Skip to main content

Contrasting effects of dams with and without reservoirs on the population density of an amphidromous goby in southwestern Japan

Abstract

Dam construction affects the distribution and abundance of freshwater fishes, particularly for diadromous species. Amphidromous fishes spawn in rivers, and hatched larvae descend to the sea (or standing water body such as lakes) immediately after hatching, grow there typically for a few months, and then ascend rivers as small juveniles. Landlocked populations of amphidromous fishes above dams and their migration between dam reservoirs and inlet streams (landlocked migration) have often been reported. In these cases, population density in the inlet streams may be increased by dam reservoirs, owing to reduced migration distance, which results in higher return rates. On the other hand, dams without reservoirs do not promote landlocked migration and merely act as a barrier to their upstream migration, resulting in decreased population density. In this study, we examined such contrasting effects of dams with and without reservoirs on the population density of an amphidromous goby, Rhinogobius fluviatilis, in two river systems (the Shigenobu and Yoshino Rivers) of Shikoku Island, Japan. In the Shigenobu River, where more than 300 sediment-control dams and weirs (barriers without reservoirs) have been installed, our analysis showed that the goby density decreased sharply with increasing the number of barriers higher than or equal to 2 m. On the other hand, in the Yoshino River, where we investigated both landlocked populations above water storage dams (with reservoirs) and non-landlocked populations in tributaries without water storage dams, the analysis showed that the goby density increased with decreasing the distance from standing water body (i.e. the sea or dam reservoirs). Furthermore, monthly surveys at sites above and below a water storage dam showed that the recruitment of ascending juveniles above the dam was more abundant and seasonally earlier than that below the dam, probably owing to reduced migration distance and increased return rates. Our results have implications for understanding how dams alter natural distribution patterns of amphidromous fishes.

Introduction

Dams can produce some inevitable consequences in physical properties of lotic systems, such as alteration of flow regimes, creation of standing water, and disruption of upstream–downstream connectivity, all of which have strong impacts on freshwater fishes (Kingsford 2000; Johnson et al. 2008; Morita et al. 2009). In particular, the barrier effect, which impedes upstream migrations, on diadromous fishes is the most obvious and widely recognised. Diadromous fishes, which migrate between rivers and the sea, are often extirpated from reaches upstream of dams (Katano et al. 2006; Fukushima et al. 2007; Cooney and Kwak 2013; LeMoine and Bodensteiner 2014). On the other hand, changes in life history of diadromous species after dam construction have also been reported. For example, rapid changes from anadromous to non-migratory, stream-resident populations during 20-30 years after dam construction have been found in white-spotted charr (Salvelinus leocmaenis) in Hokkaido, northern Japan (Morita et al. 2000; Shimoda et al. 2002). Similarly, but conversely, an occurrence of a migratory population from a stream-resident population due to dam construction has been reported for white-spotted charr in Arimine Lake (dam reservoir), central Japan: the former migrates between the reservoir and its inlet streams (Nakano et al. 1990).

Amphidromous fishes that spawn in rivers descend to the sea immediately after hatching, grow there typically for a few months, and then ascend rivers as small juveniles (McDowall 2010). During downstream migration of hatched larvae (passive drift), they have no opportunity to feed, because their food (zooplankton) is severely limited in running water (Moriyama et al. 1998; Iguchi and Mizuno 1999). Therefore, larvae of amphidromous species have to quickly reach the sea or standing water bodies, where zooplankton is abundant, to survive. In this context, reservoirs behind dams can provide a nursery habitat for larvae of amphidromous species. In fact, migration of landlocked populations between dam reservoirs and their inlet streams (hereafter landlocked migration) has frequently been reported for amphidromous species, especially gobies of the genus Rhinogobius (Tsunagawa and Arai 2008; Takagi et al. 2011, 2012, 2013).

Rhinogobius species are freshwater benthic fishes distributed around East and Southeast Asia, with more than ten species having been recorded in Japan (Yamasaki et al. 2015). Most of the species in the Japanese Archipelago are amphidromous and exhibit distinct distribution patterns along the longitudinal gradient of rivers or the size gradient of river catchments (Mizuno 1976; Uehara 1984; Kawanabe and Mizuno 1989). Among them, Rhinogobius fluviatilis (formerly called large-dark type) is characterised by occupying upper reaches of large rivers (Sone et al. 2001; Tamada 2011). Therefore, this species is particularly susceptible to high mortality during the migration at the larval and juvenile stages. Long-distance downstream and upstream migrations would raise the risk of starvation (Moriyama et al. 1998; Iguchi and Mizuno 1999) and decrease the rate of return to their adult habitats (Hitt et al. 2012), respectively.

Dams are constructed for various purposes, such as hydroelectricity, irrigation, flood control, and sediment control. When viewed from the perspective of their barrier effects on amphidromous fishes; however, dams are broadly categorised into whether these have a large reservoir or not. Dams with large reservoirs are often built in mountainous to foothill regions for multiple purposes (e.g. water supply, hydroelectricity, flood control). In general, such water storage dams completely prevent upstream migration of fishes (but see Holmquist et al. 1998). For amphidromous fishes like R. fluviatilis, however, large reservoirs behind dams can serve as larval habitats for populations above the dams. If the landlocked migration is established owing to a reservoir, the migration distance should be much reduced, resulting in lower mortality rates at early life stages. Consequently, dam reservoirs may increase the density of landlocked populations. On the other hand, dams without reservoirs do not promote landlocked migration, and merely act as a barrier to upstream migration from the sea. In most rivers in Japan, numerous sediment-control dams, which are not intended to store water, are installed in their upper reaches (e.g. Morita and Yamamoto 2002; Katano et al. 2006). These dams are small compared with water storage dams, but often arranged in series (e.g. Nakamura 2001; Kawanishi et al. 2011; Kikuchi and Inoue 2014), acting as multiple barriers for migrating fishes. Although gobies of the genus Rhinogobius are capable of passing through such relatively small dams by climbing using a suctorial disk (fused pelvic fins) (Katano et al. 2006; Takagi et al. 2011; see Holmquist et al. 1998; Cooney and Kwak 2013 for other gobies), a series of sediment-control dams may reduce their return rates to adult habitats.

Overall, dams with and without reservoirs would have contrasting consequences on the population density of amphidromous fishes: dams with reservoirs would increase the density, while dams without reservoirs would decrease it. Such effects on amphidromous fishes can be logically predicted, but have not been tested through empirical studies. In this study, we examined (1) whether dams without reservoirs negatively affect the population density of R. fluviatilis in spite of their high climbing ability (Katano et al. 2006; Takagi et al. 2011), and (2) whether dams with large reservoirs have positive effects on the population density of landlocked R. fluviatilis due to reduced migration distance.

Materials and methods

The study was conducted in the Yoshino and Shigenobu Rivers in Shikoku Island, southwestern Japan (Fig. 1). The Yoshino River is highly impounded for water supply and flood control, whereas the Shigenobu River is highly regulated by numerous sediment-control dams. Effects of dams with and without reservoirs on the population density of Rhinogobius fluviatilis were examined in tributaries of the Yoshino and Shigenobu Rivers, respectively. In addition, using a water storage dam (Ishite-gawa Dam) in the Shigenobu River (Fig. 1a), we examined effects of reduced migration distance due to the dam reservoir by comparing recruitment of ascending juveniles between landlocked (above dam) and non-landlocked (below dam) populations. We expected that juvenile recruitment of R. fluviatilis would be earlier in landlocked (above dam) populations than in non-landlocked (below dam) populations, because migration distance for the former is much shorter than that for the latter.

Fig. 1
figure1

Maps of the Shigenobu and Yoshino River systems in Shikoku Island, Japan. Open circles indicate sampling sites for non-landlocked (a, c) and landlocked (b) populations of Rhinogobius fluviatilis. Two open triangles in the Shigenobu River map (a) indicate the study sites for recruitment of ascending juveniles above and below Ishite-gawa Dam

In the Shigenobu and Yoshino Rivers, two other species of the genus Rhinogobius, Rhinogobius nagoyae (formerly called cross-band type) and Rhinogobius flumineus, are widely distributed. These species could be regarded as potential competitors of R. fluviatilis. However, we think that their effects on the density of R. fluviatilis are negligible in our study. Our study sites were limited to upper reaches (see below), which are not the main habitat of R. nagoyae (Sone et al. 2001; Tamada 2011). Although R. flumineus occupy upper reaches, their body size is smaller than that of R. fluviatilis, and thus, R. flumineus would have little competitive effect of R. fluviatilis. Rather, effects of R. fluviatilis on R. flumineus would be greater.

Effects of dams without reservoirs. The Shigenobu River (catchment area: 445 km2; main stem length: 36 km) is located near Matsuyama City, northwestern Shikoku (Fig. 1a). This river has high sediment yield in its upper reaches and exhibits an intermittent braided channel in the middle reach. Because of this situation, numerous sediment-control dams and weirs have been installed throughout the catchment, from headwater tributaries to near the sea, to prevent sediment disasters and for irrigation (more than 300 dams; see Fig. 1 in Kawanishi et al. 2011). We established 26 study sites in upper reaches of the main stem and six tributaries where there were no water storage dams (Fig. 1a, Table 1). Because all study sites were located upstream of the intermittent reaches, effects of surface water drying of the intermittent reaches as a migration barrier can be assumed to be similar among the study sites.

Table 1 General descriptions of the study sites established in the Shigenobu and Yoshino Rivers (mean ± SD, with range in parenthesis). Effects of dams without reservoirs as barriers to upstream migration were examined in the Shigenobu River while effects of dam reservoirs via landlocked migration were examined in the Yoshino River. The number of potential barrier was not quantified for the study sites in the Yoshino River

At each site, the abundance of R. fluviatilis was quantified from late July to mid-September 2007. A sampling reach was established in a riffle (reach length: 10-50 m depending on stream size), which is the main habitat of R. fluviatilis (Sone et al. 2001), at each site. Both ends of the sampling reach were blocked with 5-mm mesh nets and two removal passes were made using an electrofishing unit (Model 12 Backpack Electrofisher; Smith-Root Inc., Vancouver, WA, USA). Rhinogobius fluviatilis captured were measured for total length (TL), counted and then released alive. The total number of R. fluviatilis larger than 3 cm in TL captured by the two passes divided by the sampling area was used as an index of population density (N m−2).

Variables representing the barrier effect and other habitat characteristics were quantified at each site in 2006 and 2007. We regarded dams, weirs and natural waterfalls higher than or equal to 0.5 m without fish pass as potential barriers, and mapped all the barriers located downstream of the upper most site in each tributary (almost throughout the river system; see Fig. 1 in Kawanishi et al. 2011), with their height measured in the field. Three variables were used to represent the barrier effect: the number of barriers higher than or equal to 0.5-m, 1-m, and 2-m downstream of each site. Other habitat variables were distance from the sea (km), elevation (m), catchment area (km2), channel gradient (%), summer maximum water temperature (°C), substrate coarseness and heterogeneity, and electrical conductivity (EC: mS/cm). Distance from the sea, elevation, catchment area and channel gradient were measured using a 1: 25,000 topographic map. Summer maximum water temperature was obtained using a maximum-minimum thermometer placed at each site (from 26 July to 19 September 2007). To assess substrate conditions, we established a zigzag line (length: 50-150 m depending on stream size) with equally spaced recording quadrats (30-cm square, 0.5-1.0 m intervals) on the riverbed (including non-submerged areas). Substrate type within each quadrat was categorised and coded in order of coarseness as follows: 1 = bedrock (including concrete), 2 = sand (dominant particle size < 2 mm), 3 = gravel (2-16 mm), 4 = pebble (17-64 mm), 5 = cobble (65-256 mm), and 6 = boulder (> 256 mm). Mean and standard deviation of these coded values within each site were used as substrate coarseness and heterogeneity, respectively (Bain et al. 1985). Electrical conductivity (EC) was measured four times (December 2006, February, April and July 2007) using a portable EC meter (B-173, Horiba) and averaged as an index of water quality.

To examine effects of barriers and other habitat characteristics on population density of R. fluviatilis, the density data were analysed using generalised linear models (GLMs) with a gamma error distribution and a log link function. The response variable was the population density of R. fluviatilis at each site, with a small constant (10−5) being added to avoid problems with zero in GLM with a gamma distribution. Explanatory variables were the number of barriers higher than or equal to 0.5-m, 1-m, and 2-m downstream of each site, catchment area, elevation, channel gradient, distance from the sea, summer maximum temperature, substrate coarseness and heterogeneity, and average EC. We compared all possible models with different set of explanatory variables using Akaike’s information criterion (AIC; Burnham and Anderson 2002) and selected models with < 2 ΔAIC as the best-fit models explaining the population density. We also performed model averaging on the subset of best models, and calculated relative variable importance (RVI) of the explanatory variables for the subset of best models by summing recalculated Akaike weights across those models. In the model selection, we removed models with two or more variables on the number of barriers (i.e. ≥ 0.5-m, 1-m, or 2-m height). All analyses were performed using R version 3.1.2 (R Core Team 2014).

Effects of dam reservoirs. The Yoshino River (catchment area: 3750 km2; main stem length: 194 km), the largest river in Shikoku Island, originates in the central mountains of Shikoku and flows eastward to the Kii Channel near Tokushima City (Fig. 1). In upper reaches of this river, multiple water storage dams have been built for hydroelectricity, water supply and flood control. We established 13 study sites upstream of four water storage dams (Fig. 1b), Sameura (height: 106 m; reservoir area: 7.5 km2; elevation: 350 m; built in 1975), Ohashi (74 m; 1.0 km2; 580 m; 1939), Nagasawa (72 m; 1.4 km2; 660 m; 1949) and Omori-gawa (73 m; 0.9 km2; 780 m; 1959) Dams. These study sites were established as potential habitats for landlocked populations; R. fluviatilis populations above those dams have been confirmed to be landlocked using genetic and otolith Sr/Ca analyses (Takagi et al. 2013). For non-landlocked populations, we selected three tributaries, the Akui, Anabuki and Sadamitsu Rivers, so as to include non-landlocked sites having similar elevation to the landlocked sites (Fig. 1c, Table 1). In these three rivers, 13 study sites were established, resulting in 26 study sites in total. The Akui, Anabuki and Sadamitsu Rivers had no water storage dams, although these rivers had sediment-control dams and weirs.

At each site, the abundance of R. fluviatilis was assessed by underwater visual counts from June to September 2009. Along a stream reach including 1-3 pool-riffle sequences at each site, 5-20 equally spaced belt transects (1-m width) perpendicular to flow were established. The number of R. fluviatilis larger than 3 cm in TL encountered within the belt was recorded while snorkeling. In study sites of relatively small streams (wetted width < 5 m), the underwater counts were made along the entire stream reach, not using the belt-transect. The number of R. fluviatilis divided by the surveyed area (N m−2) was used as an index of population density.

Six habitat variables, distance from the nearest downstream standing water body (sea or reservoir), elevation, catchment area, channel gradient, and substrate coarseness and heterogeneity, were measured at each site, with methods for channel gradient and substrate being slightly different from those used in the Shigenobu River described above. The channel gradient, expressed as the percent change in the relative height to stream length (50-120 m reach depending on stream size), was measured in the field using a level and levelling rod. Substrate conditions were assessed using the transects for fish counts; seven equally spaced quadrats (30-cm square) were established along each transect. Substrate type within each quadrat was coded in the same way as described above, and mean and standard deviation of the substrate codes within each site were used as substrate coarseness and heterogeneity, respectively. In small-stream study sites where transects were not used for fish counts, transects were established for the substrate survey.

To examine effects of migration distance and other habitat characteristics on population density of R. fluviatilis, the density data were analysed using GLMs with a gamma error distribution and a log link function. The response variable was the population density of R. fluviatilis at each site. Explanatory variables were catchment area, elevation, channel gradient, distance from standing water body (sea/dam reservoir), substrate coarseness and heterogeneity. Model selection followed the same procedure as the analyses for the Shigenobu River.

Recruitment of ascending juveniles. We examined recruitment of ascending juveniles in populations above and below Ishite-gawa Dam (height: 87 m; reservoir area: 0.5 km2; built in 1972) in the Shigenobu River (Fig. 1a). Rhinogobius fluviatilis population above this dam has been confirmed to be landlocked using genetic and otolith Sr/Ca analyses (Takagi et al. 2011). Two study sites, above (1.5 km from the dam) and below the dam (14.3 km from the sea), were established, and R. fluviatilis were captured by two-pass electrofishing (Model 12 Backpack Electrofisher; Smith-Root Inc., Vancouver, WA, USA) in a riffle at each site (area: 20-80 m2) one-four times per month from early July 2008 to late September 2009 (except July 2009). All individuals captured (including < 3 cm in TL) were measured for TL, counted and then released alive. Seasonal dynamics of size structure and population density were compared between the two sites.

Results

Effects of dams without reservoirs on population density. Of 26 sampling sites in the Shigenobu River, Rhinogobius fluviatilis was found at 14 sites, with varying degrees of the population density (0.03–1.72 individuals m−2). The model selection revealed that the number of barriers higher than or equal to 2 m was the most important variable in explaining the population density: this variable was selected in all 21 top models with < 2 ΔAIC and had the highest RVI value (Table 2). The population density sharply decreased with increasing the number of barriers higher than 2 m (Fig. 2). Elevation, channel gradient, catchment area, and substrate heterogeneity were selected in 7-9 of the top models, indicating their moderate importance. The numbers of barriers higher than or equal to 0.5 m and 1 m were not included in the top models.

Table 2 Model selection results for population density of Rhinogobius fluviatilis in the Shigenobu River
Fig. 2
figure2

Population density of Rhinogobius fluviatilis in relation to elevation, distance from the sea, and the number of barriers higher than or equal to 2 m. The solid line indicates predicted values from the best model of GLM with gamma error distribution. The value of another explanatory variable in the best model (i.e. elevation; see Table 2) is fixed at its mean value to isolate the partial contribution of the number of barriers to the population density

Effects of dams with reservoirs on population density. Of 26 sampling sites in the Yoshino River, R. fluviatilis was found at 13 sites, with varying degrees of the population density (0.004–3.141 individuals m−2). The density at the landlocked sites (above dam) was generally higher than that at non-landlocked sites (Fig. 3), although no R. fluviatilis was found at six of the 13 landlocked sites. The model selection revealed that distance from standing water body (sea/reservoir) and channel gradient were the most important variables in explaining the population density: both variables were selected in all five top models with < 2 ΔAIC and had the highest RVI value (Table 3). The population density decreased with increasing distance from standing water body and channel gradient (Fig. 3).

Fig. 3
figure3

Population density of Rhinogobius fluviatilis in relation to elevation, distance from standing water body and channel gradient. Open and solid circles indicate study sites for non-landlocked and landlocked populations, respectively, in the Yoshino River. The solid lines indicate predicted values from the best model of GLM with gamma error distribution. The value of another explanatory variable in the best model (i.e. distance from standing water body or channel gradient; see Table 3) is fixed at its mean value to isolate the partial contribution of each variable to the population density

Table 3 Model selection results for population density of Rhinogobius fluviatilis in the Yoshino River

Recruitment of ascending juveniles. Population density, size structure, and timing of occurrence of ascending juveniles were clearly different between populations above and below the dam (Figs. 4, 5). The density was roughly ten times higher in the above-dam site than in the below-dam site (Fig. 4). In the below-dam site, most individuals were larger than 5 cm TL throughout the sampling period (Fig. 5). On the other hand, in the above-dam site, individuals smaller than or equal to 5 cm TL greatly contributed to the above-dam population in several months. In 2008, individuals smaller than 3 cm TL, which are regarded as ascending age-0 juveniles, had already been captured in early July in the above-dam site, whereas in the below-dam site those were not found until early September. In 2009, the recruitment of individuals smaller than 3 cm TL first occurred in early August in the above-dam site, but such small individuals were not found during the sampling period in the below-dam site.

Fig. 4
figure4

Mean population density of Rhinogobius fluviatilis above and below the Ishite-gawa Dam in each sampling month

Fig. 5
figure5

Size structure (in total body length) of Rhinogobius fluviatilis above (solid column) and below (open column) the Ishite-gawa Dam in each sampling month from July 2008 to September 2009. The number of individuals is indicated in each panel. Dashed lines are for reference. Arrows indicate the first occurrence of individuals < 3 cm TL in each year at each site. Data in December 2008, January and February 2009 are not shown, because sample sizes in both study sites were small (4-17 and 2-5 individuals above and below the dam, respectively)

Discussion

In the Shigenobu River, Rhinogobius fluviatilis density was primarily explained by the number of barriers higher than or equal to 2 m, suggesting that their densities upstream of multiple dams were decreased by barrier effects. On the other hand, in the Yoshino River, the densities of landlocked populations upstream of water storage dams were generally higher than that of non-landlocked populations. Their density was negatively related to the distance from standing water body, suggesting effects of reduced migration distance by dam reservoirs. The comparisons of seasonal dynamics of size structure and population density between populations above and below Ishite-gawa Dam in the Shigenobu River showed distinct differences consistent with the result from the Yoshino River. In the below-dam site, ascending juveniles were scarce, their recruitment was delayed, and total density was very low in comparison with the above-dam site. These results indicate contrasting effects of dams with and without reservoirs: dams with reservoirs increased the density, while dams without reservoirs decreased.

Upstream extirpation by dams or other barriers has been reported for various taxa, such as salmonids (Nakano et al. 1995; Morita and Yamamoto 2002), cyprinids (Winston et al. 1991; Katano et al. 2006), catostomids (Reid et al. 2008) and cottids (LeMoine and Bodensteiner 2014). Amphidromous gobiids, including Rhinogobius species, however, are often found upstream of dams (Holmquist et al. 1998; Katano et al. 2006; Cooney and Kwak 2013). In the Shigenobu River, the uppermost site where R. fluviatilis was captured had 46 potential barriers (including those higher than 10 m) downstream. Otolith Sr/Ca analysis by Takagi et al. (2011) showed that an adult individual of R. fluviatilis at the same site had experienced saltwater, indicating that the individual had passed through the series of dams from the sea. Katano et al. (2006) also found gobies of the genus Rhinogobius above dams of 1.5-3.9-m height, and confirmed their upstream movement over the dam of 1.5-m height using a mark-recapture survey. In Puerto Rico, gobies of the genus Sicydium have often been found above dams of > 10 m height (Holmquist et al. 1998; Cooney and Kwak 2013). Therefore, some gobies can be regarded as highly resistant to the barrier effect owing to their suctorial disk (fused pelvic fins). However, our results from the Shigenobu River indicate that, even in case of a goby having a great climbing ability, its density can be significantly reduced by the barrier effect of dams.

In contrast to the results from the Shigenobu River, the present investigation in the Yoshino River showed that the densities of non-landlocked populations were generally less than 0.5 individuals per m2 while those of landlocked populations above dams sometimes exceeded 1.5 individuals per m2, as much as more than three times higher than the former (Fig. 3). Our 13 study sites for landlocked populations above dams are more than 100 km inland from the sea, but actual migration distances are 0.9-11.6 km (i.e. distance from reservoirs; Table 1). Migration distances for non-landlocked populations of the other 13 sites are 33.5-80.5 km (i.e. distance from the sea; Table 1). Our analysis of the data from the Yoshino River suggests that such differences in migration distance affected the population density. The longer the upstream migration distance, the greater the number of barriers, which would result in lower return rate to adult habitats. Although this interpretation is reasonable, a weakness of our study design could be the spatial separation of landlocked (Fig. 1b) and non-landlocked sites (Fig. 1c). That is, possible effects of region-specific factors cannot be separated from effects of reduced migration distance by dam reservoirs. However, our investigation using Ishite-gawa Dam strongly supports positive effects of reduced migration distance by dam reservoirs on the population density of R. fluviatilis. The higher return rate of juveniles and its resultant higher population density of a landlocked population were clearly shown by the comparison between populations above and below Ishite-gawa Dam. The density in the above-dam site was roughly ten times higher than that in the below-dam site (Fig. 4). These results suggest that non-landlocked populations, which obligatorily migrate to the sea, are often far below their carrying capacity, probably owing to high density-independent mortality during their long-distance migration, as R. fluviatilis occupy upper reaches (Tamada 2011).

Although high densities of landlocked populations due to reduced migration distance may be regarded as a positive aspect of dam reservoirs for R. fluviatilis, this dam effect involves some other issues. Although dams with reservoirs may increase population density of above-dam populations, the latter are isolated by the former, owing to their barrier effects. Isolated populations above dams are reduced in total population size, and thus tend to suffer from higher extinction risks through demographic, environmental, and genetic stochasticity (Morita and Yamamoto 2002). Sakai et al. (2004) and Takagi et al. (2012) reported that landlocked populations of Rhinogobius mizunoi (formerly called cobalt type) had lower genetic diversity than non-landlocked populations. Although such a decline of genetic diversity due to isolation by dams has not been detected in our study sites (Takagi et al. 2011, 2013), isolation and fragmentation of populations involve increased extinction risk in general, especially when the size of isolated populations is small (Morita and Yamamoto 2002).

Another potential issue is competitive effects on other species. Rhinogobius species are widely distributed in various types of streams and rivers, even in reaches upstream of barriers (Katano et al. 2006; this study). However, the negative effect of barriers and the positive effect of reduced migration distance shown by our analyses suggest that population densities of amphidromous gobies of the genus Rhinogobius are often far below their carrying capacities, owing to density-independent mortality during their migrations. Increased density of landlocked populations above dams results in increased severity of intra- and interspecific competitions, which may affect other species occupying similar niches. Sone et al. (2001, 2006) showed competitive effects of R. fluviatilis on habitat use by Rhinogobius nagoyae (i.e. habitat shift). In our study sites, R. fluviatilis usually occur sympatrically with Rhinogobius flumineus, and the density of landlocked R. fluviatilis were, in some cases, extremely high in comparison to that of non-landlocked populations. In such cases, the population density, habitat use and behaviour of R. flumineus may be affected by R. fluviatilis, which remains to be investigated.

Our analyses revealed that dams with reservoirs can increase population density of a landlocked amphidromous species, while dams without reservoirs have a negative effect on that. It should be noted, however, that all amphidromous populations do not necessarily establish landlocked migration when a dam reservoir is constructed. In Puerto Rico, where amphidromous species dominate, barrier effects of dams on them have been investigated extensively (Holmquist et al. 1998; Cooney and Kwak 2013). Nevertheless, landlocked migration has not been reported (Holmquist et al. 1998; Cooney and Kwak 2013). In Kurose Dam of the Kamogawa River in Shikoku, R. nagoyae has established landlocked migration, while R. fluviatilis has not (Takagi et al. 2012). In our study, no R. fluviatilis were found in six of the 13 study sites upstream of the dam reservoirs in the Yoshino River. Probably, R. fluviatilis failed to establish landlocked migration in these six sites when the dams were built, although the reason is unclear. Dams with reservoirs cause upstream extirpation of amphidromous species, unless they establish landlocked migration. Overall, our results showed variable effects of dams: upstream extirpation by dams with reservoirs, decreased population density by dams without reservoirs, and increased density by reservoirs through landlocking. Further research is needed to identify what species have a potential for landlocked migration, and to examine factors affecting success and failure of landlocked migration. Our results have implications for understanding how dams alter natural distribution patterns of amphidromous fishes.

References

  1. Bain MB, Finn JT, Booke HE (1985) Quantifying stream substrate for habitat analysis studies. N Am J Fish Manage 5:499–506

    Article  Google Scholar 

  2. Burnham KP, Anderson DR (2002) Model selection and multimodel inference: a practical information-theoretic approach, second edition. Springer, NY

    Google Scholar 

  3. Cooney PB, Kwak TJ (2013) Spatial extent and dynamics of dam impacts on tropical island freshwater fish assemblages. BioScience 63:176–190

    Article  Google Scholar 

  4. Fukushima M, Kameyama S, Kaneko M, Nakao K, Steel EA (2007) Modelling the effects of dams on freshwater fish distributions in Hokkaido, Japan. Freshw Biol 52:1511–1524

    Article  Google Scholar 

  5. Hitt NP, Eyler S, Wofford JEB (2012) Dam removal increases American eel abundance in distant headwater streams. Trans Am Fish Soc 141:1171–1179

    Article  Google Scholar 

  6. Holmquist JG, Schmidt-Gengenbach JM, Yoshioka BB (1998) High dams and marine-freshwater linkages: effects on native and introduced fauna in the Caribbean. Conserv Biol 12:621–630

    Article  Google Scholar 

  7. Iguchi K, Mizuno N (1999) Early starvation limits survival in amphidromous fishes. J Fish Biol 54:705–712

    Article  Google Scholar 

  8. Johnson PTJ, Olden JD, Vander Zanden MJ (2008) Dam invaders: impoundments facilitate biological invasions into freshwaters. Front Ecol Environ 6:357–363

    Article  Google Scholar 

  9. Katano O, Nakamura T, Abe S, Yamamoto S, Baba Y (2006) Comparison of fish communities between above- and below-dam sections of small streams: barrier effect to diadromous fishes. J Fish Biol 68:767–782

    Article  Google Scholar 

  10. Kawanabe H, Mizuno N (1989) Freshwater fishes of Japan. Yama-kei, Tokyo, pp 584–603

    Google Scholar 

  11. Kawanishi R, Inoue M, Takagi M, Miyake Y, Shimizu T (2011) Habitat factors affecting the distribution and abundance of the spinous loach Cobitis shikokuensis in southwestern Japan. Ichthyol Res 58:202–208

    Article  Google Scholar 

  12. Kikuchi S, Inoue M (2014) Population fragmentation of a stream-resident salmonid by dams: downstream progress of extinctions from headwaters. Ecol Civil Eng 17:17–28

    Article  Google Scholar 

  13. Kingsford RT (2000) Ecological impacts of dams, water diversions and river management on floodplain wetland in Australia. Austral Ecol 25:109–127

    Article  Google Scholar 

  14. LeMoine MT, Bodensteiner LR (2014) Barriers to upstream passage by two migratory sculpins, prickly sculpin (Cottus asper) and coastrange sculpin (Cottus aleuticus), in northern Puget Sound lowland streams. Can J Fish Aquat Sci 71: 1758–1765.

    Article  Google Scholar 

  15. McDowall RM (2010) Why be amphidromous: expatrial dispersal and the place of source and sink population dynamics? Rev Fish Biol Fisheries 20:87–100

    Article  Google Scholar 

  16. Mizuno N (1976) Studies on a freshwater fish, Rhinogobius brunneus (Pisces: Gobiidae) III. Distribution of four colour types in Shikoku and Kyushu Islands. Physiol Ecol Japan 17:373–381

    Google Scholar 

  17. Morita K, Morita SH, Yamamoto S (2009) Effects of habitat fragmentation by damming on salmonid fishes: lessons from white-spotted charr in Japan. Ecol Res 24:711–722

    Article  Google Scholar 

  18. Morita K, Yamamoto S (2002) Effects of habitat fragmentation by damming on the persistence of stream-dwelling charr populations. Conserv Biol 16:1318–1323

    Article  Google Scholar 

  19. Morita K, Yamamoto S, Hoshino N (2000) Extreme life history change of white-spotted char (Salvelinus leucomaenis) after damming. Can J Fish Aquat Sci 57:1300–1306

    Article  Google Scholar 

  20. Moriyama A, Yanagisawa Y, Mizuno N, Omori K (1998) Starvation of drifting goby larvae due to retention of free embryos in upstream reaches. Environ Biol Fish 52:321–329

    Article  Google Scholar 

  21. Nakamura T (2001) Estimation of the distribution of genetically pure populations of the Japanese charr by inquiring survey. J Jpn Soc Erosion Control Eng 53:3–9

    Google Scholar 

  22. Nakano S, Inoue M, Kuwahara T, Toyoshima T, Hojyo H, Fujito E, Sugiyama H, Okuyama S, Sasa K (1995) Freshwater fish fauna in the Teshio and Nakagawa Experimental Forests and adjacent areas with reference to damming effects on their distribution. Res Bull Hokkaido Univ For 52:95–109

    Google Scholar 

  23. Nakano S, Maekawa K, Yamomoto S (1990) Change of the life cycle of Japanese charr following artificial lake construction by damming. Nippon Suisan Gakkaishi 56:1901–1905

    Article  Google Scholar 

  24. R Core Team (2014) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna. http://www.R-project.org/.

  25. Reid SM, Mandrak NE, Carl LM, Wilson CC (2008) Influence of dams and habitat condition on the distribution of redhorse (Moxostoma) species in the Grand River watershed, Ontario. Environ Biol Fish 81:111–125

    Article  Google Scholar 

  26. Sakai H, Arai T, Imai C, Sugiyama H, Sato H, Jeon SR (2004) Landlocked populations of an amphidromous goby, Rhinogobius sp. CO. Japan J Ichthyol 51:175–180

    Google Scholar 

  27. Shimoda K, Nakano S, Yamamoto S (2002) Landlocking of anadromous white-spotted charr Salvelinus leucomaenis by damming. Japan J Ichthyol 49:25–32

    Google Scholar 

  28. Sone S, Inoue M, Yanagisawa Y (2001) Habitat use and diet of two stream gobies of the genus Rhinogobius in south-western Shikoku, Japan. Ecol Res 16:205–219

    Article  Google Scholar 

  29. Sone S, Inoue M, Yanagisawa Y (2006) Competition between two congeneric stream gobies for habitat in southwestern Shikoku, Japan. Ichthyol Res 53:19–3

  30. Takagi M, Sekiya I, Shibakawa R, Shimizu T, Kawanishi R, Inoue M (2012) Genetic influence of the dam isolations in the Kamo and Nakayama River systems on the common freshwater goby populations Rhinogobius spp. Ecol Civil Eng 15:161–70

  31. Takagi M, Shibakawa R, Shimizu T, Omori K, Inoue M (2013) Genetic divergence and landlocking of a common freshwater goby Rhinogobius fluviatilis in the Yoshino River system. Ecol Civil Eng 16:13–22

    Article  Google Scholar 

  32. Takagi M, Yano S, Shibakawa R, Shimizu T, Ohara K, Sumizaki Y, Kawanishi R, Inoue M (2011) Landlocking and genetic differentiation of a common freshwater goby Rhinogobius sp. LD caused by Ishite-gawa Dam in the Shigenobu River system. Ecol Civil Eng 14:35–44

    Article  Google Scholar 

  33. Tamada K (2011) River bed features affect the riverine distribution of two amphidromous Rhinogobius species. Ecol Freshw Fish 20:23–32

    Article  Google Scholar 

  34. Tsunagawa T, Arai T (2008) Flexible migration of Japanese freshwater gobies Rhinogobius spp. as revealed by otholith Sr:Ca ratios. J Fish Biol 73:2421–2433

    Article  Google Scholar 

  35. Uehara S (1984) Distributions of four colour types of the freshwater goby, Rhinogobius brunneus, in the Tohoku District, Japan. Sci Rep Yokosuka City Mus 32:33–49

    Google Scholar 

  36. Winston MR, Taylor CM, Pigg J (1991) Upstream extirpation of four minnow species due to damming of a prairie stream. Trans Am Fish Soc 120:98–105

    Article  Google Scholar 

  37. Yamasaki YY, Nishida M, Suzuki T, Mukai T, Watanabe K (2015) Phylogeny, hybridizaion, and life history evolution of Rhinogobius gobies in Japan, inferred from multiple nuclear gene sequences. Mol Phylogenet Evol 90:20–33

    Article  CAS  PubMed  Google Scholar 

Download references

Acknowledgements

We are grateful to Shugo Kikuchi, Nobuyoshi Kotera, Yasutaka Hida, Satoshi Yano, Yuta Kudo and Yuka Fujiwara for help in the field, and Takaaki Shimizu, Masaki Shibuya, Jyun-ya Shibata, Tadao Kunihiro, Hidejiro Onishi and Hideki Hamaoka for various advice throughout the study. We also thank two anonymous reviewers for useful comments that improved the manuscript. This research was supported by a Grant-in-Aid for Scientific Research from Japan Society for the Promotion of Science (JSPS KAKENHI Grant number 19510236 to M. Takagi), a grant from the Water Resources Environment Center (to K. Omori and M. Inoue), and partly by the Environmental Research and Technology Development Fund (S9) of the Ministry of Environment, Japan. The survey in this study was conducted with the permission of Ehime Prefectural Government and complied with the current laws in Japan.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Mikio Inoue.

Additional information

Yoshifumi Sumizaki and Ryota Kawanishi equally contributed to this work.

About this article

Verify currency and authenticity via CrossMark

Cite this article

Sumizaki, Y., Kawanishi, R., Inoue, M. et al. Contrasting effects of dams with and without reservoirs on the population density of an amphidromous goby in southwestern Japan. Ichthyol Res 66, 319–329 (2019). https://doi.org/10.1007/s10228-018-00678-2

Download citation

Keywords

  • Amphidromy
  • Barrier effect
  • Landlocking
  • Migration
  • Rhinogobius