Seasonal fish community variation in headwater mangrove creeks in the southwestern Everglades: an examination of their role as dry-down refuges

Home Methods >Results Discussion Acknowledgments Literature Cited Tables and Figures PDF Version

Freshwater Flow and Salinity. --Salinity levels near our study sites increased substantially with the yearly onset of the dry season as flows from upstream freshwater marshes decreased (Fig. 2). In 2005, these increases occurred earlier in NW creeks than in RB creeks (Fig. 2A). Maximum salinities were higher in the vicinity of our NW sites than near RB sites (29 and 19, respectively). Hydroperiod was shorter in marshes upstream of NW sites than in those upstream of RB sites. Marshes upstream of NW sites were flooded for 357 d in 2004-2005; whereas marshes upstream of RB creeks were flooded for 324 d (Fig. 2B). Marshes upstream of RB also dried more frequently, but for short periods of time. In contrast, marshes upstream of the North River dried less frequently, but once dry, remained dry for a longer period of time. At our study creeks, salinity varied both among sites and across seasons (ANOVA: significant season by region interaction, F_{2, 102} = 18.6, P = 0.0001; Fig. 3). Salinities were comparable between regions during the wet season sample, but diverged as the dry season progressed, reaching 10 in NW but < 5 in RB (Tukey pairwise comparisons of RB vs NW, P < 0.0001 for both the transition period and dry season).

Estimates of Abundance. --Fish abundance in the oligohaline to mesohaline reaches of mangrove creeks, as estimated by electrofishing, gill net, and minnow trap CPUE, was consistently higher in RB creeks than in NW creeks (Fig. 4, Fig. 5). CPUE varied as a function of season (significant season by region interactions for all three CPUEs, Table 2), but season had a different effect in the two sampled regions. Across sampling gears, CPUE was highest in RB samples during the transition period. CPUE increased four-fold in electrofishing samples, eight-fold in gill nets, and nine-fold in minnow traps between the wet and transition samples (electrofishing, P = 0.009; gill nets, P = 0.002; minnow traps, P = 0.0001). The abundance of large species, such as Florida gar, bowfin, snook, largemouth bass, Mayan cichlid, and sunfishes peaked in the transition period. Among the small fishes, catches of bluefin killifish, eastern mosquitofish, coastal shiners, and smaller-bodied sunfishes also peaked during the transition period (Table 1). RB CPUE decreased significantly, returning to wet-season levels, in the dry season for electrofishing and minnow traps, but not for gill nets (electrofishing, P = 0.003; minnow traps, P = 0.0001).

Table 2. [click here to see entire table]

---

Figure 4. (A) Electrofishing and (B) gill net catch-per-unit effort (CPUE) over the three sampling seasons in the two study regions. Shown are means ± 1 SE. [larger image]

---

Figure 5. CPUE in minnow traps shown separately by trap placement: (A) top vs (B) bottom of water column) over the three sampling seasons and in the two study regions. Shown are means ± 1 SE. [larger image]

---

Figure 6. Estimates of large fish abundance (CPUE) in electrofishing samples (Log-transformed) plotted as a function of salinity. Separate least-squares regressions were fitted to the two regions: Rookery Branch (RB, solid line) and the North and Watson rivers (NW, dotted line). [larger image]

Figure 7. Two-dimensional non-metric MDS ordinations illustrating large fish (50-750 mm SL) community structure in (A) electrofishing samples, (B) gill net samples, and (C) gill net vs electrofishing samples based on Bray-Curtis similarities of log-transformed, standardized CPUE. [larger image]

All gears varied significantly among creeks within the two study regions, and this variation was affected by season (Table 2). In NW, electrofishing CPUE was higher in North River creeks than in Watson River creeks (P = 0.003), although CPUE in gill nets and minnow traps did not differ. Seasonally, electrofishing CPUE was higher in North River headwaters in the transition and dry-season samples (P = 0.06 and P = 0.0006, respectively), but not in the wet season. In RB, electrofishing CPUE was lower in Squawk Creek (RB 7, Fig. 1) than in other creeks, particularly in the transition sample (P < 0.05). Minnow trap and gill net CPUE were significantly higher in Otter Creek (RB 12, Fig. 1) than in Squawk Creek (P = 0.0006 and P = 0.02, respectively), while CPUE in other RB creeks was intermediate.

Large-fish Community Structure. --The composition of electrofishing catches varied equally between regions and among seasons (Table 3). Community structure was similar between NR and RB creeks in the wet season, but diverged considerably during the transition and dry seasons (wet, R = 0.12, P = 0.14; transition, R = 0.54, P = 0.002; dry, R = 0.74, P = 0.002; Fig. 7A). This divergence can be explained by increases in the relative contribution of freshwater taxa to the creek community. The contribution of freshwater species to CPUE was comparable between regions in the wet season (5% in NW vs 20% in RB; P = 0.225), but differed significantly in later samples (Fig. 8A). NW catches remained < 10% freshwater, whereas in RB, 80%-90% of the catch was composed of freshwater taxa during the transition and dry seasons (transition, P = 0.0011; dry season, P = 0.0001). During these drier samples, Florida gar, largemouth bass, bowfin, Mayan cichlid, and several sunfish species were almost exclusively caught in RB creeks. The relative abundance of snook was also higher in RB, whereas needlefishes and tidewater mojarras were exclusively caught in NW creeks. Composition of gill net samples was similar between regions during the wet season, but tended to differ in the dry season sample (wet, R = 0.29, P = 0.20; dry, R = 0.65, P = 0.10; Fig. 7B). Florida gar was dominant in RB gill net samples, whereas NW gill nets were dominated by a small number of striped mojarras.

Table 3. [click here to see entire table]

---

Small-fish Community Structure. --Variation in the small community structure was higher between regions than among seasons (Table 3, Fig. 9A), yet the proportion of freshwater species in traps varied as a function of both season and region (Table 4). The contribution of freshwater species to the RB small fish fauna showed no seasonal variation, averaging 96% throughout the study (Fig. 8B). In NW, however, the contribution of freshwater species decreased significantly between the wet and dry seasons, from 24% to 2% (P = 0.024). Minnow trap CPUE in NW primarily contained estuarine species (rainwater killifish, tidewater mojarra, and clown goby), whereas the RB community primarily contained freshwater species (eastern mosquitofish, sailfin molly, bluefin killifish, least killifish, dollar sunfish, and bluespotted sunfish; Table 1).

Table 4. [click here to see entire table]

---

Figure 8. Average proportion of (A) electrofishing and (B) minnow trap CPUE composed of freshwater species over the three sampling seasons in the two study regions. Only species known to occur in freshwater marshes (listed as FW in Table 1) were included in these analyses. Shown are means ± 1 SE. [larger image]

Figure 9. Two-dimensional non-metric MDS ordinations illustrating variation in small fish community structure (< 5 cm SL): (A) among seasons and sites and (B) as a function of trap placement. MDS plots are based on Bray-Curtis similarities of standardized minnow traps CPUE. [larger image]

In minnow traps, mean CPUE was similar between the top and bottom trap, but placement affected the magnitude of CPUE variation across seasons and between regions (Table 2). For instance, CPUE in RB doubled between the wet and transition samples in the top trap, but increased by 50 times in the bottom trap (P = 0.0001 for both cases, Fig. 5). Large numbers of sunfishes, bluefin killifish, and coastal shiners accounted for this increase in the bottom trap (Table 1). In NW headwaters, seasonal variation in CPUE was detected only in the top trap. In spite of very low catches, CPUE increased between the transition period and the dry season (P = 0.0001; Fig. 5A). Dissimilarity between minnow trap samples as a function of trap placement was lower than the separation observed when comparing gill nets and electrofishing samples (Table 3, Fig. 9B). The contribution of freshwater species was higher in the top than in the bottom trap (80.6% and 54.2%, respectively; Table 4). Eastern mosquitofish and least killifish were more abundant in traps placed at the top of the water column, whereas bluefin killifish, rainwater killifish, clown gobies, and dollar sunfish were more abundant in traps placed at the bottom of the water column.