Bait efficiency to monitor mandi ( Pimelodus maculatus ) in the tailrace of hydropower dams, Southeast Brazil

mandi 2021, 33, e8. Abstract: This study compared the efficiency of two baits (earthworm and cow heart) for monitoring the abundance of mandi catfish ( Pimelodus maculatus ) in the tailrace of hydropower dams, one of the commonly killed fish in turbine procedures. Sampling occurred every two months from February 2014 to December 2015 at the Camargos and Itutinga hydropower dams, located in the upper Paraná River basin, using hooks baited with earthworm and cow heart. We used generalized linear models to test the influence of dam, bait type, dam outflow discharge, water temperature, transparency, and dissolved oxygen on mandi abundance, calculated as catch per unit effort. Most mandi (73%) were caught using earthworms but catch per unit effort of mandi for both baits showed the same pattern of seasonal variation at each dam. At both dams, individuals of mandi sampled using earthworms were smaller than those sampled using cow heart, and water temperature was the only environmental variable significantly associated with the catch per unit effort of mandi. Both baits are efficient to monitor seasonal variations in mandi abundance, which is influenced by water temperature. Monitoring the abundance of mandi in the tailrace can help minimizing fish mortality by scheduling risky turbine procedures for periods of

hydropower dam of the Grande River; Itutinga (21°17′30′′S, 44°37′31′′W) is located 7.5 km downstream from Camargos. Both dams have no fishway. The tailrace of Camargos is 51 m wide and 8 m deep, while Itutinga is 52 m wide and 8 m deep. The mean and the daily outflow discharge interval (m 3 .s -1 ) during the study period was 51 and 24-182 in Camargos, We sampled fish at both dams every two months from February 2014 to December 2015. No sampling occurred at Itutinga hydropower dam in August 2014 as turbine procedures precluded the use of baited hooks. Sampling consisted of throwing the baited hook from the powerhouse into the tailrace near the exit of the draft tubes. We used 1.68 m long fiberglass rods, reels with 100 m of 0.35 mm diameter line, 35 g of lead, and size 2 hooks (270 mm long, 11 mm gape, 1 mm diameter), manufactured with high-carbon steel. We used the hook size 2 because it captured mandi of similar mean weight (118 g) of those collected in the turbine during the dewatering procedures (101 g; personal observation of the authors). When baiting with earthworms, we ensured that the bait covered the entire length of the hook. We baited hooks with strips or cubes of cow heart about the same width as the hook gape (11 mm). We used commercially raised earthworms (Annelida: Glossoscolecidae) purchased in a local market. Earthworms were approximately 54 x 3 mm, while cow heart bait cubes were 27 x 11 x 3 mm.
The fishing effort consisted of two experienced fishers, each fishing at the same time for periods of 4 h between 7:30 and 17:00 h. The catch per unit effort (CPUE) was calculated as number of mandi CPUE number of fishers x hour = . Before fishing, we randomly selected one fisher to sample with earthworm while the other used cow heart. After every hour of sampling and until the end of the 4-h sampling period, the fishers alternated the type of bait they used. We ensured that the fishers were spaced 5-15 m from each other and that the baited hooks were lying on the tailrace bottom, 3-10 m Mortality of fish in the tailrace is frequent in Brazilian hydropower dams during turbine procedures, such as start-up or dewatering, as fish in the tailrace may come into the turbine area (Rêgo et al., 2017). This mortality can range from less than a kilogram to up to tons of fish in a single turbine procedure (Andrade et al., 2012). A variety of fish species may be killed during such turbine procedures (Rêgo et al., 2017), but the mandi catfish (Pimelodus maculatus Lacepède, 1803) is by far the species most frequently killed in some basins (Andrade et al., 2012;Rêgo et al., 2017).
The abundance of mandi in the tailrace is a predictor for the number of mandi at risk of death during turbine procedure (Andrade et al., 2012). Thus, monitoring the abundance of mandi in the tailrace can help minimizing fish mortality (Loures et al., 2017) by scheduling turbine procedures for periods of lower mandi abundance. Therefore, sound understanding of the seasonal variation in mandi abundance in the tailrace is necessary for this strategy to be successful.
Sampling with gillnets is one of the most common techniques used for determining fish abundance (Oliveira et al., 2014). Gillnets can be used in many tailraces, but not in those with high turbulent flow, where baited hook may represent the only option for sampling fish (Loures et al., 2017). Sampling fish with baited hook also results in lower bycatch than gillnetting because baited hooks are more species selective than gillnets, which is important for studies that require abundance data on target species (Elliott & Beamesderfer, 1992). In this way, our objectives were to (i) evaluate differences in the number and length of sampled mandi using earthworm and cow hear -baits most commonly used by local fishermen (personal observation of the authors), (ii) compare seasonal variation in catches, and (iii) determine the abiotic factors that influence the number of sampled mandi.
from the tailrace edge, at a depth of approximately 8 m at Camargos and 6 m at Itutinga.
We recorded the standard length (SL) of every mandi captured, and released them at the capture site. Sampling had the authorization from the Ministry of the Environment, Brazil's federal environmental agency (ICMBio license #10306).
Between 9:00-16:00, we measured water temperature and dissolved oxygen (DO) using an YSI 550A probe, and water transparency using a Secchi disk. We obtained data on the dam outflow discharge from the Telemetry and Hydrometeorological Monitoring System of the hydropower dam concessionaire, the Power Company of Minas Gerais (Companhia Energética de Minas Gerais; CEMIG).
To avoid pseudoreplication, we used autocorrelation analysis to verify the temporal independence of samples for all variables used in statistical tests. We performed this analysis using the autocorrelation function (ACF) available in the R software. We used a generalized linear model (GLM) with a negative binomial distribution to determine the influence of dam, bait type, dam outflow discharge, water temperature, DO, and water transparency on the CPUE of sampled mandi. We assessed overdispersion using the ratio between the residual deviance and the degrees of freedom. We considered overdispersion appropriate when the ratio was between 0-1. In the final model, we included only significant variables. We used the Mann-Whitney test to evaluate differences in mandi SL between the two bait types for each dam. We performed all analysis in the software RStudio (R Development Core Team, 2019). We adopted the level of significance (α) of 0.05.
Mandi All autocorrelations were lower than the 95% confidence interval and, therefore, not significant. Considering that even among independent samples 1 in 20 autocorrelations might be significant (Kendall et al., 1983), we concluded that all variables we used came from independent samples (Zuur et al., 2010).
At both dams, we captured more mandi using earthworms (Camargos = 0.58; Itutinga = 1.2 mandi per fisher per hour) than using cow heart (Camargos = 0.13; Itutinga = 0.67 mandi per fisher per hour) (GLM, p = 0.04). Water temperature was the only limnological variable that explained significant variation in the number of mandi sampled through time (GLM, p < 0.01). There was no significant interaction among any pair of variables analyzed by the GLM, including the interaction between dam and water temperature. No interaction between dam and water temperature indicated that temperature had similar effects on the number of mandi sampled through time at both dams, which is also supported by graphical analysis (Figure 1).
Considering both dams, CPUE per sample ranged from 0.00 to 6.12, averaging 2.44 mandi per fisher per hour. The CPUE of sampled mandi using earthworms and cow heart showed the same pattern of seasonal variation at both dams, with higher CPUEs from December to April (Figure 1). Thus, both baits seem to indicate a period of the year when mandi abundance in the tailrace is greatest, and therefore, when turbine procedures are riskier to fish. It is well known that the behavior of Neotropical riverine fishes is influenced by environmental variables, like temperature, which affects several aspects of their biology (e.g., feeding, migration, reproduction, interactions, and life history) and abundance (Lowe-McConnell, 1987). High CPUE values occurred during warm months from November to February (Loures & Pompeu, 2012), coinciding with the spawning season of mandi in the upper Paraná River basin (Agostinho et al., 2003). Mandi is usually classified as a migratory fish (Agostinho et al., 2003), but this behavior is controversial and some authors consider mandi as sedentary (Oldani et al., 2007). Peressin et al. (2017) verified that 95% of the mandi females from the tailrace of 12 dams of the upper Paraná River basin, including Camargos and Itutinga, had undeveloped gonads. This suggests that mandi at the tailrace of dams are not migrating for spawning and could be there for other purposes, like dispersal or feeding (Godinho & Kynard, 2009).
The SL of the mandi captured using earthworms (range: 7.1-29.8 cm; mean: 14.9 cm) was significantly smaller than the SL of the mandi sampled using cow heart (range: 11.5-33.0; mean: 19.4 cm) at both dams (Mann-Whitney test; p < 0.01; Figure 2). Thus, combining both baits provides samples across a broader size range. The volume (size and shape) of bait types is also known to affect the size of fish captured with hooks (Hayes et al., 2012). We could not cut cow hearts into pieces of the same size and shape as earthworms for every sampling campaign because the size of earthworms varied slightly by season. Therefore, differences in the SL of mandi caught using the two different bait types could also be due to the differences in size and shape between the two bait types.
In conclusion, baited hooks with earthworms and cow heart can be successfully applied in tailraces with highly turbulent flow to monitor variation in mandi abundance. Both baits should be used to cover a broader range of sizes. Hooks baited with either type of bait can be used to monitor seasonal variation of mandi abundance in the tailrace of hydropower dams. This information is helpful to reduce fish mortality by scheduling turbine procedures to periods when mandi abundance is lower. Figure 2. Percentages of mandi (Pimelodus maculatus) by classes of standard length (SL) sampled with earthworms or cow heart in the tailrace of the Camargos and Itutinga hydropower dams.