Mark-recapture results and changes in bat abundance at the cave of Na Turoldu, Czech Republic

Transcription

1 Folia Zool. 51(1): 1 10 (2002) Mark-recapture results and changes in bat abundance at the cave of Na Turoldu, Czech Republic Jifií GAISLER 1 and Josef CHYTIL 2 1 Department of Zoology and Ecology, Faculty of Science, Masaryk University, Kotláfiská 2, CZ Brno, Czech Republic; gaisler@sci.muni.cz 2 Administration of the Biosphere Reserve Pálava, Námûstí 32, CZ Mikulov, Czech Republic; jchytil@palava.cz Received 25 October 2000; Accepted 9 November 2001 A b s t r a c t. Between , 3,148 records of hibernating bats were made in a natural limestone cave, 92 % of them concerning Rhinolophus hipposideros. Other species included Myotis myotis, M. mystacinus, M. brandtii, M. emarginatus, M. nattereri, M. daubentonii, Plecotus auritus, P. austriacus and Myotis blythii (which was not recorded after 1973). The abundance of R. hipposideros (and of all bats combined) decreased between , probably due to the impact of mark-recapture work, and between due to the devastation of the cave. After the cave was put under protection and the marking of bats was stopped, the hibernating assemblage recovered so that, between , numbers of bats showed significant increases resulting in numbers more than twice those at the beginning of the monitoring period. Between , 517 bats were netted and banded from spring until autumn in front of the cave. Nettings of species not found in winter (Myotis bechsteinii, Eptesicus serotinus, Vespertilio murinus, Nyctalus noctula and Barbastella barbastellus) increased the total number of species to 15. In N. noctula and V. murinus echolocation signals were also detected and, in the latter a display flight with mating calls was recorded. Samples of netted bats were compared with samples of bats observed in the same winter, and significant differences were found in species composition, diversity and in the dominance values of individual species. It is suggested that considerable numbers of bats other than R. hipposideros occur at the locality in winter, but have remained undetected due to their hidden hibernation. We analysed 1,038 individuals marked at Turold and 237 recaptured, plus 10 of bats marked elsewhere and recaptured at Turold. The longest movement recorded (66 km) was in a female M. blythii, the highest age (19.5 years) was in a male R. hipposideros. Key words: hibernating bats, fluctuation in numbers, summer bat community, results of banding Introduction The cave of Na Turoldu is the largest known bat hibernaculum in the territory of the Pálava UNESCO Biosphere Reserve, and a previous study (Gaisler et al. 1996) has shown bats to represent the most diverse community among the reserve s small mammals. While this study reported data collected between , bats in the Na Turoldu Cave have been monitored from 1958 up to the present day. This 42 year period represents the third longest continuous bat monitoring record from any site in either the Czech Republic or the former Czechoslovakia, and is exceeded only by certain localities south of Prague (44 years) and in the Moravian Karst (43 years) ( ehák 1997, ehák & Gaisler 1999). Although the cave in question has not been visited every year, the sample is large enough to be analysed with respect to shifts in numbers of hibernating bats - one of the goals of this paper. However, the previous study also showed that certain species, although common in the territory of the 1

2 reserve, were found in the cave only rarely or not at all. Therefore, the second goal of this paper was to compare species composition and diversity, as well as the dominance values of individual species, between samples obtained both inside and outside the cave. By this we intend to demonstrate just how many bat species can occur within a very small territory (i.e. <1 ha - the cave interior plus its surroundings). Material and Methods Bats hibernating inside the cave have been monitored from The cave was visited once a year in January or February except in 1963 (when it was checked in March) and in , and 1983 (no surveys made). Until 1980 bats were banded and those found with a band were recaptured. To avoid any confusion when comparing papers about batbanding and bird-ringing we make it clear that, for bats, the terms banding and ringing are synonymous. The bands used to mark bats were very similar to the rings applied to passerine birds (for methodological details see Hanák et al. (1962) and Roer (1995)). From 1981 until the present day, bats found in the cave were censused without marking or disturbance, and only those banded previously, when found, were examined closely to check the band number (Bauerová et al. 1989). Ten species were recorded, the numbers of individuals in each sample are specified in the paragraph Winter census, and total numbers are given in Table 1. Table 1. Total number of records (N) and mean number of bats (Mean) per one winter check in the Na Turoldu Cave, ; n = 35 checks. Species N Mean Rhinolophus hipposideros Myotis myotis Myotis blythii Myotis mystacinus 1 <0.1 Myotis brandtii 1 <0.1 Myotis emarginatus Myotis nattereri Myotis daubentonii Plecotus auritus Plecotus austriacus indetermined Total From , netting was performed outside the cave, up to 20 m from its entrance. This entrance, however, is closed by a solid gate which does not allow the bats to fly in and out (see the locality description). Bats were netted irregularly from April until November each year, a period termed summer for the sake of simplicity. Usually three m long nets were set one hour before sunset and lifted shortly after midnight. Netted individuals were banded and released. Thirteen species were netted (the numbers of individuals in each sample will be presented elsewhere) and the total number of banded bats can be seen in Table 2 and those of recaptures in Table 3. The cave was not visited at each netting but, when visited, no bats were seen inside. When the two samples (winter and summer) were combined, the total number of species recorded was 15, and the total number of records 3,741. In addition to the above data, a few ultrasound records of Nyctalus noctula and Vespertilio murinus were made with a Skye bat detector. 2

3 The following abbreviations of bat species names are used in the list of records from the winter census: Rhip = Rhinolophus hipposideros, Mmyo = Myotis myotis, Mbly = M. blythii, Mmys = M. mystacinus, Mbra = M. brandtii, Mema = M. emarginatus, Mnat = M. nattereri, Mdau = M. daubentonii, Paur = Plecotus auritus, Paus = P. austriacus. Calendar dates are given with date first, followed by month and year, e.g = 10 February Further explanation of symbols used in the paper: m = male, f = female, T = Turold, both inside and outside the cave, O = other localities, B = banding (ringing), R = recapture, T/T = banded and recaptured at Turold, T/O = banded at Turold, recaptured elsewhere (analogously O/T), W = winter, inside the cave, S = summer, outside the cave (netting), W/1W = banded in winter, recaptured the next winter (analogously /2W, /3W = recaptured the second, the third winter after banding; S/ = banded in summer). Study Area The cave of Na Turoldu (= at Turold) is a natural limestone cave which consists of a complicated system of corridors and small domes forming three storeys situated at elevations between m a.s.l. The total length of known underground caverns is 1,100 m. There are numerous crevices in the ceiling, the walls and in the bases of underground corridors and domes. Some of the crevices open to the surface and obviously allow bats to move in and out, but the exact locations of these bat passages are unknown. Winter air temperatures in the cave vary between C (n = 16). The cave is named after the 385 m high hill Turold, situated on the periphery of the small town of Mikulov (population = 8,000). Between , Turold was affected by the output of Ernstbrunn dolomitic limestone. During the last 66 years, however, this limestone quarry has not operated and it has become overgrown - mostly by natural vegetation. The locality has been protected since 1946 and became a nature reserve in The quarry forms part of the Pálava Biosphere Reserve, which is situated in southern Moravia, close to the Czech- Austrian border. The Pálava Reserve (area = 83 km 2 ) consists of a core Upper Jurassic limestone ridge 12 km long and 3-4 km wide, with maximum elevation of 550 m (for further details see Gaisler et al. (1996) and âtyrok et al. (1998)). Results Winter census List of records: Rhip 51 m, 33 f; Mbly 2 m, 3 f; Mema 11 m, 12 f; Paus 1 m Rhip 31 m, 13 f; Mmyo 2 f; Mbly 5 m, 9 f; Mema 1 m; Paus 1 f Rhip 40; Mema Rhip 40, Mbly 2 m, 5 f; Mema 3 m, 1 f Rhip 55 m, 27 f; Mbly 2 m, 4 f; Mema 6 m, 4 f Rhip 51 m, 16 f; Mmyo 1 m, 1 f; Mbly 3 f; Mema 3 m, 7 f Rhip 31 m, 19 f; Mmyo 1 f; Mbly 2 m, 2 f; Mema 1 f Rhip 4 m, 2 f; Mema 5 f; Paus 1 m Rhip 8 m, 7 f; Mbly 1 f; Mema 2 m; Mnat 1 m; Mdau 1 f; Paus 1m Rhip 1 f; Mema 2 f; Paus 1 f Rhip 3 m, 1 f; Paur 1 m, 1 f; Paus 1 f Rhip 21 m, 11 f; Mnat 1 m; Paus 1 m Rhip 17 m, 12 f; Mema 2 m, 3 f; Mbra 1 m Rhip 11 m, 3 f; Mmyo 1 m; Mema 1 f Rhip 16 m, 4 f; Paus 2 m R hip 16 m, 16 f Rhip 23; Mema 1 f Rhip 21; Mema 1 m Rhip 53; Mema 5; Mnat 1; Paus Rhip Rhip 36; Mmyo 1; Mnat 9; Mdau 4; Paur Rhip 48; Mmyo 1; Mema 3; Mnat Rhip 62; Mema 3

4 Rhip 116; Mema Rhip 97; Mema 4; Mnat 3; Mdau Rhip 123; Mdau Rhip Rhip 142; Mema Rhip 172; Mmyo 1; Mema 6; Mdau Rhip 208; Mmyo 2; Mema 11; Mnat 1; Mmys 1; Myotis sp Rhip 243; Mmyo 1; Mema 4; Mnat 7; Mdau Rhip 202; Mema 3; Mnat 3; Mdau 1; Paus Rhip 201; Mema 8; Mnat 5; Mdau 2; Paur Rhip 200; Mema 13, Mnat 1; Paur Rhip 214; Mema 6. Rhinolophus hipposideros was the most common bat recorded hibernating in the cave, with a mean of 83 individuals per survey for the whole monitoring period (Table 1). Myotis emarginatus was the second commonest, with an average of only four individuals per survey. With regard to other species, the biggest difference between early and subsequent years was the decline and disappearance of M. blythii. Up to 14 M. blythii per survey were recorded between , and up to four between After 1973 the species could no longer be found in the cave. The remaining species were recorded in low numbers (average one individual per survey, see Table 1). Changes in abundance The fluctuations in numbers of R. hipposideros and of all other bats combined (Fig. 1) are characterised by the following trends: (1) a decrease from , (2) a decrease from , and (3) an increase after 1983 until the end of monitoring. The first two decreasing trends are statistically insignificant due to small sample sizes (n = 3 and 6 years, respectively). In contrast, the increasing trend during the final years of monitoring is highly significant (p < 0.001). However, in species other than R. hipposideros numbers are too small to show any trends. Rhinolophus hipposideros hibernates either singly or in groups termed colonies. Fluctuations in total numbers are reflected in fluctuations within the largest winter colony. If present, this assemblage has invariably been situated at the same place in the so-called Bat Fig. 1. Fluctuations in the numbers of hibernating bats recorded in the Na Turoldu Cave, (y axis = number of bats, x axis = year). The arrow pointing downwards denotes the year the gate closing the cave was broken down, the arrow pointing upwards represents the year a new gate was installed, thus closing the cave once more. 4

5 Dome. At the beginning of research (in February 1958) the colony consisted of 40 individuals (see G aisler et al. 1990: plate 1). The colony disappeared in the early 1970s and was relocated in In 1989 it consisted of 61 individuals and continued to increase to 200 individuals by During the last survey (in January 2000) we counted 170 individuals. Possible reasons for the shifts in numbers observed, such as the breaking down of the gate closing the entrance to the cave in 1970 (Fig. 1) will be discussed. Summer netting and comparison with winter census During 31 netting sessions we banded 517 individuals from 13 species. In addition, netting resulted in 69 (13.4 %) recaptures of eight species. Three species were recorded at the locality only by netting, i.e. M. bechsteinii, E. serotinus and B. barbastellus, and two more, V. murinus and N. noctula, were detected by bat detector. During the netting on , both a display flight and the mating calls of V. murinus were recorded. The bat, obviously a male, flew in circles from till h (Central European Time) and uttered loud calls at about 15 khz in a regular sequence. Since bats were only netted between , the sample of comparable winter records was limited (Table 2). The differences between the winter and the summer samples included both data on species composition and data on the quantitative representation of each species recorded. The differences between the numbers of individuals of the eight species common to both were significant (χ 2 Goodness-of-fit test, p < 0.001). While about ten times more bats were recorded per one winter survey than per one summer netting, the overall species diversity was eight times higher in the summer than in winter. The winter sample was dominated by one superabundant species, R. hipposideros, which accounted for 95.5% of all records. In contrast, the quantitative representation of species was more balanced in summer, with M. nattereri, M. daubentonii, R. hipposideros and P. auritus each accounting for >10% of all records. The ranks of species according to their dominance values are completely different when the two samples are compared (Table 2). Table 2. Comparison of results of the winter census and summer netting, N = number of bats seen or netted; D = species dominance (%); Mean = mean number of bats per one census or netting (n = 10 winter checks, 31 summer nettings); H' = species diversity. Species Winter Summer N D Mean N D Mean R. hipposideros M. myotis M. mystacinus M. emarginatus M. nattereri M. bechsteinii M. daubentonii E. serotinus V. murinus <0.1 N. noctula <0.1 P. auritus P. austriacus B. barbastellus <0.1 Total H'

6 M ar k - r e c ap t ur e an al y sis During the entire time span , 1,038 bats of 15 species were banded at Turold (Table 3). Males were more commonly banded in all species with the exceptions of M. myotis and M. blythii. In the former, the two sexes were equally represented, but females dominated in the latter. Altogether, 237 recaptures were obtained, representing 22.8% of the total bats banded. Of these 161 (68%) were made in winter and 76 (32%) in summer. At least one recapture was obtained for nine species and 151 marked individuals. The recovery rate is higher for winter bats (30.9 %) than summer (14.7 %). There are 36 records of bat movements (15.2 % of all recaptures), including 26 concerning bats banded at Turold and recaptured outside this locality (15 R. hipposideros, 10 M. emarginatus, 1 M. blythii) and 10 of bats banded elsewhere and recaptured at Turold (6 R. hipposideros, 2 M. myotis, 1 M. blythii, 1 M. nattereri). Recapture data with respect to individual species are given below. Table 3. Survey of the number of all banded bats and of all recaptures, In the last column, the number of summer recaptures is in brackets. Species / Sex Winter Summer Total Recaptures males females males females banded R. hipposideros (23) M. myotis (1) M. blythii M. mystacinus M. brandtii M. emarginatus (15) M. nattereri (15) M. bechsteinii M. daubentonii (4) E. serotinus (1) V. murinus N. noctula P. auritus (13) P. austriacus (4) B. barbastellus Total (76) Rhinolophus hipposideros T/T recaptures: m W/1W 43 x, W/2W 14 x, W/4W 4 x, W/5W 2 x, W/6W 1 x, W/11W 1 x, W/12W 1 x, S/1W 12 x, S/2W 6 x, S/3W 5 x, S/4W 2 x, S/2S 1 x, S/3S 1 x; f W/1W 11 x, W/2W 6 x, W/3W 2 x, W/4W 1 x, W/7W 1 x, W/8W 1 x, W/12W 1 x, S/1W 5 x, S/2W 2 x, S/3W 1 x, S/4W 2 x, S/8W 1 x. In both males and females, the longest period between banding and recapture was 12 years and the individuals concerned were found in the cave (type W/W). While in type S/W the longest period was eight years in a female, and three years for an S/S type male, no recaptures of type W/S were made. There were 15 recaptures of type T/O - 10 of them W/S. The bats were recaptured in the summer of the same or subsequent year, mostly at Lednice or Bfieclav, up to 25 km east of the banding site. One record is exceptional: f, B , was repeatedly found hibernating in the cave on , and ; on it was pregnant and recaptured in a breeding colony in the loft of a building at Bulhary, Panensk ml n, (7 km NE). Another record represents the only recapture abroad: f, B

7 76, R Rabensburg, Austria, (27 km SE). The following recaptures of bats banded at Turold in summer are also worth mentioning: m, B , R artificial cave, Lednice, (13 km E); f, B, , R in a cellar at Janohrad, Lednice, (14 km E); f, B , R in a cellar at Insel, Mikulov, (4.5 km SES); m, B , R game reserve feeding trough, Klentnice, (3 km NE). Four recaptures of type O/T represent records showing the connection between the caves of the Moravian Karst and Turold: 2 m, B Jedovnické propadání Cave, , R Na Turoldu Cave, (58 km S); m, B Jedovnické propadání Cave, , R Na Turoldu Cave, (same distance, but after 19 years); f, B Ochozská jeskynû Cave, , R Na Turoldu Cave, (49 km S). Two O/T recaptures are of females who moved from breeding colonies to hibernacula: f, B Lednice, castle loft, , R Na Turoldu Cave, (13 km W); f, B Lednice, castle loft, , R Na Turoldu Cave, (same). The highest numbers of R. hipposideros were marked during winter checks in 1969 and On the former, 54 m and 27 f were banded, and 21 m and 12 f on the latter date. There were 30 recaptures (56%) of males banded in 1969 and recaptured in 1970, but only four (15%) of the females treated at the same time. This species has the second highest overall recovery rate (36.9%) i.e. 149 recaptures of 404 bats banded (Table 3). Myotis myotis Two recaptures of the O/T type: m, B B ãí skála Cave, Moravian Karst, , R Na Turoldu Cave, (53 km S); m, B Bulhary, netted in Milovice Forest, , R Turold, netted, (6 km W). As the second animal was recaptured only five weeks later (i.e. within the same season) the two localities probably belonged within its foraging range. The recovery rate was low (5.7%) for this species. Myotis blythii T/T recaptures: m W/3W 1 x, W/10W 2 x, W/11W 1 x, W/12W 1 x; f W/1W 3 x, W/2W 2 x, W/3W 1 x, W/9W 1 x, W/11W 1 x. The last three records of males and the last two of females concern only one animal each. The male was banded in 1959 and recaptured in 1969, 1970 and 1971 in the cave Na Turoldu, but it was also recaptured in Katefiinská Cave, Moravian Karst, (63 km N). The female was banded in 1959 and recaptured in 1968 and 1970 at the same place. There is one record of type O/T: f, B Erichova Cave, Moravian Karst, , R Na Turoldu Cave, (66 km S). Myotis blythii had the highest recovery rate (46.4%). Myotis emarginatus T/T recaptures: m W/1W 1 x, S/1S 1 x, S/2S 1 x, S/3S 1 x, S/7S 1 x, S/2W 1 x; f W/1W 1 x, S/2S 1 x. In addition to recaptures at the same locality, there are 10 cases of movements (T/O) - all of them concerning females banded in winter inside the cave and recaptured at summer breeding colonies at Lednice. The breeding colonies were situated in three different lofts of the central Lednice Castle, as well as in lofts of two smaller hunting castles up to 13 km away (Rybniãní and Tfii Grácie). One animal was checked six times: f, B , R Lednice, loft, , , , , It was immature in 1970 and either pregnant or lactating on subsequent recaptures. The overall recovery rate was 18 % in this species. Myotis nattereri T/T recaptures: m S/S (same season) 1 x, S/1S 8 x, S/2S 4 x, S/3S 1 x, S/4S 2 x, S/1W 2 x, S/2W 2 x, S/3W 1 x; f S/5W 1 x. 7

8 One movement (O/T) was documented between a forest habitat and a hibernaculum: m, B Pavlov, netted at Soutûska, , R Na Turoldu Cave, (5.5 km S). Myotis nattereri had a low recovery rate (11.6%), but there were large differences between the sexes: in males the recovery rate was 15.5%, but only one female was recaptured (1.8%). Myotis daubentonii T/T recaptures: m S/S (same season) 1 x, S/1W 1 x, S/1S 2 x; f S/1W 1 x. The lack of data is related to the very low recovery rate (5.1%). Eptesicus serotinus T/T recapture: f - S/2S 1 x. With only one recapture, the recovery rate for this species (3.4%) was the lowest recorded. Plecotus spp. T/T recaptures: P. auritus: m W/1S 1 x, S/1S 5 x, S/2S 3 x, S/4S 2 x, S/5S 1 x, S/2W 2 x; f S/3S 1 x. P. austriacus: m S/S (same season) 1 x, S/5S 1 x, S/6S 1 x, S/1W 4 x, S/5W 1 x; f W/2W 1 x, W/3W 1 x, S/1S 1 x. The recovery rates were high: 23.8% in P. auritus and 25.6% in P. austriacus. Discussion The results of our checks show that Na Turoldu Cave hosts large numbers of R. hipposideros but negligible numbers of bats of other species. This resembles, for example, Hermannshöhle in Austria (Baar et al. 1986) and certain caves in northern Moravia ( ehák & Gaisler 1999). Due to its exposed hibernation, R. hipposideros can be found relatively easily so that numbers ascertained at a hibernaculum are close to reality (Bezem et al. 1964). As the species is sedentary (Roer 1995, S chober & Grimmberger 1998) changes observed in its abundance at large hibernacula may reflect changes in the population density of the local population. This, however, does not seem to be the case for the decrease recorded at the beginning of our research. It seems most likely that the decreases from were the result of disturbance to the bats associated with banding (Reiter 1998). On the other hand, the decrease between was the result of disturbance by unauthorised human activity. In summer 1971, the gate closing the entrance to the cave was broken down and, during subsequent visits by intruders, the cave was devastated. The finding of only one R. hipposideros and a total of four bats in 1974 undoubtedly reflect this impact. The cave was closed and again put under protection in autumn The period (and probably until 1983, although there was no survey that year) can be viewed as representing the recovery of the hibernating assemblage, when the number of bats started to grow (Fig. 1). The absence of banding and other negative impacts upon hibernating bats seems to be only one reason for this positive trend. Even more obvious than the increase in the total numbers of R. hipposideros recorded in the cave, is the increase in the size of the winter colony from 40 in 1959 to 200 in 1996, i.e. five fold. This corresponds with data from certain other hibernacula situated close to the species northern border in Europe. After a general decrease in R. hipposideros numbers in the 1960s and 1970s (Daan et al. 1980, Ohlendorf 1995) numbers stabilised and started to increase (McAney 1994, ehák 1997, ehák & Gaisler 1999). This increase may reflect the successful breeding of local lesser horseshoe bat populations over the last fifteen years or so. Is the superabundance of R. hipposideros observed in Na Turoldu and some other caves real or an artefact of the sampling method? We do not believe that the large numbers of bats (e.g. M. nattereri, M. daubentonii or P. auritus) caught in summer in front of the cave (Table 2) do not 8

9 hibernate at the locality. These, and some other species netted at Turold, tend to hide in protected places when hibernating, and so are difficult to find (Bezem et al. 1964, S c hober & Grimmberger 1998). Baagoe et al. (1988) found that the number of M. daubentonii recorded departing from a Danish hibernaculum in spring was much greater than the number of hibernating individuals of the same species recorded by a visual census inside the same quarters in winter. Similar results were obtained in a study of M. nattereri and M. daubentonii wintering in a bunker in Poland (Jurczyszyn 1998). In both species the number of departing bats netted in late winter and spring was higher (nearly two times so in some years) than the number of bats seen inside. We therefore suspect that an unknown, yet considerable, number of bats netted outside the cave hibernate inside it. If this is the case, hibernation must occur either in protected places inaccessible to an observer, and/or they hibernate in deep slots in the rocks near to the outside the cave, which also makes them difficult to locate. If this is true, then the actual species diversity of the hibernating bat community must be greater (and the dominance of R. hipposideros lower) than that revealed from the winter samples. Irrespective of this rather speculative conclusion, the number of 15 species (nearly the half of European chiropterofauna) recorded at the locality is high and warrants Turold s the status as a nature reserve. The results of marking and recapturing bats are generally in agreement with those from previous bat banding studies in the former Czechoslovakia (H anák et al. 1962, Gaisler & Hanák 1969) - for example, the finding that R. hipposideros moves up to 27 km between summer and winter quarters. Similar movements between summer nursing quarters and a hibernaculum up to 13 km away have also been recorded in M. emarginatus (cf. Gaisler et al. 1989). New findings, however, are the movements between Turold and the Moravian Karst, i.e. between winter quarters up to 66 km, as recorded in R. hipposideros, M. myotis and M. blythii. A male R. hipposideros, banded in Moravian Karst area, was recaptured at Na Turoldu Cave after 19 years. As this individual must have been born in the summer prior to marking, its age at recapture was at least 19.5 years. Several recaptures of R. hipposideros and one in each M. myotis and M. nattereri document movements between a shelter and a foraging area or between two foraging areas at distances of up to 6 km. The overall recovery rate for banded bats of 22.8% is high (Hanák et al = 8.1%, Gaisler & Hanák 1969 = 9.1%, Hanák et al = 9.7%). Recovery was higher in winter (30.9%) than in summer (14.7%) because of the much higher representation of R. hipposideros in the winter samples. If we omit M. blythii, the actual status of which is unclear (cf. Gaisler et al. 1988), R. hipposideros has the highest recovery rate of 36.9%. This results not only from its exposed hibernation, but also from its sedentary way of life. Sedentary Plecotus spp. also have high recovery rates. R eiter (1998) analysed recapture data at a locality where R. hipposideros was absent and found recovery rates of 20 25% in Plecotus, Barbastella and Eptesicus, while the recovery rates of Myotis spp. were lower. Masing et al. (1999) showed that second year recapture rates usually varied between 20 50% in sedentary bat species in Estonia, but that values were significantly lower in migratory species. We have only few records of bats banded in winter and recaptured in summer (or vice versa) at Turold. This may (in part?) be due to the different times at which we banded bats: hibernating bats were only banded in , and bats netted outside the cave only in Another source of error may be the potentially lower trapability of rhinolophids compared to vespertilionids in summer due to their echolocation differences. The bias makes it difficult to assess differences in interseasonal recovery rates for individual species, which might contribute to elucidating the problem of their real representation at the locality in different times of year. 9

10 A c k n o w l e d g e m e n t s We thank the following colleagues and friends who took part in the fieldwork, the results of which are subject of this paper. In alphabetic order (titles and first names omitted): BaroÀ, Bartáková, Bauerová, Beljanin, Gajdo ík, Hanák, Hoat, Chytilová, KaÀuch, Klima, Koutn, Macháãek, Málek, Málková, Matu ka, Nepra, Piro, Rödl, ehák, evãík, Tau, Vla ín, Zejda, Zima and Zukal. Beljanin is Russian, Hoat is Vietnamese, KaÀuch is Slovak and Klima, though a native Czech, has German citizenship. The other people are Czech citizens and mostly live in Brno or at Mikulov. We thank also V. Jarsk for providing mark-and-recapture data on bats from the database of the Czech Agency for Nature Conservation (Prague), and two anonymous reviewers for their comments. The research was funded by GA CR Grant No 206/99/1519. L I T E R A T U R E BAAGOE, H. J., DEGN, H. J. & NIELSEN, P., 1988: Departure dynamics of Myotis daubentoni (Chiroptera) leaving a large hibernaculum. Vidensk. Meddr. dansk naturh. Foren., 147: BAAR, A., MAYER, A. & WIRTH, J., 1986: 150 Jahre Fledermausforschung in der Hermannshöhle. Ann. Nathist. Mus. Wien B, 88-89: BAUEROVÁ, Z., GAISLER, J., KOVA ÍK, M. & ZIMA, J., 1989: Variation in numbers of hibernating bats in the Moravian Karst: results of visual censuses in In: Hanák, V., Horáãek, I. & Gaisler, J. (eds.), European bat research Charles Univ. Press, Praha: BEZEM, J. J., SLUITER, J. W. & van HEERDT, P. F., 1964: Some characteristics of the hibernating locations of various species of bats in South Limburg. Proc. Koninkl. Nederl. Akad. Wet., Amsterdam, 67: âtyrok, P., DANIHELKA, J., CHYTIL, J., STUCHLÍK, J. & VIDLÁK, S., 1998: Pfiírodní rezervace Turold [Nature reserve Turold]. ARC Mikulov, 24 pp. (in Czech). DAAN, S., 1980: Long term changes in bat populations in the Netherlands: a summary. Lutra, 22: GAISLER, J., BAUEROVÁ, Z., VLA ÍN, M. & CHYTIL, J., 1988: The bats of S-Moravian lowlands over thirty years: Rhinolophus and large Myotis. Folia Zool., 37: GAISLER, J. & HANÁK, V., 1969: Ergebnisse der zwanzigjährigen Beringung von Fledermäusen (Chiroptera) in der Tschechoslowakei: Acta Sc. Nat. Brno, 3 (5): GAISLER, J., CHYTIL, J. & VLA ÍN, M., 1990: The bats of S-Moravian lowlands (Czechoslovakia) over thirty years. Acta Sc. Nat. Brno, 24 (9): GAISLER, J., VLA ÍN, M. & BAUEROVÁ, Z., 1989: The bats of S-Moravian lowlands over thirty years: small Myotis. Folia Zool., 38: GAISLER, J., ZUKAL, J., NESVADBOVÁ, J., CHYTIL, J. & OBUCH, J., 1996: Species diversity and relative abundance of small mammals (Insectivora, Chiroptera, Rodentia) in the Pálava Biosphere Reserve of UNESCO. Acta Soc. Zool. Bohem., 60: HANÁK, V., GAISLER, J. & FIGALA, J., 1962: Results of bat-banding in Czechoslovakia, Acta Univ. Carolinae Biol., 1962 (1): HANÁK, V., REITER, A. & BENDA, P., 1996: Pfiehled terestrick ch obratlovcû Ledov ch slují (The review of terrestric vertebrates of the Ledové sluje area). Pfiíroda sbor. prací z oboru ochr. pfiír., Praha, 3: (in Czech, with a summary in English). JURCZYSZYN, M., 1998: The dynamics of Myotis nattereri and M. daubentoni (Chiroptera) observed during hibernation season as an artefact in some type of hibernacula. Myotis, 36: MASING, M., POOTS, L., RANDLA, T. & LUTSAR, L., 1999: 50 years of bat-ringing in Estonia: methods and the main results. Plecotus et al., Moskva, 2: McANEY, C. M., 1994: The lesser horseshoe bat in Ireland past, present and future. Folia Zool., 43: OHLENDORF, B. (ed.), 1995: Zur Situation der Hufeisennasen in Europa. Tagungsband, Nebra, Deutschland, 182 pp. REITER, A., 1998: Is banding a real threat to bats? Vespertilio, Revúca, Praha, 3: ROER, H., 1995: 60 years of bat-banding in Europe results and tasks for future research. Myotis, 32-33: EHÁK, Z., 1997: Long-term changes in the size of bat populations in Central Europe. Vespertilio, Revúca, Praha, 2: EHÁK, Z. & GAISLER, J., 1999: Long-term changes in the number of bats in the largest man-made hibernaculum of the Czech Republic. Acta Chiropterologica, 1: SCHOBER, W. & GRIMMBERGER, E., 1998: Die Fledermäuse Europas. Kosmos, Stuttgart, 265 pp. 10

11 Folia Zool. 51(1): (2002) Diet estimation by faeces analysis: sampling optimisation for the European hare Krisztián KATONA 1 and Vilmos ALTBÄCKER 2 1 St.Stephen University, Department of Wildlife Biology and Management, Páter K. u. 1, H-2103 GödöllŒ, Hungary; katonak@ns.vvt.gau.hu 2 Eötvös Loránd University, Department of Ethology, Jávorka u. 14, H-2131 Göd, Hungary; altbac@ludens.elte.hu Received 20 June 2001; Accepted 30 October 2001 Abstract. We investigated how the sampling process of microhistological faeces analysis could be optimised for an accurate estimation. Spring diet composition of European hare (Lepus europaeus Pallas, 1778) was determined in a juniper shrubland at Bugac, Hungary. Both inter and intraobserver reliability was high permitting us to separate the components of variance due to the methodological steps in the faeces analysis. Estimates varied depending on the number of independent droppings, pellets/individual, subsamples/pellet and epidermis/subsample. The variance was much higher among than within the independent pellet groups. The cumulative frequency estimate stabilised at around 100 epidermis fragments per pellet. We conclude that the most critical steps of the sampling procedure are the collection of independent droppings and the identification of a sufficient number of epidermis fragments. We propose to collect at least 10 independent droppings, one pellet/individual, and analyse 100 epidermis fragments as an optimum for estimating the relative frequency of forage classes reliably. The importance of the individual variability in the diet should be emphasised. Key words: microhistology, faecal pellet, accuracy, minimum sample size, Lepus europaeus Introduction Feeding habits of mammals are in the centre of interest of population biology (Lodé 1996) and ecology (Mátrai et al. 1998). Several methods have already been involved in the investigation of food composition (Holechek et al. 1982). The most widely used indirect technique for determining diet composition of herbivores is the microhistological identification of epidermis fragments in the stomach content or faecal pellet (B a u m g a r t n e r 1939, Dusi 1949). The effect of sample size on the estimated diet composition has already been proved (Hanson & Graybill 1956). The greater the diet diversity of a species and the smaller the similarity of individuals, the larger the required minimum sample size is (K o v á c s & T ö r ö k 1997). Data on individual variability in the diet of herbivorous mammals, however, are very scarce (M átrai & Kabai 1989, H omolka & Heroldová 1992). Microhistological faeces analysis method includes multiple successive sampling from the individuals, pellets and epidermis fragments. Sampling size, therefore, could affect the estimate in all consecutive sampling steps. The lack of the information on the precision of the faecal sampling has already been emphasised (Holechek et al. 1982). The applied minimum sample size could vary according to study species, as for deer (A nthony & Smith 1974), hare (H omolka 1987a) and rabbit (C hapuis 1980). Sample size, 11

12 however, could also differ from study to study as the characteristics and size of study area change (Hanson & Graybill 1956, Homolka 1987b). Problems to design a valid diet composition analysis derive from the difficulties of the determination of the sample sizes required. To optimise the sampling process, a general knowledge on the variances of different sampling levels should be established. For this comparison a single basic sample of faecal pellets is needed, which has never been used by earlier investigators studying the minimum sample sizes on consecutive sampling levels. The aim of our investigation was to assess a valid sampling strategy to obtain precise results. In order to optimise faecal sampling, therefore, we compared the variances observed at the consecutive steps of faeces analysis in European hare (Lepus europaeus Pallas, 1778). Materials and Methods The study area (1200 by 1400 m, 168 ha) was situated in the Bugac Juniper Forest core area in Kiskunság National Park, central Hungary (46 38 N; E). This protected area has shrubland vegetation (for a detailed description of it see Kertész et al. (1993) and Kalapos (1989)). Fresh faecal pellets of European hare were collected on a single day, 22 April, Altogether 10 independent droppings, each containing five pellets, were collected. (Faeces of European hare is called droppings, which could include only one or several faecal pellets). Sampling points were separated by at least 200 m intervals (x ± SD = 578 ± 272 m). Spring diet composition of hares was determined by microhistological faeces analysis. Proportion of diet components was estimated in each pellet by the number of fragments for a particular forage class relative to the total number of fragments. The distinguished forage classes were monocotyledons (grasses), dicotyledons (forbs and browses), gymnosperms (evergreens), barks and seeds. For microhistological analysis the following preparation procedures were conducted (Mátrai et al. 1989). A subsample mixed from five parts of g was taken out from each pellet into test tubes. From five pellets of different droppings a subsample of one part of 0.01 g was taken parallely out into five other test tubes. After boiling in 20 % nitric acid solution for 1.5 minutes epidermis fragments were dispersed on microscopical slides into a mixture of 0.1 ml of 87 % glycerine and 0.05 ml of 0.1 % Toluidine-Blue. All fragments found on the slides were identified under 160X magnification. We had the following investigations. A) One pellet from each droppings was chosen and 200 epidermis fragments of it were identified. Just the minimum required number of fragments (100) were then categorised from the remaining pellets. Minimum sample sizes for different sampling units were determined by rarefaction analyses (Kovács & Török 1997) and by saturation curves of the proportion of forage categories. B) Intra and interobserver reliability were determined by analysing one pellet from each droppings two times. Statistical analysis of observer reliabilities was performed by Spearman-correlation (Caro et al. 1980). C) Diet composition estimated from the same pellets by subsamples including one bigger or five smaller parts were compared by Wilcoxon matched-pairs signed rank test. D) Determining diet composition of all five pellets of five droppings variances within and among droppings were tested by Kruskal-Wallis one-way analyses of variance by ranks. 12

13 To quantify diet overlaps between different samples Renkonen s proportional similarity index was always calculated (H urlbert 1978): S is = Σmin(P 1,i ; P 2,i ) where P 1,i is the proportion of prey category i in one individual, P 2,i in the other individual. Similarity and diversity indices and rarefaction curves were calculated using the software NICHE (Schluter, D., University of British Columbia, Vancouver). Results There was no significant negative correlation between the distance of sampling points of droppings and their diet similarity (Spearman-correlation: N=45, r s =0.24, p=0.12) supporting the independence of droppings. A) Rarefaction analyses and saturation curves indicated that a sample size of 100 epidermis fragments (Fig. 1), 10 independent droppings and one pellet of a single droppings were surely sufficient for obtaining unbiased estimates of diet composition. B) Intra and interobserver reliability were measured to be high (Spearman-correlation: N=10, r s >0.8 in all cases for monocots, dicots and gymnosperms; r s >0.6 for seeds and barks). C) Diet composition estimated from either a single 0.01 g subsample or from a mixture of five g subsamples seemed to be identical (Wilcoxon matched-pairs test: N=5, p>0.38 for all forage classes). D) Based on the analysis of all five pellets of five droppings, diet composition of pellets significantly differed among droppings (Kruskal-Wallis test: df=4, p<0.05 for all forage classes) (Fig. 2), but within droppings they were identical (Kruskal-Wallis test: df=4, p>0.7 for all forage classes). In the course of sampling procedure similarity was the lowest, therefore variance was the highest among independent droppings as shown in Table 1. Fig. 1. Stabilisation curves of epidermis fragments identifying to broad forage classes. Figure shows typical curves to determine an optimum sample size of the epidermis fragments based on the analysis of 200 fragments to broad forage classes in a single pellet. Relative occurrence was given by the number of fragments for a particular forage category relative to the total number of fragments. Confidence interval of five percent is indicated. 13

14 Fig. 2. Variability in the diet composition among independent hare droppings. Diet composition of a single pellet from each droppings is provided after identifying 200 epidermis fragments per pellet to broad forage classes. Relative occurrence was given by the number of fragments for a particular forage category relative to the total number of fragments. Table 1. Variability in the sampling procedure of microhistological faeces analysis. Sample size (N), value of Renkonen s proportional similarity index (Sis) and statistical significance of differences for the given sampling factors are shown. Sampling variability Sample size (N) Proportional similarity (Sis) Significance Among droppings p<0.05 Within droppings NS Between subsamples NS Interobserver reliability NS Observer1 reliability NS Observer2 reliability NS Discussion Our results suggest that sampling procedure could significantly affect the result of microhistological analysis. In our study, the largest contribution to overall variance was revealed among droppings, indicating considerable individual differences among the hares. Mátrai & Kabai (1989) and Homolka & Heroldová (1992) have already proved considerable individual differences in deer species. We have also found a low similarity among independent droppings of hares, but cumulative composition analysis indicated that 10 droppings gave a reliable estimate. This is very similar to the minimum sample size suggested by Homolka (1987a) for hares and by Chapuis (1980) for rabbit colonies inhabiting a well-defined smaller habitat. This individuality of diet composition casts doubt on the validity of many diet analyses done from a faecal pellet mixture. These investigations might reveal the pressure on the vegetation by the given herbivore population rather than the actual food choice of the individuals. Similarity of the pellet compositions was much higher within than among droppings, similarly to Chapuis (1980), suggesting the importance to collect more droppings and only one pellet/individual. 14

15 Number of identified fragments should be the other main factor of sampling design as Homolka (1987a) and C hapuis (1980) have already pointed out. There is no consensus in the literature on the minimum sample size, but it often seems to be larger than necessary. In rabbits in our study area Mátrai et al. (1998) identified fragments in different seasons. C hapuis (1980) applied a sample size of 400 fragments, but estimated 200 to be sufficient. In summary, the different steps in the sampling procedure influence the accuracy in different ways. The two major factors in faecal sampling optimisation are the sample size of droppings and epidermis fragments. Estimation of diet composition is less sensitive to the observer reliabilities and to within-droppings and within-pellet variance. We are confident that our directives will help to design future faecal analyses to gain accurate and biologically interpretable results with optimal effort and without underestimating the variety of dietary habits of animals. Acknowledgements We are indebted to K. Mátrai for her help in applying microhistological techniques and to K. Nagy for her participation in the interobserver reliability study. Two anonymous referees made constructive comments on the manuscript, while L. Sári revised the final English version. Kiskunság National Park permitted us to work in the protected study area. The work was funded by the Hungarian Science Foundation (T to V. Altbäcker). LITERATURE ANTHONY, R. G. & SMITH, N. S., 1974: Comparison of rumen and faecal analysis to describe deer diets. J. Wildl. Manage., 38(3): BAUMGARTNER, L L. & MARTIN, A. C., 1939: Plant histology as an aid in squirrel food-habits studies. J. Wildlife Manage., 3: CARO, T. M., ROPER, R., YOUNG, M. & DANK, G. R., 1980: Inter-observer reliability. Behaviour, 69 (3 4): CHAPUIS, J. L., 1980: Méthode d étude du régime alimentaire du lapin de garenne, Oryctolagus cuniculus (L.) par l analyse micrographique des fèces. Rev. Ecol.- Terre Vie, 34: (in French, with a summary in English). DUSI, J. L., 1949: Methods for the determination of food habits by plant microtechniques and histology and their application to cottontail rabbit food habits. J. Wildlife Manage., 13(3): HANSON, W. R. & GRAYBILL, F., 1956: Sample size in food-habits analyses. J. Wildlife Manage., 20(1): HOLECHEK, J. L., VAVRA, M. & PIEPER, R. D., 1982: Botanical composition determination of range herbivore diets: a review. J. Range Manage., 35(3): HOMOLKA, M., 1987a: Problems associated with investigations to the diet of the European hare. Folia Zool., 36(3): HOMOLKA, M., 1987b: The diet of brown hare (Lepus europaeus) in central Bohemia. Folia Zool., 36(2): HOMOLKA, M. & HEROLDOVÁ, M., 1992: Similarity of the results of stomach and faecal contents analyses in studies of the ungulate diet. Folia Zool., 41(3): HURLBERT, S. H., 1978: The measurement of niche overlap and some relatives. Ecology, 59(1): KALAPOS, T., 1989: Drought adaptive plant strategies in a semiarid sandy grassland. Abstracta Botanica, 13: KERTÉSZ, M., SZABÓ, J. & ALTBÄCKER, V., 1993: The Bugac Rabbit Project. Part I: Description of the study site and vegetation map. Abstracta Botanica, 17: KOVÁCS, T. & TÖRÖK, J., 1997: Determination of minimum sample size to estimate diet diversity in anuran species. Herpetol. J., 7: LODÉ, T., 1996: Polecat predation on frogs and toads at breeding sites in western France. Ethol. Ecol. Evol., 8: MÁTRAI, K. & KABAI, P., 1989: Winter plant selection by red and roe deer in a forest habitat in Hungary. Acta Theriol., 34(15): MÁTRAI, K., ALTBÄCKER, V. & HAHN, I., 1998: Seasonal diet of rabbits and their browsing effect on juniper in Bugac Juniper Forest (Hungary). Acta Theriol., 43(1):

16 Folia Zool. 51(1): (2002) Aggression in reciprocal crosses of two subspecies of wild house mouse Radka VOLFOVÁ, Pavel MUNCLINGER and Daniel FRYNTA Department of Zoology, Faculty of Science, Charles University, Viniãná 7, Praha 2, Czech Republic; Received 25 September 2000; Accepted 9 November 2001 Abstract. We studied reciprocal hybrids of Mus musculus musculus and M. m. domesticus. These two subspecies of house mouse were found to differ in their social behaviour, the former being less aggressive than the latter. The paternal effect on aggression (observed repeatedly in laboratory mice) was not found. However, F 2 generation mice were less aggressive than those from the F 1 generation, and the maternal effect was also significant in a homogenous test set. Key words: wild mice, hybrid zone, social behaviour, hybrid advantage, Y chromosome Introduction Aggression is an important component of mouse social behaviour. The level of aggression is species specific and may vary, even between populations or strains (B rain & Parmigiani 1990, Ganem & Searle 1996, Sluyter et al. 1996b). At least in part, these differences are determined by genotype, as has been made clear from model experiments on laboratory mice (Cairns et al. 1983, Sluyter et al. 1996b). It has been repeatedly demonstrated that artificial selection has a strong and immediate effect on measures of aggression recorded in standardised tests (H ood 1988a, Hood 1988b, Sluyter et al. 1996b). Special attention has been paid to paternal effects, which have been repeatedly reported in studies quantifying mouse aggression. In their study dealing with reciprocal crosses of strains DBA/1/Bg and C57BL/10/Bg, Selmanoff et al. (1975) hypothesised that the Y chromosome was responsible for the variation observed in mouse aggression. Y chromosome correlates on aggression were subsequently shown in several inbred strains and in lines bidirectionally selected for attack latency (see review by Carlier et al. 1990). Moreover, some comparisons of congenic strains confirmed the Y chromosome effect on aggression - usually the more aggressive hybrid has the Y of the more aggressive strain (Guillot et al. 1995, Carlier et al. 1990, Mohanan & Maxson 1998). Clear, fixed Y chromosome differences have been found between the subspecies Mus musculus musculus L., 1758 and Mus musculus domesticus Schwarz & Schwarz, 1943 (Boissinot & Boursot 1997) and it is of interest that these subspecies also differ in their social behaviour. The phenomenon whereby M. m. domesticus males are more aggressive than those of M. m. musculus has been shown by authors studying mouse populations from the north-western part of Europe (Thuesen 1977, Zegeren & Oortmersen 1981). We have also demonstrated the same phenomenon by comparing the social behaviour of house mice from two distant populations (i.e. Bohemia and eastern Turkey), each of which belong to different subspecies (M u n clinger & Frynta 2000). Male mice from Turkey (M. m. domesticus) were undoubtedly more aggressive than Bohemian mice (M. m. musculus). 17

17 In this paper we have attempted to demonstrate the differences between reciprocal F 1 males from crosses of two subspecies of wild house mouse to prove the paternal effect on aggression; this is the obligatory first step for testing the Y chromosome effect. In addition, we also compared their descendants (F 2 hybrids) to randomise the effects of autosomal genes. Material and Methods Parental generation mice were born in the laboratory (about 4 th - 6 th out-bred generation in captivity). Mus m. musculus stock were derived from 20 males and 20 females captured in the Bohemian villages of âerno ice, Soutice and Satalice (Czech Republic) during winter Mus m. domesticus were descendants of eight pairs of mice captured in the vicinity of the town of Kilkis, Kilkis District (Greece) in May Reciprocal hybrids of the above populations were used as experimental subjects. Twenty males of each combination of generation (F 1 and F 2 ) and cross direction, i.e. female M. m. domesticus x male M. m. musculus (hereafter MdMmF 1 males) and female M. m. musculus x male M. m. domesticus (hereafter MmMdF 1 males) were used. Experimental mice were adult (i.e. at least two months old), socially experienced males housed in heterosexual pairs (i.e. together with an adult female of the identical filial generation and cross direction). All animals were kept under an artificial 12 L: 12 D light cycle and housed in pairs in 30 x 15 x 15 cm plastic cages. Water and food (ST1 mouse and rat breeder diet, wheat etc.) were provided ad libitum. Each cage contained sawdust bedding, nesting material (paper) and shelters. The experiments were performed in January-March A standard neutral cage procedure was used. Encounters between mice were carried out in a 50 x 30 x 35 cm glass cage, divided into two equal parts by a thick card partition. During testing, the cage was illuminated by a single 40 W red light bulb. Mice were tested during the dark phase of their light-dark cycle. At the beginning of each experimental session, two mice were placed in the pen on the opposite sides of the partition, and left for five minutes. The central partition was then removed and video recording by a single VHS camera commenced. The video camera was stopped at the end of the session (ten minutes after the moment at which one or both animals paid attention to the other for the first time). The cage was thoroughly cleaned using 96% ethanol after each session. Repeated tests on the same individual were undertaken at a minimum interval of 24 hours. In the first set we tested mice in homogenous dyads, i.e. with opponents of the same generation and cross direction. In this set 120 dyadic encounters were staged (i.e. 30 encounters for every combination of generation and cross direction). Each animal was tested with different opponents three times. In the second set, the same individuals were tested in heterogeneous dyads consisting of individuals belonging to the same generation, but different cross direction. 20 and 17 heterogeneous dyads were performed on the F 1 and F 2 generations, respectively. The video records of the encounters were subsequently observed, and the duration of behavioural elements was quantified using the computer program package ACTIVITIES (Vrba & Donát 1993). We used the standard catalogue of 34 behavioural elements for the purpose of data collection (see â iháková & Frynta 1996 for details). Selmanoff et al. (1976) and M a x s o n et al. (1979) suggested the any aggression score as being a more robust and reliable index of fighting behaviour than any single 18

18 agonistic act. Therefore, only the following four variables were derived from the primary data and used in further analyses: 1. Aggressive behaviour the sum of the time intervals spent in the following elements of aggressive behaviour: Threat, Aggressive upright, Attack, Chase, Roll-over fight. 2. Defensive behaviour the sum of the time intervals spent by defensive elements: Defensive upright, Defensive threat, Avoid, Retreat, Flee, Jump avoid, Freeze, Submissive posture. 3. Agonistic behaviour the total time spent in aggressive, defensive, and neutral (Neutral upright, Box, To-fro - repeated approach and avoidance, Tail rattling) elements of behaviour. We believe this variable is the best indicator of hostile motivation in interacting mice. This well-defined and continuous variable shows nearly log-normal distribution, so we used it as first choice in our testing. 4. Attack latency latency of the first attack expressed in seconds. This variable was highly correlated with time spent in agonistic behaviour (Spearman s r = -0.78, n = 157). The behaviour of both interacting animals in a particular dyad is obviously intercorrelated. Thus, in the first set of experiments dealing with comparisons of homogenous dyads we summed the scores of agonistic and aggressive behaviour recorded in both members of a single dyad. These were log-transformed and further treated by analyses of variance/covariance (Statgraphics v. 5.0). The scores for defensive behaviour as well as attack latency were treated by non-parametric statistics. Defensive behaviour was used solely for comparisons within heterogeneous dyads and analysed with Wilcoxon matchedpair tests. Results The mean time spent in agonistic behaviour in the first set of experiments is shown in Fig. 1. The proportion of time spent in agonistic behaviour by MdMmF 1 males (F 1 generation = 22%, F 2 = 12%) was higher than in MmMdF 1 males (F 1 generation = 11%, F 2 = 4%). Fig.1. Average total time spend in agonistic behaviour by homogenous dyads. The results of parental population tests are shown for comparison. Key: Md M. m. domesticus males, Mm - M. m. musculus males, F1: MdMm, F1: MmMd, F2: MdMm, F2: MmMd MdMm different generations and types of hybrids, see Material and Methods for more details. 19

19 Analysis of variance, as controlled for body weight (covariate F = 8.59, p = ), revealed significant differences in the time spent in agonistic behaviour between reciprocal hybrids (F = 18.27, p < ) and between the F 1 and F 2 generations (F = 6.51, p = ). The effect of interaction between the above factors was also important (F = 5.63, p = ). The differences between reciprocal hybrids were also supported when aggressive behaviour was treated separately (F = 3.99, p = ). Attack latency was not affected by the cross direction in the F 1 generation (Mann- Whitney test: z = -1.29, p = ) and only a marginal effect was found in the F 2 (z = -2.05, p = ). Attack latency was considerably longer in both MmMdF 1 and MdMmF 1 hybrids in the F 2 than in the F 1 generation (cross direction pooled: z = -3.84, p = ; average rank = 72.0 and 49.0, respectively). Unlike homogenous set tests, further experiments involving both reciprocal hybrids in a single dyad enabled matched pair comparisons. No effect of reciprocal hybridisation was found, either on the time spent in defensive behaviour (Wilcoxon matched-pair tests: z = , p = , n = 37 dyads, F 1 and F 2 generations pooled) or on attack latency (z = 0.28, p = ). The F 1 generation also remained more agonistic than the F 2 generation in this experiment (ANOVA: F = 5.79, p = ). Discussion We did not find any paternal effect on agonistic behaviour in male mice; moreover, we found a significant maternal effect. Surprisingly, the more aggressive hybrid did not carry the Y chromosome of the more aggressive subspecies. Studies carried out on laboratory mice have not led to consensus on how the Y chromosome determines the level of aggression in males. Roubertoux et al. (1994) failed to demonstrate the effect of the non-pseudoautosomal region of the Y on mouse aggression. Other studies have suggested influences from the pseudoautosomal region, genetic background and the maternal environment (Carlier et al. 1991, Roubertoux et al. 1994). However, there is some evidence that artificial selection for male aggression leads to a correlated response in female aggression (H ood & Cairns 1988). This finding provides indirect evidence for the effect of genes not located on the Y chromosome, because the Y chromosome is absent in females. To simulate natural conditions we kept mice in pairs, and the offspring had equal chances to be influenced by their fathers and their mothers. In spite of this, we found a significant maternal effect in the first set of experiments. We have no cross-fostering data to demonstrate that the differences between reciprocal hybrids are of genetic nature, and therefore prenatal and/or postnatal environmental effect cannot be excluded. Such environmental effects were previously revealed by studies of laboratory mice (Carlier et al. 1991, but see Sluyter et al. 1995, 1996a). It should be noted that aggression is a context-specific phenomenon (e.g. Suchomelová & Frynta 2000) and therefore details of experimental procedure can considerably influence results. In order to reduce stress, we studied mice in a standard neutral cage environment, i.e. outside the context of territoriality in which mice fight more seriously (G r a y & Hurst 1995), and we avoided the use of inbred laboratory mice as standard opponents. We suppose that this design is simply not applicable in wild mouse studies. Moreover, Guillot et al. (1995) showed that the standard opponent test is less 20

20 sensitive to parental (Y chromosome) effects than homogenous set tests. Nevertheless, social behaviour is determined by both interacting animals, so it is not surprising that the results of our first and second set were not the same. Unlike the total time spent in agonistic behaviour, attack latency was found to be only slightly effected by cross direction even though these two variables were highly correlated. This may suggest that the total time spent in agonistic behaviour is a more sensitive indicator of hostility than attack latency is. Despite differences between reciprocal hybrids, all the F 1 hybrids were highly aggressive and their mean scores for agonistic behaviour (201 s) even exceeded those previously recorded in the more aggressive (M. m. domesticus) parental population (191 s). We suggest that the fact that the F 1 hybrids consistently exhibited more agonistic behaviour than the F 2 in both experimental sets may be explained by recombination of the loci affecting agonistic behaviour. We conclude that paternal effects are not easily demonstrable in our wild mice crosses. The differences observed between reciprocal hybrids, as well as between F 1 and F 2 generations, in their social behaviour cannot be attributed to the Y chromosome as a single factor, and the possible effects of genes on other chromosomes, hybrid advantage, and even non-genetic maternal effects cannot be rejected. Acknowledgements We thank Dr. Radka D a n d o v á and Mr. Milan Kaftan for technical assistance in keeping the experimental animals. The project was supported by Ministry of Education, Youth and Sport of the Czech Republic (grant VZ ). The participation of P.M. was also supported by Ministry of Education, Youth and Sport of the Czech Republic (grant no. VS 97102). LITERATURE BOISSINOT, S. & BOURSOT, P., 1997: Discordant phylogeographic patterns between the Y chromosome and mitochondrial DNA in the house mouse: selection on the Y chromosome? Genetics, 146: BRAIN, P. F. & PARMIGIANI, S., 1990: Variation in aggressiveness in house mouse populations. Biol. J. Linn. Soc., 41: CAIRNS, R. B., MACCOMBIE, D. J. & HOOD, K.E., 1983: A developmental-genetic analysis of aggressive behaviour in mice: I. behavioural outcomes. J. Comp. Psychol., 97: CARLIER, M., ROUBERTOUX, P. L., KOTTLER, M. L. & DEGRELLE, H., 1990: Y chromosome and aggression in strains of laboratory mice. Behav. Genet., 20: CARLIER, M., ROUBERTOUX, P. L. & PASTORET, C., 1991: The Y chromosome effect on intermale aggression in mice depends on the maternal environment. Genetics, 129: âiháková, J. & FRYNTA, D., 1996: Intraspecific and interspecific behavioural interactions in the wood mouse (Apodemus sylvaticus) and the yellow-necked mouse (Apodemus flavicollis) in a neutral cage. Folia Zool., 45: GANEM, G. & SEARLE, J. B., 1996: Behavioural discrimination among chromosomal races of the house mouse (Mus musculus domesticus). J. Evol. Biol., 9: GRAY, S. & HURST, J.L., 1995: The effects of cage cleaning on aggression within groups of male laboratory mice. Anim. Behav., 49: GUILLOT, P.-V., CARLIER, M., MAXSON, S.C. & ROUBERTOUX, P.L., 1995: Intermale aggression tested in two procedures, using four inbred strains of mice and their reciprocal congenics: Y chromosomal implications. Behav. Genet., 25: HOOD, K. E., 1988a: Selective breeding - selective rearing interactions and the ontogeny of aggressive behavior. Behav. Brain Sci., 11:

21 HOOD, K. E., 1988b: Female aggression in mice: development, social experience, and the effects of selective breeding. J. Comp. Psychol., 2: HOOD, K. E. & CAIRNS, R. B., 1988: A developmental-genetic analysis of aggressive behavior in mice: II. Cross-sex inheritance. Behav. Genet., 18: MAXSON, S. C., GINSBURG, B. E. & TRATTNER, A., 1979: Interaction of Y-chromosomal and autosomal gene(s) in the development of intermale aggression in mice. Behav. Genet., 9: MOHANAN, E. J. & MAXSON, S. C., 1998: Y chromosome, urinary chemosignals, and an agonistic behaviour (offense) in mice. Physiol. Behav., 64: MUNCLINGER, P. & FRYNTA, D., 2000: Social interactions within and between two distant populations of house mouse. Folia Zool., 49: 1-6. ROUBERTOUX, P. L., CARLIER, M., DEGRELLE, H., HAAS-DUPERTUIS, M.-C., PHILLIPS, J. & MOUTIER, R., 1994: Co-segregation of intermale aggression with the pseudoautosomal region of the Y chromosome in mice. Genetics, 135: SELMANOFF, M. K., JUMONVILLE, J. E., MAXSON, S. C. & GINSBURG, B. E., 1975: Evidence for a Y chromosomal contribution to an aggressive phenotype in inbred mice. Nature, 253: SELMANOFF, M. K., MAXSON, S. C. & GINSBURG, B. E., 1976: Chromosomal determinants of intermale aggressive behavior in inbred mice. Behav. Genet., 6: SLUYTER, F., MEIJERINGH, B. J., VAN OORTMERSSEN, G. A. & KOOLHAAS, J. M., 1995: Studies on wild house mice (VIII): Postnatal maternal influences on intermale aggression in reciprocal F 1 s. Behav. Genet., 25: SLUYTER, F., VAN DER VLUGT, J. J., VAN OORTMERSSEN, G. A., KOOLHAAS, J. M., VAN DER HOEVEN, F. & DE BOER, P., 1996a: Studies on wild house mice. Prenatal maternal environment and aggression. Behav. Genet., 26: SLUYTER, F., VAN OORTMERSSEN, G. A., DE RUITER, A. J. H. & KOOLHAAS, J. M., 1996b: Aggression in wild house mice: Current state of affairs. Behav. Genet., 26: SUCHOMELOVÁ, E. & FRYNTA, D., 2000: Intraspecific behavioural interactions in Apodemus microps: a peaceful mouse? Acta Theriol., 45: THUESEN, P., 1977: A comparison of the agonistic behaviour of Mus musculus musculus L. and Mus musculus domesticus Rutty (Mammalia, Rodentia). Vidensk. Meddr. dansk naturh. Foren., 140: VAN ZEGEREN, K. & VAN OORTMERSSEN, G.A., 1981: Frontier disputes between the West- and East- European house mouse in Schleswig-Holstein, West Germany. Z. Saugetierk., 46: VRBA, I. & DONÁT, P., 1993: Activities version 2.1. Computer programme for behavioural studies. 22

22 Folia Zool. 51(1): (2002) Food intake, feeding behaviour and stock losses of cormorants, Phalacrocorax carbo, and grey herons, Ardea cinerea, at a fish farm in Arcachon Bay (Southwest France) during breeding and non-breeding season Jesús M. LEKUONA c/ Virgen del Puy n 5, 7 D, E Pamplona, Navarra, Spain; jmlekuona@wanadoo.es Received 15 May 2000; Accepted 6 August 2001 Abstract. The feeding ecology of cormorants and grey herons were investigated at a fish farm in Arcachon Bay (southwest France) during both breeding and non-breeding season. Cormorants were mainly recorded during winter and grey herons during both breeding and wintering seasons. Adult cormorants and herons were the most abundant age clas at the fish farm. Adult cormorants and herons were more successful at feeding than first-years and although younger birds spent more time feeding and their biomass intake rates remained lower than those of adults birds. Cormorants and herons took the same biomass intake per feeding session at the fish farm during the non-breeding season, about 200 g. The impact of the two ichthyophagous birds (cormorant and grey heron) was estimated as 53.0% (average predation of cormorant per year) and 10.8% (mean predation of heron per year) of the annual yield of the fish farm. This imposed a significant economic loss due to low productivity of the farm. Key words: aquaculture, diet, foraging parameters, impact, piscivorous birds Introduction In French coastal and inland waters the cormorant (Phalacrocorax carbo) and the grey heron (Ardea cinerea) are commonly perceived as two great predator of fish stocks at fish farms (Hafner & Moser 1980, I m & Hafner 1984, Marion 1983, 1990, Marion & Marion 1983, 1987, Perennou 1987, T rolliet 1993a, 1993b). In other countries of Europe most studies with these two fish-eating birds have focused on extensive aquacultures (D raulans et al. 1985, C arss & Marquiss 1991, Janda 1993, C a r s s 1994, D irksen et al. 1995, K eller 1995, S uter 1995, Van Eerden & Gregersen 1995, Dekker 1997) rather than intensive systems (C a r s s 1990, C a r s s 1993a,1993b, L e k u o n a 1998) where fish stocks area reared at high densities. Their feeding activities had been often considered to result in important economic losses (Meyer 1981, M arion 1983,1990, U tschick 1983, D raulans & Van Vessem 1985, Carss 1991, Osieck 1991, Dirksen et al. 1995, Keller 1995, Warke & Day 1995, Pérez-Hurtado et al. 1997, Lekuona 1998). Recently, C a r s s (1993a) found that these losses are small compared with other forms of fish mortality. A wide variety of deterrent techniques (acoustic, visual and biological) has been tried in attempts to reduce cormorant and heron impact (Barlow & Bock 1984, Draulans 23

23 1987, M oerbeek et al. 1987, F eare 1988, O sieck 1991, T rolliet 1993a, 1993b, Kirby et al. 1996, V a n D a m & Asbirk 1997, Lekuona 1998). The present study gives some data on heron and cormorant predation at a fish farm in terms of piscivorous bird numbers during breeding and non-breeding seasons, food, feeding parameters and stock losses. The study had several aims: 1) to know heron and cormorant abundance and age structure of their population in Arcachon bay and at a fish farm, 2) to determine the food and some feeding parameters of two species of piscivorous birds and 3) to quantify their impact on fish stocks. Study Area, Material and Methods The study was carried out in Arcachon Bay (Fig. 1), southwest France (44 40 N, 1 10 W), a coastal lagoon with variations in water level due to tides, triangular in shape, approximately 20 km wide and km 2 in area (106.6 km 2 of mud and 54.7 km 2 of deep channels, Lekuona 1997). A colony of Ardeids breed there (Lekuona 1993, Lekuona & Campos 1995, Lekuona 1997). Fig. 1. Arcachon bay study area showing location of grey heron colony ( ), winter roosts of cormorants ( ) and the study fish farm (Certes). The farm was located in the oriental extreme of Arcachon bay. The fisf farm of Certes had an area of 142 ha, with two kind of fish ponds: deep ponds (1.5-2 m of depth and 3-4 m of width) and shallow ponds ( m of depth and m 2 of area, Lekuona 1993). Fish stocks (eel Anguilla anguilla, mullets Quelon labrosus and Liza ramada, gilthead Sparus aurata and bass Dicentrarcus labrax) were mainly reared at low densities in shallow ponds. The total yield per year of the fish farm was 9.1 Tm (35 kg.ha -1 for eel, for mullets 100 kg.ha -1, for gilthead 5 kg.ha -1 and for bass 30 kg.ha -1 ). 24

24 The observations on the both cormorant and grey heron s activities were carried out between February 1992 and December 1993, using a x telescope. One day a week, between h and h the bay and the fish farm were visited to obtain an index of abundance of two piscivorous birds by transect counts on a standard route of land. The cormorant s and heron s indeces of abundance were each the maximum of four counts made each month throughout the study period. During non-breeding season cormorant and heron roosts were censused fortnightly at dusk. Two age-classes were taken into account both for cormorants and herons (adult and first-year birds). Birds were categorized on their plumage characteristics (C ramp & Simmons 1977). Most of birds could not be recognized individually (only few cormorants had coloured rings) and some of observations on different study days could have been of the same individuals. However, the data were recorded over several months (February 1992 to December 1993) and the foraging sessions were assumed to be independent. Cormorants and herons were always observed at a distance <100 m. Each day the following data were noted for each bird in both natural feeding areas (Arcachon bay) and at fish farm (Certes): 1) Total time spent (Tt) in a feeding site, as the time elapsed from a bird s arrival to its departure (defined as a foraging session). 2) Foraging time (Ff), as the time used only for foraging within Tt by the two piscivorous bird species. 3) The number of feeding attempts during Ft and their results (successful and unsuccessful). 4) Dive time, only for cormorants, as the average of all dives made during each foraging bout in seconds. 5) Number, type and size of the captured prey. For herons four broad taxonomic groups of prey were used: amphibians, fish, insects, and crustaceans. The size of the prey was calculated only for fish (which were the most abundant prey, see below) in relation to the size of the both cormorant and grey heron s beak (8 cm and 12 cm, respectively, Cramp & Simmons 1977, Voisin 1991). In this way, for grey heron five size classes of fish were established: very small (<12 cm), small (12-18 cm), medium (19-25 cm), large (26-32 cm) and very large (>32 cm), and for cormorant five size classes were also established: very small (<8 cm), small (8-14 cm), medium (15-21 cm), large (22-28 cm) and very large (>28 cm). The biomass (g wet weight) of fish was calculated according to the formulae of R i c h n e r (1986) for eel and flounder Pleuronectes flesus. For mullets, Thick-lipped grey mullet and thin-lipped grey mullet the equation W = L (W = wet weight in g, L = total length in mm) was calculated according to the data of A rne (1938). The biomasses (g wet weight) of gilthead and bass taken were calculated according to the average weight of similar-sized specimens captured in the study area. Cormorant and heron predation (the estimated biomass of fish removed, in kg) was calculated on monthly and per species bases as: Σ Y ij = N j x C x P ij (Suter 1995) where N j is the number of cormorant days during the month j, C is the daily food consumption (g/day) and P ij is the proportion of the species i in the diet (% in biomass) recorded in the month j. Two averages were compared using the Student s t-test (parametric data) and Mann- Whitney Z-test (non-parametric data) and frequencies were compared using the χ 2 test, with Yates correction as necessary (S okal & Rohlf 1979, Fowler & Cohen 1992). The Bonferroni sequential correction test was made in multiple comparisons to avoid 25

25 making the Type I error (reject the null hypothesis when, in fact, it is true) and for keeping the overall error of the test set below 5%. Chandler (1995) showed that sometimes this kind of correction was extremely conservative. Thus, our statistical results have been explained basically without correction, when their biological meaning was very evident. Results Heron and cormorant counts and age ratio Grey heron numbers increased in spring (Fig. 2) due to start of the breeding season in the near ardeid colony located in a riparian wood near the fish farm (<10 km) and rose during the summer (post-fledgling period of juvenile grey herons). During the breeding season adult herons were the most important age class in the study area. During summer juvenile numbers were more abundant in Arcachon bay, falling during autumn when adults showed a high level. Cormorant roost numbers increased in autumn, remained relatively high during the winter, before falling during the spring to a low level in summer (Fig. 3). Adult birds were more abundant during autumn and winter than during spring and summer, when first-year birds had a high level. Seasonal changes were found in the abundance index of two species of fish-eating birds at the fish farm during the present study (Fig. 4): for herons due to management desiccation of extensive ponds (second week of March 1922 and first week of may 1993) and for cormorants due to a change in the feeding behaviour at the farm: very often cormorants foraged solitarily but in November 1993 social fishing (>300 birds) was observed in several shallow ponds, increasing the predation on fish stock (see below). Diet, feeding behaviour and fish stock losses The eel was the main fish species captured by both herons and cormorants in Arcachon bay and at the fish farm (Table 1). Flounders were also taken by two species of piscivorous birds but only in natural habitat. Gilthead and bass were taken by cormorants at low numbers at the fish farm. Other prey species taken by herons were not important (crustaceans, amphibians and insects). The main amphibian and invertebrate taxa identified as a heron food were: frogs Rana perezi, insects (mainly beetles, Coleoptera) and shrimp, Crangon spp. and shore crab, Carcinus maenas. During the breeding season, grey heron spent less total time at the fish farm (t=13.6, p<0.001), more foraging time (t=18.4, p<0.001) and had a higher biomass intake (t=3.2, 0<0.01) than in Arcachon bay (Table 2). During winter some feeding parameters showed the same pattern as the breeding period: grey heron spent less total time at fish farm (t=4.44, p<0.001) and took more biomass (t=1.94, p<0.05) with less feeding time (t=2.79, p<0.01). The Bonferroni sequential test gave the same statistical results. The agression rate observed at the fish farm was higher than in the bay during the breeding season (Z=9.2, p<0,01). In the area under study agressions were not found during winter season. During winter cormorant (Table 3) spent more time fishing in the bay than at the fish farm (t=7.2, p<0.001) and the dive time was also greater in bay than at fish farm (t=16.2, p<0.001). No differences were found between feeding habitats in the biomass intake of 26

26 F-92 A J A O D F A J A O D Fig. 2. Number of grey herons (upper) and age-structure population (lower) during the study period in Arcachon Bay (southwest France). Data are percentage numbers, black: adult birds, white: of first-year birds. Table 1. Food of cormorants and grey herons in natural feeding habitats and at the Certes fish farm in Arcachon Bay (southwest France) during the breeding (B) and non-breeding season (NB). Piscivorous bird Grey heron Cormorant Feeding habitat Bay Fish farm Bay Fish farm Prey B NB B NB NB NB Eel Mullet Flounder Gilthead Bass Unidentified fish Other prey species Total

27 F-92 A J A O D F A J A O D F-92 A J A O D F A J A O D Fig. 3. Number of cormorants (upper) and age-structure population (lower) during the study period in Arcachon Bay (southwest France). Data are percentage numbers, black: adult birds, white: first-year birds. wintering cormorants (t=0.35, NS). Cormorants and herons had the same biomass intake per feeding session at the fish farm during non-breeding period (t=0.47, NS). Both adult grey heron and cormorants (Table 4) were more successful at feeding than first-year birds and younger birds (χ 2 =41.7, 1df, p<0.001 and χ 2 =26.4, 1df, p<0.001, respectively), spent less time feeding (t=16.3, p<0.001 and t=11.7, p<0.001, respectively) and their biomass intake remained higher (t=9.6, p<0.001 and t=10.8, p<0.001, respectively) than those of juveniles at fish farm. Great differences were found in the length of fishes captured by both cormorants and herons in the bay and at the fish farm (χ 2 =32.1, 4df, p<0.001 and χ 2 =27.9, 4df, p<0.001, respectively, Fig. 5). At the fish farm cormorants caught very large fishes whereas grey heron captured large and very large fishes. In Arcachon bay grey heron took significantly more small and medium size fishes than at the fish farm. During breeding season fish losses by grey herons were 12.7% in 1992 and 8.9% in 1993 of the total fish farm yield (the average impact per year was 10.8%). During the study 28

28 F M A My J F-92 A J A O D F A J A O D Fig. 4. Census of grey herons at the fish farm during the breeding season in 1992 and 1993 (upper) and number of cormorants during the study period (lower). Table 2. Foraging studied parameters in feeding bouts made by grey herons in Arcachon bay and at Certes fish farm during breeding and non-breeding season. N: number of foraging bouts, Tt: total time spent in the habitat (min), %: percentage of success, Ft: foraging time (min), B: biomass intake (g wet weight) and A (aggression rate, Ag min -1 ). Tt, Ft, B and A as average±standard deviation. Habitat Bay Fish farm Bay Fish farm Season Breeding Non-breeding N Tt 136.4± ± ± ±23.0 % Ft 27.7± ± ± ±10.3 B 197.7± ± ± ±61.7 A 0.086± ± ± ±0.00 period cormorant losses were 38.4% in 1992 and 67.6% in 1993 (the average predation per year was 53.0%). The habit of mass fishing by cormorants observed in winter 1993 significantly increased the annual impact. 29

29 Table 3. Foraging studied parameters in feeding bouts made by cormorants in Arcachon bay and at fish farm during non-breeding season. N: number of foraging bouts, %: percentage of success, Ft: foraging time (min), D: dive time (s) and B: biomass intake (g wet weight). Ft, D and B as average±standard deviation. Feeding habitat Bay Fish farm N % Ft 27.9± ±2.6 D 24.3± ±3.3 B 195.8± ±58.9 Table 4. Foraging parameters of cormorants and grey herons at a fish farm in southwest France taking into account the age classes. N: number of foraging bouts, %: percentage of success, Tt: total time spent in the feeding habitat, Ft: feeding time (min), D: dive time (s), B: biomass taken (g wet weight). Tt, Ft, D and B are given as average±standard deviation. Species N % Tt Ft D B Grey Heron Adult ± ± ±46.2 First-year ± ± ±34.7 Cormorant Adult ± ± ± ±41.4 First-year ± ± ± ±19.9 Fig. 5. Length of prey (cm) captured by cormorants (upper) and grey herons (lower) at Certes fish farm (black) and in Arcachon bay (white). Data are percentages. 30

30 Discussion As a consequence of the rapid increase of both cormorant (V a n Eerden &Gregersen 1995) and heron populations (V o i s i n 1991) in Europe, interference with fisheries has risen (Marion & Marion 1983, 1987, Moerbeek et al. 1987, Marion 1990, Carss 1994, V a n D a m & Asbirk 1997, Callaghan et al. 1998, Leopold et al. 1998). Arcachon bay area was located within a region with high heron and cormorant density (Campos & Lekuona 1994, L e kuona 1999), where adult birds were the most abundant age class in coastal and inland waters (L e k u o n a 1997). In the present study cormorant numbers increased at the fish farm during winter and the heron number increased during breeding season due to some management techniques. Both adults and juveniles birds visited the fish farm during all the study period, but they showed significant agerelated differences in feeding performance (C arss 1993b). Young birds were able to compensate their lower success by attempting more feeding strikes or dives (Desgranges 1981, L ekuona 1997). In this study both age classes took fish of a similar size as did those studied by D raulans (1987) and C arss (1993b), but juvenile birds fed for more time at the fish farm than did adults. Draulans (1987) also found that first-year herons were less successful at foraging thatn adults at low prey densities; similar data were found in the present study. Cormorants and herons could not be recognized individually and some observations could have been made on the same bird, and this fact may affect the comparisons between two age classes. However, the data were recorded over 24 months with great differences always between age classes so that the foraging sessions were assumed to be independent and representative of the impact made by cormorants and herons at the fish farm. An adult (heron and cormorant) needs a biomass intake per day of g of fish (Cramp & Simmons 1977, M arion 1990, C arss 1993b, S uter 1995, Lekuona & Campos 1997, V a n D a m & A sbirk 1997). The biomass intake rates at the farm in relation to the feeding time of the two species were considerably higher than other published rates (C o o k 1978, R i c h n e r 1986), but similar data were found for grey heron at cage fish farms in Argyll (Carss 1993b). Thus, the daily requirements of food could be met at the fish farm making a second foraging session and many of the cormorants and herons probably caught all their prey there. At the fish farm, food requirements of the two ichthyophagous species were satisfied within a shorter foraging time than in Arcachon bay. Foraging activity of breeding grey herons and wintering cormorants in Arcachon bay was strongly influenced by variations in sea water level due to tides (L ekuona 1997, 1999), but no significant differences were found in their feeding activity at the fish farm during tides (L e k u o n a 1997) because the water level was constant during most of the study period. At the study fish farm the cormorant predation on fish stocks was higher than heron impact, but these rates represented a considerable economic loss. Similar date were found in other regions of Europe (I m & H afner 1984, P erennou 1987, M arion 1990, Osieck 1991, Lekuona 1998). Most studies on fish-eating birds have been focused in estimate direct loss of fish stocks (Marion 1990, O sieck 1991), but some other factors must be taken into account (parasites, physiological poor condition, blindness, wounded fish...) (C a r s s 1990, 1993b). Also in some regions of Europe available data at fish farms are anecdotal or based on 31

31 second-hand information (C a r s s 1993b), also this author found that not all of the fishes taken by grey heron were healthy. Parasited trouts moved slowly just below the surface and they formed a high proportion of the prey caught by herons. Thus, such predation did not represent a direct loss. It is obvious that very often it is extremely complicated to quantify the interaction between piscivorous birds and fisheries. Acknowledgements I would like to thank to the Asociación de Amigos de la Universidad de Navarra (1992) and to the Caja de Ahorros Municipal de Pamplona (1993) for their grants awarded to me for the completion of this study. My thanks go to all those who helped with the fieldwork, especially Alain Fleury, Claude Feigné and Veronique Hidalgo (Parc Ornithologique du Teich), Alberto Artazcoz and Itziar Gastón. Iam very grateful to the Certes s fish managers and their staff who kindly allowed me free access to the fish farm during the study period. Dr. P. Blahák, Professor R.H.K M a n n and an anonymous reviewer gave useful comments on a earlier draft of this paper. LITERATURE ARNE, P., 1938: Contribution à l étude de la biologie des muges du Golfe de Gascogne. P.V. Reun. Com. Explor. Scient. Mer Mediterranée, 11: BARLOW, C. G. & BOCK, K., 1994: Predatioin of fish in farm dams by cormorants, Phalacrocorax spp. Australian Wildl. Res., 11: CAMPOS, F. & LEKUONA, J. M., 1994: La población invernante de Cormorán Grande (Phalacrocorax carbo) en el Norte de España y Suroeste de Francia. Ardeola, 41: CALLAGHAN, D. A., KIRBY, J. S., BELL, M. C. & SPRAY, C. J., Cormorant Phalacrocorax carbo occupancy and impact at stillwater game fisheries in England and Wales. Bird Study, 45: CARSS, D. N., 1990: Beak-prints help in war againts aerial invaders. Fish Farmer, 13: CARSS, D. N., 1991: Avian predation at farmed and natural fisheries. Proceedings of the Institute of Fisheries Management. September 1991, University of Aberdeen. CARSS, D. N., 1993a: Cormorants Phalacrocorax carbo at cage fish farms in Argyll, western Scotland. Seabirds, 15: CARSS, D. N., 1993b: Grey Heron, Ardea cinerea L., predation at cage fish farms in Argyll, western Scotland. Aquaculture and Fisheries Management, 24: CARSS, D. N., 1994: Killing of piscivorous birds at Scottish fin fish farms, Biol. Conserv., 68: CARSS, D. N. & MARQUISS, M., 1991: Avian predation at farmed and natural fisheries. In: Lucas, M. C., Diak, I. & Laird, L. (eds.), Interaction between fisheries and environment. Proc. Institute of Fisheries Management, 22nd Annual Study Course, University of Aberdeen. CHANDLER, C. R., 1995: Practical considerations in the use of simultaneous inference for multiple test. Anim. Behav., 49: COOK, D. C., 1978: Foraging behaviour and food of grey herons on the Ythan Estuary. Bird Study, 54: CRAMP, S. & SIMMONS, K. E. L., 1977: The birds of the Western Palearctic. Vol. I. Oxford University Press, Oxford, U.K. DEKKER, W., 1997: The impact of cormorants and fykenet discards on the fish yield from Lake Ijsselmeer, The Netherlands. In: Van Dam, Ch. & Asbirk, S. (eds.), Proceedings of the Workshop an International Conservation and Management Plan for the Great Cormorant (Phalacarcrocorax carbo). October 1996, Lelystad, The Netherlands: DESGRANGES, J. L., 1981: Observations sur l alimentation du grand heron (Ardea herodias) en Quebec (Canada). Alauda, 49: DIRKSEN, S., BOUDEWIJN, T.J., NOORDHUIS, R. & MAERTEIJN, E. C. L., 1995: Cormorants Phalacrocorax carbo sinensis in shallow eutrophic freshwater lakes: prey choice and fish consumption in the non-breeding period and effects of large-scale fish removal. Ardea, 83: DRAULANS, D., 1987: The effectiveness of attempts to reduce predation by fish-eating birds: a review. Biological Conservation, 41:

32 DRAULANS, D. & VAN VESSEM, J., 1985: The effect of disturbance on nocturnal abundance and behaviour of Grey Herons (Ardea cinerea) at a fish farm in winter. Journal of Applied Ecology, 22: DRAULANS, D. & VAN VESSEM, J. & DE BONT, A. F., 1985: The effects of killing and removal on the abundance of Grey Heron at fish farms. XVIIth Congress of the International Union of Game Biologists, pp. FEARE, C. J., 1988: Cormorants as predators at freshwater fisheries. Inst. Fisheries Management Ann. Stud. Course, 18: FOWLER, J. & COHEN, L., 1992: Statistics for Ornithologists. BTO Guide 22. HAFNER, H. & MOSER, M., 1980: Les hérons et la pisciculture en Camargue. Colloque sur les zones humides et littorales, l aquaculture et la faune sauvage. France. Bull. de l office National de la Chasse, p IM, B. H. & HAFNER, H., 1984: Impact des oisesaux piscivores et plus particulièrement du grand cormoran Phalacrocorax carbo sinensis sur les exploitations piscicoles en Camargue, France. Le Sambuc, Station Biologique de la Tour du Valat. JANDA, J., 1993: La situation en Tchécoslovaquie. Bulletin Mensuel Office National de la Chasse, 178: KELLER, T., 1995: Food of Cormorants Phalacrocorax carbo sinensis wintering in Bavaria, southern Germany. Ardea, 83: KIRBY, J. S., HOLMES, J. S. & SELLERS, R. M., 1996: Cormorants Phalacrocorax carbo as fish predators: a appraisal of their conservation and management in Great Britain. Biol. Conserv., 75: LEKUONA, J. M., 1993: Uso de hábitat, alimentación y reproducción de la Garza Real (Ardea cinerea) en la Bahía de Arcachon (Francia). Tesis de Licenciatura, Universidad de Navarra. LEKUONA, J. M., 1997: Importancia de las aves ictiófagas: Cormorán Grande (Phalacrocorax carbo) y Garza Real (Ardea cinerea) en el norte de España y suroeste de Francia. Tesis Doctoral, Universidad de Navarra. LEKUONA, J. M., 1998: Impacto del Cormorán Grande Phalacrocorax carbo sinensis, la Gaviota Reidora Larus ridibundus y la Garza Real Ardea cinerea en un piscifactoría de Navarra durante la época invernal. Ardeola, 45: LEKUONA, J. M., 1999: Food and foraging activity of grey herons, Ardea cinerea, in a coastal area during the breeding season. Folia Zool., 48: LEKUONA, J. M. & CAMPOS, F., 1995: Le succès de reproduction du Héron cendré (Ardea cinerea) dans le Bassin d Arcachon. Alauda, 63: LEKUONA, J. M. & CAMPOS, F., 1997: Foraging ecology of cormorants (Phalacrocorax carbo) wintering in northern Spain. Folia Zool., 46: LEOPOLD, M. F., VAN DAMME, C. J. G. & VAN DER VEER, H. W., 1998: Diet of cormorants and the impact of cormorant predation on juvenile flatfish in the Ducht Wadden Sea. Journal of Sea Research, 40: MARION, L., 1983: Etat de nos connaissances sur les relations oiseaux piscivores et populations piscicole naturelles et artificielles et propositions de recherches prioritaires pour la période Rapport du Groupe de Travail sur les oiseaux piscivores, Ministère de l Environnement, Paris. MARION, L., 1990: Les oiseaux piscivores et les activités piscicoles: impact et protection. Secrétariat d Etat chargé de l environnement & Ministère de l agriculture et de la forêt, Paris. MARION, P. & MARION, L., 1983: Etude de la predation par le Héron cendré sur les piscicultures intensives et sur le milieu naturel. Rapport Laboratoire d Evolution des Systèmes Naturels et Modifiés. Ministère de l Environnement, Paris. MARION, P. & MARION, L., 1987: Impact de la predation exercée par le Héron cendré sur une pisciculture. Revue d Ecologie (Terre et Vie), 4: MEYER, J., 1981: Room for birds and fish. Fish Farmer, 4: MOERBEEK, D. J., VAN DOBBEN, W. H., OSIECK, E. R., BOERE, G. C. & BUNGENBERG DE JONG, C. M., 1987: Cormorant damage prevention at a fish farm in the Netherlands. Biol. Conserv., 39: OSIECK, E. R., 1991: Prevention of Cormorant damage at the Lelystad fish farm. In: Van Eerden, M. R. & Ziljstra, M. (eds.), Proceedings Workshop 1989 on Cormorants (Phalacrocorax carbo). Rijkswaterstaat Directorate Flevoland, Lelystad: PERENNOU, C., 1987: L impact de Héron cendré Ardea cinerea en bassin d alevinage. L Oiseau et R. F. O., 57: PÉREZ-HURTADO, A., MASERO, J. A., ARROYO, G., PIZARRRO, S R., HORTAS, F. & CASTRO, M., 1997: Estudio de la incidencia de la Avifauna Ictiófaga en los cultivos marinos y otras especies orníticas en el Parque Natural de la Bahía de Cádiz. Consejería de Medio Ambiente de la Junta de Andalucía, Cádiz. RICHNER, H., 1986: Winter feeding strategies of individually marked herons. Anim. Behav., 34: SOKAL, R. R. & ROHLF, F. J., 1979: Biometría. Blume, Madrid, Spain. 33

33 SUTER, W., 1995: The effect of predation by wintering cormorants Phalacrocorax carbo on grayling Thymallus thymallus and trout (Salmonidae) populations: two case studies from Swiss rivers. J. of Applied Ecology, 32: TROLLIET, B., 1993a: Moyens préventifs de limitation dl impact du grand cormoran sur la pisciculture extensive. Bulletin mensuel Office National de la Chasse, 178: TROLLIET, B., 1993b: Un nouveau moyen d effarouchement: le fusil laser. Bulletin mensuel Office National de la Chasse, 178: UTSCHICK, H., 1983: Abwehrstrategie und Abwehrmassnahmen gegen den Graureiher (Ardea cinerea) und Fischgewässern. Garm. Vogelk. Ber., 12: VAN EERDEN, M. R. & GREGERSEN, J., 1995: Long-term changes in the northwest European population of Cormorants Phalacrocorax carbo sinensis. Ardea, 83: VAN DAM, C. & ASBIRK, S., 1997: Cormorants and human interests. Proceedings of the Workshop towards an International Conservation and Management Plan for the Great Cormorant (Phalacrocorax carbo), 3 4 October 1996, Lelystad, The Netherlands. VOISIN, C., 1991: The Herons of Europe. T & DA Poyser, London. WARKE, G.M.A. & DAY, K. R., 1995: Changes in abundance of cyprinid and percid prey affect rate of predation by Cormorants Phalacrocorax carbo carbo on Salmon Salmo salar smolt in Nothern Ireland. Ardea, 83:

34 Folia Zool. 51(1): (2002) Importance of colony size and breeding synchrony on behaviour, reproductive success and paternity in house sparrows Passer domesticus Radovan VÁCLAV 1* and Herbert HOI 2 1 Institute of Zoology, Slovak Academy of Sciences, Dúbravská cesta 9, SK Bratislava, Slovakia; uzaevacl@savba.sk 2 Konrad Lorenz Institut für Vergleichende Verhaltensforschung, Österreichische Akademie der Wissenschaften, Savoyenstraße 1a, A-1160 Vienna, Austria; h.hoi@klivv.oeaw.ac.at Received 20 March 2001; Accepted 12 November 2001 A b s t r a c t. Several comparative studies have previously identified breeding density and synchrony as potential determinants of reproductive success and extra-pair mating. However, the mechanisms and interaction of these two factors are poorly known. Here, we examined the effects of breeding density and synchrony on the behaviour, reproductive success and paternity losses in house sparrows. In order to test the effects of colony size, we created nest sites with varying numbers of nest-boxes. Our results show that there is an interaction between breeding synchrony and density, namely that breeding synchrony decreased with colony size. Neither colony size nor breeding synchrony seemed to influence brood size at fledging, although birds in larger colonies laid larger clutches. Moreover mate guarding behaviour was not influenced substantially by these two factors. Only nest guarding was significantly related to colony size and breeding synchrony. Paternity losses were not significantly related to colony size but they appeared to decrease with increasing synchrony. This finding supports the idea that extra-pair fertilisations are under male rather than female control. Key words: breeding density, synchrony, house sparrows, reproductive success, paternity Introduction Ecological factors like food availability, weather or predator pressure are well known to influence life history and reproductive success of a species (Lack 1968). In addition, they were recognised to effect parental effort (H o i et al. 1995), mating system (Emlen & Oring 1977, Hoi et al. 1995), sexual strategies (see Birkhead & M ø ller 1992) and paternity (Hoi & Hoi-Leitner 1997). In respect of the latter two, colony size and breeding synchrony have been recently suggested as important socio-ecological determinants (Birkhead & Biggins 1987, Birkhead & Mø ller 1992). Both factors may determine intraspecific competition for resources such as food, nest sites or mates (W ittenberger & Hunt 1985, Brown 1987). In this way, breeding density and synchrony may directly influence breeding success, but may also effect sexual strategies like extra-pair behaviour or paternity guards and consequently the realised reproductive success. M ø ller & Birkhead (1993) reported that cuckoldry is higher in colonial than solitary breeding bird species. They suggested that this might be due to insufficient mate guarding or increased possibility to encounter sexual partners. W agner (1993) hypothesised that females may even promote breeding in higher densities in order to facilitate extra-pair matings. Nevertheless, only some studies support this prediction (e.g. M ø l l e r * Corresponding author 35

35 1985, Birkhead et al. 1987, H ill et al. 1994, M ø ller & Ninni 1998). A comparative study by W estneat & Sherman (1997) did not provide general support for this idea between species (but see M ø ller & Ninni 1998) although the authors showed that cuckoldry increases with density within species. W estneat & Sherman (1997) further suggested that density might perhaps be linked with additional factors such as habitat structure, female behaviour affecting encounter rate with extra-pair males or breeding synchrony. Indeed, breeding synchrony was found to favour extra-pair mating in some species (Stutchbury et al. 1994). However, the role of synchrony for extra-pair behaviour is still controversial (see W eatherhead 1997). Namely, it is not clear whether extra-pair paternity (EPP) should increase (Stutchbury & Morton 1995) or decrease with synchrony of breeding (Birkhead& Mø ller 1992). A recent comparative study by M ø ller & Ninni (1998), however, suggests that among species paternity decreases with breeding asynchrony. Most of the non-comparative studies on extra-pair paternity tend to examine only one factor at the time, whether it is breeding density or synchrony (T husius et al. 2001). In this study, therefore, we try to investigate simultaneously the importance of both colony size and breeding synchrony on reproductive success, paternity assurance, nest guarding behaviours and paternity losses in house sparrows Passer domesticus. The house sparrow is an ideal species for such a study because it nests solitarily as well as in colonies of variable sizes (S ummer-smith 1954, Cramp 1994, T ost 1994). Moreover, sperm competition is intense in this species as indicated by high within-pair copulation frequency (M ø ller 1987) and occurrence of EPP (e.g. W etton & Parkin 1991, Cordero et al. 1999, V áclav et al., in press). In order to study the effect of colony size and breeding synchrony, we offered house sparrows breeding situations in the plots with varying number of nest-boxes and experimentally altered start of egg-laying. Material and Methods Our study was conducted in the Schönbrunn Zoo in Vienna, Austria during the breeding seasons 1999 and In these two years, house sparrows nested in 80 and 69 nest-boxes, respectively. Five to ten nest-boxes were hung on buildings in 1999 and three nest-boxes in In this way, we created four plots with five nest-boxes and ten plots with ten nest-boxes in 1999 and 15 plots with three nest-boxes in All plots were occupied in 1999 but eight plots were unsettled or settled only during one breeding attempt in The nest-boxes were placed m apart in all plots. Birds were usually trapped after completing their clutches. We weighed all birds and measured the length (mm) of tarsus, wing and tail. We found no difference in the size of the male ornament (mean badge size: mm 2 in 1999 and mm 2 in 2000; t-test: t = 0.39, p = 0.70, n = 16, 14). Since badge size is considered to be a condition-dependent trait (Griffith 2000), we think that the phenotype of birds did not markedly influence breeding patterns between years. In both breeding seasons, we started to observe activity around nest-boxes several months before the first eggs were laid (25 March in both years). Clutch and brood size were regularly monitored in all nest-boxes at least every second day during the whole breeding season. Brood size at fledging refers to an estimate of fledging numbers when chicks were days old. Accordingly, fledging success was calculated as the number of chicks in the 36

36 nest reaching the age of days divided by the number of laid eggs. We analysed the behaviour of 28, 29 and 25 pairs during three consecutive breeding attempts during the egg laying in Every nest site was observed daily during 15 min when we recorded the presence of birds inside and around nest-boxes, the time mates stayed together and withinpair copulation frequency. In 2000, we manipulated breeding synchrony experimentally. To achieve asynchrony, we swapped up to the half of nest contents from the nest of a focus pair to the nest of a neighbouring pair that was at a similar stage of breeding. On the other hand, we attained synchrony by equalising nest contents between pairs at apparently different breeding stages. The focus pair was, in this case, one that nested formerly at a later breeding stage. This was not the most suitable technique in the second and mainly third breeding attempt when the volume of nest material was more or less stable and was hence an unreliable estimate of breeding stage. For this reason, we used behavioural observations to estimate breeding stage and manipulated nest contents in the same logic as described above. Moreover, in the third breeding attempt, we delayed more advanced breeders by reducing their clutch size by taking up to two eggs and hence extending their laying period by laying additional eggs. In this way, we could increase synchrony between pairs with reduced clutches and focus pairs by more than 40% (mean clutch size in our study was 4.6 eggs in 1999 and 4.3 in 2000). Consequently, synchrony refers in this part of the results to the cases when we achieved either full or at least 80% breeding synchrony, whereas asynchrony means that the focus and neighbouring pairs did not overlap their laying periods. Synchrony from the correlational part of results refers to the proportion of the laying period that overlapped between pairs within a nesting site. For analysing paternity losses, we used multi-locus DNA fingerprinting. We examined genetic parentage of 34 families. Laboratory procedures followed those previously described by Krokene et al. (1996). We used band-sharing and number of novel bands (i.e. bands present in chicks but not in the putative parents) when excluding parents (e.g. Krokene et al. 1996). The method and assumptions for paternity exclusion are presented in more detail in W etton et al. (1987) and W estneat (1990). Paternity losses refer for each pair to the proportion of extra-pair young in the brood. Parametric tests were used in our analyses only when the requirements for these tests were met. Normality was tested with the Liliefors and Shapiro-Wilk tests. In order to achieve normality, data on paternity losses were log transformed. In parametric tests, post hoc comparisons were calculated with the Newman-Keuls test. If not stated otherwise, presented values are means and standard errors. The majority of studied pairs (94 % in 1999 and 93 % in 2000, respectively) contributed more than one clutch within breeding season. Hence, in order to avoid pseudo-replication, we examined seasonal and annual variation of reproductive success and behaviour with repeated ANOVAs. In addition, since the number of pairs at nest sites fluctuated between breeding cycles, variation of different behaviours was analysed using mean values for each nest site. Synchrony decreased with increasing size of colony (multiple regression: r 2 = 0.22, F 2,27 = 3.83, p = 0.034, effect of colony size ß = -0.47, p = 0.011, effect of set-up ß = -0.02, p = 0.90). To control for this interrelationship and the effect of nest-box set-up, we included in multiple regression analyses all of the three factors. Hence, colony size, breeding synchrony and the type of setup (5 and 10 nest-box sites) were entered as independent variables and reproductive success, behaviour and paternity losses as dependent variables. 37

37 Results Seasonal and annual variation in reproductive success and behaviour during egg laying Although the first eggs were recorded on the same day, 25th March, pairs nesting in 2000 tended to initiate their clutches later than pairs in 1999 and this difference increased and was significant in the two subsequent breeding attempts (Fig. 1). Fig. 1. Start of egg laying (means ± s.e.) for the three breeding attempts in 1999 and Median-test of the annual difference between first breeding attempts: χ 2 = 3.43, p = 0.068, n = 78; second breeding attempts: χ 2 = 12.36, p < 0.01, n = 81; and third breeding attempts: χ 2 = 20.13, p < 0.01, n = 73. There was seasonal variation in reproductive success (Fig. 2). Clutch size, hatching size and brood size at fledging varied significantly among the three breeding attempts (Fig. 2a, b and c). Reproductive success in terms of brood size at fledging peaked in the second breeding attempt in 1999, whereas in 2000 it appeared to decline with the season. Importantly, when taking clutch size into account, efficiency of fledging chicks did not change significantly between the three breeding cycles (Fig. 2d). Examining annual variation in reproductive success, we found that larger clutches were laid in 1999 but hatching size and brood size at fledging were similar between years (Fig. 2a, b and c). In addition, pairs achieved similar fledging success in both years (Fig. 2d). Interactions between the effect of year and the time of season for clutch and brood size at fledging, and mainly fledging success (Fig. 2a, c and d) suggest that pairs in 1999 allocated most reproductive effort to the second breeding attempt rather than the first as in Seasonal variation of colony size, synchrony and behaviour Depending on the number of available nest-boxes in the plots (plots with 3, 5 and 10 nestboxes), 87%, 77% and 46% of the nest-boxes were occupied (Fig. 3). While the number of breeding pairs did not differ between the plots with 5 and 10 nest-boxes, fewer pairs nested 38

38 Fig. 2. Seasonal and annual variation in four measures (means ± s.e.) of reproductive success. (a) Clutch size (2-way repeated measures ANOVA, seasonal effect: F 2,96 = 13.54, p < 0.001, year effect: F 1,48 = 15.29, p < 0.001, interaction: F 2,96 = 5.90, p < 0.01), (b) Hatching size (seasonal effect: F 2,96 = 5.51, p < 0.01, year effect: F 1,48 = 1.99, p = 0.17, interaction: F 2,96 = 2.67, p = 0.07), (c) Brood size at fledging (seasonal effect: F 2,96 = 5.49, p < 0.01, year effect: F 1,48 = 0.04, p = 0.84, interaction: F 2,96 = 6.24, p < 0.01), (d) Fledging success (seasonal effect: F 2,96 = 0.85, p = 0.43, year effect: F 1,48 = 0.35, p = 0.55, interaction: F 2,96 = 4.50, p = 0.013). in the plots with 3 nest-boxes (the mean number (± se) of pairs in plots: with 10 nest-boxes: 4.56 ± 0.42; 5 nest-boxes: 3.83 ± 0.34; 3 nest-boxes: 2.5 ± 0.1; F 2,69 = 23.25, p < 0.01; post hoc 10-3** and 5-3**; ** p < 0.01). In contrast to the set-ups with 5 and 10 nest-boxes, there was a seasonal decline in the colony occupancy with breeding season for the set-up with 3 nest-boxes (Fig. 3). On average, 25% of the laying period (days) overlapped between pairs within a nesting site. We found no change in breeding synchrony between the three successive breeding attempts (first attempt: 26.45% ± 5.31; second attempt: 21.09% ± 3.27; third attempt: 26.43% ± 7.51, Kruskal-Wallis: H 2,30 = 0.51, p = 0.75). We did not detect significant differences between three breeding attempts in the time that mates stayed at the nest together during the time of egg laying, although mate guarding 39

39 Fig. 3. Seasonal variation of nest site occupancy (means ± s.e.) in relation to the number of installed nest-boxes. (10 nest-boxes: Kruskal-Wallis H 2,18 = 1.83, p = 0.40; 5 nest-boxes: H 2,12 = 1.18, p = 0.55; 3 nest-boxes: H 2,42 = 6.02, p = 0.049). appeared to decline towards the end of season (Fig. 4a). Copulation frequency, however, significantly declined with the breeding season (Fig. 4b). In contrast to males, the time females stayed during egg laying at the nest alone decreased with the time of season (Fig. 4c). Moreover, males and females seemed to reduce the time they stayed inside the nestboxes during egg laying after the first breeding attempt (Fig. 4d). Paternity losses (proportion of extra-pair chicks/brood) did not significantly differ between breeding attempts (first attempt: 6.7%; second attempt: 28.8%; third attempt: 24%; Kruskal-Wallis H 2,25 = 3.78, p = 0.15). The role of colony size for reproductive success, behaviour during egg laying and paternity losses After controlling for the effect of season, multiple regression analyses revealed a relationship only between clutch size and the number of breeding pairs. Specifically, clutch size increased with colony size (Fig. 5) (including the 5 and 10 nest-boxes set-up only). In addition, the partial regression coefficient suggests that clutch size was larger in five than in ten nest-box plots (see Table 1). We found a relationship between nest guarding behaviour and colony size (Table 2). Nevertheless, birds of both sexes breeding in the plots with five nest-boxes stayed in their nests during egg laying significantly longer than birds in the plots with ten nest-boxes (see Fig. 6, Table 2). The time males and females stayed at the nest alone was not related to colony size or the type of set-up (see Table 2). Similarly, paternity assurance behaviours do not seem to be largely related to colony size (Table 2). There is, however, indication of a link between copulation frequency and colony size (see Table 2). Finally, the relationship between colony size and paternity losses seemed to be positive but not significantly (r = 0.18, p > 0.4, n = 24). 40

40 Fig. 4. Seasonal variation of mate guarding behaviour (means ± s.e.) in terms of (a) time mates stayed at nests together (1-way repeated measures ANOVA: F 2,14 = 2.34, p = 0.13) and, (b) copulation frequency (F 2,14 = 5.02, p = 0.02; 1-3* and 2-3*), and nest guarding behaviour (means ± s.e.) in terms of (c) time males stayed at nests alone (F 2,14 = 0.90, p = 0.43; time females alone (F 2,14 = 4.09, p = 0.04; 1-3*), (d) time males inside the nest (F 2,14 = 6.32, p = 0.01; 1-2* and 1-3*) and time females inside the nest (F 2,14 = 2.81, p = 0.09). After tests are showed post hoc comparisons (* p < 0.05). The role of breeding synchrony for reproductive success, behaviour during laying and paternity losses After controlling for the effect of season, a multiple regression analysis revealed a relationship only between hatching success and breeding synchrony (Fig. 6a). Specifically, the number of hatched chicks increased when egg laying was more synchronous at a colony. However, the influence of breeding synchrony on brood size at fledging (Fig. 6b) and fledging success was not significant (see Table 1). Experimental manipulation of the laying synchrony in 2000 supported our result from the natural condition about the existence of a relationship between hatching size and laying synchrony. Although we found no significant difference in clutch size between synchronous 41

41 Fig. 5. The relationship between clutch size and the number of pairs at nest site. Values of clutch size were standardised for the time of season. Lines are linear regressions for each types of nest site. Table 1. Partial regression coefficients (p) from stepwise multiple regression models between socio-ecological factors and parameters of reproductive success. Whole model values are r 2, (n) and corresponding p. Variables of reproductive success were controlled for the time of season (we used residual values). Socio-ecological variables Parameters of reproductive breeding type of set-up whole model success colony size synchrony clutch size 0.50 (0.013) 0.20 (0.28) 0.45 (0.01) 0.32 (30)* hatching size 0.38 (0.07) 0.44 (0.03) 0.27 (0.13) 0.24 (30) (0.07) hatching failure 0.36 (0.05) (30)* brood size at fledging 0.22 (0.25) (30) (n.s.) fledging success 0.21 (0.27) (30) (n.s.) * p < 0.05 Table 2. Partial regression coefficients (p) from stepwise multiple regression models between socio-ecological factors and behaviours related to mate and nest guarding. Whole model values are r 2, (n) and corresponding p. All variables referring to behaviour were controlled for the time of season (we used residual values). Socio-ecological variables Mate and nest guarding breeding type of set-up whole model behaviours colony size synchrony copulation frequency 0.21 (0.32) (26) (n.s.) time mates together a 0.11 (0.63) 0.07 (0.77) 0.04 (0.87) 0.01 (26) (n.s.) time males in the nest 0.28 (0.12) 0.45 (0.02) 0.46 (0.01) 0.41 (26)** time females in the nest 0.33 (0.08) 0.38 (0.04) 0.30 (26)* time males at the nest alone 0.35 (0.10) 0.27 (0.20) 0.30 (0.14) 0.20 (26) (n.s.) time females at the nest alone 0.29 (0.13) 0.33 (0.09) 0.23 (26)* a none of the factors was significant and therefore we performed standard multiple regression p < 0.05, ** p <

42 Fig. 6. Hatching size (a) and brood size at fledging (b) in relation to breeding synchrony. Values of hatching size and brood size at fledging were standardised for the time of season. Lines are linear regressions for each types of nest site. and asynchronous pairs (t = 1.09, p = 0.28, n = 28, 20), pairs nesting synchronously hatched significantly more chicks (t = 2.37, p = 0.022, n = 28, 20). Moreover, differences in brood size at fledging and fledging success between synchronously and asynchronously laying pairs were marginally significant (brood size at fledging: t = 1.97, p = 0.054, n = 28, 20; fledging success: t = 1.86, p = 0.069, n = 28, 20). With respect to behaviour, synchrony seems to have a large influence only on nest guarding behaviour. Both males and females spent a longer time in the nest with decreasing synchrony (Fig. 7a and b, Table 2). We did not find any stronger relationship between synchrony and paternity assurance behaviours (see Table 2). Similarly, there was no significant relationship between breeding synchrony and paternity losses (Fig. 8). However, 43

43 Fig. 7. Nest guarding behaviours in relation to breeding synchrony. (a) Time males stayed in the nest. (b) Time females stayed in the nest. Duration of nest guarding was standardised for the time of season. the shape of the relationship between synchrony and paternity losses indicates that males suffered a loss of their paternity less often with increasing synchrony. Discussion Our results suggest that colony size does not have a strong effect on reproductive success, behaviour and paternity losses. Clutch size seemed to increase with colony size but finally there was no difference in the reproductive success in relation to colony size. Surprisingly, we did not find a substantial effect of colony size on investment in paternity guards or nest guarding behaviours and consequently only a weak effect on paternity losses. This is in contrast to some other studies, which found a density dependent effect on paternity losses (Gowaty & Bridges 1991, Hill et al. 1994, H o i & H o i - L e i t n e r 1997) 44

44 Fig. 8. The relationship between paternity losses and breeding synchrony (r = -0.21, p = 0.31, n = 24). Paternity losses were log transformed to achieve normality. as well as on investment in paternity assurance tactics (B irkhead et al. 1987, Hatchwell 1988, for house sparrows see T o s t 1994). Hence, although sperm competition occurred in our study population, it did not seem to be directly linked to colony size. An explanation could be that extra-pair copulations occurred at the foraging sites and colony size would hence not reflect the level of sperm competition. This is, however, unlikely since W etton et al. (1995) showed specifically for house sparrows and Birkhead & M ø ller (1992) for other colonially breeding species that extra-pair males are often neighbours. There is still a possibility that density may be important for extra-pair fertilisations since we manipulated only colony size but not density and the distance between nest-boxes remained similar in all colonies. In contrast to T o s t (1994), who found that house sparrows adjust their paternity guards in relation to colony size, we may not have found such a relationship for another reason. T ost s (1994) population nested under natural conditions, whereas our study provides an example of an unnatural nest-box population. There are several important differences between natural and nest-box populations that need to be stressed. Firstly, in nest-box populations, quality of nesting sites does not necessarily reflect territory quality, such as food abundance. Secondly, if females seek extra-pair fertilisations (EPF) according to the male territorial quality, equal quality of nest-boxes disables females to judge the quality of males because male dominance is no longer reflected in nest site quality. Thirdly, the cost of searching extra-pair males may be too high in uniform and high quality nesting habitats (Slagsvold & Dale 1991, W estneat 1992). Our results are, in respect of the latter two points, similar to the study by V eiga & Boto (2000), who did not find in their nest-box colony higher levels of EPP than those reported from lower density nest-box colonies of house sparrows. Nonetheless, other factors like breeding synchrony (see later, Stutchbury & Morton 1995), territory quality (H o i et al. 1995) or intense intraspecific competition (V eiga 1992) might have outweighed the role of breeding density in our study. Also, it is important to mention that the phenotype of birds was not considered in our study due to a small sample size and hence we might lose an important component of variation in our analyses. 45

45 The missing relationship between colony size, behaviour, reproductive success and achieved paternity may also be due to a nonlinear relationship between them. T o s t (1994) found a considerable variation of breeding sociality, ranging from rare cases of solitarily breeding pairs up to colonies of 200 breeding pairs. Yet, even within these huge colonies, he could usually distinguish clusters with two to seven pairs. This suggests that birds avoided to nest solitarily as well as in large groups and the observed aggregations reflected the preferred colony size. In line with this, we found that, regardless of the nest-box availability, there was no significant difference in the colony size between plots with 5 and 10 nestboxes. Pairs nested most frequently in groups of four to five pairs in both plots with 5 or 10 nest-boxes, whereas in the plots with 3 nest-boxes, naturally, in smaller groups of two to three pairs. Although our results about the role of nesting site type need to be treated with caution because plot size was confounded to the year of study, it seems that breeding in groups of four to five pairs at the nest site may have been beneficial. We have two indications for this assumption. First, pairs in the plots with 3 nest-boxes started breeding at approximately same time as the pairs in the plots with 5 and 10 nest-boxes, but it took them longer to fledge their chicks (Fig.1). This may be disadvantageous since it is known that offspring produced late in the season have a shorter time to build up their energetic reserves for the winter, reach a lower hierarchical status and suffer higher mortality (L a c k 1954, Klomp 1970, see D aan & Tinbergen 1997). Second, despite the positive relationship between clutch size and colony size, pairs in 5 nest-box plots had significantly larger clutches than pairs in 10 nest-box plots (Table 1). Therefore, it seems that availability of nest-boxes could serve as a habitat cue reflecting future colony size. Consequently, birds may try to avoid nesting sites that deviate from the preferred colony size. In contrast to colony size, we found that breeding synchrony influenced hatching success. Namely, more chicks hatched under both natural and experimental conditions in situations when pairs laid eggs at the nest site more synchronously. Again, breeding synchrony did not seem to largely influence paternity assurance behaviour but paternity losses appeared to decrease with higher breeding synchrony (Fig. 8). Hence, this finding supports Birkhead & Biggins (1987) hypothesis that EPP decreases with breeding synchrony due to the incompatibility of males to guard their mates and seek extra-pair copulations simultaneously. This is in contrast to the hypothesis of S tutchbury & Morton (1995), which implies an increase in extra-pair paternity with breeding synchrony because females should have better possibilities to compare the quality of possible copulation partners that all breed at the same time. In contrast to paternity assurance behaviour, we found that breeding synchrony significantly influenced investment in nest guarding behaviour. Synchrony is usually thought to be advantageous because it decreases the risk of nest depredation (e.g. W ittenberger & Hunt 1985, I m m s 1990) but may also reflect the time when the environmental conditions for breeding are at best (Brown & Brown 1996). However, in our population house sparrows had three breeding attempts with fluctuating reproductive success, but without obvious fluctuations of breeding synchrony. Synchrony may, however, mediate different levels of intraspecific competition. Indeed we found that nest guarding behaviour apparently increased with decreasing synchrony. V e i g a (1990, 1992) showed that competition between females over the superior males, eventually leading to egg destruction, decreases with breeding synchrony. Since we have shown that egg losses are in general very high in house sparrows and substantially influence reproductive success also in our population, nest guarding seems to be very important in this species. Moreover, our previous works (V á c l a v & H o i in 46

46 press, V áclav et al. in press) show that egg losses were highest in the nests of males that were cuckolded most often, possibly reflecting male harassment upon females. Hence, since males are cuckolded less often and females suffer lower clutch losses when breeding synchronously, intra-sexual competition in both sexes may favour breeding synchrony and consequently genetic monogamy. The reason why house sparrows do not prefer to breed in large colonies and prefer to stay in loose groups might therefore be rooted in avoidance of breeding asynchronously. To summarise, our study showed that colony size does not seem to be a factor substantially explaining reproductive success, extra-pair behaviour and paternity losses in house sparrows. Although clutch size increased with colony size, birds preferred nesting in rather small groups. Higher hatching success and possibly also higher fledging success and lower risk of paternity losses increased when birds nested more synchronously. Since synchrony decreased with colony size in our study population, this might be a reason why house sparrows of both sexes prefer to breed in small groups. A c k n o w l e d g e m e n t s We thank the staff of Vienna Zoo in Schönbrunn for their permission to carry out this study and their field assistance during two years. We thank Donald B l o m q v i s t and an anonymous referee for their comments on this manuscript. R.V. was funded from the exchange programme between Austrian Academy of Sciences and Slovak Academy of Sciences and by grants from the Jubiläumsfonds der Stadt Wien STI L I T E R A T U R E BIRKHEAD, T. R., ATKIN, L. & MØLLER, A. P., 1987: Copulation behaviour of birds. Behaviour, 101: BIRKHEAD, T. R. & BIGGINS, J. D., 1987: Reproductive synchrony and extra-pair copulations in birds. Ethology, 74: BIRKHEAD, T. R. & MØLLER, A. P., 1992: Sperm competition in birds: evolutionary causes and consequences. Academic Press, London. BROWN, J. L., 1987: Helping and Communal Breeding in Birds: Ecology and Evolution. Princeton University Press, Princeton. BROWN, C. R. & BROWN, M. B., 1996: Coloniality in the cliff swallow: the effect of group size on social behaviour. University of Chicago Press, Chicago. CORDERO P. J., WETTON, J. H. & PARKIN, D. T., 1999: Extra-pair paternity and male badge size in the House Sparrow. J. Avian Biol., 30: CRAMP, S., 1994: The birds of the Western Palearctic, Vol. IX.: Buntings and New World Warblers. Oxford University Press, Oxford. DAAN, S. & TINBERGEN, J. M., 1997: Adaptations of Life Histories. In: Krebs, J. R. & Davies, N. B. (eds.), Behavioural ecology: an evolutionary approach. Blackwell Science, Oxford: EMLEN, S. T. & ORING, L. W., 1977: Ecology, sexual selection and the evolution of mating systems. Science, 197: GOWATY, P. A. & BRIDGES, W. C., 1991: Nestbox availability affects extra-pair fertilisations and conspecific nest parasitism in eastern bluebirds, Sialia sialis. Anim. Behav., 41: GRIFFITH, S. C., 2000: A trade-off between reproduction and a condition-dependent sexually selected ornament in the house sparrow Passer domesticus. Proc. R. Soc. Lond., B 267: HATCHWELL, B. J., 1988: Intraspecific variation in extra-pair copulation and mate defence in common guillemots Uria aalge. Behaviour, 107: HILL, G. E., MONTGOMERIE, R., ROEDER, C. & BOAG, P. T., 1994: Sexual selection and cuckoldry in a monogamous songbird: implications for sexual selection theory. Behav. Ecol. Sociobiol., 35: HOI, H. & HOI-LEITNER, M., 1997: An alternative route to coloniality in the bearded tit: females pursue extrapair fertilizations. Behav. Ecol., 8:

47 HOI, H., KLEINDORFER, S., ILLE, R. & DITTAMI, J., 1995: Prey abundance and male parental behaviour in Acrocephalus warblers. Ibis, 137: IMMS, A. D., 1990: On the adaptive value of reproductive synchrony as a predator-swamping strategy. Am. Nat., 136: KLOMP, H., 1970: Regulations of the size of bird populations by means of territorial behaviour. Netherlands Journal of Zoology, 22: KROKENE, C., ANTHONISEN, K., LIFJELD, J. T. & AMUNDSEN, T., 1996: Paternity and paternity assurance behaviour in the bluethroat, Luscinia s. svecica. Anim. Behav., 52: LACK, D., 1954: The Natural Regulations of Animal Numbers. Clarendon Press, Oxford. LACK, D., 1968: Ecological Adaptations for Breeding in Birds. Chapman & Hall, London. MØLLER, A. P., 1985: Mixed reproductive strategy and mate guarding in semi-colonial passerine, the swallow Hirundo rustica. Behav. Ecol. Sociobiol., 17: MØLLER, A. P., 1987: House sparrow (Passer domesticus) communal displays. Anim. Behav., 35: MØLLER, A. P. & BIRKHEAD, T. R., 1993: Cuckoldry and sociality a comparative-study of birds. Am. Nat. 142: MØLLER, A. P. & NINNI, P., 1998: Sperm competition and sexual selection: a meta-analysis of paternity studies of birds. Behav. Ecol. Sociobiol., 43: SLAGSVOLD, T. & DALE, S., 1991: Mate choice models: can cost of searching and cost of courtship explain mating patterns of female pied flycatchers? Ornis Scand., 22: STUTCHBURY, B. J. & MORTON, E. S., 1995: The effect of breeding synchrony on extra-pair mating systems in songbirds. Behaviour, 132: STUTCHBURY, B., RHYMER, J. M. & MORTON, E. S., 1994: Extra-pair paternity in the hooded warbler. Behav. Ecol., 5: SUMMER-SMITH, D., 1954: The communal display of the house sparrow Passer domesticus. Ibis, 96: THUSIUS, K.J., DUNN, P. O., PETERSON, K. A. & WHITTINGHAM, L. A., 2001: Extrapair paternity is influenced by breeding synchrony and density in the common yellowthroat. Behav. Ecol., 12: TOST, J., 1994: The effects of colony size and male quality on paternity assurance strategies in house sparrows (Passer domesticus L.). Unpublished M.Sc. thesis. University of Vienna, Vienna. VÁCLAV, R. & HOI, H., 2002: Different reproductive tactics in house sparrows signalled by badge size: is there a benefit of being average? Ethology (in press). VÁCLAV, R., HOI, H. & BLOMQVIST, D., 2002: Badge size, paternity assurance behaviours and paternity losses in male house sparrows. J. Avian. Biol. (in press). VEIGA, J. P., 1990: Sexual conflict in the house sparrow: interference between polygynously mated females vs. asymmetric male investment. Behav. Ecol. Sociobiol., 27: VEIGA, J. P., 1992: Why are house sparrows predominantly monogamous? A test of hypotheses. Anim. Behav., 43: VEIGA, J. P. & BOTO, L., 2000: Low frequency of extra-pair fertilisations in House Sparrows breeding at high density. J. Avian Biol., 31: WAGNER, R., 1993: The pursuit of extra-pair copulations by female birds: a new hypothesis of colony formation. J. Theor. Biol., 163: WEATHERHEAD, P. J., 1997: Breeding synchrony and extra-pair mating in red-winged blackbirds. Behav. Ecol. Sociobiol., 40: WESTNEAT, D. F., 1992: Do female red-winged blackbirds pursue a mixed-mating strategy? Ethology, 92: WESTNEAT, D. F. & SHERMAN, P. W., 1997: Density and extra-pair fertilizations in birds: a comparative analysis. Behav. Ecol. Sociobiol., 41: WESTNEAT, D. F., SHERMAN, P. W. & MORTON, M. L., 1990: The ecology and evolution of extra-pair copulations in birds. In: Power, D. M. (ed.), Current ornithology. Vol. 7. Plenum Press, New York: WETTON, J. H., BURKE, T., PARKIN, D. T. & CAIRNS, E., 1995: Single-locus DNA fingerprinting reveals that male reproductive success increases with age through extra-pair paternity in the house sparrow (Passer domesticus). Proc. R. Soc. Lond B, 260: WETTON, J. H, CARTER, R. E., PARKIN, D. T. & WALTERS, D., 1987: Demographic study of a wild House Sparrow population by DNA fingerprinting. Nature, 327: WETTON, J. H. & PARKIN, D. T., 1991: An association between fertility and cuckoldry in the house sparrow, Passer domesticus. Proc. R. Soc. Lond B, 245: WITTENBERGER, J. L. & HUNT, G. L. Jr., 1985: The adaptive significance of coloniality in birds. In: Farner, D. S., King, J. R. & Parkes, K. C. (eds.), Avian Biology. Vol. 8. Academic Press, Orlando:

48 Folia Zool. 51(1): (2002) Bagrichthys majusculus, a new catfish from Indochina (Teleostei, Bagridae) Heok Hee NG Fish Division, Museum of Zoology, University of Michigan, 1109 Geddes Avenue, Ann Arbor, Michigan , USA; heokheen@umich.edu Received 20 November 2000; Accepted 29 August 2001 A b s t r a c t. Bagrichthys majusculus, a new species of bagrid catfish from Indochina, is very similar to B. macracanthus and B. vaillantii, and has been previously identified as the former species. It differs from congeners in having a unique combination of the following characters: relatively large and broad mouth, well-developed oral dentition with homodont teeth, gill rakers, moderately-long dorsal spine with serrations, 9 pectoral-fin rays, inner and outer mandibular barbels with straight margins, pectoral-spine length % SL (standard length), dorsal-spine length % SL, length of adipose-fin base % SL, adipose maximum height % SL, depth of caudal peduncle % SL, and head depth % SL. Key words: new species, B. macracanthus, B. vaillantii, Southeast Asia Introduction The genus Bagrichthys consists of highly-specialised bagrid catfishes found in large rivers throughout Southeast Asia. They are characterised by their elongate and laterally compressed caudal peduncle, dorsally-directed serrations on the posterior edge of the dorsal-fin spine, gill membranes united at the isthmus, and a long adipose fin without a free posterior margin. The taxonomy of Bagrichthys has been the subject of recent work (R o b e r t s 1989, N g 1999, 2000), and six species are currently recognised, viz. B. hypselopterus (Bleeker, 1852), B. macropterus (Bleeker, 1853), B. macracanthus (Bleeker, 1854), B. vaillantii (Popta, 1906), B. micranodus Roberts, 1989, and B. obscurus Ng, Whilst comparing specimens identified as B. macracanthus from Indochina with those from Sumatra and Borneo, several significant differences in morphology were observed between the two populations. The population from Indochina is thus described in this study as belonging to a new species, B. majusculus. Material and Methods Measurements were made point to point with dial callipers and data recorded to 0.1 mm. Counts and measurements were made on the left side of specimens whenever possible. Subunits of the head are presented as proportions of head length (HL). Head length itself and measurements of body parts are given as proportions of standard length (SL). Measurements and counts were made following N g (2000). Fin rays were counted under a binocular dissecting microscope using transmitted light. Vertebral counts were taken from radiographs following the method of R o b e r t s (1994). Numbers in parentheses following a particular fin-ray, branchiostegal-ray, gill-raker or vertebral count indicate the number of specimens with that count. Institutional codes follow Eschmeyer (1998). 49

49 Bagrichthys majusculus, new species (Fig. 1) Pseudobagrichthys macracanthus (non Bleeker, 1854) Bleeker 1865a: 34, 1865b: 175. Bagroides macracanthus (non Bleeker, 1854) Sauvage 1883: 161, Durand 1940: 30, S m i t h 1945: 378, Suvatti 1936: 77, 1950: 292, Kuronuma 1961: 6, T a k i 1968: 20, 1974: 50, Fig. 51, Orsi 1974: 161, D e s outter 1975: 458, K o ttelat 1985: 270, M a i & Nguyen 1988: 49, Maiet al. 1992: 185. Bagrichthys hypselopterus (non Bleeker, 1852) Chevey 1932: 17, Chaux & Fang 1949: 342, Kuronuma 1961: 6. Bagroides hypselopterus (non Bleeker, 1852) K o ttelat 1989: 13. Bagrichthys macracanthus (non Bleeker, 1854) Jayaram 1968: 378 (in part), Roberts 1993: 33, Rainboth 1996: 139. Holotype: UMMZ , male, mm SL; Thailand: Ubon Ratchathani province, Khong Chiam district, Mun River at Ban Dan, 3 km upstream from Mekong River confluence; W. J. Rainboth, R. E. Arden & Y. Dhammigbavorn, 12 June Paratypes: CAS 92487, 1 male, mm SL; 1 female, mm L; Laos: Mekong at Ban Hang Khone, just below Khone Falls; T. R. Roberts, June July MCZ 51318, 1 male, mm SL, 1 female, mm SL; Thailand: Nong Khai market; T. R. Roberts, 13 May UMMZ , 1 female, 96.8 mm SL; Vietnam: Phung Ding province, S side of Can Tho island, 3 km SE of Can Tho; E. D. Buskirk, UMMZ , 1 female, mm SL; Thailand: Khon Kaen province, Nam Pong Reservoir, 50 m from E shore, 3 km S of fish landing; W. J. Rainboth et al., September UMMZ , 1 male, mm SL; 1 female, mm SL; Cambodia: Stung Treng morning market; W. J. Rainboth, N. van Zalinga & C. Rotha; 26 January UMMZ , 1 female, mm SL; Cambodia: Kandal province, Tonle Sap; W. J. Rainboth & C. Rotha, 13 February UMMZ , 1 male, mm SL; Laos: Mekong River at Ban Hang Khone, just below Khone Falls; I. Baird, date unknown. USNM , 1 female, mm SL; Thailand: Roi Et province, Lam Chi, 1.5 km below highway 23 bridge, 4 km W of Selaphum; WBD- Mekong expedition, 31 January USNM , 1 female, mm SL; Thailand: Ubon Ratchathani province, market in Ubon at edge of Mun River; WBD-Mekong expedition, 14 September Diagnosis: Bagrichthys majusculus differs from other congeners in having a unique combination of the following characters: relatively large and broad mouth, well-developed oral dentition with homodont teeth, gill rakers, moderately long dorsal spine with Fig. 1. Bagrichthys majusculus, paratype male, MCZ 51318, mm SL; Thailand: Nong Khai. 50

50 15 27 serrations, 8 9 pectoral-fin rays, inner and outer mandibular barbels with straight margins, pectoral-spine length % SL, dorsal-spine length % SL, length of adipose-fin base % SL, adipose maximum height % SL, caudal peduncle depth % SL, and head depth % SL. Description: Body long and compressed. Dorsal profile rising evenly and steeply from tip of snout to origin of dorsal fin, then sloping ventrally from there to end of caudal peduncle. Ventral profile flat to anal-fin base, then sloping dorsally from there to end of caudal peduncle. Anus and urogenital openings located at vertical through middle of adpressed pelvic fin. Skin smooth. Lateral line complete and midlateral, with long tubular epidermal extensions of the sensory pores. Head somewhat compressed and narrow, anterior part produced into a prominent protruding snout with somewhat truncate margin when viewed laterally. Mouth inferior and relatively small, with papillate lips. Gill openings wide, extending from exposed surface of posttemporal to beyond isthmus. Gill membranes free from, and not attached across, isthmus. Eye ovoid, horizontal axis longest; located entirely in dorsal half of head. Orbit with free margin. Barbels in four pairs. Maxillary barbel moderately long, extending to opercular flap. Nasal barbel slender and short, extending to posterior margin of orbit (females) or to midway between posterior margin of orbit and base of supraoccipital (males). Inner mandibular-barbel origin close to midline; barbel thicker and longer than nasal barbel and extending to level of anterior margin of orbit. Outer mandibular barbel originates posterolateral of inner mandibular barbel, extending to level of centre of orbit. Oral teeth small and in irregular rows on all tooth-bearing surfaces. Premaxillary tooth band rounded, of equal width throughout. Dentary tooth band broader than premaxillary tooth band, widening laterally and then tapering to a sharp point poterolaterally. Vomerine tooth band unpaired, continuous across midline; strongly arched along anterior margin; band width broader than premaxillary band. Dorsal fin located at middle of body; origin nearer tip of snout than caudal flexure. Dorsal-fin margin convex, usually with anterior branch of fin rays longer than other branches. Last dorsal-fin ray without posterior membranous connection to body. Dorsal-fin spine short, straight and robust, with serrations on posterior margin. Adipose fin with margin convex for entire length; posterior end deeply incised. Caudal fin deeply forked; upper and lower lobes pointed, with outermost principal fin-rays produced into filaments. Procurrent rays symmetrical and extending only slightly anterior to fin base. Anal-fin base ventral to posterior half of adipose fin. Fin margin curved or straight. Last anal-fin ray without posterior membranous connection to body. Pelvic-fin origin at vertical through posterior end of dorsal-fin base. Pelvic-fin margin slightly convex, tip of adpressed fin not reaching anal-fin origin. Pectoral fin with stout spine, sharply pointed at tip. Anterior spine margin smooth; posterior spine margin with serrations along entire length. Pectoralfin margin straight anteriorly, convex posteriorly. In % SL: head length , head width , head depth , predorsal distance , preanal length , prepelvic length , prepectoral length , body depth at anus , length of caudal peduncle , caudal peduncle depth , pectoralspine length , pectoral-fin length , dorsal-spine length , length of dorsal-fin base , pelvic-fin length , length of anal-fin base , caudal-fin length , length of adipose-fin base , adipose maximum height , post-adipose distance ; in % HL: snout length , interorbital distance , eye diameter , nasal barbel length (females) or 51

51 (males), maxillary barbel length (females) or (male), inner mandibular barbel length (females) or (males), outer mandibular barbel length (females) or (males). Branchiostegal rays 5 (7) or 6 (1). Gill rakers 3+7 (2), 3+9 (2), 4+8 (1) or 3+10 (1). Vertebrae 17+25=42 (1), 18+24=42 (1), 18+25=43 (3), 18+26=44 (2), 19+25=44 (4) or 19+26=45 (1). Fin ray counts: dorsal II,7 (8); pectoral I,8 (1) or I,9 (7); pelvic i,5 (8); anal iii,10 (1), iv,9 (1), iv,10 (2), iii,11 (1), iv,11 (2) or v,10 (1); caudal 8/9 (8). Colour in alcohol: dorsal surface of head and body uniform brown or dark brown, with a cream-coloured midlateral streak running along entire length of body. Large regularly spaced cream-coloured patches occasionally present on nape and sides of body below anterior third of adipose fin and immediately above posteriormost anal-fin ray; patches sometimes coalescing to form bands. Ventral surfaces of head and body cream; adipose fin and fin rays of all fins brown; inter-radial membranes of all fins with scattered melanophores. Distribution: Known from the Mekong and Chao Phraya River drainages in Indochina. Ecology: Bagrichthys majusculus feeds on crustaceans and other small benthic animals (T a k i 1978). It spawns at the beginning of the rainy season in flooded riparian forests, with juveniles beginning to appear in August (R a i n b o t h 1996). Etymology: From the Latin majusculus, meaning somewhat greater, in reference to the relatively larger adipose fin base, pectoral and dorsal spines of this species when compared to B. macracanthus, its closest congener. Remarks: Bagrichthys majusculus differs from B. hypselopterus in having fewer gill rakers on the first gill arch (10 13 vs ), serrations on the posterior margin of the pectoral spine (15 27 vs. 60 or more) and pectoral-fin rays (8 9 vs ), the nape and dorsal-fin base slightly (vs. very strongly) elevated, inner and outer mandibular barbels with straight margins (vs. convoluted anterior margins), and jaw teeth homodont, i.e. of one kind (vs. markedly heterodont, i.e. of different kinds). It differs from B. macropterus, B. micranodus, and B. obscurus in having a relatively large and broad (vs. small and narrow) mouth with well-developed, i.e. with exposed teeth forming well defined tooth bands (vs. extremely reduced, i.e. with few scattered teeth deeply buried in soft tissue) oral dentition, and a longer pectoral spine ( % SL vs ). Bagrichthys majusculus most closely resembles B. macracanthus and B. vaillantii, but can be differentiated from both species in having a longer pectoral spine (pectoral-spine length % SL vs ), the former in having a larger adipose fin (length of adipose-fin base % SL vs ), deeper caudal peduncle (depth of caudal peduncle % SL vs ), and the latter in having a longer dorsal spine (dorsal-spine length % SL vs ), more slender head (head depth % SL vs ) and a deeper adipose fin (adipose maximum height % SL vs ). From the results of this study and a previous one by N g (1999), it is clear that there are two species of Bagrichthys in Indochina that are closely related to morphologically similar Sundaic species. Although R a i n b o t h (1996) postulated that there may be more than one species of Bagrichthys with long dorsal spines in the Mekong River, the results here indicate that B. majusculus is the only such species present in the Mekong. The historical biogeography of Indochinese Bagrichthys is probably similar to that for Hemisilurus proposed by B ornbusch & Lundberg (1989), i.e. an ancestral Bagrichthys in the North Sunda River system splitting into the B. hypselopterus + B. macracanthus + B. majusculus (diagnosed as having long dorsal spines with 18 or more serrations in adults, and exposed jaw and palatal teeth forming well defined tooth bands) 52

52 and B. macropterus + B. micranodus + B. obscurus lineages (diagnosed as having short dorsal spines with 15 or less serrations in adults, and scattered jaw and palatal teeth deeply buried in soft tissue) not later than the Pleistocene, with subsequent divergence by the post- Pleistocene isolation of the Indochinese and Sundaic populations. Comparative material: Bagrichthys hypselopterus: CAS 49366, 2 ex., mm SL; Borneo: Kalimantan Barat, fish market at Sintang. CAS 49367, 1 ex., mm SL; Borneo: Kalimantan Barat, Sungai Tawang near Pengembung. ZRC 40472, 1 ex., mm SL; ZRC 41532, 15 ex., mm SL; ZRC 41898, 1 ex., mm SL; Sumatra: Jambi, Pasar Angso Duo (fish market). Bagrichthys macracanthus: CMK 11278, 2 ex., Sumatra: Jambi, Danau Arang Arang. ZMA , 3 ex., mm SL; Sumatra: Jambi, Batang Hari. ZRC 14251, 1 ex., mm SL; Peninsular Malaysia: Selangor, North Selangor peat swamp forest, pools at Sungai Bernam headworks. ZRC 38547, 2 ex., mm SL; Sumatra: Jambi, Sungai Kumpeh in Arang Arang. ZRC 38633, 1 ex., mm SL; Sumatra: Jambi, Danau Arang Arang. ZRC 39034, 1 ex., mm SL; Sumatra: Riau, Sungai Bengkwan, tributary of Indragiri River, 4 hours downstream from Rengat. ZRC 41533, 1 ex., mm SL; ZRC 41897, 1 ex., mm SL; ZRC 44118, 12 ex., mm SL; Sumatra: Jambi, Pasar Angso Duo (fish market). ZRC 42601, 1 ex., 52.8 mm SL; Sumatra: Jambi. ZRC 44161, 4 ex., mm SL; Sumatra: Jambi, Sungai Alai. ZRC 46030, 2 ex., mm SL; Borneo: Sarawak, Kapit market. Bagrichthys macropterus: USNM , 4 ex., mm SL; Borneo: Kalimantan Barat, Kapuas mainstream 58 km NE of Sintang and 1 km downstream from Sebruang. ZRC 38997, 1 ex., mm SL; ZRC 41534, 5 ex., mm SL; Sumatra: Jambi: Pasar Angso Duo (fish market). Bagrichthys micranodus: CAS 49369, 2 paratypes, mm SL; USNM , 3 paratypes, mm SL; Borneo: Kalimantan Barat, Kapuas River 58 km NE of Sintang and 1 km downstream from Sebruang. CMK 7015, 1 ex., mm SL; Borneo: Kalimantan Barat, Kapuas River at Nanga Embaloh. Bagrichthys obscurus: USNM , holotype, mm SL; Thailand: Roi Et province, Lam Chi, 1.5 km below highway 23 bridge, 4 km W of Selaphum. CAS 92567, 1 paratype, mm SL; Laos: Mekong below Khone Falls. CAS 92580, 2 paratypes, mm SL; Laos: Mekong at Ban Hang Khone, just below Khone Falls. UMMZ , 1 paratype, mm SL; UMMZ , 1 paratype, mm SL; Thailand: Ubon Ratchathani province, Khong Chiam district, Mun River at Ban Dan, 3 km upstream from confluence with Mekong River. UMMZ , 2 paratypes, mm SL; Cambodia: Stung Treng morning market. USNM , CAS 61918, 22 ex., mm SL; Thailand: fish market at Ubon Ratchathani. Bagrichthys vaillantii: RMNH 7839, 1 ex., 90.6 mm SL; Borneo: Tepoe. CAS 93398, 1 ex., mm SL; eastern Borneo. CAS 93403, 1 ex., mm SL; Borneo: Kalimantan Timur, Sungai Belayan from mouth of Sungai Sentekan southwards for about 6 km. CMK 7635, 5 ex., mm SL; Borneo: Kalimantan Timur, stream connecting Mahakam River to Danau Semajang. Acknowledgements I am grateful to the following for permission to examine material under their care: William E s c h m e y e r (CAS), Maurice K o t t e l a t (CMK), Karsten H a r t e l (MCZ), Martien van O i j e n (RMNH), Douglas Nelson (UMMZ), Lynne Parenti (USNM), Isaäc I s brücker (ZMA), Peter N g (ZRC). I also thank H. H. T a n for assistance in obtaining data. This work was funded by support from the Rackham School of Graduate Studies, University of Michigan. 53

53 LITERATURE BLEEKER, P., 1865a: Nouvelle notice sur la faune ichtyologique de Siam. Ned. Tijdschr. Dierk., 2: BLEEKER, P., 1865b: Sixième notice sur la faune ichtyologique de Siam. Ned. Tijdschr. Dierk., 2: BORNBUSCH, A. H. & LUNDBERG, J. G., 1989: A new species of Hemisilurus (Siluriformes, Siluridae) from the Mekong River, with comments on its relationships and historical biogeography. Copeia, 1989: CHAUX, J. & FANG, P. W., 1949: Catalogue des Siluroidei d Indochine de la collection du laboratoire des pêches coloniales au muséum, avec la description de six espèces nouvelles. (Suite et fin). Bull. Mus. Natn. Hist. Nat., 2e sér., 21: CHEVEY, P., 1932: Inventaire de la faune ichtyologique de l Indochine. Deuxième liste. Notes Inst. Océanogr. Indochine, 19: DESOUTTER, M., 1975: Étude de quelques Bagridae (Siluriformes, Pisces) du Cambodge. Description d une espèce nouvelle : Mystus aubentoni. Bull. Mus. Natn. Hist. Nat., 3e sér., 206: DURAND, J., 1940: Notes sur quelques poissons d espèces nouvelles ou peu connues des eaux douces cambodgiennes. Notes Inst. Océanogr. Indochine, 36: ESCHMEYER, W. N., 1998: Catalog of Fishes. California Academy of Sciences, San Francisco. JAYARAM, K. C., 1968: Contributions to the study of bagrid fishes (Siluroidea: Bagridae). 3. A systematic account of the Japanese, Chinese, Malayan, and Indonesian genera. Treubia, 27: KOTTELAT, M., 1985: Fresh-water fishes of Kampuchea A provisory annotated check-list. Hydrobiologia, 121: KOTTELAT, M., 1989: Zoogeography of the fishes from Indochinese inland waters with an annotated check-list. Bull. Zoöl. Mus., 12: KURONUMA, K., 1961: A Check List of the Fishes of Vietnam. Division of Agriculture and Natural Resources United States Operations Mission to Vietnam. MAI, D. Y. & NGUYEN, V. T., 1988: Species composition and distribution of the freshwater fish fauna of southern Vietnam. Hydrobiologia, 160: MAI, D. Y., NGUY ~ EN, V. TRỌNG, NGUY ~ EN, V. THIỆN, LÊ, H. Y. & HÚ A, B. L., 1992: Dịnh loại các loàicá nu ó c ngọt Nam Bộ. [Identification of the Fresh-water Fishes of South Vietnam]. Nhà Xuâ t Bán Khoa Học Và Ky ~ Thuật, Hanoi. NG, H. H., 1999: Bagrichthys obscurus, a new species of bagrid catfish from Indochina (Teleostei: Bagridae). Rev. Biol. Trop., 47: NG, H. H., 2000: Bagrichthys vaillantii (Popta, 1906), a valid species of bagrid catfish from eastern Borneo (Teleostei: Siluriformes). Zool. Med., 73: ORSI, J. J., 1974: A check list of the marine and freshwater fishes of Vietnam. Pub. Seto Mar. Biol. Lab., 21: RAINBOTH, W. J., 1996: Fishes of the Cambodian Mekong. FAO Species Identification Field Guide for Fishery Purposes, FAO, Rome. ROBERTS, T. R., 1989: The freshwater fishes of Western Borneo (Kalimantan Barat, Indonesia). Mem. California Acad. Sci., 14: ROBERTS, T. R., 1993: Artisanal fisheries and fish ecology below the great waterfalls of the Mekong river in southern Laos. Nat. Hist. Bull. Siam Soc., 41: ROBERTS, T. R., 1994: Systematic revision of the Asian bagrid catfishes of the genus Mystus sensu stricto, with a new species from Thailand and Cambodia. Ichthyol Explor. Freshwaters, 5: SAUVAGE, H. E., 1883: Sur une collection de poissons recueillis dans le Mé-Nam (Siam) par M. Harmand. Bull. Soc. Philom. Paris, Ser. 7, 7: SMITH, H. M., 1945: The freshwater fishes of Siam or Thailand. Bull. U. S. Natn. Mus., 188: SUVATTI, N. C., 1936: Index to Fishes of Siam. Bureau of Fisheries, Bangkok. SUVATTI, N. C., 1950: The Fauna of Thailand. Department of Fisheries, Bangkok. TAKI, Y., 1968: Notes on a collection of fishes from lowland Laos. United States Agency for International Development Mission to Laos Agriculture Division. TAKI, Y., 1974: Fishes of the Lao Mekong basin. United States Agency for International Development Mission to Laos Agriculture Division. TAKI, Y., 1978: An analytical study of the fish fauna of the Mekong basin as a biological production system in nature. Research Institute of Evolutionary Biology Special Publications, 1, 77 pp. 54

54 Folia Zool. 51(1): (2002) Movements of barbel, Barbus barbus (Pisces: Cyprinidae) Dedicated to the late Professor Antonín Lelek - our friend and colleague Milan PE ÁZ, Vlastimil BARU, Miroslav PROKE and Miloslav HOMOLKA Institute of Vertebrate Biology, Academy of Sciences of the Czech Republic Kvûtná 8, Brno, Czech Republic; penaz@brno.cas.cz Received 10 May 2001; Accepted 7 August 2001 A b s t r a c t. Altogether 701 adult barbel, Barbus barbus were captured by electrofishing and individually tagged to study their local displacement and movements in a stretch of the River Jihlava (Czech Republic). A total of 149 fish were recaptured and 105 of them (70.47 %) were considered as resident because they were always recaptured in the same, relatively restricted ( m) stream section, which always contained a pool and was demarcated naturally by riffles on both edges. The remaining 44 recaptured specimens (29.53 %) belonged to the mobile part of population, their movements encompassing two (or exceptionally more) adjacent stream sections and at maximum a distance of 1680 m downstream or 2020 m upstream. The proportion of mobile barbel, relatively low in smaller and middle size classes, increased in the largest size classes ( mm of SL). A rather limited extent of movements also suggests a relatively small area of home range in the studied stretch, which nevertheless provides satisfactory resources and favourable conditions required by barbel over their entire life cycle. The extent of movements and corresponding proportion of mobile fish appear to be increasing with diminishing habitat patchiness. In the stretch of River Jihlava studied, with a rich patchy heterogenous habitat and well developed riffle-pool-raceway structure, each section (pool) can be considered as a more or less isolated spatial unit containing its own, and in a certain degree, isolated component of a metapopulation. Key words: tagging, resident and mobile fish, territoriality, metapopulation, riffle-pool development Introduction The barbel Barbus barbus (Linnaeus, 1758) is a popular and much studied species whose biology is still insufficiently understood. Unfortunately it is also a species currently vanishing from many European rivers. Already four International Round Tables (Montpellier France in 1989, Liége - Belgium in 1993, Liblice Czech Republic in 1995 and Thessaloniki Greece in 1997) have been devoted to this and other Barbus species, providing a good information tool of actual knowledge and a solid base stimulating their further research. Within this context, the barbel is also the subject of an extensive research project on the River Jihlava, Czech Republic. The construction and operation of hydroelectric complex on this stream and its close vicinity created conditions that interfered with natural environmental stimuli (temperature, photoperiod) triggering gonad maturation and spawning of the barbel (PeÀáz et al. 1999). The population appeared to be suppressed for a period after the sudden change in habitat caused by the start of hydroelectric operation operating energy complex. Subsequently, the adaptation processes seems to have been completed and the barbel population was found again viable, prosperous and dominating the fish community (P eàáz et al. 1999). The aim of present partial study was to analyse and 55

55 quantify the movements of barbel as well as to assess the results relative to situation existing in some other rivers and barbel populations studied in a similar way (Steinmann et al. 1937, Hunt & Jones 1974, Pelz & Kästle 1989, Baras 1995, 1997, Lucas & Batley 1996). Study Area Our survey took place in a 3.0 km stretch of the River Jihlava (Czech Republic), a fifth-order stream (tributary to the River Svratka, Danube basin) near the village of Hrub ice between river kilometers (RK) and (49 07 N, E) (Fig. 1). The study stretch is characteristic of submountain streams, running through an incised valley, morphologically heterogenous, corresponding to a barbel zone (H u e t 1949) and actually influenced by an upstream located energy operating hydro- and nuclear complex. The main environmental impacts exerted on the river, described in particular by PeÀáz et al. (1999), were a regulated discharge and reduced temperature variations. These, rather favourable habitat changes were followed by considerable changes in composition of fish assemblage. The former fish community dominated by barbel and other stream cyprinids was replaced by a salmonid community dominated by brown trout Salmo trutta m. fario, however, the barbel maintained a strong and in fact increased presence in present fish community, which consists of at least 25 species. Except of barbel, species recorded during experimental electrofishing were rainbow trout Oncorhynchus mykiss, brook trout Salvelinus fontinalis, grayling Thymallus thymallus, roach Rutilus rutilus, chub Leuciscus cephalus, dace L. leuciscus, rudd Scardinius erythrophthalmus, tench Tinca tinca, nase Chondrostoma nasus, gudgeon Gobio Fig. 1. Map of the study area. 56

56 gobio, bleak Alburnus alburnus, spirlin Alburnoides bipunctatus, crucian carp Carassius carassius, common carp Cyprinus carpio, eel Anguilla anguilla and, in addition, by sport fishing were recorded also common bream Abramis brama, Eurasian perch Perca fluviatilis, pike-perch Stizostedion lucioperca, asp Aspius aspius, ide Leuciscus idus, goldfish Carassius auratus and grass carp Ctenopharyngodon idella (PeÀáz et al. 1999). The habitat of river under study was very heterogenous and had a typical patchy riffle pool raceway structure, with fully natural and stabilised river bed and banks, with abundant spawning grounds but with no tributaries habitable for barbel (Table 1). The study stretch was divided into seven sections (A G), each containing a complete set of various microhabitats, mainly a deeper pool, long raceway and separated between adjacent sections by riffles as natural boundaries. These patches differ mainly by depth and flow velocity: riffles were characterised by a water velocity of m.s -1 and a depth of cm, raceway by m.s -1 and cm while the pools by m.s -1 and a depth of cm. The shallow parts of raceways are often overgrown by dense macrophytes (Batrachium fluitans, Callitriche sp.) during spring and summer. Table 1. Sections of the stretch of River Jihlava under study and their environmental characters (RK = river kilometers). Section RK Area Length Proportion of patches, % hectares m riffle pool raceway A B C D E F G Total Methods and Material A gasoline powered electroshocker (DC, 250 V, A, 50 Hz) was used to collect the fish. Electrofishing excursions were carried out on 15 dates: , , , , , , , , , , , , , and All fish except of barbel 120 mm of SL were after capture immediately released. After being measured, weighed and examined for sex according to external features (mainly running sexual products) recognisable only during reproduction season, 701 barbel were individually marked by means of the anchor full plastic tags (Floy Tag - type FD-94). Tags were fixed on the left body side into the dorsal musculature, just below the anterior edge of dorsal fin. Different colours of tags were applied in consecutive years (yellow 1999, white 2000, red 2001). The effect of tags on fish behaviour, survival and growth was minimised by the type of tags used and their careful application. The tags proved to be durable and easy to recognise in the field even after long periods (2 years). The tagged and recaptured fish were always released in centres of the same particular sections (see Table 1). The distance between centres of sections where the fish were released 57

57 and centres of sections where they were recaptured, was considered as distance that a recaptured fish had moved. All recaptures were obtained by the experimental fishing exclusively. Barbel do not represent a favourite coarse fish in sport fishery conducted in the stretch of River Jihlava under study, where presently the brown trout and other salmonid fishes have been preferably appraised, nevertheless, 30 to 87 individuals were caught by anglers annually during in the fishing ground Jihlava 5B located between RK 45.9 and 55.6 (19 hectares) and encompassing also the river stretch under study. Despite the fact that the fishermen were informed about the tagging project, none have so far reported capturing a tagged barbel. Occasionally, the fishing ground concerned is also stocked by artificially reared one-yearold barbel in a quantity amounting individuals annually. The fish were classed according to their recapture rate and location as: 1) specimens with restricted home range (i.e. inhabiting permanently only the same single stream section where they were captured, after tagging released and eventually recaptured); they are designated as resident (also home or stationary) fish; and 2) specimens with a larger home range (extending over the two or more neighbour stream sections, i.e. those recaptured at least once beyond the boundaries of the home section where they were originally released); these specimens are designated as mobile or stray fish. The differences in frequencies of recaptures of individual size classes and stream sections were analysed by means of chi-square test (S okal & Rohlf 1981). The unequal numbers of fish tagged in individual size categories and subsequent stream sections, as well as different length and area of these sections, were respected when computing the expected frequencies. The stream section G, where only one fish was tagged, was excluded from the statistical analyses. Results Recapture rate In total 149 of 701 tagged barbel (21.26 %) were recaptured (Table 2). Of them, 117 were recaptured once, 24 twice, seven specimens three times and one fish was recaptured four times. The distribution of recaptures among subsequent stream sections fluctuated % (Table 2). Recapture frequencies plotted against consecutive recapture categories showed a distinct exponential relation (Fig. 2), which may have resulted from a higher natural mortality, eventual loss of tags. However, most likely it is caused by the fact that always only a minor part of population has been caught and the probability of recapture was thus relatively low. The expected values of successive multiple recaptures, calculated by means of the recapture coefficient C r were not significantly different from the observed values in total (χ 2 = 1.465; P<0.05) nor in all subsequent recapture categories (Table 3). The C r coefficient can be thus considered, under these environmental conditions and using the fishing methods described, valid for calculating the quantities of the single or multiple recaptures of marked fish. Resident and mobile fish and their home range Of the 149 barbel recaptured and analysed for their movements, 105 specimens (70.47 %) could be classed as resident, i.e. steadily living in a single stream section - a fairly confined 58

58 Table 2. Tagging and recapture results with barbel in the River Jihlava; proportion of resident and mobile fish. Section Number of Recaptured fish, N/% tagged fish resident 1 mobile 2 all A B C D E F G Total/Mean fish recaptured in the same of successive stream section as they were released after tagging 2 fish recaptured at least once outside of the home section where they were captured and released when tagged 3 S.D. = ±7.26 Fig. 2. Recapture rate of the barbel from the River Jihlava, Czech Republic, in successive multiple recaptures. Recapture category 1-4 explained in the legenda to Table 3. ( m) area demarcated by the two riffles (Table 2). This group of fish did not move beyond stream riffles and respected them as boundaries of their home range. The remaining 44 (29.53 %) of recaptured barbels moved in various, however rather restricted, extent from their home section where they were first caught and after tagging again released. According to our arbitrary and perhaps rather rigorous criteria, these fish were categorised as mobile part of barbel population under study. The home range of mobile barbel is thus formed by the two or even more of stream sections. Mobile barbel crossed the riffles, however, 93.2 % of them have been recaptured within a distance (1000 m from the place of their release after tagging. No difference was found between the numbers of mobile fish wandering upstream and those wandering downstream (Table 4). 59

59 Table 3. Recapture frequencies of the barbel in the River Jihlava and the difference between actual and expected values in consecutive recapture categories (1 number of fishes recaptured 1 4 times; 2 = fish recaptured 2 4x; 3 fish recaptured 3 4x; 4 fish recaptured 4 times; 5 fish recaptured 5 times); number of tagged fish = 701. Recapture Number of recaptured fish Recapture χ 2 P df category actual theoretical expected 1 coefficient 2 N a N t N e C r > > > Sum > Expected frequency N e = N t /C rn, where n = recapture category (1 4). 2 Recapture coefficient was computed as weighted mean of shares N t /N o of successive recapture categories (1 4) Table 4. The distribution of mobile barbel according to the distance of movement realised in the River Jihlava, Czech Republic. Distance from Fish recaptured downstream Fish recaptured upstream Home section (m) number % number % All An above-average number of barbel were recaptured in that stream section exhibiting a lower proportion of pools (D), whereas the lower recapture rate was found in a section with a higher pool proportion and low area of riffles (A). Resident fish were more frequently recaptured, compared to the mobile ones, in the section with low proportion of pools (D) whereas the mobile fish dominated the resident ones in the section with the above-average proportion of pools and riffles and low proportion of raceways (Table 5). Table 5. Chi-squared test for significant difference (P < 0.05; denoted by a < or > symbol) between prevailing habitat patches and the numbers of tagged and recaptured fish in subsequent stream sections (< = observed values less than expected; > = observed value greater than expected; R = resident > mobile; M = mobile > resident; NS = nonsignificant difference, P > 0.05). Stream Riffle Pool Raceway No of fish Resid/mobile section tagged recaptured fish A < > NS > < NS B NS NS NS > NS NS C < < < > NS NS D NS < NS < > R E > > < < NS M F NS NS NS < NS NS 60

60 Influence of fish size All tagged fish as well as all recaptured fish were divided into 50 mm size (SL) classes and the recaptured fish additionally subdivided into the resident and mobile parts (Table 6). Only one tagged fish was recaptured within the smallest size class ( mm), however, the proportion of recaptured fish increased in consecutive size classes within the size range mm of SL (Fig. 3) and this relation showed to be significant (Pearson correlation coefficient, r = 0.915; P<0.001; N = 10). The probability of recapture thus increases with the size of fish, which may also result from higher susceptibility of larger fish towards the D.C. electricity. With an apparent fluctuation in individual size classes, the proportion of resident barbels predominated in the size range until SL of 450 mm, whereas the proportion of mobile fish exceeded that of resident ones in the two largest, less frequented size classes ( mm SL) (Table 6). A further analysis, which followed after the size classes were collected by a 100 mm interval into four size categories only, resulted in following conclusions: 1) The number of recaptured individuals belonging to subsequent size categories was not equal in the total sample of fish (χ 2 = 10.49; P < 0.025; d.f. = 3) whereby an above-average number of mobile individuals was captured in the smallest size category (p < 0.005). 2) Neither it was equal in the subsample of resident individuals (χ 2 = ; P < 0.025; d.f. = 3) in which the above-average number was found in the size category mm (p < 0.025). 3) The unequal number of recaptures existed also in mobile fish (χ 2 = ; P < 0.005; d.f. = 3) and they were higher than the expected one in the highest size category mm (P < 0.005). The comparison of frequencies of recaptured mobile and resident fish (contingency table 2x4) showed that they were not equal (χ 2 = 15.51; P < 0.05; df = 3). Against expectations, more resident fish were caught than mobile ones in the size category mm (P < 0.025), whereas more mobile than resident individuals were recaptured in the largest size category mm (p < 0.005). Table 6. Recapture data of Barbus barbus analysed according t the size (SL) and residentiality; (size related with the date of tagging). Size No of No of recaptured fish Proportion* of class tagged all fish repeatedly recaptured resident fish mobile fish mm, SL fish n % 1x 2x 3x 4x n n Total * Proportion expressed in per cent of total number of all recaptured fish 61

61 Fig. 3. Dependence of recapture rate and mobility of barbel upon their size in the River Jihlava, Czech Republic. (Expressed as percentages of the fish recaptured against fish tagged in consecutive size classes). Discussion General nature of barbel s movements According to G e r k i n g (1953), many stream fishes live in very restricted areas during most, if not all, of their life. The problem of their local movements has to be recognised from the two viewpoints. Firstly, home range (sensu G e r k i n g 1953) is understood as the area over which the fish normally travels, searching for the optimum element of habitat (patch, microhabitat) required at given time and phase of its life cycle (foraging, spawning, nursery, resting, hiding, wintering etc.). In the gregarious barbel, home range selection and occupation is conditioned not only by the availability of suitable physical habitat, mainly feeding areas, but also by the actual status and abundance of the population and by the presence of sufficiently numerous shoals (B a r a s 1997). In our study, home range and its area were considered only indirectly, based on the location where tagged fish have been released and repeatedly recaptured and on the extent of movements exerted. Secondly, territory is any area that is defended, mostly with elements of aggressive behaviour. Generally, territorial behaviour concerns mainly reproductive territories and is markedly pronounced in monogamic species spawning in couples, in nest-building and in the egg and offspring protecting or guarding species and as such the territoriality could rather be a seasonal phenomenon. Nothing of this seems to be the case in barbel, which spawn collectively in numerous aggregates, and are neither hiding nor guarding their eggs and not exhibiting any kind of parental care for young. However, some features of territorial behaviour could be recognised in barbels spawning behaviour. The chase-away interactions between males, mainly between the courting ones, have a character of spatial competition and defence (Hancock et al. 1976, Baras 1994, P o ncin et al. 1996). Furthermore, the territoriality may be a true behaviour pattern of the barbel population under study because of its sound structure, characterised by a high reproduction potential and recruitment whilst simultaneously subjected to a low predation and fishing pressure. This 62

62 situation is expected to increase competition for territories and food resources, eventually creating aggressive behaviours and the subsequent emigration into underpopulated areas of submissive (smaller, younger) individuals, which may appear as strays (mobile fish) in our observations. A similar pattern has been reported for by-passed sections of the upper River Rhône, where reductions in deep-water habitat due to flow to a hydroelectric plant resulted in smaller and younger nase to emigrate because the available deep-water habitat was occupied by older and larger nase (P ersat & Chessel 1989). Such aspects of territorial behaviour, however, still were not satisfactorily understood in the barbel. Strong recruitment of young fish, together with behavioural dominance of old fish would substantiate the expectation that rather young fish would become mobile and search for new territories. However, rather the opposite situation seems to be true in the River Jihlava, where the mobility rate increased with increasing size (and age) of fish. Similar fact was generally stated also by Gerking (1953) and specially for the barbel found by Hancock et al. (1976). This pattern may be associated with a higher reproduction activity in larger barbel as well as their greater susceptibility to being caught by electrofishing. However, B aras (1995) found no relation between fish size and patterns of locomotory activity. The barbel s locomotory activity and extent of movements exhibit clear diel and seasonal patterns and are mainly related to the temperature and length of day. Foraging was found the main stimulus determining the localised activity in the diel scale, whereas spawning is the main motivation for movements of the seasonal scale (Lucas & Batley 1996). B a r a s (1995) found the diel and seasonal locomotory activity patterns in the barbel thermal-dependent and a temperature of 4.0 C, when barbel entered a dormancy period, as a thermal lower limit for its activity. However, water temperatures below 4 C are less common in our study area during winter months due to the influence of energy operating complex (see PeÀáz et al. 1999). Spatial dimension of movements in different rivers Bibliographic sources report much variation in the barbel s movements in various riverine habitats and often this species is considered as a wandering fish. Some of barbels tagged by Steinmann et al. (1937) were re-captured in a distance more than 300 km from their release site. Regardless of ability of some individuals to move over long distances, most barbel populations have been characterised as resident, of course often with different criteria used for the distance fixed as the boundary between the residential and mobile movement patterns. For example Steinmann et al. (1937) considered as resident all fish recaptured within a ± 5 km distance from the site where they were released. These authors found that resident were 48 % of 216 recaptured individuals in the German stretch of the Danube, 54 % of 130 recaptures in the Austrian Danube, whereas 73.7 % of 38 recaptured fish in the River Main. In the River Severn, the barbel population is also divisible into the mobile and resident components (Hunt & Jones 1974b), however, in a much larger scale than found in the River Jihlava: 86 % of 531 recaptured barbels, considered as resident, moved within a 5 km distance from the point of release in the Severn (of them 54 % did not move at all and permanently occupied the same place). The remaining 14 % wandered more widely, both upstream and downstream, up to a maximum of 34 km from the site of tagging (the mobile fish). Total range over which the barbel were recaptured was 54 km. Similarly, in the River Nidd (a rather deep stream, sectioned by weirs, NE England, tributary to Ouse) some individuals moved nearly 20 km upstream. The radiotracking carried out in this river 63

63 brought also the evidence that some tracked barbels moved regularly between the Rivers Nidd and Ouse and it was demonstrated that each of these rivers is used by them at different times of the year (Lucas & Batley 1996, Lucas & Frear 1997). A short-term (12 days) radiotracking survey conducted on eight barbel in the River Nidda (tributary to the Main, Germany) showed that the fish behaved as significantly resident moving within a distance of m only (Pelz & Kästle 1989). In large rivers, the habitats required by stream fishes are often separated by very large distances relative to a species dispersal abilities and considerable movements must be thus exhibited by the fish. Migration behaviour of the barbel could be also inferred from their seasonal occurence in fish ladders. For instance in a fish ladder built adjacent to the Stfiekov weir on the River Labe, the barbel was the most frequently (19 49 %) entering fish species during a survey conducted in and a successful passage has been confirmed (Novotn & Punãocháfi 1996). When compared to most of published data, the extent of movements and consequently the home range of barbel in the River Jihlava seem to be very limited and the resident character of behaviour is probably the most profoundly expressed of all studied populations despite the fact that we have fixed the maximum limit distance for movements of resident specimens to a relatively very low value of m. The very heterogenous and patchy character of the study stretch (and in each of the stream sections as defined in our survey) may contribute to the observed patterns by ensuring very favourable conditions for all phases of the life cycle (gonad maturation, spawning, larval and juvenile nursery, foraging, hiding, wintering etc.) within a relatively very small area compared to other locations. Especially, the richness of available spawning sites reduces the need for extended spawning migrations and this is reflected in home range fidelity as well as their behaviour patterns. Patchy habitat structure is followed by patchy population dispersion According to Gerking (1953), in streams with a well developed riffle pool structure, such as the River Jihlava, each pool (or better each stream section separated by riffles) can be considered as a more or less isolated unit containing a natural population of its own. The fish population of a small stream may thus be thought as a series of discrete, natural units rather than as a single, homogenous and freely missing group. This approach is, of course, near to the theory of metapopulation dynamics (H anski 1996, Hanski & Simberloff 1997). Consequently, the set of local, more or less isolated, barbel population of the River Jihlava, exhibiting a dynamic turnover (extinction and colonisation; Harrison 1998) could thus be considered a case of metapopulation. However, all this does not mean a fragmentation either of the physical habitat, or of the gene pool of the given population. Despite restricted extent of home range and well expressed residential behaviour, a sufficient gene exchange undoubtedly exists there as only a small genetic differentiation exists among the barbel populations of the Czech Republic ( lechtová et al. 1998). Importance of aquatic system connectivity Riffles, despite not being crossed by majority of barbel population, i.e. by resident specimens, do not fragment the stream physically, in fact they easily could be crossed by the fish. This 64

64 connectivity plays an important role in the formation and functioning of metapopulation (Schmutz & Jungwirth 1999). Barbel movements are notably restricted or impeded by the construction of dams, weirs and other water retention structures. This problem seems to be especially serious in those streams where natural habitat physical heterogeneity and riffle-pool development are in decline and the spawning migrations to remote sites are necessary. Longitudinal and/or lateral connectivity is thus crucial, as is instream and riparian cover (Copp & Bennets 1996), to the continued existence of barbel populations (Baras et al. 1994, Lucas & Batley 1996, Lusk 1996, Lucas & Frear 1997, Schmutz & Jungwirth 1999). Acknowledgements We much appreciate valuable comments to the earlier version of manuscript by Dr. Juraj H o lãík and Gordon H. Copp as well the field assistance by Mr. Jan K o cián and Mr. Josef Melkus. This research was funded through grants No 206/01/0586 by the Grant Agency of the Czech Republic and Nos AV and KSK by the Grant Agency of the Academy of Sciences of the Czech Republic. LITERATURE BARAS, E., 1994: Constraints imposed by high-densities on behavioural spawning strategies in the barbel, Barbus barbus. Folia Zool., 43 (3): BARAS, E., 1995: Seasonal activities of Barbus barbus effect of temperature on time-budgeting. J.Fish.Biol., 46 (5): BARAS, E., 1997: Environmental determinants of residence area selection by Barbus barbus in the River Ourthe. Aquatic Living Resources, 10 (4): BARAS, E., LAMBERT, H. & PHILIPPART, I.C., 1994: A comprehensive assessment of the failure of Barbus barbus spawning migrations through a fish pass in the canalized river Meuse (Belgium). Aquatic Living Resources, 7 (3): COPP, G.H. & BENNETTS, T.A., 1996: Short-term effects of removing riparian and instream cover on barbel (Barbus barbus) and other fish populations in a stretch of English chalk river. Folia Zool., 45 (3): GERKING, S.D., 1953: Evidence for the concepts of home range and territory in stream fishes. Ecology, 34: GUNNING, G.E., 1963: The concepts of home range and homing in stream fishes. Ergebnisse der Biologie, 26: HANCOCK, R.S., JONES, J.W. & SHAW, R., 1976: A preliminary report on the spawning behaviour and the nature of sexual selection in the barbel, Barbus barbus (L.). J.Fish.Biol., 9: HANSKI, I., 1996: Metapopulation ecology. In: Rhodes, O.E. Jr., Chesser, R.K. & Smith, M.H. (eds): Population dynamics in ecological space and time. University of Chicago Press, Chicago: HANSKI, I. & SIMBERLOFF, D., 1997: The metapopulation approach, its history, conceptual domain, and application to conservation. In: Hanski, I. & Gilpin, M. (eds): Metapopulation biology: ecology, genetics and evolution. Academic Press, New York: HARRISON, S., 1998: Do taxa persist as metapopulations in evolutionary time? In: McKinney, M.L. & Drake, J. A. (eds), Biodiversity dynamics: Turnover of populations, taxa, and communities. Columbia University Press, New York: HUET, M., 1949: Aperçu de la relation entre la pente et les populations piscicoles des eaux courantes. Schweitz. Z. Hydrol., 11: HUNT, P.C. & JONES, J.W., 1974: A population study of Barbus barbus (L.) in the River Severn, England: II. Movements. J.Fish.Biol., 6: LUCAS, M.C. & BATLEY, E., 1996: Seasonal movements and behaviour of adult barbel Barbus barbus, a riverine cyprinid fish: Implications for river management. Journal of Applied Ecology, 33 (6):

65 LUCAS, M.C. & FREAR, P.A., 1997: Effects of a flow-gauging weir on the migratory behaviour of adult barbel, a riverine cyprinid. J.Fish.Biol., 50: LUSK, S., 1996: Development and status of populations of Barbus barbus in the waters of the Czech Republic. Folia Zool., 45 (Suppl. 1): NOVOTN, V. & PUNâOCHÁ, P. (eds), 1996: Die Fischfauna der Elbe. International Commission zum Schutz der Elbe, Magdeburg, pp Tables. PELZ, G.R. & KÄSTLE, A., 1989: Ortsbewegungen der Barbe Barbus barbus (L.) radiotelemetrische Standortbestimmungen in der Nidda (Frankfurt/Main). Fischökologie, 1/2: PE ÁZ, M., BARU, V. & PROKE, M., 1999: Changes in the structure of fish assemblages in a river used for energy production. Regul. Rivers: Res. Mgmt., 15: PERSAT, H. & CHESSEL, D., 1989: Typologie de distributions en classes de taille: intéret dans l étude des populations de poissons et d invertébrés. Acta Oecologia, 10 (2): PONCIN, P., PHILIPPART, J.C. & RUWET, J.C., 1996: Territorial and non-territorial behaviour in the bream. J.Fish.Biol., 49 (4): SCHMUTZ, S. & JUNGWIRTH, M., 1999: Fish as indicators of large river connectivity: the Danube and its tributaries. Archiv f. Hydrobiologie, 3 (Suppl. 115): SOKAL, R.R. & ROHLF, F.J., 1981: Biometry. W.H. Freeman and Company, 859 pp. STEINMANN, P., KOCH, W. & SCHEURING, L., 1937: Die Wanderungen unserer Süsswasserfische dargestellt auf Grund von Markierungsversuchen. Z. f. Fischerei, 35: (Viz Heuschman: Handbuch d. Binnenfischerei Mitteleuropas). LECHTOVÁ, V., LECHTA, V., LUSKOVÁ, V. & LUSK, S., 1998: Genetic variability of common barbel, Barbus barbus populations in the Czech Republic. Folia Zool., 47 (Suppl. 1):

66 Folia Zool. 51(1): (2002) Morphometric study of the Iberian Aphanius (Actinopterygii, Cyprinodontiformes), with description of a new species Ignacio DOADRIO 1, José A. CARMONA 1 and Carlos FERNÁNDEZ-DELGADO 2 1 National Museum of Natural Sciences, Department of Biodiversity, c/josé Gutiérrez Abascal 2, Madrid, Spain; mcnd147@mncn.csic.es 2 Faculty of Sciences, Department of Animal Biology, University of Córdoba, Av. San Alberto Magno s/n, Córdoba, Spain Received 30 June 2000; Accepted 13 November 2001 A b s t r a c t. Morphometric variation in Aphanius iberus was analysed to demostrate the remarkable genetic divergence between Mediterranean and Atlantic population of Iberia. Four discrete morphotypes in males and three in females were distinguished. Morphometric data discriminated the Atlantic and Mediterranean populations, but revealed the Villena population as the morphologically most differentiated. Atlantic populations were described as a new species, Aphanius baeticus sp. nov., which differs from A. iberus in the overall shape, coloration pattern and number of branched rays on the dorsal and anal fins. The Villena population was retained in A. iberus because, despite of its morphological differentiation, show high genetic introgression at the nuclear genome with neighbouring populations. The range of A. baeticus sp. nov. is restricted to the eight localities of the Atlantic slope of the Iberian Peninsula. This new species should be considered Critically Endangered (CR) according to the IUCN Red List Categories. Key words: killifishes, taxonomy, Lebias Introduction The Aphanius Nardo, 1827 cyprinodontid killifishes are likely descendants of the Tethys fauna inhabiting coastal lagoons, marshes and salty rivers around the circum-mediterranean area as well as the Arabian Peninsula as far as Iran and Pakistan (V illwock 1999). As a consequence of the paleogeographic history of the region, which supported dramatical changes during the Cenozoic period, species of Aphanius show an irregular distribution pattern with isolated populations. Hence, the genus Aphanius in the eastern Mediterranean basin and adjacent regions (the Middle and the Near East) is highly diverse, with at least ten different species (H u b e r 1996). In contrast, in the western Mediterranean area, only three species are known Aphanius apodus (Gervais, 1853), A. fasciatus (Valenciennes in Humboldt & Valenciennes, 1821) and A. iberus (Valenciennes in Cuvier & Valenciennes, 1846) (Villwock 1999, Doadrio 1994). A. iberus has been reported in a wide area along the Mediterranean coast of southern France (Arnoult 1957), and Spain (Paracuellos & Nevado 1994, Fernández-Pedrosa et al. 1995), some localities of the Atlantic Iberian coast (Hernando 1975) and in North Africa in Algeria and Morocco (P ellegrin 1921, Villwock & Scholl 1982). A. iberus now appears to be extinct in France (Allardi & K e i t h 1991), and the African populations are both morphologically and genetically different from those of Iberia (V illwock & Scholl 1982) and might deserve distinct taxonomic recognition (Doadrio et al. 1996). 67

67 In the course of recent genetic studies of different populations of A. iberus using mitochondrial DNA and allozyme markers, the Atlantic populations were found to be genetically differentiated from the Mediterranean ones (D oadrio et al. 1996, Perdices et al. 2001). Similarly, restriction fragment length polymorphism analysis of mitochondrial DNA markers showed that the population of the endorrheic Villena Lagoon is highly differenciated within the Mediterranean populations (Fernández-Pedrosa et al. 1995). Since diagnostic loci and high genetic distances discriminate the Atlantic and Mediterranean populations, morphometric variation analysis is needed to characterise these two discrete gene pools. In this paper, we conduct a morphometric study of several Atlantic and Mediterranean populations to characterise the genetically differentiated populations of A. iberus. Both genetic and morphological data are used in order to revise the taxonomic status of the populations of A. iberus. Materials and Methods The specimens originated from varios locations in Spain and one location in Greece (Table 1). The samples housed in the Museo Nacional de Ciencias Naturales of Madrid (listed in the Appendix) were also revised as a comparative material. Twenty-two morphometric variables were measured. All measurements are in millimetres and were log-transformed for morphometric analysis. Only adult specimens of the 0+ and 1+ age groups were used for comparative morphometric analysis. The following abreviations were used for morphometric and meristic characters (Tables 2, 3 and 6): SL, standard length; HL, head length; HD, head depth; PrOL, preorbital length; ED, eye diameter; PrDD, predorsal distance; PrPD, prepectoral distance; PrVD, preventral distance; PrAD, preanal distance; CPL, caudal peduncle length; APL, anal peduncle length; PVL, pectoral-ventral length; VAL, ventral-anal length; DFL, dorsal fin length; DFH, dorsal fin height; PFL, pectoral fin length; VFL, ventral fin length; AFL, anal fin length; AFH, anal fin height; CFL, caudal fin length; BD, body depth; BLD, body least depth; D, dorsal fin rays; A, anal fin rays; P, branched pectoral fin rays; V, branched ventral fin rays; C, branched caudal fin rays; LLS, lateral line scale rows; Abd. vert., abdominal vertebrae; Cau. vert., caudal vertebrae; GR, gill rakers. Only branched caudal, pectoral and ventral fins rays were counted. The osteological characters were studied Table 1. Localities and sample size of Aphanius populations used for morphological, osteological and meristic analysis. Aphanius iberus morphometry osteology & meristics Locality males females males females Salinas de San Pedro del Pinatar, Murcia, Spain Albufera de Valencia, Valencia, Spain River Adra, Adra, Almeria, Spain Villena Lagoon, Villena, Alicante, Spain River Salado, Lebrija, Sevilla, Spain Iro River, Chiclana de la Frontera, Cádiz, Spain River San Pedro, Paterna de la Ribera, Cádiz, Spain Aphanius fasciatus males females males females Messolongi Marshes, Greece

68 Appendix Aphanius iberus. France. MNHN Lectotype, MNHN Syntypes. Spain. MNCN , , , , , , Albufera de Valencia, Valencia. MNCN , , 28787, Albufera de Valencia, Silla, Valencia. MNCN , , 15343, Prat de Cabanes, Cabanes, Castellón. MNCN , Albufera de Torreblanca, Torreblanca, Castellón. MNCN , , , , , , , , , Salinas de San Pedro del Pinatar, Murcia. MNCN Salinas de Marchamalo, Murcia. MNCN , Segura River, Guardamar de Segura, Alicante. MNCN , Salinas de Santa Pola, Alicante. MNCN Albatera, Alicante. MNCN , La Cava, Tortosa, Tarragona. MNCN Pantano del Hondo, Elche, Alicante. MNCN Delta del Ebro, La Tancada, Tarragona. MNCN Aiguamolls del Empordá, Ampurias, Gerona. MNCN , Albuixec, Valencia. MNCN Balsas de riego, Adra, Almería. MNCN Adra River, Adra, Almeria. Aphanius baeticus sp. nov. MNCN Sevilla. from cleared and stained specimens (W a s s e r s u g 1976). Caudal vertebres nomenclature follows Coelho et al. (1998). To test for sexual dimorphism and population variation, we used analysis of variance (ANOVA) for unbalanced design to test for differences in means (for groups or variables) because it is more statistically powerful than the simple t test, permitting the test of each variable whilst controlling for all others as well as detection of interaction effects between variables. A multivariate analysis of variance (MANOVA) was performed to test for two dependent variables at a time (i.e. between species and sexes) Differences in body shape among populations were analysed using Principal Component Analysis (PCA). Burnaby s method was used to correct for size effects (B u r n a b y 1966) using an iml procedure written in SAS-pc by Dr. L.F. Marcus (corrected from version in Appendix in R holf & Bookstein 1987) (see B ekele et al. 1993). Canonical variate analyses (CVA) and Mahalanobis distances between group centroids were used to summarise differences among populations, which were clustered in a phenogram using the Unweighted Pair Group Method (UPGMA). All the computations were performed using SAS 11, (SAS Institute Inc. 1996) and NTSYS-pc version 1.8 (Rohlf 1993). We maintain the use of Aphanius against Lebias until the International Commission on Zoological Nomenclature publishes the decision on the submitted application referring to Lazara s Lebias type species (see K ottelat 1997). Institutional acronyms: MNHN Museum National d Histoire Naturelle, Paris. MNCN Museo Nacional de Ciencias Naturales, Madrid. UCOBA Departamento de Biología Animal, Universidad de Córdoba, Córdoba. Results Significant differences (ANOVA) among populations for the morphometric variables were mainly due to differences in SL (Table 2), with sexual dimorphism statistically significant in practically all characters. These results suggest that females are larger than males and show overall smaller ventral fins. There was interaction between sexes and populations for most of the variables, which should be due to size differences between sexes for each population. Significant differences (MANOVA) were found in all characters among populations and between sexes and between populations by sexes, justifying the separate treatment of males and females. 69

69 Table 2. Two-way analysis of variance testing for sexual dimorphism, population variation and their interaction (Pop*Sex). Mean squares from SAS type III glm procedure for unbalanced designs. Except ventral fin length for sexual dimorphism and anal fin height for interaction between sexes and species, all variables were significant (P< 0.01). Variables are described in Methods. (n.s.= not significant). Variable Population Sex Pop*Sex SL HL HH PrOL ED PrDD PrPD PrVD PrAD CPL APL PVL VAL DFL DFH PFL VFL n.s AFL AFH n.s. CFL BD BLD Table 3. Eigenvectors and eigenvalues for the first four principal components for 22 variables for males and females. Variable codes are given in the Method section. Males Females Eigenvectors I II III IV I II III IV SL CFL HL HH PROL ED PRDD PRPD PRVD PRAD CPL APL PVL VAL DFL DFH PFL VFL AFL AFH BD BLD Eigenvalue (53.35) (68.24) (77.30) (82.75) (71.09) (77.74) (83.69) (87.29) 70

70 Fig. 1. A) Plots of first two principal components for 22 Burnaby corrected variables (males-left and femalesright). B) Plots from canonical variates analysed (males-left and females-right). River Salado, Lebrija, Sevilla. River Iro, Chiclana de la Frontera, Cádiz. River San Pedro, Paterna de la Ribera, Cádiz. River Adra, Adra, Almeria. Albufera de Valencia, Valencia. Salinas de San Pedro del Pinatar, Mar Menor, Murcia. Villena, Alicante, and A. fasciatus Messolongi Marshes, Greece. The first four principal components accounted for 88.5% of variance in males and 92.6% in females. The areas of the scores of each component diferred significantly (MANOVA) among populations for these first four components. Normalized eigenvectors on the first principal components showed close values for both sexes and with the same sign, suggesting an influence of body size. Variation in body size could be due to local adaptation to different habitats like marshes, irrigation chanels, lagoons or salty springs. A PCA with 22 Burnaby s corrected variables was carried out to remove size effect. The first two principal components accounted for 68.2% of variation in males and 77.7% in females (Table 3). For both sexes, the highest eigenvectors were dorsal fin length and caudal peduncle length with five different groups of males (Fig. 1). This five groups represent: Adra population, Villena population, the remaining Mediterranean populations, Atlantic populations and A. fasciatus. In females, this structure is less evident and the Adra population overlaps with Atlantic populations. Significant differences in morphology were found between the populations (Fig. 1), but differentiation was more evident in males. Mahalanobis distances D2 and its posterior cluster (Fig. 2) were congruent with previous genetic studies (F e r n á n d e z- P e d r o s a et al. 1995, Doadrio et al. 1996), although Adra population was not studied in there. 71

71 Fig. 2. UPGMA clusters on Mahalanobis distances between groups of centroids. Top Tree for males: A. baeticus sp. nov. A. iberus. Adra population. A. fasciatus. Villena population. Bottom Tree for females: A. baeticus sp. nov. and Adra population. A. iberus. Villena population. A. fasciatus. Discussion Taxonomic considerations Morphological analyses revealed the existence of four morphotypes in the Iberian populations. Eventhough morphotypes have previously been used as the base of species description, many species show a high degree of phenotypic plasticity and therefore it is necesary to corroborate the presence of a well-differentiated gene pool for taxonomical purposes. Hence, we consider the presence of diagnostic loci as a strong criterion for species recognition (B eerli et al. 1995, M endoza-qijano et al. 1998) since, in a concordant geographic framework, we can distinguish populations that may be isolated ephemerally only from those that have been isolated long enough to differentiate unlinked characters (B aum & Donoghue 1995). Therefore, the two morphotypes corresponding to the Villena and Adra populations can not be considered as new species due to the lack of diagnostic loci. The Adra population showed a different morphotype in males and currently is isolated from other Aphanius populations by at least for 350 kilometers of coastal line, nevertheless, its low genetic divergence seems to indicate a recent speciation of this populations. On the contrary, the drainage channel in Villena Lagoon connected this population with the neighbouring Mediterranean populations favouring a nuclear DNA introgression in this century. The introgression process and the persistent small population size could have propitiated the high morphological divergence of this population. 72

72 The Atlantic populations showed such genetic and morphological divergence with respect to other populations that merit to be distinguished at the species level. No available name can be applied to the Atlantic populations, and therefore, we choose to designate these populations A. baeticus sp. nov. Hence, A. iberus is maintained for the Mediterranean populations. Aphanius baeticus sp. nov. is characterized by morphological characters that refer to mensural and meristic characters (Tables 4 and 5). Moreover, using molecular markers A. baeticus sp. nov. showed diagnostic loci in different enzymatic systems (Doadrio et al. 1996), and highly differentiated haplotypes of mitochondrial DNA (P e rdices et al. 2001). An interesting finding was that males showed a higher number of differentiated morphotypes than females, indicating that sexual dimorphism varies from one species to another. Table 4. Frequency distribution of Peduncle indeces (CPL/BLD) in Aphanius from Iberia. Caudal peduncle index Species male A. iberus (Villena) A. iberus A. baeticus sp. nov females A. iberus (Villena) A. iberus A. baeticus sp. nov Table 5. Summary of diagnostic characters of Iberian Aphanius. Characters A. iberus A. baeticus sp. nov. Number of branched rays on dorsal fin 8-9(10) 8(9) Number of branched rays on anal fin 8-9 (9)10(11) Preorbital distance long short Coloration pattern in males narrow silver wide silver transversal bars transversal bars Coloration pattern in females numerous and few large black small black spotts spotts in flanks Unique alleles in allozymes IDHP-1*100 IDHP-1*73 IDHP-2*100 IDHP-2*110 smdh-2*100 smdh-2*110 The high degree of differentiation among populations of A. iberus is similar to that detected in the closely related species A. fasciatus. In the latter species, population differentiation corresponds to its naturally fragmented distribution and to a restricted gene flow among populations, which favours a genetic structure through isolation by distance (Maltagliati 1998,1999, Maltagliati & Camilli 2000). The main divergence between the Atlantic and Mediterranean populations of the Iberian Aphanius probably occurred after the Mediterranean dessication and subsequent interruption of water communication during the Lower Pliocene (D oadrio et al. 1996). The same pattern of speciation, based on a paleogeographical event, was proposed for the Anatolian species Aphanius asquamatus (Sözer, 1942) which probably diverged from the central 73

73 Anatolian cyprinodonts during the Upper Miocene-Lower Pliocene (see V illwock 1999). During this period, Anatolia was raised several meters above sea level whilst the entire region was separated in two large systems. Afterwards, the central Anatolian cyprinodonts gradually split during the Pliocene and Pleistocene periods, rendering distinguishable species, such as A. chantrei Gaillard, 1895 and A. anatoliae Leidenfrost, 1912, besides many subspecies and micro populations (V illwock 1963). Although Aphanius are euryhaline and a coastal route of colonization should be used, the strait of Gibraltar appears to act as an important barrier. Thus, for instance, the exotic species Fundulus heteroclitus, which quickly colonized the southern Atlantic area of the Iberian Peninsula, has not been able to colonize the Mediterranean basin through the strait of Gibraltar. Natural geographic barriers and life-history traits, which are similar to those observed in A. fasciatus e. g. benthic eggs and absence of larval stages (Maltagliati & C a m i l l i 2000), should prevent gene flow between Atlantic and Mediterranean populations. Aphanius iberus is critically endangered (Blanco & González 1992, Maitland & Crivelli 1996, Council Directive 92/43/ECC). The distribution range of A. baeticus sp. nov. is relatively narrow and pollution, exotic species and hydraulic alterations render the species critically endangered too. Description of new species Aphanius baeticus sp. nov. Diagnosis. Differs from all other known species of Aphanius in the combination of: ten branched rays in the anal fin, eight branched rays in the dorsal fin, long and low caudal peduncle, short preorbital length, males show light wide transversal bars and females large black spots. Presence of unique alleles at IDHP-1*, IDHP-2* and smdh-2* (Doadrio et al. 1996). Holotype. (Fig. 1). MNCN Male, 31.3 mm SL. Salado River, Lebrija, Sevilla. Guadalquivir basin, Spain. 20 m altitude. Leg. C. Fernández-Delgado. 6. VI Paratypes. MNCN , 29 individuals. Salado River, Lebrija, Sevilla. Guadalquivir basin, Spain. Leg. C. Fernández-Delgado. 6.VI MNCN , 45 individuals. Salado River, Lebrija, Sevilla. Guadalquivir basin, Spain. Leg. C. Fernández- Delgado. 17.VII MNCN , 34 individuals. Salado River, Lebrija, Sevilla. Guadalquivir basin, Spain. Leg. C. García Utrilla. 25.I UCOBA , 55 individuals. Salado River, Lebrija, Lebrija, Sevilla, Spain. Guadalquivir basin. Leg. Fernández-Delgado. MNCN , 65 individuals. Hondón Lagoon, Parque Nacional de Doñana, Huelva, Spain. Leg. L. Dominguez. 5.XI MNCN , 12 individuals. Salinas de San Carlos, Sanlúcar de Barrameda, Cádiz, Spain. Leg. J.A. Hernando. 15.VI MNCN , 28 individuals. Salado de San Pedro River, Paterna de la Ribera, Cádiz, Spain. Leg. C. Fernández-Delgado. UCOBA 1-96, 96 individuals. Salado de San Pedro River, Paterna de la Ribera, Cádiz, Spain. Leg. C. Fernández-Delgado. UCOBA , 30 individuals. Roche River, Roche, Cádiz, Spain. Leg. C. Fernández-Delgado. MNCN , 35 individuals. Iro River, Chiclana de la Frontera, Cádiz, Spain. Leg. Fernández-Delgado. 21.IX UCOBA , 50 individuals. Iro River, Chiclana de la Frontera, Cádiz, Spain. Leg. C. Fernández-Delgado. Description. D I-II 8(9), A I-(9)10(11), P (8)9-10(11), V (3)4-5, C (13) LLS (24)25(26), Abd. vert. (10)11, Cau. vert. (14,15)16(17), G. R. 8-10(11). The body form of 74

74 Fig. 3. Aphanius baeticus sp. nov. Top: Holotype, male, MNCN Bottom. Paratype, female, MNCN Salado River, Lebrija, Sevilla. Spain. Scale 1 cm. Drawing by M. Merino. Aphanius baeticus sp. nov. (Fig. 3, Table 6) is characterised by a deep but more elongated body than in A. iberus. Minimal body depth is (1.7) times in the caudal peduncle in females and (1.8) in males. Minimum body depth (1.2) times in the anal peduncle in females and (1.2) in males. Short head comprising (3.5) times the standard length in females and (3.4) in males. Preorbital length is small, reaching (1.7) times the eye diameter in females and (1.8) in males. Insertion of the ventral fin is anterior to dorsal fin origin. Preventral length is (1.1) times the predorsal length in females and (1.2) in males. Fin size moderately large. Robust premaxille (maximun length / maximum high = (2.3)) with (7)8-10(11) teeth. Dentary with (7)8-11(12,13) teeth. Pigmentation pattern. Males. Light wide vertical bars along the body side, being more evident in juveniles (Fig. 4). The number of bars on the flanks is typically and 4-5 on the caudal fin. Four rows of dark spots extend onto the dorsal and anal fins. Pectoral and ventral fins are ligth orange. Females. Light brown with few and large dark spots on the body flanks. Two rows of spots are shown, one of them along the lateral line and another in the ventral side. Frequently, the spots are arranged as bars on the middle of the body. An additional row of spots is shown on the caudal peduncle. Caudal fin has 4-6 dark vertical bars. Sexual dimorphism. Males are smaller overall than females with proportionally larger ventral fins. Distribution. A. baeticus sp. nov. occurs in the lower reaches of the River Guadalquivir and streams located on the southern Atlantic slope of Spain (Fig. 5). The species is currently found in the following areas: River Santiago, El Palmar de Troya, Sevilla. River Salado, Lebrija, Sevilla. Parque Nacional de Doñana, Huelva. Salinas de Sanlúcar de Barrameda, Sanlúcar de Barrameda, Cádiz. River Salado de San Pedro, Paterna de la Ribera, Cádiz. 75