Patterns of Distribution of Bivalve Populations in a Mediterranean Temporary River

: In the south of the Iberian Peninsula, many rivers are intermittent, a state most likely to be exacerbated by climate change, strongly a ﬀ ecting river biota. An additional challenge for native biota in this area is the arrival of new species, frequently aided by humans, and bivalves are particularly at risk. Here we assessed whether the native ( Unio delphinus ) and invasive ( Corbicula ﬂuminea ) bivalves di ﬀ ered in habitat use. To address this question, we sampled populations of both species in six isolated permanent pools in the same river during summer in three consecutive years. U. delphinus occurred in all pools, while C. ﬂuminea occurred only in the two most downstream pools. U. delphinus , but not C. ﬂuminea , was found preferentially in patches under riparian vegetation cover. Both species were found in similar sediment types (coarse and ﬁne gravel respectively). Although U. delphinus was present in all pools, recruitment was detected only in 2016, in one pool. We concluded that both species have the potential to compete for space, but a well-developed riparian vegetation cover may provide U. delphinus some advantage against C. ﬂuminea .


Introduction
Bivalves have a pivotal importance in freshwater, filtering phytoplankton, bacteria and fine particulate organic matter from the water column and sediment [1]. With the exception of Invasive or exotic freshwater bivalves, bivalve biodiversity is declining rapidly at a global level [2], and most native bivalves are highly endangered [3][4][5]. This decline is caused mainly by habitat degradation and biological invasions [6].
Bivalve introductions were fostered by the globalization of economic trade routes, increased watershed connectivity and recreational transport [7] and most likely will continue to occur at a greater pace. In many systems the decline of native bivalves occurs concomitantly with the spread of the invasive Asian clam (Corbicula fluminea) or Zebra and Quagga mussels (Dreissena sp.) [8]. Invasive bivalves may often become dominant (by attaining high density and biomass rapidly [9,10]), and therefore alter the community structure and function of invaded systems [11][12][13][14].
The Oeiras (Guadiana basin, South Portugal; Mediterranean climate) is an intermittent river with superficial flow after winter rains but reduced to persistent isolated pools in the summer. These pools

Qualitative Sampling
We firstly searched for bivalves with a batiscope (adapted from [12]) until the total pool area was explored or to a maximum of one hour of searching. Specimens were identified, measured, weighed and Concomitantly we characterized the pools in terms of riparian vegetation and possible human-impacted riparian features (e.g., presence of artificial structures and livestock access). We also measured turbidity and conductivity.

Qualitative Sampling
We firstly searched for bivalves with a batiscope (adapted from [12]) until the total pool area was explored or to a maximum of one hour of searching. Specimens were identified, measured, weighed and placed back at the same sampling location. For each pool, a capture-per unit of effort value (C.P.U.E.) was calculated, and results were expressed as the number of bivalves captured per person, per hour [23].

Quantitative Sampling
After the qualitative sampling, we established perpendicular transects starting at the tip of the submerged area, with a minimum distance of five meters, making sure that all micro-habitats were included. As pools length vary yearly with different submerged areas this procedure was repeated each year. In each transect, we established 0.25 m 2 quadrats with one quadrat randomly taken within the first meter of each margin (left and right) and two random quadrats in the middle of each transect to a maximum depth of 1.5 m [12,24,25]. The minimum distance between quadrats was three meters. Bivalves were identified, measured and weighed. We also recorded the water depth, distance to the margins, type of substrate and micro-habitat (see below) within the quadrat. Sediment grain size in each quadrat was classified according to American Society for Testing and Materials (ASTM) [26] as boulders (>300 mm), cobblers (75-300 mm), coarse gravel (19-75 mm), fine gravel (4.75-19 mm), sand (0.075-4.75 mm) and silt/clay (<0.075 mm).
Micro-habitats were defined by taking into consideration the sediment type, presence, type and extension of vegetation as well as whether quadrats were sampled under riparian vegetation shade. All native bivalves were left in their location while C. fluminea specimens were removed from the pools. Yearly population estimates were obtained multiplying population density by pool size.

Data Analysis
When appropriate, environmental data was tested for normality (Kolmogorov-Smirnov normality test) and homogeneity of variances (Levene's test) [27]. If normality and homogeneity of variances were not achieved, non-parametric tests were used.
Sediment data was expressed as percentages of cover at the sampling quadrats. A Multiple Correspondence Analysis (MCA) from FactoMineR package in R (Version 3.5.0, R. Core Team, 2018), was used to relate the presence/absence of bivalves and the environmental variables: sediment type, riparian gallery cover (presence or absence), distance to the nearest margin (m) and water depth (m).
We evaluated the correlation between the estimated densities and C.P.U.E. values and environmental variables using Spearman's rank correlation (SPSS, version 22.0, IBM, Armonk, New York, NY, USA). For C. fluminea the data used corresponded to the two most downstream pools. To assess habitat preferences in terms of sediment types we applied the Ivlev's electivity index, E i (Ivlev, 1961) adapted for habitat preferences (e.g., [28,29]), given by: where r i is the proportion of individuals in a habitat with a specific sediment type (i), and P i is the relative abundance of that habitat in the study area. Values of E i = −1 indicate avoidance, E i = 0 represents non-selective use of habitat type, and E i = 1 indicates exclusive use of habitat type. Population structure (sum of all pools) were compared using a G-log likelihood ratio (SPSS, version 22.0) to test for differences among years (using the post hoc z test). Six size classes were considered for U. delphinus: <30 mm, [30-40[ mm, [40-50[ mm, [50-60[ mm, [60-70[ mm and ≥70 mm. Juveniles were grouped in the first size class according to Smith [30]. Corbicula fluminea were also grouped in five size classes: <10 mm, [10-20[ mm, [20-30[ mm, [30-40[ mm and ≥50 mm.

Sites Description
The six pools differed in their physical conditions. Some locations had a well sustained riparian vegetation cover, mainly with ash and willow trees (Fraxinus angustifolia and Salix atrocinerea) and stable margins (e.g., pool C; Table 1), while others had discontinuous riparian corridors, unstable banks, some non-native tree species, modified river channels or cattle presence. Water conductivity (>600 µs cm −1 ), and turbidity (5.69-57.9 NTU's) were high. Conductivity was high in all pools as sampling occurred in late summer with low water availability and varied between years due to shifts in water volume and depth. Turbidity varied because intrinsic pool characteristics were also different. Some pools are shallower, some pools have more fish fauna, some pools might experience increased anthropogenic pressure, all are variables that might influence turbidity differently.

Bivalve Abundance and Distribution
Four bivalve species occurred in the Oeiras river. The native U. delphinus and the invasive C. fluminea were the most abundant, while Anodonta anatina and Unio tumidiformis were scarcer. Unio delphinus catch per unit effort (C.P.U.E.) ranged from zero (pools B and E in 2017) to 50.5 per researcher h −1 (pool A in 2017; Table 2). Unio delphinus densities attained up to 6.0 individuals per m 2 (pool D in 2017) while maximum densities of C. fluminea were estimated in 12.7 individuals per m 2 (pool E; Table 2, Figure 2). For C. fluminea C.P.U.E. and estimated densities were related (Spearman's rank correlation: n = 6; r s = −0.829; p = 0.042) unlike for U. delphinus (n = 18; rs = −0.273; p = 0.274). C. fluminea and U. delphinus density was not correlated (n = 170; r s = 0.110; p = 0.154). In general U. delphinus density tended to increase from 2015 to 2017 in most pools (except B and E), while C. fluminea density increased only in pool E within the same period.
At patch scale, the presence of bivalves was explained by the distance to the margins and depth (more individuals of both species in the shallow margins) and fine gravel (C. fluminea) or coarse gravel substrates (U. delphinus) (Multiple correspondence analysis; Table 3 and Figure 3). These results were partially consistent with Spearman's rank correlation: Densities of U. delphinus were correlated with the presence of fine gravel (r s = 0.126, p < 0.01) coarse gravel (r s = 0.160, p < 0.001) and riparian vegetation cover (r s = 0.153; p < 0.01). Densities of C. fluminea were positively correlated with fine gravel (n = 170; r s = 0.260; p < 0.001); sand (n = 170; r s = 0.170; p = 0.027) and silt/clay (n = 170; r s = 0.210; p < 0.01) and negatively correlated with depth (n = 170; r s = −0.207; p = 0.007) and rock presence (n = 170; r s = −0.302; p < 0.001). Although significant, all these correlations were low.     The previous findings were also mostly consistent with the Ivlev's electivity index results regarding the native Unionidae ( Figure 4; Table A1  The previous findings were also mostly consistent with the Ivlev's electivity index results regarding the native Unionidae ( Figure 4; Table A1 Table 4. Results from the post hoc z test after the G likelihood ratio was performed comparing size structure for both bivalve species. Same letters from a, b and c denote a subset of Year categories whose column proportions do not differ significantly from each other at the 0.05 level.

Species
Size

Discussion
The native U. delphinus and the introduced C. fluminea, when co-occurring, they roughly coexist in the same substrate types, preferring coarse and fine gravel, respectively. Nevertheless, U. delphinus was more abundant in locations under riparian tree cover, unlike C fluminea. The importance of vegetation for U. delphinus is unclear, but it may be related to protection against high

Discussion
The native U. delphinus and the introduced C. fluminea, when co-occurring, they roughly coexist in the same substrate types, preferring coarse and fine gravel, respectively. Nevertheless, U. delphinus was more abundant in locations under riparian tree cover, unlike C fluminea. The importance of vegetation for U. delphinus is unclear, but it may be related to protection against high temperatures caused by direct sunlight. Other authors reported U. delphinus burrowing in banks between tree roots in hydraulically more stable locations [16]. Related species, such as U. tumidiformis, U. mancus, and U. ravoisieri seem to have the same preference for river locations shaded by riparian vegetation [31,32].
Densities of U. delphinus were higher than the reported for the northern Portuguese rivers Tua and Sabor (0.015-0.121 ind/m 2 ; [33]), although sampling techniques differed, and those rivers were permanent. In the Oeiras river, while U. delphinus occurred in all the sampled pools, C. fluminea was only present in the two most downstream pools. It is possible that C. fluminea had fewer capabilities to expand upstream due to the river intermittency despite studies suggesting C. fluminea can move upstream up to 2.4 km per year [34]. In our study site the closest pool is approximately 4 km upstream (with the section in between pools being dry most of the year). Native bivalves are very adapted to summer conditions, unlike C. fluminea, which is known to be sensitive to summer environmental stress, suffering mass mortality events [35] making upstream dispersal more difficult. Some studies also suggest that the success of C. fluminea invasion decreases with increasing abundance of adult native mussels, probably due to lack of space for the invaders, physical displacement by actively burrowing mussels and locally reduced food and oxygen resources [21]. C. fluminea is a hermaphroditic species. Larvae are incubated until being released as juveniles into the water column, settle and bury [9]. Before settling they can be dragged through currents or move attached to other organisms [9]. Unlike native mussels this invasive species does not need a fish host to successfully reproduce, which is an enormous competitive advantage. On the contrary, glochidium larvae of freshwater Unionidae need to find suitable fish hosts to attach themselves to and metamorphose into free-living juveniles [17].
Our results were consistent with previous studies reporting C. fluminea preference for sediments with large organic matter content (lower grain sized sediments) such as sand mixed with silt and clay [10]. Organic material in sediments may be important for C. fluminea as this species is known for high filtration and growth/turnover rates, exacerbated during summer conditions [10,36]. Nonetheless in our pools sand was very uncommon, only detected in 2016. Therefore, the Ivlev's index avoidance and the association observed in Spearman's rank correlation have to be interpreted very carefully. Karatayev and co-workers [37], found a correlation between C. fluminea occurrence and of other three unionids. The same study reported that both were mostly found in coarse detritus (as U. delphinus in our study) and clay substrates (C. fluminea), similar to this study, due to the higher organic matter content and at the same depth.
Corbicula fluminea presence can exacerbate the pressure on U. delphinus by competing for food [21,38] and reducing available habitats for juvenile unionids. Suspension and deposit feeding negatively impact unionid recruitment, and the ingestion of unionids sperm, glochidia and juveniles may contribute to population declines [9,20]. Also, the introduction of new parasites and diseases [39], and increased ammonia toxicity as a result of massive C. fluminea die-offs, especially in the summer, may also increase native bivalves' mortality [40].
Changes in environmental conditions due to global warming may enable the colonization by new fish, which may not be suitable hosts for some native bivalve's glochidia [41]. Additionally, invasive invertebrate species may predate or compete with U. delphinus [1,9,[42][43][44][45]. Several invasive predatory species are already abundant in this river, such as the Pumpkinseed sunfish (Lepomis gibbosus), the Chameleon cichlid (Australoheros facetus) and the red swamp crayfish (Procambarus clarkii), which may partially explain the low number of juveniles detected during this three-year survey.
Finally, as a consequence of global warming, and water deviation for irrigation and livestock, a reduction in water availability is expected in pools in the coming years [14] further increasing the pressure on U. delphinus.

Conclusions
In conclusion, the knowledge about U. delphinus preference for locations under riparian vegetation, with coarse and fine gravel, may aid conservation efforts. In this context, improvement of riparian vegetation should lead to better-quality habitat, potentially decreasing suitability for invasive species such as C. fluminea. Mitigation or protection measures should start by protecting or increasing riparian areas, enabling protection from high summer temperatures, and maintaining areas with coarse and fine gravel. Since our results suggest that native and invasive species prefer similar sediment types and may compete for space, extreme caution should be taken not to allow C. fluminea expansion upstream.

Acknowledgments:
We thank SOMINCOR-Sociedade Mineira de Neves Corvo for financing this research. Data analysis was performed within the scope of the project MUSSELFLOW (PTDC/BIA-EVL/29199/2017). F. Banha and M. Gama were supported by a Post-doc grant from the University of Évora within the scope of MARE (UID/MAR/04292/2019). We thank Godinho and Alfredo for their valuable field support, suggestions and patience when field work lasted forever.

Conflicts of Interest:
The authors declare no conflict of interest.