© J. Gelder et al., Published by EDP Sciences 2023

1 Introduction

Freshwater ecosystem fragmentation is recognised as one of the most impactful on the aquatic resources, affecting habitat connectivity on multiple spatial and temporal scales and leading to reduced species geographical distribution and/or communities and populations isolation (Carpenter et al., 2011; Romão et al., 2018; Legrand et al., 2020; Ovidio et al., 2020; Consuegra et al., 2021). As freshwater fish must disperse or migrate throughout the year to access breeding, feeding and refuge habitats, populations are largely impacted in terms of their structure, migration, recruitment or spawning success by physical obstructions (Weibel and Peter, 2013; Mameri et al., 2019; Ovidio et al., 2021; Benitez et al., 2022; Grimardias et al., 2022). Spawning activity is one of the most common motivators for long-distance migration, but other movements may occur outside the spawning period for ontogenetic and trophic reasons (Benitez et al., 2015, 2018). Therefore, the restoration of river longitudinal connectivity is a management restoration action that has to be associated with the presence of qualitative functional habitats and a sufficient physicochemical water quality (Bernhardt and Palmer, 2007; Fullerton et al., 2010; Tummers et al., 2016; Ovidio et al., 2020: 2023).

Scientists and river managers have succeeded in facilitating the passage of fish around or through obstructions using fishways, bypass channels and fish elevators. The ability to use fishways depends on the species and their life stage but also their ability to swim; consequently, fishways designs may vary depending on the target species (Noonan et al., 2012; Silva et al., 2018; Grimardias et al., 2022). Over the last years, progress has been made to improve fishway access and performance, combining knowledges of hydraulics and fish ecology. Fishways design tend to become predominantly adapted to different species, sizes and migratory strategies (Benitez et al., 2015; Ovidio et al., 2017, 2020; Romão et al., 2019; Grimardias et al., 2022).

When new fishways are installed in rivers, there is also a real interest to perform a monitoring programme to evaluate their seasonal use by different species and to quantitatively evaluate the extent to which fish will have access to newly opened river sections. As humanely and/or logistically costly, very few studies on the use of fishways have been done during several consecutive years (Tummers et al., 2016; Legrand et al., 2020; Benitez et al., 2022; Grimardias et al., 2022). Such long-term monitoring is, however, interesting to highlight the between years variability in the use of the fishways for different species under fluctuating environmental conditions (Belliard et al., 2018; Benitez et al., 2022). The use of capture traps as a monitoring method is relatively fastidious because it requires regular human passage. However, this method makes it possible to obtain precise and qualitative information on fish such as species taxonomic determination, individuals weight, size or sex, and to employ tagging for different scientific purposes (Prchalová et al., 2011; Benitez et al., 2022). Moreover, monitoring during several consecutive years since the opening of the migratory axis allows to analyse the temporal processes of colonisation of newly re-opened habitats by fish communities, which is an important, but yet purely informed, scientific key-point for following restoration of longitudinal continuity.

In order to restore connectivity, multi-species vertical slot fishways were installed in three medium size rivers in the south of Belgium. Theses fishways were intensively monitored by capture traps during several consecutive years after setup to obtain data on their use by different fish species and on the evolution and changes of fish species using the fishways over time. Such long-term monitoring is particularly adapted to analyse the colonization dynamic of migratory axes, just after the reestablishment of rivers longitudinal connectivity. In order to meet these objectives, we analysed: (1) the diversity, abundance, biomass and size of species captured in the three fishways; (2) the evolution of the dynamic pattern of capture over consecutive years, at species and ecological guild levels; (3) the periodicity of capture and its variation over years of monitoring; and (4) the environmental conditions (water temperature and flow conditions) associated with species capture.

2 Material and methods

2.1 Study site and fishways characteristics

The study was conducted on three rivers belonging to the Belgian Meuse River basin: the Berwinne, a tributary of the Meuse; the Amblève a tributary of the Ourthe; and the Ourthe (Fig. 1). Each of these rivers have a fishway (FW) built in 2002 (Berwinne River: FW1-B), 2007 (Ambleve River: FW2-A) and 2009 (Ourthe River: FW3-O) to restore connectivity. Before that, no device was present at these physical barriers (concrete ramp dam at FW1-B and FW3-O and hydropower dam at FW2-A). The average annual discharge is 1.9 m³/s for the Berwinne, 19.3 m³/s for the Amblève and 67.4 m³/s for the Ourthe. The ecological status of rivers as defined by biological, physicochemical and hydro morphological indicators is medium for the Berwinne and good for the Amblève and Ourthe Rivers (i.e. Public Service of Wallonia − DEE). According to Huet (1949), the downstream parts of the Berwinne and Amblève Rivers belong to the grayling/barbel fish zone and the Ourthe river is characterised as a barbel fish zone (Huet, 1949). In total, 23 species are potentially present in the Berwinne and Amblève Rivers and 24 species in the Ourthe River (Electrofishing data, University of Liège). All fishways are vertical-slot pool multi-specific types (height between pools ≤0.3 m) equipped with 4 (FW1-B), 15 (FW2-A) and 16 (FW3-O) pools. The three fishways have a constant operating flow and are not influenced significantly by river flow fluctuations. In addition, a capture trap was installed in the three fishways. The first one (FW1-B) was equipped with a grid located in the upper pool with 3 cm of space in the upstream opening and a cone in the downstream opening. The second (FW2-A) and the last one (FW3-O) had a cage in the upstream pool with a grid of 1 × 1 × 1 cm and 5 × 5 × 5 cm, respectively (Tab. 1). The minimum capture size is 50 mm (FW1-B), 25 mm (FW2-A) and 150 mm (FW3-O).

Fig. 1 Locations of the Berneau fishway in the Berwinne (FW1-B), the Lorcé fishway in the Amblève (FW2-A) and the Grosses-Battes fishway in the Ourthe River (FW3-O) and pictures showing fishway configurations.

2.2 Fish capture and environmental variables

The three fishways (FW) were monitored for several consecutive years: from October 2002 to October 2008 for FW1-B, from October 2007 to October 2015 for FW2-A and from September 2009 to September 2012 for FW3-O. The monitoring period ranged from 2 to 5 times per week, depending on the capture intensity with a total of 730 monitoring events at FW1-B, 1311 at FW2-A and 286 at FW3-O. Individuals in the capture trap were caught with a dip net after placing a grid just downstream, which prevents the passage of other individuals during the monitoring.

Captured fishes were anesthetised in a solution of 4-allyl-2-methoxyphenol (Eugenol: 0.1 ml/L), identified at the species level, counted, measured (±1 mm, fork length) and weighed (±1 g). Following biometric analyses, fish were released upstream of the dam after a recuperation period of a few minutes. Fish caught were grouped into different guilds according to their ecological preferences (Benitez et al., 2022):

• Rheophilic species: trout (Salmo trutta), sea trout (Salmo trutta), rainbow trout (Oncorhynchus mykiss), brook trout (Salvelinus fontinalis), barbel (Barbus barbus), chub (Squalius cephalus), nase (Chondrostoma nasus), spirlin (Alburnoides bipunctatus), asp (Aspius aspius), dace (Leuciscus leuciscus), grayling (Thymallus thymallus), loach (Barbatula barbatula) and bullhead (Cottus rhenanus).

• Eurytopic species: common bleak (Alburnus alburnus), common bream (Abramis brama), silver bream (Blicca bjoerkna), roach (Rutilus rutilus), gudgeon (Gobio gobio), european catfish (Silurus glanis), minnow (Phoxinus phoxinus), European eel (Anguilla Anguilla) and three-spined stickleback (Gasterosteus aculeatus).

• Limnophilic species: pike (Esox lucius), perch (Perca fluviatilis), tench (Tinca tinca), ide (Leuciscus idus), common rudd (Scardinius erythrophthalmus), koi (Cyprinus rubrofuscus), common carp (Cyprinus carpio) and leather carp (Cyprinus carpio nudus).

This separation in ecological guilds allows to have a more synthetic view of the colonisation process for species having closer habitat preference.

Environmental variables were continuously recorded (every hour) during the monitoring of the fishways. Data on water temperature (°C) were recorded by data loggers (Tidbit Onset) installed at the inlet of the fishways, and the flow data (m³/s) were granted by SETHY (Wallonia Public Service of Hydrological Studies) located 3 km downstream of the FW1-B, 16 km downstream of the FW2-A and 0.2 km of the FW3-O.

2.3 Data and statistical analysis

Firstly, we produced a global view of the fish diversity (i.e. by species and by ecological guild and in terms of abundance, biomass and size) observed in each FW. We presented the results by year of monitoring in order to take in account the reproductive periods. It should be noted that some species were not present every year in the different fishways. For species with at least 5 individuals captured per year of monitoring, we compared the sizes of individuals between the different monitoring years for the three FW with non-parametric Kruskal-Wallis test.

We investigated dynamic patterns of capture per year of monitoring for each FW, using cumulative curves for the three ecological groups (including all individuals) and for species with a minimum of 10 captured individuals. The number of captures was computed as a percentage, with 100% corresponding to the total number of individuals captured during the entire monitoring, namely during 6 years at FW1-B, 8 years at FW2-A and 3 years at FW3-O. We used the chi² test to determine if the observed distribution of captures during each year of monitoring for each FW was homogeneous or heterogeneous compared to a theoretical number of captures (corresponding to the total capture divided by the number of years of monitoring). We also used chi² test to compare (i) the number of captures between rheophilic and eurytopic guilds during the first three years of monitoring for each fishways and between the three fishways since the lowest monitoring time is 3 years at FW1-B; (ii) the number of captures of rheophilic and eurytopic guilds during the first year compared to the sum of captures in the second and third year. Only species with at least 5 individuals captured per year were considered for these tests. This last constrain excluded the limnophilic guild from these tests.

We analysed the periodicity of capture (by month) per year of monitoring for species with at least 5 individuals captured for each year of monitoring using violinplots. We compared temporal trends in capture periodicity between the years of monitoring with non-parametric Kruskal-Wallis tests. The post hoc pairwise comparison of the Mann-Whitney (U) test was used when the Kruskal-Wallis test was significant.

The environmental values were transformed into daily data, and each fish captured was linked with the environmental data of the previous day's capture (Benitez et al., 2015). The temperature and flow data were analysed by species with a minimum of 3 individuals for each fishway. Since the rivers have different sizes, the flow values were divided by the average flow of each river. We calculated the 25 and 75 percentiles of index flow values during capture (i.e. river flow the day before the capture divided by the average annual flow) to determine 3 migration flow categories:

• Low flow migration: < percentile 25.

• Mean flow migration: between percentile 25 and percentile 75.

• High flow migration: > percentile 75.

The proportion of individuals (%) per species captured for each category was further calculated at the three FW.

The significance level was set at p < 0.05 for all statistical tests (chi² test, Kruskal-Wallis and Mann-Whitney) and was performed using a R statistical program.

3 Results

3.1 Capture diversity (abundance, biomass and size)

A total of n = 1504 individuals from 13 different fish species were captured in the FW1-B from October 2002 to October 2008. In the FW2-A, n = 4507 individuals belonging to 23 species were monitored from October 2007 to October 2015. In the FW3-O, n = 1403 fish from 21 species were captured from September 2009 to September 2012 (Tab. 2).

The most abundant ecological guild at the FW1-B and FW2-A in terms of number of individuals was the rheophilic guild with 82% and 53% of individuals captured, respectively, and the eurytopic guild in FW3-O with 63% of individuals captured. At FW2-A and FW1-B, eurytopic species were the second most abundant guild with 47% and 18% of individuals captured, respectively, and the rheophilic guild with 35% in FW3-O (Tab. 2).

During the first year of monitoring, 397 individuals were captured at FW1-B, 540 at FW2-A and 898 at FW3-O. The number of individuals over the monitoring time varied from 163 to 397 in the FW1-B, from 161 to 1333 in the FW2-A and from 117 to 898 in the FW3-O. This represents 5–10 species, 11–17 species and 11–18 species, respectively. At FW1-B, the greatest number of species was captured between 2004 and 2005 with 10 species, between 2012 and 2013 at FW2-A with 17 species, between 2009–2010 and 2011–2012 at FW3-O with 18 species captured. New species were still captured during the fourth and fifth years of monitoring at FW1-B, during the second, fifth, sixth and eight years at FW2-A and during the second year of monitoring at FW3-O (Fig. 2). In terms of number of individuals per species, the spirlin (rheophilic) was the most abundant at FW1-B (n = 548 individuals), the minnow (eurytopic) at FW2-A (n = 1837) and the bream (eurytopic) at FW3-O (n = 833) (Tab. 2).

Regarding the biomass, rheophilic species were dominant at FW1-B and FW2-A, representing 95% (139 kg) and 96% (608 kg), respectively, of the total biomass and eurytopic species at the FW3-O with 52% (1275 kg) of the total biomass. The most represented species in terms of biomass was the trout at FW1-B (97 kg), the barbel at FW2-A (276 kg) and the bream at FW3-O (1038 kg). The biomass over year of monitoring varied from 14 to 53 kg at FW1-B, from 32 to 161 kg at FW2-A and from 197 to 1778 kg at FW3-O (Tab. 2).

The largest and smallest individuals captured at FW1-B were an eel (765 mm) during the first year and a minnow (31 mm) during the fourth year of monitoring, respectively; a barbel (640 mm) during the last year and a minnow (39 mm) during the first year of monitoring at FW2-A; an European catfish (1160 mm) during the last year and a spirlin (46 mm) during the first year of monitoring at FW3-O (Tab. 2). The KW statistical test showed no trend between the different monitoring year regarding the size of individuals captured in the three FW (KW test, all p> 0.05).

Fig. 2 Histograms of the number of species and new species in the Berwinne (FW1-B), the Amblève (FW2-A) and the Ourthe River (FW3-O), depending on the year of monitoring.

3.2 Dynamic pattern of capture over consecutive years

The rheophilic species were the first to be captured in the three fishways. Species of this guild were captured regularly throughout the year of monitoring at FW1-B and FW2-A, with 50% of individuals captured during the third and fourth years of monitoring, respectively. At FW3-O, rheophilic species showed an earlier capture with 50% of the capture rate during the first year of monitoring; the same trend was observed for the eurytopic species. At FW1-B, the eurytopic species reached 50% of capture rate during the fourth year of monitoring, and during the third year at FW2-A. We observed 50% of capture rate of limnophilic species during the first year of monitoring at FW1-B and FW3-O, and during year sixth at FW2-A (Fig. 3a).

The cumulative frequency of fish capture during years of monitoring changed according to the species (Fig. 3b). The grayling at FW2-A and the trout at FW3-O showed a homogeneous distribution of captures throughout the entire monitoring period (Chi² test, p > 0.05). The trout at FW1-B and FW2-A; the chub at FW1-B and FW3-O; the barbel at FW2-A and FW3-O; the spirlin at FW1-B; the gudgeon at FW2-A; and the nase and the bream at FW3-O had heterogeneous capture frequencies (Chi² test, all p < 0.05). The distribution of the number of captures of rheophilic and eurytopic guilds is significantly different between the first 3 years for the three fishways (Chi² test, all p < 0.05). The number of captures of rheophilic species in the first year was significantly greater than the number of captures of eurytopic species at FW1-B and FW2-A and the reverse trend was observed at FW3-O (Chi² test, p < 0.05). Moreover, the number of captures of rheophilic species during the first year was significantly greater than the sum of the second and third year captures at FW2-A and FW3-O. The same trend was observed for the eurytopic species at FW1-B and FW3-O (Chi² test, p < 0.05) (Fig. 3a). Some species were quickly captured: the barbel at FW1-B and FW3-O, for which 54% and 50% of individuals were captured after 30 and 34 days of monitoring respectively, the sea trout (50% of individuals after 60 days), the chub (51% of individuals after 57 days), the nase (72% of individuals after 24 days), the spirlin (69% of individuals after 6 days) and the bream (51% of individuals after 46 days) at FW3-O (Fig. 3b).

Fig. 3 (a) Cumulative frequency of the three ecological guilds with the shaded area corresponding to the first three years of monitoring common to the three FW and (b) cumulative frequency of captured individuals per species (belonging to the three guilds only) in the Berwinne (FW1-B), Amblève (FW2-A) and Ourthe River (FW3-O) according to the monitoring days.

3.3 Periodicity of capture

The periodicity of capture for the trout at FW1-B, the barbel and the chub at FW3-O showed no significant difference between years of monitoring (KW test, p > 0.05). The periodicity was significantly different between years for the other species: the chub and the spirlin at FW1-B, the trout at FW2-A and FW3-O, the barbel, the gudgeon and the grayling at FW2-A, and the bream, the roach and the nase at FW3-O (KW test, all p < 0.05). The bream at FW3-O showed a significant difference in the periodicity of capture between all the years of monitoring. Some species had only two years with a different periodicity: the barbel at FW2-A (2007–2008 and 2014–2015) and the trout and the nase at FW3-O (2009–2010 and 2011–2012). The chub and the spirlin at FW1-B had a similar periodicity between years 2005–2006 and 2006–2007 and between years 2005–2006 and 2007–2008. The trout, the gudgeon and the grayling at FW2-A had at least 3 years of similar capture periodicity (Fig. 4).

Fig. 4 Violin plots of the periodicity of capture (month) by year of monitoring, and the median represented by a point in the Berwinne (FW1-B), the Amblève (FW2-A) and the Ourthe (FW3-O) River. Species sharing at least one common letter (above each violin plot) did not differ at the 0.05 level of significance.

3.4 Environmental factors

Temperature and flow values during individuals captures varied by species and by FW. The median capture temperature varied from 10 °C (trout) to 22.1 °C (minnow) for the FW1-B, from 7.4 °C (stickleback) to 25.8 °C (minnow) for FW2-A and from 7.9 °C (grayling) to 19.2 °C (common carp) for FW3-O. The river median index flow at which individuals were captured varied from 0.26 (gudgeon) to 1.3 (eel) for FW1-B, from 0.16 (spirlin) to 2.03 (stickleback) for FW2-A and from 0.15 (spirlin) to 1.08 (nase) for FW3-O. The trout was the species captured at the highest water flow index value for the three fishways, with 6 at FW1-B, 5.1 at FW2-A and 3.4 at FW3-O. The minimum water flow index value was 0.10 (minnow) at FW1-B, 0.08 (spirlin) at FW2-A and 0.13 (trout) at FW3-O (Tab. 3).

Most of captures took place at mean flow (flow index values between 0.17 and 0.64) for all FW with 53% of captures at FW1-B, 58% at FW2-A and 73% at FW3-O. The spirlin was the only species that had most of its individuals captured at low flow index values (flow index <0.17) at FW2-A (63.4%) and FW3-O (92.3%). However, at FW1-B 54.7 % of individuals were captured at mean flow index value. The gudgeon at FW1-B, the brook trout at FW2-A, the common bleak and the European catfish at FW2-0 had 100 % of their capture at mean flow index. Other species had most individuals that were captured under different flow index conditions depending on the river (Tab. 4).

4 Discussion

Measures to restore the free movement of fish at physical barriers are generally based on the installation of fishways, as the full removal of these barriers is most often not possible (Silva et al., 2018). Long-term scientific monitoring of fishways is not frequent, and most studies focus on the reproductive period of a few target species or during a limited time period (synthesis in Noonan et al., 2012 and Benitez et al., 2022). In this study, we performed long-term manual monitoring of three multi-species fishways equipped with capture devices as soon as they were installed in order to analyse their progressive use by fish and to perform analysis on the dynamic of colonisation of the re-opened migratory axis, at a multi-species level and over a long period of time.

Our results show that the three fishways were used by a wide diversity of fish species, as the number of species captured represents 58% of the species potentially present in the Berwinne (species absent: the grayling, the dace, the stone loach, the bullhead, the common bleak, the bream, the pike, the tench and the common rudd), 100% of species in the Amblève and 70% in the Ourthe River (species absent: the loach, the bullhead and the minnow). In terms of representativeness of captures in the fishways, the dominant ecological guild was the rheophilic guild in the Berwinne (FW1-B) and in the Amblève (FW2-A) Rivers. These rivers have low mean annual temperatures, coarse substrate and a high current velocity which correspond to rheophilic preferences in terms of habitats (Huet, 1949). The captures in the lower Ourthe River (FW3-O), with higher mean temperature, higher flow and finer substrate, were dominated by the eurytopic species. The important fish diversity sampled in the three fishways attests of their proper functioning through their use by fish species presenting different ecological exigences (Epler et al., 2004; Thiem et al., 2013; Benitez et al., 2015) and swimming capacities (Baudoin et al., 2015). We observed that the number of new fish species captured in the three fishways was variable and gradual from the beginning (axis opening) to the end of the monitoring. Indeed, new species were still captured after 5 years of monitoring at FW1-B, 8 years at FW2-A and 2 years at FW3-O. To obtain 100% of the species captured in the fishways, it took 220 days of monitoring at FW1-B, 935 days at FW2-A and 87 days at FW3-O. Therefore, while lengthening the monitoring time, we succeeded in detecting species which would have been considered absent on a shorter timescale. This underlines the pertinence of long-term monitoring to have a complete view of the fishway use after the opening of a migratory axis as the migratory impulse may vary depending on the species, their functional habitat requirements, or the environment. Lamouroux et al. (2006) observed in a fishway of the Rhône River that the number of species varied from 16 to 26 over the 9 years of monitoring while 32 species were counted in total. The variations in terms of species presence over time between the different rivers could originate from potential seasonal biotic and abiotic variations such as environmental factors that may or may not trigger movements, or pressures present in the rivers that will impact movements in fish populations (Veiga et al., 2006; Costa et al., 2007; De Leeuw and Winter, 2008). As the main goal of installing a fishway is to allow species to move through newly opened habitats, our results underline that the colonisation may be a long process in some instances for some species. But, the important point is that in the long term, the connectivity between river stretches is restored.

The greatest number of individuals were captured during the first year of monitoring at FW1-B (n = 399 individuals) and FW3-O (n = 897 individuals), and during the third year at FW2-A (n = 1333 individuals). Results at FW2-A suggest that even if the fishway was used by fishes just after its opening, the fish capture peaks take some time to appear. Sun et al. (2022) showed a marked increase in trout abundance 4 years after restoration of a migratory axis in the river Deerness in England. The maximum fish biomass was observed during the first year of monitoring for the three fishways. During the first year of monitoring, larger species identified as roach, barbel, grayling or common carp, increased the biomass despite a small number of individuals. Concerning the Ourthe River, a larger number of bream (n = 479) were captured during the first year for a weight of 903 kg with a strong influence on the repartition of the biomass. When assessing the effect of the reopening of a migratory axis by means of fishway monitoring, it is, therefore, important not to extrapolate trends of a single year of monitoring. The size diversity of individuals captured showed that the three fishways are used by individuals of different age classes, both juveniles and adults (Prchalová et al., 2011; Benitez et al., 2015).

Our results on the dynamic pattern showed that the rheophilic species were the first to be captured at the three fishways. These species are very exigent in terms of habitats suggesting that they migrate first in order to find new suitable habitats for their needs (De Leeuw and Winter, 2008; Pander et al., 2013; Benitez and Ovidio, 2017). In addition, as rheophilic species tend to be attracted by higher flows, it is possible that they found the input of fishways more easily (Britton and Pegg, 2011; Benitez and Ovidio, 2017; Benitez et al., 2018). Rheophilic species were regularly captured at FW1-B and FW2-A throughout the year of monitoring and had an early capture peak at FW3-O, while the eurytopic species showed later peaks for the first two fishways and an earlier peak for FW3-O. In addition, our results showed that the number of captures during the first year of opening of the migratory axis was overall higher than the total captures during the second and third years after opening suggesting post-opening effect of migratory axis. We observed that the same species may colonise fishways at different time steps, depending on the river. For example, the barbel migrated at FW1-B and FW3-O (with 50% of the individuals captured during the first year after opening), while much later at FW2-A (50% of the individuals having been captured during the third year of monitoring). This species is known for its important mobility, moving regularly between its resting and feeding habitats but also at the time of the spawning period (Baras et al., 1994; Ovidio et al., 2007; Le Pichon et al., 2016). The sea trout, the chub, the nase, the spirling and the bream at FW3-O migrated early at FW3-O (with 50% of the individuals captured during the first year of monitoring). This tendency may be associated with a quick colonisation process of the migratory axis since, subsequently, the number and biomass of individuals captured for these species decreased (Benitez et al., 2015). Other species reached 50% of capture rate after more than two years of monitoring like the minnow at FW1-B and FW2-A or the roach and the dace at FW2-A with strong variations between years, as previously shown in the Elbe River in Czech Republic (medium flow conditions = 160 m³/s) where the abundance of captures varied from one year to another depending on temperature and flow conditions (Prchalová et al., 2011). These results underline that the temporal dynamic of colonisation of a newly opened river stretch is quite variable between species but also for the same species living in different habitats, and that a complete vision of the process requires multi-year monitoring from the opening.

In terms of periodicity of movements between monitoring periods, we observed that the majority of species (except the trout at FW1-B, the barbel and the dace at FW3-O) had a trend of periodicity that varied over time. Variations of recruitment rates and differences in terms of environmental conditions over monitoring time are important factors that influence movement periodicity over time (Ovidio and Philippart, 2008; Tummers et al., 2016; Pachla et al., 2022). In addition, it could also be expected that movement of individuals from downstream areas to the newly open upstream river stretch may influence the population dynamic and define new biological exchanges that influence mobility patterns of the different size classes in the river (Roscoe and Hinch, 2010). Despite variations of movement periodicity over time, the main peaks were observed during spawning periods for the barbel, the gudgeon, the nase, the grayling, the chub and the bream, which is consistent with the literature (Philippart, 1989; Lucas and Batley, 1996; Fredrich et al., 2003; Epler et al., 2004; Ovidio et al., 2007; Ovidio and Philippart, 2008; Benitez et al., 2015; Romão et al., 2019; Winter et al., 2021). The spirlin at FW1-B showed main peaks outside of its migration period, as also observed by Benitez et al. (2015).

Most of the captures were observed above 8 degrees for the three fishways, although some captures of individuals took place at lower temperatures (e.g. trout captures between 5 and 7 °C or the grayling captures at FW2-A and FW3-O between 6 and 7 °C). In the Odra River in Poland (mean annual flow = 168 m³/s), similar results were obtained with fish captures starting/increasing when temperature reached 8 °C (Kotusz et al., 2006). Temperature ranges of captures for a single species was variable between fishways but with close median values. Some species had wide temperature capture ranges in some fishways and limited in others like the roach with temperatures ranging from 14 to 16 °C at FW1-B (median = 16), 8 to 26 °C at FW2-A (median = 14) and 7 to 18 °C at FW3-O (median = 10.5). The spawning period strongly influenced the temperatures at which individuals of most species were captured (Prchalová et al., 2011; Benitez and Ovidio, 2017). In addition, movement of individuals of a species can vary not only with temperature but also with flow, and sometimes both together (Ovidio et al., 1998; Slavík et al., 2009; Boavida et al., 2018). As for temperature, the flow rate at capture was very variable from one fishway to another as observed by Benitez and Ovidio (2018). The trout was captured at both low and high flow index values. Salmonids are known for their great swimming ability to cope with higher flow conditions (Slavík et al., 2009). The large difference in flow at which trout were captured could be explained by different types of movements (reproduction, habitat change). We observed that during some peaks of flow index values, large rheophilic species were captured (trout, rainbow trout, barbel, chub and nase) while small species were preferentially captured at relatively lower flow values like the minnow and the spirlin (Prchalová et al., 2011). Since the ability to swim against current velocity is related to the size of the individuals, large species would be more adapted to move during important flows, contrary to smaller individuals (Rasmussen and Belk, 2017; Mameri et al., 2019; Stoffers et al., 2022). These differences in the influence of environmental factors on the period of movement must be considered when assessing the effect of river connectivity restoration.

Our study based on multi-annual multi-species analysis of the dynamics of fish colonisation of three fishways in three rivers in Belgium showed a wide temporal diversity of species moving upstream through the devices. We detected the presence of different capture peaks and the arrival of new species, sometimes long time after the opening of the migratory axis. The dynamic of captures varied according to the year of monitoring showing that periodicity may fluctuate over time and depending on the river for some species. In the future, to determine the ecological benefit of the opening of new axis for fish populations, it would be interesting to (i) realize an exhaustive fish sampling downstream of the obstacle (before the opening of the migratory axis) in order to obtain information on the species likely to migrate; (ii) incorporate active telemetry monitoring data of individuals that crossed fishways to analyse their capacity to reproduce and to develop adapted behavioural tactics to exploit new habitats.

Acknowledgements

The monitoring of the fishways was financed by Public Service of Wallonia, non-navigable watercourses. We thank SPW-SETHY and the Aqualim plateforms for the river flow and temperature data. We thank an anonymous reviewer and the Editor in Chief of KMAE for their valuable comments on earlier versions of the manuscript, and the members of the team and the students of UGERAA-ULIEGE for their help to monitor the three fishways.

References

All Tables

All Figures

	Fig. 1 Locations of the Berneau fishway in the Berwinne (FW1-B), the Lorcé fishway in the Amblève (FW2-A) and the Grosses-Battes fishway in the Ourthe River (FW3-O) and pictures showing fishway configurations.
In the text

	Fig. 2 Histograms of the number of species and new species in the Berwinne (FW1-B), the Amblève (FW2-A) and the Ourthe River (FW3-O), depending on the year of monitoring.
In the text

	Fig. 3 (a) Cumulative frequency of the three ecological guilds with the shaded area corresponding to the first three years of monitoring common to the three FW and (b) cumulative frequency of captured individuals per species (belonging to the three guilds only) in the Berwinne (FW1-B), Amblève (FW2-A) and Ourthe River (FW3-O) according to the monitoring days.
In the text

	Fig. 4 Violin plots of the periodicity of capture (month) by year of monitoring, and the median represented by a point in the Berwinne (FW1-B), the Amblève (FW2-A) and the Ourthe (FW3-O) River. Species sharing at least one common letter (above each violin plot) did not differ at the 0.05 level of significance.
In the text