Discussion
The serial discontinuity concept describes how riverine systems are expected to respond to major disruptions along the course of a river (Ward and Stanford, 1989; Ward & Stanford, 1995; Stanford & Ward, 2001). The expectations vary depending on the geomorphology of the river and the responses are described for three different reach types – constrained, braided, and meandering (Ward & Stanford, 1995). The portions of the Tallapoosa River sampled in our study fit between the constrained and braided reach types, characterized by a stable channel, a narrow riparian corridor, and lotic habitats throughout. As such, the biodiversity of the Tallapoosa was expected to increase with distance downstream (Ward & Stanford, 1995). Interestingly, we measured no such increase in Shannon’s H or species/family richness within the river section from Harris Dam to the headwaters of Lake Martin. However, we documented variation in the assemblage structure that is likely influenced by the presence and operation of Harris Dam.
Sunfishes and minnows were generally the most common fish families in this part of the Tallapoosa River, and variation in diversity from upstream to downstream was neither large nor systematic. Catostomids, centrarchids, and cyprinids were dominant in catches above Harris Dam, similar to the findings of Travnichek and Maceina (1994) who conducted a survey (prior to the implementation of the water release management Green Plan in 2005) of the broader Tallapoosa River from the upper reaches near Heflin, AL downstream to the Coastal Plain. Overall species diversity index values for this study area were slightly higher and more variable in 1994 (1.98 - 3.53) versus our study (2.49 - 2.88), though this difference may have been driven in part by differences in sampling techniques (prepositioned electrofishing grids in 1994 versus boat electrofishing in our study). Trends in fish diversity upstream to downstream were similar between our findings and those of Travnichek and Maceina (1994), who found some evidence that river regulation diminished the number of obligate fluvial specialist species in the Tallapoosa River. It is important to note that centrarchids were not historically dominant in the Tallapoosa River (Irwin & Hornsby, 1997). Considering that catch rates of centrarchids in both our study and in Travnichek and Maceina (1994) were high downstream of Harris Dam, this supports the idea that generalist species (such as many centrarchids) may be less affected by river regulation (Scott, 1951; Swingle, 1953; Kinsolving & Bain, 1993; Travnichek & Maceina, 1994), and that this trend had not been altered by implementation of the more recent “Green Plan” flow modification. While Travnichek and Maceina (1994) observed an increase in species richness from upstream to downstream across a much broader study area, their results from within the bounds of this study area did not yield such a trend, again consistent with our findings.
Irwin and Hornsby (1997) compared rotenone surveys conducted at Horseshoe Bend in 1951 (pre-Harris Dam) versus 1996 (post-Harris Dam) to assess the effects of river regulation due to Harris Dam on downstream fish assemblages. Differences in species composition between these rotenone studies also suggested that the pre- versus post-dam fish assemblage at Horseshoe Bend may have shifted from one dominated by cyprinids and ictalurids to one dominated by centrarchids (Irwin & Hornsby, 1997). Our results showed a larger relative proportion (by numbers) of centrarchids versus the 1951 rotenone sample (0.01 in 1951, 0.46 in our study), but a similar proportion to the 1996 sample (0.51 in 1996). In addition, the proportion of cyprinids and catostomids in our sample was higher than in the 1996 rotenone sample (0.11 in 1996, 0.43 in our study), but similar to the 1951 findings (0.50 in 1951; Irwin & Hornsby, 1997). Differences in sampling method (electrofishing versus rotenone), sampling frequency (bimonthly here versus a single sample historically), and seasonality likely limits direct comparisons. However, the continued prevalence of centrarchids in the Tallapoosa River downstream of Harris Dam in our study, Travnichek and Maceina (1994), and Irwin and Hornsby (1997) suggests that Harris Dam has affected and continues to affect the downstream fish assemblage.
The presence of significant differences in fish assemblage composition across sites in ordinal space and the formation of upstream versus downstream site blocks on the Tallapoosa River suggests that discharge regulation or habitat modification by Harris Dam was responsible for much of the variation in fish assemblage structure. Although riverine fish assemblages naturally vary longitudinally, regulation of the Tallapoosa River by Harris Dam changes the quality of habitat by altering temperatures and flow regimes, and by fragmenting the river, changing natural patterns of fish movement, persistence, and colonization (Vannote et al., 1980; Irwin & Freeman, 2002; Kiraly et al., 2014, Irwin et al., 2019). Our study describes patterns that are influenced by Harris Dam and the resulting flow regulation. The first multidimensional scaling axis (MDS-1) correlated strongly with the CPE of ictalurids, percids, and centrarchids, and the tailrace had the highest scores of any site for MDS-1. The species within those families that largely influenced this trend tended to be smaller bodied ictalurids and darters – species that tend to be more rheophilic. This suggests that the tailrace is primarily occupied by species that are able to persist in the variable flow conditions, potentially at the expense of other species including minnows and suckers, that might be less well-suited for conditions in the tailrace. The number of indicator species at a site can be considered a measure of how unique that site is relative to the other sites considered. The presence of more indicator species at the upstream site supports the conclusion that the Lee’s Bridge fish assemblage was distinct from the sites downstream of the dam.
The MDS analysis indicated both spatial and temporal variation in fish assemblage structure throughout the mainstem Tallapoosa River. Several previous studies did not include seasonal variation when quantifying the Tallapoosa fish assemblage (Travnichek & Maceina, 1994; Irwin et al., 2019). Our results indicate that conclusions drawn from Travnichek and Maceina (1994) and Irwin et al. (2019) need to be limited to the time scales they encompassed. Our inability to sample at Lee’s Bridge in winter unfortunately hampered even broader generalization given that a model of complete annual variation in fish assemblage might identify additional patterns.
It is difficult to determine exactly how much Harris Dam has affected the fish assemblage in the Tallapoosa River given the paucity of data predating its construction. Longitudinal variation in fish assemblage structure is inevitable in river systems as energy and nutrient inputs change (Vannote et al., 1980). Overinterpreting the single rotenone sample taken before Harris Dam began operation is also questionable given the habitat heterogeneity of the Tallapoosa River and the limited spatial and temporal coverage of rotenone sampling (Swingle, 1953). Outside of this study area, literature indicates that flow regulation, and especially peaking flow, negatively impacts fish recruitment and spawning (Weyers et al., 2003; Rolls et al., 2013). The persistent regulation of the Tallapoosa River by Harris Dam likely resulted in a changes in the fish assemblage driven by the inability of certain species to adapt their spawning and feeding habits to the rapid temperature and flow fluctuation (Rolls et al., 2013).
Releases of water from dams can strongly affect habitat conditions for fish and other aquatic organisms (Freeman et al., 2005; Young et al., 2011). Impacts that affect fish at the individual scale can also be manifested at the population and assemblage scales. Our sampling spanned a longitudinal gradient that included a site above Harris Dam and three sites at increasing distances downstream of the dam, allowing us to examine whether patterns in fish communities are consistent with expected effects of the dam, namely a recovery gradient in the diversity or assemblage composition (Travnichek & Maceina, 1994; Ward & Stanford, 1995; Ward & Stanford, 2001), as well as whether the implementation of the modified flow regime has had any effects on the downstream fish community. Previous studies (see below) have quantified assemblage structure and responses of particular fish populations across this same reach, allowing comparisons that span a range of temporal scales.
Across the sampled sites downstream of Harris Dam, we expected to observe an increase in biodiversity as hypothesized under the serial discontinuity concept (Ward & Stanford, 1995). Interestingly, while we did see a recovery gradient in multivariate space downstream of the dam (where site assemblages became increasingly similar to the upstream, unregulated site), there was no such gradient in species richness or Shannon’s H. These species diversity results are similar to those from other previous studies of the Tallapoosa River fishes. Clearly the use of simple diversity indices did not capture the changes in fish assemblage structure that we observed downstream of the dam. Quantifying more than just simple diversity indices will be required to capture the recovery gradient in response to disturbance.