Introduction

About one-third of the known European freshwater fish species are endangered or are facing the risk of extinction (Kottelat and Freyhof 2007; Vincenzi et al. 2012) and nearly all salmonid species have endemic populations that went extinct or became endangered (Williams et al. 1989; Behnke 1992). Among trout species, marble trout Salmo marmoratus is of particular conservation interest (Vincenzi et al. 2012; Lorenzoni et al. 2012), being endemic to the Adriatic River Basin of Northern Italy (Turin et al. 2006; Pujolar et al. 2011) and the Adriatic River system of the Western Balkans (Povž 1995; Povž et al. 1996; Crivelli et al. 2000; Fumagalli et al. 2002). Many authors have reported a progressive restriction of the original distribution area (Sommani 1961; Tortonese 1967; Crivelli et al. 2000; Turin et al. 2006: Meraner et al. 2007; Meraner et al. 2008) and the species is considered one of the most endangered freshwater fish of the Adriatic Basin (Povž et al. 1996; Crivelli et al. 2000), as listed in Annex II of the European Union Habitats Directive (92/43/EEC) and recently included as Critically Endangered in the IUCN Red List (Rondinini et al. 2013). In fact, hybridization with brown trout Salmo trutta (see Povž 1995; Berrebi et al. 2000; Delling et al. 2000; Meraner et al. 2008; Meraner et al. 2010), displacement by alien rainbow trout Oncorhynchus mykiss (see Vincenzi et al. 2011) and habitat alteration (Crivelli et al. 2000; Chiesa et al. 2016) are the most serious causes of concern for the survival of S. marmoratus. Due to these reasons, some rehabilitation projects have been led since the late 1980s, both in Italy (Specchi et al. 2004) and Slovenia (Crivelli et al. 2000; Vincenzi et al. 2012), and several studies have been carried out on the biology and ecology of S. marmoratus (see Povž et al. 1996; Delling et al. 2000; Meldgaard et al. 2007; Vincenzi et al. 2007a, b, 2008a, b, 2010a, b, 2011; Simčič et al. 2015) and on genetics (Berrebi et al. 2000; Snoj et al. 2000; Fumagalli et al. 2002; Jug et al. 2005; Pustovrh et al. 2014; Sušnik Bajec et al. 2015). Nevertheless, despite this great number of works, specific studies including characterizations of the wild marble trout reproductive cycle are still missing, even though this kind of research could give important information in the context of repopulation and management policies. In addition, the development of any useful management tool that could assist in conserving populations would be advantageous for the survival of the species, as suggested by Lorenzoni et al. (2012). Several published works include informations about sex steroid hormone levels for S. trutta (see Estay et al. 2003; Fregeneda-Grandes et al. 2013), O. mykiss (see Tyler et al. 1990; Tyler and Sumpter 1996; Wilkinson et al. 2010; Estay et al. 2012; Kagawa 2013) and other salmonids (Taranger et al. 1998; King and Pankhurst 2003; Jordanova et al. 2013; Taranger et al. 2015), almost all regarding farmed fish, but none addressed the marble trout.

In this context, it was deemed of interest to analyse the reproductive cycle of wild S. marmoratus females, within a community living in its typical habitat. Investigations have been led in a prealpine watercourse close to the boundary between Italy and Slovenia with the aim (i) to obtain a one-year dataset regarding variations of gonadosomatic index, oocyte diameters and plasma level of sex hormones (17β-oestradiol and testosterone) of wild S. marmoratus females; (ii) to check the correlations among the mentioned variables and between these variables and the chemical–physical features; (iii) to validate useful information for the identification of the reproductive stage without killing the specimens.

Materials and methods

Study area. The study was carried out in the Uccea Brook, a right side tributary of the Soča River with a total length of 14 km (Fig. 1); the Uccea spring is placed at 1098 m above sea level in Northeast Italy, on the slopes of the Musi Mount, within the Special Area of Conservation called “Northern Julian Pre-alps” (SAC Site Code: IT3320012), which is included in the Special Protection Area identified as “Julian Alps” (SPA Site Code: IT3321002). The Uccea Brook flows in the west–east direction, partially within the SAC mentioned above; after 9 km from the spring, the watercourse crosses the state border between Italy and Slovenia and then flows into Soča River 5 km after the boundary. The sampling operations were always performed in a brook section of 1.7 km length, between the confluence with the Bianco Brook and the state border limit, outside the SPA and SAC zones.

Fig. 1
figure 1

Study area: samplings were performed in the Uccea Brook, alongside a watercourse stretch between the confluence with the Bianco Brook (UTM coordinates: 33T 376548.00 E, 5129328.00 N) and the state border limit (UTM coordinates: 33T 378168.00 E; 5129188.00 N)

Full size image

All activities described in the present study were performed by permission of the Safeguard Fisheries Authority of the Region Friuli Venezia Giulia, as requested by local laws. Experimental procedures were carried out according to the Guidelines of the European Directive 2010/63/EU on the protection of animals used for scientific purposes and the principle of the 3Rs was applied. The Italian law (D. Lgs 26–14), which applies the European Directive, states that the used protocol does not require an authorization to the use of animals (art. 2, d).

At each sampling event, the values of the main chemical and physical data were recorded in the water column to support biological surveys: in particular, conductivity (µS cm−1), pH, temperature (°C) and dissolved oxygen (mg l−1 and percentage of saturation) were registered using field meters (HI 9033 conductivity meter; HI 9125 pH/ORP meter; HI 9147 dissolved oxygen meter; all instruments are manufactured by Hanna Instruments Inc., Woonsocket, Rhode Island, USA). Four values were measured approximately at mid-depth across a representative brook section, paying attention to cover different microhabitats. The mean values and relative standard deviations (sd) observed during the study period are reported in Table 1.

Table 1 Bimonthly mean values ± sd of chemical and physical parameters observed within the Uccea Brook during the bimonthly study period (February–December 2015)
Full size table

Fish sampling. Captures were performed by wading a watercourse section with a battery-powered backpack electrofisher (Model IG200–2: 15–25 A, 150–200 V; manufactured by Hans–Grassl GmbH, Schönau am Königssee, Germany).

During December 2014, a preliminary sample collection was carried out to check if the chosen population was sufficiently well structured to support the study without consequences for its conservation. As Salmo marmoratus is an annual spawning fish in nature field, preliminary sampling allowed to detect its spawning period within the Uccea Brook. Captures were carried out along a 180 m brook stretch, which was proportional to the riverbed width (Forneris et al. 2005), using the removal method (Seber and Le Cren 1967; Seber 1973) to obtain data about the marble trout community structure. Specimens of S. marmoratus were identified on the basis of the phenotypic characters. At the end of the preliminary sampling, all fish were released at the same site without consequences for their vitality.

As the Uccea marble trout community was sufficiently well structured to support the study, fish samples were bimonthly collected (February, April, June, August, October and December 2015) during the first week of the month. The last two samplings were performed in the middle of November and in the middle of December 2015 to gain information immediately before and after the spawning. Four mature female (20 cm length at least) were captured each time, moved into a tank and anesthetized using tricaine methane-sulfonate (MS-222) (Topic Popovic et al. 2012), to limit damages during the manipulation. Total length (L T ; x ± 0.1 cm) and weight (W; x ± 0.1 g) were measured for each specimen, and a blood sample was obtained from the caudal vein for hormone determinations. Then, specimens were killed by anaesthetic overdose and placed in frozen bags for the transport to the laboratory, where ovaries were dissected. Fulton condition factor (K) was calculated for each specimen (Ricker 1975) and the age was determined by scales lecture.

Gonadosomatic index and oocyte growth. Fresh gonad weights were measured immediately after dissection (x ± 0.0001 g) with an analytical balance, and the gonadosomatic index (I G %) was calculated.

Then, oocyte diameters were measured, using a technique based on the work of Thorsen and Kjesbu (2001) and used by several authors with some difference in details (Murua et al. 2003; Friedland et al. 2005; Klibansky and Juanes 2007, 2008; Alonso-Fernández et al. 2009; Gray et al. 2014; Diaha et al. 2016). For each specimen, an ovary subsample (approximately, 2 cm length) was removed from the centre of the left gonad and placed in 70% ethanol. Samples were then dissected to separate individual follicles, which were placed in glass Petri dishes filled with 33% glycerol solution and submitted to moderate stirring until they were adequately separated among each other. Subsequently, groups of four Petri dishes were captured as images using a flatbed scanner (Epson Perfection V500 Photo). Every image was 2400 dpi (6479 pixels × 22392 pixels) in 8-bit greyscale and saved in uncompressed tagged image file format (TIFF). Once scanned, each 44 MB image was analysed using the ImageJ Version 1.49 program (Schneider et al. 2012). The system was length calibrated (mm units) and diameter measurements were performed for 15 oocytes for each fish, using the greyscale images; two values were taken for each examined oocyte. Oocyte diameters were measured from February to November and fish caught in December (after deposition) were not considered.

Plasma level determination of sex steroids. To determine the sex steroid profiles, blood samples were obtained in 2 ml heparinized tubes, transferred in Eppendorf tubes, brought to the laboratory and kept at ambient temperature for less than two hours. Samples were immediately centrifuged at 3000 rpm for 5 minutes to obtain the plasma fractions, which were then frozen at −20 °C until hormone analysis. Competitive chemiluminescent immunoassays for the quantitative determination of 17β-oestradiol (E2; ng ml−1) and testosterone (T; ng ml−1) plasma levels were used to obtain steroid profiles (Access Immunoassay Systems, Beckman and Coulter Inc., Brea, CA, USA). Samples were assayed following the instructions provided by the manufacturer. The minimum detectable quantity was 0.02 ng ml−1 for E2 and 0.1 ng ml−1 for T; the cross-reactivity to other hormones was less than 2% in both analyses.

Statistical analysis. Mean values and relative standard deviations (sd) of condition factor (K), gonadosomatic index (I G), oocyte diameter, plasma level of 17β-oestradiol (E2) and plasma level of testosterone (T) were calculated for each sampling event. As the data were not normally distributed, analyses were performed using the non-parametric Kruskal–Wallis test. The Conover–Iman test (Conover and Iman 1979; Conover 1999) was used as post hoc test, to assess the presence of significant differences between values observed in two consequent sampling events. The correlations among biotic variables and between these variables and chemical and physical features were investigated with the Spearman rank correlation coefficient (ρ). In addition, the relations between sex steroids an oocyte size, expressed as oocyte diameter, were examined by non-linear regression. All statistical analyses were performed using StatSoft STATISTICA 7.1 (StatSoft 2005), except for the Conover–Iman test, which was performed using R software, version 3.3.1 (R Core Team 2013), with a P level of 0.05 used to interpret significance for all tests.

Results

During the bimonthly samplings, 28 mature females were captured, ranging from 21.0 cm to 40.5 cm length and from 89.4 g to 491.7 g weight (Table 2); most specimens were assigned to the 3+ age group (19 individuals), six specimens were 4+, two were 2+ and only one trout belonged to the 6+ group. Condition factor (K) mean values did not differ significantly during the study period, except in February, when the observed mean value was lower than that at the other months (Kruskal–Wallis test: H 6,28 = 14.00, P < 0.05; Conover–Iman test: P < 0.05 for all comparisons).

Table 2 Mean values and standard deviations (sd) for total length, weight, condition factor, gonadosomatic index, plasma levels for 17β-oestradiol, plasma levels for testosterone and oocyte diameters measured during the bimonthly sampling period (February–December 2015)
Full size table

Gonadosomatic index and oocyte growth. The pattern observed for the gonadosomatic index (I G ) is reported in Figure 2a; significant variations were highlighted during the study period (Kruskal–Wallis test: H 6,28 = 23.71, P < 0.05). The I G values remained nearly at the same level from February to June, ranging from 0.87 ± 0.41% to 1.00 ± 0.27%; then values increased significantly in August (3.73 ± 0.90%) and October (7.72 ± 2.15%) (Conover–Iman test: P < 0.05 for both comparisons). I G increased further in November just prior to spawning, when the maximum mean value was recorded (10.91 ± 3.37%), but no significant difference was highlighted with the previous sample (Conover–Iman test: P = 0.39). The increasing pattern is quite linear from June to November. I G drastically decreased in December, immediately after spawning, as the minimum mean value was observed (0.44 ± 0.25%; Conover–Iman test: P < 0.001).

Fig. 2
figure 2

Bimonthly variations of gonadosomatic index I G (a) and oocyte diameters (b) observed for the female marble trouts sampled in the Uccea Brook [n = 4 for all cases; black dots correspond to mean values; boxes correspond to standard errors (se); whiskers correspond to standard deviations (sd); asterisks indicate significant differences with the previous bimonthly value; *: P < 0.05; **: P < 0.01; ***; P < 0.001]

Full size image

Oocyte growth pattern is reported in Figure 2b and Figure 3. The minimum mean diameter was observed in February (0.692 ± 0.031 mm), after the end of the previous reproductive period and the beginning of the new cycle; then, the oocyte size increased always significantly from month to month (Kruskal–Wallis test: H 5,24 = 22.25, P < 0.05; Conover–Iman test: P < 0.01 for all comparisons). The maximum average size was observed in November (4.624 ± 0.208 mm), just prior to the spawning season (Figs. 2b, 3). A positive correlation between I G values and oocyte growth pattern was highlighted (Spearman rank correlation coefficient ρ = 0.874, P < 0.0001, n = 24). A positive correlation was found also between I G values and water temperature (Spearman rank correlation coefficient: ρ = 0.431, P < 0.05, n = 28) and between oocyte growth and water temperature (ρ = 0.463, P < 0.05, n = 24).

Fig. 3
figure 3

Changes of the oocyte diameters observed for the female marble trouts collected in the Uccea Brook (for each graph n = 120; size class step = 0.1 mm)

Full size image

Plasma level of sex steroids. Plasma steroid patterns observed between February and December 2015 are shown in Fig. 4. Significant bimonthly variations were detected both for 17β-oestradiol (E2) and testosterone (T) (Kruskal–Wallis test: H 6,28 = 24.97, P < 0.01 for E2; H 6,28 = 24.28, P < 0.01 for T). During the February–June period that corresponds to the early vitellogenesis phase, low levels of E2 were detected, ranging between 0.214 ± 0.015 ng ml−1 and 0.442 ± 0.119 ng ml−1; no significant differences were recorded in this period (Conover–Iman test: P > 0.11 for all comparisons). E2 levels increased slightly but significantly in August, reached a peak in October (78.090 ± 23.882 ng ml−1), decreased clearly but not significantly in November and then significantly declined in December (spawning season) (Conover–Iman test: P < 0.001 for all comparisons except between October and November, where P = 0.11).

Fig. 4
figure 4

Bimonthly levels of 17β-oestradiol (E2, continuous black line), testosterone (T, continuous grey line) and variations of water temperature (broken black line) observed during the reproductive cycle of female marble trout in the Uccea Brook (February–December 2015; mean ± sd; n = 4 for each case). Asterisks (black for E2, grey for T) indicate significant differences with the previous bimonthly value; *: P < 0.05; **: P < 0.01; ***; P < 0.001)

Full size image

The seasonal pattern of T levels lagged behind and followed that of E2 (Spearman rank correlation coefficient between E2 and T: ρ = 0.833, P < 0.0001, n = 28), with low values which did not differ significantly between February and June (Conover–Iman test: P > 0.05 for all comparisons), slightly increased in August and rapidly increased in October; the levels remained high at the beginning of the spawning period, when the maximum mean value was detected (71.800 ± 29.406 ng ml−1 observed in November) and then drastically decreased after the spawning (Conover–Iman test: P < 0.03 for all comparisons except between October and November, where P = 011).

A correlation was found also between steroid profiles and water temperature (Spearman rank correlation coefficient: ρ = 0.463, P < 0.02, n = 28 for E2; ρ = 0.421, P < 0.05, n = 28 for T). Sex steroid profiles were also correlated with I G (Spearman rank correlation coefficient: ρ = 0.671, P < 0.0001, n = 28 for E2; ρ = 0.757, P < 0.0001, n = 28 for T) and were strongly linked with oocyte growth pattern (Spearman rank correlation coefficient: ρ = 0.846, P < 0.0001, n = 24 for E2; ρ = 0.932, P < 0.0001, n = 24 for T): the relationship between oocyte diameters and E2 plasma levels and the relationship between oocyte diameters and T plasma levels both resulted in a highly significant non-linear regression, with r 2 values equal to 0.890 and 0.947, respectively (Fig. 5; Table 3).

Fig. 5
figure 5

Relationship between mean oocyte diameter and plasma levels of 17β-oestradiol (E2) (a) and testosterone (T) (b), where x = E2 or T and y = oocyte diameter. The curve for E2 was fitted by: y = 0.5772ln(x) + 1.7711 (r 2 = 0.890, P < 0.0001, n = 24); the curve for T was fitted by: y = 0.6871ln(x) + 1.5368 (r 2 = 0.947, P < 0.0001, n = 24)

Full size image
Table 3 Non–linear regression between sex steroids (E2, T) and oocyte size, expressed as oocyte diameter
Full size table

Discussion

The present study reports bimonthly patterns for variations of gonadosomatic index, oocyte diameters and plasma levels of sex steroids (17β-oestradiol and testosterone) for wild female Salmo marmoratus, living in its typical habitat.

The patterns observed in the Uccea Brook are generally in line with those reported in other studies concerning the biology and reproduction of salmonids. From the beginning of the analysis to the spawning phase, the I G mean values observed in the present work underwent an 11- or 12-fold increase (Table 2; Fig. 2a). This range is comparable with those provided by Jelli et al. (1991), which reported I G mean values for S. marmoratus captured in different river basins of the Po Plain area, ranging between 0.609 ± 0.204% and 15.948 ± 1.854% during the March–December period (Adige River basin) and between 0.556 ± 0.033% and 9.674 ± 2.170% between March and October (Piave River). The I G mean values observed in the Uccea Brook showed a quite constant level from February to June, then rapidly increased in August (at the beginning of the secondary vitellogenesis), reached its peak in November just prior the deposition and finally rapidly decreased after spawning. This pattern is generally in agreement with those observed by other authors with reference to other salmonids such as brown trout Salmo trutta (Linnaeus 1758) (Billard 1987; Estay et al. 2003; Arslan and Aras 2007; Nusrat and Ulfat 2013), Caspian brown trout Salmo trutta caspius (see Jamalzadeh et al. 2013), rainbow trout Oncorhynchus mykiss (see Tyler et al. 1990), amago salmon Oncorhynchus rhodrus (see Kagawa 2013), Ohrid trout Salmo letnica (see Jordanova et al. 2013) and Atlantic salmon Salmo salar (see King and Pankhurst 2003; Taranger et al. 2015). However, although the general pattern is the same, the decrease observed in the early post spawning period (December) is faster than that in some other studies (Jordanova et al. 2013; Nusrat and Ulfat 2013), which reported more gradually declining values, and the maximum values recorded just prior to the spawning are generally lower than those indicated for other salmonids (Tyler et al. 1990; Estay et al. 2003; Jordanova et al. 2013; Kagawa 2013; Taranger et al. 2015), where maximum I G values were equal to 20–22%. However, these differences could depend on climate variations, watercourse characteristics and could occur from zone to zone and/or from year to year (Jelli et al. 1991). An example of this variability could also be found in the work of Arslan and Aras (2007) on the reproductive cycle of two brown trout populations in Turkey, as the authors reported values of I G which differed in two subsequent years for both populations.

Concomitant with the variations in I G , the mean oocyte diameter underwent a 7-fold increase from the beginning of the study until the spawning period (Table 2; Figs. 2b, 3). Diameters recorded during the mature phase are in agreement with those reported for S. marmoratus in the Piave River basin (4.430 ± 0.242 mm), but lower than those obtained in the Adige River area (5.580 ± 0.380 mm and 5.508 ± 0.565 mm) during a work carried out in Northern Italy by Jelli et al. (1991). The observed mature oocyte diameters are also in line with values reported for wild brown trout in Central Italy (4.810 ± 0.028 mm) by Bicchi et al. (2010). The oocyte growth pattern reported in the present study is also comparable with those provided by Estay et al. (2003) for farmed brown trout in Chile and with data reported in literature for O. mykiss (see Tyler and Sumpter 1996; Kwon et al. 2001; Gülsoy 2007) and for S. salar (see King and Pankhurst 2003). Bimonthly values are generally slightly lower than those observed in other studies regarding salmonids (Tyler et al. 1990; Kagawa 2013; Taranger et al. 2015), but this could be due to several factors, such as female fish size (Wootton 1998; King and Pankhurst 2003; Heinimaa and Heinimaa 2004) which influences the timing of the first sexual maturation stage and the duration of the vitellogenesis phases (Bromage and Cumaranatunga 1988; Estay et al. 2003).

The patterns of plasma levels observed herein for gonadal steroid E2 and T are strictly related to the gonadal/oocyte development and were similar and broadly consistent with those reported for salmonids elsewhere. In particular, the trends observed in the present study are in agreement with those described by Kagawa (2013) and with those reported by other authors for other trout species and generally for salmonids (Taranger et al. 1998; Estay et al. 2003; King and Pankhurst 2003; Estay et al. 2012; Hoogenboom et al. 2012; Fregeneda-Grandes et al. 2013; Taranger et al. 2015), though the values provided in the present work for the Uccea Brook population show a wider range (0.214 ± 0.015–78.090 ± 23.882 ng ml−1 for E2; 0.327 ± 0.086–71.800 ± 29.406 ng ml−1 for T) and lower values in the initial samplings (Table 2; Fig. 4). In fact, E2 and T levels remained very low from February to June (≤0.442 ± 0.119 ng ml−1 for E2 and ≤0.820 ± 0.238 ng ml−1 for T), slightly increased in August (Table 2; Fig. 4) and then increased greatly. These differences in plasma concentration of the steroids could be due to several factors such as body size, age, nutritional condition, dominance status and environmental condition, or could be related to the different methodologies used, the geographical zones studied or the different species analysed (Fregeneda-Grandes et al. 2013). In addition, differences could be due to the fact that literature datasets are often referred to farmed/cultured salmonids, or to experimental treatments carried out under simulated conditions on fish collected from hatcheries, while the present study considered only wild fish. The specimen origin from a natural environment could also explain the intra-sample variability, expressed as standard deviations (sd), observed in the present work for I G , oocytes growth and especially for steroid levels (in particular, during the mid- and late vitellogenesis in the August–November period), which is generally higher than in other datasets. Despite these differences, general patterns are consistent with literature data. Moreover, non-linear relationships have been highlighted between oocyte size (expressed as diameter) and E2 concentration, and between oocyte size and T concentration (Table 3; Fig. 5), with a very strong correlation in both cases (r 2 = 0.890 and r 2 = 0.947 respectively). A strong non-linear relationship between E2 and oocyte size (r 2 = 0.834) was highlighted in a study on S. salar by King and Pankhurst (2003), which reported that an exponential relationship could reflect the increase in E2 levels, and that I G and oocyte size also increased following an exponential pattern. However, the relationship between T and oocyte diameter presented herein is stronger and could likely better reflect the progress of the oocyte maturation in terms of temporal scale. In fact, T has been detected in the plasma of females in almost all teleost fish species during the reproductive season (Fregeneda-Grandes et al. 2013) and its role as a precursor for the synthesis of E2 by aromatase activity in the ovary is well known (Fostier et al. 1983; Kagawa et al. 1983; Norberg et al. 1989; King and Pankhurst 2003; Kagawa 2013). Follicles produce E2 in response to gonadotropin stimulation, but as this production increases during oocyte growth, it decreases rapidly in association with the ability of the oocyte to mature (Kagawa et al. 1983; Young et al. 1983; Kanamori et al. 1988; Kagawa 2013). Moreover, T production increases during oocyte growth and peaks during the postvitellogenetic phase. These high plasma T levels observed in salmonid mature females are related to the decrease of the aromatase activity which converts T to E2. Using in vitro experiments, Kagawa (2013) described trends for Oncorhynchus rhodurus, indicating that E2 levels increase during vitellogenesis, reach a peak and then rapidly decrease in mature and ovulated fish, while the pattern of T levels follows that of E2 but remains high in mature and ovulated specimens. Estay et al. (2003) reported that in farmed brown trout, T levels remain high during most of oogenesis and then fall down during ovulation, while the E2 level falls down in the spawning period, when the oocyte size is greater. T levels reported herein for S. marmoratus remained high at the beginning of the spawning period, when the maximum mean value was detected, and then drastically decreased after the deposition. These differences could conceivably be due to the factors mentioned above. In particular, differences in plasma steroid could be due to the source of fish, i.e. farmed/cultured or wild (Fregeneda-Grandes et al. 2013). Norberg et al. (1989) found that differences in plasma levels between wild and farmed female brown trout might be due to differences in stress susceptibility, environmental conditions, nutritional status or genetic divergences. In addition, some datasets were obtained from experimental treatments carried out under simulated conditions (Taranger et al. 1998; Estay et al. 2003; Wilkinson et al. 2010; Estay et al. 2012; Fregeneda-Grandes et al. 2013; Taranger et al. 2015) on fish collected from hatcheries or from in vitro experiments (Kagawa et al. 1983; Kagawa 2013), instead of analysing wild specimens and/or studying another species.

However, the observed patterns imply that higher levels of E2 do not correspond to the maximum oocyte size (Table 3; Fig. 5a) at the end of the reproductive cycle in mature fish, even if the correlation between E2 and the oocyte diameter is strong, while this problem did not occur with T plasma concentrations, which can always correspond to a homologous oocyte size. Therefore, it could be likely possible to obtain information about the oocyte maturation stage from this relationship. This fact is of pivotal importance, especially in the case of an endangered species like S. marmoratus, which needs the development of several non-invasive management tools.

In conclusion, the patterns observed for I G , oocyte growth, E2 and T for S. marmoratus in the Uccea prealpine brook are consistent with literature data and could represent a general reproduction pattern for S. marmoratus. In addition, the significant relationship between oocyte growth and plasma steroid concentrations (especially for testosterone) is clearly validated by the data of the present study and the quantification of plasmatic testosterone levels will be a valuable tool for the identification of reproductive females, without requiring their sacrifice, in future programs for the management of marble trout.