1 Introduction

The Brazilian rivers have been sorely segmented by dams. Fish stocking is one of the management actions used to minimize the impacts of dams on native fish fauna (^{Agostinho et al., 2008}). It consists of the breeding and cultivation of fish in captivity for later release into natural water bodies. It is normally applied when the management goal is to supplement the native populations (^{Agostinho et al., 2007}). However, despite being historically applied, the efficiency of stocking, as a management measure, has been not satisfactory in Brazil, since the target stocks have not been recovered (^{Agostinho et al., 2007}). This can be explained by poor habitat quality of the stocked rivers (^{Pompeu, et al., 2012}) and by differences between hatchery and wild environment.

Hatchery tanks are characterized by a high degree of environmental homogeneity (^{Pakkasmaa et al., 1998}). They are normally made of concrete, with no substrate, obstacles or shelter and have constant food availability, high population density and species monoculture. As a result, farmed fish differ from wild fish in several aspects, especially in the morphological ones, exhibiting distinct colorations (^{Maynard et al., 1995}); different body, head and fin proportions (^{Swain et al., 1991}; ^{Fleming et al., 1994}; ^{Hard et al., 2000}; ^{Solem et al., 2006}; ^{Belk et al., 2008}) and less morphological variability (^{Taylor, 1986}). Moreover, density-dependent processes probably influence early growth and survival (^{Winemiller and Dailey, 2002}). On the other hand, natural environments are marked by many environmental variations, whether physical, chemical or biological ones. Marginal lagoons inhabited by many neotropical migratory fish species in the early stages of development, are rich in food and shelter provided by aquatic plants, wood debris, and other structures (^{Agostinho and Júlio-Junior, 1999}), and have high species richness with diverse social interactions (^{Pompeu and Godinho, 2003}). As a result of these environmental differences, the survival rates of released fish are low, because they are adapted to the rearing environment and not to the wild (^{Maynard et al., 1995}). The selection of phenotypes adapted to the captivity is called domestication, which can affect the physiology, behavior and morphology of fish (^{Araki et al., 2008}).

Specifically in relation to morphology, polymorphisms between alternative environments may occur in a same species, in response to different selective pressures (^{Robinson and Wilson, 1996}; ^{Langerhans et al., 2003}). In this respect, a range of studies conducted, mainly in temperate countries, has demonstrated the existence of wide morphological differences between wild and farmed fish (^{Taylor, 1986} ; ^{Swain et al., 1991}; ^{Pakkasmaa et al., 1998}; ^{Hard et al., 2000}; ^{Ojanguren and Brana, 2003}; ^{Cramon-Taubadel et al., 2005}; ^{Berejikian and Tezak, 2005}; ^{Solem et al., 2006}; ^{Belk et al., 2008}).

Fisheries management in South America has been based on stocking, fisheries regulation and construction of fish passes (^{Pompeu et al., 2012}). However, although fish stocking have been performing for years, we lack basic studies evaluating survival rates of released fish, and the efficiency of this strategy in recovering populations (^{Vieira and Pompeu, 2001}). Until now, there are no studies comparing wild and hatchery fish in respect to morphological aspects. Based on this, the primary goal of this study was to verify if hatchery fish are morphologically different from wild individuals of the same species and size class.

2 Material and Methods

3 Results

The analyzed fish of wild and hatchery populations showed a similar standard length (t(1;131) = 0.1441; p = 0.7048), allowing comparison between the two groups.

The Principal Component Analysis (PCA) indicated that each population occupies completely distinct morphological spaces (see Figure 1). The first two PCA axes were the most important in explaining the total variation of the sample (PCA1 = 34.73% and PCA 2 = 12.08%), jointly explaining 46.81% of variance existing between the two populations and due to the ecomorphological attributes analyzed.

Figure 1 Projection of two populations of Prochilodus lineatus [H = hatchery; W = wild] in the first two axes of the Principal Component Analysis. 

The T test conducted between the scores of the first two PCA axes showed significant differences only for the first axis (PCA 1: p = 0.0001; MS = 453.7531; t = 640.2777; PCA 2: p = 0.9113; MS = 0.0181; t = 0.0124). Thus, the ecomorphological attributes that more contributed to the differentiation between populations in the first axis of the PCA, make up the main source of morphological differentiation among populations. The most important attributes in PCA 1 (considering contributions higher than 70%) were compression index (CI), relative body height (RH), relative area of the pelvic fin (RAPlF) and aspect ratio of the caudal fin (ARCdF).

The discriminant analysis was also significant between the two analyzed populations (Wilks’ Lambda: 0.10259; F(12,119) = 86.748 p = 0.0001). The ecomorphological attributes that contributed significantly to the differentiation between them were relative height (RH) (F (3,176) = 72.03539, p = 0.00000), compression index of the caudal peduncle (CICP) (F (3,176) = 17.33353, p = 0.00006), relative area of the pectoral fin (RAPtF) (F (3,176) = 17.22258, p = 0.00006), aspect ratio of the caudal fin (ARCdF) (F (3,176) = 6.37739, p = 0.01287) and relative length of the head (RLH) (F (3,176) = 4.76763, p = 0.03096) (see Table 4).

Considering the attributes individually, all the important ones to the differentiation in PCA, as well as in the DA analysis performed, were significantly different between wild and hatchery fish (Wilks’ Lambda: 0.10259; F (12,119) = 86.748; p = 0.0001). Two attributes, RH and ARCdF, were pointed out by the two analyses as the most important. These two attributes showed higher values for the wild population, indicating that the wild fish had a higher body (RH: KW(1; 132) = 97.71758, p = 0.0001) and larger caudal fin (ARCdF: t(1; 130) = 90.4828, p = 0.0001) (see Figure 2).

Figure 2 Most important ecomorphological attributes in morphological differentiation between populations of Prochilodus lineatus in PCA1 and DA concomitantly. 

The morphological variability of the fish was analyzed by comparing the distance from the centroid and closest neighbor (see Figure 3). The morphological variability of the wild population was higher than that of the hatchery population, because the wild fish presented higher dispersion in morphological space (ANOVA log(DC+1): F(1; 131) = 11.9207, p = 0.0007) and thus occupied a larger morphological hypervolume (Kruskal-wallis’ test CDN: KW(1; 133) = 18.7552; p = 0.0001).

Figure 3 Log of the distance from the centroid plus one [log (DC +1)] and the distance from the closest neighbor. The ANOVA performed for the distance from the centroid was significantly higher for the population of wild fish, as was the Kruskal-Wallis test conducted for the distance from the nearest neighbor. 

4 Discussion

The results show that the fish hatcheries creates a less diverse population and individuals morphologically different from its wild counterparties of the same size class. Selective pressures present in the fish farming tank are quite different from those existing in the natural environment, which may result in very different morphologies generated by phenotypic plasticity (^{Taylor, 1986}; ^{Swain et al., 1991}; ^{Fleming et al., 1994}; ^{Hard et al., 2000}; ^{Belk et al., 2008}). Therefore, in the present study, the morphological differences between wild and hatchery fish could have been generated by phenotypic plasticity caused by adaptation to the growth environment, or may have been caused by genetic differences between populations.

The absence of genetic data does not allow separating the effects of these two sources of variation. But as the hatchery population also has a wild origin, it is concluded that the morphological differences detected were mostly due to the environmental component. One evidence of this, it is that the differentiation between the two studied populations can be easily understood considering the environmental differences between the lagoons and hatchery tanks. For example, the relative body height (RH) was higher for wild fish. This attribute is directly related to the ability of the fish to develop vertical displacements (^{Gatz Junior, 1979}). Therefore, the higher RH of wild fish is consistent with the need to perform maneuvers to avoid obstacles, considering the structural richness of the lagoons. Wild fish also had higher aspect ratio of the caudal fin (ARCdF), attribute that is directly proportional to the amount of swimming that the fish does (^{Gatz Junior, 1979}). This can be an indication that wild fish swim more than the hatchery, probably because they have a larger and more complex environment to explore. Morphological differences between populations of wild and hatchery fish have been previously highlighted, and environmental factors have been suggested as the principal source of variation (^{Taylor, 1986}; ^{Fleming et al., 1994}; ^{Ellis et al., 1997}; ^{Cramon-Taubadel et al., 2005}; ^{Belk et al., 2008}). However, the role of intraspecific interactions in wild environments as a morphological differentiation mechanism cannot be excluded. During interspecific competition, for instance, individuals from one species suffer a reduction in fertility, growth or survival as a result of resource exploitation or interference by another species (^{Begon et al. 2007}).

In relation to the morphological variability, the explanation for the lower value found for the hatchery population may lie in the lower genetic variability of this population (^{Vuorinen, 1982}; ^{Agostinho et al., 2007}), which may consequently result in minor morphological variability. However, recent studies (^{Gomes et al., 2008}; ^{Chittenden et al., 2010}) have shown the presence of minor genetic differences when comparing wild specimens with hatchery, since currently genetic diversity of stocked fish has been expanded from the use of a high number of breeder individuals of wild origin. Thus, the low morphological variability found for the hatchery population likely is due to the high homogeneity of the tanks (^{Hynes et al., 1981}; ^{Taylor, 1986}), where environmental pressures are constant, due to low disturbance frequency. The intraspecific polymorphism is typically caused by different selective pressures (^{Langerhans et al., 2003}). Thus, in heterogeneous environments such as lagoons, subject to numerous pressures and disturbances, various morphological types can coexist, as each may be favored at different times, depending on the disorder or pressure acting on the population. On the other hand, in homogeneous environments, such as farming tanks, the constant selective pressure favors a basic type of morphology, so that individuals tend to be morphologically closer.

Although stocking is intended to replenish stocks that have experienced genetic and/or demographic losses, the can carry environmental risks, even when performed with native species. These potential impacts include dissemination of pathogens and parasites and deleterious effects related to the genetic quality of matrices and fingerlings (bottleneck effects, loss of genetic variability and fitness, domestication) (^{Agostinho et al., 2010}). The release of a less variable and morphologically distinct population does not represent itself an impact. However, these aspects must represent lower probability to survive, hindering the use of this strategy as a conservation tool.

In summary, our results demonstrate that hatchery populations of Prochilodus lineatus are morphologically different and less diverse when compared with wild ones. However, for the actual consequences for hatchery management it is necessary to carry out specific experiments and studies that aim to elucidate how much and for how long these morphological differences could affect performance and survival of fish released in the natural environment. Additionally, considering the lower morphological variability of hatchery populations and the concept of environmental heterogeneity, the enrichment of hatchery tanks can be proposed as a way of increasing the morphological variability of the produced fingerlings.