8 5.5–6.5

>25 >25 0–34

Medium sand Mud Medium sand and mud

a

From Pomfret et al. (1991). Unpublished data.

ANOVA) leads to greater possibility of type I and II statistical errors, particularly with sample sizes >100 (Wilson, Grenfell, & Shaw, 1996). Generalised linear modelling (GLM) of untransformed data was conducted utilising a negative binomial error structure and log link function in S-Plus 6 (Insightful Corp., 2001) (see Crawley (1993, Chap. 10) for a good introduction to GLM analysis). Maximal models (i.e. initial models with all predictors included) included fish abundance per trawl as response against predictors that included the main effects of month (five levels corresponding to the periods described above), station (three levels, as above) and tide (HW/LW), as well as all two-way main effect interactions and the three-way month–station–tide interaction. The null hypothesis in all cases was that there would be no significant relationship between any predictor and fish abundance in each trawl. Minimal models (i.e. final models with only statistically significant predictors left) were determined by stepwise deletion that assessed the increase in deviance (discrepancy between observed and model-fitted values) attributable to removal of individual predictors through v2 goodness of fit tests comparing the deviance of the models with and without the particular term included (Crawley, 1993). Predictors were deemed significant at the P < 0:05 probability level. Significant

interaction terms were excluded from models if any of the associated main effects were not statistically significant (Nelson & Leffler, 2001). Models were assessed to have represented the data adequately when an estimate of the dispersion parameter (i.e. ratio of residual deviance to residual degrees of freedom) was close to one (Crawley, 1993; Wilson et al., 1996). The influence of significant main effects on model parameters was assessed by inspection of plots of their contribution to the linear predictor of the minimal models (Goni, Alvarez, & Adlerstein, 1999), while significant interactions’ features were assessed by examining plots of mean abundance.

3. Results Details of all benthic and demersal species captured in the trawl studies are given by Greenwood et al. (2002). Null hypotheses of no relationship between predictors and fish abundance data were rejected to varying degrees according to species (Table 3). There was rejection of the null hypothesis for five of seven predictors in the cases of whiting and pogge, while it was accepted in four out of seven predictors for ÔgobiesÕ and fatherlasher. The minimal models obtained in the present study generally explained a relatively low proportion of the null models’ deviance in each case, ranging from approximately 11% in plaice to 38% in dab (Table 3, column 2). Data were adequately represented by the negative binomial models, however, with estimates of dispersion parameters generally close to one (range: 0.54–1.16). Main effects accounted for the greatest proportion of deviance in each model. Month was significant in all 10 models, and explained the greatest deviance per degree of freedom in whiting, dab, pogge,

Table 3 Results of GLM of Agassiz trawl-caught fish, lower Forth estuary, 1982–2001 Terms in models Main effects

Interactions

Station (2 d.f.)

Tide (1 d.f.)

Month (4 d.f.)

Station–tide (2 d.f.)

Station–month (8 d.f.)

Species

Residual deviance (null deviancce)

Deviance

Sig.

Deviance

Sig.

Deviance

Sig.

Deviance

Sig.

Deviance

Sig.

Whiting Cod Plaice Flounder Dab Pogge Fatherlasher Eelpout Gobies Sea snail

668.67 453.59 608.71 623.51 359.88 652.74 417.61 635.25 529.24 314.69

8.33 80.38 30.36 154.71 17.56 22.42 128.56 62.65  25.30

* *** *** *** *** *** *** *** NS ***

6.11   14.88  14.47  10.61 4.15 

* NS NS *** NS *** NS ** * NS

24.83 23.31 30.98 81.32 144.76 56.68 24.06 78.29 98.21 141.96

*** *** *** *** *** *** *** *** *** ***

8.04     7.44 ()  () 

* NS NS NS NS * () NS () NS

28.83 25.15 19.98 35.56 27.76 28.75  53.65  2.04

*** ** * *** *** *** NS *** NS *

(755.99) (575.02) (682.42) (961.84) (584.03) (783.21) (568.34) (821.80) (631.60) (499.33)

All models are based on negative binomial error structure. Null models possessed 593 d.f. Abbreviations: resid. dev., residual deviance in minimal model; null dev., deviance in null model; Sig., statistical significance (*P < 0:05; **P < 0:01; ***P < 0:001; NS, P > 0:05). See Figs. 2–11 for plots of relationships between response and predictors. Tide–month and tide–month–station interactions not significant in any model, so not shown in table. ( ) Indicates significant interaction dropped from model due to lack of significance of a main effect in the interaction.

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225

ÔgobiesÕ and sea snail, while ranking second behind the station factor in cod, plaice, flounder, fatherlasher and eelpout (Table 3). Station was statistically significant in all but one minimal model (ÔgobiesÕ), and generally ranked first or second in importance, while tide was present in models of whiting, flounder, pogge, eelpout and ÔgobiesÕ. The station–month interaction was present in eight models, and usually explained least deviance per degree of freedom when compared to main effects; the only other significant predictor was station–tide in whiting and pogge minimal models. Various features of the mid-lower Forth estuarine fishes’ distribution were evident from examination of main effects remaining in minimal models following stepwise deletion. GLMs involving the station factor suggested highest abundances of cod, pogge, fatherlasher and eelpout at Port Edgar (Figs. 2a, 3a, 4a and 5a); whiting numbers were similarly high at this site and at Tancred (Fig. 6a), while sea snail abundance was greatest at Port Edgar and Longannet (Fig. 7a). Only plaice were assessed to be most abundant at Tancred (Fig. 8a), though dab showed high abundance at both Tancred and Longannet (Fig. 9a). Flounder was the sole species to be most numerous at Longannet (Fig. 10a). Clearer distinctions were evident for minimal abundances, with Tancred being the site least favoured by cod, flounder, pogge, fatherlasher, eelpout and sea snail. Port Edgar possessed lowest abundances of plaice and dab, while whiting were least numerous at Longannet. Fish abundances generally appeared to be enhanced in the sampling area during LW, as shown in the models of pogge, eelpout, flounder and ÔgobiesÕ (Figs. 3b, 5b, 10b and 11a), with the exception being whiting, whose numbers significantly increased at HW (Fig. 6b). Seasonal influences on fish abundance in the whole mid-lower estuarine study area were generally clear from the GLMs: species were able to be classified as being of high winter/low summer abundance (pogge, fatherlasher, whiting, sea snail, dab and ÔgobiesÕ; Figs. 3c, 4b, 6c, 7b, 9b and 11b) or else low winter/high summer abundance (eelpout, plaice and flounder; Figs. 5c, 8b and 10c). Only cod did not show clear seasonal progressions involving 6-month separation of annual maximum and minimum numbers—lowest abundance was in spring (March/April), with the rest of the year being similar (Fig. 2b). Significant interaction terms in the GLMs enabled some of the associations between station and tide or month to be examined more closely, as summarised in Table 4.

4. Discussion The generalised linear models produced in the present study explained on average just over 20% of the total variation in the data. Higher values have been obtained in

215

Fig. 2. Results of GLM of cod data. Main effect plots (a,b) illustrating the contribution of the different factor levels (fitted to average zero) to the linear predictor (Goni et al., 1999); bars are proportional to the number of observations at each factor level, broken lines are 2 SE. Interaction plots (c) showing the mean abundances for each combination of factors; data points and error bars omitted for clarity. Level codes for plots: station: 1, Port Edgar; 2, Tancred; 3, Longannet; tide: 1, LW; 2, HW; month: 1, January/February; 2, March/April; 3, June/July; 4, September/October; 5, December.

216

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225

Fig. 3. Results of GLM of pogge data. See legend to Fig. 2 for explanation of terms.

analyses of commercial CPUE data, e.g. 39% for sole (Solea solea (L.)) (Large, 1992) and 63% for hake (Merluccius merluccius (L.)) (Goni et al., 1999). Comparison of the values obtained in these studies must take into account the inherent differences in the type of data involved, for the present study was based on low intensity scientific sampling and so was liable to greater variation. Undertaking replicate trawling near each station would have been likely to reduce sampling error. Changes in fish

utilisation of the mid-/lower estuary over the 20 years of sampling, attributable to environmental factors (e.g. reductions in effluent emissions and increasing water temperature), may also have increased data variability. Proportion of variation explained in the present study compares well with values of 15–35% in models of faecal egg counts in Soay sheep (Ovis aries L.) of differing sex, age and year of sampling (Wilson et al., 1996), especially given that both studies utilised a negative binomial error

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225

217

Fig. 4. Results of GLM of fatherlasher data. See legend to Fig. 2 for explanation of terms.

structure for residuals. Proportion of total variation explained in the present study would have been further increased had ÔyearÕ been introduced into the models; this was not undertaken as trends in abundance over the 20year study period were assessed elsewhere (Greenwood et al., 2002), and the many levels of the ÔyearÕ factor would

have introduced more parameters into models (Crawley (1993) recommends no more than n/3 parameters in maximal models, where n is the number of samples). The objective of the present study was to examine within-year influences on fish abundance (cf. Nelson & Leffler, 2001), and the models produced fulfilled this purpose.

Fig. 5. Results of GLM of eelpout data. See legend to Fig. 2 for explanation of terms.

218

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225

Fig. 6. Results of GLM of whiting data. See legend to Fig. 2 for explanation of terms.

Seasonal trends of abundance in the species studied were generally similar to those in other northeast Atlantic estuaries (Table 5). The present study utilised some of the same data as previous investigations in the Forth estuary (i.e. Elliott et al., 1990; Elliott & Taylor, 1989), though omitting the additional lower estuarine stations of Blackness and Bo’ness (trawled until 1998) and data from additional months when monthly sampling was undertaken (1981–1986). This selective

use of some of the previously analysed data undoubtedly contributed to differences between the present study’s seasonal trends and those highlighted by Elliott et al. (1990). Whiting and dab changes in abundance with season were similar to the 1980s data, whereas those of cod and plaice differed somewhat. Plaice numbers peaked in the summer, as shown by Elliott et al. (1990), but minima were more clearly observed from January to April in the present study; there were no

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225

Fig. 7. Results of GLM of sea snail data. See legend to Fig. 2 for explanation of terms.

well-defined minima in the previous study. Elliott et al. (1990) showed cod abundance to peak in February and to be high from November to March, while being considerably lower from April to September. The present study showed only a clear minimum in March/ April, with abundances in the other four months being relatively similar. This difference arose from increased

219

Fig. 8. Results of GLM of plaice data. See legend to Fig. 2 for explanation of terms.

abundance of cod that occurred in the Forth from 1990 onwards, and which was more evenly spread through the remaining months of the year (Greenwood et al., 2002). Use of the Forth estuary by juvenile marine species is likely to be for several reasons: shelter from predators in a turbid area where predators are less common, high food concentrations, and lower osmotic stress (Blaber & Blaber, 1980; Elliott et al., 1990). The

220

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225

Fig. 9. Results of GLM of dab data. See legend to Fig. 2 explanation of terms.

for

high abundance of flounder in the mid-estuary during June/July was exactly that noted by Elliott and Taylor (1989) in both subtidal and intertidal areas. Of particular note was the information provided on the seasonality of lesser-studied non-commercial species. The large number of pogge from autumn to late winter was similar to observed trends in the mid-Thames estuary, suggested to be evidence of an inshore spawning migration (Power & Attrill, 2002). Thames pogge observe a second peak in abundance during March/ April, possibly to take advantage of water temperatures higher than the sea in order to increase growth rates and

associated fecundity (Power & Attrill, 2002). No such trend was obvious from the Forth data, indeed abundance was at its lowest in March/April. A small population of pogge may be resident in the Forth throughout the year, with numbers being augmented in the colder months by immigration of spawning individuals. Abundance of fatherlasher also observed clear seasonality with maxima in winter (December–January) and a low in summer, suggesting that pogge and fatherlasher abundances may be influenced by similar mechanisms. Sea snail in the mid-lower Forth estuary become more numerous in the estuary in June/July and are most abundant in September/October, then decline through December and January to very low values in March/April. This is somewhat earlier than the trend shown in the outer Severn estuary, where the first sea snail arrive in October, peak in abundance from December to January and remain until March at the latest (Henderson & Seaby, 1999). The Severn sea snail are semelparous and therefore somewhat different to those of the Forth, and have been shown to grow more quickly at lower temperatures. This may be due to several factors including greater penetration into shallow water at lower temperatures leading to increased feeding opportunities on brown shrimp (Crangon crangon L.) in intertidal and adjacent areas, decreased swimming ability of prey, and possibly physiological adaptation to enhanced growth at low temperature (Henderson & Seaby, 1999). The clear summer/winter contrast in abundance of ÔgobiesÕ may be explained by the semelparous nature of the species (Wheeler, 1969), assuming most individuals captured were sand goby. Minimum abundance was in June/July, prior to presumed hatching of young of the year. By September/ October there was an increase in abundance, which was elevated to very high levels in December/January as that year’s new recruits were caught in trawls. In late spring (April), numbers were reduced due to overwintering mortality and the onset of spawning. Death of adults following spawning would have led to very low abundances in June/July, completing the cycle. Distribution of the fish within the estuary appeared to reveal several significant differences between species. Only ÔgobiesÕ were distributed relatively evenly throughout the study area, shown by station not being a significant factor in GLM of this taxon. Data from the earlier Forth work of Elliott and Taylor (1989) also suggested such a trend. Pomfret, Turner, and Phillips (1988) found sand goby in greatest abundance at the most seaward site at Jarrow in the Tyne, an estuary in the same geographic region as the Forth; 5 km upstream at Hebburn numbers were only a tenth of the lower estuarine values. Such a difference between estuaries may be due to the Tyne possessing a large middle estuary of some 25 km length and only 3.5 km of marinedominated lower section, compared with 18 and 14 km,

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225

221

Fig. 10. Results of GLM of flounder data. See legend to Fig. 2 for explanation of terms.

respectively, for the middle and lower sections of the Forth estuary (Pomfret et al., 1991). Middle sections of estuaries have the greatest salinity range, so sand gobies may prefer the more seaward sites, as suggested by Wheeler (1969). In the Forth only the Longannet station

was in the middle estuary, and salinities at this site are often fully marine at HW. Six of the nine species that included station as a significant factor in GLMs were generally in very low abundance at Tancred. Such a trend is curious, given the significant richness of the

Fig. 11. Results of GLM of ÔgobiesÕ data. See legend to Fig. 2 for explanation of terms.

222

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225

Table 4 Summary of features illustrated by significant interactions from GLM of mid-lower Forth estuary Agassiz trawl data, 1982–2001 Species

Station–tide

Station–month

Whiting

Cod

LW: little difference between stations; HW: large numbers at PE, decreasing uniformly through T to L (see Fig. 6d) –

Plaice



Flounder



Dab



Pogge

LW: numbers at T much lower than at PE and L (PE > L); HW: similar trend as at LW, less pronounced (L > PE) (see Fig. 3d)

Eelpout



Sea snail



Similar levels at all stations in January/February; # at L in March/April, at PE and T in June/July; " from summer to winter evident, most pronounced at PE (Fig. 6e) Sharp # from January/February to low in March/April at L, followed by gradual " to plateau in autumn/winter; sharp # from March/April to June/July at PE, followed by sharp " to peak in winter; T similar to PE but less pronounced (Fig. 2c) Highest abundance at all stations in summer; sharp # at all stations from summer to winter, annual lows in March/April, except at T which shows earlier " (Fig. 8c) Relatively low abundance at PE and T throughout the year; major summer peak in abundance at L (Fig. 10d) L and T similar: # from low abundance in January/February to minima in summer, then sharp " through autumn to winter peak; PE levels low, with moderate " at same time as other stations (Fig. 9c) All show # from January/February to March/April (similar minima for T and L), PE has a summer minimum; abundance " sharply at L to autumn, then falls off in December, while a more moderate " at T continues through December; PE shows sharper summer–autumn " and similar " through December as at T (Fig. 3e) Lowest abundances at all stations in winter (December–January/ February); sharp " in January/February to peak in March/April at PE and T, then pronounced # from summer to winter; " is delayed until March/April at L, otherwise similar decline towards winter (Fig. 5d) All stations similar: low abundance to summer, then sharp " to autumn peak (particularly at L), then pronounced # in December (Fig. 7c)

LW/HW: low/high water; PE, T, L: Port Edgar, Tancred, Longannet, respectively; ", #: increase, decrease, respectively (in abundance).

benthos at this site, with macrofaunal abundances >10 those of the other trawl stations (Elliott & Kingston, 1987). Consideration of the proximity of dredge and spoil disposal sites (Fig. 1) to the Longannet and Tancred stations, respectively, yields no clues to any potential anthropogenic influence on abundance for the former is closer to disturbance than the latter. Motile species are often not adversely affected by dredge activities (Kennish, 1992), though assessments may be based on overall stock levels as opposed to distributions in relation to disturbance (e.g. Polger, Summers,

Cummins, Rose, & Heimbuch, 1985). In any case, the considerable abundance of whiting, plaice and dab at Tancred suggests that it is suitable for occupation. Substratum preferences of whiting and plaice are similar (Table 1), and are not inconsistent with the muddy bottom at Tancred (Table 2), but dab prefer a sandy substrate and so high abundance at this site appears anomalous. Whiting do not seem to penetrate very far up estuaries in great numbers (Elliott & Taylor, 1989; Maes, van Damme, Taillieu, & Ollevier, 1998) and this explains relatively low abundance at Longannet. The

Table 5 Comparison of seasonal trends assessed in the present study with others from selected northeast Atlantic estuarine systems Forth

Severn

Thames

Zeeschelde

Tyne

Loire

Species

Outer estuary Elliott et al. (1990), Mid-estuary (Henderson & Holmes, Elliott and (Claridge, Potter, 1991; Henderson & Taylor (1989) & Hardisty, 1986) Seaby, 1999)

Mid-estuary (Power & Attrill, 2002; Mid-estuary Power, Attrill, & (Maes et al., Thomas, 2000) 1998)

Lower estuary (Pomfret Mid-estuary et al., 1991) (Robin, 1991)

Whiting Cod Plaice Flounder Dab Eelpout ÔGobiesÕ Pogge Sea snail

3   3 3  – – –

– – (3) 3 3 – – (3) –

  3 3 (3) (3) – – –

(3) – – 3 – –  – (3)

– – –  (3) – – – (3)

– – –  – – (3) – –

– – – 3 – – – – –

3, Compares well, similar duration and timing of peaks/troughs of abundance; (3), reasonable comparison, differs only in timing of peak abundance; , timing and/or number of peaks in abundance differ from present study; –, no data for temporal trends available.

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225

potential for differing water depths influencing fish abundances at the three trawl stations also exists. Pihl and Wennhage (2002) showed that significant declines in total fish density were observed over sandy substrates in a Swedish outer estuarine region when water depth increased from <3 to >6 m; this was attributed to fish preference for shallower areas with greater vegetation cover. Dominant species included several examined in the present study (ranked by biomass: flounder, whiting, fatherlasher, cod and plaice), but inter-estuarine comparison for individual species is not possible due to the pooling of the data for all species together in the Swedish study. Flounder and plaice were shown to be relatively scarce at the deepest trawl station in the mid-/ lower Forth, while whiting, cod and fatherlasher were abundant at this site. Tide height has been shown to influence fish abundance in subtidal estuarine areas where large mudflats exist, by concentrating fish in the subtidal at LW (e.g. Morrison, Francis, Hartill, & Parkinson, 2002). While many abundant species present in the Forth estuary have been shown to use the mudflats (Elliott & Taylor, 1989), tide was shown to be a significant main effect in only five of the 10 models evaluated. There was evidence of a subtidal concentration effect in the cases of flounder, pogge, eelpout and ÔgobiesÕ, with abundances at HW being significantly lower than at LW. Gibson, Robb, Burrows, and Ansell (1996) found significantly more fish captured at LW than HW on a west Scottish sandy beach, including several of the species examined in the present study (whiting, cod, pogge, plaice, fatherlasher, sand goby). Salinities reduced to below optimum levels in the regions of the mid-estuary upstream of those sampled could also have caused a downstream migration into the study area. The significant tide–station interaction in the pogge GLM provides evidence for both a subtidal concentration effect (enhanced abundance at LW at Longannet) and downstream migration to the mouth of the estuary facilitated by the ebb tide (high abundance at Port Edgar at LW). Low abundance of pogge at Tancred at both LW and HW further emphasises the avoidance of this site by this species. Flounder were found at Longannet site in greatest numbers at LW. This agrees with the observation that the species is the most frequently encountered on the estuarine mudflats at Kinneil and Skinflats, having been captured in 94% of intertidal trawls by Elliott and Taylor (1989). Whiting abundance was significantly increased at HW, in contrast to other species. Examination of the nature of the significant tide–station interaction in the whiting model may clarify this effect. Whiting abundance at HW tended to decrease slightly at the Longannet site, in accordance with migration onto the mudflats (Elliott & Taylor, 1989). The very substantial increases in abundance at Port Edgar (and, to a lesser extent, Tancred)

223

from LW to HW may be caused by upstream movement of fish from the lower estuary and Firth of Forth on flood tides. Since shallow, turbid areas are likely to also serve as nursery habitat for MJ species (Blaber & Blaber, 1980), net movement of water into the estuary may facilitate immigration of fish from surrounding areas, followed by emigration on the ebb tide. Clupeids have been shown to move with tidal flow in the Forth estuary (Welsby, Dunn, Chapman, Sharman, & Priestley, 1964), as have rainbow smelt (Osmerus mordax (Mitchill)) in the Mystic River, USA (Haedrich & Haedrich, 1974). Significant month–station interactions occurred in many species’ GLMs and provide insight into spatial differences in abundance that changes with season. Whiting were abundant at all stations in January, with abundances decreasing at the most upstream site (Longannet) by April, with similar decreases in abundance at the lower sites being delayed until June/July. This is consistent with a gradual downstream movement as emigration commenced following habitation of the estuary in the winter months. The species is present in the estuary year-round, something that has been observed in other locations (Humber—Marshall & Elliott, 1998; Severn—Henderson, Seaby, & Somes, 2002), though there is still marked seasonality. While only moderate increases in abundance occurred at Tancred and Longannet following the summer months, there was a great increase in whiting catches at Port Edgar from summer through autumn to winter. Henderson and Holmes (1989) suggested that whiting follow their Crangon crangon prey in the Severn estuary. Reduced salinities in winter lead to enhanced abundance in the lower Severn estuary in winter, for C. crangon prefer salinities >12. Information on seasonal changes in abundance of C. crangon by site in the Forth is lacking, though the mid-lower estuary was noted as having more shrimp than the most seaward areas (Jayamanne & McLusky, 1997). Data from the present study are not dissimilar to those of Henderson and Holmes (1989), for the significant interaction of station– month suggested very high whiting abundance in December at Port Edgar. This could have been explained by whiting moving to areas of greater salinity that C. crangon would prefer in winter. Lack of shrimp and water quality data in the present study means this hypothesis cannot be tested. The initial rise in whiting abundance at Longannet in summer may have consisted of the newly recruited O-group feeding on smaller prey, with subsequent movement to other areas to feed on larger prey. Whiting switch diet from smaller prey items such as the mysid Neomysis integer (Leach) to C. crangon as they grow (Henderson & Holmes, 1989). Feeding is of potential importance in determining species distribution, for species show ontogenetic shifts in diet (Costa & Elliott, 1991) and often migrate as they grow older and larger. Similar mechanisms may operate

224

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225

in cod as in whiting, for the two are ecologically close. Cod appear to exhibit similar trends to whiting, at least at Longannet and Port Edgar, though with differences in timing of seasonal peaks of abundance. Flounder abundance peaked during June/July at Longannet; little seasonal difference in abundance at Port Edgar and Tancred site emphasises the general low utilisation of these sites by this species (see above). Eelpout abundance was relatively low at all sites in January, but then increased sharply at the Port Edgar and Tancred stations in March/April. Evidence for the individuals from Tancred then continuing to migrate up to Longannet in June/July exists, as the increase at the latter station matched the decrease at the former closely. Eelpout feed on the mudflats near Longannet (Poxton, 1987) and this would enhance abundance at the upper site. Numbers at Port Edgar remained high until a sharp decrease after June/July that was mirrored at the Longannet site. Winter abundance at all sites was generally rather low, possibly due to emigration of adults for breeding purposes (Elliott et al., 1990). Haedrich (1983) suggested that differences in the timing of seasonal peaks of occupation of estuaries by fish species would allow the carrying capacity of the environment to be more fully realised. Some evidence for this exists in the Forth, based on the data from the present study. Peaks in abundance of potentially competitive juvenile plaice and dab are offset by some 6 months. Where peaks in abundance collide between competitors for similar resources, e.g. plaice and flounder, there is evidence that spatial segregation may occur, for plaice tend to be more abundant at the Tancred site which flounder occupy least. Dab and flounder also use similar food resources; winter peaks in abundance of dab suggest that they could utilise the Longannet site in the relative absence of flounder, which are most abundant in summer. Poxton (1987) cites unpublished works showing that the two common juvenile gadoid species both fed mostly on abundant Crangon crangon in the estuary, but alternative prey were Neomysis integer for whiting and pink shrimp Pandalus montagui Leach in cod. This illustrates how cooccurrence of ecologically very similar species can be facilitated by partitioning of differing food resources. In conclusion, the present study reinforces the great importance of seasonality in estuarine fish species, whether residents or opportunistic MJs (Elliott et al., 1990). Distribution within the lower Forth estuary was also of importance, while tidal influences were least important of the main effects studied. Clearly, the precise distribution of the fish species studied in the lower estuary depends on a suite of factors that have complex interactions. The role of food type and availability was not addressed in detail by the current study but is likely to play an important part in determining distributions. The effect on fish of substrate disturbance by dredging

and dumping in the Forth is not well understood but is also likely to have important consequences for fish communities in terms of habitat alteration and food availability. Given the importance of fish to estuarine functioning and the use of fish in measuring estuarine quality (see recent volume by Elliott Hemingway, 2002 for review), both of these areas are worthy of more detailed study. Acknowledgements Thanks to the crew of S/V Forth Ranger. Special thanks to Darren Shaw for provision of a new negative binomial fitting function. References Blaber, S. J. M., & Blaber, T. G. (1980). Factors affecting the distribution of juvenile estuarine and inshore fish. Journal of Fish Biology 17, 143–162. Claridge, P. N., Potter, I. C., & Hardisty, M. W. (1986). Seasonal changes in movements, abundance, size composition and diversity of the fish fauna of the Severn estuary. Journal of the Marine Biological Association of the United Kingdom 66, 229–258. Costa, M. J., & Elliott, M. (1991). Fish usage and feeding in two industrialised estuaries—the Tagus, Portugal, and the Forth, Scotland. In M. Elliott, & J.-P. Ducrotoy (Eds.), Estuaries and coasts: Spatial and temporal intercomparisons (pp. 289–297). Fredensborg: Olsen and Olsen. Crawley, M. J. (1993). GLIM for ecologists (379 pp.). Oxford: Blackwell. Elliott, M., & Dewailly, F. (1995). The structure and components of European fish assemblages. Netherlands Journal of Aquatic Ecology 29, 397–417. Elliott, M., & Hemingway, K. L., (Eds.) (2002). Fishes in estuaries (658 pp.). Oxford. Blackwell. Elliott, M., & Kingston, P. F. (1987). The sublittoral benthic fauna of the estuary and Firth of Forth, Scotland. Proceedings of the Royal Society of Edinburgh 93B, 449–465. Elliott, M., O’Reilly, M. G., & Taylor, C. J. L. (1990). The Forth estuary: a nursery and overwintering area for North Sea fishes. Hydrobiologia 195, 89–103. Elliott, M., & Taylor, C. J. L. (1989). The structure and functioning of an estuarine/marine fish community in the Forth estuary, Scotland. In R. Z. Klekowski, E. Styczynska, & L. Falkowski (Eds.), 21st European marine biology symposium, Gdansk (pp. 227–240). Warsaw: Polish Academy of Sciences. Gibson, R. N., Robb, L., Burrows, M. T., & Ansell, A. D. (1996). Tidal, diel and longer term changes in the distribution of fishes on a Scottish sandy beach. Marine Ecology Progress Series 130, 1–17. Goni, R., Alvarez, F., & Adlerstein, S. (1999). Application of generalized linear modeling to catch rate analysis of Western Mediterranean fisheries: the Castello´n trawl fleet as a case study. Fisheries Research 42, 291–302. Greenwood, M. F. D., Hill, A. S., & McLusky, D. S. (2002). Trends in abundance of benthic and demersal fish populations of the lower Forth estuary, East Scotland, from 1982–2001. Journal of Fish Biology 61(Suppl. A), 90–104. Haedrich, R. L. (1983). Estuarine fishes. In B. Ketchum, & J.-P. Ducrotoy (Eds.), Ecosystems of the world 26 estuaries and enclosed seas (pp. 183–207). Amsterdam: Elsevier. Haedrich, R. L., & Haedrich, S. O. (1974). A seasonal survey of the fishes in the Mystic River, a polluted estuary in downtown Boston, Massachusetts. Estuarine, Coastal and Marine Science 2, 59–73.

M.F.D. Greenwood, A.S. Hill / Estuarine, Coastal and Shelf Science 58 (2003) 211–225 Henderson, P. A. (1989). On the structure of the inshore fish community of England and Wales. Journal of the Marine Biological Association of the United Kingdom 69, 145–163. Henderson, P. A., & Holmes, R. H. A. (1989). Whiting migration in the Bristol Channel: a predator–prey relationship. Journal of Fish Biology 34, 409–416. Henderson, P. A., & Holmes, R. H. A. (1991). On the population dynamics of dab, sole and flounder within Bridgwater Bay in the lower Severn estuary, England. Netherlands Journal of Sea Research 27, 337–344. Henderson, P. A., & Seaby, R. M. H. (1999). Population stability of the sea snail at the southern edge of its range. Journal of Fish Biology 54, 1161–1176. Henderson, P. A., Seaby, R. M. H., & Somes, R. (2002). Fish and crustacean captures at Hinkley Point B Nuclear Power Station: Report for the year April 2001 to March 2002 (21 pp.). Lymington: Pisces Conservation Ltd. URL: http://www.powerstationeffects. co.uk/reports/hinkfish2002.pdf. Insightful Corporation (2001). S-plus 6 for windows. Seattle: Insightful Corporation. Jayamanne, S. C., & McLusky, D. S. (1997). Biology of the shrimps in the Forth estuary. In D. S. McLusky (Ed.), Coastal zone topics: Process, ecology and management. 3. The estuaries of Central Scotland (pp. 147–156). Peterborough: Joint Nature Conservation Committee. Kennish, M. J. (1992). Ecology of estuaries: anthropogenic effects (499 pp.). Boca Raton, FL: CRC Press. Kupschus, S., & Tremain, D. (2001). Associations between fish assemblages and environmental factors in nearshore habitats of a subtropical estuary. Journal of Fish Biology 58, 1383–1403. Large, P. A. (1992). Use of a multiplicative model to estimate relative abundance from commercial CPUE data. ICES Journal of Marine Science 49, 253–261. Loneragan, N. R., Potter, I. C., Lenanton, R. C. J., & Caputi, N. (1986). Spatial and seasonal differences in the fish fauna in the shallows of a large Australian estuary. Marine Biology 92, 575–586. Maes, J., van Damme, P. A., Taillieu, A., & Ollevier, F. (1998). Fish communities along an oxygen-poor salinity gradient (Zeeschelde estuary, Belgium). Journal of Fish Biology 52, 534–546. Maes, J., Taillieu, A., van Damme, P. A., Cottenie, K., & Ollevier, F. (1998). Seasonal patterns in the fish and crustacean community of a turbid temperate estuary (Zeeschelde estuary, Belgium). Estuarine, Coastal and Shelf Science 47, 143–151. Marshall, S., & Elliott, M. (1998). Environmental influences on the fish assemblage of the Humber estuary, U.K. Estuarine, Coastal and Shelf Science 46, 175–184. McLusky, D. S. (1982). The impact of petrochemical effluent on the fauna of an intertidal estuarine mudflat. Estuarine, Coastal and Shelf Science 14, 489–499. Morrison, M. A., Francis, M. P., Hartill, B. W., & Parkinson, D. M. (2002). Diurnal and tidal variation in the abundance of the fish

225

fauna of a temperate tidal mudflat. Estuarine, Coastal and Shelf Science 54, 793–807 (doi:10.1006/ecss.2001.0857). Nelson, G. A., & Leffler, D. (2001). Abundance, spatial distribution, and mortality of young-of-the-year spotted seatrout (Cynoscion nebulosus) along the Gulf Coast of Florida. Gulf of Mexico Science 19, 30–42. Pihl, L., & Wennhage, H. (2002). Structure and diversity of fish assemblages on rocky and soft bottom shores on the Swedish west coast. Journal of Fish Biology 61(Suppl. A), 148–166. Polger, T. T., Summers, J. K., Cummins, R. A., Rose, K. A., & Heimbuch, D. G. (1985). Investigation of relationships among pollutant loadings and fish stock levels in northeastern estuaries. Estuaries 8, 125–135. Pomfret, J. R., Elliott, M., O’Reilly, M. G., & Phillips, S. (1991). Spatial and temporal patterns in the fish communities in two UK North Sea estuaries. In M. Elliott, & J.-P. Ducrotoy (Eds.), Estuaries and coasts: Spatial and temporal intercomparisons (pp. 277–284). Fredensborg: Olsen and Olsen. Pomfret, J. R., Turner, G. S., & Phillips, S. (1988). Beam trawl surveys as a monitoring tool in polluted estuaries in north-east England. Journal of Fish Biology 33(Suppl. A), 71–77. Potter, I. C., Claridge, P. N., Hyndes, G. A., & Clarke, K. R. (1997). Seasonal, annual and regional variations in ichthyofaunal composition in the inner Severn estuary and inner Bristol Channel. Journal of the Marine Biological Association of the United Kingdom 77, 507–525. Power, M., & Attrill, M. J. (2002). Factors affecting long-term trends in the estuarine abundance of Pogge (Agonus cataphractus). Estuarine, Coastal and Shelf Science 54, 941–949 (doi:10.1006/ ecss.2001.0866). Power, M., Attrill, M. J., & Thomas, R. M. (2000). Environmental factors and interactions affecting the temporal abundance of juvenile flatfish in the Thames estuary. Journal of Sea Research 43, 135–149. Poxton, M. G. (1987). Fishery studies in the estuary and Firth of Forth, Scotland. Proceedings of the Royal Society of Edinburgh 93B, 495–508. Robin, J.-P. (1991). Assessment of juvenile flounder catches at the Cordemais Power Station water intake in the Loire estuary, France. Netherlands Journal of Sea Research 27, 317–323. Webb, A. J., & Metcalfe, A. P. (1987). Physical aspects, water movements and modelling studies of the Forth estuary, Scotland. Proceedings of the Royal Society of Edinburgh 93B, 259–272. Welsby, V. G., Dunn, J. R., Chapman, C. J., Sharman, D. P., & Priestley, R. (1964). Further uses of electronically scanned sonar in the investigation of the behaviour of fish. Nature 203, 588–589. Wheeler, A. (1969). The fishes of the British Isles and North-West Europe (618 pp.). London: MacMillan. Wilson, K., Grenfell, B. T., & Shaw, D. J. (1996). Analysis of aggregated parasite distributions: a comparison of methods. Functional Ecology 10, 592–601.