Introduction
The latitudinal biodiversity gradient peaking in the tropics has been of key interest in macroecology for decades (Hillebrand, 2004, Kinlocket al. , 2018) both at regional (Edgar et al. , 2017, Saeediet al. , 2019) and global scales (Chaudhary et al. , 2016, Gagné et al. , 2020, Righetti et al. , 2019, Tittensoret al. , 2010). In particular, the various possible environmental drivers of latitudinal gradients have been discussed at length (Fieldet al. , 2008, Gagné et al. , 2020, Wang et al. , 2009). Understanding the impacts of changing environments on such large-scale diversity patterns is of growing importance (Gagné et al. , 2020, Pimm et al. , 2014).
Because data on large geographic scales are required for these studies, recent research has focused on using composite datasets rather than detailed field collections (for example; Gagné et al. , 2020, Menegotto and Rangel, 2018, Miller et al. , 2018). However, there are some large-scale fieldwork schemes such as the Reef Life Survey (Barneche et al. , 2019, Edgar et al. , 2017, Edgar and Stuart-Smith, 2014) and some terrestrial compilation efforts (Cerezeret al. , 2020). Historically, species distributions have been based on range data, layering distributions to generate richness estimates (McKinney and Kark, 2017, Roy et al. , 1998, Tsianouet al. , 2016). Range-through, more often used in temporal studies, tends to artificially increase richness estimates and similarities towards the centre of distributions both spatially and temporally (Boltovskoy, 1988), but has been used to test many latitudinal gradient hypotheses (Hughes et al. , 2013, Royet al. , 1994, Roy et al. , 1998). Atlases have similar problems and are generally only available for well-studied taxa (Donald and Fuller, 1998, Robertson et al. , 2010). Uneven sampling and techniques that vary between countries also have an impact, even in better-surveyed groups like mammals and birds (Robertson et al. , 2010, Whittaker et al. , 2005). Both range-through and atlas approaches result in a loss of abundance information and assume even sampling across species ranges, which makes diversity estimates misleading compared to those produced by routine field surveys (Robertson et al. , 1995) despite successful modelling studies using atlas data (Sadoti et al. , 2013).
In modern studies, large composite databases are often used to generate diversity estimates and taxonomic ranges. Although many such databases exist, the largest and most heavily cited are the Global Biodiversity Information Facility (GBIF) for terrestrial and marine studies and the Ocean Biogeographic Information System (OBIS) for marine studies. Despite the obvious benefits of these datasets, their properties can create significant issues in global-scale studies. Data included are haphazard, idiosyncratic and, like atlases, unevenly distributed across countries (Beck et al. , 2014, Boakes et al. , 2010) and globally (Menegotto and Rangel, 2018). Despite large numbers of records, they also do not give as much range information as many manual compilation methods (Beck et al. , 2013), and without true abundance information, they are of limited usefulness in estimating true diversity. In addition, sampling effort varies temporally (Boakeset al. , 2010), in intensity (Ballesteros-Mejia et al. , 2013), and between countries (Mora et al. , 2008), resulting in a lack of understanding of local-scale processes.
In this study we investigate a latitudinal diversity gradient in marine bivalves using a field dataset spanning 2,667 km of Australia’s eastern coastline. Although marine bivalves are underrepresented in large datasets (Troudet et al. , 2017), they are well-studied in other continents (Jablonski et al. , 2013, Roy, 2001, Roy et al. , 1994) and easy to collect in large abundance in the field. Bivalves are not included in major field surveys of Australian waters, such as the Reef Life Survey. Thus, diversity estimates have been limited to small numbers of specimens and observations in OBIS. We will describe variation in bivalve diversity with respect to varying environmental conditions along the coastline, as well as across a previously identified major biogeographic transition. Finally, the strength, variation, and environmental response of this gradient will be compared to the latitudinal gradient found in analyses of a macroecological dataset generated using bivalve occurrence data from OBIS.
Data and Methods