Reproduction of two loricariid species in a confined river and implications for environmental impacts of dams

1Departamento de Ecologia, Universidade Federal do Rio Grande do Sul (UFRGS). Av. Bento Gonçalves, 9500, Setor 4, Prédio 43422, CP 15007, 91501-970 Porto Alegre, RS, Brazil. (LDF) lucas.defries@yahoo.com.br, http://orcid.org/0000-0001-8361-2608 (corresponding author), (GR) gilberto.rosa@ufrgs.br, (FGB) fgbecker@ufrgs.br 2JPSBIO Consultoria Ambiental, Estrada Francisca de Oliveira Vieira, 796, 91780-050 Porto Alegre, RS, Brazil. josefpezzi@gmail.com 3Simbiota Consultoria Ambiental, Rua Emiliano de Macedo, 1425, 95800-000 Venâncio Aires, RS, Brazil. fabio@simbiota.com.br Introduction


Reproduction of loricariids and the impact of dams
In confined rivers, the advantage of producing offspring in time to benefit from the expanded and productive flooded habitats (Winemiller et al., 2008) is less evident because flooded habitats are much smaller in area and less persistent compared to large floodplain rivers.In confined rivers, growth and refuge habitats for fish offspring are probably located in pools or at tributary confluences (Reynalte-Tataje et al., 2008, 2012;Hermes-Silva et al., 2009;Ávila-Simas et al., 2014).Thus, it is plausible to expect that in confined rivers relations between abiotic factors and fish reproduction will differ from those observed in floodplain rivers, particularly in subtropical latitudes where temperature and daylight are seasonally variable.
Knowledge about reproductive biology of fishes inhabiting confined rivers is particularly important in Brazil, because the physical characteristics of confined channels are usually favorable for damming and hydroelectric power generation (Reynalte-Tataje et al., 2012).In fact, there are more than 700 large dams (>15 meters high) in large rivers of Brazil (Agostinho et al., 2008).This continuous and widespread process of environmental change threatens the spawning of habitats and recruitment of many fish species.Correctly predicting impacts on fish species, and eventually designing adequate conservation plans and management actions, depends on basic life-history information (Agostinho et al., 2007(Agostinho et al., , 2008)).
In this study, we have two objectives.The first one is to provide a description of fundamental reproductive characteristics of two armored catfish species (Hemiancistrus fuliginosus Cardoso, Malabarba, 1999 andHypostomus isbrueckeri Reis, Weber, Malabarba, 1990; Loricariidae) in a confined river just before the closing of a hydroelectric reservoir.The second is to investigate how abiotic factors (flow, temperature, day-length and rainfall) relate to reproduction in a subtropical confined river, under the hypothesis that these relations should be different from those known for floodplain river species.

Material and Methods
This study was carried out in the Passo Fundo River and its main tributary, the Erechim river, in Southern Brazil.The Passo Fundo River is a major tributary of the Uruguay river, which joins the Paraná river to form the La Plata estuary.It is a confined river with no clear floodplain, flowing along a basaltic plateau, in elevations from ca 750 m to 250 m above sea level (drainage area = 4077 km 2 ; length = 247 km).The upper course of the Passo Fundo River was dammed for hydropower production in 1971 upstream from the confluence with the Erechim River (UHE Passo Fundo hydropower plant; 27°33'11''S, 52°44'29''W) and in 2009 another dam was built downstream of the confluence with the Erechim River (UHE Monjolinho; 27°19'43''S, 52°44'05''W) (Fig. 1).River discharge is seasonally variable and predictable, with two typical high flow periods in the late fall and in Fig. 1.Geographical situation of the study area and sampling points.Note that fish samples for this study were taken before river damming by the UHE Monjolinho.spring, and a lower flow in summer.Climate in the region is humid temperate (Cfa, Köppen classification), with mean annual temperature of 20 ºC and annual rainfall of 2000 mm.Summer temperature (January-March) can reach 38 ºC and negative temperatures may occur in winter.In contrast to tropical parts of South America, there is no clear dry season, although rainfall is higher in fall and spring (Fig. 2).We sampled Hem.fuliginosus and Hyp.isbrueckeri approximately every two months, with eight samples encompassing a 12 month period, between February 2008 and January 2009.We placed gillnets in two sites (P1 and P2) at the Erechim River and in three sites (P3, P4 and P5) at the Passo Fundo River (Fig. 1).Erechim River is the main tributary of the Passo Fundo River, and the five sampling points encompassed a 168 km-long stretch of free-flowing confined river upstream from the confluence with Uruguay river.
Our samples were taken before UHE Monjolinho was built, in May 2009.In each sampling site we used one 100-m gillnet set (mesh-sizes: 12 mm, 20 mm, 30 mm, 40 mm e 50 mm, bar measure between knots), which was set in mid-afternoon and collected in the next morning (ca.12 hours in water).The sampled sites can be described as channels with mainly rocky substrate and occasional sandy patches; water flow is heterogeneous (fast flowing waters and low flowing waters along the same stretch) and riparian vegetation integrity is relatively good.All collected specimens were fixed in 10% formalin.Voucher specimens for Hem.fuliginosus and Hyp.isbrueckeri were deposited in the Department of Zoology of Universidade Federal do Rio Grande do Sul (UFRGS 11396 and UFRGS 22553, respectively).
From each specimen, we measured total length (to the nearest mm) and total weight (g).Gonads were removed, weighted (0.01 g) and macroscopically inspected for maturity stage.We initially applied a classification with five gonadal development stages (immature, maturing, mature, spawning, spent), but macroscopical distinction of maturing, spawning and spent ovaries was difficult because specimens were initially stored in formalin.We then reallocated gonads to only three categories: immature, mature, and not mature.The mature category included females in early spawning stages (i.e,macroscopical characteristics of both mature and spawning gonads).Oocytes from mature ovaries were counted and measured under stereomicroscope, and these data were used to describe fecundity and oocyte development type.
Gonadosomatic index values (GSI) was calculated as GSI = (W g.W t -1 )*100, where W g is the gonad weight and W t is the total weight (Vazzoler, 1996).Condition factor (K) (Le Cren, 1951) , where W c = W t -W g and b is the coefficient of the length-weight relation (W t = a.L t b ).We analyzed the monthly frequency of mature females and mean gonadosomatic index values for determining the reproductive season.Monthly differences in mean K and GSI were tested by one-way non-parametrical ANO-VA (Kruskall-Wallis test).We pooled individual data from all five sampling sites to calculate the GSI mean, K mean, and frequency of mature females in a given month.
The reproductive strategy of the studied species was assessed by examining total and relative fecundity, GSI and size of oocytes.Fecundity was defined as the number of oocytes in mature ovaries (n = 28, Hem.fuliginosus; n = 12, Hyp.isbrueckeri).Relative fecundity was expressed as the number of oocytes.Wt -1 (after testing for linear relationship between number of oocytes and W t ).The type of oocyte development was determined by inspecting the pattern of oocyte size frequency distribution (Vazzoler, 1996).
Size at first maturity (L 50 ) was defined as the total length where 50% of all individuals are predicted to be adults (Vazzoler, 1996).For each species, L 50 was estimated by logistic regression (Roa et al., 1999), with total length as independent variable and life stage (either adult or non-adult) as dependent variable.Individuals were classified as either adult or non-adult according to the gonadal maturity stage (non-adult = immature gonads; adults = all other gonadal stages).
To investigate the relation between river flow and reproductive timing, we obtained historical data (1969 to 2006) on river flow (discharge, m 3 /s) from the fluviometric station "Ponte do rio Passo Fundo" (Agência Nacional de Águas;  .For analysis purposes, all variables were averaged over the period one month (mean river flow in month m is the average of daily flow values in that month).
The relation of reproductive period and condition factor with seasonality of abiotic factors was examined by calculating Pearson correlations between mean monthly values of GSI and K (average of individual values over each sampled month) and river flow (2008-9), historical river flow , rainfall, day length, and mean monthly air temperature.Because reproductive events may be time lagged (e.g., Torres-Mejia, Ramírez-Pinilla, 2008), correlations were also calculated between GSI at month m and values of abiotic factors one month (m-1) or two months (m-2) before.This procedure resulted in 11 pairwise correlation tests for female GSI, therefore we corrected for multiplicity of tests using the false discovery rate (FDR), i.e, we corrected p-values to control the proportion of false positives (Benjamini, Hochberg, 1995;Verhoeven et al., 2005;Waite, Campbell, 2006).We also tested the correlation of fish condition with reproduction and abiotic factors.

Results
A total of 216 Hem.fuliginosus and 312 Hyp.isbrueckeri were captured.The length frequency distribution in the samples suggests that males of both species tend to reach larger sizes than females (Figs.3a, b).Sex ratio was not significantly different from 1:1 (Figs.3c, d).There was no significant difference in the length-weight relation for males and females (Tab.1).
Reproductive characteristics.Mature ovaries of Hem.fuliginosus contained a lower number of oocytes in comparison to Hyp. isbrueckeri, but oocyte diameter was similar for both species (Tab.2).Relative fecundity was higher in Hyp.isbrueckeri.The number of oocytes presented positive linear relations to total length (Fig. 4), as well as to gonad weight (n oocytes = 13.79Wg+ 21.88; p < 0.01; r² = 0.80 to Hem. fuliginosus and n oocytes = 7.03Wg + 96.38; p < 0.01;   fuliginosus probably spawning between January and February and Hyp.isbrueckeri between December and January.Immature females were found all year round, except in late fall (June; Fig. 6).Apparently, there were no mature males     of either species in our samples, but this could have been a bias resulting from our procedure for maturity stage classification (only macroscopical inspection).Length at first maturity was 96.5 mm (females) and 125 mm (males) for Hem.fuliginosus, and 141 mm (females) and 140.7 mm (males) for Hyp.isbrueckeri (Fig. 8).
Reproduction and Abiotic factors.The GSI of both species was significantly correlated only to time-lagged river flow (Tab.3).For Hem. fuliginosus females, mean GSI was correlated with river flow after a two-month delay, while for Hyp.isbrueckeri females there was one-month delay (Tab.3, Fig. 9).No other environmental variable was correlated to GSI variation during the study period, although the increase in female GSI generally coincided with increasing temperature and daylight hours.Therefore, for both species, the peak gonadal maturity and the subsequent spawning (steep decrease in GSI, Fig. 7) seem to have occurred during low river flow period in summer, after the spring high flow period (which is a historically consistent flow pattern, Fig. 9).Condition factor varied significantly along the sampling period (except for male Hyp.isbrueckeri) (Fig. 10), but it was not related to the gonadal cycle or to abiotic variables (Pearson correlations were not significant after correction for FDR).
The reproductive period of Hem.fuliginosus (October to February) and Hyp.isbrueckeri (October to January) is similar to that of other loricariids from tropical latitudes, although armored catfishes encompass a wide diversity of forms and behaviors (Burgess, 1989;Agostinho et al., 1995;Suzuki et al., 2000).Present knowledge about loricariid life-history indicates that reproduction takes place in spring and summer (Agostinho et al.,1987(Agostinho et al., , 1991;;Mazzoni, Caramaschi, 1995, 1997a, 1997b;Ramos, Konrad, 1998, 1999;Hirschmann et al., 2011;Gomes et al., 2015), as is also the case for many other Neotropical groups (Lampert et al., 2004;Dala-Corte, Azevedo, 2010).However, in contrast to species from tropical latitudes and floodplain rivers, we observed that GSI values of Hem.fuliginosus and Hyp.isbrueckeri females were strongly correlated to flow in the months preceding spawning (r ≥ 0.87; two-month lag for Hem.fuliginosus and one-month lag for Hyp.isbrueckeri); and a weaker positive relation with daylight and temperature (0.5 < r < 0.7).Other studies in upper Uruguay River suggest that temperature increase is the trigger to fish reproductive activity (Hermes-Silva et al., 2009;Correa et al., 2011;Lopes et al., 2014), but we observed that GSI variation was highly related to two-month previous river flow, while temperature and daylight were only mildly related to GSI variation.Fish eggs and larvae studies have revealed that fish reproduction is highly seasonal in the Upper Uruguay River, being related to temperature and water flow (Correa et al., 2011).Flow regime has a strong relationship with fish life history (Te-9 e170163[9] desco, Hugheny, 2006;Olden et al., 2006;Tedesco et al., 2008), especially with predictability and variability of river flow (Olden, Kennard, 2010).We then suggest that temperature and day-light are probably related to the acceleration of gonadal development, but spawning (as measured by steep decrease in GSI) is probably related to seasonally predictable decreases in river flow.
We also found that peak reproduction of Hem.fuliginosus and Hyp.isbrueckeri was delayed in relation to river flow, occurring in summer, just after the decrease of peak river flow, differing from the reproductive pattern of floodplain species, which tend to reproduce during rising waters, preceding flood peaks (Agostinho et al., 1991;Paugy, 2002;Agostinho et al., 2004;Suzuki et al., 2004;Rutaisire, Booth, 2005;Andrade, Braga, 2005;Bailly et al., 2008).In confined rivers, discharge increase may result in unfavorable conditions for reproduction (for example, higher mortality or lower recruitment due to physical removal of eggs, larvae and spawning substrate), thus, reproducing during low flow would represent a more favorable condition (Pusey et al., 2001).Humphries et al. (1999) proposed that in rivers with relatively unpredictable flood dynamics or where flood does not coincide with adequate temperatures, recruitment should occur during low-flow (Low Flow Recruitment hypothesis, LFR).According to Humphries et al. (1999), periods of low flow during summer are more predictable from year to year.In our study, the historic flow 1969-2006 and flow 2008-9 showed a predictable low flow pattern during summer in the Passo Fundo River.In situations where LFR is plausible, fish reproduction should coincide with predictable low flows, when temperatures are warm and food resources are more concentrated (Lytle, Poff, 2004;Zeug, Winemiller, 2007).
Additionally, we speculate that suitable physical conditions for oviposition (water depth and velocity) possibly become more available during the recession limb of the hydrograph.Such coincidence between life-history (reproductive) and low river flow in summer may result in higher fitness (Lytle, Poff, 2004).In the upper Uruguay River, ichthyoplankton studies have shown that pool habitat formation during summer -in low flow and high temperature conditions -is relevant for development of loricariid species (Hypostomus spp., Loricariichthys spp.and Rineloricaria spp.), because pool habitats provide both favorable physical habitat conditions and concentration of food resources for larvae growth (Ávila-Simas et al., 2014;Lopes et al., 2014).According to Correa et al. (2011), the particular topography of the upper Uruguay River favors spawning in the rapids, and pools are used as nursery areas for larvae.We thus suggest that, in confined rivers, reproduction of fish -both migratory and non-migratory -has a distinct relation with abiotic factors in comparison to reproduction in floodplain rivers.In confined rivers, because of the absence of floodplain habitats (floodplain lakes, back waters and wetlands), fish reproduction must take place within the river channel, so that mortality of early-life stages is much more dependent on flow variation.In this sense, predictability of flow (particularly of low-flow) and its coincidence with historical river flow patterns, would be key factors for reproductive success and population persistence in the long term (Lytle, Poff, 2004).Under these circumstances, species tending to equilibrium life-history strategies, with reproduction synchronized with low river flow, might be relatively more common than in other environments.Both these later ideas should be tested in the future, by assessing reproduction of other species in a wider taxonomic scope (families, orders) and in comparison to other environments.
Our results have implications for assessing and mitigating the impacts of river damming on fish populations in confined rivers.Because river damming and flow management affect the predictability of low flow in confined rivers, fish reproduction could be severely affected, particularly for species that evolved to cope with natural flow regimes (seen Lytle, Poff, 2004).The fish species studied here presented an equilibrium life history strategy (Winemiller, 1989;Winemiller, Rose, 1992) correlated to a historical pattern of low river flow in summer.Equilibrium strategists are expected to be positively related with predictability of water flow and to be favored in more stable environments (Mims, Olden, 2012, 2013;Gomes et al., 2015).The formation of the UHE Monjolinho reservoir could affect the spawning of Hem.fuliginosus and Hyp.isbrueckeri, which are cumulatively affected by flow regulation by the upstream Passo Fundo Dam, resulting in the disappearance of favorable environments for recruitment of loricariids species and of the hydrological regime associated to the reproductive cycle.Drastic reduction in loricariids after reservoir formation has been previously reported (see Agostinho et al. (1999) for Hypostomus sp. and Hirschmann et al. (2008) for Hemiancistrus punctula-Tab.3. Pearson correlations (r) between monthly mean of gonadosomatic index (GSI) and monthly values (n = 8) of environmental variables in the Passo Fundo river (Brazil).Significance of the correlations was determined using the false discovery rate (FDR) procedure.Significant correlations (p ≤ 0.05) are in bold and underlined.Lagged variables are indicated by i-1 and i-2 for one and two-month shifts respectively (see text).tus).Our results thus add support to the need for managing river flow in accordance to the historical patterns under which fish populations evolved.This means not only that seasonal flow patterns should be taken into account in reservoir operation, but also that short-term (few hours) extreme variations in flow should be avoided.Short-term extreme variation in discharge is common in dammed confined rivers as a result of high demand for power generation.Ecologically oriented management of low flow during summer could be a useful strategy for conservation of fish species in dammed confined rivers.Clearly, information about fish life-history and its relation to abiotic factors should be obtained before dam construction and fish monitoring programs demanded by environmental agencies should objectively assess how fish reproduction is affected.Flow management should than be regulated in a multiobjective manner that includes not only power generation, but also fish life-cycles, according to the Natural Flow Paradigm (Poff et al., 1997;Lytlle, Poff, 2004).A complementary conservation strategy would be the preservation of undammed or, at least, unregulated confined free-flowing rivers or river segments.

Fig. 2 .
Fig. 2. Historical and observed variation of abiotic factors in the Passo Fundo river basin.a. river flow; b. rainfall; c. air temperature and daylight hours.
drainage area = 3709 km 2 ).Flow data for the study period (February 2008 to January 2009) was obtained from the flow monitoring program at the UHE Monjolinho (drainage area 3770 km 2 ).Temperature and rainfall data for the nearest meteorological station (Iraí) were provided by the National Meteorological Institute (INMET -8° Distrito), including historical rainfall data

Fig. 8 .
Fig. 8. Logistic regressions for estimating size at first maturity Hemiancistrus fuliginosus and Hypostomus isbrueckeri in the Passo Fundo and Erechim rivers, southern Brazil.
Reproduction of loricariids and the impact of dams