Does habitat otherness affect weatherfish Misgurnus fossilis reproductive traits?

Abstract

The fecundity and sexual structure of weatherfish (Misgurnus fossilis) population, an inperilled and protected freshwater fish with a poorly known life history, was studied in three waterbodies: the River Ner, the Południowy canal and the Nowy Rów canal (Poland) differing in hydrological character. We compared reproductive traits; i.e. sex ratio, absolute and relative fecundity, oocyte size, gonado-somatic index and body condition. In all study sites, the sex ratio of weatherfish did not differ from parity (1: 1), though in the River Ner, the highest GSI values were recorded for females whilst male GSI did not differ among sites. The absolute (FA = 2860 ± 2065) and relative fecundity (FR = 120 ± 55 eggs per g of female body weight) in the River Ner were significantly lower than in the other two sites. In the River Ner the frequency distribution of oocyte diameter, decomposed using the Bhattacharya method, showed two distinct groups in equal numbers. Oocytes that were ready to spawn were larger in the River Ner than in the Południowy and Nowy Rów canals. Fish in the River Ner were also characterized by the lowest Fulton condition index (mean K = 0.36 ± 0.1). The trade-off between impaired fecundity and increased egg diameter may result from the different nature of the studied watercourses or levels of endocrine-disrupting chemicals (EDCs), such as steroid compounds.

Keywords:

• Sex ratio
• GSI
• oocyte diameter
• fecundity
• weight-length relationship

Introduction

The weatherfish Misgurnus fossilis is a small, freshwater, benthic cobitid, naturally distributed through Central and Eastern Europe, from the northern border of the Alps (North France) to western Russia (Kottelat & Freyhof 2007). Weatherfish are demersal and have the ability to burrow into soft mud during dry periods or frosts (Boroń et al. 2002; Kottelat & Freyhof 2007). Because they are able to use atmospheric oxygen (intestinal breathing) the species can tolerate low dissolved oxygen levels (Jakubowski 1958). This fish species can tolerate unfavourable environmental conditions and a relatively high level of pollution (Jakubowski 1958; Drozd et al. 2009; Pyrzanowski et al. 2019). It prefers shallow waterbodies with sandy substrates covered with a thick layer of organic deposits, often densely overgrown with macrophytes. Typical habitats for weatherfish are slow-flowing rivers, canals and drainage ditches, oxbow and unmanaged lakes and fish ponds (Meyer & Hinrichs 2000; Pekarik et al. 2008; Mazurkiewicz 2012). Basic information on the life-history of M. fossilis, which is essential for conservation, is scarce (Kottelat & Freyhof 2007). Apart from this, M. fossilis is considered endangered due to habitat loss and was listed in the European Fauna-Flora-Habitat and Natura 2000 directives (Annex II of the Council Directive 92/43/EEC), representing species of European Community interest (E.U. 1992) and is included in many national red lists of endangered and protected fish species (Drozd et al. 2009; Hartvich et al. 2010). In Europe, the weatherfish has been classified as species of low concern (LC) (Freyhof & Brooks 2011), but observed genetic diversity is the lowest reported for any European freshwater fish, and it has been proposed that its threat level should be raised (Bohlen et al. 2007). In the light of a widespread decline in weatherfish populations, it is necessary to obtain baseline data on their life history traits, which are a prerequisite for rational conservation planning (Wootton & Elvira 2000). The aim of this study was to provide detailed information on the reproductive biology of M. fossilis populations inhabiting three waterbodies in west-central Poland.

Materials and methods

Study area and sampling site

Study sites were in three watercourses: the River Ner (52°08ʹ83.76ʹ’ N; 18°87ʹ70.17ʹ’ E), the Południowy drainage canal (52°13ʹ14.86ʹ’ N; 19°48ʹ03.62ʹ’ E) and the Nowy Rów drainage canal (51°12ʹ38.29ʹ’ N; 16°43ʹ17.34ʹ’E). All sites were located in the Nature 2000 areas, i.e. the River Ner and Południowy canal in Pradolina Bzury-Neru (PLH100006) and the Nowy Rów canal in Łęgi Odrzańskie (PLB020008). The River Ner is lowland, right-hand tributary of the River Warta, that receives treated wastewater from Łódź, the third largest city in Poland, and is characterized by a high level of water pollution (Kostrzewa 1999; Mosiej et al. 2007; Penczak et al. 2010; Pyrzanowski et al. 2015). The main source of pollution is municipal and industrial wastewater (Mosiej et al. 2007). The main characteristic of examined sited are given in Table I. Along its entire length, the River Ner flows mainly among farmlands, wastelands and meadows. Both drainage canals, along their entire length, are located in Natura 2000 areas. Their catchment areas are typically agricultural, mainly dominated by meadows, pastures and peat bogs. The Nowy Rów canal is part of the drainage system that discharges water to the River Odra, whereas the Południowy canal drains water directly to the River Bzura. In the River Ner M. fossilis co-exists with bream (Blicca bjoerkna), roach (Rutilus rutilus), ruffe (Gymnocephalus cernua), perch (Perca fluviatilis), gudgeon (Gobio gobio) and bleak (Alburnus alburnus). In both drainage canals the weatherfish is the dominant fish species. In case of the Południowy canal it co-exists with pike (Esox lucius), crucian carp (Carassius carassius), roach and tench (Tinca tinca), while in the Nowy Rów canal it occurs together with tench (Pyrzanowski et al. 2015, 2020).

Table I. Characteristics of weatherfish M. fossilis sampling sites: Południowy canal, the River Ner and the Nowy Rów canal

	Południowy canal	River Ner	Nowy Rów canal
Water depth [m]	0.3–0.8	0.5–1	0.3–0.8
Width [m]	3	20–25	3
Bottom	Organic sediments	Sand and silt	Organic sediments
Vegetation[% of watercourse bottom covered]	70	>1	70
		Concentration [mg/L]
Nitrates (NO3^−)	0.02	0.04	nd.
Nitrites (NO2^−)	3.21	27.28	0.13
Phosphates (PO4^3−)	4.55	3.84	2.84
Sulphureous (SO4^2−)	152.61	333.37	139.70
Fluorides (F^−)	0.23	1.90	0.41
Chlorides (Cl^−)	267.50	468.94	116.40
Bromides (Br^−)	0.33	0.35	0.10
		Concentration [ng/L]
Oestrone (E1)	20.5	46.9	nd.
17β-oestradiol (E2)	25.2	46.0	nd.
Oestriol (E3)	23.4	12.2	nd.
17α-ethynyloestradiol (EE2)	17.3	14.0	nd.

nd. - not detectable concentration.

Sampling

The fish samples were performed at the peak of reproductive season (April) 2014 and 2015 at three sampling sites, and to maintain the uniformity of gonad developmental conditions, sampling dates were adjusted to the pattern of prior local temperature. Fish were caught by electrofishing (EFGI 650; BSE Specialelektronik Bretschneider, Germany), immediately euthanized with an overdose of clove oil (Javahery et al. 2012), chilled and subsequently frozen in the laboratory.

Analysis

In the laboratory, all specimens were measured for total length (TL) and standard length (SL) to the nearest 1 mm and weighed (W) to the nearest 10 mg before dissection. The relationship between SL and TL was described by the equation: SL = 0.889 × TL – 1.728; r² = 0.99; p < 0.001. Fish sex (female, male or juvenile) was determined by visual examination of secondary sexual characteristics and confirmed by dissection of the gonads. The sex-ratio of all populations was calculated as the number of males divided by the total number of individuals (males plus females) in the population. For each sampling date and site, the observed proportion of males to females was tested for deviations from a 1:1 sex ratio using a binomial test (Wilson & Hardy 2002). Fish gonads were removed, weighed (W_G) (nearest 10 mg) and preserved in glycerine. To determine reproductive allocation, the gonado-somatic index (GSI) (Wootton 1998) was calculated using: GSI = 100W_G/W. From each ovary, three subsamples (anterior, central and posterior part of left gonad) were collected and weighed (nearest 0.1 mg). Oocytes were photographed under a stereomicroscope (Nikon SMZ1000), counted and their diameters measured to the nearest 0.001 mm using LUCIA 5 image analysis software. The gravimetric method was used to evaluate absolute fecundity (F_A) – oocytes number per female and relative fecundity (F_R); i.e. number of oocytes per 1 g of female body weight was calculated (Bagenal 1978). For each site, oocyte diameters data were pooled and used to obtain their size-frequency histograms. If the frequency of oocyte size distribution was polynomial the Bhattacharya method followed by modal class progression analysis was used to decompose observed distributions into Gaussian components (Bhattacharya 1967). This procedure allows separation of discrete normal distributions in the data based on a separation index (SI) in the case that the difference between successive Gaussian means divided by the difference between their standard deviations exceeded 2 (Gayanilo et al. 2005). For each fish the Fulton condition factor (K) was calculated using (Le Cren 1951; Ricker 1975):

$\mathrm{K}=100\times \mathrm{W}\times \mathrm{T}{\mathrm{L}}^{-3}$

where:

W– fish weight (g)

TL – total length (cm)

One-way analysis of variance (ANOVA II) was used to test for differences among sites for the F_A, F_R and oocytes size and two-factor ANOVA (site × sex as factors) was performed for GSI and K. When significant differences were observed, multiple comparison tests (post-hoc HDS Tukey test) were used (Zar 2010). All values were reported as mean ± standard error. Before running ANOVA, the data were examined for normality (Shapiro-Wilk test) and homogeneity of variance (Levene’s test).

The relationships between body weight (W), gonad weight (W_G), absolute fecundity (F_A) and total length (TL) were determined by linear regression (log-transformed data). The weight-length relationship was tested against isometry; i.e. slopes b = 3, or allometry (positive; b > 3 or negative; b < 3) with Bailey’s t-test for sex and sites. To identify differences among sites, analysis of covariance (ANCOVA) was used. Firstly, differences in slopes (b) were tested and if the null hypothesis was rejected, the common slope (b_c) was calculated and differences in intercepts (a coefficient) were tested (Zar 2010). Tukey’s HSD post-hoc test was also used to identify which sites were responsible for differences in linear regressions (Zar 2010).

Results

Sex ratio

A total of 242 individual weatherfish (117 females, 97 males and 28 juveniles) were collected, of which 76 individuals were from the Południowy canal (collected in 2014), while 84 individuals were from the River Ner and 82 from the Nowy Rów canal (both collected in 2015) (Table II). In all sites the sex ratio did not differ from parity; the Południowy canal: f_M = 0.426 (p = 0.047); the River Ner: f_M = 0.471 (p = 0.086); the Nowy Rów canal: f_M = 0.450 (p = 0.060).

Table II. Characteristics of weatherfish M. fossilis from the Południowy canal, the River Ner and the Nowy Rów canal study sites. Mean values with standard deviation and range for total length (TL), body weight (W) and gonad weight (W_G). Dates in brackets indicate sampling year

			Południowy canal (2014)	River Ner (2015)	Nowy Rów canal (2015)
Females		n	39	35	43
	TL (mm)	Mean ± SD	136.9 ± 40.8	152.3 ± 36.5	154.3 ± 29.7
		Min – max	88.9 – 199.5	100.35 – 233.0	114.5 – 247.0
	W (g)	Mean ± SD	16.1 ± 14.1	14.4 ± 9.8	14.1 ± 14.1
		Min – max	2.7 – 41.7	2.8 – 42.9	6.1 – 65.2
	WG (g)	mean ± SD	2.117 ± 2.787	2.298 ± 2.587	1.772 ± 4.495
		Min – max	0.015 – 9.075	0.013 – 9.926	0.013 – 18.065
Males		n	29	32	36
	TL (mm)	Mean ± SD	124.7 ± 29.1	135.9 ± 24.8	144.2 ± 21.8
		Min – max	87.6 – 171.0	96.7 – 173.0	109.9 – 200.5
	W (g)	Mean ± SD	10.5 ± 7.8	10.1 ± 4.8	13.1 ± 6.5
		Min – max	2.7 – 23.3	3.1 – 18.5	5.6 – 29.0
	WG (g)	Mean ± SD	0.087 ± 0.086	0.083 ± 0.073	0.082 ± 0.087
		Min – max	0.002 – 0.262	0.003 – 0.234	0.006 – 0.392
Juveniles		n	8	17	3
	TL (mm)	Mean ± SD	95.0 ± 7.9	104.1 ± 5.9	124.9 ± 9.6
		Min – max	87.5 – 107.6	90.5 – 116.0	115.5 – 138.3
	W (g)	Mean ± SD	3.4 ± 1.0	4.0 ± 0.7	7.8 ± 2.2
		Min – max	2.3 – 5.1	2.7 – 5.0	5.5 – 10.8

Fish condition

The Fulton condition index (K) (Figure 1) varied significantly among sites (F_{2, 208} = 16.80, p < 0.001) but not between males and females (F_{1, 208} = 1.89, p = 0.171). Moreover, there is no significant interaction between sex and site (F_{2, 208} = 1.93, p = 0.148). A post-hoc HDS Tukey test revealed significantly lower values for condition for both sexes in the River Ner and for females in the Nowy Rów canal. In contrast, the highest condition values were found in the Południowy (both sexes) and for males in the Nowy Rów canal (Figure 1).

Figure 1. Fulton’s condition index (mean ± standard error) for weatherfish, Misgurnus fossilis; the same letters denote groups that did not differ statistically (the HDS Tukey post-hoc test)

GSI conditions

The gonado-somatic index varied significantly with sex (F_{1, 208} = 70.643, p < 0.001) and site (F_{2, 208} = 4.154, p = 0.017). In general, females GSI was over ten times higher than males GSI (Figure 2) and post-hoc HDS Tukey test revealed a lack of differences in males GSI among sites. For females there were differences between the Południowy canal and the River Ner (p = 0.039) and the Nowy Rów canal and the River Ner (p = 0.003) (Figure 2). In all study sites, females and males GSI increased significantly with length (Table III). Analysis of covariance (ANCOVA) showed a lack of difference in slopes among sites (F_2;113 = 0.114, p = 0.891) and a common slope was b_c = 5.053, (s.e.b_c = 0.277). Nevertheless, differences between estimates of the coefficient a (intercept) were detected (F_2;115 = 14.022, p < 0.001) and multiple comparisons (post-hoc HDS Tukey test) showed differences among all the study sites. For males, a regression of GSI on TL also showed a lack of difference in b among sites (F_2,93 = 0.140, p = 0.867, b_c = 2.382 (0.349)), though there were differences in intercepts (a coefficients) (F_2,95 = 3.149, p = 0.044). However, multiple comparison (post-hoc HDS Tukey test) failed to indicate which sites differed from each other. In this case, the null hypothesis was not rejected and the lack of a difference in the regression of GSI on TL was accepted.

Table III. The gonadosomatic - total length regression parameters and their standard error (SE) (log-transformed data) for weatherfish M. fossilis males (M) and females (F) collected in the Południowy canal (A), the River Ner (B) and the Nowy Rów canal (C)

Site	Sex	a	SE a	b	SE b	r^2	n	p
A	F	−10.1273	0.4820	4.9787	0.2272	0.9285	39	< 0.001
B	F	−10.2073	0.8405	5.0541	0.3868	0.8339	35	< 0.001
C	F	−11.0884	1.7372	5.3122	0.8056	0.5209	43	< 0.001
A	M	−5.4409	1.6416	2.4493	0.7866	0.2642	29	< 0.01
B	M	−6.2161	1.2249	2.8009	0.5757	0.4410	32	< 0.001
C	M	−6.5351	0.9114	2.8836	0.4229	0.5777	36	< 0.001

Figure 2. Average GSI (mean ± standard error) for female and male weatherfish, Misgurnus fossilis from the Południowy canal, the River Ner and the Nowy Rów canal; the same letters denote groups that did not differ statistically (the HDS Tukey post-hoc test)

W-TL relationship

The relationship between body weight and total length (Table IV) was isometric for females in the River Ner (t₃₃ = 0.095, p = 0.925), but in both canals the relationship was positively allometric (t₃₇ = 7.177, p < 0.001 in the Południowy and t₄₁ = 2.675, p < 0.011 in the Nowy Rów). For males this relationship was positively allometric only in the Południowy canal (t₂₇ = 6.214, p < 0.001), but in the River Ner and the Nowy Rów canal males showed isometric growth (t₃₀ = 1.179, p = 0.248 and t₃₄ = 1.303, p = 0.201, respectively). Weight-length relationships differed among all sampling sites, for both females (F_2;111 = 6.727, p = 0.002) and males (F_2;91 = 5.532, p < 0.001) (Table IV). Multiple comparisons (post-hoc HDS Tukey test) showed differences among all the study sites for both sexes.

Table IV. Weight-length regression parameters and their standard error (SE) (log-transformed data) of weatherfish M. fossilis males and females collected in the Południowy canal (A), the River Ner (B) and the Nowy Rów canal (C)

Site	Sex	a	SE a	b	SE b	r^2	n	p
A	F	−6.0878	0.1905	3.3510	0.0489	0.9922	39	< 0.001
B	F	−5.4416	0.2164	2.9917	0.0878	0.9724	35	< 0.001
C	F	−5.9974	0.1339	3.2681	0.1002	0.9629	43	< 0.001
A	M	−6.1773	0.1320	3.3930	0.0632	0.9907	29	< 0.001
B	M	−5.0314	0.3415	2.8108	0.1605	0.9109	32	< 0.001
C	M	−4.5763	0.6232	2.6233	0.2891	0.7077	36	< 0.001

W_G was significantly associated with TL (Table V). For females, the slope of the relationship (coefficients b) between TL and W_G did not differ among sites (F_2;109 = 0.283, p = 0.754, b_c = 8.182 (0.287)) but among sites differences were found for the coefficient a (F_2;115 = 14.565, p < 0.001). A post-hoc HDS Tukey test showed differences among all study sites. A similar situation was observed in the case of males, with no among sites differences in coefficient b (F_2,91 = 0.0714, p < 0.930, b_c = 8.182 (0.287)), but significant differences for coefficient a (F_2,91 = 49.349, p < 0.001). Post-hoc comparisons (HDS Tukey test) showed differences only between the drainage canals; i.e. the Południowy canal (A) and the Nowy Rów canal (C). Differences between the River Ner (B) and both drainage canals were not found.

Table V. Regression of gonad weight on total length (log-transformed data) of weatherfish M. fossilis males and females collected in the Południowy canal (A), the River Ner (B) and the Nowy Rów canal (C)

Site	Sex	a	SE a	b	SE b	r^2	n	p
A	F	−18.2151	0.5160	8.3298	0.2432	0.9694	39	< 0.001
B	F	−17.7287	0.9051	8.0843	0.4171	0.9192	35	< 0.001
C	F	−19.2592	1.6972	8.6624	0.7859	0.7477	43	< 0.001
A	M	−13.6182	1.6513	5.8423	0.7912	0.6688	29	< 0.001
B	M	−13.2475	1.2139	5.6117	0.5705	0.7633	32	< 0.001
C	M	−13.1114	1.0306	5.5068	0.4782	0.7960	36	< 0.001

Fecundity

Fecundity was estimated for 48 mature females from all sites (17 from the Południowy canal, 21 from the River Ner and 10 from the Nowy Rów canal). In females from the Południowy canal absolute fecundity (F_A) ranged from 2828 to 12,336 eggs (mean ± SD: 6153 ± 2812.8), relative fecundity (F_R) varied from 43 to 222 eggs/g of body weight (1184 ± 51.77). In females from the River Ner, F_A ranged from 355 to 5734 eggs (2859 ± 2065.0), F_R from 75 to 320 eggs (124 ± 54.68), while in females from the Nowy Rów canal F_A ranged from 459 to 23,776 eggs (8438 ± 10,116.2), F_R from 39 to 423 eggs (148 ± 148.07).

Weatherfish absolute fecundity varied among sites (F_2,45 = 4.557, p = 0.016), and a post-hoc HDS Tukey test showed differences between the Południowy canal and the River Ner (p = 0.012). Similarly, there were differences among sites in relative fecundity (F_2,45 = 4.869, p = 0.012). A post-hoc Tukey’s test showed differences between the Południowy canal and the River Ner (p = 0.0415), the Nowy Rów canal and the River Ner (p = 0.030). However, relative fecundity (F_R) did not differ between canals (p = 0.834).

Across all sites F_A increased significantly with fish length, though coefficients b and a did not differ among sites. The relationship between TL and F_A for all sites took the form: F_A = 5.494 (0.633) × log TL - 8.816 (1.423), r² = 0.621, p < 0.001.

Oocyte size

Oocyte diameter of females from the Południowy canal ranged from 0.30 to 1.30 mm (0.99 mm ± 0.63), while their mean (±SD) wet weight was 0.78 mg (± 0.331). For females from the River Ner oocyte diameter ranged from 0.54 to 1.64 mm (mean 1.16 mm ± 0.97) and mean weight was 1.51 mg (± 0.893). For females from the Nowy Rów canal oocyte diameter ranged from 0.29 to 1.25 mm (mean 0.91 mm ± 0.97) and mean weight was 1.11 mg (± 0.671). In all study sites, egg diameter histograms were polymodal and Bhattacharya’s method revealed two distinctive size groups in each population. In the Południowy and the Nowy Rów canals a small fraction of small oocytes were observed (Figure 3). These small oocytes reached a size (mean ± SD) of 0.58 mm (± 0.081) and 0.59 mm (± 0.059) in Południowy and Nowy Rów canal and constituted 4.0 and 5.8% of the oocyte population, respectively. In the two canals, larger oocytes dominated in the gonads and measured: 1.00 mm (± 0.064) and 0.91 mm (± 0.098), respectively. Meanwhile, in the River Ner the number of small oocytes was similar to the number of larger oocytes, and differences in their size were less distinct (Figure 3). In the River Ner, the diameter of small oocytes average 0.89 mm (± 0.094) and the average size of the larger class of oocytes was 1.17 mm (± 0.097). In the River Ner, oocytes ready to spawn were larger than those at the corresponding stage from the canal sites (F_2,726 = 39.69, p < 0.0001). A post-hoc Tukey’s test revealed differences between the Południowy canal and the River Ner (p < 0.001), the Nowy Rów canal and the River Ner (p < 0.001) and with no difference between canals (p = 0.998).

Figure 3. Size-frequency distribution of oocyte diameters for weatherfish, Misgurnus fossilis, from the Południowy canal, the River Ner and the Nowy Rów canal

Discussion

For the weatherfish populations the sex ratio did not differ from the 1: 1 ratio, typical for this species (Meyer & Hinrichs 2000; Pyrzanowski et al. 2020), though in wild populations different levels of ploidy may occur (Raicu & Taisescu 1972; Drozd et al. 2010). In the case of other cobitids, especially the genera Cobitis and Sabanejewia, the sex ratio is usually biased towards females due to polyploidy (Bohlen & Ritterbusch 2000). The mechanisms of sex determination and sex reversal in fish not only vary among species, but is also complex and still not well understood (Leet et al. 2011; Wootton & Smith 2015). Many factors in the aquatic environment may affect the proportion of the sexes in wild fish populations because of the sensitivity of gonad tissue to hormonal treatments, especially in the early life stages (Smith & Wootton 2016).

In the case of weatherfish there are no data on the condition of individuals of this species. In our study, Fulton’s condition coefficient varied significantly among sites but surprisingly not between sex. Fulton’s and other coefficient are an attempt to realize from relation of body weight and length (Le Cren 1951; Bolger & Connolly 1989; Blackwell et al. 2000) but if the weight-length relationship is allometric (b ≠ 3) Fulton’s coefficient index is not applicable. As our result showed, only males in the River Ner and the Nowy Rów canal display isometric growth pattern. Similar like condition, the weight-length regression data are also scarce. As we know such regression was published only for Belgium and also proved positive allometric weight-length relation (Verreycken et al. 2011). Also relative body weight (Wr) suggested as the index of condition (Blackwell et al. 2000), can not be calculated due to small dataset. In such a situation slope of weight-length regression (Bolger & Connolly 1989) or residual analysis (Fechhelm et al. 1995) could be a measure of the condition for population. Being aware of these problems Fulton’s factor was calculated because of its popularity in fish biology studies.

Gonadosomatic index (GSI), which relates gonad weight to body weight, is commonly used in fish ecology to quantify reproductive development in fish (Wootton 1998). Variation in the GSI fluctuation over time reflects not only gonad development but it is also a proxy to assess the energy allocation between reproduction and growth; i.e. as a surrogate of reproductive effort (Stearns 1992; Wootton 1998). Like other loaches, weatherfish showed common differences in GSI between the sexes, with female GSI over 10 times larger than that of males. Among cobitid loaches that are multiple spawners, female GSI usually varies between 18% and 26% immediately prior to reproduction (Mills & Eloranta 1985; Saat et al. 2003; Mousavi Sabet et al. 2011). The gonad weight of both sexes is usually significantly associated with fish size (Table V). Thus, the GSI suffers from the disadvantage of all such indices (e.g. conditions factors) and the index reflects the appropriate regression of gonad size on body length (Bolger & Connolly 1989; Wootton 1998). However, in our study variation in weatherfish GSI among sites showed a similar pattern to that seen for the regression between gonad weight and TL.

In both drainage canals, weatherfish had relatively higher absolute fecundity (on average of 6100 oocytes in the Południowy canal and 8400 in the Nowy Rów canal), compared with the River Ner with an absolute fecundity of 2800 eggs per female. A study by Drozd et al. (2009) for weatherfish from a floodplain area in the Czech Republic, showed the total number of eggs per female ranged between 5800 and 7900, a figure comparable to the Południowy and Nowy Rów canals. Under natural conditions, the reproductive potential of weatherfish can be double this figure, as reported by Podubski and Štedronský (1954) for a population from southern Bohemia, Opalatenko (1974), Kouril et al. (1996) and Adamkova-Stibranyiova et al. (1999).

The absolute maximum fecundity might be determined in the process that controls spawning. In German fish farms, fecundity was artificially increased from 15,900 to 25,800 eggs (Geldhauser 1992). In case of related species, such as the oriental weatherfish (Misgurnus anguillicaudatus), females can produce up to 1800–15,500 eggs per batch (Berg 1949; Suzuki 1983). Clearly, the absolute fecundity of females may vary depending on size. A similar pattern was observed in case of relative fecundity. The highest value was recorded in the Południowy canal (189 eggs per g b. m.), second in the Nowy Rów canal (148 eggs per g b. m.) and lowest in the River Ner (124 eggs per g b. m.). Our results are comparable to those obtained by Drozd et al. (2009), where the stripped fecundity (number of eggs/g b. m.) averaged 121.6 eggs per female, as well as for Mazurkiewicz (2012), who reported a figure of 250 eggs per g b. m. In contrast, Podubski and Štedronský (1954), Opalatenko (1974), Geldhauser (1992) Kouril et al. (1996) and Adamkova-Stibranyiova et al. (1999) obtained values two or three times higher.

The diameter (without envelopes) of mature weatherfish eggs (in metaphase of the second meiotic division) ranges between 1.17 and 1.30 mm (Kostromarova 1991). After laying, the eggs swell to the diameter of 1.7–1.9 mm (Mazurkiewicz 2012). Oocyte diameter for female weatherfish from a floodplain area in the Czech Republic measured on average 1.42 mm with an average weight of 0.88 mg (Drozd et al. 2009). Slightly greater mass of wet eggs was reported by Kouril et al. (1996) with an average weight of one egg obtained during artificial propagation of about 1.07 mg and by Adamkova-Stibranyiova et al. (1999) of 1.00 mg. In the case of the related but smaller loaches, the oocyte diameter is slightly similar, i.e., for M. anguillicaudatus it ranges from 0.72 to 0.85 mm (Suzuki 1983; Zheng 1985), for Misgurnus mizolepis it measures on average 1.10 mm (Kim et al. 1987), for stone loach: 0.85 to 0.96 mm (Mills et al. 1983; Mills & Eloranta 1985; Saat et al. 2003) and for Cobitis taenia on average 1.14 mm (Bohlen 1998). In the present study, the observed average size of unspawned oocytes from females from the Południowy canal (0.99 mm) and Nowy Rów canal (0.91 mm) were smaller than in the River Ner (1.16 mm). Consequently, the average weight of eggs in the River Ner (1.51 mg) was almost twice as high as in the Południowy and Nowy Rów canals (0.78 and 1.1 mg, respectively). The average egg diameter and weight may vary during one individual season and between populations (Drozd et al. 2009). Two groups of oocyte sizes were observed in the ovaries of females from both canals and river. In the canals large, mature oocytes were typical. The dominance of mature oocytes in the gonads may indicate that all eggs were to be released once during the spawning season, suggesting a lack of batch spawning. In contrast, in the River Ner both size classes of oocytes were equally numerous. This difference may indicate another physiological response (at the mating system level) as an investment in iteroparity (Warner 1998), potentially leading to higher lifetime reproductive success.

Differences in reproductive traits of the weatherfish inhabiting the three watercourses presented in this study may have many reasons, but most probably is high level of pollution in aquatic ecosystems. The relatively large number of chemicals polluting the environment is the cause of numerous problems arising from their physiological side effects. The main elements are industrial chemicals, pharmaceuticals and endocrine-disrupting chemicals (EDCs), such as steroid compounds, including natural products, as well as their synthetic forms, which are widely used in the form of oral birth-control pills and in human disease therapy (Pojana et al. 2004; Wang et al. 2018). Despite their usual low concentrations in open waterbodies they have been classified as a potential environmental factor affecting the metabolic processes of living organisms (Vallejo-Rodríguez et al. 2018). EDCs increase the risk of disruption to the reproductive biology of vertebrates (Lange et al. 2012), directly impacting the fertility of fish and contributing to the decline of some fish species (Kim et al. 1997; Jobling et al. 2003, 2006; Mills & Chichester 2005; Lü et al. 2012). Common effects of oestrogen exposure include increased oestrogen and vitellogenin concentrations in plasma, reduced gonad development and disruption of gonad function. Steroid oestrogens also contribute to changes in the ratio of reproductive cell types, pathological changes in gonads, decreased sperm counts and, consequently, to a reduction in fertility. They might also cause changes to male secondary sex characteristics and potentially leading to the complete feminization of males (Leet et al. 2011; Huang et al. 2016; Hassell et al. 2016; Czarny et al. 2017), changes in sexual behaviour (Dammann et al. 2011; Reyhanian et al. 2011) and ultimately contributing to local extinctions (Czarny et al. 2017). The study conducted by Kim et al. (1997) shows that in related species, such as the M. mizolepis, higher concentrations and/or longer exposure to oestrogen 17β-oestradiol (E2) can alter the sex ratio of young individuals. In addition to significant classical feminizing effects, morphological changes within the ovaries and size and morphology of the pectoral fins, being the expression of sexual dimorphism in weatherfish, were also observed by Kim et al. (1997), which also contradicts our observations. The presence of oestrogens in the water, which as Mills et al. (2003) reported, not only affects gamete quality, but are also responsible for decreases in the amount of eggs and sperm produced by fish. Reduced fertility after exposure to oestrogens, particularly to 17α-ethynyloestradiol (EE2) has also been observed by Voisin et al. (2016). Considering that the River Ner is associated with a large urban agglomeration, the level of pollution, including EDCs, is higher than in the case of the other two watercourses, which are located in rural, non-urbanized areas with a lower potential to generate pollution (Table I, Pyrzanowski, unpublished data). This assumption however, requires detailed study on the possible impact of endocrine-disrupting chemicals, especially oestrogenic steroids, on endangered weatherfish populations and contributing therefore to the better understanding of the impact of pollutant on fish life history traits in general.

Additional information

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

Special thanks are extended to C. Smith from Nottingham Trent University (UK) for English correction and the useful suggestions on an earlier version of this manuscript. We are greatly indebted to S. Szklarek from European Regional Centre for Ecohydrology of the Polish Academy of Sciences for water sample analysis. We also thank two anonymous reviewers for valuable improvements to the manuscript. The weatherfish is protected in Poland, therefore all procedures were carried out under permission from the Local Ethics Committee (66/ŁB729/2014) and the Regional Directorate of Environmental Protection (WPN-II.6401.268.2014.KW2).

Disclosure statement

No potential conflict of interest was reported by the authors.