Intraspecific and Interspecific Variations in Habitat Usage by Hemimyzon formosanus and Sinogastromyzon nantaiensis in Riffle Areas of a Mountain Stream

Abstract

In this study, we explored the structure of two migratory fish species and their variations along the longitudinal gradient of a stream by examining abiotic factors such as spatial scale elements and environmental factors, and biotic factors such as competition. The studied fish species, i.e., Hemimyzon formosanus and Sinogastromyzon nantaiensis, were sampled from the middle and upper reaches of the Nanzihsian Stream using the electrofishing technique, and data were collected on the physical environment covered by the electric grid. Statistical analyses confirmed that the interspecific and the intraspecific hydraulic habitat environments of the two species were different, i.e., H. formosanus was distributed relatively upstream and S. nantaiensis was distributed relatively downstream. Therefore, the habitat use of the two benthic fish species was different. The Nanzihsian Stream experienced an extreme flood event in 2009. The densities of adult and juvenile H. formosanus and S. nantaiensis are significantly higher than those before the extreme flood event in 2021, so we suggest that the H. formosanus and S. nantaiensis populations have recovered to those of the stages before the extreme flood event. We provide an integrated approach for applying engineering and biology to the context of future projects involving river dredging after extreme floods.

1. Introduction

Given that each individual in a group must make use of the resources that are available and that these resources may be limited, individuals of the same species are forced to compete with each other in order to obtain a sufficient share of these resources. In cases where different species use the same resources, if two competing species are able to coexist in a specific variable environment, this is the result of niche differentiation [1]. Two main operating mechanisms underlie the distribution of a fish population: one mechanism involves biotic factors and the other mechanism involves abiotic factors [2,3]. Competition is one major biotic factor. In addition, elements of the physical habitat act as abiotic factors, such as water depth, flow velocity, and substrate type, and these elements have been identified and used as research variables that affect the composition and structure of the fish community in a stream [4]. The Rhinogobius sp. LD (the large dark type) inhabits the upper reaches of streams, while the Rhinogobius sp. CB (the cross-band type) inhabits the lower reaches, with no overlap in their respective niches. The distribution of stone sizes along a riverbed seems to affect the upstream migration and reproduction of these species [5]. Furthermore, laboratory experiments have shown that these species both engage in interspecific and intraspecific competitions. The two benthic fish species in this study also show upstream and downstream distributions. The difference is that the fish species in this study are herbivorous and non-territorial, and the population distribution overlaps in the relatively middle reaches of the Nanzihsian Stream. Hemimyzon formosanus is the most abundant fish species in the stream (maximum density is 178/m²), followed by Sinogastromyzon nantaiensis (maximum density is 67/m²), and they can coexist in one square meter (the smallest unit of measurement for this study). Because they have very similar feeding habits and inhabit the same visually qualitative habitats, they are potential competitors. Research conducted by the Endemic Organisms Research and Conservation Center of the Council of Agriculture at the Executive Yuan in Taiwan [6] integrated findings of research done on ecosystem health monitoring [7] and on marine biomonitoring [8]. They also used a stepwise decision-making framework to select target fish species to be used as indicator species for investigating environmental variation before and after hydraulic engineering projects are carried out, which may impact the life history of the species. Yang et al. [9] rated H. formosanus as a nationally vulnerable (NVU) freshwater fish species and S. nantaiensis as a nationally near-threatened freshwater fish species according to the recommended categories and criteria published by the International Union for Conservation of Nature (IUCN) in its Red List of Threatened Species™. Both species are endemic to Taiwan, and the latter is a conservation Class III species [10]. Furthermore, H. formosanus and Sinogastromyzon puliensis, as migratory species, have exhibited the same upstream migration behavior during the same season [11]. Fish from the Balitoridae family of fish inhabit gravel-bottomed areas where they cling to the gravel to scrape algae and eat small aquatic insects in order to survive [12]. One particular species in this family, i.e., H. formosanus, is a bottom-dwelling fish that inhabits fast-moving shallow water. The body size of an adult fish is less than 10 cm. Its streamlined and compressed body and its specialized ventral fins equip it with the ability to stick to rocks and contributes to its low drag coefficient that allows it to thrive in strong currents. H. formosanus is a herbivorous fish with a downward-sloping mouth that feeds mainly on algae [13]. S. nantaiensis also belongs to the Balitoridae family. Its body shape is very similar to that of H. formosanus, making it difficult for the general population to distinguish these species. In addition, they also have similar feeding habits. Finally, they inhabit the same qualitative features of the mesohabitat, in particular riffle habitats; therefore, they tend to coexist in the middle and upper reaches of a mountain river. H. formosanus and S. nantaiensis populations are indicator species for the success of riffle habitat restoration, but the factors that affect the distribution of fish populations that they both belong to and of their individual populations are rarely discussed. In addition, the hydraulic habitat environments inhabited by adults and juveniles of these two benthic species have rarely been discussed. Therefore, this study examined the hydraulic habitat environments inhabited by two benthic fish species, i.e., H. formosanus and S. nantaiensis, and the density for the two species from relative upstream to downstream of the Nanzihsian Stream.

To satisfy the different requirements of research on distinct ecological subjects, in particular the study of the interactions between individual organisms and their habitats, it is necessary to determine the most suitable spatial scale to use [14]. In the context of research on fish, on the one hand, the use of certain elements of the mesohabitat by two different benthic fish species, i.e., Crossostoma lacustre and Rhinogobius candidianus, has been found to overlap [15]. On the other hand, these species exhibit distinct patterns of use of microhabitat features, which has been thought to potentially reduce the pressure resulting from interspecific competition. In another study that assessed the impact of the physical features of a riverine ecosystem, two riffles were visually classified as being of the same type, but after quantitative measurements were taken, they were found to differ in terms of such characteristics as their hydraulic conditions and bed stability, which had an effect on the composition of the invertebrate populations inhabiting the surface of the substrate [16]. In aquatic mesohabitats, complex riffles have the greatest density of fish, with large numbers of small fish species; deep pool fronts have the greatest biomass, with large numbers of larger fish. The observation that different fish species require different mesohabitats, made in the context of an examination of segment scale environmental variables, has led to the conclusion that optimization of ecological design strategies for stream restoration may result in an increase in hydraulic habitat diversity [17]. Interactions among individuals of a species such as competition and predation usually occur at a small spatial scale and a short time scale [18].

In this study, we used the hierarchical framework to classify stream habitat systems, ranging in decreasing size from streams through segments, reaches, pool⁄riffle systems down to microhabitat proposed by Frissell et al. [19]. We explored whether the two benthic species mentioned above showed evidence of differences in intraspecific or interspecific hydraulic quantifying habitat use at all different spatial scales, even though they inhabited the same descriptive and qualitative habitat. Then, we assumed that the adults and juveniles of these species make use of different types of quantifying habitat environments at different spatial scales to reduce competition for spatial utilization. In addition, an extreme flood event occurred, in August 2009, that affected the Nanzihsian Stream, causing devastating damage to fish habitats. The H. formosanus and S. nantaiensis populations plummeted. The Taiwan government departments continued to carry out river dredging projects for many years after 2009 for the purpose of river protection. The stream was excavated into a single deep groove. A large number of small boulders and big boulders were removed from their habitats. Such human engineering interference has prolonged the period of time for H. formosanus and S. nantaiensis populations to recover to the stage before extreme flood events. This study attempted to add spatial elements in a hierarchical system of habitat classification and biological factors of habitat use and the distribution structure of these two benthic and landlocked migratory fish species in order to provide an integrated approach for applying engineering and biology to the context of future projects involving river dredging after an extreme flood event.

2. Materials and Methods

2.1. Sampling Sites

Sampling for this study was conducted in 2021, in the upper-middle section of the Nanzihsian Stream, a tributary of the Kaoping River in southern Taiwan. The Nanzihsian Stream is about 117 km long and 20–50 m wide (water surface), with a watershed area of about 741 square kilometers. The highest elevation in the watershed is 3750 m, its lowest point is 50 m, and the average slope of the stream bed is 1:15 [20]. The unaltered annual mean flow is 30 m³/s, as measured at the Nan-Feng Bridge hydrologic station (Water Resources Agency, Ministry of Economic Affairs, 2009). Streams in the sampling area meander through open valley floors without shade from woody riparian vegetation [13] (Figure 1).

The 5 sampling sites used in this study were located along river reaches which have been relatively undisturbed by human activity. In order, from the site located the farthest upstream to the site located farthest downstream, the sampling sites were the Mincyuan Bridge (altitude 480 m), the Dubuluan Bridge (altitude 420 m), the Sianshan Bridge (altitude 400 m), the Dawulong Bridge (altitude 380 m), and the Siaolin Bridge (altitude 360 m). The distances between each site and the following site along the longitudinal gradient of the river are 5.6 km between the Mincyuan Bridge and the Dubuluan Bridge, 3.4 km between the Dubuluan Bridge and the Sianshan Bridge, 3.6 km between the Sianshan Bridge and the Dawulong Bridge, and 3.8 km between the Dawulong Bridge and the Siaolin Bridge. The straight-line distance between the Mincyuan Bridge and the Siaolin Bridge is 12 km. Finally, it is worth mentioning that the section of the stream above the Sianshan Bridge has been designated a stream-type national wetland and wildlife reserve [21].

2.2. Samples and Sampling Methods

We collected samples of two benthic fish species, H. formosanus and S. nantaiensis, between January and April 2021 (which is the dry season in the hydrological year) with the help of prepositioned areal electrofishing devices (PAEDs). We considered the safety of our staff and conducted fish sampling in the dry season. This sampling method reduces the risk of frightening the fish [22], and it allows researchers to study fish densities in particular types of microhabitats. Therefore, the fish–microhabitat relationship can be described in terms of the interaction between interspecific competition and the physical microhabitat [13].

2.3. Habitat Environment Investigation

The steps we followed to collect data involved, first, finding a deep pool system at each sampling site, since there are riffle systems at both the upstream and the downstream sections of such a system. The length of the riffle systems which were identified in this study varied between approximately 30 m and 50 m. Next, fish samples were randomly collected from these riffle systems following the direction from downstream sections to upstream sections. The area affected by each PAED electrode (i.e., the sampling site) was 1 m × 1 m. After collecting the fish, we assessed the environmental factors such as flow velocity, water depth, and substrate type. The flow velocity and water depth had an average value of 9 points, as measured by the electrode. We used a propeller current meter to measure the flow velocity at 0.4 times the water depth below the water surface, and we used an iron ruler to vertically insert it into the water to the bottom of the stream. The approach taken to codify the components of the substrate was the surface-visual method proposed by Platts et al. [23]. Then, we used a transparent acrylic sheet marked with a 1 m × 1 m grid consisting of 100 modules and placed it at the sampling site to calculate the distribution of various particle sizes in the substrate. The code applied to the classification of the various types of substrate particles was a modification of the system proposed by Bovee and Milhous [24]. We also modified the formula for calculating the average value of the code proposed by Statzner et al. [25] (see

Table 1. Substrate type and code (modified from Bovee and Milhous, 1978 [24]).

Substrate Code	Substrate Type	Substrate Particle Size Range	Percentage
1.5	Sand	<2 mm	α1
3.5	Gravel	2–64 mm	α2
5.5	Pebble	64–256 mm	α3
7	Small boulder	256–512 mm	α4
8	Big boulder	>512 mm	α5

Note: Average value of substrate code = α1 × 1.5 + α2 × 3.5 + α3 × 5.5 + α4 × 7 + α5 × 8 (modified from Statzner et al., 1988 [25]).

).

2.4. Fish Species Identification and Distinguishing Maturity

We identified fish species visually and measured their body size before releasing them back into the stream. The posterior edge of the pelvic fin of H. formosanus is completely separated and completely healed for S. nantaiensis. We referred to the research on H. formosanus and S. nantaiensis species dissected by Yang et al. [26], who calculated the 50% mature body length (ML50) based on a regression analysis of the body length and the gonad development stage of H. formosanus and S. nantaiensis. According to their results, the 50% mature body length for the population of H. formosanus was 4.7 cm, and for S. nantaiensis it was 4.9 cm. In another study, the body length of an adult H. formosanus was measured at more than 5 cm, and that of a juvenile fish was less than 4 cm [27]. In the current study, we decided whether the fish samples were adults or juveniles based on the total length data. The body length of the H. formosanus was calculated at 5 cm or longer for adults and at 3.9 cm or shorter for juveniles, whereas the body length of the S. nantaiensis was at least 5.1 cm for adults and at the most 4.0 cm for juveniles.

2.5. Riffle in Different Spatial Scales

As mentioned above, we sampled fish from the riffle systems located both upstream and downstream of a deep pool, which we designated the pool/riffle system spatial scale. Next, we considered both the upstream and the downstream riffle systems together, which we designated the reach spatial scale. The sampling sites at the Mincyuan, Dubuluan, Dawulong, and Siaolin Bridges all had two types of riffle systems, named Riffle I and Riffle II, but there was only one type of riffle system at the Sianshan Bridge sampling sites. In summary, there were 9 riffle systems at the pool/riffle system spatial scale and 4 riffles (combining Riffles I and II) at the reach spatial scale. In addition, we identified the stream spatial scale as encompassing all the sampling sites. Finally, the PAED electrode was designated the microhabitat spatial scale (Figure 2).

2.6. Analysis of Quantitative Habitat Environment Difference

At the microhabitat spatial scale, we used data on environmental factors from all sampling sites to produce habitat suitability index curves in order to determine whether the habitat use of the two target fish species was the same.

For each riffle at the pool/riffle system spatial scale, we used the independent samples t-test procedure (SPSS version 17.0) to examine whether these species differed in terms of intraspecific (at different maturity) and interspecific hydraulic quantifying habitat use at the same maturity and at different maturity (the requirement for passing the t-test is p ≤ 0.05). The same analytical procedure that was followed at the pool/riffle system spatial scale was applied at the reach and stream spatial scales. The statistical analyses were performed with adult and juvenile fish of the H. formosanus and S. nantaiensis species as the comparison groups, which gave 4 groups with comparisons based on the developmental stages of the fish.

At the stream spatial scale, the independent sample t-test was also used to compare the average value of the physical parameters (water depth, flow velocity, and substrate type) and the fish density between sampling sites.

2.7. Previous Sampling Data

An extreme flood event (Typhoon Morakot) occurred, in August 2009, that affected the Nanzihsian Stream. We conducted fish sampling and habitat environment surveys before (December 2008 to March 2009) and after the extreme flood event (from December 2011 to March 2012). The sampling data in 2008–2009 can be regarded as the samples of the habitat environment for the H. formosanus and S. nantaiensis populations before the extreme flood event. The sampling data in 2011–2012 can be regarded as the habitat environment for the H. formosanus and S. nantaiensis populations which experienced disturbances from water conservancy dredging projects and have not yet recovered after the extreme flood event. The sampling data of this study in 2021 can be regarded as the samples of the habitat environment for the H. formosanus and S. nantaiensis populations which have recovered after the extreme flood event.

3. Results

This study placed a total of 153 PAED electrodes at five sampling sites to collect samples. The coordinates, the number of fish species, the number of sampling sites, and the sampling month for each sampling site are shown in

Table 2. The coordinates, the numerical abundance, the number of sampling sites, and the sampling month for each sampling site, in 2021.

Sampling Sites	Coordinates	Hemimyzon formosanus (Number)	Sinogastromyzon nantaiensis (Number)	Sampling Month in 2021	Site (Number)
Mincyuan Bridge	120.69343, 23.24787	1333	13	February, March	33
Dubuluan Bridge	120.68024, 23.20863	1149	165	February, April	33
Sianshan Bridge	120.66514, 23.20257	734	400	March, April	25
Dawulong Bridge	120.65121, 23.18734	562	357	January, March	32
Siaolin Bridge	120.63575, 23.15787	492	553	February, March	32

.

At the microhabitat spatial scale, fish density was examined in relation to a number of environmental factors. The density of both the adult and the juvenile fish of H. formosanus was 3.22 individuals/m² and 14.39 individuals/m², respectively, at a flow velocity of 1.4–2.2 m/s, which was above average. In addition, the maximum fish densities for both groups were observed at a flow velocity of 1.8–2.0 m/s and at a water depth of 40–45 cm (see Figure 3a and Figure 4a). Regarding the substrate type, the average density of H. formosanus adults was 3.22 number/m², and the maximum density corresponded to a substrate code of 7.5–8.0, whereas for the juveniles of this species, the average density was 14.39 individuals/m² and the maximum density corresponded to a substrate code of 6.5–7.0 (see Figure 5a). In the case of S. nantaiensis, the average density of adult fish in relation to the flow velocity and the substrate type was 2.44 individuals/m², with the maximum density corresponding to a flow velocity of 1.8–2.0 m/s and a substrate code of 7.5–8.0. For the juveniles of this species, the average density was 5.19 individuals/m², with the maximum density corresponding to a flow velocity of 1.2–1.4 m/s and a substrate code of 7.0–7.5 (see Figure 3b and Figure 5b). Finally, the average densities of the S. nantaiensis adults and juveniles in relation to the water depth were 2.40 individuals/m² and 5.24 individuals/m², respectively, with the maximum density corresponding to a water depth of 35–40 cm (see Figure 4b).

At the riffle system spatial scale, first, the analysis focused on comparing the environmental factors and the developmental stage of the H. formosanus and S. nantaiensis species fish inhabiting the riffles located at each sampling site (see

Table 3. Differences in habitat environmental factors according to the developmental stage of Hemimyzon formosanus and Sinogastromyzon nantaiensis in Riffle I and Riffle II at each sampling site.

		Habitat Environment Sampling Sites	Riffle I				Riffle II
Fish Species and Maturity			Velocity (m/s)	Depth (cm)	Substrate Code	Number	Velocity (m/s)	Depth (cm)	Substrate Code	Number
Mincyuan Bridge	Hemimyzon formosanus	Adult	1.28	42.8	7.299	78	1.65	34.0	7.343	57
		Juvenile	1.26	40.9	7.156	287	1.63	33.4	7.274	366
	Sinogastromyzon nantaiensis	Adult	1.03	37.9	7.708	3	-	-	-	0
		Juvenile	1.06	34.6	7.650	1	1.86	43.1	7.300	1
Dubuluan Bridge	Hemimyzon formosanus	Adult	1.43	28.4	6.421	27	1.63	35.6	7.563	202
		Juvenile	1.41	28.4	6.396	181	1.73	43.1	7.670	280
	Sinogastromyzon nantaiensis	Adult	1.45	24.4	6.708	2	1.62	40.3	7.644	61
		Juvenile	1.52	30.5	6.586	5	1.60	50.6	7.814	26
Sianshan Bridge	Hemimyzon formosanus	Adult	1.48	34.2	7.226	54	-	-	-	-
		Juvenile	1.41	32.5	7.147	431	-	-	-	-
	Sinogastromyzon nantaiensis	Adult	1.53	34.0	7.325	90	-	-	-	-
		Juvenile	1.55	35.3	7.308	155	-	-	-	-
Dawulong Bridge	Hemimyzon formosanus	Adult	1.46	39.1	7.344	11	1.54	30.7	7.553	29
		Juvenile	1.34	28.4	6.795	221	1.49	31.0	7.516	140
	Sinogastromyzon nantaiensis	Adult	1.60	36.9	7.409	10	1.58	33.8	7.746	67
		Juvenile	1.52	32.7	7.346	65	1.50	34.2	7.623	66
Siaolin Bridge	Hemimyzon formosanus	Adult	1.25	32.0	7.325	12	1.77	35.1	7.477	21
		Juvenile	1.29	33.0	7.434	114	1.47	33.3	7.320	177
	Sinogastromyzon nantaiensis	Adult	1.33	32.9	7.528	38	1.51	36.7	7.615	25
		Juvenile	1.28	34.9	7.419	149	1.61	34.7	7.452	157
Significance of mean difference (t-test p-value)			0.482	0.468	0.050	-	0.665	0.078	0.280	-
			-	-	-		-	-	-
			0.716	0.983	0.738		0.002	0.000	0.014
			-	-	-		0.811	0.010	0.005
			0.095	0.155	0.186		-	-	-
			0.434	0.237	0.549
			0.180	0.000	0.019		0.382	0.837	0.692
			0.320	0.073	0.544		0.099	0.752	0.041
			0.425	0.659	0.099		0.000	0.075	0.182
			0.044	0.110	0.004		0.075	0.006	0.001

). Another comparison was made between the developmental stages reached by fish of both species (see

Table 4. Differences in habitat environmental factors between Hemimyzon formosanus and Sinogastromyzon nantaiensis both at the same and at different developmental stages in Riffle I and Riffle II at each sampling site (t-test p value).

Habitat Environment			Riffle I						Riffle II
			Velocity		Depth		Substrate Code		Velocity		Depth		Substrate Code
Fish Species and Maturity			Hemimyzon formosanus
			Adult	Juvenile	Adult	Juvenile	Adult	Juvenile	Adult	Juvenile	Adult	Juvenile	Adult	Juvenile
Dubuluan Bridge	Sinogastro-myzon nantaiensis	Adult	-	-	-	-	-	-	0.697	0.014	0.034	0.297	0.203	0.678
		Juvenile	-	-	-	-	-	-	0.584	0.044	0.000	0.053	0.000	0.000
Sianshan Bridge		Adult	0.204	0.000	0.931	0.101	0.106	0.000	-	-	-	-	-	-
		Juvenile	0.057	0.000	0.366	0.000	0.165	0.000	-	-	-	-	-	-
Dawulong Bridge		Adult	0.154	0.000	0.630	0.083	0.784	0.217	0.560	0.016	0.010	0.000	0.037	0.000
		Juvenile	0.460	0.000	0.052	0.001	0.992	0.000	0.581	0.709	0.034	0.004	0.455	0.093
Siaolin Bridge		Adult	0.076	0.084	0.617	0.973	0.006	0.016	0.003	0.563	0.135	0.000	0.194	0.000
		Juvenile	0.457	0.813	0.158	0.036	0.163	0.595	0.043	0.000	0.588	0.002	0.757	0.006

). The number of individuals of S. nantaiensis captured in Riffles I and II at the Mincyuan Bridge and in Riffle I at the Dubuluan Bridge was insufficient, so the analysis of habitat differences could not be carried out. Finally, a comparison of the environmental factors and the density of fish at different developmental stages was carried out, and differences between the riffles were examined at each sampling site (see

Table 5. Comparison of habitat environmental factors and fish density according to the species and developmental stage of the fish and according to the riffle type at each sampling site.

	Habitat Environment and Fish Density	Velocity (m/s)	Depth (cm)	Substrate Code	Hemimyzon formosanus (number/m2)		Sinogastromyzon nantaiensis (number/m2)		Sampling Site Number
Riffle Type					Adult	Juvenile	Adult	Juvenile
Mincyuan Bridge	Riffle I	1.15	29.6	6.781	3.5	13.5	0.2	0.07	15
	Riffle II	1.59	40.7	7.278	6.3	40.7	0.0	0.1	9
Dubuluan Bridge	Riffle I	1.35	29.6	6.350	4.7	18.1	0.2	0.5	10
	Riffle II	1.63	40.5	7.630	10.5	14.0	3.1	1.4	19
Sianshan Bridge	Riffle I	1.42	32.7	7.177	2.2	17.2	3.6	6.2	25
Dawulong Bridge	Riffle I	1.30	30.6	6.880	0.7	13.8	0.6	4.1	16
	Riffle II	1.52	33.4	7.510	1.8	8.8	4.2	4.1	16
Siaolin Bridge	Riffle I	1.27	33.6	7.440	0.8	7.6	2.5	9.9	15
	Riffle II	1.41	33.8	7.350	1.2	10.4	1.5	9.2	17
Significance of Mean Difference (t-test p-value)		0.001	0.002	0.050	0.086	0.002	-	-	-
		0.037	0.048	0.000	0.065	0.296	0.000	0.241	-
		-	-	-	-	-	-	-	-
		0.068	0.429	0.009	0.017	0.049	0.027	0.969	-
		0.185	0.930	0.553	0.432	0.184	0.141	0.778	-

).

At the reach spatial scale, we repeated the analysis that was performed on the data collected at the riffle system spatial scale (see Figure 6a–d,

Table 6. Differences in habitat environmental factors according to the developmental stage of Hemimyzon formosanus and Sinogastromyzon nantaiensis in the riffles at each sampling site at the reach spatial scale.

	Habitat Environment and Fish Species	Hemimyzon formosanus				Sinogastromyzon nantaiensis
Sampling Sites and Fish Maturity		Velocity (m/s)	Depth (cm)	Substrate Code	Number	Velocity (m/s)	Depth (cm)	Substrate Code	Number
Mincyuan Bridge	Adult	1.43	37.4	7.284	138	1.03	37.9	7.708	3
	Juvenile	1.46	37.1	7.178	287	1.46	38.8	7.475	2
Dubuluan Bridge	Adult	1.61	34.8	7.428	229	1.61	39.8	7.614	63
	Juvenile	1.61	37.3	7.170	461	1.58	47.3	7.616	31
Dawulong Bridge	Adult	1.52	33.0	7.496	40	1.59	34.2	7.703	77
	Juvenile	1.40	29.4	7.075	361	1.51	33.4	7.486	131
Siaolin Bridge	Adult	1.58	34.0	7.422	33	1.40	34.5	7.563	63
	Juvenile	1.40	33.2	7.364	291	1.45	34.8	7.436	306
Significance of mean difference (t-test p-value)		0.367	0.689	0.032	-	-	-	-	-
		0.858	0.036	0.000		0.722	0.043	0.986
		0.016	0.011	0.000		0.402	0.372	0.000
		0.008	0.451	0.461		0.112	0.676	0.000

and

Table 7. Differences in habitat environmental factors between Hemimyzon formosanus and Sinogastromyzon nantaiensis both at the same and at different developmental stages in the riffles at each sampling site at the reach spatial scale (t-test p-value).

Habitat Environment			Velocity		Depth		Substrate Code
Fish Species and Maturity			Hemimyzon formosanus
			Adult	Juvenile	Adult	Juvenile	Adult	Juvenile
Dubuluan Bridge	Sinogastromyzon nantaiensis	Adult	0.980	0.936	0.032	0.271	0.007	0.000
		Juvenile	0.632	0.726	0.000	0.001	0.064	0.000
Dawulong Bridge		Adult	0.252	0.000	0.417	0.000	0.017	0.000
		Juvenile	0.877	0.000	0.785	0.000	0.909	0.000
Siaolin Bridge		Adult	0.011	0.984	0.661	0.109	0.021	0.000
		Juvenile	0.052	0.041	0.417	0.001	0.779	0.017

). At the reach scale, the analysis of intraspecific and interspecific habitat differences was not performed because there was only one riffle at the Sianshan Bridge.

By combining the data collected at both the riffle system and the reach spatial scales, it was possible to compare both fish species in terms of their preferences for environmental features (see Table 3, Table 4, Table 6 and Table 7). This method allowed us to create groups combining various fish based on species and developmental stages. One possible group consisted of adults and juveniles of the H. formosanus species, which were compared based on intraspecific habitat differences. At the Mincyuan Bridge Riffle I, the ratio of the number of significant differences to the number of groups compared (hereafter called the significant difference-to-comparison group ratio) was 1/1 (see

Table 8. Intraspecific and interspecific habitat differences at the riffle system scale and at the reach spatial scale.

			Sampling Sites	Mincyuan Bridge	Dubuluan Bridge	Sianshan Bridge	Dawulong Bridge	Siaolin Bridge	Intraspecific Habitat Differences Frequency Divided by Total		Interspecific Habitat Differences (Frequency/Total)
Riffle Type and Coefficient of Variation									Hemimyzon formosanus	Sinogastromyzon nantaiensis
the riffle system spatial scale	Riffle I	Intraspecific habitat differences	Hemimyzon formosanus	1/1	0/1	0/1	1/1	0/1	4/9 (44%)	4/6 (67%)	21/24 (87.5%)
			Sinogastromyzon nantaiensis	-	-	0/1	0/1	1/1
		interspecific habitat differences		-	-	3/4	3/4	3/4
	Velocity	Coefficient of variation		0.2228	0.1888	0.2076	0.2474	0.1251
	Depth			0.2261	0.2073	0.2474	0.3773	0.2382
	Substrate Code			0.0883	0.0630	0.0603	0.1109	0.0328
	Riffle II	Intraspecific habitat differences	Hemimyzon formosanus	0/1	1/1	-	0/1	1/1
			Sinogastromyzon nantaiensis	-	1/1		1/1	1/1
		Interspecific habitat differences		-	4/4		4/4	4/4
	Velocity	Coefficient of variation		0.1946	0.2679		0.2206	0.2437
	Depth			0.2218	0.3859		0.2391	0.1381
	Substrate Code			0.0721	0.0927		0.0619	0.0797
the reach spatial scale	Riffle	Intraspecific habitat differences	Hemimyzon formosanus	1/1	1/1	-	1/1	1/1	4/4 (100%)	3/3 (100%)	10/12 (83.3%)
			Sinogastromyzon nantaiensis	-	1/1		1/1	1/1
		Interspecific habitat differences		-	4/4		2/4	4/4
	Velocity	Coefficient of variation		0.252	0.255		0.243	0.206
	Depth			0.254	0.382		0.308	0.188
	Substrate Code			0.0842	0.113		0.097	0.061

Note: At the riffle system spatial scale, the values marked in red mean that Riffle I or Riffle II had larger coefficient of habitat environment variation at the same sampling sites.

). On the one hand, the denominator of this ratio indicates that only one group of H. formosanus adults and one group of juveniles were compared. The numerator, on the other hand, indicates that only one group showed significant differences in habitat use. In other words, the adults and juveniles of the H. formosanus, which were sampled at the Mincyuan Bridge Riffle I, were found to inhabit distinct ecological environments. The intraspecific differences in habitat use of H. formosanus were not significant at the Mincyuan Bridge Riffle II, giving a ratio of 0/1.

In total, nine comparison groups of H. formosanus samples from all five sampling sites were examined to identify intraspecific differences in habitat use. Four of these groups showed significant differences at the riffle system spatial scale, for a significant difference-to-comparison group ratio of 4/9, or 44% (see Table 8). The S. nantaiensis was omitted from the analysis of intraspecific differences in habitat use because of the insufficient samples collected at the Mincyuan Bridge and Dubuluan Bridge sampling sites.

Regarding interspecific differences in habitat use, members of the H. formosanus and S. nantaiensis species at both the same and different developmental stages were divided into four groups. For fish sampled at Riffle I at the Sianshan Bridge sites, the significant difference-to-comparison group ratio was 3/4 (see Table 8), indicating that three of the four groups exhibited significant differences in habitat use. In total, twenty-four comparison groups of fish from all five sampling sites were analyzed, with significant interspecific differences found for twenty-one of them. When all groups were considered, the significant difference-to-comparison group ratio reached as high as 87.5% at the riffle system spatial scale.

The results of the analysis of differences in the interspecific and intraspecific distribution structure of the two fish species at the stream spatial scale are shown in

Table 9. Differences in interspecific and intraspecific distribution structure for the two benthic fish species at the stream spatial scale.

	Fish Species, Habitat Environment	Hemimyzon formosanus (number/m2)		Sinogastromyzon nantaiensis (number/m2)		Velocity (m/s)	Depth (cm)	Substrate Code
Sampling Sites		Adult	Juvenile	Adult	Juvenile
Mincyuan Bridge		4.333	20.79	0.09	0.06	1.295	33.2	7.029
Dubuluan Bridge		6.939	13.97	1.91	0.094	1.481	36.8	7.142
Sianshan Bridge		2.160	17.24	3.60	6.20	1.420	32.7	7.177
Dawulong Bridge		1.250	11.28	2.406	4.093	1.407	32.0	7.197
Siaolin Bridge		1.031	9.09	1.967	9.562	1.349	33.7	7.397
Significance of Mean Difference (t-test p-value, versus the Bridge to next (below))		0.181 0.014 0.047 0.549	0.083 0.209 0.005 0.191	0.000 0.028 0.215 0.614 (0.033)	0.008 0.000 0.131 0.000	0.036 0.512 0.875 0.462	0.226 0.196 0.781 0.427	0.529 0.831 0.899 0.180

and Figure 7a–d. In addition, using a statistical analysis t-test and the big data of 155 electric grids, which contained 494 adults and 2220 juveniles of H. formosanus and 296 adults and 625 juveniles of S. nantaiensis, and we confirmed the differences in both interspecific and intraspecific habitat use of the two benthic fish species at the stream spatial scale. In terms of interspecific habitat preference comparisons, adults of the H. formosanus prefer habitats with significantly faster flow velocity than juveniles of S. nantaiensis(p value = 0.034), and in terms of intraspecific habitat preference comparisons, adults of H. formosanus prefer habitats with significantly larger substrate particle size than juveniles (p value = 0.000). The substrate particle size of the preferred habitat of S. nantaiensis adults was the largest significantly (p value = 0.000, vs. by the latter), followed by juveniles of S. nantaiensis, adults of H. formosanus, and juveniles of H. formosanus that were the smallest significantly (p value = 0.000, vs. by the former) (See Figure 8). The flow velocity of the preferred habitat of adults of H. formosanus was the fastest, followed by adults of S. nantaiensis, juveniles of S. nantaiensis (p value = 0.010, vs. by the former), and juveniles of H. formosanus was the slowest significantly (p value = 0.010, vs. by the former). We again confirmed that the interspecific and intraspecific habitat use of the two benthic fish species was different at the stream spatial scale.

4. Discussion

On the one hand, the principle of mutual exclusivity holds that competing species cannot live in close proximity indefinitely. On the other hand, they may successfully coexist if the ecological niche that each species inhabits is sufficiently differentiated from all the others. Because interspecific competition is common [1], we assumed that it existed between the H. formosanus and S. nantaiensis species. In addition, we also assumed that intraspecific competition of spatial occupation existed between fish at different developmental stages.

By examining different spatial scales and different environmental factors, we were able to show the existence of a difference in interspecific and intraspecific habitat use for these two fish species despite their strong visual similarities in body size.

Density is a quantitative measurement of the number of individuals per unit area and it requires a large amount of data for an affirmative conclusion. Density can be useful to detect preference [13]. H. formosanus and S. nantaiensis have been reported to be widely distributed species with large population sizes. The density of H. formosanus was maximum in habitats with a flow velocity of 0.8–1.0 m/s or water depth of 40–70 cm [28]. However, there is no literature discussing the habitat environment differences between adults and juveniles of H. formosanus and S. nantaiensis. Literature on the habitats of S. nantaiensis populations is quite sparse and lacks quantitative data. Our research results at the microhabitat spatial scale filled this gap. Adults of the H. formosanus were more numerous in habitats with larger substrate particle size (peak density = 7.5–8.0) than juveniles (peak density = 6.5–7.0). The same result was found for S. nantaiensis adults (peak density = 7.5–8.0 versus peak density = 7.0–7.5 for juveniles). The main difference between the adults of both species was the flow velocity of their habitats. H. formosanus adults were found to inhabit faster flowing water (peak density = 1.8–2.0 m/s) than the S. nantaiensis adults (peak density = 1.6–1.8 m/s). The same difference was observed between the juveniles of both species, with H. formosanus inhabiting faster flowing water (peak density = 1.8–2.0 m/s) than S. nantaiensis (peak density = 1.2–1.4 m/s). The H. formosanus adults and the S. nantaiensis juveniles differed primarily with regard to two features of their habitats: flow velocity and substrate type. The former inhabited faster flowing water and larger substrate particle size. The environmental factor that accounted for the differences between the H. formosanus juveniles and the S. nantaiensis adults was substrate type, with the former inhabiting smaller substrate particle size. In summary, the interspecific and intraspecific habitat use of the two benthic fish species are different at the microhabitat spatial scale. (see Figure 3, Figure 4 and Figure 5)

Habitat use was also distinguished at the riffle system spatial scale and the reach spatial scale. The proportions of fish displaying statistically significant intraspecific differences in relation to these spatial scales were 44% and 100%, respectively, for H. formosanus and 67% and 100%, respectively, for S. nantaiensis. The corresponding proportions with regard to interspecific differences were 87.5% and 83.3%, respectively. (see Table 6, Table 7 and Table 8)

At the riffle system spatial scale, it was easier to assess the statistically significant differences in habitat use between comparison groups when the variation of the environmental factors present at the riffles at each sample site was high. Relatively large coefficients of variation for the environmental factors at Riffle II at both the Dubuluan Bridge and the Siaolin Bridge were obtained, and statistically significant differences in habitat use were found for all six comparison groups (see Table 8). For seven of the nine riffles that we studied at the riffle system spatial scale, it was observed that the larger the variation of the environmental factors, the larger the number of comparison groups that exhibited significant differences in habitat use. Conversely, the smaller the variation of the environmental factors, the smaller the number of groups that exhibited significant differences in habitat use (see Table 8).

At the reach spatial scale, statistically significant differences in habitat use were found for eighteen of the nineteen comparison groups for the five sampling sites, with a statistically significant difference-to-comparison group ratio as high as 95.8%, which is very reasonable. When the term riffle is used to discuss the reach spatial scale, it covers a larger area than Riffles I and II at the riffle system spatial scale, which means greater environmental variation and a more diverse environment. Therefore, the results of this study proved that the H. formosanus and S. nantaiensis fish species’ interspecific and intraspecific habitat use is different at the riffle system and reach spatial scales.

At the stream spatial scale, the distribution curve of the density of H. formosanus adults shows a peak at the Dubuluan Bridge, which is located in a relatively upstream part of the river. The density decreases slightly upstream toward the Mincyuan Bridge and more significantly further downstream, with the lowest density recorded at the Siaolin Bridge. The density of H. formosanus adults at the relatively upstream Dubuluan Bridge is significantly higher than that at the relatively downstream Sianshan Bridge (p-value = 0.014); the density of H. formosanus adults at the relatively upstream Sianshan Bridge is significantly higher than that at the relatively downstream Dawulong Bridge (p-value = 0.047). In addition, the distribution curve of the density of the juveniles of the same species follows a smoothly descending trend line, with the highest density at the Mincyuan Bridge and the lowest density at the Siaolin Bridge. The density of H. formosanus juveniles at the relatively upstream Sianshan Bridge is significantly higher than that at the relatively downstream Dawulong Bridge (p-value = 0.005). The distribution boundary for the data on the S. nantaiensis population suggests that adult fish have colonized the upstream portion of the reach at the Dubuluan Bridge. Because the density of the juveniles of the same species is statistically significantly lower than that of the adults at the Dubuluan Bridge Riffle II (p-value = 0.033), we assumed that the juveniles may have colonized the upstream portion of the reach at the Sianshan Bridge. The distribution curve of the density of S. nantaiensis adults shows a peak at the Sianshan Bridge, located in the middle course of the river, with the density decreasing both upstream and downstream. On the one hand, the density of S. nantaiensis adults at the relatively upstream Dubuluan Bridge is significantly lower than that at the relatively downstream Sianshan Bridge (p-value = 0.028); the density of S. nantaiensis adults at the relatively upstream Sianshan Bridge is significantly higher than that at the relatively downstream Siaolin Bridge (p-value = 0.033). On the other hand, the distribution curve of the density of the S. nantaiensis juveniles follows a rising trend line which peaks at the Siaolin Bridge, with a significant increase in density from the upstream part of the river to the sections farther downstream. The density of S. nantaiensis juveniles at the relatively upstream Dubuluan Bridge is significantly lower than that at the relatively downstream Sianshan Bridge (p-value = 0.000). Although it has never been proven that S. nantaiensis migrates upstream, our results for data at the stream spatial scale show that adults of the species were found in larger numbers than juveniles in sections of the stream farther upstream. Overall, our results suggest that the adults of both the S. nantaiensis and the H. formosanus species tend to migrate upstream, which suggests that there is a difference in intraspecific distribution structure for each species at the stream spatial scale.

The density of H. formosanus adults was found to be statistically significantly higher than that of S. nantaiensis adults in the stretch of river from the Mincyuan Bridge to the Dubuluan Bridge (p-value = 0.000 at the Mincyuan Bridge and 0.010 at the Dubuluan Bridge). However, the opposite was true from the Sianshan Bridge to the Dawulong Bridge, where S. nantaiensis adults were denser, with a statistically significant difference at the Siaolin Bridge (p-value = 0.041). Between the Mincyuan and Siaolin Bridges, the densities of adults of both species reached their peaks which were staggered, with H. formosanus found near the Dubuluan Bridge and S. nantaiensis more numerous near the Sianshan Bridge. The density of S. nantaiensis adults is significantly higher than that of H. formosanus adults at the Sianshan Bridge (p-value = 0.050); the density of S. nantaiensis adults is significantly lower than that of H. formosanus adults at the Dubuluan Bridge (p-value = 0.010). The distributions of the density of juvenile fish of these species show opposite tendencies from sections farther upstream moving downstream, with S. nantaiensis following a rising trend line and H. formosanus following a decreasing trend line. S. nantaiensis juveniles reached their highest density at the Siaolin Bridge, which coincidentally is where the H. formosanus reached their lowest density. The density of S. nantaiensis juveniles is significantly lower than that of H. formosanus juveniles at the relatively upstream Mincyuan Bridge and Dubuluan Bridge (p-value = 0.000 at the Mincyuan Bridge and Dubuluan Bridge). The peak density of fish of each species at the same developmental stage was observed at different sampling sites, which suggests that there was a difference in interspecific distribution structure between these species at the stream spatial scale.

The results of the fish sampling survey conducted in the Nanzihsian Stream, in 2018, by the Kaohsiung City Government [29] were roughly comparable to ours. They found that H. formosanus tended to inhabit sections of the river at higher altitudes and with lower water temperatures than S. nantaiensis; H. formosanus is distributed relatively upstream of the Nanzihsian Stream, and S. nantaiensis is distributed relatively downstream of the stream. Our results showed that H. formosanus adults were more capable of living in habitats with a faster water flow than S. nantaiensis adults at different spatial scales. At the microhabitat spatial scale, the H. formosanus density peaked in areas with a flow velocity of 1.8–2.0 m/s, while the highest flow velocity inhabited by S. nantaiensis was 1.6–1.8 m/s. At the riffle system spatial scale, the flow velocity in the area near the Siaolin Bridge where H. formosanus adults were found was 1.765 m/s, which was statistically significantly faster than the flow velocity of 1.506 m/s for S. nantaiensis (p-value = 0.003, see Table 4). At the reach spatial scale, the flow velocity in the area near the Siaolin Bridge where H. formosanus adults were found was 1.58 m/s, which was significantly faster than the flow velocity of 1.40 m/s for S. nantaiensis (p-value = 0.011, see Table 7).

We also examined the body size of the two fish species. Although they share visual similarities, H. formosanus is more slender than S. nantaiensis, and its body shape makes it less susceptible to strong water currents. Therefore, H. formosanus is more capable of colonizing the upper reaches of the stream than S. nantaiensis. Abiotic factors have been shown to be more important determinants of the structure of fish communities at large spatial scales [18]. Consistent with this finding is our suggestion that the interactions among water temperature and flow velocity were the abiotic factors both of the structure of the fish community consisting of the H. formosanus and S. nantaiensis species and of its variation along the longitudinal gradient of the Nanzihsian Stream. Based on the above statistical verification and our other research data, we also confirmed the distribution ranges of H. formosanus and S. nantaiensis species. H. formosanus is distributed relatively upstream of the Nanzihsian Stream, and S. nantaiensis is distributed relatively downstream of the stream, and there is an overlap between them.

In an examination of differences in microhabitat preferences, Lee and Suen [13] indirectly showed that Hemimyzon formosanus, Rhinogobius nantaiensis, Onychosoma alticorpus, and Acrossocheilus paradoxus inhabited distinct ecological niches. This suggests the existence of interspecific competition in fish communities at the level of the riffle ecosystem. In addition to the similarities between their appearances and feeding habits, H. formosanus and S. nantaiensis also inhabit the same descriptive and qualitative geomorphological features of the mesohabitat, such as rapids and riffles where the flow velocity is high and the substrate particles are large. This study suggests that even with the same preferences for food and the same qualitative habitat characteristics, the two fish species tended to inhabit areas with different quantitative features at different spatial scales. The different habitat environments of these species may reduce interspecific competition for spatial resources.

Begon et al. [1] mentioned that sometimes two individuals involved in a situation of intraspecific competition do not have the opportunity to interact directly with each other. As long as some of the limited resources that are available are used by one individual, the other individual has fewer resources to use, which adversely affects their life in a variety of ways, including the effectiveness of survival, growth, and reproductive mechanisms. This study suggests that the adults and juveniles of both the H. formosanus and the S. nantaiensis species inhabited areas with different quantitative features at different spatial scales; thus, this demonstrates that, at different spatial scales, there is a difference in intraspecific habitat use between fish at different developmental stages, which may reduce intraspecific competition for spatial resources.

Studies on typhoon disturbances in mountain streams of Taiwan have indicated that typhoons cause significant modifications of stream habitat and fish populations, but that fish populations recover rapidly after typhoons [30,31]. Recolonization by individuals from nearby colonies or survivors from refuge areas suggests that fish populations may return to pre-typhoon condition in <2 years [30]. This response indicates that mountain stream fish in Taiwan have adapted to episodic flood events. Although flash floods reduce fish populations, their rapid recovery ensures that interspecific competition soon exerts pressure upon the fish community assemblage [13]. The Nanzihsian Stream experienced an extreme flood event in 2009, and fish habitats suffered extensive damage from landslides. In 2012–2013, due to the disturbance of substrate dredging projects, the adult densities of H. formosanus and S. nantaiensis were still significantly lower than those before the extreme flood event. Another reason is that the habitat lacked the small and large boulders preferred by H. formosanus and S. nantaiensis [32,33]. After the extreme flood event and continued disturbances from dredging projects, the H. formosanus and S. nantaiensis populations have not recovered to the stage before the extreme flood event for nearly 10 years [28,32,33]. In 2021, the densities of adults and juveniles of H. formosanus and S. nantaiensis were significantly higher than those before the extreme flood event, so we suggest that the populations of H. formosanus and S. nantaiensis populations have recovered to the stages before the extreme flood event [32,33].

5. Conclusions

The phenomenon of niche differentiation when different species coexist in a given community can be seen everywhere in nature, but it is by no means necessary. In many cases, the resources used by ecologically similar species are spatially isolated [1], which largely precludes competition. Likewise, although we found that the difference between interspecific and intraspecific habitat use was common in the Nanzihsian Stream, it tended to occur at different spatial and time scales.

The first contribution of our study is the finding that intraspecific and interspecific habitat use and distribution structure along the longitudinal gradient the river are different for H. formosanus and S. nantaiensis, in particular, with regard to such characteristics as the similarities in body size, feeding habits, and the same qualitative features of habitats. H. formosanus and S. nantaiensis were capable of coexisting in long stretches of rivers, such as the section between the Dubuluan Bridge and the Siaolin Bridge. Unlike studies on territorial fish species such as the Rhinogobius species [5] and the Salvelinus species [34], their analyses revealed the existence of obvious distribution boundaries as a result of differences in substrate particle size and water temperature.

Providing multiple-channel sections in dredging activities could improve fish habitat environments [35]. The final contribution of this study is that we provide an integrated approach for applying engineering and biology to the context of future habitat restoration and projects involving river dredging after extreme flood events. We further illustrate the formation of a deep groove with a uniform depth, and all horizontal structures in the river channel with excessively high drops should be avoided. There should be an alternating and continuous system of deep pool and riffle. The riffle system should maintain a water depth of 25–50 cm, a flow velocity of 0.8–2.2 m/s, and a variety of substrate components, from small pebbles to large boulders along the bed of a turbulent river. The diversity of habitat environments should be maintained from the reach spatial scale to the stream spatial scale (see coefficient of variation in Table 8) in order to satisfy the habitat-related requirements of both fish species that inhabit the river during their entire life cycle.