Looking for a needle in a haystack : molecular detection of larvae of invasive Corbicula clams

The invasive bivalves Corbicula spp. and Limnoperna fortunei predominate in South American rivers. They can be sympatric in distribution, and because their larval stages are morphologically similar, monitoring them in zooplankton using microscopy protocols is often inefficient, producing ambiguous results. We designed a pair of primers to amplify a fragment of the mtDNA cytochrome c oxidase subunit I of Corbicula species. A multiplex reaction, containing the specific primer pair and a pair of universal primers (to control for the quality of the DNA in the sample) was tested with regards to specificity and ability to detect Corbicula spp. larvae in plankton samples that also contain other species in different proportions. Our molecular protocol allows for fast and accurate detection of Corbicula spp. even when concentrations of these species are low in samples, which is useful when examining large volumes of ballast/piped water. Further, the protocol is valuable for the monitoring/prospecting of early stages of the life cycle of Corbicula spp. in watersheds that have been invaded, or which are considered at risk of invasion by these species.


Introduction
Invasive species are often associated with loss of biodiversity (Rosa et al. 2011;Sousa et al. 2013;Pigneur et al. 2014), changes in native communities (Schlaepfer et al. 2005) and even accelerated extinction of native species (Clavero and Garcia-Berthou 2005).Additionally, some invasive species damage artificial structures and impact economic activities (Rosa et al. 2011).Successful invasive organisms, for instance the South American bivalves Corbicula fluminea (Müller, 1774) and Limnoperna fortunei (Dunker, 1857), often disperse efficiently using a combination of natural and human-mediated mechanisms (Cox 2004).
Corbicula clams and L. fortunei (the "golden mussel") were accidentally introduced to South America, most likely by ballast water (Darrigran and Pastorino 1993).These invasives often occur sympatrically (Darrigran 2002) and are still expanding their distribution in this continent (Oliveira et al. 2010).According to Pigneur et al. (2014), Corbicula spp.are particularly efficient invaders of river systems, reaching densities of up to several thousands of individuals per square metre in the Rio Paraná, Argentina.Corbicula spp.clams can reduce phytoplankton density and compete with native bivalve species of Mycetopodidae and Hyriidae (Santos et al. 2012).The golden mussel, on the other hand, has caused many problems for South American hydroelectric power plants by fouling in cooling ducts (Darrigran and Damborenea 2009;Belz et al. 2012).Management and control strategies need to be implemented for these species where they are present, and should include continuous evaluations of propagule pressure in new habitats (Darling and Blum 2007).
Active search for individuals is frequently used to monitor invasive bivalves.Adult specimens can be found in the substrate, and larval stages present in plankton are detected using optical  (2,4,6,7,9,10,11,12,14,16,18,20) and L. fortunei (1,3,5,8,13,15,19) are at these stages in their life cycle.microscopy (Pestana et al. 2008;Lopes and Vieira 2012).However, zooplankton monitoring is often inefficient (Mansur et al. 2012a).According to Darrigran et al. (2009), this is in part because larvae of Corbicula clams and the golden mussel are very similar (Figure 1), which makes species determination under the microscope difficult, repetitive and tedious.One (untested) strategy that is often adopted by surveyors is to assume that all free-living bivalve larvae found in freshwater plankton samples are golden mussels.This strategy is based on the premise that all larval stages of native bivalves are exclusively parasitic (glochidia of Mycetopodidae and Hyriidae species; Mansur et al. 2012b;Gatlin et al. 2013) and/or that some species of Corbicula incubate their initial larval stages in the gills of their parents (Martins et al. 2006;Houki et al. 2011;Mansur et al. 2012b).
A molecular protocol for monitoring golden mussel larvae (Pie et al. 2006) has been widely used in hydroelectric power plants in Southern Brazil (Boeger et al. 2007).Combined with microscopic procedures, this protocol assists with larval identification and informs decision-makers regarding the need for management interventions such as the chemical control of larval settlements in the cooling system of turbines.However, this molecular protocol has failed to detect L. fortunei larvae in zooplankton samples in the past, even when large numbers of bivalve larvae were detected under the microscope.These results have prompted us to ask whether the early developmental stages present in these samples were in fact larvae of Corbicula spp., since no other freshwater species of bivalves in the freshwater environs of South America release larvae in the plankton.
We believe that markers are needed for species of Corbicula because the taxonomy of the genus is uncertain and species determination is difficult due to morphological plasticity/variability (Lee et al. 2005;Pigneur et al. 2011).As a consequence, the species composition of invasive Corbicula clams in South America and in other continents is largely questionable (Pfenninger et al. 2002;Lee et al. 2005;Hedtke et al. 2008;Pigneur et al. 2011).We therefore developed a molecular protocol for the detection of Corbicula species, similar to the one available for L. fortunei (Pie et al. 2006), with the following goals in mind: (i) to provide a tool to monitor the temporal and spatial availability of bivalve larvae; (ii) to facilitate identification of adults and larvae; (iii) to investigate whether free early larval stages of Corbicula (outside their parents' gills) are common in plankton samples, and (iv) to ascertain whether Corbicula spp.occur sympatrically with L. fortunei larvae in South America.

Sampling and DNA extraction
Zooplankton samples and adult specimens of Corbicula spp.and L. fortunei were collected from reservoirs and Hydroelectric power stations (UHE) in southern Brazil (Table 1).Zooplankton samples were collected by filtering 4,000 L of water through a plankton net (64 µm mesh size), following Tschá et al. (2012).Two independent zooplankton samples were obtained from each collecting point and were preserved in 96% ethanol and taken to the laboratory.One zooplankton sample from each collecting point was processed under the dissecting scope and each identified bivalve larva was transferred to a microscope slide.Larval stage determination, based on Santos et al. (2005), was performed under a light microscope.Whole DNA extracts of zooplankton samples (from the second zooplankton sample per collection point) and DNA extracts from individual larvae were subjected to molecular protocols for the identification of Corbicula spp.and L. fortunei.
The concentration of all DNA was measured using a NanoDrop 3300 (Thermo Scientific).

Design of specific COI primers for Corbicula spp
A mtDNA fragment (700 bp approx.)from the cytochrome oxidase subunit I (COI) gene was amplified from the DNA of adult specimens of Corbicula spp.using a universal primer pair (LCO and HCO, Table 2).DNA was amplified in 25 μL reactions with 2-3 ng/μL of template DNA, 2 mM of MgCl 2, 0.4 mM of dNTPs, 1X buffer, 1.25 U of AmpliTaq DNA Polymerase and 0.5-1 mM of each primer.The following program protocol was used to obtain products: initial denaturation at 95°C for 5 min, followed by 35 cycles of 30s at 92°C, 30s at 48-51°C, 30s at 68°C, and final extension at 68°C, for 2 min.Amplified fragments were sequenced in laboratory, in both directions, using Applied Biosystems 3130 automatic sequencer and the same amplification primers.Sequences were assembled, edited and a consensus was generated using Geneious® 6.1.2(Biomatters; Available at http://www.geneious.com/).We compiled 25 COI sequences (600 bp approx.after trimming) derived from ten adults of Corbicula spp., as well as sequences from closely related species available on GenBank (Table 3), and aligned all these sequences based on the frequency of mismatches between them.Transversions and gaps were given more weight.Based on this alignment, unique regions were identified in the COI sequences of Corbicula spp., and a primer pair was designed to amplify 400 bp of their mtDNA (

Development of a Multiplex PCR assay
A multiplex PCR assay was developed using a pair of invertebrate universal primers that amplify an 800 bp fragment of nuclear 18S rDNA (Table 2) in addition to the specific COI marker that we developed for Corbicula spp.The universal 18S rDNA primer pair serves as positive control to account for variable DNA quality (Pie et al. 2006;King et al. 2009;Ludwig et al. 2011) and inhibition of the PCR of each individual sample.The Multiplex reaction was optimized by changing primer concentrations, DNA template concentration, and annealing temperature and time.
Specificity tests were conducted by testing the designed primers against samples of other mollusk species found in South America: L. fortunei, C. gigas, M. brasiliensis, Thais sp. and M. tuberculatus (some of these species were chosen also to test against species that could be found in ballast water).These tests ensured that the designed primers were specific to Corbicula spp., and that DNA from other species in a multiplex reaction would not result in cross-amplification.
Also, a sensitivity and specificity test was performed by adding the equivalent of the DNA content of one, two, four and sixteen bivalve larvae to total DNA aliquots derived from a zooplankton sample that did not contain Corbicula spp.(confirmed by the absence of the specific band in the application of the multiplex  PCR designed herein).The DNA content of a single larva of Corbicula spp. was estimated from larvae of L. fortunei (Pie et al. 2006), since they have similar size (approximately 28.5 ng of DNA per larva).The equivalent volume of the simulated number of larvae was added to an extract of 50µL of the full genomic DNA of the zooplankton sample (500 ng/µL).This plankton sample was primarily composed of cyclopoid copepods, bivalve larvae, insect larvae, cladocerans, tardigrades, nauplii larvae and mites.
An additional test involved the use of environmental plankton samples obtained from distinct regions of Brazil (see Table 1).In order to identify each larva collected to the species level, and to evaluate the presence of the species in plankton samples, molecular markers specific for L. fortunei (e.g.Pie et al. 2006) were used in parallel with the molecular markers for Corbicula spp.developed in this study (Table 2).

Results
We designed a pair of primers to amplify a 400 bp fragment of the mitochondrial DNA COI gene of Corbicula spp.(Table 3).The optimized conditions for the Multiplex PCR assay are the following: initial denaturation of 5 min at 95°C, 35 cycles of 94°C for 30s, 44°C for 30s and 72°C for 40s, and final extension of 5 min at 72°C.The amplification reaction (25 μL) consisted of 3 mM of MgCl 2, 0.4 mM of dNTPs, 1X buffer, 2.5 U of AmpliTaq DNA Polymerase, and 4 mM of specific primers.PCR products were analyzed using electrophoresis in 1.5% agarose gel to compare the size of each amplified fragment with a marker of known size (Ladder 1Kb Invitrogen®).All sequences obtained by us from these fragments matched 100% with Corbicula spp.sequences in BLASTn (at https://www.ncbi.nlm.nih.gov/).The specificity test revealed that the designed primers did not amplify the DNA of any other taxon tested (Figure 2A).The combined tests of sensitivity and specificity with plankton samples spiked with specific amounts of Corbicula spp.DNA, demonstrating that the protocol above is capable of detecting the DNA of a single larva when this DNA is pooled with DNA extracted from a plankton sample (Figure 2B).
As a whole, the quality of the DNA in the environmental plankton samples processed in this study was adequate (non-degraded), as indicated by the positive amplification of the 18S rDNA fragment (Figure 2A).After identifying the early bivalve larval stages in the zooplankton of Southern Brazilian rivers (Table 1) we separated 160 larvae for molecular identification, of which only 129 had adequate amounts of DNA for molecular analysis.After application of the molecular identification protocol on both species, 49 larvae were identified as Corbicula spp.and 80 as L. fortunei (Table 1).D-shaped larva (see Santos et al. 2005) was the most common stage found for both species.Larvae of Corbicula and L. fortunei were detected in sympatry by the Multiplex reaction applied to zooplankton samples and to individual larvae in one location, UHE Jauru (Table 1).

Discussion
Implementation of the specific molecular protocol to detect/identify Corbicula spp.larvae in environmental samples demonstrated that one or more species of this genus were present in the plankton of 4 out of 7 locations sampled.Since the current protocol cannot distinguish among Corbicula species, positive results may indicate the presence of one or more of the species recorded from Brazil (C.fluminea, C. largillierti and C. cf.fluminalis) (see Martins et al. 2004;Mansur et al. 2012a).Efforts are currently being made in our laboratory to improve this molecular protocol in order to differentiate among all species of Corbicula that occur in South America.
Our results contradict the notion that all species of Corbicula incubate in the demi-gills (e.g.Cataldo and Boltovskoy 2000;Martins et al. 2006) and suggest that their larval stages occur in sympatry with early larvae of L. fortunei (Table 1).Therefore, the widespread strategy currently used in South America for the microscopic detection and quantification of L. fortunei larvae, which assumes that all free-living bivalve larvae found in freshwater plankton samples are golden mussels, is inappropriate.While most freshwater Corbicula are hermaphrodites and ovoviviparous, with incubation in the maternal gill (e.g.Glaubrecht et al. 2006), some species do employ different reproduction modes, as reported by Byrne et al. (2000) for Corbicula australis (Deshayes, 1830).Corbicula australis is dioecious and incubates veliger to pediveliger larvae in the inner demibranchs.There are published records of C. fluminea being hermaphroditic, incubating juveniles in outer demibranchs and releasing planktonic veliger larvae (McMahon 2002).Corbicula fluminalis is not known to incubate larvae in its gills (e.g.Korniushin 2004).
Although the molecular detection protocol designed herein reveals only the presence/absence of Corbicula spp.larvae in plankton, this information can aid in studies on propagule pressure, which will allow for rapid effective management response and preparedness (e.g.Darling and Blum 2007).However, in order to increase the usefulness of the results derived from this method, the present protocol needs to be improved to allow quantification of larvae by real-time PCR, as proposed by Endo et al. (2009).Quantification of larvae of L. fortunei and Corbicula spp. is fundamental to guide methods of control, especially in the definition of dosages of anti-fouling and/or molluscicide products usually applied to semi-closed water systems such as cooling systems of hydroelectric power plants.
Early detection, which allows for rapid response, is crucial for integrated programs of management and control of invasive species (e.g.Molnar et al. 2008).However, early detection is often difficult when the invasive organism is small, inconspicuous and/or difficult to identify.Detecting invasive species during the first phases of an invasion, when they are still in low concentrations, is important for successful intervention.Therefore, our protocol represents an important tool to monitor and understand the biology and larval dispersal capacity of Corbicula species in continental waters.Similar monitoring has been applied systematically to L. fortunei since 2006 by technicians of the Instituto Lactec (http://www.institutoslactec.org.br/),COPEL (http://www.copel.com),

Table 1 .
Pie et al. 2006for the collection of zooplankton and specimens of Corbicula spp.andL.fortunei.The use of specific molecular protocols (this work;Pie et al. 2006) determined the presence (+) or absence (-) of larvae of both species in each location.

Table 2 )
. Primer sequences were tested with GenBank's Basic Local Alignment Search Tool (BLASTn) (at https://www.ncbi.nlm.nih.gov/) to ensure that the designed primers would match only COI sequences of Corbicula spp.

Table 2 .
List of primers used in the development of the molecular detection protocol for Corbicula spp.

Table 3 .
Species list and their GenBank accession numbers used in the development of Corbicula spp.molecular markers.