2012 NORTHEASTERN NATURALIST 19(3):445–460 Variability in Beach Seine Samples at Small Spatial and Temporal Scales in a Near-shore Estuarine Environment Kevin Teather1,*, Pamela MacDonald2, and Christina Pater1 Abstract - fish communities at nine sites in three estuaries, all emptying into Charlottetown Harbour, PE, Canada, were sampled with beach seines to assess variability at small spatial (at the same sites, between sites within estuaries, between adjacent estuaries) and temporal (minutes, hours, days, months, ebb and flood tides) scales. A total of 11 species were identifi ed, of which two (Fundulus heteroclitus [Mummichog] and Menidia menidia [Atlantic Silverside]) made up more than 90% of the individuals captured. Samples from the same sites taken 20–30 min apart did not differ with respect to number of individuals, number of species caught, or species diversity. However, the cumulative number of species continued to increase over the fi rst fi ve of six samples with repeated sampling at the same location. On larger spatial scales, communities (as measured by the Global R coeffi cients) differed more between sites within estuaries than between adjacent estuaries. Temporal variability in fi sh community composition was minimal and increased with increasing time between seine hauls. Time of day was weakly but positively correlated with the number of species captured, fi sh communities captured during flood and ebb tides did not differ signifi cantly, and slightly more species were captured in June than in either July or August. An understanding of variability in beach seine samples taken at small spatial and temporal scales is important before implementing sampling programs to look at broad-scale patterns in fi sh communities. Introduction Beach seining is extensively used to evaluate near-shore fish populations and communities. Topics investigated include population estimates of individual species that are of interest for conservation or commercial reasons (Wilson and Weisberg 1993) and environmental monitoring (Lowell et al. 2003), as well as for determining diurnal and tidal influence on fish community composition (e.g., Godefroid et al. 1998), community composition in different habitats (e.g., De Troch et al. 1996), and spatio-temporal patterns of distribution and abundance (e.g., Lekve et al. 1999, Paperno and Brodie 2004). Despite being widely used, sources of variation from sampling within small areas over a short time period are poorly understood. While some of this variation may reflect important underlying biotic and abiotic conditions, some may arise due to the difficulty in obtaining an accurate assessment of highly mobile animals in a restricted time and space. This study was designed to assess variability in beach seine samples taken over a series of small spatial and temporal scales using a potentially disruptive sampling technique. 1Department of Biology, University of Prince Edward Island, Charlottetown, PE, Canada C1A 4P3. 2Stantec, 165 Maple Hills Avenue, Charlottetown, PE, Canada C1C 1N9. *Corresponding author - kteather@upei.ca. 446 Northeastern Naturalist Vol. 19, No. 3 Sampling at the smallest spatial and temporal scales (at or near the same time and place) provides a measure of the repeatability of the sampling method. While repeated sampling of the same area may increase the reliability of measures of fi sh community structure, the act of sampling itself may result in signifi cant changes in community composition. Indeed, Wilson and Weisberg (1993) advised against using multiple samples from each site when beach seining, at least when sampling individual species. They found that repeated samples from the same site within a short time frame were not true replicates in the sense that the fi rst haul caught consistently more fi sh than the second. Sampling at progressively larger spatial scales (different sites within estuaries, adjacent estuaries) and longer temporal scales (hours, days, weeks) provides additional information on the sources of variability in beach seine samples. Larger distances between sampling sites and longer times between sampling are expected to result in greater differences in fish community structure (Remmert 1983). For example, Desmond et al. (2002) found that fish assemblages varied considerably among different estuaries, while sites sampled within estuaries were generally more similar. However, this relationship may not hold at all scales; Selleslagh et al. (2009) found that certain fish community parameters differed more between sites within estuaries than between estuaries. On larger temporal scales, fish communities can change dramatically due to seasonal migration and juvenile recruitment (Hagan and Able 2003, Lazzari et al. 1999). However, smaller-scale movements, often related to feeding and predator avoidance, may lead to changes in fish communities at shorter intervals, including diurnally (e.g., Hagan and Able 2008), between morning and afternoon (Willis et al. 2006), and with tidal changes (Gibson et al. 1996, Hampel et al. 2003, Jovanovic et al. 2007). Given the importance of beach seining in providing information on fish communities, it is necessary to establish the reliability of samples in estimating various measures of community structure. Although a number of studies have examined larger-scale spatial and temporal patterns of fish community structure, few studies have combined these efforts with a repeated sampling technique to assess the dependability of their endeavours over short distances and time intervals. Methods Study area and beach seining This study was conducted on the North (N), West (W), and East (E) rivers that empty into Charlottetown Harbour on the south shore of Prince Edward Island, Canada (fig. 1). Three estuarine sites on each river were sampled repeatedly during the period between 8 June and 12 August 2005. Sites were chosen based on similarity in depth, type of sediment, and vegetation as well as ease of access. The average distance between these sites was 4.7 ± 2.9 km. Beach seining was conducted 22 times at each site with the exceptions of W3 (18 samples) and W1 (20 samples) for a total of 192 samples over 65 days. 2012 K. Teather, P. MacDonald, and C. Pater 447 fish communities were sampled using a 30-m x 2-m beach seine having a bar mesh diameter of 0.635 cm and a central bag measuring 2 m x 2 m x 1 m. The seine was walked out at an angle of 120° from shore (in the direction of the tide) for 15 m and then brought back to shore in a gradual arc walking against the tide. The total area sample was estimated to be 225 m2 for each sweep. fish were removed from the net, placed in a large water-fi lled tub, identifi ed, counted, and returned to the water. While young-of-the-year of some species were captured (primarily sticklebacks and Fundulus heteroclitus L. [Mummichog]), they were not counted because many were observed slipping through the mesh. After each sample, we recorded water temperature (ºC), dissolved oxygen concentration (mg/L), and salinity (ppt) using a YSI 650 MDS multi-meter. For each site we also visually estimated the amount of vegetation (% cover) in a 0.5-m2 quadrat at 1-m intervals along a transect that extended approximately 12 m perpendicular from the shore. Sampling design We assessed the repeatability of sampling in two ways. first, for all samples with the exception of those taken during the fi rst week, sites were sampled figure 1. Beach seining sites on the three rivers leading into Charlottetown Harbour, PE, Canada–West (W), North (N), and East (E). 448 Northeastern Naturalist Vol. 19, No. 3 twice, with the second sample being taken immediately after processing the fi rst (approximately 20 min) at a location approximately 20 m away. There were no obvious differences in vegetation or substrate between the two sites. Secondly, at each site we took six consecutive samples separated by 30 minutes at the same location. This procedure was done once at each of the nine sites between 14 June and 20 July, during either ebb or flood tides. The sampling schedule was designed to assess variability in fi sh abundance, species richness (number of taxa), and diversity (Shannon index [H']; Krebs 1999) as it relates to increasing spatial and temporal scales. During one day, all three sites along one of the three rivers were sampled four times, twice as the tide was going out (ebb tide) and twice as the tide was coming in (flood tide), avoiding either high or low tide periods. The fi rst two samples were collected between 06:00 and 12:00 while the third and fourth samples were collected in the same way between 11:00 and 17:00. Exact sampling times depended on the stage of the tide; we attempted to sample individual sites when water depth was similar for ebb and flood tides. The six sites on the other two rivers were sampled in the same way on the following two days. One week later, the same sites were sampled again; this method permitted us to reverse the sampling by tidal stage during morning and afternoon periods. This entire procedure was repeated in June, July, and August. Data analyses We fi rst determined whether fi sh communities differed spatially and/or temporally using PRIMER®, version 6.1.6 (PRIMER-E Ltd, Plymouth, UK). Abundance data (fourth root transformed) were used to generate Bray-Curtis resemblance matrices. Two dimensional non-metric multidimensional scaling (nMDS) plots were used to visually examine the spatial and temporal similarity between fi sh communities. Analysis of similarity (ANOSIM) tests were then carried out to determine whether there were signifi cant differences between groups of community samples. The ANOSIM test compares similarities between species assemblages using permutation/randomization tests and provides a Global R metric ranging from 0 (communities are identical) to 1 (communities are completely different) (Clarke and Gorley 2006). Global R values greater than 0.75 indicate that the groups are well separated, those between 0.50 and 0.75 indicate that the groups are overlapping but still different, those between 0.25 and 0.50 suggest that the groups are overlapping but somewhat different, and values below 0.25 infer that the groups are insuffi ciently different to separate them (Clarke and Gorley 2001). To provide more specific information concerning potential differences between fish communities, we examined three measures: total number of species captured, total number of individuals captured, and species diversity as determined by the H' (Krebs 1999). All variables were assessed for normality using Kolmogorov-Smirnov tests. When variables were not normally distributed, we attempted to normalize them using appropriate transformations. When this 2012 K. Teather, P. MacDonald, and C. Pater 449 was unsuccessful, differences were tested using nonparametric procedures (Kruskall-Wallis or Mann-Whitney U tests). Otherwise, standard parametric procedures were used (paired t-tests or one-way ANOVA with Tukey’s posthoc tests). Results Environmental variability The nine sites were largely similar with respect to abiotic variables (Table 1) and vegetation cover with a few notable exceptions. While there was no difference between sites with respect to temperature (range between sites: 17.8 ºC ± 2.9 [SD] to 19.5 ± 3.3, ANOVA: P = 0.767) or dissolved oxygen concentrations (7.1 mg/L ± 2.7 to 8.9 ± 2.1, P = 0.513), there was a signifi cant difference in salinity with N3 (23.6 ppt ± 3.7) being lower than N1 (27.2 ± 0.9), E1 (27.0 ± 0.6), or E2 (26.5 ± 0.5) (ANOVA: P = 0.026). In general, sites had little or no rooted vegetation, with the exception of N1, where it was estimated that 60% of the sampling area was covered by Zostera marina L. (Eelgrass). fish species composition In total, eleven species of fi sh were captured: Gasterosteus wheatlandi Putnam (Blackspotted Stickleback), Pungitius pungitius L. (Ninespine Stickleback), Apeltes quadracus Mitchill (Fourspine Stickleback), Gasterosteus aculeatus L. (Threespine Stickleback), Mummichog, Menidia menidia L. (Atlantic Silverside), Pleuronectes putnami Gill (Smooth Flounder), Pseudopleuronectes americanus Walbaum (Winter Flounder), Salvelinus fontalis Mitchill (Brook Trout), Myoxocephalus aenaeus Mitchill (Grubby), and Scomber scombrus L. (Atlantic Mackerel) (Table 2). However, two species (Mummichog and Atlantic Silverside) made up more than 90% of fi sh captured. Only two Grubby and three Table 1. Abiotic conditions and vegetation cover (mean ± SD) at each study site and for each river. n = 22 at each site with the exceptions of W3 (n = 18) and W1 (n = 20). Riversite Vegetation cover (%) Temperature (°C) Salinity (ppt) Dissolved oxygen (mg/L) North N1 63.2 17.8 ± 2.9 27.2 ± 0.9 8.9 ± 2.1 N2 0.3 18.0 ± 3.0 26.0 ± 1.7 7.9 ± 1.1 N3 0.0 18.4 ± 3.1 23.6 ± 3.7 8.2 ± 1.4 All N sites 18.1 ± 2.9 25.6 ± 2.8 8.4 ± 1.6 West W1 5.9 19.5 ± 3.3 26.1 ± 2.2 8.0 ± 1.5 W2 0.0 18.6 ± 3.0 25.5 ± 1.2 8.2 ± 0.9 W3 0.0 19.4 ± 2.7 26.3 ± 0.7 7.7 ± 0.6 All W sites 19.2 ± 2.9 26.0 ± 1.5 8.0 ± 1.0 East E1 0.0 18.5 ± 2.6 27.0 ± 0.6 7.9 ± 1.1 E2 2.1 19.2 ± 2.6 26.5 ± 0.5 7.5 ± 1.4 E3 10.0 18.8 ± 2.3 26.0 ± 0.7 7.1 ± 2.7 All E sites 18.8 ± 2.5 26.5 ± 0.7 7.5 ± 1.8 450 Northeastern Naturalist Vol. 19, No. 3 Table 2. Number of individuals of each species caught (n ranged from 18 to 22). bs = blackspotted stickleback, 3-sp = threespine stickleback, 4-sp = fourspine stickleback, 9-sp = ninespine stickleback, mum = mummichog, silv = Atlantic silverside, sm fl= smooth flounder, wi fl= winter flounder, bt = brook trout, grub = grubby, mack = mackerel. River/site bs 3-sp 4-sp 9-sp mum silv sm flwi flbt grub mack total North N1 11 81 277 6 353 105 29 20 0 0 0 882 N2 9 37 103 26 4784 4784 18 22 1 0 0 6929 N3 13 28 7 16 356 356 184 63 0 0 0 3867 Total 33 146 387 48 5482 5245 231 105 1 0 0 11,678 West W1 39 3 33 3 18981 497 33 20 0 0 0 19,611 W2 11 431 28 400 3344 1997 0 6 1 2 482 66,702 W3 11 270 16 131 1418 1020 33 22 1 0 310 3232 Total 61 706 77 534 23743 3514 66 48 2 2 0 28,753 East E1 9 8 6 6 1894 1111 24 5 0 0 0 3063 E2 5 5 82 8 5436 690 39 7 0 0 0 6272 E3 4 4 128 38 2046 468 24 8 0 0 0 2720 Total 18 17 216 52 9376 2269 87 20 0 0 0 12,055 Total 112 869 680 634 38601 11028 384 173 3 2 792 53,278 2012 K. Teather, P. MacDonald, and C. Pater 451 Brook Trout were captured, while Atlantic Mackerel were captured at only two sites during the fi nal sampling period. The smallest spatiotemporal scale: Repeatability of samples We fi rst assessed the repeatability of beach seining by comparing samples taken consecutively (within 20–30 min of each other from two locations approximately 20 m apart). The total number of species captured did not differ between samples (3.89 ± 1.70 [SD] vs. 3.73 ± 1.55, paired t-test: P = 0.361, n = 88). The total number of individuals captured did not differ between the fi rst and second samples (305.7 ± 540.9 vs 271.2 ± 550.8, paired t-test on transformed data: P = 0.369), although the large amount of variation in catches may have masked potential differences. finally, species diversity (H') of the fi rst (0.515 ± 0.330) and second (0.530 ± 0.445) samples did not differ signifi cantly (Mann-Whitney: P = 0.608). Thus, for subsequent analyses, the mean number of species, individuals captured, and species diversity from the two samples combined was used. To further examine sample repeatability, we compared the results from six consecutive beach seines, each separated by 30 minutes, carried out once at each of the nine study sites. Consecutive samples revealed no difference in the total number of species captured (one-way ANOVA: P = 0.261) nor in the number of individuals captured (examined as a proportion of the total individuals captured at each site to control for overall differences in numbers at each site) (P = 0.986). However, the cumulative number of species increased over the fi rst fi ve of the six samples collected (fig. 2). Variability at increasing spatial scales Sites within estuaries had varying differences in fi sh communities (fig. 3a). On the North River, N1 and N3 exhibited the largest difference in fi sh community assemblage (Global R = 0.423), while N2 and N3 were relatively similar (Global R = 0.067). fish communities along the West River were substantially different figure 2. Cumulative number of species (mean ± se, n = 9 sites) captured from six consecutive samples, separated by 30 min, at the same site. 452 Northeastern Naturalist Vol. 19, No. 3 between W1 and W3 (Global R = 0.539), while W2 and W3 are the most similar (Global R = 0.151). fish assemblages in both these estuaries differed largely on a geographic scale between the upstream and downstream sites. All sites on the East River were relatively similar (Global R = 0.044 to 0.104). Similarities in fi sh communities among the three estuaries, based on samples combined from the three sites on each, was generally higher than that between sites within estuaries with Global R values of 0.203 (West-North), 0.179 (North-East) and 0.135 (West-East). To further assess the possible relationship between distance and fi sh community differences, we correlated Global R between each pair of the nine sites and absolute distances between sites along river courses. There was no signifi cant relationship (Spearman’s rho = -0.122, P > 0.05). Comparison of fi sh communities using more specifi c metrics provides further information concerning how fi sh communities differed between sites within and between rivers (Table 3). There were no differences in the number of species captured at sites on the North River (ANOVA: P = 0.189), although there were slightly fewer individuals captured at N1 than either of the other two sites (ANOVA: P = 0.061). N1 also exhibited greater species diversity than the other two sites (Kruskal-Wallis: P = 0.048). On the West River, there were more species captured at W3 than at W1 (ANOVA: P = 0.005). However, signifi cantly more individuals were captured at W1 (ANOVA: P = 0.007), which also exhibited signifi cantly lower diversity than either of the other two sites on the West River (Kruskal-Wallis P = 0.007). Among the East River sites, there were no differences in the number of species captured (ANOVA: P = 0.955), in the total number of individuals captured (P = 0.196), or in the species diversity (Kruskal- Wallis P = 0.943). Table 3. Total species captured, number of species captured per sample, number of fi sh captured per seine, and species diversity (Shannon index [H']) for each of the nine sites. Numbers (mean ± sd) combine samples taken at incoming and outgoing tide, morning and afternoon, and in June, July, and August (n = 11 for North and West River sites and n = 10 for East River sites). Statistical differences between rivers and sites are described in the text. River/Site Total species Species/seine Individuals/seine H' North N1 8 6.33 ± 0.98 73.5 ± 68.4 1.250 ± 0.213 N2 9 5.50 ± 1.08 577.4 ± 1258.2 0.705 ± 0.547 N3 8 5.16 ± 1.70 322.3 ± 422.9 0.786 ± 0.413 10 5.67 ± 0.60 324.4 ± 251.9 0.917 ± 0.298 West W1 8 4.10 ± 2.13 1961.1 ± 1542.6 0.254 ± 0.427 W2 10 5.16 ± 1.52 558.5 ± 424.4 0.596 ± 0.520 W3 10 6.00 ± 1.49 323.2 ± 250.1 0.785 ± 0.332 11 5.09 ± 0.95 947.6 ± 885.5 0.545 ± 0.269 East E1 8 3.91 ± 1.78 255.3 ± 394.0 0.397 ± 0.407 E2 8 4.16 ± 1.52 522.6 ± 476.8 0.413 ± 0.416 E3 8 3.75 ± 1.06 226.7 ± 224.3 0.417 ± 0.334 9 3.94 ± 0.209 334.9 ± 163.3 0.409 ± 0.010 2012 K. Teather, P. MacDonald, and C. Pater 453 There were also a number of differences in fi sh communities between adjacent rivers. Signifi cantly fewer species were captured per seine haul at the East River than either the North River or West River (ANOVA: P < 0.01; Table 3). In addition, there were signifi cantly more fi sh captured per beach seine haul at the West than either the North or East rivers (ANOVA on transformed data: P < 0.01). However, species diversity of the North River was higher than that of either the West or East rivers (ANOVA on transformed data: P < 0.01). Variability at increasing temporal scales There was little or no difference in fi sh community assemblages captured in the morning or afternoon (Global R = 0.020), or during ebb and flood tides (Global R = 0.000). Beach seine samples taken at progressively larger time intervals generally showed correspondingly greater differences in fi sh communities (fig. 3b). Global R’s from the one-week sampling intervals in June (8–10, 13–15) and July (6–10, 13–15) were 0.004 and 0.052, respectively, indicating almost identical communities. However, the fi sh community changed signifi cantly during the last sampling period (Aug 10–12), primarily because of a large number of Atlantic Mackerel captured at this time. The mean Global R for between-month differences was 0.155, with the most pronounced differences between August and both June (Global R = 0.259) and July (Global R = 0.211). We expected that differences in water parameters for flood and ebb tides might contribute to differences in fi sh communities between these times. However, figure 3. Multidimensional scaling (MDS) Plot showing variation in fi sh communities between (a) nine sampling sites on the three rivers and (b) different sampling times. Points closer together are indicative of more similar communities. 454 Northeastern Naturalist Vol. 19, No. 3 differences in most measured variables were relatively minor and probably did not affect fi sh distribution considerably. For example, water temperature (18.80 ± 2.70 [SD] °C, n = 49) and salinity (26.26 ± 1.46 ppt) of flood tides were slightly higher than those of ebb tides (18.24 ± 2.58 °C and 25.79 ± 2.22 ppt, paired t-tests P = 0.074 and 0.019, respectively). There was no difference in the concentration of dissolved oxygen for flood (8.11 ± 1.51 mg/L) and ebb (7.82 ± 1.16 mg/L) tides (P = 0.252). Similarly, fi sh communities were similar for both tidal periods. There was no difference in the number of species captured (4.40 ± 1.47 vs. 4.04 ± 1.57; paired t-test: P = 0.102), in the number of individuals captured (289.1 ± 441.3 vs. 247.7 ± 411.9; paired t-test done on transformed data: P = 0.273), or in species diversity (0.511 vs. 0.421; Mann-Whitney: P = 0.491) during flood and ebb tides. To eliminate the extreme variability in numbers of fi sh captured between sites and to facilitate temporal comparisons, we converted “number of individuals captured” to proportional values. Thus, values from each sample are subsequently provided as a “proportion of all individuals captured at that site”. Time of day was weakly correlated with the number of species caught (r = 0.217, P = 0.034), but did not correlate with the proportion of individuals captured (as a function of the 10–11 samples taken from each site) (r = 0.034, P = 0.743), or diversity (r = 0.095, P = 0.357). The mean number of species caught was slightly higher in June than in July or August, although this difference was not signifi cant (ANOVA: P = 0.068; Table 4). The proportion of fi sh captured each month did not differ signifi cantly (ANOVA: P = 0.461) and there was no difference in the diversity of species captured each month (P = 0.251). finally, we examined which environmental variables were most highly correlated with fi sh community structure. Using all sites, for all three months, only temperature was correlated with the variation in fi sh communities between sites (BEST analysis: Spearman’s rho = 0.239, P < 0.01). When months were analyzed separately, none of the abiotic variables was correlated with differences between sites in either June or July; in August, temperature was again correlated with fi sh community structure (Spearman’s rho = 0.229, P < 0.01). Discussion To better understand factors influencing the structure of fi sh communities, it is important to quantify variability at small scales when habitat variability is expected to be low as well as situations in which it is high. Ideally, fi sh communities sampled at approximately the same time and place will be identical; Table 4. Monthly comparisons (mean ± SD, n = 9) of number of species captured per seine, the proportion of individuals captured, and mean species diversity (H'). Month Species/seine Proportion of individuals captured Mean H' June 5.21 ± 1.45 0.331 ± 0.173 0.772 ± 0.548 July 4.92 ± 1.89 0.297 ± 0.182 0.511 ± 0.458 August 4.58 ± 1.70 0.392 ± 0.158 0.616 ± 0.446 2012 K. Teather, P. MacDonald, and C. Pater 455 however, variability due to sampling is unavoidable, particularly for highly mobile animals that may move in groups. As spatial sampling scales increase, variability in fi sh communities is also expected to increase, reflecting real differences due to changes in their environment; fi sh communities are also expected to be more variable with increasing temporal scales, within a year at least, largely due to changes in life-history stages. An understanding of variability at different spatial scales is particularly important for monitoring programs where sampling is not randomized to account for time of day (Willis et al. 2006) or that assume seine hauls done in close proximity or within an estuary are considered replicates (Weldon et al. 2008, Wilson and Weisberg 1993). While the capture efficiency of seines is known to be associated with habitat (Parsley et al. 1989, Pierce et al. 1990), net length (Vanderklift et al. 1998), and position of fish in the water column (Lyons 1986), few studies have examined the repeatability of samples taken from the same site at the same time. Wilson and Weisberg (1993) found, when sampling Morone saxatilis Walbaum (Striped Bass), second samples at the same site contained consistently fewer fish and thus could not be considered replicates. While recommending against repeated sampling, Wilson and Weisberg did not indicate the time interval between first and second samples or whether second samples were taken at exactly the same location. Both of these will no doubt influence the independence of the two samples. Based on 79 paired comparisons at nine different sites, with the second sample offset from the first spatially by 20 m and temporally by about 20–30 minutes, we found no differences in number of species caught, number of individuals caught, or species diversity. During six consecutive samples at each of the sites, we repeatedly sampled the same spot, but temporally separated samples by about 30 minutes. Again, no differences were apparent in the number of fish captured. However, repeated samples at the same location were more likely to include uncommon species, and ultimately provided a more accurate census of community composition. Clearly, whether repeated samples should be used will depend on habitat structure, types of species captured, and the kind of information that is required. We subsequently assessed variability in fish communities that could be attributed to progressively larger spatial (different sites within estuaries and different estuaries) and temporal (morning vs. afternoon, ebb vs. flood tides, different weeks and months) scales. We found little difference in the fish communities between rivers. Rivers were adjacent to each other and opened into a common harbor; combining the samples collected from various sites probably ensured that we sampled most nearshore fish in each river. For some parameters, fish community differences between sites within rivers were greater than those between rivers. Selleslagh et al. (2009) also found that fish assemblages were more variable between sites within estuaries than between estuaries and attributed this to salinity gradients for each site. While salinity levels were similar for most of our sites, some of the variation in community assemblage between sites was likely due to differences in the amount of submerged aquatic 456 Northeastern Naturalist Vol. 19, No. 3 vegetation. The first site (N1) along North River, for example, contained about 60% vegetation, far more than any of the other eight sites. While the number of individuals captured there was low, the number of species, and thus species diversity (H'), was significantly greater than that of other sites. Temperature was the lone abiotic factor correlated with fish community structure. This correlation was apparent over the course of the study period as well as for between-site variation, at least during the last sampling period. Adams (1976), while studying a similar habitat in North Carolina, found similar differences in fish communities due to both vegetation and temperature. fish community differences in beach seine samples increased as temporal intervals between sampling increased. On a daily basis, slightly more species were captured in the afternoon than in the morning. Previous studies have found little variation in composition of fish assemblages during daylight hours (Spyker and Van den Berghe 1995, Thompson and Mapstone 2002, Willis et al. 2006), although Willis et al. (2006) noted differences in the relative abundance of some reef fishes in the morning and afternoon. The effects of time of day are typically more obvious between nocturnal and diurnal assemblages of fish when movements may be related to feeding and/or predator avoidance (Dulčić et al. 2004, Morrison et al. 2002, Nash et al. 1994, Pessanha and Araújo 2003, Pierce et al. 1990). We found no differences in species numbers, the number of individuals captured, or the species diversity during ebb or flood tides. Studies examining the influence of tidal differences on variability in capture rates have generally compared high versus low tides, rather than ebb and flood tides (Gibson et al. 1996, Kneib and Wagner 1994, Morrison et al. 2002). For example, Morrison et al. (2002) found that low tide samples had higher species diversity and abundance than high tide samples. fish associated with intertidal habitats and associated vegetation must cope with regularly shrinking and expanding environments due to the tidal variability and thus move inshore and offshore with the changing of tides. Movement of fi sh into the intertidal zone is usually related to feeding on intertidal organisms that are only available at high tide (Gibson et al. 1996). It has also been suggested that movement offshore at these times may be to avoid predators that move inshore at high tides (Abou-Seedo et al. 1990). While habitat differences experienced by fi sh at high and low tides may be primarily responsible for differences in fi sh communities, habitats will largely remain the same for ebb and flood tides. Because water depth was similar for both sampling periods in this study, access to different habitats was unimportant. We also found few differences in salinity and temperature that might affect species movements or distribution. We acknowledge, however, that tidal range is relatively low along the southwestern coast of Prince Edward Island; differences in fi sh communities due to tidal influence may be much greater in regions, such as the Bay of Fundy, where tidal range is much greater. Although fi sh communities at the nine sites remained relatively constant over the three months of the study, the changes that did occur probably reflect seasonal 2012 K. Teather, P. MacDonald, and C. Pater 457 movements or changes in habitat utilization. For example, slightly more species were captured in June, than either July or August. A similar trend was noted by Weldon et al. (2008), who found that the number of species captured in most estuaries sampled around the Gulf of St. Lawrence peaked in June and declined slightly in the following months. Dulčić et al. (2004) documented seasonal differences in near-shore fi sh communities in the middle Adriatic, with species diversity and evenness being highest between April and July and then decreasing in August. Longer-term studies examining fi sh communities with beach seines have noted pronounced seasonal differences in one or more of species numbers, diversity, or evenness (e.g., Dulčić et al. 2004, Lazzari et al. 1999, Pombo et al. 2002). In our study, almost 800 Atlantic Mackerel were captured at two sites during the last week of the study, resulting in a dramatic difference in fi sh community composition during this time. Factors responsible for variation in species composition of a sample of fi sh can be divided into two groups. The fi rst are real biotic or abiotic factors of interest to most researchers, including habitat structure, seasonality, tidal influence, etc. The second group includes factors that blur this variation by introducing perceived variability where none may exist. These include differences resulting from the biases of specifi c sampling methods as well as the repeatability of individual samples taken using the same method. In this study, we examined variation in beach seine samples at different spatial and temporal scales in an effort to separate these two sources of variation. The results are of particular importance when developing sampling regimes to analyze potential differences in fi sh community structure separated in time and space. In eastern Canada, for example, fisheries and Oceans Canada has implemented the Community Aquatic Monitoring Program (CAMP) to help assess the health and productivity of local estuaries using methods almost identical to the ones in this study to evaluate fi sh communities. Identifi cation and analyses of both temporal and spatial sources of variation are crucial to the interpretation of those data (DFO 2011). Although conducted at a relatively small scale, the results of this study suggest that some factors that sampling protocols should take into account are: 1) the number of times a site should be sampled to obtain an accurate picture of the fi sh community, 2) the time of day sampling takes place (particularly when comparing sites sampled at different times), 3) weekly and seasonal differences in community composition and species numbers, 4) minor differences in vegetation and water temperature between sampling sites that may have important effects on fi sh community composition, and 5) variation at sites within estuaries that may be more extensive than differences between neighboring estuaries. Acknowledgments Thanks to Colette Cheverie for her dedicated assistance with fi eld work throughout this project. Delephina Keen of fisheries and Oceans Canada provided advice, fi eld assistance, and technical support, while Daryl Guignion periodically offered a strong back for pulling seines. The manuscript was signifi cantly improved with the comments of 458 Northeastern Naturalist Vol. 19, No. 3 Simon Courtenay and three anonymous reviewers. Funding and support was provided by the National Jobs for Youth Program, the University of Prince Edward Island, and an anonymous donor. Literature Cited Abou-Seedo, F., D.A. Clayton, and J.M. Wright. 1990. Tidal and turbidity effects on the shallow-water fi sh assemblage of Kuwait Bay. Marine Ecology Progress Series 65:213–223. Adams, S.M. 1976. The structure of Eelgrass, Zostera marina (L.), fi sh communities. I. Structural analysis. Journal of Experimental Marine Biology and Ecology 22:269–291. Clarke, K.R., and R.N. Gorley. 2001. Primer V5: User manual/tutorial. Primer-E Ltd. West Hoe, Plymouth, UK. 91 pp. Clarke, K.R., and R.N. Gorley. 2006. Primer V6: User manual/tutorial. Primer-E Ltd. West Hoe, Plymouth, UK. 190 pp. Department of fisheries and Oceans Canada (DFO). 2011. Proceedings of a regional advisory process meeting to review the Community Aquatic Monitoring Program (CAMP) and its use to infer the ecological health of bays and estuaries in the southern Gulf of St. Lawrence. March 17–18, 2010. DFO Canadian Science Advisory Secreteriat Proceedings Series 2011/029. Available online at http://www.dfo-mpo.gc.ca/ csas-sccs/Publications/Pro-Cr/2011/2011_029-eng.html. Desmond, J.S., D.H. Deutschman, and J.B. Zedler. 2002. Spatial and temporal variation in estuarine fi sh and invertebrate assemblages: Analysis of an 11-year data set. Estuaries 25:552–569. De Troch, M., J. Mees, I. Papadopoulos, and E.O. Wakwabi. 1996. fish communities in a tropical bay (Gazi Bay, Kenya): Seagrass beds vs. unvegetated areas. Netherlands Journal of Zoology 46:236–252. Dulčić, J., B. Grbec, L. Lipej, G. Beg Paklar, N. Supić, and A. Smirčić. 2004. The effect of the hemispheric climatic oscilations on the Adriatic ichthyofauna. Fresenious Environmental Bulletin 13:293–298. Gibson, R.N., L. Robb, M.T. Burrows, and A.D. Ansell. 1996. Tidal, diel, and longerterm changes in the distribution of fi shes on a Scottish sandy beach. Marine Ecology Progress Series 130:1–17. Godefroid, R.S., M. Hofstaetter, and H.L. Spach. 1998. Moon, tidal, and diel influences on catch composition of fi shes in the surf zone of Pontal do Sul Beach, Parana. Revista Brasileira de Zoologia 15:697–701. Hagan, S.M., and K.W. Able. 2003. Seasonal changes of the pelagic fi sh assemblage in a temperate estuary. Estuarine, Coastal, and Shelf Science 56:15–29. Hagan, S.M., and K.W. Able. 2008. Diel variation in the pelagic fi sh assemblage in a temperate estuary. Estuaries and Coasts 31:33–42. Hampel, H., A. Cattrijsse, and M. Vincx. 2003. Tidal, diel, and semi-lunar changes in the faunal assemblage of an intertidal saltmarsh creek. Estuarine, Coastal, and Shelf Science 56:795–805. Jovanovic, B., C. Longmore, A. O’Leary, and S. Mariani. 2007. fish community structure and distribution in a macro-tidal inshore habitat in the Irish Sea. Estuarine, Coastal, and Shelf Science 75:135–142. Kneib, R.T., and S.L. Wagner. 1994. Nekton use of vegetated marsh habitats at different stages of tidal inundation. Marine Ecology Progress Series 106:227–238. 2012 K. Teather, P. MacDonald, and C. Pater 459 Krebs, C.J. 1999. Ecological Methodology. Addison Wesley Longman, Menio Park, CA. Lazzari, M.A., S. Sherman, C.S Brown, J. King, B.J. Joule, S.B. Chenoweth, and R.W. Langton. 1999. Seasonal and annual variations in abundance and species composition of two nearshore fi sh communities in Maine. Estuaries 22:636–647. Lekve, K., N.C. Stenseth, J. Gjøsæter, J.M. Fromentin, and J.S. Gray. 1999. Spatiotemporal patterns in diversity of a fi sh assemblage along the Norwegian Skagerrak coast. Marine Ecology Progress Series 178:17–27. Lowell, R.B., S.C. Ribey, I.K. Ellis, E.L. Porter, J.M. Culp, L.C. Grapentine, M.E. Mc- Master, K.R. Munkittrick, and R.P. Scroggins. 2003. National assessment of pulp and paper environmental effects monitoring data. National Water Research Institute (NWRI) contribution No. 03-521. Fountain Valley, CA. 124 pp. Lyons, J. 1986. Capture effi ciency of a beach seine for seven freshwater fi shes in a northtemperate lake. North American Journal of fisheries Management 6:288–289. Morrison, M.A., M.P. Francis, B.W. Hartill, and D.M. Parkinson. 2002. Diurnal and tidal variation in the abundance of the fi sh fauna of a temperate tidal mudflat. Estuarine, Coastal, and Shelf Science 54:793–807. Nash, R.D.M., R.S. Santos, A.J. Geffen, G. Hughes, and T.R. Ellis. 1994. Diel variability in catch rate of juvenile flatfi sh on two small nursery grounds (Port Erin Bay, Isle of Man and Porto Pim Bay, Faial, Azores). Journal of fish Biology 44:35–45. Paperno, R., and R.B. Brodie. 2004. Effects of environmental variables upon the spatial and temporal structure of a fi sh community in a small, freshwater tributary of the Indian River Lagoon, Florida. Estuarine, Coastal, and Shelf Science 61:229–241. Parsley, M.J., D.E. Palmer, and R.W. Burkhardt. 1989. Variation in capture effi ciency of a beach seine for small fi shes. North American Journal of fisheries Management 9:239–244. Pessanha, A.L., and F.G. Araújo. 2003. Spatial, temporal, and diel variations of fi sh assemblages at two sandy beaches in the Sepetiba Bay, Rio de Janeiro, Brazil. Estuarine, Coastal, and Shelf Science 57:817–828. Pierce, C.L., J.B. Rasmussen, and W.C. Leggett. 1990. Sampling littoral fi sh with a seine: Corrections for variable capture effi ciency. Canadian Journal of fisheries and Aquatic Science 47:1004–1010. Pombo, L., M. Elliott, and J.E. Rebelo. 2002. Changes in the fi sh fauna of the Ria de Aveiro estuarine lagoon (Portugal) during the twentieth century. Journal of fish Biology 61:167–181. Remmert, H. 1983. Studies and thoughts about the zonation along the rocky shores of the Baltic. Zoologica 22:121–125. Selleslagh, J., R. Amara, P. Laffargue, S. Lesourd, M. Lepage, and M. Girardin. 2009. fish composition and assemblage structure in three Eastern English Channel macrotidal estuaries: A comparison with other French estuaries. Estuarine, Coastal, and Shelf Science 81:149–159 Spyker, K.A., and E.P. Van den Berghe. 1995. Diurnal abundance patterns of Mediterranean fi shes assessed on fi xed transects by scuba divers. Transactions of the American fisheries Society 124:216–224. Thompson, A.A., and B.D. Mapstone. 2002. Intra- versus inter-annual variation in counts of reef fi shes and interpretations of long-term monitoring studies. Marine Ecology Progress Series 232:247–257. 460 Northeastern Naturalist Vol. 19, No. 3 Vanderklift, M.A., T.J. Ward, and J.C. Phillips. 1998. Use of assemblages derived from different taxonomic levels to select areas for conserving marine biodiversity. Biological Conservation 86:307–315. Weldon, J., S. Courtenay, and D. Garbary. 2008. The community aquatic monitoring program (CAMP) for measuring marine environmental health in coastal waters of the southern Gulf of St. Lawrence: 2007 overview. Canadian Technical Reports of fisheries and Aquatic Sciences 2825:viii + 75 pp. Willis, T.J., F. Badalamenti, and M. Milazzo. 2006. Diel variability in counts of reef fi shes and its implications for monitoring. Journal of Experimental Marine Biology and Ecology 331:108–120. Wilson, H.T., and S.B. Weisberg. 1993. Design considerations for beach seine surveys of Striped Bass. North American Journal of fisheries Management 13:376–382.