Introduction

Freshwater organisms are currently facing several threats, which include water pollution and removal, changes in land use, introduction of exotic species, and obstruction and flow regulation by dams and weirs (Dudgeon et al. 2006; Vörösmarty et al. 2010). In this last scenario, rapid variations in flow caused by hydropower dams (e.g., hydropeaking releases) can be particularly harmful for fish species, especially at young life stages. Additionally, more intense rainfall events under the ongoing climate change scenario (Milly et al. 2002) can also lead to abrupt changes in river flow and to an increasing risk of fish drift downstream (Lechner et al. 2016).

Fish are particularly vulnerable to these threats. Depending on the stage of their life cycle, they may need to perform upstream migrations along the river to reach suitable habitat areas for growth, feeding, refuge, reproduction, and spawning, meaning they will have to swim against the current. Furthermore, these populations are highly dependent on the annual recruitment, which is in turn related to the survival of larvae and young of the year (Magalhães et al. 2003; Schmutz et al. 2016). In this context, species’ swimming abilities need to be included in dispersal models, providing managers with information for decisions on flow regulation (Romão et al. 2012; Boavida et al. 2013) and fishway design (Adams et al. 2009; Peake 2008; Katopodis et al. 2019).

Since the 1960s, several studies addressed swimming performance in freshwater fish species (e.g., Brainbridge 1960; Brett et al. 1964; Cai et al. 2019; George et al. 2019; Tudorache et al. 2007), but until recently, most target species had relatively large ranges and low levels of conservation concern (e.g., common carp Cyprinus carpio Linnaeus, 1758 in Bainbridge 1960; roach Rutilus rutilus (Linnaeus, 1758) in Tudorache et al. 2007; bighead, silver and grass carps, invasive to North America, in George et al. 2018), despite some work already done in sturgeons (Deslauriers & Kieffer 2012; Katopodis et al. 2019; Verhille et al. 2014). Data on the swimming performance of fishes in biodiversity hotspots, such as the Mediterranean basin, considered one of the priority areas for biodiversity conservation (Myers et al. 2000), is still scarce.

Specifically for the Leuciscidae, a fish family that includes endemic Iberian fish species formerly included in Cyprinidae (Schönhuth et al. 2018), 24 out of the 31 species (77%) found in the Iberian Peninsula are endemic (IUCN 2019). Swimming performance studies conducted in this group have been restricted to a small number of species: the Iberian barbel Luciobarbus bocagei (Steindachner, 1864) (in Mateus et al. 2008; Silva et al. 2011; Alexandre et al. 2013), the Tagus nase Pseudochondrostoma polylepis (Steindachner, 1864), the bordello Squalius caroliterti (Doadrio, 1988) (in Romão et al. 2012), the southern straight-mouth nase Pseudochondrostoma willkommii (Steindachner, 1866) (in Alexandre et al. 2016), and the ruivaco Achondrostoma oligolepis (Robalo, Almada, Doadrio & Kottelat, 2005) (in Silva 2016). It should be noted that all these studies used adult individuals, meaning that the swimming performance of developmental stages of Iberian fish species remains yet unknown.

In this study, we sought to measure the critical swimming speed (Ucrit) in three-threatened Iberian leuciscid species: the western ruivaco Achondrostoma occidentale (Robalo, Almada, Sousa Santos, Moreira & Doadrio, 2005), the Portuguese nase Iberochondrostoma lusitanicum (Collares-Pereira, 1980), and the southwestern nase Iberochondrostoma almacai (Coelho, Mesquita & Collares-Pereira, 2005). Swimming performance was tested by measuring the critical swimming speed, Ucrit, defined by Brett (1964) as the maximum velocity that a fish can maintain for a certain period of time, variable among species and among developmental stages. Several authors have suggested that Ucrit is as an ecologically relevant measurement (e.g., Plaut 2001; Faria et al. 2014) and a good indicator of their maximum aerobic performance (Hammer 1995; Drucker 1996), allowing fish to perform rapid vertical movements to face different water currents (Fisher and Leis 2009). Additionally, the role of morphometric variables as predictors of Ucrit was also evaluated, as their importance in determining individual swimming performance has already been demonstrated in other freshwater fish species (minnows: Bestgen et al. 2010; Iberian leuciscids: Mateus et al. 2008).

Material and methods

Test subjects

A total of 49 fish were retrieved from captive stocks currently undergoing a captive breeding program for Portuguese native leuciscids (Sousa-Santos et al. 2014a): 17 A. occidentale, 20 I. lusitanicum, and 12 I. almacai, from the rivers Safarujo, Grândola, and Odelouca, respectively. These are small river basins, with main river courses ranging between 18 km (Safarujo) and 92 km (Odelouca). Due to different availability of early juvenile fish in these captive stocks, A. occidentale and I. lusitanicum were tested in November 2017, while I. almacai juveniles were tested in June 2018. Fish stocks in this captive breeding program are reared following a “naturalistic approach,” aiming to preserve the natural behavioral patterns of the species, such as escape from predators and search for food (Sousa-Santos et al. 2014a). To achieve this, four conditions were established at the start of the program: (1) naturally occurring spawnings; (2) availability of refuge areas for fry and juveniles; (3) natural conditions of light and temperature; and (4) minimal human intervention (Sousa-Santos et al. 2014a).

Critical swimming speed was measured in a swimming chamber made of clear Perspex with 6 parallel swimming lanes, each 30 mm wide, 50 mm high, and 180 mm long. (Fig. 1). Water was retrieved from a PVC tank (200 L capacity) through a pump and redistributed equally in the six lanes composing the swimming chamber (Fig. 1; for further details see Faria et al. 2009). Fish were placed in an aerated container kept at the same temperature as the swimming chamber (20.0 ± 1.0 °C) 1 hour before each trial, to acclimate and recover from any potential handling stress. Fish were then introduced into the swimming chamber (Fig. 1), one fish per lane and left to acclimate at a constant velocity of 5 cm s−1 for 5 min. If a fish was pulled over to the back wall of the lane during this period, it was removed from the experiment.

Fig. 1
figure 1

Schematics of the swimming chamber: (A) water input, equally distributed to the six lanes (I–VI), each 180 mm long and 30 mm wide; (B) outflux; (C) source tank to where the water drops; (D) pump connected to the tank and the pipe circuit, redistributing the water back to the swimming chamber

Full size image

Each trial consisted of a 4-step series of constant water speed increments (5, 10, 15, and 20 cm s-1), lasting 5 min each. Whenever a fish was in the back wall and unable to swim against the current, the time and the water velocity were recorded. If during the trials fish displayed unsteady swimming, i.e., moved back and forward in the chamber (Mateus et al. 2008), they would not be considered for the analyses. After each trial, fish were photographed for subsequent morphometric measurements and placed in an aerated container before returning to their original tanks. No fish died during the experimental procedures. Morphometric measures were taken on calibrated photographs (one photograph per fish, lateral view). Pictures were taken after putting each fish in a transparent aquarium of 15 × 5 × 10 cm (camera model: Canon Powershot SX160 IS), and measures were obtained using the software ImageJ (version 1.52d) and included standard length (“SL”), fork length (“FL”), total fish length (“TL”), body depth (“BD,” measured from the insertion of the dorsal fin to the ventral edge of the fish profile), and caudal peduncle height (“CP”).

Ucrit determination

Ucrit was determined following the formula proposed by Brett (1964): Ucrit = U + [(t/ti) × Ui)], where U is the penultimate speed a fish is able to maintain, Ui the velocity increment, t the time swam in the final velocity increment, and ti the time interval for each velocity increment. As 20 cm s−1 was the maximum velocity step that was possible to attain in this experimental design (due to the pump and swimming chamber constraints), a Ucrit of 25 cm s−1 was assumed for all fish that successfully concluded the trials (as velocity increment = 5 cm s−1).

Data analyses

Data normality was assessed with the Shapiro-Wilk test and distribution observation through histograms, while heteroscedasticity was evaluated using the F-test implemented in the R software (function var.test, R Core Team 2018). Comparison of the Ucrit obtained for the three target fish species was performed with an ANOVA test, followed by a Tukey post hoc test to account for pairwise comparisons. Additionally, a generalized linear model (GLM) for Ucrit was built taking into account the morphometric variables (TL, BD, and CP). To reduce bias related to individual body length, we standardized Ucrit for species comparison and model fit according to total body length: relative critical swimming speed (TL s−1), expressed in body lengths per second (cm/cm s−1). FL and SL measures were excluded due to their strong correlation with TL (r = 0.994), while CP was removed from the analyses due to its correlation with BD, thus preventing collinearity issues in fitted models (see below).

A backwards stepwise regression was computed from a GLM following a Gaussian distribution, with TL, BD, and CP to retrieve the most adequate model, i.e., the one with the lowest Akaike Information Criterion (AIC) score and an AIC difference (Δi) of 2.0, following the recommendations of Burnham and Anderson (2002). For the GLM and stepwise analyses, Ucrit values obtained for the three groups were pooled together. A significance of 0.05 was considered for all analyses, which were performed in R (R Core Team 2018, version 3.5.1), using the function step. AIC from the package MASS (Vernables and Ripley 2002).

Results

A total of 5 fish (2 A. occidentale, 2 I. lusitanicum, and 1 I. almacai) were removed after the acclimation period due to pulling over to the back wall, while 8 fish (2 A. occidentale, 3 I. lusitanicum, and 3 I. almacai) displayed poor rheophilic behavior (i.e., did not try to swim against the current) and were not considered for analyses. Overall, 36 tested fish were considered for analyses: 13 A. occidentale, 15 I. lusitanicum, and 8 I. almacai, ranging from 24 to 53 mm total length (see Table S1 in Electronic Supplementary Material for further details). Twenty-two fish could not withstand the final velocity step (20 cm s−1): 5 A. occidentale (38%), 11 I. lusitanicum (73%), and 7. almacai (88%). The measured morphometric variables and Ucrit (absolute and relative) values for the three species are summarized in Table 1.

Table 1 Minimum, maximum, and mean values with standard deviations for total fish length (TL), body depth (BD), caudal peduncle height (CP), and absolute and relative Ucrit for A. occidentale, I. lusitanicum, and I. almacai
Full size table

Significant differences in Ucrit values were found when comparing the three species (F = 6.347, p = 0.009), with I. almacai having the lowest average relative critical swimming speed value: 4.5 TL s−1 (Table 1). Post hoc comparisons revealed significant differences of the Ucrit values between I. almacai and both A. occidentale (mean deviation = 1.523, p = 0.017) and I. lusitanicum (mean deviation = 1.527, p = 0.014); in contrast, the Ucrit values obtained for these last two species did not differ (mean deviation = 0.004, p = 0.999).

The fitted model for Ucrit retained both morphometric variables entered in the model and significantly related to the response variable (Table 2), total fish length and body depth; with the second being positively related with swimming performance, more “robust” fish (i.e. with higher body depths) were more likely to attain higher relative Ucrit (Fig. 2). In contrast, relative Ucrit was negatively correlated with total fish length, but absolute Ucrit was higher in longer fish (Fig. 2).

Table 2 AIC scores and differences (Δi) for the tested models for relative Ucrit (TL s−1). The most fit model included the two morphometric traits that were considered for these analyses (TL and BD)
Full size table
Fig. 2
figure 2

Correlation plots for (a) total length (mm) and absolute Ucrit (cm s−1); (b) total length (mm) and relative Ucrit (TL s−1); (c) body height (mm) and absolute Ucrit (cm s−1), and (d) body height (mm) and relative Ucrit (TL s−1); different point labels refer to the three species considered: A. occidentale (red triangles); I. lusitanicum (green dots) and I. almacai (blue crosses)

Full size image

Discussion

This study provided the first data on the swimming performance of young developmental fish stages of three small leuciscid species, endemic to the Iberian Peninsula. As this geographic area is part of the Mediterranean basin biodiversity hotspot, the protection of the endemisms is crucial for the maintenance of biodiversity within this diverse fish family.

Our results show that most fish could not withstand the last water velocity step (20 cm s−1), which can be surpassed under natural conditions throughout the year in Portuguese rivers, such as the Odelouca River, where I. almacai occur (Pires et al. 2008). Particularly in early summer, water velocities in this river can vary between 10 and 50 cm s−1 (Pires et al. 2008). Even when regulation by dams and subsequent discharges (e.g., hydropeaking events) are not taking place, larvae and young juvenile fish must rely on refuges to avoid higher velocities, namely, during flood events (Lechner et al. 2016). I. almacai was the species that showed the lowest average relative Ucrit among the three target species, which may indicate that it may be more susceptible to higher water velocities. In a study conducted by Pires et al. (2008) in the Odelouca River, field surveys revealed a decrease in abundance of I. almacai following a flood event in late summer-early fall. Though we are aware that some individuals may be able to surpass the Ucrit threshold of 25 cm s−1 assumed in this study, the information gathered here can be used, for instance, to determine the total habitat suitability, taking into account suitable water velocity (Santos and Ferreira 2008; Boavida et al. 2013), and the definition of criteria for ecological flow discharges in dams and weirs in regulated rivers (Romão et al. 2012).

For the three species studied, at least for captive bred individuals, fish body length and body depth seem to be determinant traits for Ucrit of early juvenile stages, accordingly with previous literature targeting other small minnows (Bestgen et al. 2010) and also the Iberian leuciscid species L. bocagei (Mateus et al. 2008). Indeed, body size is among the most important characteristics that influence swimming performance (Beamish 1978; Hammer 1995, Plaut 2001). Previous studies have also shown that absolute swimming speed (i.e., Ucrit expressed in cm s−1) increases with size; contrastingly, the relative speed (Ucrit expressed in body lengths per second) tends to decrease as size increases (Brett 1965; Hammer 1995; Mateus et al. 2008). The results obtained in this study for three different leuciscid species are concordant with these patterns, which were also found for the Iberian barbel L. bocagei (Mateus et al. 2008).

Following these results, future studies should also address intraspecific morphological differences between populations of the same species, as morphological features may be responsive to local habitat conditions (Alexandre et al. 2013; Haas et al. 2015). A previous study on the growth rate of A. occidentale (Mameri et al. 2018) revealed differences in growth and length at age between the three wild populations of this species, and also an overall higher growth rate in the stocks bred in captivity compared with the wild ones. This is most likely due to the fact that fish experience much less stressful conditions in captivity than in their natural habitat. Studies using captive bred Iberian leuciscids have so far provided valuable insights for the knowledge of their ecology and reproduction (Gil et al. 2010; Sousa-Santos et al. 2014b) but may not fully represent the response that wild individuals would have to certain stressors, such as water velocity increases. Thus, it would be interesting to assess if the pattern of swimming performance herein described is similar for wild-caught individuals, to clarify how differently captive and wild individuals cope with increasing water velocities. Additionally, although we tried to select individuals with similar body length, I. almacai was tested in the early summer, while A. occidentale and I. lusitanicum were tested in late fall, after the breeding season. Therefore, a possible effect of different somatic condition between these seasons, which was not assessed in this study, should be taken into consideration in future studies.

The information gathered in this study can provide important guidelines for the conservation and management of wild stocks of these endemic species, by allowing the establishment of regional water velocity criteria for fishways and habitat restoration measures that promote refuge availability for juveniles and fry (e.g., plant aquatic macrophytes and native riparian tree species with submerged roots acting as shelters), so that sudden instream flow changes would be negligible stressors and species recruitment would not be compromised.