Fish population size and movement patterns in a small intermittently open South African estuary more

Estuarine, Coastal and Shelf Science 67 (2006) 10e20 www.elsevier.com/locate/ecss Fish population size and movement patterns in a small intermittently open South African estuary J.R. Lukey a, A.J. Booth b, P.W. Froneman a,* a Coastal Research Group, Department of Zoology and Entomology, Rhodes University, Grahamstown 6140, South Africa b Department of Ichthyology and Fisheries Science, Rhodes University, Grahamstown 6140, South Africa Received 30 March 2005; accepted 10 October 2005 Available online 27 January 2006 Abstract The population size and movement patterns of small fish (>50 mm SL) in a small intermittently open estuary (Grant’s Valley estuary: 33  400 12.100 S, 26  420 12.600 E) situated on the south-east Cape coast of South Africa were examined during the closed phase over the period May and August 2004. The estuary was subdivided into four discrete areas and the fish within each area sampled using a 30 m seine net (15 mm mesh). Fish captured were marked by fin clipping according to the area of capture. Fish population size was estimated by using three methods: the Schnabel estimator, the Hilborn estimator, and a derived estimator. A total of 12 species was captured and marked during the study. The total number of fish in the estuary was estimated at ca. 12 000 individuals (11 219e13 311). Marine-breeding species (Rhabdosargus holubi, Monodactylus falciformis, and two mullet species) numerically dominated the ichthyofauna, possibly as a result of their effective use of overtopping events, when seawater washes over the sandbar, to enter the estuary during the closed mouth phase. The two mullet species, Myxus capensis and Liza richardsonii, and the Cape stumpnose, R. holubi moved extensively throughout the estuary, while the remaining species exhibited restricted movement patterns possibly due to the preference for refuge and foraging areas associated with reed beds. The observed movement patterns of individual fish species appeared to be associated with both foraging behaviour and habitat selection. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: estuarine fish; intermittently open estuary; South Africa; mark-recapture; fish movement modelling; population size estimates; habitat selection 1. Introduction Intermittently open (IO) estuaries account for approximately 70% of all South African estuaries (Whitfield, 1992). These estuaries are characterised by a sandbar across the mouth that forms a temporary barrier between the marine and estuarine environment. Horizontal gradients in temperature and salinity within these systems are generally absent. The absence of these gradients has been linked to the small catchment size that limits freshwater input and coastal winds that increased horizontal and vertical mixing of the water column (Froneman, 2002). Results of several studies indicate that the ichthyofaunal * Corresponding author. E-mail address: w.froneman@ru.ac.za (P.W. Froneman). 0272-7714/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ecss.2005.10.021 component of IO estuaries in southern Africa are dominated, both in number and biomass, by marine-breeding species with an obligate estuarine phase (Griffiths, 2001a; Vorwerk et al., 2001; Vivier and Cyrus, 2002). The recruitment of these species into IO estuaries occurs either during mouth-opening events or during the overtopping of seawater across the sandbar during spring high tides or during severe storms (Neira and Potter, 1992a; Cowley et al., 2001; Kemp and Froneman, 2004). Fish species diversity within IO estuaries in South Africa and Australia has been shown to increase with overtopping and breaching events, and with the size of the estuary (Bennett, 1989; Whitfield et al., 1989; Griffiths and West, 1999). With the exception of the vegetation free, shallow, marine sediment dominated mouth region, fish within IO estuaries appear to be well-mixed with no clear spatial patterns in the distribution of species (Vorwerk et al., 2001, 2003). J.R. Lukey et al. / Estuarine, Coastal and Shelf Science 67 (2006) 10e20 11 Information on the population sizes of organisms within an estuary is crucial for understanding the ecology of the system and relates directly to the management of fish populations (Cowley and Whitfield, 2001). Mark-recapture studies in larger IO estuaries indicate high population size variability. This has been attributed to differences in recruitment opportunities (i.e., length of open phase, number of overtopping events, and habitat availability) (Blaber, 1973; Cowley and Whitfield, 2001). The processes of movement and habitat selection of organisms within heterogeneous landscapes can be studied using ecological models that address population dynamics and spatial distributions (Lima and Zollner, 1996). Local fish movement can be seen as a measure of habitat selection and foraging behaviour in fish species with immigration rates as an indicator of habitat quality (Gilliam and Fraser, 1987; ´ ´ Belanger and Rodrıguez, 2002). At present, very little is known about fish population sizes and movement within southern African intermittently open estuaries. Most studies on ichthyofaunal population sizes in IO southern African estuaries have been restricted to larger systems (Blaber, 1973; Bennett et al., 1985; Cowley and Whitfield, 2001; Vorwerk et al., 2001). By contrast, information on fish populations in the much smaller systems, with surface area of less than 5 ha, is absent. Therefore, the aims of this study were to examine fish population size within small IO estuaries and for comparisons to larger IO estuaries in South Africa and IO estuaries in Australia, together with examining habitat selection, foraging and localised fish movement patterns. 2. Materials and methods 2.1. Study site Grant’s Valley estuary (33  400 12.100 S, 26  420 12.600 E) is located approximately 5 km east of Kenton-on-Sea in the Eastern Cape, South Africa. This small estuary (length 900 m; maximum width 14 m) has a catchment area of about ca. 13 km2. The estuary is shallow (<1.5 m) with a surface area of ca. 3 ha. The mouth of the estuary is predominantly closed, opening infrequently during the rainy summer season and remaining open for duration of days to a few weeks. Freshwater flow into the estuary comes from an intermittent stream that is impounded by several small farm dams. The streambed remained dry during the study period. Freshwater input therefore only occurs from direct runoff from the surrounding area and during periods of high rainfall when the dams overflow. The catchment area of the estuary is mainly covered in coastal thicket, although in the upper reaches the estuary coastal grassland predominates. Submerged reeds, mainly Phragmites australis, occur on the east bank in the lower reaches, closest to the mouth and along the west bank in the middle and upper reaches. 2.2. Sampling procedure This study was conducted during the closed phase of the estuary between April 2004 and July 2004. Sampling occurred four times: in April, June and twice in July. Block nets (10e 20 m long  1.5 m, 50 mm mesh) were placed across the estuary at three different areas dividing the estuary into four distinct areas and preventing inter-area movement. Area 1 was the mouth region, closest to the ocean (Fig. 1). It is a shallow (0.5 m maximum depth) region characterised by the virtual absence of aquatic macrophytes. The mean salinity for this area was 16.75 (Æ0.5), and water temperature ranged from 14.0  C in July 2004 to 18.8  C in May 2004. Area 2 is a deeper region of the estuary (1.5 m maximum depth) with a bed of the reed Phragmites australis located along the upper western bank. The mean salinity for this area was 16.5 (Æ0.6), and water temperature ranged from 13.3  C in July 2004 to 18.8  C in May 2004. Area 3 was of medium depth (1 m maximum depth), and was devoid of reed beds, but with 5e10% cover of the submerged macrophyte Potamogeton pectinatus. The mean salinity for this area was 16.75 (Æ0.5), and water temperature ranged from 13.5  C in July 2004 to 18.5  C in May 2004. The mean depth of area 4 was approximately 1.5 m. An extensive reed bed was located on the upper western bank. The mean salinity for this area was 16.75 (Æ0.5), and water temperature ranged from 13.2  C in July 2004 to 19.4  C in May 2004. Fish in the four areas were sampled during daylight hours (between 09:00 and 15:00 h) with a seine net (30 m  2 m with a 15 mm mesh). The net was deployed in a semi-circle and hauled along the area by three to four people, ensuring the footrope was dragged along the bottom to minimize fish escape. Each haul across the complete width of the estuary encompassed ca. 300 m2 in area 1 and ca. 400 m2 for the other areas. Two to three net hauls were conducted within each area for each of the four sampling periods. Fish captured were transferred to 200 L well aerated polycarbonate containers filled with estuarine water from the region sampled. All fish that were caught were identified, placed on a fish board, measured for standard length (SL) to the nearest millimetre, and marked by means of clipping a single fin depending on the region where the fish was originally sampled (left pelvic fin for area 1, right pelvic fin for area 2, left pectoral fin for area 3, and right pectoral fin for area 4). The fin rays were clipped off in one straight cut using stainless steel scissors, leaving the base of the fin for regrowth. For sole, Heteromycteris capensis, there are no pectoral fins to clip; therefore, the fringing dorsal fin was clipped and only the total population size was estimated. If a fish was recaptured in an area different from where it was originally captured, it was returned to the recapture area. To minimize stress all fish captured were returned to the estuary within 60 min of capture. To maximise survival only fish >50 mm SL were fin clipped. This size represents the length of fish that could not escape through the 15 mm mesh size of the seine net. No mortality from clipping and handling was recorded during the four surveys. 2.3. Data analysis Population size estimates were determined using three different mark-recapture estimators. The first two, the Schnabel 12 J.R. Lukey et al. / Estuarine, Coastal and Shelf Science 67 (2006) 10e20 N Coastal thicket Coastal grassland 4 3 2 Sand Bar 1 Indian Ocean 30ºS South Africa 0 150 300 m 20ºE Indian Ocean 30ºE Fig. 1. Position of sampling areas in the intermittently open Grant’s Valley estuary located on the south-east coast of South Africa. (1938) and derived (Cowley and Whitfield, 2001) estimators, did not assume movement within the estuary, while the third, based on a Hilborn (1990) estimator, assumed that fish moved within the four demarcated areas. In all models, the following assumptions were made: (1) there was no recruitment into, or emigration out of the population; (2) marked fish did not lose their marks and were easily recognisable on recapture; (3) marked and unmarked fish suffered the same mortality; (4) marked fish randomly mixed with unmarked fish; and (5) marked fish and unmarked fish are equally vulnerable to sampling. 3. Non-movement models 3.1. Schnabel estimator The maximum likelihood estimate of the total number of ^ fish, N, from the Schnabel (1938) estimator is (Seber, 1965, 1986): PT mM t¼2 ^ ¼ PT t t ; N n t¼2 t Variance estimates were obtained from the methods of Robson and Regier (1964). 3.2. Derived method The derived method was used to obtain a population size estimate for those species where no recaptures or few recaptures were obtained. The percentage catch representation of these species was compared with a ‘control species’ from which calculated population size estimates were obtained (Cowley and Whitfield, 2001). The basic assumptions of the derived method were (1) that all species have an equal probability of capture, and (2) that the species under investigation have the same distribution as the ‘control species’ (Cowley and Whitfield, 2001). For example, if 50 individuals of a non-recaptured species were caught, 1000 individuals of the ‘control species’ (with a calculated population estimate of 2000) were caught during the same mark-recapture period, then the derived population size for the species with no recaptures would be 100 [i.e., 50/1000  2000] (Cowley and Whitfield, 2001). Multiple control species were considered for different species of fish, and the species with the highest recapture rate for that group was used. The use of multiple control species gave estimates closer to the modelled estimates for well-recaptured species. Liza richardsonii was used as the control species for all the mullet species, Glossogobius callidus was used as the control species for the benthic fish (gobies and the sole), and Rhabdosargus holubi was used as a control ð1Þ where nt is the number of fish sampled on the tth occasion, mt is the number of marked fish in the tth sample, and Mt is the cumulative number of marked fish in the population, and T is the total number of time periods in the study. J.R. Lukey et al. / Estuarine, Coastal and Shelf Science 67 (2006) 10e20 13 species for the remaining species (sparids, moonies, and tilapia). 3.3. Movement model A modified Hilborn (1990) estimator was used to estimate both the number of fish in the estuary, and the probability of moving from one area to another. Species with good recapture rates (>10%) and all the species combined were modelled and estimated using this method, while species with lower recapture rates were combined into families when possible for movement analysis. 3.3.1. Predicted number of marked fish The predicted number of marked fish (with area-specific marks) that were marked in area i and moved to area j in ^ the tth sampling occasion, M ij;t , is calculated from the surviving number of marks in that area and the newly marked fish introduced to all areas that move to or stay in area i. It is assumed that fish that were marked in each area move instanta^ neously according to the estimated movement matrix Fij . Therefore, ^ ^ M ij;tþ1 ¼ M ij;t þ A X j¼1 ÿln L ¼ ÿ T A YY t¼1 i¼1 ÿ Á ^ ^ Ri;t ÿ Ri;t ln Ri;t ÿ lnðRi;t !Þ: ð6Þ The movement matrix was simplified by noting that the last column vector is calculated from all the column vectors as: Aÿ1 X i¼1 FA;j ¼ 1 ÿ Fi;j : ð7Þ It is assumed that the row vectors of the movement matrix sum to unity and that all parameters are positive. 3.3.5. Parameter variability Parameter variability was estimated using parametric bootstrapping (Efron, 1979) as it is noted that the observed recaptures are Poisson-distributed. During each bootstrap iteration the observed number of fish marked in each area i that moved to area j were Poisson deviates drawn from the original observed number of recaptures that were marked in area i and moved to area j. The bootstrapping procedure was iterated 500 times, and the 100(1 ÿ a) % confidence intervals calculated using the percentile method (Buckland and Garthwaite, 1991). 4. Results ^ Mj;t Fij : ð2Þ 3.3.2. Predicted recaptures If there is a constant catchability over all areas, then the number of recaptures per area j that were originally marked in area i during the previous sampling occasion is: ^ Rij;t ¼ ^M ji;tÿ1 : p^ ð3Þ 4.1. Population estimates A total of 3498 fish from six families and 12 species were marked during the study. In total, 448 recaptures were made, a recapture rate of 12.8%. While seven species were recaptured, most recaptures were from three species: Rhabdosargus holubi, Monodactylus falciformis, and Glossogobius callidus (Table 1). No recaptures were recorded for Oreochromis mossambicus, Mugil cephalus, Diplodus sargus capensis, Caffrogobius gilchristi, and Psammogobius knysnaensis. For those species where no recaptures were recorded, population size was estimated using the derived method. The estimated total number of fish using the Schnabel method (S) was 12 262 (95% CI ¼ 11 219e12 922; CV ¼ 4.2%) individuals, and 12 258 (95% CI ¼ 11 373e13 311; CV ¼ 4.2%) using the Hilborn method (H) (Table 2). Rhabdosargus holubi (50e180 mm SL; mean ¼ 74.9 mm) was the most abundant species captured in the estuary during the study and accounted for 40.6% of the captures. Population estimates for this species were 3970 (S), and 4162 (H). The Cape moony (Monodactylus falciformis) (50e126 mm SL; mean ¼ 66.1 mm) was the second-most-caught species in the study with 602 individuals, accounting for 17.2% of the fish captures. Population estimates were 1875 (S) and 1617 (H). Freshwater mullet (Myxus capensis) (52e270 mm SL; mean ¼ 107.1 mm) was the most common mullet species in the estuary. Recaptures occurred on all the sampling trips, with 21 recaptures recorded. The population estimates for this species were 1113 (S). 3.3.3. Total number of fish The total number of fish in area j at time t is calculated from the total number of fish examined for marks at time t in area i and the estimate of catchability such that: 1 ^ N j;t ¼ ^ p A X i¼1 nij;t : ð4Þ The average number of fish in the study area for the entire study is:  1 N¼ T A T XX j¼1 t¼2 ^ N j;t ; ð5Þ where T is the total number of time periods in the study. 3.3.4. Parameter estimation The probability of capture, ^, and a movement matrix from p ^ area i to area j, Fij , were estimated by minimising a Poisson likelihood of the form: 14 J.R. Lukey et al. / Estuarine, Coastal and Shelf Science 67 (2006) 10e20 Table 1 Total fish captures in the Grant’s Valley estuary between April and July 2004. Sampling effort (E ) is represented by the number of seine hauls in each area. (EUC: Ib e estuarine species that breed manly in estuaries; IIa e marine species with juveniles dependent on estuaries as nurseries; IIb e marine species with juveniles occurring in estuaries; IIc e marine species with juveniles occasionally occurring estuaries; IV e freshwater species; Vb e facultative catadromous species) Species Number caught and marked 1 (E ¼ 8) Sparidae Rhabdosargus holubi Lithognathus lithognathus Diplodus sargus capensis Monodactylidae Monodactylus faliciformes Mugilidae Myxus capensis Liza richardsonii Mugil cephalus Juvenille Mugilidae Gobiidae Glossogobius callidus Caffrogobius gilchristi Psammogobius knysnaensis Cichlidae Oreochromis mossambicus Soleidae Heteromycteris capensis Total 67 3 0 2 0 0 1 285 2 0 1 19 3 383 2 (E ¼ 10) 545 26 4 165 37 49 3 108 98 1 0 13 4 1053 3 (E ¼ 9) 286 51 0 29 62 50 1 43 118 0 0 20 10 670 4 (E ¼ 8) 524 30 0 406 164 53 3 84 103 1 0 21 2 1391 Total 1422 110 4 602 263 152 8 520 321 2 1 73 20 3498 Percent recaptures (%) Estuarine utilization category (EUC) (Whitfield, 1998) 17.2 13.6 0.0 13.1 8.3 8.5 0.0 1.0 26.5 0.0 0.0 0.0 15.0 12.8 IIa IIa IIc IIa Vb IIc IIa e Ib Ib Ib IV IIb The River goby (Glossogobius callidus) (50e133 mm SL; mean ¼ 81.1 mm) was the third most abundant species of fish captured in the estuary with 321 captures, composing 9.2% of the overall abundance. Glossogobius callidus also had the highest recapture rate (26.5%) with 85 recaptures resulting in population estimates of 660 (S), and 655 (H). Juvenile Mugilidae, those mullet too small and immature to identify accurately, were most likely juvenile Myxus capensis Table 2 Estimates of abundances for fish species in the Grant’s Valley estuary using three mark-recapture techniques, with 95% confidence intervals (CIs) and coefficients of variation (CV) in parentheses Species Sparidae Rhabdosargus holubi Lithognathus lithognathus Diplodus sargus capensis Monodactylidae Monodactylus faliciformes Mugilidae Myxus capensis Liza richardsonii Mugil cephalus Unidentified juvenile mullet All mullet Gobiidae Glossogobius callidus Caffrogobius gilchristi Psammogobius knysnaensis Cichlidae Oreochromis mossambicus Soleidae Heteromycteris capensis All species Schnabel 3970 369 e 1875 1113 546 e 3276 7626 660 e e (CIs; CV) (3602e4203; 4.3%) (250e724; 32.1%) Hilborn 4162 386 e 1617 (CIs; CV) (3781e4659; 5.2%) (234e667; 25.6%) Derived e 307 11 1681 945 e 29 1868 e (1641e2216; 7.9%) (844e1790; 24.8%) (385e985; 50.2%) (2180e6558; 32.3%) (5916e11 496; 17.2%) (591e780; 7.7%) (1355e1974; 10.0%) e e e e 8042 655 e e (6151e11 683; 17.0%) (572e782; 8.2%) e 4 2 207 e 28 12 262 (21e111; 51.7%) (11 219e12 922; 4.2%) e 29 12 258 (15e67; 45.4%) (11 373e13 311; 4.2%) 41 e J.R. Lukey et al. / Estuarine, Coastal and Shelf Science 67 (2006) 10e20 15 Table 3 Estimated movement probabilities for total fish population, using the Hilborn (1990) method, in the Grant’s Valley estuary with the 95% confidence intervals in parentheses. Probability that a captured fish is a recapture ¼ 0.073 (0.066e0.080) Source area Destination area Area 1 Area Area Area Area 1 2 3 4 0.00 0.01 0.00 0.00 (0.00e0.01) (0.01e0.02) (0.00e0.01) (0.00e0.01) Area 2 0.28 0.46 0.20 0.19 (0.15e0.40) (0.43e0.56) (0.13e0.30) (0.12e0.25) Area 3 0.66 0.46 0.39 0.21 (0.52e0.78) (0.36e0.48) (0.27e0.48) (0.20e0.35) Area 4 0.06 0.08 0.40 0.60 (0.02e0.14) (0.04e0.12) (0.29e0.52) (0.45e0.62) and Liza richardsonii with occasional Mugil cephalus. A total of 511 juvenile mullet (50e123 mm SL; mean ¼ 62.9 mm) were caught and 5 recaptures were recorded throughout the study. The resulting estimate of total numbers was 3276 (S). The mullet were combined into the one group for increased accuracy of results. The total mullet population was estimated at 7626 (S), and 8042 (H). The remaining species (Liza richardsonii, Lithognathus lithognathus, Heteromycteris capensis, Oreochromis mossambicus, Mugil cephalus, Diplodus sargus capensis, Caffrogobius gilchristi and Psammogobius knysnaensis) captured accounted for ca. 10% of the total numbers with 371 captures and 29 recaptures. Of these, only L. richardsonii, L. lithognathus and O. mossambicus seem to have any significant contribution to the population. However no conclusions can be made from the limited recaptures. One longfin eel (Anguilla mossambica) was captured during the April sampling period. Occasional schools of estuarine roundherring (Gilchristella aestuaria) and Cape silverside (Atherina breviceps) were caught during this study, but these fish were too small for mark-recapture and were not marked. 4.2. Movement data For the movement model, only two recaptures occurred in area 1 and both were fish marked from area 2. The other three areas all had many recaptures, from all the areas (area 2 ¼ 147; area 3 ¼ 183; area 4 ¼ 116). Distinct patterns in movement were observed for the fish community sampled (Table 3). Of the 383 fish marked in area 1, 47 were recaptured with the model estimating 66% of their movement into area 3, with less movement into area 2 (28%) and minimal movement to area 4 (6%). No fish were predicted to remain in area 1, as no recaptures from this area were found in this area. Fish from area 2 (1053 marked, 212 recaptured) were equally estimated to remain in the area and move to area 3 (both 46%) with minimal movement to area 4 (8%) and area 1 (1%). Fish from area 3 (610 marked, 84 recaptured) remained in the area (39%), moved to area 4 (40%), while some movement to area 2 (19%), and no movement to area 1 was estimated. Fish marked in area 4 (1398 marked, 105 recaptured) tended to remain in the area (60%), while some movement was estimated into areas 2 and 3 (19% and 21%), and no movement was estimated into area 1. The Mugilidae, a combined category of all species in the family, tended to move widely through the estuary (Table 4), but recapture rates were particularly low (1.8%). Mugilidae were caught throughout the estuary with some residency patterns shown for areas 2, 3 and 4. Area 1 had many captures in April (N ¼ 279) and very few captures (N ¼ 4) during the other sample periods. The benthic species, Glossogobius callidus was a particularly resident species (Table 5) with high proportions remaining in areas 3 and 4 (74 and 75%). Those marked in area 2 did show movement to areas 1 and 3 (10 and 52%), but none were found to move to area 4. Monodactylus falciformis also exhibited resident behaviour and was particularly dominant in vegetated areas (Table 6). Vegetated areas 2 (N ¼ 165) and 4 (N ¼ 406) had much higher captures then areas 1 (N ¼ 2) and 3 (N ¼ 29). Movement estimates also show Myxus capensis remained in area 2 (66%) and area 4 (67%). On the other hand those M. capensis marked in area 3 showed complete movement to area 4 (100%). Only one M. capensis marked in area 1 was recaptured, and it was recaptured in area 3. Cape stumpnose, Rhabdosargus holubi, was captured throughout the estuary, and recaptures occurred in all areas, except area 1. Rhabdosargus holubi showed some resident behaviour in area 2 (43%) and area 4 (41%), but movement from area 2 to area 3 (52%), and movement from area 4 to area 3 (36%) and area 2 (23%) was considerable. Those captured in area 1 were found to move to area 2 (33%) and area 3 (59%), while those captured in area 3 had a tendency to move to area 2 (48%), with Table 4 Estimated movement probabilities for the Mugilid population, using the Hilborn (1990) method, in the Grant’s Valley estuary with the 95% confidence intervals in parentheses. Probability that a captured fish is a recapture ¼ 0.018 (0.012e0.025) Source area Destination area Area 1 Area Area Area Area 1 2 3 4 0.00 0.00 0.00 0.00 (0.00e0.50) (0.00e0.00) (0.00e0.00) (0.00e0.00) Area 2 0.33 0.62 0.00 0.17 (0.00e1.00) (0.25e1.00) (0.00e0.00) (0.00e0.42) Area 3 0.33 0.25 0.42 0.25 (0.00e1.00) (0.00e0.60) (0.14e0.75) (0.00e0.52) Area 4 0.33 0.12 0.58 0.58 (0.00e1.00) (0.00e0.40) (0.25e0.86) (0.27e0.86) 16 J.R. Lukey et al. / Estuarine, Coastal and Shelf Science 67 (2006) 10e20 Table 5 Estimated movement probabilities for the Glossogobius callidus population, using the Hilborn (1990) method, in the Grant’s Valley estuary with the 95% confidence intervals in parentheses. Probability that a captured fish is a recapture ¼ 0.156 (0.123e0.189) Source area Destination area Area 1 Area Area Area Area 1 2 3 4 0.00 0.10 0.00 0.00 (0.00e0.00) (0.00e0.24) (0.00e0.00) (0.00e0.00) Area 2 0.67 0.38 0.00 0.13 (0.00e1.00) (0.16e0.62) (0.00e0.00) (0.03e0.25) Area 3 0.33 0.52 0.74 0.11 (0.00e1.00) (0.31e0.77) (0.52e0.91) (0.02e0.21) Area 4 0.00 0.00 0.26 0.76 (0.00e0.50) (0.00e0.00) (0.09e0.48) (0.63e0.89) significant numbers remaining in area 3 (26%) or moving to area 4 (26%) (Table 7). White steenbras, Lithognathus lithognathus, were captured in all areas. The majority of the captures occurred in area 3 (N ¼ 51). All recaptures occurred on the final sampling date. Those fish captured in area 1 and area 2 showed a strong movement pattern to area 3 (45 and 55%), while fish captured in area 3 showed residency (48%) and movement to area 2 (52%). No L. lithognathus originally captured in area 4 were recaptured, so movement from this area cannot be estimated (Table 8). 5. Discussion 5.1. Population estimates Mark-recapture techniques are widely used to estimate animal population sizes. These techniques have been utilised for fish populations since the early 20th century, using various methods and resulting in estimates of survival, movement and population size (Schnabel, 1938; Darroch, 1958, 1961; Seber, 1965; Buckland, 1980, 1982). Closed populations, described as populations that remain unchanged during the period of investigation, tend to have simpler estimation models since the effects of the migration, mortality and recruitment are considered negligible (Seber, 1986). With immigration restricted by the mouth closure, and mortality minimized by the virtual absence of piscivorous fish and birds, recruitment was the only potential source of error for the mark-recapture population estimates. Recruitment of ichthyofauna into the estuary likely occurred during the overtopping events that occurred during the study (unpublished data). This recruitment is, however, unlikely to have introduced a high degree of error as we only considered fish >50 mm SL during the study and overtopping typically moves post-flexion larvae (SL < 10 mm) and juvenile fish (SL 10e40 mm) into the estuaries (Kemp and Froneman, 2004). Tag loss and fish mortality related to handling stress have also been identified as important sources of variability in mark-recapture studies (Hansen, 1988; Moffett et al., 1997). Fin clipping probably reduced these problems as the study was conducted over a relatively short period of time and little or no fin regrowth was observed. In the absence of any recaptures, the derived method used to estimate fish population sizes should be considered with caution (Cowley and Whitfield, 2001). Population estimates for those species (Mugil cephalus, Diplodus sargus capensis, Caffrogobius gilchristi, and Psammogobius knysnaensis) where no recaptures were made are therefore likely to be inaccurate. Many Oreochromis mossambicus were captured, but none were recaptured. Previous research has shown that adult O. mossambicus exhibit mass mortality in low temperature waters (e.g., during winter) (Jubb, 1979). As a consequence, the estimates of the population size of O. mossambicus are likely to be low. It is notable that a similar study conducted in IO estuaries within the same geographic region, O. mossambicus contributed <2% of the total catch (Vorwerk et al., 2001). Species richness during this study was similar to that of Cowley and Whitfield (2001) in the nearby IO East Kleinemonde estuary, where 12 and 10 species were found in two different surveys, using similar gear and sampling strategy, compared to the 15 species found in this study. Larger IO estuaries and estuaries connected to the sea more often, such as the East Kleinemonde, should show a greater species richness than the smaller, isolated Grant’s Valley estuary (Neira and Potter, 1992a; Vivier and Cyrus, 2002). Vorwerk et al. (2001) found 20 species of fish in the East Kleinemonde, including Atherina breviceps, Oreochromis mossambicus, Gilchristella aestuaria, Glossogobius callidus, Psammogobius knysnaensis and Heteromycteris capensis. By contrast, in larger, permanently open systems within the same geographic region, up to 30 species may be recorded (Vorwerk et al., 2001). Although that study employed both seine nets and Table 6 Estimated movement probabilities for the Monodactylus falciformis population, using the Hilborn (1990) method, in the Grant’s Valley estuary with the 95% confidence intervals in parentheses. Probability that a captured fish is a recapture ¼ 0.127 (0.101e0.156) Source area Destination area Area 1 Area Area Area Area 1 2 3 4 0.00 0.00 0.00 0.00 (0.00e0.25) (0.00e0.00) (0.00e0.00) (0.00e0.00) Area 2 0.00 0.66 0.00 0.33 (0.00e0.25) (0.51e0.79) (0.00e0.00) (0.16e0.52) Area 3 1.00 0.17 0.00 0.00 (0.25e1.00) (0.07e0.29) (0.00e0.00) (0.00e0.00) Area 4 0.00 0.17 1.00 0.67 (0.00e0.25) (0.06e0.30) (1.00e1.00) (0.48e0.84) J.R. Lukey et al. / Estuarine, Coastal and Shelf Science 67 (2006) 10e20 17 Table 7 Estimated movement probabilities for the Rhabdosargus holubi population, using the Hilborn (1990) method, in the Grant’s Valley estuary with the 95% confidence intervals in parentheses. Probability that a captured fish is a recapture ¼ 0.095 (0.083e0.108) Source area Destination area Area 1 Area Area Area Area 1 2 3 4 0.00 0.00 0.00 0.00 (0.00e0.00) (0.00e0.00) (0.00e0.00) (0.00e0.00) Area 2 0.33 0.43 0.48 0.23 (0.12e0.52) (0.36e0.51) (0.29e0.65) (0.11e0.37) Area 3 0.59 0.52 0.26 0.36 (0.41e0.77) (0.43e0.59) (0.11e0.43) (0.20e0.51) Area 4 0.07 0.05 0.26 0.41 (0.00e0.17) (0.02e0.09) (0.12e0.42) (0.26e0.57) gill nets, the use of the block nets with the small area of the Grant’s Valley estuary probably increased the probability of this study capturing all species within the system. Australian studies (Griffiths and West, 1999; Griffiths, 2001b) found similar fish diversity (16 species) in the small (2 ha) Australian IO Shellharbour Lagoon with similar families dominating (sparids, mullets and gobies) while the smaller Bellambi Lagoon (1.4 ha) contained only five species. The lower fish species diversity in small IO estuaries can be linked to limited recruitment and lower habitat availability (Bennett, 1989; Whitfield et al., 1989; Griffiths and West, 1999). However, juvenile fish of economic importance are often amongst represented species (Griffiths and West, 1999). Estuarine utilization categories divide fish species into groups based on their level of estuarine dependence (Wallace et al., 1984; Whitfield, 1998). The two most numerically abundant fish species found in the Grant’s Valley estuary (Rhabdosargus holubi and Monodactylus falciformis) are classified by Whitfield (1998) as category IIa fish, marine spawners dependent on estuaries for nursery areas. Of the 12 fish species sampled in this study, eight of the species spawn in the marine environment and have an obligate estuarine phase (type II or type V). The remaining four species, of which only the River Goby, Glossogobius callidus, makes a considerable contribution to overall numbers, are all species that can breed within the estuary (type I or type IV) (Whitfield, 1998). In small IO Australian estuaries, marine-breeding fish do not appear to gain access to estuaries during overtopping events due to the general absence of marine fish larvae within the marine waters adjacent to the systems. As a consequence, estuarine breeding fish numerically dominate the ichthyofauna of those systems (Neira and Potter, 1992b; Potter et al., 1993; Young et al., 1997; Griffiths, 2001c). Marine spawned species have, however, been shown to enter these estuaries during opening events (Griffiths and West, 1999; Griffiths, 2001b,c). In the Grant’s Valley estuary, it is only overtopping that maintains the dominance of marine-breeding species during the extended closed periods. The numerically dominant fish species within the Grant’s Valley estuary, namely Rhabdosargus holubi, Monodactylus falciformis and the mullet species, are all characterised by their ability to withstand a wide range in salinity conditions (Blaber, 1974; Day et al., 1981; Bennett, 1985; Branch et al., 1985), and demonstrate an extended breeding period (Wallace, 1975; Bok, 1979; van der Horst and Erasmus, 1981; Lasiak, 1984). These fish have also been shown to utilise overtopping events to recruit into the estuary during the extended closed phase (Kemp and Froneman, 2004). Another abundant species, Glossogobius callidus, which prefers freshwater but can breed in estuaries, is found over a wide range of salinities, but are almost never found outside the river/estuary environment (Whitfield, 1998). Glossogobius callidus was found to be the dominant goby species, and abundant, in other nearby IO estuaries (Vorwerk et al., 2001). Fish species highly tolerant to environmental changes may be the only species that can thrive in small IO estuaries as these systems have lower buffering capabilities and are more likely to experience rapid changes than larger systems (Griffiths and West, 1999). There have been no similar population studies conducted in small IO estuaries. Studies conducted in the larger, IO East Kleinemonde estuary indicated a population estimate of ichthyofauna equivalent to 18 000 fish (z0.10 fish mÿ2) in a study with a similar recapture rate and 133 000 (z0.76 fish mÿ2) in a second study characterised by an increased frequency of overtopping events, but with a much lower recapture rate (Cowley and Whitfield, 2001). By contrast, during this study the mean fish density was estimated at z0.41 fish mÿ2. The higher densities in the Grant’s Valley estuary, as compared to the first study, can be linked to extended mouth closure and the recruitment of fish into the estuary during overtopping events that leads to a build-up of fish biomass in the system. In contrast in the East Kleinemonde estuary, mouth-breaching Table 8 Estimated movement probabilities for the Lithognathus lithognathus population, using the Hilborn (1990) method, in the Grant’s Valley estuary with the 95% confidence intervals in parentheses. Probability that a captured fish is a recapture ¼ 0.095 (0.083e0.108) Source area Destination area Area 1 Area Area Area Area 1 2 3 4 0.25 0.06 0.00 0.00 (0.00e0.25) (0.00e0.25) (0.00e0.25) (0.00e0.00) Area 2 0.25 0.39 0.52 0.00 (0.00e0.25) (0.00e0.50) (0.27e1.00) (0.00e0.00) Area 3 0.45 0.55 0.48 0.00 (0.25e1.00) (0.31e1.00) (0.00e0.71) (0.00e0.00) Area 4 0.05 0.00 0.00 0.00 (0.00e0.25) (0.00e0.10) (0.00e0.03) (0.00e0.00) 18 J.R. Lukey et al. / Estuarine, Coastal and Shelf Science 67 (2006) 10e20 events occur more frequently than in the Grant’s Valley estuary (Cowley and Whitfield, 2001; Kemp and Froneman, 2004), resulting in emigration of fish from the estuary to the marine environment. During the second study the increased overtopping could account for the larger fish abundance. Cowley and Whitfield’s (2001) mullet numbers are more conservative since they included mullet >100 mm while this study included mullet >50 mm. In an Australian study (Young et al., 1997), densities ranged from 0.53 to 11.87 fish mÿ2 in the much larger IO Moore River estuary. Densities in permanently open systems tend to be higher than IO estuaries as recruitment can constantly occur, but little quantification has been shown (Whitfield and Kok, 1992). Rhabdosargus holubi, Monodactylus falciformis, and Myxus capensis are dominant species found in both the study by Cowley and Whitfield (2001) in the East Kleinemonde and in the current study on the Grant’s Valley estuary. With respect to total estimated abundance, R. holubi were found to be less dominant in the present study, where they accounted for 30% of the total, than in the East Kleinemonde where they accounted for between 70 and 80% of the total ichthyofauna. The absence of Pomadasys commersonnii and the piscivorous Lichia amia in this study emphasized the differences between the two estuaries (Cowley and Whitfield, 2001). The presence of R. holubi, Lithognathus lithognathus, and the mullet show the importance of IO estuaries as significant fish habitats for a number of ecologically important fish (Bok, 1979, 1984; Bennett, 1993). 5.2. Fish movement studies Habitat selection by fish species is an important influence on fish assemblage structure in estuaries (Whitfield, 1999). Channel depth, seagrass beds, rocky outcroppings, light penetration levels, sediment type, and the presence of aquatic macrophytes have all been shown to influence distribution of fish in estuaries worldwide (Connolly, 1994; Whitfield, 1999; Griffiths, 2001a). Immigration rate has been shown to be a good indicator of habitat quality for those fish exhibiting exploratory ´ ´ behaviour (Belanger and Rodrıguez, 2002). Furthermore, estuary size also tends to structure fish assemblages (Anganuzzi et al., 1994) as fish habitat preference and food source vary with size (Whitfield, 1998). Movement of the fish community within the estuary during the present study was seen to be minimal as the majority of fish were found to remain within the area of the estuary where they were captured although a slight upstream movement can been seen. This upstream movement may be a seasonal shift to deeper waters during the colder winter. Notable exceptions to this observation were the mullet, and Cape stumpnose, Rhabdosargus holubi, which appeared to demonstrate a high degree of inter-area movement. For the mullet species, gut content analysis studies indicate that mullet consume mainly microphytobenthic algal and diatoms (Blaber, 1987). The mullet, as detritivores make an imˇ portant link in overall fish production (Ray and Straskraba, 2001). Maximum biomass of microphytobenthic algae is generally recorded in the mouth region of the estuary where optimum conditions for growth prevail (Perissinotto et al., 2003; Nozais et al., 2005). The high degree of movement demonstrated by mullet may therefore be attributed to food availability and foraging behaviour. The most dominant mullet species in the estuary is the catadromous Myxus capensis. Catadromous mullet have been shown to show a general upstream movement pattern in estuaries (Almeida, 1996), which may also play a factor in the movement of mullet even though they cannot reach freshwater due to impoundments and restricted fluvial flow. Rhabdosargus holubi did show some preference for areas with high densities of submerged vegetation, as large juveniles feed on epiphytic diatoms covering aquatic macrophytes (Whitfield, 1998). The absence of macrophytes in shallow mouth region during the study would make that area a less suitable habitat for these fish. The remaining species appeared to be restricted to specific areas, particularly those areas characterised by submerged macrophytes or reed beds. A number of previous studies have demonstrated that fish biomass is greatest in those regions of the estuary where reeds or submerged macrophytes persist (Weis and Weis, 2003; Adams et al., 2004). The increase in biomass is thought to be a result of the reed beds providing refuge against predators coupled with their role as detritus traps providing improved foraging regions (Griffiths, 2001b; Weis and Weis, 2003; Nagelkerken and van der Velde, 2004). Small IO estuaries can sustain a large number of fish, with the dominant species relying on overtopping and bar-breaching events as a major mode of entering the system. Fish species richness is limited due to estuary size and the extended period of mouth closure. Freshwater and estuarine fish species are commonly found, however, estuarine-dependent marine species generally dominate small IO systems in South Africa. With lack of sandbar breaching, the numbers of the marine fish species accumulate by recruitment via overtopping events and are available to populate coastal waters upon opening. Movement of fish within these estuaries appears to be species specific and dependent on habitat selection and foraging behaviour. Acknowledgements We are grateful to Rhodes University and BP South Africa for providing funds and facilities to conduct this research. Our thanks also go out to Paul Vorwerk, Anthony Bernard, Justin Blake, Russell Gerber, Mike Jennings, Peet van Rooyen, and Alistair Green for providing field assistance. References Adams, A.J., Locascio, J.V., Robbins, B.D., 2004. Microhabitat use by postsettlement stage estuarine fish: evidence from relative abundance and predation among habitats. Journal of Experimental Marine Biology and Ecology 299, 17e33. Almeida, P.R., 1996. Estuarine movement patterns of adult thin-lipped grey mullet, Liza ramada (Risso) (Pisces, Mugilidae), observed by ultrasonic tracking. Journal of Experimental Marine Biology and Ecology 202, 137e150. J.R. Lukey et al. / Estuarine, Coastal and Shelf Science 67 (2006) 10e20 Anganuzzi, A., Hilborn, R., Skalski, J.R., 1994. Estimation of size selectivity and movement rates from mark-recovery data. Canadian Journal of Fisheries and Aquatic Science 51, 734e742. ´ ´ Belanger, G., Rodrıguez, M.A., 2002. Local movement as a measure of habitat quality in stream salmonids. Environmental Biology of Fishes 64, 155e164. Bennett, B.A., 1985. A mass mortality of fish associated with low salinity conditions in the Bot River estuary. Transactions of the Royal Society of South Africa 45, 437e447. Bennett, B.A., 1989. A comparison of the fish communities in nearby permanently open, seasonally open and normally closed estuaries in the southwestern cape, South Africa. South African Journal of Marine Science 8, 43e55. Bennett, B.A., 1993. The fishery of white steenbras Lithognathus lithognathus off the Cape coast, South Africa, with some considerations for its management. South African Journal of Marine Science 13, 1e14. Bennett, B.A., Hamman, K.C.D., Branch, G.M., Thorne, S.C., 1985. Changes in the fish fauna of the Bot River estuary in relation to opening and closure of the estuary mouth. Transactions of the Royal Society of South Africa 45, 449e464. Blaber, S.J.M., 1973. Population size and mortality of juveniles of the marine teleost Rhabdosargus holubi (Pisces: Sparidae) in a closed estuary. Marine Biology 21, 219e225. Blaber, S.J.M., 1974. Osmoregulation in juvenile Rhabdosargus holubi (Steindachner) (Teleostei: Sparidae). Journal of Fish Biology 6, 797e800. Blaber, S.J.M., 1987. Factors affecting recruitment and survival of Mugilids in estuaries and coastal waters of southeastern Africa. American Fisheries Society Symposium 1, 507e518. Bok, A.H., 1979. The distribution and ecology of two mullet species in some fresh water rivers in the Eastern Cape, South Africa. Journal of the Limnological Society of South Africa 5, 97e102. Bok, A.H., 1984. Extensive culture of two mullet species in freshwater impoundments in the Eastern Cape. South African Journal of Zoology 19, 31e36. Branch, G.M., Bally, R., Bennett, B.A., De Decker, H.P., Fromme, G.A.W., Heyl, C.W., Willis, J.P., 1985. Synopsis of the impact of artificially open¨ ing the mouth of the Bot River estuary: implications for management. Transactions of the Royal Society of South Africa 45, 465e483. Buckland, S.T., 1980. A modified analysis of the Jolly-Seber captureerecapture model. Biometrics 36, 419e435. Buckland, S.T., 1982. A mark-recapture survival analysis. The Journal of Animal Ecology 51, 833e847. Buckland, S.T., Garthwaite, P.H., 1991. Quantifying precision of mark-recapture estimates using the bootstrap and related methods. Biometrics 47, 255e268. Connolly, R.M., 1994. Removal of seagrass canopy: effects on small fish and their prey. Journal of Experimental Marine Biology and Ecology 184, 99e110. Cowley, P.D., Whitfield, A.K., 2001. Fish population size estimates from a small intermittently open estuary in South Africa, based on mark-recapture techniques. Marine and Freshwater Research 52, 283e290. Cowley, P.D., Whitfield, A.K., Bell, K.N.I., 2001. The surf zone ichthyoplankton adjacent to an intermittently open estuary, with evidence of recruitment during marine overwash events. Estuarine, Coastal and Shelf Science 52, 339e348. Darroch, J.N., 1958. The multiple-recapture census: I. Estimation of a closed population. Biometrika 45, 343e359. Darroch, J.N., 1961. The two-sample captureerecapture census when tagging and sampling are stratified. Biometrika 48, 241e260. Day, J.H., Blaber, S.J.M., Wallace, J.H., 1981. Estuarine fishes. In: Day, J.H. (Ed.), Estuarine Ecology with Particular Reference to Southern Africa. Balkema, Cape Town, pp. 197e221. Efron, B., 1979. Bootstrap methods: another look at the Jackknife. The Annals of Statistics 7, 1e26. Froneman, P.W., 2002. Response of the plankton to three different hydrological phases of the temporarily open/closed Kasouga Estuary, South Africa. Estuarine, Coastal and Shelf Science 55, 535e546. Gilliam, J.F., Fraser, D.F., 1987. Habitat selection under predation hazard: test of a model with foraging minnows. Ecology 68, 1856e1862. 19 Griffiths, S.P., 2001a. Factors influencing fish composition in an Australian intermittently open estuary. Is stability salinity-dependent? Estuarine, Coastal and Shelf Science 52, 739e751. Griffiths, S.P., 2001b. Diel variation in the seagrass ichthyofaunas of three intermittently open estuaries in south-eastern Australia: implications for improving fish diversity assessments. Fisheries Management and Ecology 8, 123e140. Griffiths, S.P., 2001c. Recruitment and growth of juvenille yellowfin bream, Acanthopagrus australis Gunther (Sparidae), in an Australian intermittently ¨ open estuary. Journal of Applied Ichthyology 17, 240e243. Griffiths, S.P., West, R.J., 1999. Preliminary assessment of shallow water fish in three small intermittently open estuaries in south-eastern Australia. Fisheries Management and Ecology 6, 311e321. Hansen, L.P., 1988. Effects of Carlin tagging and fin clipping on survival of Atlantic salmon (Salmo salar L.) released as smolts. Aquaculture 70, 391e394. Hilborn, R., 1990. Determination of fish movement patterns from tag recoveries using maximum likelihood estimators. Canadian Journal of Fisheries and Aquatic Science 47, 635e643. Jubb, R.A., 1979. Cichlid fish mortality: Kasouga River estuary. Lammergeyer 27, 51. Kemp, J.O.G., Froneman, P.W., 2004. Recruitment of ichthyoplankton and macrozooplankton during overtopping events into a temporarily open/ closed southern African estuary. Estuarine, Coastal and Shelf Science 61, 529e537. Lasiak, T.A., 1984. The reproductive biology of the moony, Monodactylus falciformes, in Algoa Bay. South African Journal of Zoology 19, 250e 252. Lima, S.L., Zollner, P.A., 1996. Towards a behavioral ecology of ecological landscapes. Trends in Ecology and Evolution 11, 131e135. Moffett, I.J.J., Crozier, W.W., Kennedy, G.J.A., 1997. A comparison of five external marks for Atlantic salmon, Salmo salar L. Fisheries Management and Ecology 4, 49e53. Nagelkerken, I., van der Velde, G., 2004. A comparison of fish communities of subtidal seagrass beds and sandy seabeds in 13 marine embayments of a Caribbean island, based on species, families, size distribution and functional groups. Journal of Sea Research 52, 127e147. Neira, F.J., Potter, I.C., 1992a. The ichthyoplankton of a seasonally closed estuary in temperate Australia. Does an extended period of opening influence species composition? Journal of Fish Biology 41, 935e953. Neira, F.J., Potter, I.C., 1992b. Movement of larval fishes through the entrance channel of a seasonally open estuary in western Australia. Estuarine, Coastal and Shelf Science 35, 213e224. Nozais, C., Perissinotto, R., Tita, G., 2005. Seasonal dynamics of meiofauna in a South African temporarily open/closed estuary (Mdloti Estuary, Indian Ocean). Estuarine, Coastal and Shelf Science 62, 325e338. Perissinotto, R., Nozais, C., Kibirige, I., Anandraj, A., 2003. Planktonic food webs and benthic-pelagic coupling in three South African temporarilyopen estuaries. Acta Oecologica 24, S307eS316. Potter, I.C., Hyndes, G.A., Baronie, F.M., 1993. The fish fauna of a seasonally closed Australian estuary. Is the prevalence of estuarine-spawning species high? Marine Biology 116, 19e30. ˇ Ray, S., Straskraba, M., 2001. The impact of detritivorous fishes on a mangrove estuarine system. Ecological Modelling 140, 207e218. Robson, D.S., Regier, H.A., 1964. Sample size in Peterson mark-recapture experiments. Transactions of the American Fisheries Society 93, 215e226. Schnabel, Z.E., 1938. The estimation of total fish population of a lake. The American Mathematical Monthly 45, 348e352. Seber, G.A.F., 1965. A note on the multiple-recapture census. Biometrika 52, 249e259. Seber, G.A.F., 1986. A review of estimating animal abundance. Biometrics 42, 267e292. van der Horst, G., Erasmus, T., 1981. Spawning time and spawning grounds of mullet with special reference to Liza dumerili (Steindacher, 1869). South African Journal of Science 77, 73e78. Vivier, L., Cyrus, D.P., 2002. Ichthofauna of the sub-tropical Nhlabane Estuary, KwaZulu-Natal: drought-related changes in the fish community during extended mouth closure. Marine and Freshwater Research 53, 457e464. 20 J.R. Lukey et al. / Estuarine, Coastal and Shelf Science 67 (2006) 10e20 Whitfield, A.K., 1998. Biology and ecology of fishes in southern African estuaries. Ichthyological Monographs of the J.L.B. Institute of Ichthyology, No. 2, 223 pp. Whitfield, A.K., 1999. Ichthyofaunal assemblages in estuaries: a South African case study. Reviews in Fish Biology and Fisheries 9, 151e186. Whitfield, A.K., Beckley, L.E., Bennett, B.A., Branch, G.M., Kok, H.M., Potter, I.C., van der Elst, R.P., 1989. Composition, species richness and similarity of ichthyofaunas in eelgrass Zostera capensis beds of southern Africa. South African Journal of Marine Science 8, 251e259. Whitfield, A.K., Kok, H.M., 1992. Recruitment of juvenile marine fishes into permanently open and seasonally open estuarine systems on the southern coast of South Africa. Ichthyological Bulletin of the JLB Smith Institiute of Ichthyology 57, 1e39. Young, G.C., Potter, I.C., Hyndes, G.A., de Lestang, S., 1997. The ichthyofauna of an intermittently open estuary: implications of bar breaching and low salinities on faunal composition. Estuarine, Coastal and Shelf Science 45, 53e68. Vorwerk, P.D., Whitfield, A.K., Cowley, P.D., Paterson, A.W., 2001. A survey of selected Eastern Cape estuaries with particular reference to the ichthyofauna. Ichthyological Bulletin of the JLB Smith Institute of Ichthyology 72, 1e52. Vorwerk, P.D., Whitfield, A.K., Cowley, P.D., Paterson, A.W., 2003. The influence of selected environmental variables on fish assemblage structure in a range of southeast African estuaries. Environmental Biology of Fishes 66, 237e247. Wallace, J.H., 1975. The estuarine fishes of the east coast of South Africa. Part III. Reproduction. Investigational Report of the Oceanographics Research Institute 40, 1e72. Wallace, J.H., Kok, H.M., Beckley, L.E., Bennett, B.A., Blaber, S.J.M., Whitfield, A.K., 1984. South African estuaries and their importance to fishes. South African Journal of Science 80, 203e207. Weis, J.S., Weis, P., 2003. Is the invasion of the common reed, Phragmites australis, into tidal marshes of the eastern US an ecological disaster? Marine Pollution Bulletin 46, 816e820. Whitfield, A.K., 1992. A characterisation of southern African estuarine systems. South African Journal of Science 18, 89e103.