Introduction
The spatial and temporal processes that generate and maintain marine biodiversity are not fully understood (Bowen et al. , 2013; Orsiniet al. , 2013; De Meester et al. , 2016; Costello and Chaudhary, 2017). Marine populations display diverse patterns of genetic structure, such as isolation-by-distance (Wright, 1943; Chaves-Fonnegraet al. , 2015; Pérez-Portela, Noyer and Becerro, 2015), regional clustering (Selkoe et al. , 2014; Brown, Davis and Leys, 2017; Riesgo et al. , 2019), isolation-by-environment (Orsini et al. , 2013; Giles et al. , 2015), as well as patterns that are not clearly linked to spatial or environmental structuring (Cornwellet al. , 2016; Miller et al. , 2018; Taboada et al. , 2018). However, barriers to dispersal and isolating mechanisms over small spatial scales, such as the range of marine protected areas, remain elusive especially for sessile marine organisms with a dispersive larval stage (Liggins, Treml and Riginos, 2013). Sponges, integral but often underappreciated assets of benthic communities (Diaz and Rützler, 2001; Bell, 2008; Bell et al. , 2013; De Goeij et al. , 2013; Dunn, Leys and Haddock, 2015; Webster and Thomas, 2016), are generally considered to be poor dispersers as their larvae have limited swimming capacity and are short-lived (Maldonado, 2006). Sponges are therefore excellent candidates to investigate marine population genetic structure on small scales. However, despite the recognized relevance of sponges in benthic ecosystems, results are ambiguous on sponge genetic diversity, degrees of gene flow between populations and drivers of divergence.
Different studies have explored how geography, oceanography and environmental factors may influence gene flow within sponge species. Due to their restricted dispersal, the spatial scale of sponge gene flow should be limited. The majority of studies investigating genetic structure in sponges have revealed species complexes with divergence among morphologically cryptic lineages (Oppen, Wörheide and Takabayashi, 2002; Wörheide, Solé-Cava and Hooper, 2005; Uriz and Turon, 2012; Pérez-Portela and Riesgo, 2018). However, studies investigating within-lineage divergence are scarce. This may be a result of many most studies using relatively small amounts of genetic data and a single type of genetic marker (Selkoe et al. , 2016; Timm, 2020). An increase in number of molecular markers is expected to advance inferences on demography and structure (Felsenstein, 2004; Allendorf, Hohenlohe and Luikart, 2010; Kelley et al. , 2016; Pérez-Portela and Riesgo, 2018). Molecular markers commonly used to assess sponge phylogeography and population structure include mitochondrial markers (mtDNA) such as Cyochrome c oxidase I (COI ) and ATP6, and nuclear markers such as introns, internal transcribed spacers (ITS ) and microsatellites (Oppen, Wörheide and Takabayashi, 2002; Wörheide, Solé-Cava and Hooper, 2005; Uriz and Turon, 2012; Pérez-Portela and Riesgo, 2018). Though widely used in phylogeographic and population genetic studies (Avise, 2000, 2009), mitochondrial markers exhibit low mutation rates in sponges (Wörheide, Solé-Cava and Hooper, 2005; Huang et al. , 2008). As a result, the majority of studies using mtDNA find panmixia among sponge populations across broad geographic ranges (e.g. Duran, Pascual and Turon, 2004; Whalan et al. , 2008; De Bakker et al. , 2016; Ekins et al. , 2016). ITS markers can show more structure (Bentlage and Wörheide, 2007; Becking et al. , 2013; Ekinset al. , 2016), but generally at larger spatial scales and are hampered by intra-genomic polymorphisms (Frankham, Briscoe and Ballou, 2002). Microsatellites could be reliable and sufficiently variable to detect population structure, yet are time-consuming to design de novo for each species (Frankham, Briscoe and Ballou, 2002; Pérez-Portela and Riesgo, 2018), and generally relatively few markers have been used per study (<20), again limiting the molecular marker panel. Furthermore, microsatellites can be confounded by homogenizing forces of evolution, making them less effective in detecting genetic divergence (Oppen, Wörheide and Takabayashi, 2002). Hence, there is a need for increasing genetic resolution in order to reassess assumptions of panmixia within sponge populations at fine spatial scales.
Recently, there has been an increase in the use of reduced representation genomic methods and Single Nucleotide Polymorphisms (SNPs) for population studies on non-model organisms (Baird et al. , 2008; Peterson et al. , 2012; Puritz et al. , 2014; Catchen et al. , 2017). Genome-wide SNP data increases the number of loci compared to traditional mitochondrial or nuclear markers and a larger marker panel is expected to small scaled population structure when compared to single marker studies. The effect of an increased marker panel has been shown for example in mussels (Becking et al. , 2016; Maas et al. , 2018; de Leeuw et al. , 2020), and fish (Bradbury et al. , 2015; Lemopoulos et al. , 2019; D’Aloia et al. , 2020; Sunde et al. , 2020). However, high resolution studies on sponges are lagging behind (Pérez-Portela and Riesgo, 2018), with notable exception of Brown et al. (2014), Brown et al. (2017) and Leiva et al. (2019). Using the SNPs generated by Brown et al. (2014), Brown et al. (2017) genotyped 67 SNPs for the deep-sea glass sponge Aphrocallistes vastus and found high differentiation between geographic regions (average FST= 0.25), but no structure at distances <275km, indicating connectivity at this scale. Leiva et al. (2019) observed panmixia for the Antarctic sponge Dendrilla antarctica over 900km when analyzing 389 neutral SNPs. However, 140 SNPs under putative positive selection did show genetic differentiation (global FST = 0.20) over 100km. Potentially the number of SNPs used in these studies are still too low to detect small-scaled population divergence. Using RADseq techniques such as ddRAD (Petersonet al. , 2012) may increase the number of retained SNPs to thousands and provide the necessary resolution.
Another challenge to unveiling sponge population genetic patterns is that sponges are considered true holobionts, associations between the host and its microbes (Webster and Thomas, 2016), and may not evolve as single units. Due to the propensity of sponges to harbor dense communities of microbes, there is a potential of including associated microbial material in extractions, therefore clouding host specific patterns. Studies into the sponge holobiont have suggested that microbial communities are highly specific to sponge host identity (Easson and Thacker, 2014; Reveillaud et al. , 2014), and communities to be stable across gradients in geography (Taylor et al. , 2005), time (Hardoim and Costa, 2014), and, for tropical reef sponges, depth (Steinert et al. , 2016). However, these expectations do not always hold true (Swierts, Cleary and de Voogd, 2018; Cleary et al. , 2019; Easson et al. , 2020; Ferreiraet al. , 2020). For example, Easson et al. (2020) concluded that microbe community structure is influenced by the interplay of geographic, environmental and host factors, with a potential effect of even small population-level genetic structure. Since patterns of microbial diversity can differ from sponge host diversity (Noyer and Becerro, 2012), it is important to understand how microbial community patterns are related to sponge host population genetics.
Islands, and other insular systems, provide ideal models to test factors that underlie population structure since they are well-defined and are of lower complexity than open areas (Warren et al. , 2015). Marine lakes are insular systems of bodies of seawater surrounded completely by land that maintaining a connection with the surrounding sea through caves or porous rock (Holthuis, 1973; Hamner, Gilmer and Hamner, 1982; Dawson et al. , 2009; Becking et al. , 2011). Clusters of marine lakes are present in the Caribbean, Vietnam, Palau, and Indonesia, particularly in East Kalimantan and in West Papua (Dawsonet al. , 2009; Becking et al. , 2011; Becking, de Leeuw and Vogler, 2015). Marine lakes were formed de novo when depressions in karstic rock were filled with sea water after the Last Glacial Maximum (approximately 20,000 years ago) (Tomascik and Mah, 1994), and house clearly defined populations (Gotoh et al. , 2011; Itescu, 2018). Sponges are usually well-represented in marine lakes, having high diversity and abundance (Azzini et al. , 2007; Becking et al. , 2011, 2013; Cleary et al. , 2013). Having originated roughly at the same time marine lakes represent relatively controlled biotopes where each lake can be seen as an independent replicate of eco-evolutionary dynamics over time.
Marine lakes have been used before to study population genetic and microbial community patterns studies (Becking et al. , 2013; Cleary et al. , 2013; Cleary, Polónia and de Voogd, 2018; Ferreiraet al. , 2020). The sponge Suberites diversicolor(Porifera, Demospongiae, Suberitidae, Becking and Lim, 2009) has been found to occur extensively in marine lakes, and also in brackish coastal areas (Becking and Lim, 2009; Cleary et al. , 2013). UsingCOI and ITS genetic markers, Becking et al. (2013) studied the phylogeography of S. diversicolor from multiple marine lakes and lagoon populations in the Indo-Pacific. They identified two distinct genetic lineages and regional structuring yet did not observe subtle levels of structuring at smaller spatial scales. The lack of structure could be explained by recurrent gene flow among lakes, or by lack of resolution of genetic markers used by Becking et al.(2013), as they recovered a low number of haplotypes (4 for ITSand 3 for COI ). Analyzing the microbes of the same S. diversicolor populations, Cleary et al. (2013) found that the associated microbial community did not differ among sponges sampled from marine lakes and open sea habitats within one region. Between broad geographic regions (>1,400km) the associated microbial communities were significantly different (Ferreira et al. , 2020). Clearly, there is a need to further elucidate population genetic patterns and see how host and microbe patterns contrast.
Studying a priori defined sponge populations from nine marine lakes and two lagoon locations in Indonesia (Berau, East-Kalimantan and Raja Ampat, West-Papua) and Australia, we aim to assess the population structure in S. diversicolor and associated drivers. Selecting marine lakes on different spatial scales (1-1,400km), along a gradient of connection to the surrounding sea and with different environmental regimes allows the opportunity to assess effects of geographic distance, permeability of barriers and local environments in shaping genetic structure. In order to assess the effect of level of genetic resolution, we compared results of our genome-wide sequencing strategy (double-digest restriction-site associated DNA sequencing, (ddRAD, Peterson et al., 2012)) to previously published results on the same individuals using single markers (COI and ITS ) (Beckinget al. , 2013). Furthermore, we compare host population structure with previously published structure from associated microbial communities in S. diversicolor (Cleary et al. , 2013; Ferreira et al. , 2020) to assess whether sponge host and microbes are on similar evolutionary tracts.