Introduction
Diatoms are photoautotrophic unicellular organisms present in almost every freshwater habitat (Zimmermann, Glöckner, Jahn, Enke, & Gemeinholzer, 2015). They are one of the main components in the nitrogen, phosphorus, silicon and carbon biochemical cycles, being responsible of at least the 25% of the global carbon dioxide fixation and the 20% of the global net primary production (Wilhelm et al., 2006; Zimmermann, Jahn, & Gemeinholzer, 2011; Zimmermann et al., 2015). In addition, benthic diatoms are very important ecological indicators as they are highly sensitive to environmental conditions and have relatively short generation times (Vasiljević et al., 2014; Visco, Apothéloz-Perret-Gentil, Cordonier, Esling, Pillet, & Pawlowski, 2015; Round, Crawford, & Mann, 1990). Consequently, analyzing their communities is important to provide an overview of water quality and allow the detection of environmental changes in freshwater assemblages (Kermarrec et al., 2014).
Researches efforts have historically been focused on understanding how ecological communities are assembled in space and time and how they vary at local scale (Bishop, Robertson, van Rensburg, & Parr, 2015; Leibold et al., 2004). In this vein, the metacommunity concept, first defined by Leibold et al. (2004) as “a set of local communities linked by dispersal of multiple potentially interacting species”, offers the possibility to examine the complex ecological mechanisms, since it takes into account the regional-scale processes in addition to local ones (Leibold et al., 2004). From the very beginning (Brown, Sokol, Skelton, & Tornwall, 2016), four not mutually exclusive paradigms -patch dynamics, species sorting, mass effect and neutral perspective- have been associated with metacommunity organization. Each of the four paradigms is characterized by the amount of weight they place on a combination of regional and local processes, disturbance, and the degree to which species are equivalent in their functional biology (Brown et al., 2016). More specifically, in neutral perspective, speciation, extinction, migration and immigration processes participate structuring ecological communities, while in patch dynamics there is a colonization-competition trade off. In species sorting perspective, environmental filtering and biotic interactions filter the occurrence of the species, whereas in mass effect paradigm, high rates of dispersal in addition to environmental factors are taking into account (Heino et al., 2015; Göthe, Angeler, Gottschalk, Löfgren, & Sandin, 2013). Taking in consideration a metacommunity perspective, the β-diversity concept (i.e. variation in community composition and structure among sites in a geographical area) is essential to understand ecosystem functioning, since it provides important information about the patterns of diversity and processes that modify the ecosystems (Bonecker et al., 2013; Florencio, Díaz-Paniagua, Gómez-Rodríguez, & Serrano, 2014). As a consequence, several studies examining diatoms β-diversity at various spatial scales (Green et al, 2004; Smucker & Vis, 2011; Leboucher et al., 2019; Rodríguez-Alcalá et al., 2019) have reported that spatial structure is responsible of a significant proportion of the community variance, which suggest that diatoms lack strict ubiquitous dispersal and thereby exhibit clear biogeographical patterns (Soininen, 2007).
Morphological identification of diatoms species is subjected to a bottleneck since the number of expert taxonomists is decreasing whereas the number of identified taxa is increasing rapidly (Pečnikar and Buzan, 2013). Traditional taxonomic identification relies in morphological traits and light or electron microscopy, so is time-consuming and demands specialized knowledge (Blanco, 2020). In addition, the presence of cryptic species and phenotypic plasticity found in some species may hinder classical species identification (Hadi et al., 2016). In this vein, several studies have recently revealed the presence of hidden diversity on diatoms (Trobajo et al., 2010; Rovira, Trobajo, Sato, Ibáñez, & Mann, 2015), suggesting that its diversity has like been underestimated (Mann & Vanormelingen, 2013). As a consequence, exclusive reliance on morphological traits may lead to ambiguous identification of taxa (Zimmermann et al., 2011; Kowalska, Pniewski, & Latała, 2019). To overcome the biases related with morphological identifications, alternative techniques such as DNA barcoding have been developed in recent years. The term DNA barcoding was first used by Arnot, Roper, & Bayoumi (1993) to refer the possibility of differentiate stocks and lineages of Plasmodium falciparum based on targeting tandem repeats of circumsporozoite gene. DNA barcoding relies in a short DNA fragment, which is sequenced and compared with a reference library, allowing the identification of all taxa independently of its life stage (Zimmermann et al., 2015; Rimet, Vasselon, Keszte, & Bouchez, 2018; Hebert, Cywinska, Ball, & deWaard, 2003). In addition, DNA metabarcoding combine DNA barcoding methods with high-throughput techniques (HTS) allowing the sequencing of millions of DNA fragments from many samples simultaneously (Rivera et al., 2017). Metabarcoding has been applied regularly to inventory taxonomic diversity of freshwater diatoms since the pioneering study of Kermarrec et al. (2013). Since then, several studies have highlighted the usefulness of metabarcoding approach in combination with morphological methods for diatom biomonitoring (Mora et al., 2019; Zimmermann et al., 2015). However, the incompleteness of reference database still remains as one of the main biases constraining the accuracy of molecular-based inventories.
Here, we used a comprehensive dataset of 22 ponds located in a Mediterranean landscape to (i) assess if morphological and metabarcoding approaches could be comparable methods in β-diversity studies of benthic diatoms, and (ii) test whether environmental filtering and dispersal limitation, predictably assemble diatom communities. Importantly, we also assessed the influence of taxonomic resolution (genera and species level) on the observed community-level patterns and processes. We focused our study exclusively on incidence values, avoiding the use of abundance values in order to prevent the biases related with the copy number variation of the marker gene per cell (Vasselon, Domaizon, Rimet, Kahlert, & Bouchez, 2017). Following recent works (Rimet et al., 2018; Medlin, 2018; Kowalska et al., 2019), we hypothesized that morphological methods based on light microscopy and metabarcoding approach provide similar results when it comes to the β-diversity patterns of diatom metacommunities (Hyphothesis 1 ). On the other hand, we expected to find a relationship between environmental template and compositional variation, since freshwater diatoms have usually been found to be strongly related to environmental dissimilarity at regional extents (Hyphotesis 2 ; Declerk, Coronel, Legendre, & Brendonck, 2011).