EvalUaTION Of THE TEmpORal DEvElOpmENT Of THREE INTRODUCED pOpUlaTIONs Of bROOk TROUT, SalvelinuS fontinaliS, IN JIzERské mOUNTaINs, CzECH REpUblIC

Three populations of brook trout, Salvelinus fontinalis, established in streams and reservoirs in the Jizerské Mountains, Czech Republic, were studied between 1995 and 2006. The maximum age was usually 3+ to 4+, but the age 7+ was also recorded. The inflows of all reservoirs are populated mostly by 0+ to 2+ brook trout, while older specimens live in a lentic environment, migrating to lotic habitats only to spawn. The density of brook trout in streams ranged from 8 to 95 individuals per 100 m2 during spring and summer, while that for reservoirs ranged between 31 and 290 individuals per ha. A strong negative correlation between the size of adult fish and population density in reservoirs was observed. The type of management influenced both density and growth of brook trout. Further introduction of this species in the central European area should be restricted to habitats where native salmonids cannot survive, such as acidified waters. key words: growth; population density; age structure; non-native species Received February 15, 2014; Revised October 23, 2014; accepted October 24, 2014


INTRODUCTION
For centuries human activities have been, and still are, breaking down natural barriers to species dispersal, resulting in gradual homogenization of the earth's biota.However, both unintentional and intentional species introduction are now occurring at unprecedented rates (D' Antonio and Vitousek, 1992).Fish, in particular, have been deliberately introduced worldwide, usually to improve local fisheries or enhance recreational activities (Billington and Hebert, 1991).However, knowledge of the biology and ecology of introduced species, necessary for the evaluation of their effect on native species and their potential socio-economic usefulness, is still inadequate (Allendorf, 1991).
Brook trout, Salvelinus fontinalis, native to eastern North America, from Newfoundland to Hudson Bay and south to Georgia, is an example of a freshwater fish that has reached an almost worldwide distribution (MacCrimmon and Campbell, 1969).In the Czech Republic brook trout was first introduced in 1883 (Frič, 1884), since when it has been reared in many hatcheries and distributed to suitable localities all over the country.Despite this, self-reproducing populations have very rarely been established (Dyk, 1963;Lohniský, 1963).
There is a wealth of literature on North American populations of brook trout, but information on populations in Europe is scarce (Cavalli et al., 1997), including in the Czech Republic, where no detailed study of its biology and ecology has been conducted.The aim of this study was to investigate acclimatization of three populations of brook trout, introduced in the 1990s into the Protected Landscape of the Jizerské Mountains, northern Czech Republic, to evaluate the effect upon them of the type of management and further utilization of the species in the central European area.

maTERIals aND mETHODs
The study was conducted within the catchments of three mountain reservoirs (Table 1) in the Jizerské Mountains (50°50ʹ N; 15°10ʹ E).All catchments have similar water characteristics: conductivity is <75 μS cm -2 ; dissolved oxygen concentration is high, 8 to 11 mg l -1 ; pH ranges between 5 and 7.However, during snowmelt and high rainfall, acidity can increase to pH 4. Summer temperature in streams does not exceed 15°C, whereas the surface temperature in reservoirs can reach 24°C.Brook trout is the only fish species to have been established successfully in all three studied catchments.Details about stocking of the reservoirs with brook trout and angler catch are summarized in Table 2.
Research was conducted from 1995 to 2006.Sampling was carried out in May (further referred to as spring), the second half of August (summer) and the second half of October (autumn).In 1995 and 1999, sampling was conducted only in spring and summer, while between 2003 and 2006, it was undertaken only in autumn, with the exception of Bedřichov Reservoir in 2004 and 2005, when sampling was carried out each month from May to October.On each occasion, two to five stream transects were sampled in each catchment.Fish were collected with either a LENA portable battery pulse electrofisher or BMA engine electrofisher (Radomír Bednář, Olomouc, Czech Republic).Prior to sampling, each transect was isolated using barrier nets (mesh size 6 mm).Fish that were caught were held in cages placed in the stream and anesthetized using 2-phenoxy ethanol before measuring and marking procedures.Standard length (S L ) of all specimens was measured to the nearest millimeter.For marking and tagging, Visible Implant Elastomers, Visible Implant Alphanumeric Tags (both Northwest Marine Technology), T-Bar Anchor Tag (Floy Tag) or fin clips were used.Scales were collected for age determination.After recovery from anesthesia, the fish were released.
The frequency distribution of the length of individuals was used to divide them into age groups.The age estimated by this method was then compared with that determined from the scales.So as not to damage them, newly hatched fry were not collected during spring sampling, even though they were present.
Due to the small width of the sampled transects (maximum distance 6 m) and high transparency of the water, population density was estimated both from the total number of fish caught in a transect and those recorded escaping downstream during sampling.

Estimation of population size
The model of temporal change in capture probability (Mt), as implemented in the Simply Tagging program (Pisces Conservation Ltd.), was used for estimation of total population size.This program uses a maximum likelihood estimator of Chapman (1951) for two sampling occasions.As the population estimates were based on one-year intervals (autumn at time t to autumn at time t+1), the number of initially marked fish was reduced to allow for mortality.For Souš and Josefův Důl populations, a natural annual mortality of 50%, widely reported for brook trout populations (Saunders and Power, 1970;Quinn et al., 1994;Curry et al., 2003), was applied.For Bedřichov Reservoir, where it was necessary to allow also for fishery catch, a reduction of 60% was used.Fingerlings at time t+1 were excluded from the calculation.

statistical analysis
Statistical analyses were performed in Statistica 6.0.Analysis of variance (ANOVA; α = 0.05) was used for comparison of S L of brook trout.Kruskal-Wallis and Mann-Whitney U tests (α = 0.05) were used for comparison of fish density in the streams.Spearman's rank correlation coefficient (α = 0.05) was used to investigate any correlation between brook trout density and growth.

REsUlTs age structure
The maximum age of brook trout determined from scales was 4+ in the Bedřichov and Josefův Důl populations, and 5+ in the Souš population.However, individual marking in the Souš catchment showed that some specimens could reach a considerably greater age, at least 7+.
The most abundant age groups in the inflows of all reservoirs were 1+ and 2+ in spring, and 0+, 1+ and 2+ in summer (Fig. 1).At spawning time, older (≥2+) and thus bigger brook trout, migrating from reservoirs to the inflows, were abundant.In Bedřichov, it was not possible to separate accurately spawning fish into age groups for most of the years.Therefore, these were grouped together in a class, "2+ and older".In the inflows of Josefův Důl Reservoir, large older fish (≥3+) prevailed only in autumn 2000, while in the other years, the age composition in autumn was rather similar to that of summer (Fig. 1).This could indicate delayed spawning or a much lower abundance of adult fish in the reservoir in comparison with Bedřichov and Souš reservoirs.For several years following the brook trout introduction in 1996 to Souš Reservoir, the age structure of its population in the inflows was different from that of the other reservoirs.In 1997 and spring 1998, the cohort originating from the fry stocked in spring 1996 predominated in the catch, while during summer 1998, fingerlings from successful spawning in 1997 predominated (Fig. 1).Since autumn 1999, the dynamics of the population structure has been similar to that of the other two reservoirs.

population density
Densities of brook trout in the inflows of the all reservoirs are reported in Table 3. Population densities in the streams fluctuated throughout the course of the study and no overall trend was apparent.Spring and summer densities were not significantly different within the same locality (Mann-Whitney U test).However, due to the presence of spawning fish, densities in the inflows of Bedřichov and Souš reservoirs were significantly higher in autumn than in spring and summer (Mann-Whitney U test; Bedřichov: U = 306.5,p <0.001; Souš: U = 136, p <0.001), while autumn densities in the inflows of Josefův Důl Reservoir were not significantly different from those of spring and summer.
Although there was no statistically significant difference in spring and summer brook trout densities among the three catchments, the autumn densities did differ (Kruskal-Wallis test: H (df 2, N 90) = 7.49; p = 0.02).
The estimated population sizes were in the same order of magnitude for all reservoirs (Table 4), though density decreased with increase in size of reservoir: the highest densities were in Bedřichov, while the lowest were in Josefův Důl.
Statistically significant differences among years in the S L of spawning fish were observed in each reservoir catchment (ANOVA; Souš: F (9;4185) = 61.83,p<0.001;Bedřichov; F (11;4443) = 39.13,p<0.001;Josefův Důl: F (6;678) = 17.65, p<0.001) (see Fig. 2).The overall average S L of mature fish, calculated from the data pooled for all seasons, differed significantly among populations (ANOVA; F (2;9332) = 412.78,p<0.001).These lengths were 190.1, 206.4 and 218.8 mm for Bedřichov, Souš and Josefův Důl, respectively.The growth of the 0+ cohort was weakly negatively correlated with the mean spring and summer density of trout in the streams (Spearman R = -0.52,p = 0.02).No significant correlation was found between the growth increment of 1+ brook trout (length increment of 0+ cohort over a one-year period, from autumn to autumn, time t to t+1) and the mean spring and summer brook trout density.There was a strong negative correlation between size of adult fish and estimated overall density for the reservoirs (Spearman R = -0.885,p < 0.001).

DIsCUssION
The determination of age of brook trout from scales is a common ichthyological tool (e.g.Cooper, 1951;Saunders and Power, 1970;Frenette and Dodson, 1984;Cavalli et al., 1997).This method was found here to be reliable for small (and therefore young) specimens (S L ≤150 mm), which have a relatively stable growth rate.However, for larger specimens, whose growth rates vary and can rapidly decrease, this method can result in underestimation of age.This problem was reported by Dutil and Power (1977)  and Kozel and Hubert (1987).Individual marking of brook trout at Souš Reservoir revealed considerably higher maximal age (at least 7+) than was determined from scales (5+).It is possible that the maximum age of brook trout in the studied populations was seven to eight years.However, most specimens probably reach the age of 3+ to 4+; older fish are rather rare.This is in agreement with data published on native and introduced populations of brook trout from North America, where age 7+ or higher has been recorded in some populations (Saunders and Power, 1970;Donald et al., 1980;Frenette and Dodson, 1984;Kozel and Hubert, 1987;Donald and Alger, 1989), while for the majority of brook trout populations the maximum age is 3+ or 4+ (McFadden, 1961;Wydoski and Cooper, 1966;Flick and Webster, 1976;Quinn et al., 1994).
The inflows of all reservoirs are populated mostly by young specimens, 0+ to 2+.This indicates that most brook trout leave a lotic environment at latest during their third winter and spend the rest of their life in a lentic environment, apart from during the spawning season when mature specimens migrate from reservoir to stream.This underlines the importance of inflows for the maintenance of brook trout populations, both as a spawning ground and as a habitat where they spend the first two or three years of their life.The importance of stream habitats has been pointed out even for lake-spawning populations of brook trout, where up to 80% of fingerlings move to streams during their first year and stay there for two years (Curry et al., 1997).
The population structure in Bedřichov Reservoir has been affected by stocking.Released fingerlings are on average twice as large as native ones.This could explain the difficulties in reliably dividing larger fish caught in autumn into age groups.It often resulted in a necessity to combine the bigger specimens in a "2+ and older" group, while in other reservoirs it was possible to determine at least one more age group.Only during the last two years of the investigation, i.e., three years after the stocking of fingerlings and one year after the stocking of 1+ brook trout stopped, were we able to identify 2+ and "3+ and older" cohorts in the autumn catch.
Already in the first year of the study, the age structure of the brook trout population from Josefův Důl suggested a well-developed population.It is noteworthy that the reservoir, although stocked several times previously (Table 2), was believed to be fishless, as no survivors from the stocking had ever been reported; the population density had been presumably too low.
Development of the age structure of the population in Souš was rapid.It was already well developed in autumn 1999 (Fig. 1), the fourth year after introduction.
Spring and summer densities of brook trout in the streams of the Jizerské Mountains fall within the range reported for North American lotic populations of this species (Ensing et al., 1990;Thompson and Rahel, 1996;Clarke and Scruton, 1999).However, the densities estimated here for the reservoirs are higher than those reported elsewhere for brook trout from lentic habitats (Quinn et al., 1994, Curry et al., 2003).This suggests that the recruitment of brook trout is probably high and the capacity of redds is sufficient for reproduction at the localities studied here.
The growth of 0+ and 1+ brook trout cohorts in the inflows of the studied catchments was slow when compared to the growth of these two age groups reported for North America (Carlander, 1969).Interestingly, it was similar to that reported for brook trout from high altitudes (Kozel andHubert, 1987, Cavalli et al., 1997), and oligotrophic or acidified waters (Wydoski and Cooper, 1966;O'Connor and Power, 1976;Frenette and Dodson, 1984;Curry et al., 1997).Indeed, the studied catchments are affected by acidification (Hůnová et al., 2004) and have low productivity.
A comparison of size of adult brook trout from different populations is difficult because the size range of the same age group can be very different, even over a relatively small geographical area (Donald et al., 1980).Nevertheless, the size of adults in our study is smaller than that of other lentic populations (Donald et al., 1980;Quinn et al., 1994;Curry et al., 2003).The largest mature brook trout were in Josefův Důl, and the smallest in Bedřichov.Our data indicate that the type of management influences the growth of brook trout.Regular stocking of the Bedřichov catchment over a long period resulted in overcrowding of the reservoir, negatively affecting growth.Negative correlation between salmonid density and growth has been reported in numerous studies (McFadden et al., 1967;Rabe, 1970;Grant and Kramer, 1990;Rieman and Myers, 1992;Curry et al., 2003).The compensatory effect of ex-ploitation, which can improve growth (Pechlaner and Zaderer, 1985;Langeland, 1986;Donald and Alger, 1989), was not evident in Bedřichov Reservoir, probably because the angling pressure is too low to effect a reduction in population size.Increased growth of adult fish in 2004, recorded when the stocking program stopped, was caused most probably by a decrease in brook trout density in the reservoir.Thus the food supply in the reservoir, previously shared by many individuals, was sufficient to support an increase in size (Pechlaner and Zaderer, 1985).
This study shows that brook trout can easily become established in suitable habitats in central Europe and can become important for recreational fishing.Stocking with an appropriate number of fry could be an efficient means of introduction.However, although brook trout have been stocked in many areas of the Czech Republic, only a very few populations have become established (Šanda, 2006).Moreover, such selfsustaining populations inhabit localities with no brown trout, Salmo trutta (Šanda, 2006).This suggests that, at least in the Czech Republic, brown trout replaces brook trout.It is known that brown trout out-compete brook trout for habitat and food resources (Waters et al. 1983).Furthermore, in waters where brook trout and brown trout coexist, the presence of upwelling water is believed to be a parameter segregating the spawning habitat of the two species (Witzel and MacCrimmon, 1983).Where such spatial separation does not exist, which is likely in most Czech streams, many brook trout redds are excavated by subsequently-spawning brown trout, thereby reducing brook trout reproductive success (Sorensen et al., 1995).Moreover, Grant et al. (2002) suggested that interspecific sexual interactions between brook trout and brown trout might play a role in species replacement.
We recommend that (i) the further introduction of brook trout into the freshwaters of the central European area should be restricted to acidified waters, where native salmonids cannot survive, and where any local economic benefit that a recreational fishery may provide will be lost; (ii) when stocking lentic waters with brook trout, it is essential to incorporate stream habitats into management plans; (iii) when the decision to introduce brook trout is taken, the possible negative influence of this species on native fauna should always be taken into account.The same recommendations also apply to other areas in Europe with similar climate.

fig. 2 .
fig. 2. Average standard length of 0+, 1+ and mature brook trout in Jizerské Mountain catchments at the end of a growing season.Error bars are standard deviations.Arrows in the lower figure indicate an important change in Bedřichov catchment: (a) end of stocking with fingerlings; (b) end of stocking with 1+; (c) initiation of improved growth of mature fish.

Table 1 .
Basic description of the studied localities.

Table 2 .
Stocking activities at the studied catchments.For Bedřichov, number of brook trout caught by anglers is also given.T L is a total length.

Table 3 .
Average densities of brook trout (no.specimens per 100 m 2 ) in lotic habitats of all studied catchments.
a pooled data for May and June; b pooled data for July to September

Table 4 .
Summary of data used to determine population size, based on a maximum likelihood (ML) estimate, of brook trout in studied catchments of Jizerské Mountains.Population estimates were based on one-year intervals (autumn to autumn).The estimate includes all cohorts with exception of cohort 0+.
a population estimate based on a one-week interval between marking and recapture.Sampling was undertaken in May.Only fish >150 mm were included; b population estimate based on individual marking with T-Bar Anchor Tags.Only fish >150 mm were included; c population estimate based on individual marking with Visible Implant Alphanumeric Tags.Included are also fish marked in spring and summer 2002.Only fish >120 mm were included; d population estimate based on individual marking with Visible Implant Alphanumeric Tags.Only fish >120 mm were included.Estimation is based on fish marked in autumn 2001, and spring and summer 2002.