Recruitment and population structure of the non-indigenous brackish-water mytilid Xenostrobus securis ( Lamark , 1819 ) in the Kino River , Japan

The recruitment and population structure of the non-indigenous brackish water mytilid Xenostrobus securis (Lamark, 1819), native to Australasia, were studied in the lower reaches of the Kino River, central Japan. Post-larvae > 2 mm in shell length were found exclusively among barnacle stands of Fistulobalanus kondakovi (Tarasov and Zevina, 1957) and algal fronds of Gloiopeltis tenax (Turner) Decaisone, 1842 from January to May and from July to October. Density of the post-larvae was much greater in the latter period than in the former. Small mussels with shell lengths of 2–4 mm settled to the conspecific mussel beds in April and September. Primary and secondary settlements onto different substrates were observed in the early stages of the life cycle. Two or three cohorts were identified in conspecific mussel beds, and they grew at a rate of ca. 1 mm per month. Density in the beds fluctuated monthly and was greater in the lower than in the middle intertidal zone, suggesting greater spatial heterogeneity in density due to predation in the former.


Introduction
Biological invasions of mytilid and dreissenid bivalves have had serious impacts on native aquatic ecosystems and human activities in the world (Nalepa and Schloesser 1993;Minchin et al. 1998;Crooks 2009).Several biological and population characteristics contributing to the success of the invasion and population expansion by the bivalves have been suggested: high fecundity, genetic variability, great tolerance to environmental fluctuations, short life span, rapid growth, early maturity, and gregarious behaviour (Morton 1996;Sakai et al. 2001).Many population studies have been conducted to determine the characteristics that are significant to such successful invasions (e.g., Morton 1975;Seed and Suchanek 1992;Crooks 1996;Strayer and Malcom 2006).
The axe-head mussel Xenostrobus securis (Lamarck, 1819) is a brackish water mytilid native to Australasia (Wilson 1968).In Japan, this species was discovered first in 1972 in Kojima Bay, Okayama Prefecture (Kimura et al. 1999), and it has since expanded its distribution on the temperate coasts of the Seto Inland Sea, Pacific Ocean, East China Sea, and Japan Sea (Iwasaki et al. 2004a;Iwasaki 2006) at an average rate of 23.9 km year -1 (Iwasaki et al. 2004b).X. securis has a negative impact on native brackish water ecosystems, covering and smothering native barnacles and causing serious changes in the zonation patterns of native sessile organisms (Iwasaki, unpubl. data).It has also been introduced to the southern coasts of the Republic of Korea (Shirafuji and Sato 2003), Adriatic and Western coasts of Italy (Lazzari and Rinaldi 1994;Barbieri et al. 2011), Atlantic coasts of Spain (Garci et al. 2007;Pascual et al. 2010;Adarraga and Martinez 2012), and the Mediterranean lagoons of France (Gofas and Zenetos 2003).Multiple divergent genetic lineages have been observed both within the native and invasive populations (Colgan and da Costa 2013).Wilson (1968) demonstrated that X. securis grows rapidly, matures early, and has great tolerance to variations in salinity for the populations in the Swan River Estuary, Western Australia.In Japan, the life cycle, early life history, and population characteristics of this species have been reported for lentic populations in Lake Hamana, central Japan (Abdel-Razek et al. 1993a, b;Kimura andSekiguchi 2009, 2012), and Dokai Bay, western Japan (Kohama et al. 2001).No studies have, however, reported on lotic populations in river estuaries or on the life cycle and population structure of this species in other countries.To evaluate the invasive success of this species, population characteristics variability in a wide range of habitats must be examined.
In this study, we described the recruitment of post-larval mussels, population structure, growth, and life cycle of the population of X. securis in the Kino River estuary, central Japan.These results were compared with those for lentic populations in Japan and lotic populations in native regions.Causes of the spatiotemporal variation are discussed in terms of invasion success in a wide range of habitats for this species.

Study site
The Kino River is 136 km long and runs westward through Nara and Wakayama Prefectures, central Japan, into Kii Channel, which is connected with the Pacific Ocean (Figure 1).It covers 1750 km 2 and has a mean discharge of 37.4 m 3 s −1 at its lower reach.
Field sampling was conducted on the vertical concrete wall at Arimoto, Wakayama City, Wakayama Prefecture, which is ca. 5 km upstream from the river mouth (34°14'N, 135°11'E, Figure 1).The Kinokawa Estuary Barrage is located 200 m upstream from the study site and intercepts the inflowing tide.The study site was thus located at the upper limit of the brackish water zone in the river.The maximum tidal range at the site was 1.8 m.

Field sampling
Preliminary surveys were conducted during the summer and autumn of 2005.The results suggested that X. securis smaller than 2 mm (hereafter referred to as 'post-larvae') were much more abundant among barnacle stands and algal fronds than in mussel beds on the concrete wall.Accordingly, monthly sampling for barnacles, algae, and small mussels (including post-larvae) was performed from May 2006 to May 2007 during low tides.The samples were taken from four permanent quadrats, each of 100 cm 2 , in the middle part of the intertidal zone at ca. 10 cm below mean tide level (hereafter referred to as the 'upper level') and the lower part of the intertidal zone at ca. 50 cm below mean tide level (hereafter referred to as the 'lower level').These quadrats were set in the bare substrate formed after sampling of mussel beds in April 2006.The permanent quadrats were photographed, and the barnacles and algae were then collected monthly and preserved in 5% formalin.
In the laboratory, post-larvae among the barnacles and algae were collected, identified, and counted under a binocular microscope at a magnification of 40×.When more than 100 postlarvae had been collected in each monthly survey, 100 individuals were chosen randomly for measurement under a binocular microscope at a magnification of 40× with an electronic micrometer to the nearest 0.01 mm.The post-larvae were identified according to the criteria outlined in Kimura and Sekiguchi (1996).The percentage coverage by sessile organisms in the permanent quadrats was measured by counting 169 equidistant dots on the photographs on a personal computer.
Monthly sampling for X. securis in mussel beds was conducted from April 2006 to May 2007 on a vertical concrete wall.Four samples, each of which was bounded by a 100-cm 2 quadrat and 100% covered with mussels, were taken for both upper and lower levels.All sediments, shell debris, and byssal threads of the mussels within the quadrats were collected and preserved in 5% formalin.
In the laboratory, mussels were separated from sediment particles and byssal threads and counted.Mussel shell lengths were measured with Vernier calipers to the nearest 0.05 mm.Post-larvae were examined under a binocular microscope at a magnification of 40×, and measured with an electronic micrometer to the nearest 0.01 mm.
Water temperature and salinity were measured in the surface water close to the shoreline at the study site during low tide, every month from April 2006 to May 2007, using a digital thermometer (AD-5625, A&D Ltd., Tokyo, Japan) and a refracting salinity meter (S-milE, ATAGO Ltd., Tokyo, Japan), respectively.

Data analyses
The population of X. securis at the study site was composed of several cohorts.Each cohort was identified monthly and recorded in the form of length-frequency histograms.The mean shell length, standard deviation, and number of individuals were estimated using Bhattacharya's method of modal progression analysis and FiSAT II (Gayanilo et al. 2005), a computer program (ver.1.2.2) provided by the Food and Agriculture Organization (http://www.fao.org/fishery/topic/16072/en,Accessed on 18 December 2013).Difference in mean density of mussels between the higher and lower levels was analyzed using t-test after confirmation of homogeneity of variance.

Water temperature and salinity
Water temperature at the study site changed markedly, increasing from April to August 2006, decreasing to January 2007 and again increasing to May 2007 (Figure 2).The maximum water temperature was 30°C in August 2006 and the minimum was 11°C in January 2007.
Salinity changed irregularly within the range from 2 to 19.In April, July, and September 2006 and April 2007, salinity decreased below 5 due to heavy rainfall 1 or 2 days prior to measurement.Without the rainfall, salinity ranged from ca. 10 to 20 during the study period.

Organisms at the study site
Throughout the study period, X. securis covered most of the concrete wall from 20 cm above mean tide level (MTL) down to the subtidal zone.Several sessile organisms such as the indigenous barnacle Fistulobalanus kondakovi (Tarasov and Zevina, 1957), the non-indigenous barnacles Amphibalanus amphitrite (Darwin, 1854) and A. eburneus (Gould, 1841), the indigenous red alga  Gloiopeltis tenax (Turner) Decaisne, 1842 and the green algae Ulva spp.were scattered in patches on the concrete wall.During the field surveys, the carp species Cyprinus carpio (L.1758) was frequently observed foraging on the mussels in the lower level during flood tides.

Post-larvae in permanent quadrats
The barnacle F. kondakovi, the red alga G. tenax, and the green algae Ulva spp.settled in large numbers on the bare substrate of the permanent quadrats during summer, autumn, and winter, respectively, forming barnacle stands or algal mats covering ca.55%-75% of the quadrats (Figure 3).Small specimens of X. securis (including post-larvae) were abundant in the permanent quadrats, which were covered with barnacles and G. tenax from July to October 2006.Mean density of the small mussels at the lower level was significantly higher than that at the upper level in August (t = 4.89; df = 4; P < 0.01), September (t = 3.04; df = 4; P < 0.05) and October (t = 4.85; df = 4; P < 0.01).From February to May 2007, small mussels were collected among algal fronds of Ulva spp. at the lower level, but the mean densities were much lower than those from July to October 2006.
Most of the small mussels collected in the permanent quadrats were post-larvae throughout the year (Figure 4).They were not collected in the monthly samples of mussel beds (see Figures 5 and 6).Polymodal size frequency distributions were observed from July to October 2006 for both upper and lower levels and no marked increase in mode was detected during this period.From February to May 2007, size-frequency distributions showed no distinct modes.

Size distribution for mussel beds
In April 2006, three cohorts were detected in mussel beds at the upper level (Cohorts Z, A and B in Figure 5).Their mean shell lengths were 18.2 mm for the large-sized cohort (Cohort Z), 11.6 mm for the middle-sized cohort (Cohort A) and 4.5 mm for the small-sized cohort (Cohort B).All of these cohorts increased  Throughout the study period, very few postlarvae were collected in the mussel beds where a large amount of mud accumulated throughout the year.Most byssal threads, which seem to be the sites most suitable for primary settlement of post-larvae, were buried within the mud.

Growth
Mean shell lengths of Cohort Z at the upper level were ca.20 mm from April to August 2006, showing no monthly change (Figure 7A).Mean shell lengths of the Cohorts A, B and C increased monotonically by ca. 1 mm per month (Figure 7A).Increases in size for three cohorts at the lower level, Cohorts α, β and γ, were similar to those for Cohorts A, B and C at the upper level (Figure 7B).However, the increase in size for Cohorts β and γ slowed in winter

Monthly change in abundance
Mean densities (N 100 cm −2 ) of Cohorts Z, A and α decreased from spring to autumn of 2006 (Figure 8).Mean densities of Cohorts B and β increased in the spring of 2006 and then tended

Periods of spawning and recruitment
Table 1 shows the periods during which spawning females, larvae, and post-larvae of X. securis were present in this and all other available studies.Recruitment of small mussels (2-5 mm) occurred at seven localities in Australia and Japan.In the Swan River Estuary of Western Australia, periods of spawning and spatfall (recruitment of spat with shell lengths of 2-4 mm) were variable according to the locality (Wilson 1969).Recruitment periods   and 6.Table 1.Periods during which spawning females, planktonic larvae and post-larvae smaller than 2 mm were present and recruitment of small mussels with 2-5 mm occurred into conspecific mussel beds, the number of cohorts consisting of mussel beds and the maximum life span of the cohorts.

Locality
Period of occurrence were much earlier in the downstream (Crawley) than the upstream localities (Barker Bridge) in the Austral spring and summer (water temperature 20°C-30°C; Wilson 1968Wilson , 1969)).For the upstream population (Barker Bridge) at least, the periods differed from one season to another.Wilson (1969) stated that variability in population characteristics (e.g., the periods of spawning and post-larval recruitment, growth, size structure, and longevity) may be interpreted largely in terms of the species' physiological limitations and responses to variable salinity conditions.
Temperature appears to have only a secondary effect on the biology of this species.
In Lake Hamana, in the middle of Japan, spawning females were found from January to February (water temperature 2°C-8°C) in low numbers and from June to November (20°C-30°C) in large numbers (Abdel-razek et al. 1993a, b).Planktonic larvae were found from March to April (10°C-15°C) and from July to January in the lake (10°C-33°C) (Kimura and Sekiguchi 2012).Post-larvae were abundantly present in summer and autumn (July to October, 26°C-33°C) in the sediment of Lake Hamana (Kimura and Sekiguchi 2012), on the ropes suspended in the water column in Dokai Bay (Kohama et al. 2001), and among the barnacle stands or foliose algae in the Kino River (present study).In Lake Hamana and Kino River, however, a small number of postlarvae were found even during winter and spring (December to May, 5-20°C) (Kimura and Sekiguchi 2012;present study).In all four Japanese localities, recruitment of small mussels 2-5 mm long onto conspecific beds occurred abundantly during summer and autumn (from July to October, 26°C-33°C).Results of the present study showed, however, that recruitment of small mussels occurred even in April (17°C).
The results in Japanese waters indicate variability in periods of breeding and recruitment of post-larvae or small mussels though the main seasons for breeding and recruitment were summer and autumn.As suggested by Wilson (1969), water temperature may not be a dominant factor in the occurrences of spawning females, larvae, and post-larvae in Japanese waters.Instead, stochastic success in recruitment of larvae or survival of post-larvae due to the instability of brackish environments, in which salinity fluctuates markedly, may have caused the regional differences shown in Table 1 (Wilson 1968;Kimura and Sekuguchi 2012).Tolerances to variations in salinity and temperature by X. securis have been examined only for adult mussels >20 mm (Wilson 1968;Kimura et al. 1995).

Primary and secondary settlements
Some species of mytilid bivalves have a period of post-larval dispersal by planktonic drift.Larvae of the mussels preferentially settle onto filamentous substrata such as bryozoans, hydroids, and algae (primary settlement, Bayne 1964;Seed 1969;Petersen 1984a, b;Eyster and Pechenik 1987).After metamorphosis and growth on such substrates, post-larvae move onto the byssal threads or shells of conspecific adults (secondary settlement, Bayne 1964) by drifting with single, long, byssal threads (Sigurdsson et al. 1976;Lane et al. 1985).The differences in primary and secondary settlement sites and post-larval dispersal from the former to the latter site are considered to be adaptive because larval mortality due to filter-feeding by large mussels is reduced at the former site (Seed and Suchanek 1992).
At the site of the present study, the post-larvae of X. securis occurred abundantly among barnacle shells and algal fronds.In contrast, few were collected within conspecific mussel beds.Kimura andSekiguchi (2009, 2012) also observed that small mussels >2 mm long were abundant within adult beds on artificial rocks, but scarce in the sediments at the bottom of Lake Hamana.As suggested by Kimura and Sekiguchi (2012), these results suggest strongly that small mussels > 2 mm in length opted for secondary settlement onto conspecific beds, i.e. larvae of X. securis settled primarily onto substrates other than conspecific beds, metamorphosed there, and then moved into X.securis beds after growing to a size >2 mm.The avoidance of the larvae to settle onto the conspecific beds seems to be due to the burying of byssal threads by the mud.
Results of this study suggest a precautionary measure to control this invasive species.As small post-larvae invisible to the naked eye may attach to barnacle stands or algal fronds, transportation of these sessile organisms between sites may promote range expansion of X. securis.Accordingly, removal of sessile organisms that may harbor small X. securis from materials conveyed to uncolonised water bodies is essential to prevent the range expansion.

Population structure, growth and life span
The number of cohorts found within mussel beds and the maximum life span of the cohorts at the seven localities in Australia and Japan ranged from one to three in the Swan River Estuary, Lake Hamana, and Dokai Bay (Table 1).The present study showed that mussel beds were composed of two to four cohorts at both upper and lower levels.As suggested by Wilson (1969), the spatiotemporal variation in the number of cohorts seems to have been due to stochastic success in the breeding of adult mussels, recruitment of larvae, or survival of post-larvae under the unstable brackish water conditions at each locality.Abdel-Razek et al. (1993b) reported that it takes 60-70 days for newly-hatched larvae to grow to post-larvae length of 5 mm at 20°C.Considering the duration of this period of larval and post-larval growth, the mean shell lengths, the timing of recruitment into mussel beds, and growth rates; cohorts A and α in the present study must have been born in the summer of 2005, Cohorts B and β in the late winter of 2006, Cohorts C and γ in the summer of 2006, and Cohort D and Cohort δ in late winter of 2007.As Cohorts A and α decreased in density from April to October and disappeared in November 2006, the maximum life span of one cohort may have been 15 or 16 months at the present study site.A decrease in density of large mussels from spring to summer and disappearance of the largest cohort in autumn were shown in Lake Hamana by Abdel-Razek et al. (1993b) and Kimura and Sekiguchi (2009) and in Dokai Bay by Kohama et al. (2001).A similar maximum life span to that at the present study site (15 or 16 months) was suggested for Dokai Bay by Kohama et al. (2001), whereas longer life spans (21 or 24 months) were reported for the Swan River (Wilson 1969) and Lake Hamana (Kimura and Sekiguchi 2009) (Table 1).Taken together, these results suggest that the maximum life span of this species is 24 months.
Each cohort in the present study tended to grow faster during spring to autumn than in winter as was also found in Lake Hamana (Abdel-Razek et al. 1993;Kimura and Sekiguchi 2009).The growth rate tended to be greater at the lower than at the upper levels.Similar differences in growth rate between upper and lower intertidal zones have been reported for several species of marine mussels (Griffiths and Hockey 1987;Yamada 1989).These differences have been attributed to shorter exposure to air, greater food availability, and milder environmental conditions in the lower intertidal than in the upper zones (Seed and Suchanek 1992).
Mean densities for Cohorts β, γ and δ at the lower level of the present study site fluctuated much more than those for Cohorts B, C and D at the upper level.Predation may have been responsible for the greater variation in mussel density in the lower level because large fish such as the carp C. carpio were observed foraging on the mussels in the lower level.
In conclusion, the pattern and period of recruitment by post-larvae or small mussels and the population structure for X. securis varied according to locality.This variability may have been attributable to the physiological responses and limitations of this species under the unstable environmental conditions of a brackish water system.Experimental studies of salinity tolerance by small mussels and post-larval dispersal to the site of secondary settlement should be conducted in order to elucidate the cause for the spatiotemporal variations reported herein.

Figure 2 .
Figure 2. Monthly changes in water temperature (blue, filled circles) and salinity (red, filled triangles) at the study site.

Figure 3 .
Figure 3. Monthly changes in mean percentage cover of sessile organisms in four permanent quadrats (A: upper level, C: lower level) and mean density (± SD, vertical bar) of Xenostrobus securis (100 cm −2 ) collected among barnacle stands or algal fronds in four permanent quadrats (B: upper level, D: lower level).
in size from May to July 2006.Numbers in Cohorts Z and A decreased towards August 2006; cohort Z disappeared by September 2006, and cohort A by November 2006.A new cohort ca. 2 mm long on average (Cohort C) appeared in August 2006, increasing in number and size during the autumn and winter of that year.From November 2006 to March 2007, mussels at the upper level consisted of two cohorts (Cohorts B and C), the mean shell lengths of which increased during the period.In April 2007, a new cohort ca. 4 mm long on average (Cohort D) appeared, making the mussel beds at the upper level consist of three cohorts again.Their mean shell lengths in April 2007 were similar

Figure 4 .
Figure 4. Size frequency distributions of X. securis collected among barnacles or algal fronds in four permanent quadrats for each upper (red bars) and lower (blue bars) level.

Figure 5 .
Figure 5. Size-frequency histograms of X. securis collected from four quadrats (each 100 cm 2 ) at the upper level for 14 months.Locations of letters Z, A, B, C and D indicate the mean shell lengths of five different cohorts identified by modal progression analysis using FiSAT II (Gayanilo et al. 2005).
(from December 2006 to March 2007).Considering the similarities in size and growth pattern, Cohorts A and α must have settled into this population in the summer of 2005, and Cohorts B and β must have done so in the winter of 2006.

Figure 6 .
Figure 6.Size-frequency histograms of X. securis collected from four quadrats (each 100 cm 2 ) at the lower level for 14 months.Locations of letters α, β, γ and δ indicate the mean shell lengths of four different cohorts identified and calculated by a modal progression analysis using FiSAT II (Gayanilo et al. 2005).

Figure 7 .
Figure 7. Monthly changes in mean shell length (± SD, vertical bar) of each cohort identified and calculated by a modal progression analysis using FiSAT II (Gayanilo et al. 2005).Letters Z, A, B, C, D, α, β, γ and δ indicate the different cohorts shown in Figures 5 and 6.

Figure 8 .
Figure 8. Monthly changes in densities (400 cm −2 ) of several cohorts at the upper (A) and lower (B) levels.Letters Z, A, B, C, D, α, β, γ and δ indicate the different cohorts shown in Figures 5 and 6.