Population sampling of the golden mussel , Limnoperna fortunei ( Dunker , 1857 ) , based on artificial ceramic substrate

The ceramic substrate (21cm in length, 6cm in width and 1.3cm in depth) was tested for the Limnoperna fortunei population, sampling at two localities in the Jacuí delta (Jacuí Canal (Canal do Jacuí – CJ) and Port Docks (Cais do Porto – PO)) in Rio Grande do Sul state, Brazil. The individuals were quantified through the superimposition of a squared (1cm2) and segmented (sI, sII e sIII) sheet on the substrate. Using Kruskal-Wallis and Mann-Whitnney, the recruit and adult average densities were compared in each segment, surface (smooth and pipe-shaped) and sampling locality (α = 0.05). In CJ, the extreme and intermediate (adult) segments differed statistically (p < 0.0001), as well as the ceramic substrate surfaces (recruits) (p = 0.04). The recruit and adult densities between the CJ and PO localities also differed between themselves (p < 0.0001). The method was efficient for the invasive population sampling.


Introduction
The golden mussel, an invasive bivalve of Asian origins, was brought to the continental waters of South America via ballast water (Pastorino et al., 1993).In Brazil, it was first recorded in the Guaíba Lake, in Porto Alegre, Rio Grande do Sul state (Mansur et al., 1999).The environmental (landscape, aquatic flora and fauna) and economic (constructed systems) damages, resulting from the proliferation of the invasive species, were widely recorded (Darrigan and Mansur, 2006;Mansur et al., 2003;2004).
The golden mussel (Mytiloida) produces byssus threads, which allow the formation of compact macroclusters on hard substrates.This feature differentiates them from other native bivalves of the south of Brazil (unionids and venerids) which live burrowed in sandy substrates, with the exception of the byssus-bearing species Eupera Bourguinat, 1854, which do not create macroclusters (Mansur and Pereira, 2006).
Due to the variability of substrates to which the mussel adheres, the comparison of samplings carried out through traditional gathering methods (dredges and square frames) becomes very difficult.As for the artificial substrates, they standardize sampling, reduce variability and sample processing time (Gibbons et al., 1993).Artificial substrates were widely used for the monitoring of the invasive species Dreissena polymorpha (Pallas, 1771), the zebra mussel, in limnic ecosystems in North America (Claudi and Mackie, 1994).
Various types of artificial substrates have been used in the monitoring of L. fortunei populations, such as three-dimensional asbestos structures (Morton, 1977), PVC adaptations (Boltovskoy and Cataldo, 1999), wood (Fontes et al., 2002;Mansur et al., 2009) and PVC (Darrigran et al., 2007) both in the shape of an "X" and PET bottles (Faria et al., 2006).Mansur et al. (2003) used ceramic substrates (six-hole commercial bricks), for the first time, to sample L. fortunei populations in the Guaíba Lake, in Porto Alegre.Santos (2004) and Santos et al. (2008) continued the monitoring of the Lake with ceramic substrate, identifying and quantifying different life stages of the invasive species, on the substrate.Later, the use of ceramic substrate (commercial brick) was adopted by FURNAS Central Hydroelectric Plants and by FEPAM (Fundação Estadual de Proteção Ambiental -State Foundation of Environment Protection) (Terra et al., 2007).
Monitoring golden mussel population densities is one of the main goals of the invasive species handling projects, but monitoring efficacy requires adequate methods of sampling and subsampling.
Aiming to test a sampling and subsampling method which would result in significant information on the life cycle stages of recruits and adults of L. fortunei, experiments were carried out in the Jacuí River Delta (Municipality of Porto Alegre, Rio Grande do Sul state), using ceramic artificial substrate.

Material and Methods
The study area (Figure 1) is located in the Jacuí River delta, in the Rio Grande do Sul State Central Depression physiographic unit, in the Porto Alegre metropolitan region.The Jacuí River debouches into the Guaíba Lake through an inner delta created by a series of distributaries, among which the Jacuí Canal stands out (average depth: 8m).In the Port Docks, near the Porto Alegre urban area, the peripheral landscape has been significantly altered with land filling.Steep shores (concrete walls) and deep waters (5m) are presently found in this locality.
The sampling localities are in the Jacuí River delta, on the right shore of the Jacuí Canal (CJ), on Pintada Island and in the Port Docks (PO), downtown Porto Alegre.After analyzing the historical data of larval densities in Guaíba Lake (Santos, 2004) and in the lower Paraná River (Boltovskoy and Cataldo, 1999), the period from December 2005 to March 2006 was determined for the execution of the artificial substrate exposure experiments, the period in which the highest larval production and recruitment occurs.The period of exposure (three months) is the minimum necessary for colonization (recruits) and development of specimens (adults).
In the same period, monthly samplings of larvae (30L) were made in both sampling localities with a Population sampling of the golden mussel plankton net (30μm), in accordance with Santos et al. (2005).The monitoring of larval densities was used as a potential recruitment indicator.
The chosen substrate consisted of a commercial refractory brick, 21cm in length, 6cm in width, and 1.3cm in thickness (Figure 2).One side of the substrate presented a smooth surface (Figure 2a) and the other side was pipe-shaped (2b).The bricks were perforated at 2cm from their extremities (Figure 2).They were attached to the frames through this hole, with plastic rings.
In December (2005), nine replicas of ceramic substrate were exposed in each sampling locality, attached to metal structures (iron frame) through plastic rings.In CJ, the frames were suspended with ropes on the lower part of a wharf.Therefore, both surfaces remained equally exposed to the water column.In PO, the iron frames were suspended with ropes near the dock's concrete wall.The smooth surface of the substrate was positioned in front of the concrete wall, while the pipeshaped surface remained exposed to the water mass.In   March, all substrates were collected from the water and transported to the laboratory.
The larva samples were fixed with alcohol 96°GL and transported to the laboratory.Monthly water samplings were made in each of the sampling localities in order to analyze physical, chemical (water temperature, pH, turbidity, oxygen chemical demand, dissolved oxygen, total phosphorus and nitrate) and microbiological (fecal coliforms) indicators, in accordance with the described APHA ( 2004) methods.
The larvae were quantified (ind.m -3 ) in accordance with Santos et al. (2005).The bricks were dehumidified in a Pasteur oven (60°C) and then examined under a stereoscopic microscope for quantification.
According to Santos et al. (2008), recruits and adults presented, respectively, <0.5 and ≥ 0.5mm in length.The quantification took place in these two stages, through a stereoscopic microscope with a millimetric lens and transparent grid (Figure 3) superimposed on both substrate surfaces.In each segment of the grid, five squares (Q ij ) of 1cm 2 were randomly picked, on which the recruit and adult individuals were quantified.Through equation 1 (Ds n = (ΣQ ij /5)) the individual densities were calculated (ind.cm - ) for the three segments (Ds n; n = I, II or III ) of each surface.When no individuals were detected in the squares of a segment, the total quantification of individuals was made in the segment, dividing the resulting quantification value by the segment area (42cm 2 ).Then, surface densities were obtained (Dl n; n = smooth, I or pipe-shaped, II ) through equation 2 (Dl n = μDs n ).The average density (D, ind.cm -2 ) of individuals on the ceramic substrate was obtained through the sum of both surface densities, using equation 3 (D = ΣDl n ).
The average density in the segments was considered as a replica.In order to verify significant differences between average densities in different segments (Ds I , Ds II e Ds III ) of the brick, the non-parametric Kruskal-Wallis test was used, followed by multiple comparisons using the Dunn method (α = 0.05).Through the Mann-Whitnney test (α = 0.05), the average densities (ind.cm - ) of substrate surfaces (H 0 : Dl 1 = Dl 2 ), sampling localities (H 0 : D (CJ) = D (PO)) and population components (H 0 : D (recruits) = D (adults)) were compared.Substrate segments, as well as surfaces, were tested using only CJ data, where surfaces remained equally exposed to the water column.Non-parametric tests were chosen after verifying that the data was not normal, through the Kolmogorov-Smirnov test for one sample (H 0 : data follows normal distribution), with a 95% significance level, according to Zar (1999).The water quality indicators were compared between localities using the Mann-Whitnney test (α = 0.05).The Kruskal-Wallis test was obtained through the BIOESTAT 3.0 software, while the Mann-Whitnney and the Kolmogorov-Smirnov tests were obtained using the SPSS 11.5 software.Population sampling of the golden mussel

Larval density
The Jacui Canal presented the highest larval densities (Figure 4), with maximum values in the months of December and January.Significant differences were verified (p = 0.08) among CJ and PO, considering α < 0.1.In the period of the experiment, the average density (ind.m -3 ), the standard error, and the minimum and maximum larval densities in CJ were, respectively, 7208.3 ± 2026.1 (1500.0-11066.6)and in PO, 2000.0 ± 751.7 (700.0 -3866.7).

Ceramic substrate
Recruit and adult individuals showed a differentiated pattern of distribution on the tested substrate surfaces (Figure 5).
The sampling number, the average adult density (ind.cm - ), the standard error and the minimum and maximum values for the three ceramic substrate segments (Figure 6a) were, respectively: s I , n = 38.00 and 1.25 ± 0.32(0.00-5.60); s II , n = 38.00 and 0.30 ± 0.11 (0.00 -2.80); and s III , n = 38.00 and 1.13 ± 0.31 (0.00 -6.80).The Kruskal-Wallis test (Dunn a posteriori) did not present significant differences (p = 0.057) between average densities of adult individuals on the three segments of the brick, even considering a predominance of adult individuals in the segments s I and s III (Figure 5b, 5c).On the other hand, when compared only to the average densities of the exposed substrates in CJ, a significant difference was verified, with segments s I and s III differing from s II (p < 0.0001).The respective average values for density, standard error, minimum and maximum were (Figure 6b): s I , n = 12.00 and 2.90 ± 0.50 (0.60 -5.60); s II , n = 12.00 and 0.47 ± 0.26 (0.00 -2.80) and s III , n = 12.00 and 2.53 ± 0.58 (0.00 -6.80).
Probably, at the beginning of the settling process, recruits were distributed evenly on the substrate, and later, during size increase, they moved to the extremities.The adult specimens stayed in the extreme segments (s I and s III ) and near the ring which fixes the substrate on the frames, where they clustered permanently, losing the movement of displacement.Similar behavior has been described for Mytilus edulis Linnaeus, 1758 (blue mussel), observed mainly in the adult individuals that aggregate and form dense beds.They have a limited capacity of movement, using their byssal threads (NRC, 2009).On the other hand, Perna perna (Linnaeus, 1758) (brown mussel) show a higher capacity of movement, using their byssal threads following the tide on coastal stones (Marques, 1988), but not forming mussel beds like M. edulis.This last species occurs on many kinds of substrate (bedrock, large to very large boulders, small boulders, muddy gravel, sandy mud, muddy sand, in rockpools, under boulders, in caves, etc.) (MARLIN, 2009).Limnoperna fortunei shows a similar behavior to M. edulis, forming large mussel beds mainly between rushes on sandy shores (Santos et al., 2008).
According to Uryu et al. (1996), the presence of clusters with large individuals indicates that the locality is probably advantageous for mussel settlement and growth.
Similar macrocluster formation and individual displacement (recruits and adults) patterns were also observed by Santos et al. (2008) in (ceramic and natural) substrate, on Veludo Beach (PV), at Guaíba Lake, Porto Alegre.The authors also verified that the individuals displacement movements decreased with the increase of size, and that the recruits sought protection under and among adult shells.According to Uryu et al. (1996)  Population sampling of the golden mussel cease and stimulates secretion of byssus threads by the smaller specimens.
The displacement of recruits seeking small clusters has been observed in some marine mytilids (Tan, 1975).Morton (1960) described a similar behavior for the bivalve Lasaea rubra (Montagu, 1803), due to the positive chemotaxis.

Smooth versus pipe-shaped surface
The sampling number, average recruit density (ind.cm - ), standard error and minimum and maximum values (Figure 7a) were, respectively: smooth surface, n = 45.00 and 7.19 ± 1.59 (0.00 -44.00); pipe-shaped surface, n = 45.00 and 11.40 ± 2.20 (0.00 -50.00).The Mann-Whitney test did not present any significant differences (p = 0.286) between the average densities of golden mussel recruits on smooth and pipe-shaped surfaces.When compared only to the average densities of the exposed substrates in CJ (Figure 7b), a significant difference (p = 0.04) between the smooth and pipeshaped surfaces of the artificial ceramic substrate was verified.The sampling number, recruit average density, standard error and minimum and maximum values were, respectively: smooth surface, n = 18 and 16.58 ± 2.77 (0.00 -44.00); pipe-shaped surface, n = 18 and 26.99 ± 2.72 (8.80 -50.00).
The predominance of recruits on the substrate's pipe-shaped surface may be related to greater protection.On the smooth surface, the recruits remained exposed to predators and to the water current.The pipe-shaped surface must offer similar protection to that found by recruits under and among the adults shells, at the start of settlement.
The Mann-Whitney test did not present any significant differences (p = 0.925) between the average densities of L. fortunei recruits on smooth and pipeshaped surfaces (Figure 7a).When compared only to the average densities of the exposed substrates in CJ (Figure 7b), no significant difference (p = 0.406) was observed between the surfaces.The sampling number, average density, standard error and minimum and maximum values were, respectively: smooth surface, n = 18 and 1.66 ± 0.40 (0.00 -5.60); pipe-shaped surface, n = 18 and 2.28 ± 0.49 (0.00 -6.80).P. E. A. Bergonci et al.Santos et al. (2008) evaluated macroclusters in ceramic substrates (six-hole bricks) on PV, at Guaíba Lake, and verified that the mussels would rather encrust on the walls of the holes than on the substrate's outer surface.
Adequate surfaces, such as slits and grooves, are preferred by L. fortunei, determining the magnitude of the invasive species settlement (Morton, 1977).Morton (1977) confirmed that, in Hong Kong reservoirs, the invasive species first colonized surfaces with slits or grooves, and only later, smooth or exposed surfaces.This behavior was also verified by Uryu et al. (1996), mainly in younger individuals.In this study, a preference for surfaces similar to those reported by the above authors was verified.This preference is called "positive thigmotactism".Similar behaviour patterns were verified by Santos et al. (2008) for a lacustrine population of L. fortunei growing on marginal vegetation and by MARLIN (2009) for M. edulis settling in crevices and fissures on hard substrate.

Sampling localities
The average values, standard error, maximum and minimum values of water quality indicators are listed in Table 1.
The nitrate concentration (p = 0.04) and the fecal coliform densities (p = 0.03) were distinct in the sampling localities.In PO, the nitrate concentration and fecal coliform densities were higher due to the sewage release from the downtown Porto Alegre sewage system and the polluted waters of the Gravataí River, contaminated by the neighboring municipalities' sewage systems.Faria and Lersch (1998) and Bendati et al. (2003) carried out extended monitoring in the Jacuí River Delta and in the Guaíba Lake, respectively, and verified better quality waters (Class 2, according to CONAMA n.20, 1986) in CJ and inferior quality waters (Class 4) in PO.
Recruit and adult maximum densities were observed in CJ, where maximum values of larval densities occurred.In PO, some factors such as the significant formation of biofilms and the adherence of organic debris on the substrates, as well as the presence of oils deriving from ship leaks, must have contributed to the lowest recruit and adult densities (Figure 8).Ind.cm -2 Ind.cm -2 Ind.cm -2 Ind.cm -2

FIGURE 3 :
FIGURE 3: Squared and segmented sheet (sn; n = I, II or III) for golden mussel subsampling on the ceramic substrate (21 x 6cm).The area of each square (Q ij ) is 1cm -2 .

FIGURE 7 :
FIGURE 7: Average densities (ind.cm - ) of Limnoperna fortunei adults and recruits, with the respective standard error for smooth and pipeshaped surfaces, on artificial ceramic substrate.A, in the Jacuí Canal and in the Port Docks (PO); B, in CJ only.

TABLE 1 :
Water quality indicators (December 2005 to March 2006), on Canal, in the Pintada Island (CJ) and in the Port Docks (PO) (Jacuí River delta, Rio Grande do Sul State, Brazil): average standard error (minimum and maximum).Fecal coliforms (CF).Probability (p) obtained through the Mann-Whitney test when comparing to the sampling localities.Significant differences*.
FIGURE 8: Average density (ind.cm - ) of adults and recruits with the respective standard error in the Jacuí Canal, on Pintada Island (CJ) and in the Port Docks (PO) (Jacuí River delta, Rio Grande do Sul state, Brazil).