Fisheries Management and Ecology, 2002, 9, 95±103
Diel drift behaviour of ®sh eggs and larvae, in particular barbel, Barbus barbus (L.), in an English chalk stream G. H. COPP & H. FAULKNER Department of Environmental Sciences, University of Hertfordshire, Hat®eld, UK
S. DOHERTY Troon, Ayrshire, UK
M. S. WATKINS Aggregate Industries UK Ltd, Coalville, UK
J. MAJECKI Department of Experimental Zoology and Evolutionary Biology, University of LoÂdz¢, Poland
Abstract To estimate the potential loss of ®sh larvae to downstream stretches, the downstream drift behaviour of ®sh eggs and larvae, in particular barbel, Barbus barbus (L.), was examined in the River Lee, a small, nutrient-rich chalk stream in England, using drift nets over nine consecutive 24-h periods in June 1993 at one location and over ten 24-h periods, once a week for 10 weeks from May to July 1995 at a location slightly more upstream. The density of drifting ®sh larvae was not correlated with river discharge in 1993 or 1995. A clear diel pattern was found in the drift of ®sh eggs and larvae, with barbel being a predominant species. Almost all ®sh larvae drifted at night both in 1993 and 1995, but the drift of eggs in 1995 occurred regardless of luminosity, although most eggs drifted during the day and at dawn. The highest densities of drifting ®sh larvae (in particular barbel) were found in the nets set in the highest water velocities, with the opposite pattern observed for ®sh eggs, suggesting either active response to the water current or shape-related dierences in the drift behaviour of passive particles. K E Y W O R D S : Alburnus, bleak, dispersion, early development, migrations, net clogging, non-salmonids, ontogeny, particle hydrodynamics.
Introduction The drift of young ®sh from spawning grounds to downstream nursery and over-wintering sites is an important behavioural mechanism in the early ontogeny of riverine ®shes, assuring the dispersion of ®sh populations within ¯uvial ecosystems (PenaÂz, Roux, Jurajda & Olivier 1992). The ultimate reasons for migration are probably adaptations to increase survival and growth, as well as to maximize ®tness,
which is increased by moving between dierent habitats that support variable needs during ontogeny (Jonsson 1991). Migration allows the colonization of alternative feeding or nursery areas and the dispersion of species, which ultimately leads to increased survival (Pavlov, Pakhorukov, Kuragina, Nezdoliy, Nekrasova, Brodskiy & Ersler 1978). Drift mechanisms in riverine ®shes are thus linked to growth, survivorship and recruitment success.
Correspondence: Dr Gordon Copp, Station d'Hydrobiologie lacustre-INRA, 75 avenue de Corzent, B.P. 511, F-74203 Thonon-les-Bains CeÂdex, France (e-mail:
[email protected]) Ó 2002 Blackwell Science Ltd
95
96
G. H. COPP ET AL.
The degree of downstream movement in young ®sh, the role of drift in the life cycle, and the mechanisms by which migration takes place vary between species and between water courses for the same species (Pavlov 1994; Jurajda 1998). The barbel, Barbus barbus (L.), is likely to be an important component in drift, it being a lithophilous (Kryzhanovsky 1949) and rheophilous (PenaÂz & Jurajda 1993; Jurajda 1998) river ®sh that exhibits seasonal and diel spawning periodicity (Baras 1995). In England, the distribution of barbel has increased because of introductions, transfers and translocations (Wheeler & Jordan 1990). However, in some areas of its original range such as in the River Lee (Hertfordshire), barbel reproductory success appears to be impeded by other factors, as young-ofyear (0+) and 1+ are infrequently encountered (Copp & Bennetts 1996; Pilcher & Copp 1997; Watkins, Doherty & Copp 1997). This recruitment problem is very likely linked to water quality problems (Tyler & Everett 1993), as the upper Lee is in many places a geomorphologically natural river, but up to 80% of its discharge during low ¯ows is treated euent from a sewage treatment works (Copp & Bennetts 1996). Additionally, ®sheries management practices by angling clubs may contribute to these circumstances through stocking practices that favour the chub Leuciscus cephalus (L.), an opportunistic omnivore known to prey on barbel eggs (M. Carter & M. Pilcher, Environment Agency, personal communications) and young-of-the-year (Mann 1976). The aim of the present study was to determine the intensities and patterns of drift by ®sh eggs and larvae in a small chalk stream, in particular those of young barbel, to estimate the loss of ®shes to downstream stretches, which is a particular concern to ®sheries managers. The drift of young ®sh in smaller, non-salmonid water courses has received little study in Europe (e.g. Jurajda 1998; Carter & Reader 2000), so an underlying objective was to consider the patterns and densities of ®sh drift in the River Lee with respect to published data for larger European rivers. Materials and methods The River Lee has a chalk spring source, but its discharge is in¯uenced by patterns of euent discharge from a treatment plant near its source (Faulkner & Copp 2001). The study sites were situated within two contiguous private estates, Woolmer's Park and Water Hall Farm (National Grid Reference: TL 288 100). Management of the river channel at Woolmer's Park has been limited to the removal of overhanging or fallen trees and bushes (see Fig. 1 of Copp & Bennetts
Figure 1. Water temperature and discharge (a), and number of ®sh larvae per 1000 m3 (b) captured in two drift nets during every 12-h over 10 days in June 1993 from the River Lee at Water Hall Farm, Hertfordshire (England). Day 7:00±19:00 hours (D); night 19:00±07:00 hours (N).
1996). The channel in the stretch that encompasses the two study sites varies between 3 and 12 m wide, in places is over 2 m deep, and has a riparian border width that varies from 2 to > 40 m, with the widest riparian strip occurring along the downstream part of Woolmer's Park. The river contains rie, pool and run sequences, and as it leaves Woolmer's Park estate to Water Hall Farm, the river enters pasture land with little riparian cover. At Water Hall Farm, the river gradually begins to deepen as it approaches a weir, which lies at the bottom end of the estate. The substratum along the entire stretch encompassing both sites was composed primarily of gravel, pebbles and sand, with the extent of silt and mud accumulation increasing in proportion as the river approaches the weir at the downstream end of Water Hall Farm. In 1993, a preliminary study of downstream drift behaviour of ®sh in the River Lee was carried out using two drift nets placed in the river over nine consecutive days. The nets had an opening of 0.26 m ´ 0.52 m, with a 1 m long net (1 mm mesh) of square-to-conical
Ó 2002 Blackwell Science Ltd, Fisheries Management and Ecology, 2002, 9, 95±103
DRIFT BEHAVIOUR OF RIVER FISH EGGS AND LARVAE
shape leading to a 5 cm diameter opening to which a 1-L neoprene bottle was attached using a strong elastic band (PenaÂz et al. 1992). The nets were ®xed to the bottom of the river about 5 m upstream of a gravel ford at Water Hall, approximately 1 km downstream of the study stretch later used in 1995. One net was placed in mid channel (mean water depth 74.5 cm) and one at 1 m from the right bank (mean water depth 54.9 cm), oriented facing upstream. The nets were set at 19:00 hours on 19 June 1993 and their contents were emptied once every 12 h (07:00 hours, 19:00 hours), beginning at 07:00 hours on 20 June and ending at 19:00 hours on 29 June 1993. The bottle created a zero-velocity zone at the end of the net, reducing the amount of damage to the captured ®sh larvae, which may be crushed by the force of the water passing through the nets (Copp & Cellot 1988; PenaÂz et al. 1992). This approach also reduced the amount of time needed to collect the sample from the net, as the used sample bottle could be quickly replaced by an empty one. In 1995, drift samples were collected over a 24-h period each week from 10 May to 13 July, with drift nets set in the River Lee at Woolmer's Park (about 1 km upstream of the 1993 site) in order to be just downstream of suspected ®sh spawning areas (G.H. Copp, unpublished data). Each week, three drift nets (GB Nets, Todmorden, Lancashire, UK) were placed in the stream at 09:00 hours, two nets about 1 m from each bank (mean depth 0.79 m) and one net in the middle of the channel (mean depth 0.85 m), which was approximately 6 m wide and was bordered on the left bank by a thick reed bed, Phragmites sp., and by dense brambles, Rubus fruticosus, and nettles, Urtica sp., on the right bank; thus, nets placed at this point in the river stretched across approximately 12% of the river transect. The nets had a mouth opening of 0.24 m ´ 0.40 m, with a 1-mm mesh net of squareto-conical shape (0.5 m total length) leading to a 3-cm diameter ®tting at the end to which a 1-L neoprene bottle was attached. The standard screw-type attachments supplied with the drift nets were ®tted with neoprene screws caps to permit direct attachment of the 1-L bottle to the net (Gale 1975). The nets were emptied once every 3 h, starting at 12:00 hours and ending at 09:00 hours. At each sampling time in both 1993 and 1995, water temperature, light intensity (lux) and water velocity (centre of net opening) were recorded. Fish and ®sh eggs, as well as invertebrates (V. Edmonds-Brown, G. Levebre, P. Mortlock & G.H. Copp, unpublished data), were sorted from the large amounts of drifting debris (leaves, ®lamentous algae, grass clippings, twigs,
etc.) collected in each sample period. Sorting of samples took place within 1±6 h after sample collection, and sorting was restricted to 30 min per sample so as to provide a common unit of inspection per sample; this was deemed necessary in view of the relatively large number of samples collected, the need to sort them whilst still fresh (which requires less than 20% of the time needed for preserved samples because living, moving and naturally pigmented animals are easier to spot amongst the debris than dead animals). The ®sh specimens extracted from the debris were preserved in 4% formaldehyde. In the laboratory, the diameter of ®sh eggs was measured to the nearest 0.01 mm and tentatively identi®ed from their diameters using bibliographic sources (Maitland 1972; PenaÂz & Prokes 1978). Fish larvae were identi®ed using Koblickaja (1981), with attention to local variation in the number of myomeres and pigmentation. The standard length (SL) in mm and the weight in g was measured for each specimen, which was categorized according to their ontogenetic step using Krupka (1988) for barbel, Prokes & PenaÂz (1979) for gudgeon, Gobio gobio (L.), Prokes & PenaÂz (1980) for chub, Copp (1990) for roach, Rutilus rutilus (L.), and Koblickaja (1981) for all other species. Drift densities in each case were calculated per 1000 m3 of ¯owing water, using estimated volumes of water ®ltered for each sample interval, because the absolute numbers of ®sh eggs and larvae in the drift are spuriously related to water velocity. The volume of water ®ltered could not be derived directly from the area of the net opening and water velocity (at the start and end of sampling), as has been undertaken in many previous studies, because of excessive amounts of drifting organic and mineral matter in the river. Also, resources to obtain `in-mouth' ¯owmeters to measure directly the velocity, and hence volume, of the water ®ltered by the net (e.g. Gadomski & Barfoot 1998) were not available. Thus, estimates of volume ®ltered were generated from the starting water velocity in front of the net using a pre-calibrated, non-linear model of ¯ow retardance, optimized using ®eld velocity readings and drift data in a separate experiment (Faulkner & Copp 2001). Because of the large number of zeros in the ®sh data set, dierences in drift densities (i.e. numbers of ®sh larvae and eggs per volume) between sampling nets were tested using Wilcoxon's (1993) data or Friedman's (1995) data signed rank tests, between periods of day using the Mann±Whitney (day, night in 1993) or Kruskal±Wallis (day, dusk, night, dawn in 1995) tests, and between sample dates using the Kruskal±Wallis test. Comparisons for 1995 were
Ó 2002 Blackwell Science Ltd, Fisheries Management and Ecology, 2002, 9, 95±103
97
98
G. H. COPP ET AL.
made with data from all ten 24-h drift densities combined according to daytime (09:00±18:00 hours), dusk (21:00 hours), night (24:00±03:00 hours) and dawn (06:00 hours). To permit comparisons between 1993 and 1995, drift densities for the latter were combined as day (09:00±18:00 hours) and night (21:00±06:00 hours), respectively. Drift densities and net entrance water velocities over the entire 10-week study were examined according to net position to determine if ®sh larvae attempt to avoid drifting. Correlation analysis was used to examine relationships between environmental variables and the drift densities of ®sh larvae and eggs and analysis of variance was used to compare water velocity values. Results A total of 31 ®sh larvae were captured in the two drift nets during the 1993 part of the study (Fig. 1, Table 1), mostly bleak, Alburnus alburnus (L.), roach and barbel. The net closest to the right bank, where water velocities were highest, captured a higher density of ®sh larvae (bank mean 2.12 ®sh ´ 1000 m±3, middle mean 0.76 ®sh ´ 1000 m±3), which was almost signi®cant (Wilcoxon's, P < 0.059). Similar to other studies of ®sh drift (e.g. PenaÂz et al. 1992), over-night samples, i.e. those collected between 19:00 and 07:00 hours, had signi®cantly higher densities of ®sh (all species, Mann±Whitney, P < 0.002) than day samples (collected between 07:00 and 19:00 hours); a similar pattern was observed with respect to barbel
only (Mann±Whitney, P < 0.02). The drift densities (numbers per volume) of all species combined were not correlated with water temperature, river discharge (Fig. 1) or illumination, but barbel numbers per volume were signi®cantly correlated with light levels (r 0.536, P 0.05) but not with water temperature or river discharge (Fig. 1); note that mean river discharge and temperature were inversely correlated (r ±0.616, P < 0.005). The higher incidence of older specimens in the net in 1993 (Table 1) probably resulted from the longer period (12 h versus 3 h in 1995) that the nets remained in the river, increasing the chance of larger benthic specimens, e.g. gudgeon, stone loach, Barbatula barbatula (L.), wandering inadvertently into the net or using it as cover. Only one specimen of the three-spined stickleback, Gasterosteus aculeatus L., was collected. In 1995, a total of 67 eggs were captured in the drift nets (Table 2). These data suggest that most of the eggs captured were nine-spined stickleback, Pungitius pungitius (L.), stone loach and gudgeon, of which the former did not occur in the drift nets as larvae and the latter two both spawn on sand or on plants above sand bottoms (Balon 1984). The highest mean egg drift was observed on 10, 18 and 25 May, and the number of eggs captured on these dates was signi®cantly greater (Kruskal±Wallis, P 0.0001) than on any other date in 1995 (Fig. 2). Over the 10 sampling dates, the drop in egg numbers correlated with the drop in river discharge (r 0.781, P 0.008, d.f. 8), and were inversely correlated with the rise in water tem-
Table 1. Number of specimens, mean and standard error (SE) of standard length (SL) in mm, and the proportion (in %) of each step [Free-Embryo (FE), Larva 1±Larva 6 (L1±L6), Juvenile (J), see Methods] of ®shes collected in two drift nets from the River Lee (England) in nine consecutive 24-h samples during June 1993, and in three drift nets in weekly 24-h samples during May±July 1995 Species 1993 A. alburnus R. rutilus B. barbus G. gobio P. phoxinus B. barbatula G. aculeatus 1995 B. barbus P. phoxinus G. gobio L. leuciscus B. barbatula L. cephalus C. gobio
n
Mean SL
SE
8 8 6 4 2 2 1
8.69 11.71 11.50 5.75 7.00 22.00 12.50
0.69 1.15 0.22 0.95 0.20 2.00 ±
23 9 6 6 6 5 1
12.16 10.07 12.67 12.75 13.83 10.80 6.00
0.14 0.62 0.40 0.31 1.13 0.88 ±
FE
L1 75 12.5
L2
L3
L4
L5
25
12.5 37.5
12.5
12.5
25
50
12.5 100 25
L6
J
100 50
50 100
22 33
60
18 22 50 33 16 20
33 17 34 50 40
12 17 33 34 40
16
100
Ó 2002 Blackwell Science Ltd, Fisheries Management and Ecology, 2002, 9, 95±103
DRIFT BEHAVIOUR OF RIVER FISH EGGS AND LARVAE
Table 2. Number and proportion (in %) of 67 eggs captured at dierent diameters (in mm)*, and the proportion (in %) collected in three small drift nets (1 mm mesh, 0.5 m long, mouth 0.24 m ´ 0.4 m) from the River Lee (England) in weekly 24-h samples during May±July 1995. The highest percentages are given in bold Nb Diameter n %
0.8 1 1.5
1.0 2 3.0
Gg
Pu 1.1 15 22.4
1.15 2 3.0
1.2 22 32.8
1.25 3 4.5
1.3 12 17.9
Rr 1.35 2 3.0
1.4 3 4.4
Pp, Ll Aa, Ga 1.5 1 1.5
Gg 1.8 2 3.0
Bb, Lc 1.9 1 1.5
2.0 1 1.5
*Maitland (1972) gave 0.9 mm for B. barbatula (Nb), 1.2 mm for P. pungitius (Pu), 1.4 mm for R. rutilus (Rr), 1.5 mm for P. phoxinus (Pp), L. leuciscus (Ll), A. alburnus (Aa), and G. aculeatus (Ga), 1.8 for G. gobio (Gg), and 2.0 mm for B. barbus (Bb) and L. cephalus (Lc). PenaÂz & Prokes (1978) gave 1.29 mm for G. gobio (Gg).
perature (r ±0.632, P 0.05). These correlations were worse but more signi®cant (P 0.0001) on a sample by sample basis (all three nets combined): egg density versus river discharge (r 0.501, P 0.0001, d.f. 78) and versus temperature (r 0.327, P 0.003, d.f. 78). When the sampling hours were grouped according to day, dusk, night and dawn, the mean density of drifting eggs was highest in the day and at dawn, but with no signi®cant dierences (Fig. 3). A total of 56 ®sh larvae were captured in three drift nets over the ten 24-h periods in 1995, with barbel being the most abundant species (Table 1) and mostly of larva step L2. Neither bleak nor three-spined stickleback were found in drift samples of 1995. Species present in 1993 but not observed in 1996 were bleak, roach and three-spined stickleback. Species occurring in 1996 but not in 1993 were minnow, Phoxinus phoxinus (L.), dace, Leuciscus leuciscus (L.), bullhead, Cottus gobio L., and chub. Minnow were captured at various stages of development. Most gudgeon were of step L3. Dace and chub ranged in development from steps L3 to L5; the majority of stone loach were step L4, and the lone bullhead was step L6. When the data were examined on a weekly basis (Fig. 2), ®sh larvae drift densities were not signi®cantly correlated with mean river discharge (r ±0.425, d.f. 8) or water temperature (r 0.425). Note that river discharge and water temperature were again inversely correlated (r ±0.88, P 0.0008, d.f. 8). The highest mean densities of drifting ®sh larvae occurred on 22 June and 5 July, respectively (Fig. 2), these being signi®cantly higher than on any other sampling date (Kruskal±Wallis, P 0.0001). With respect to diel drift patterns, there was a signi®cant dierence in the numbers of ®sh caught at the dierent hourly time intervals (Kruskal±Wallis, P 0.0001), with the highest mean density corresponding to 03:00 hours. As in 1993,
signi®cantly higher ®sh densities occurred at night in 1995 (Mann±Whitney, P 0.0001), with a small number of all species migrating during the day and at dawn, and none at dusk (Fig. 3, Kruskal±Wallis, P 0.0001). As in 1993, ®sh drift densities were not correlated with water velocity, nor were the densities of barbel larvae. The pattern of drift by 0+ barbel larvae was similar in both years, with signi®cantly higher densities at night and dawn than during the day and at dusk (Figs 1 & 3). Drift densities of ®sh eggs and larvae in 1995 diered according to net position (Fig. 4), with signi®cantly higher densities of eggs occurring in the left bank net (Friedman's test, P 0.02), where the water velocity was signi®cantly lower (ANOVA, F 3.00, d.f. 239, P 0.0528). By contrast, the densities of drifting ®sh larvae were in general higher (but not signi®cantly so) in the centre and right bank nets; similarly, the densities of barbel were signi®cantly higher in the middle net (Friedman's, P 0.04), where water velocity was highest (Fig. 4). Discussion Unlike in many larger rivers (Gale & Mohr 1978; Pavlov et al. 1978; Clark & Pearson 1980; Brown & Armstrong 1985; PenaÂz et al. 1992; Schmutz, Matheisz, Pohn, Rathgeb & Unfer 1994), the numbers of drifting ®sh larvae in the River Lee was comparatively low in both years of study. In terms of drift density (numbers per volume of water ®ltered), ®sh drift in the River Lee under low ¯ow conditions was about a tenth of that observed under similar conditions in larger water courses such as the River RhoÃne, France, at Jons (Copp & Cellot 1988), and the River Morava, Czech Republic (Jurajda 1998). The present River Lee drift densities were, however, similar to those in the RhoÃne at BreÂgnier-Cordon (PenaÂz et al. 1992) and those in the River Deschutes, USA (except
Ó 2002 Blackwell Science Ltd, Fisheries Management and Ecology, 2002, 9, 95±103
99
100
G. H. COPP ET AL.
Figure 3. Cumulative mean and SE bars for the number of ®sh eggs and larvae captured per 1000 m3 in weekly drift samples in the River Lee between 10 May and 13 July 1995 at Woolmer's Park, with samples combined as daytime (09:00, 12:00, 15:00, 18:00 hours), dusk (21:00 hours), night (24:00, 03:00 hours) and dawn (06:00 hours).
Figure 2. Mean and standard error (SE) bars for water temperature and discharge (a), number of ®sh larvae per 1000 m3 (b) and number of ®sh eggs (c) and captured in three drift nets (sampling every 3 h from 09:00 to 09:00 hours) in the River Lee, Hertfordshire (England) on each sampling date during May±July 1995 at Woolmer's Park, Hertfordshire (England).
for Catostomidae, which were about twice as abundant in July) near the Dalles Reservoir in Washington State (Gadomski & Barfoot 1998). Note, however, that more detailed knowledge of the particular hydrological regimes of these rivers is required to make sensible comparisons between rivers. The high quantities of drifting debris in the Lee decreased the ®ltering eciency of the drift nets, a
problem not accounted for in other studies of ®sh and invertebrate drift except those using in-mouth ¯owmeters with the drift nets (e.g. Gadomski & Barfoot 1998). Despite the importance of using drift densities (number per volume) to avoid spurious relations between numbers of drifting organisms and ¯ow variables, hydrological model-based calibration of drift nets, such as used here to correct for ®ltration eciency errors caused by debris, could not be found in any of the available literature on ®sh and invertebrate drift for which `in-mouth' ¯owmeters were not used; this sampling bias must either be negligible elsewhere (i.e. little or no non-animal drifting material) or it is being ignored so as to simplify matters. The good relationship between the passive eggs and ¯ow-related variables (discharge and temperature) suggests that the role of hydrology in theoretical drift research is being overlooked (Faulkner & Copp 2001). Densities of drifting ®sh eggs, which are by nature entirely passive, were found to be signi®cantly higher in the net with the lowest water velocity (Fig. 4), thus egg entrainment may be related to turbulence rather than velocity, given that suspended sediment
Ó 2002 Blackwell Science Ltd, Fisheries Management and Ecology, 2002, 9, 95±103
DRIFT BEHAVIOUR OF RIVER FISH EGGS AND LARVAE
Figure 4. Cumulative mean and SE bars for the number of ®sh eggs and larvae captured per 1000 m3 in weekly drift samples in the River Lee between 10 May and 13 July 1995 at Woolmer's Park, with samples combined according to the net in which they were collected. The mean water velocity (with SE bars) is also given as open dots.
con- centration in the river cross-section is maximal at points of maximum turbulence rather than those of maximum velocity. The spherical shape of the eggs may predispose them to being pushed towards the low ¯ow areas. Hanson & Jonsson (1985) carried out release experiments with oranges in the River Isma to establish the rate of passive transport, and more than 80% of the oranges never reached the river mouth, having been stopped by obstacles or settled out in low ¯ow areas. Of the three recognized forms of downstream migration (passive, active-passive, active), the drift of ®sh free-embryos and larvae is normally attributed either to passive (no orientation to current) or activepassive (orientation to current). Disagreement exists, however, over which of these forms is predominant under dierent environmental circumstances (Fortier & Leggett 1982; Pavlov, Barus, Nezdoliy & Gajdusek 1987). Passive drift, the most common form, is connected either with the physical inability of young ®sh (free-embryos, young larvae) to resist a current, or loss or orientation because of low visibility (older larvae, 0+ juveniles). The distribution and timing of ®sh larvae, once drifting, are in¯uenced by secondary ¯ow patterns, turbulence, rheogradient, hydrostatic reaction and buoyancy (Pavlov 1994), as well as factors that aect water temperature, velocity and to a certain extent, daytime light intensity (Jonsson 1991; Fuiman & Ottey 1992; Pavlov, Sbikin & Mochek 1968; Gadomski & Barfoot 1998), all of which can in¯uence rheo-reaction. Fish larvae drifting in the
River Lee occurred almost entirely in the nets with the fastest ¯ows (Fig. 3); whereas in the upper River Volga, young roach and perch Perca ¯uviatilis L. drifted along the shorelines, whilst pikeperch Stizostedion lucioperca (L.) and smelt Osmerus eperlanus (L.) drifted within the stream ¯ow (Pavlov 1994). These dierences could not be explained from a hydraulics point of view, but dierences in particle shape may be at least partially responsible. Increases in discharge will cause a more rapid downstream passive migration of young ®shes (Pavlov 1994), but there is evidence that even the larvae of some rheophilous cyprinids are able to respond actively to resist local ¯ow hydraulics (Gozlan 1998), with dispersion in young juveniles linked to the interacting factors of discharge and light (Vilizzi & Copp 2001). For instance, the densities of drifting ®sh larvae in the River Lee were correlated with water velocity in 1993, and discharges were also found to be higher at night (Faulkner & Copp 2001) (suggesting an external velocity control), which could be related to the propensity of ®sh larvae to drift at night (Fig. 1). This pattern was not seen in 1995, when the day/night water velocity and drift of larvae were inversely correlated, suggesting the eect of some other control such as predator avoidance. In view of the importance that illumination has on the drift of ®shes in general, and barbel in particular (Fig. 1; also Pavlov 1994), any diel pattern in the drift of debris (and hence water clarity) could be expected to in¯uence the active drift behaviour of ®sh larvae and perhaps that of invertebrates (Edmonds-Brown, Levebre, Mortlock & Copp, unpublished data). Although less territorial than salmonids, some European cyprinids are facultative piscivores, such as the chub (Mann 1976). In the River Lee, the density of large chub is high (Copp & Bennetts 1996; Pilcher & Copp 1997; Watkins et al. 1997), and thus predation and/or cannibalism risk could be a factor leading to the movement of ®sh larvae away from lentic areas into the ¯ow at night and dawn (Fig. 3). Results from a study of diel foraging and microhabitat use by ®shes in the same site at Woolmer's Park (Copp, Spathari, Turmel & Enot, unpublished data) support this assumption, indicating that bankside shelter is used by ®sh larvae and juveniles during the day, when avian and ®sh predation risk would be greatest, and dispersion into the channel at night. The propensity to drift diers between species and may vary between years, being density dependent (Economou 1991), such as recently observed in emerging barbel larvae under experimental conditions (Vilizzi & Copp, unpublished data).
Ó 2002 Blackwell Science Ltd, Fisheries Management and Ecology, 2002, 9, 95±103
101
102
G. H. COPP ET AL.
Barbel was the most frequently captured species in drift samples from the River Lee in 1995, and amongst the most numerous in 1993, with 100 and 60% being at the second step (L2) of larva development (Krupka 1988) in 1993 and 1995, respectively. A small percentage of barbel in the present study were found to be L1 and L3, but this could be explained by natural variations on the individual ontogeny the ®sh (Balon 1990). At L2, barbel became positively photoreactive, this coinciding with the development of the anterior gas chamber, which is known to induce vertical `jumps' at dawn and at dusk (PenaÂz 1971; Krupka 1988) that result in them leaving the gravel beds and entering the ¯ow. This diel behaviour was also evident in all of the other species of ®sh larvae (not illustrated), with almost all ®sh larvae drifting at night (and to a lesser extent at dawn) and only a few specimens of barbel and minnow occurring in day samples in both 1993 and 1995 (Fig. 3). The presence of bleak larvae in the drift in 1993 but not 1995 may re¯ect a decline in the species, which is known to occur at only one location upstream of Hertford (Lemsford Mill, about 15 km upstream of the study sites; M. Pilcher, Environment Agency, personal communication). This suggests an intermittent but marked dispersion phenomenon, intermittent spawning success, or habitat unsuitability, as bleak have not been observed in the Lee between Lemsford Mill and Hertford (Copp & Bennetts 1996; Watkins et al. 1997; Pilcher & Copp 1997). In conclusion, drift of ®sh larvae in the River Lee in general, and barbel in particular, occurred predominately at night, although there was some variation between species. The densities of drifting ®sh in the River Lee were in some cases much lower than, and in other cases similar to, those observed in larger European rivers under low ¯ow conditions. But then, chalk streams tend to have more uniform hydrological regimes, which may result in a low intensity of drift, but yielding drift patterns more re¯ective of species behaviour than hydrological variations. To aid the management of coarse ®sheries, further studies are needed on emergence and post-emergence displacement behaviour (e.g. Vilizzi & Copp 2001), incubation success with respect to water quality levels, incubation and early life mortality rates as a result of predation, and over-winter survivorship to elucidate these recruitment problems. Acknowledgments We express our gratitude to the family of the late Mrs Lucas for her permission to open and free access to the
River Lee at Woolmer's Park, to T. Bennetts for assistance in the ®eld, to A. Green for help with the modelling, to P. Jurajda for the use of his drift nets in 1993, as well as to M. Carter and M. Pilcher of the Environment Agency for providing unpublished information. The visit to the UK by JM was funded by the European Commission (TEMPUS). References Balon E.K. (1984) Additions and amendments to the classi®cation of reproductive styles in ®shes. In: E.K. Balon (ed.) Early Life Histories of Fishes: New Developmental, Ecological and Evolutionary Perspectives. Dordrecht: Dr W. Junk Publishers, pp. 59±72. Balon E.K. (1990) Epigenesis of an epigeneticist: the development of some alternative concepts on the early ontogeny and evolution of ®shes. Guelph Ichthyological Reviews 1, 1±48. Baras E. (1995) Thermal related variations of seasonal and daily spawning periodicity in Barbus barbus. Journal of Fish Biology 46, 915±917. Brown A.V. & Armstrong M.L. (1985) Propensity to drift downstream among various species of ®sh. Journal of Freshwater Ecology 3, 3±17. Carter K.L. & Reader J.P. (2000) Patterns of drift and power station entrainment of 0+ ®sh in the River Trent, England. Fisheries Management and Ecology 7, 447±464. Clark A.L. & Pearson W.D. (1980) Diurnal variations in ichthyoplankton densities at Ohio River Mile 571. Transactions of the Kentucky Academy of Sciences 41, 116±121. Copp G.H. (1990) Shifts in the microhabitat of larval and juvenile roach Rutilus rutilus (L.) in a ¯oodplain channel. Journal of Fish Biology 36, 683±692. Copp G.H. & Bennetts T.A. (1996) Short-term eects of removing riparian and instream cover on barbel (Barbus barbus) and other ®sh populations in a stretch of English chalk stream. Folia Zoologica 45, 283±288. Copp G.H. & Cellot B. (1988) Drift of embryonic and larval ®shes, especially Lepomis gibbosus (L.) in the Upper Rhone River. Journal of Freshwater Ecology 4, 419±424. Economou A.N. (1991) Is dispersal of ®sh eggs, embryos and larvae an insurance against density dependence? Environmental Biology of Fishes 31, 313±321. Faulkner H. & Copp G.H. (2001) A model for accurate drift estimation in streams. Freshwater Biology 46, 723±733. Fortier L. & Leggett W.C. (1982) Fickian transport and the dispersal of ®sh larvae in estuaries. Canadian Journal of Fisheries and Aquatic Sciences 39, 1150±1163. Fuiman L.A. & Ottey D.R. (1992) Temperature eects on spontaneous behaviour of larval and juvenile red drum,
Ó 2002 Blackwell Science Ltd, Fisheries Management and Ecology, 2002, 9, 95±103
DRIFT BEHAVIOUR OF RIVER FISH EGGS AND LARVAE
Sciaenops ocellatus, and implications for foraging. Fisheries Bulletin 91, 23±35. Gadomski D.M. & Barfoot C.A. (1998) Diel and distributional abundance patterns of ®sh embryos and larvae in the lower Columbia and Deschutes rivers. Environmental Biology of Fishes 51, 353±368. Gale W.F. (1975) A quick-opening bucket for plankton and larval ®sh nets. Progressive Fish Culturist 37, 164. Gale W.F. & Mohr H. W. Jr (1978) Larval ®sh drift in a large river with a comparison of sampling methods. Transactions of the American Fisheries Society 107, 46±55. Gozlan R.E. (1998) Environmental biology of the so®e Chondrostoma toxostoma (Cyprinidae), with emphasis on early development. PhD Thesis, University of Hertfordshire, Hat®eld, 210 pp. Hanson L.P. & Jonsson B. (1985) Downstream migration of hatchery-reared smolts of Atlantic salmon (Salmo salar) in the River Imsa, Norway. Aquaculture 45, 237±248. Jonsson N. (1991) In¯uence of water ¯ow, water temperature and light on ®sh migrations in rivers. Nordic Journal of Freshwater Research 66, 20±35. Jurajda P. (1998) Drift of larval and juvenile ®shes, especially Rhodeus sericeus and Rutilus ruitlus, in the River Morava (Danube basin). Archives fuÈr Hydrobiologie 141, 231±241. Koblickaja A.P. (1981) Key for Identifying Young Freshwater Fishes. Moscow: Light and Food Industrial Publishing House, 208 pp. Krupka I. (1988) Early development of barbel, Barbus barbus (Linneaus 1758). Prace Ustavu RybaÂrstva a HydrobioloÂgie 6, 115±138. Kryzhanovsky S.G. (1949) Eco-morphological principles and patterns of development among minnow, loach and cat®sh. Part II: Ecological groups of ®shes and patterns of their distribution. Fisheries Research Board of Canada Translation Series No. 2945 (1974). 197 pp. Maitland P.S. (1972) A Key to the Freshwater Fishes of British Isles. Freshwater Biological Association. Occasional Publication No. 27, 139 pp. Mann R.H.K. (1976) Observations on the age, growth, reproduction and food of the chub Squalius cephalus (L.) in the River Stour, Dorset. Journal of Fish Biology 8, 265±288. Pavlov D.S. (1994) The downstream migration of young ®shes in rivers: Mechanisms and distribution. Folia Zoologica 43, 193±208. Pavlov D.S., Sbikin Y.N. & Mochek A.D. (1968) The eects of illumination in running water on the speed of ®shes in relation to features of their orientation. Journal of Ichthyology 12, 703±711.
Pavlov D.S., Pakhorukov A.M., Kuragina G.N., Nezdoliy U.K., Nekrasova N.P., Brodskiy D.A. & Ersler A.L. (1978) Some features of the downstream migrations of juvenile ®shes in the Volga and Kuban Rivers. Journal of Ichthyology 19, 363±374. Pavlov D.S., Barus V., Nezdoliy U.K. & Gajdusek J. (1987) Downstream migrations from Mostiste and Vestonice reservoirs. Acta Scientiarum Naturalium Academiae Scientiarum Bohemoslovacae Brno 21, 1±64. PenaÂz M. (1971) Dierences in mortality rate and development in feeding and starving larvae of Chondustroma nasus and B. barbus (Pisces). Zoologica Listy 20, 85±94. PenaÂz M. & Jurajda P. (1993) Fish assemblage of the Morava River: longitudinal zonation and protection. Folia Zoologica 42, 317±329. PenaÂz M. & Prokes M. (1978) Reproductive and early development of the gudgeon, Gobio gobio, I. Spawning and embryonic period. Folia Zoologica 27, 257±267. PenaÂz M., Roux A.-L., Jurajda P. & Olivier J.-M. (1992) Drift of larval and juvenile ®shes in a by-passed ¯oodplain of the Upper River Rhone, France. Folia Zoologica 41, 281±288. Pilcher M. & Copp G.H. (1997) Winter distribution and habitat use of ®sh in a regulated lowland river system of South-East England. Fisheries Management and Ecology 4, 199±215. Prokes M. & PenaÂz M. (1979) Reproductive and early development of the gudgeon, Gobio gobio, I. Larval and juvenile periods. Folia Zoologica 28, 55±64. Prokes M. & PenaÂz M. (1980) Early development of the chub, Leuciscus cephalus. Acta Scientiarum Naturalium Academiae Scientiarum Bohemicae Brno 14, 1±40. Schmutz S., Matheisz S., Pohn A., Rathgeb J. & Unfer G. (1994) Erstbesiedelung des Marchfeldkanals aus ®shoÈkologischer Sicht. OÈsterreichs Fischerei 47, 158±178. (German with English summary). Tyler C.R. & Everett F. (1993) Incidences of gross morphological disorders in barbel (Barbus barbus) in three rivers in southern England. Journal of Fish Biology 43, 739±748. Vilizzi L. & Copp G.H. (2001) Behavioural responses of 0+ barbel in an arti®cial stream: distribution and velocity use. Animal Behaviour 61, 645±654. Watkins M.S., Doherty S. & Copp G.H. (1997) Microhabitat use by 0+ and older ®shes in a small English chalk stream. Journal of Fish Biology 50, 1010±1024. Wheeler A. & Jordan D.R. (1990) The status of the barbel, Barbus barbus (L.) (Teleostei, Cyprinidae), in the United Kingdom. Journal of Fish Biology 37, 393±399.
Ó 2002 Blackwell Science Ltd, Fisheries Management and Ecology, 2002, 9, 95±103
103