Size-mass relationships of Melanoides tuberculatus ( Thiaridae : Gastropoda ) in a eutrophic reservoir

This study evaluated the relationships of certain allometric measurements in Melanoides tuberculatus Muller, 1774, in order to develop a statistical model to estimate the biomass of this mollusc species. We measured the total length and aperture of 70 shells. These measurements were correlated with the biomass values to construct exponential and power-function models, and both models showed high coefficients of determination. The exponential model was the better biomass predictor, with a coefficient of determination over 93%. These proposed models may be an effective tool to determine the biomass of M. tuberculatus in eutrophic Brazilian reservoirs.

Reservoirs are artificial ecosystems that functions mainly to generate electricity and supply water for domestic and industrial processes.From the ecological point of view, reservoirs produce considerable alterations in aquatic ecosystems due to changes in the time of water residency, habitat fragmentation and exotic species invasion (TUNDISI 2008, AGOSTINHO et al. 2005).Eutrophication is also a serious threat, since the excessive growth of algae resulting from enrichment of the water, mainly with phosphorus and nitrogen, can modify the biodiversity and the distribution of organisms in these environments (FIGUEIRÊDO et al. 2007).
In tropical countries, both native mollusc species and exotic invasive species such as Melanoides tuberculatus Muller, 1774 occur in reservoirs.The Afro-Asian M. tuberculatus was originally reported in Brazil in 1967 in Santos (São Paulo), and spread to Brasília (Distrito Federal) and the states of Rio de Janeiro, Goiás, Paraíba and Espírito Santo (FERNANDEZ et al. 2003, VAZ et al. 1986).Melanoides tuberculatus was first reported in the state of Minas Gerais in 1986, from Pampulha Reservoir in the city of Belo Horizonte (CARVALHO 1986, FREITAS et al. 1987).In a study begun in September of that same year at Soledade Lake in the Ouro Branco region, but not published until 1994, some individuals were also found (SILVA et al. 1994).In 1996, the species was reported from Dom Helvécio Lake in the Rio Doce State Park (DE MARCO JR 1999).
This invasive mollusc threatens the biodiversity of ecosystems where it is found, and it also represents a risk for human health since it hosts the trematodes Paragonimus westermani Kerbert, 1878 and Clornorchis sinensis Looss, 1907 species endemic to Asia that are able to parasitize humans (SOUZA & LIMA 1990).The species is an r strategist, with parthenoge-netic reproduction and the potential to maintain high population densities for long periods of time.It is easily transported and highly adaptable, establishing on all kinds of substrates (POINTIER et al. 1993).
Because of all these reasons, M. Tuberculatus is found in many aquatic ecosystems with different levels of pollution and ranging from oligotrophic to hypereutrophic (ROCHA-MIRANDA & MARTINS-SILVA 2006, CALLISTO et al. 2005).Since M. Tuberculatus is a potential competitor of native planorbid species that host human parasites, it has been introduced into the American continent for purposes of biological control (GUIMARÃES et al. 2001).
In studies on benthic aquatic insects, biomass estimates provide information to estimate growth, production, and feeding ecology (BENKE et al. 1999, JOHNSTON & CUNJAK 1999).Similarly, information on biomass can be used to support inferences about the adaptation of M. tuberculatus in eutrophic environments and its effect on other species of molluscs (POINTIER & AUGUSTIN 1999).
However, biomass estimation takes time and is prone to errors, especially when the organisms are preserved.Preservatives can produce significant losses of biomass and most are toxic (WETZEL et al. 2005).Because of these and other problems, that many studies on insects have used the size of body structures as biomass predictors (GENKAI- KATO & MIYASAKA 2007, GONZÁLEZ et al. 2002, CRESSA 1999).However, the relationships between size and weight must be used cautiously since they may not account for environmental and geographical variations and are not exact for every taxon, i.e., there is a different relationship for each species (JOHNSTON & CUNJAK 1999).
Since M. tuberculatus was recorded in Brazilian waters, few studies have assessed the spread of this invasive mollusc in this country.The status of the species is still poorly known, as is its distribution and ecology (FERNADEZ et al. 2003).The size structure of a population of M. tuberculatus in a eutrophic reservoir might be an important tool to aid in the management of this alien species in southeastern Brazil.
The objective of the present study was to establish, through allometric relationships, a statistical model to estimate the biomass of M. tuberculatus populations.

MATERIAL AND METHODS
The Ibirité Reservoir (19°07'00"-20º02'30"S, 44°07'30-44º07'30"W) is fed mainly by the Ibirité River.The dam was construted in 1967 in order to supply water for the Gabriel Passos refinery (REGAP), one of the Petrobras refineries constructed along highway BR-381.The reservoir has a surface area of 2.8 km 2 , a volume of 15.423.000m 3 and a mean depth of 16 m (GARCIA et al. 2009).
The reservoir has undergone rapid artificial eutrophication because of human impacts in its drainage basin.This eutrophic state is characterized by high primary production, algal bloom episodes, and the presence of aquatic macrophytes (MORENO & CALLISTO 2006).During the study period the mean depth at the sampling stations was 2.6 m (± 1.65).Mean dissolved oxygen concentrations were about 7.3 mg/l (± 0.68), with a mean temperature of 27.83 °C (± 0.52), a slightly basic pH (7.4 ± 0.32) and high electrical conductivity (237.6 µS cm -1 ± 23.16).
The molluscs were collected from the sediments during April 2009, by means of an Ekman-Birge sampler with a sampling area of 0.0225 m 2 (FREITAS et al. 1987).In all, 15 sampling points were selected along the shore of the reservoir, and three replicates were taken at each point (Fig. 1).The collected material was placed in plastic bags and transported, with no the added of preservatives, to the Laboratório de Ecologia de Bentos, Universidade Federal de Minas Gerais.At the laboratory, the samples were washed on sieves (0.5 mm mesh size) to separate the molluscs, on the same day of their collection.
The measurements of the total length of the shell (from the vertex to the farthest point on the opposite end) and the shell aperture were taken using a Vernier caliper with a precision of 0.05 mm.Each snail was weighed to obtain the total wet weight.The specimens were dried in the oven at 60°C until they reached a constant weight, approximately 48 hours, and then incinerated in a furnace at 500°C for four hours in order to estimate the weight of the mineral fraction.The biomass was obtained from the difference between the total dry weight and the weight of the mineral fraction.The results were expressed in milligrams and correlated with the length and shell aperture measurements in millimeters, in order to establish the relationship between weight and length (ELKARMI & ISMAIL 2007).
Dispersion diagrams were constructed, correlating the biomass (B) with the measurements of total length (TL) and shell aperture (SA).The method of least squares was used to produce the regression models.The validity of the equations was based on the regression significance (p), the coefficient of determination (R 2 ) and on the analysis of the residues (VIEIRA 2003).

RESULTS
The relationships between the measurements and the biomass of 70 individuals were used.The shell lengths ranged from 3.2 to 18.95 mm, with a mean of 10 mm (± 4.62).The shell apertures ranged from 0.75 to 5.9 mm, with a mean of 2.8 mm (± 1.47).The biomass ranged from 0.4 to 22 mg, with a mean of 5.76 mg (± 5.91).
The exponential and power-function models presented significant correlations (p < 0.001).The coefficients of determination of the regressions (R 2 ) varied from 0.9206 to 0.9685 (Tab.I).The results indicated that the shell length produced the highest coefficient of determination in the regression analyses of the exponential and power-function models (R 2 ), 0.9685 and 0.9414, respectively.

DISCUSSION
The models showed high coefficients of determination, explaining from 92 to 96% of the biomass variation as a function of the measurements used.The coefficients of determination of the exponential model were slightly higher than the coefficients of the power-function model, showing that the exponential model better explained the data variation.The power-function model showed an irregular dispersal pattern of residuals for the measurements of both shell length and aperture (Figs 2 and 3) (VIEIRA 2003).
In practice, the interpretation of the exponencial model shows that M. tuberculatus has rapid growth, is well adapted to the eutrophic environment, and probably exerts competitive pressure on other species of molluscs, meeting the requirements of an invasive species (ELKARMI & ISMAIL 2007, EVERETT 2000, POINTIER et al. 1993).Previous studies have used these relationships to propose equations for the estimation of the growth rate and/or secondary production of benthic aquatic insects (GENKAI- KATO & MIYASAKA 2007, GONZÁLEZ et al. 2002, CRESSA 1999).
Some of the specimens collected were excluded from this study because of significant losses of the shell apex.A study in Pampulha Reservoir found it difficult to obtain allometric relationships because each individual had lost on average 20.53% of its estimated theoretical length, with larger individuals having lost more of the shell (FREITAS et al. 1987).This breakage may occur due to factors such as water temperature, pH, total hardness, alkalinity, availability of calcium ions, and dissolved salts.These factors can influence the demineralization process and consequently lead to the loss of the entire shell (ORONSAYE 2002, LANZER & SHAFER 1988).Individuals with shell length and aperture greater than 10 and 3 mm, respectively, showed greater biomass variations as a function of their size (Figs 4 and 5).This may be related to the onset of sexual maturity and the stage of the reproductive period, since embryos can remain in the brood pouch for three to five months before being released (POINTIER et al. 1993, DUD-GEON 1986).
Variations in the relationship between size and biomass in populations of a species from different geographical regions can be caused by differences in the physical and chemical factors and by fluctuations in their environmental conditions (GONZÁLEZ et al. 2002).This highlights the importance of selecting geographical regions with similar characteristics and preferably specific taxa, when carrying out studies that involve these kind of regressions.On the other hand, a study by GENKAI-KATO & MIYASAKA (2007) showed that seasonal variations did not significantly affect these relationships.
In addition, the regression equations must be constructed using the data from fresh individuals, since preserved individuals can lose as much as 73.8% of their biomass during the first weeks, and this may lead to underestimation of the weight.The choice of preservative, ethanol or formaldehyde, does not seem to have a significant effect on loss of biomass, since both substances produce a large loss, mainly during the first three weeks following preservation (WETZEL et al. 2005).
The models presented here can be used to determine the biomass of M. Tuberculatus.The exponential model better described the relationship between size and biomass and the biological characteristics of the species.Because of the presence of numerous individuals lacking the shell apex, the use of the shell aperture measurement is a good alternative for the construction of the models to describe snail populations in Brazilian eutrophic reservoirs.

Figure 1 .
Figure 1.Distribution of sampling points in the Ibirité Reservoir, in the basin of the Paraopeba River, Minas Gerais.Modified from GARCIA et al. 2009.

Table I .
Regression equations and values found for each variable.The biomass is represented by B, and the shell total length (TL) and shell aperture (SA) of M. tuberculatus are represented by L. (a and b) regression constants, (R²) coefficient of determination, (e) Euler's number (2.718).