Corals, like most colonial animals, are prone to asexual reproduction via fragmentation that creates clonal copies. Genetic studies have consistently shown that asexual reproduction is very common; consequently, coral colonies from the same reef (or nearby reefs) often have identical genotypes. A growing number of studies have formally quantified the genetic similarity among proximate colonies (i.e., positive spatial autocorrelation, for example: Miller 1998; Costantini et al. 2007; Underwood et al. 2007; Miller and Ayre 2008; Underwood et al. 2009; Gorospe and Karl 2013; Smilansky and Lasker 2014). Given that clonality appears to be ubiquitous for corals and that closely related colonies (including clonal copies) tend to be nearby, how does spatial genetic structure (especially small-scale positive spatial autocorrelation) affect common metrics of genetic diversity? And, how difficult is positive spatial autocorrelation to detect?

In the present paper, Gorospe et al. (2015) demonstrate that spatial genetic structure within reefs should be a concern to coral geneticists seeking to describe genetic diversity either within reefs or between reefs (that is, to estimate population genetic structure or gene flow). Moreover, for a number of realistic scenarios, they show that the power to detect small-scale positive spatial autocorrelation is quite low; thus, failure to detect a spatial autocorrelation pattern will not constitute strong evidence for the absence of such a pattern. These issues are tangibly illustrated using their own data (Gorospe and Karl 2013) where they had exhaustively genotyped all 2352 colonies of the “cauliflower” coral, Pocillopora damicornis (using 6 microsatellite loci), from a small patch reef in Kāne`ohe Bay, Hawai’i (Fig. 1). Simulations that vary the degree and proportions of clones within hypothetical populations augment these empirical observations to highlight the generality of the findings.

The results are sobering for any empirical geneticist reflecting upon either past work in corals or embarking upon future studies. First, sampling that is truly random is the most reliable: the more expedient strategies of saturation and nearest neighbor sampling yield diversity estimates with large variance and in some instances bias. Thus, semi-haphazard sampling that typifies many studies—collecting fragments from colonies here and there—is likely to yield biased or unreliable estimates of genetic diversity. Second, huge numbers of individuals need to be genotyped to produce accurate estimates, with lower bounds of 500 to estimate the true proportions of unique genotypes (G/N) and lower bounds of 1000 or more to detect within-reef spatial autocorrelation patterns.

Why do these results matter? Because corals are the foundational species of coral reefs and sensitive to changes in temperature, sediments, pH, etc., many studies seek to identify populations containing genetic variation that could buffer corals against such stressors or might even be candidate sources for assisted migration. Within-population diversity might also be used to infer populations that have been historically stable and environmental factors that promote stability and persistence. Such inferences based on empirical observations of genetic diversity will be unreliable if sampling lacks the power to estimate true population values. Furthermore, most metrics that describe population genetic structure rely on within-population diversity estimates. If diversity within populations is overestimated, the between-population diversity will be underestimated (and gene flow overestimated) and vice versa (Edwards and Beerli 2000). Thus, typical population genetic measures could be especially error prone (and perhaps biased) for corals. Therefore, statistical analyses and interpretation of coral population genetic diversity should, ideally, consider how spatial genetic structure might affect outcomes (and concomitantly implement appropriate field sampling methods including random sampling and large sample sizes).

Although the emphasis of Gorospe et al. (2015) is on corals, intriguingly the authors suggest that positive spatial autocorrelation at short distances might be common in marine organisms, albeit subtle and difficult to detect. Positive spatial autocorrelation at very small distances (<100 m) has been reported for a number of marine taxa, notably those low dispersal ability (e.g., seagrasses: Alberto et al. 2005; Migliaccio et al. 2005; Ruggiero et al. 2005, sponges: Calderon et al. 2007; Blanquer et al. 2009, bryozoans: Pemberton et al. 2007, and ascidians: Yund and O’Neil 2000; David et al. 2010), but whether positive spatial autocorrelation at relatively short distances is also common for taxa with extended larval durations is unknown. As demonstrated by all the aforementioned studies, incorporating spatially explicit approaches adds a valuable dimension to marine genetic research and can reveal subtle but important patterns of genetic diversity.

Fig. 1
figure 1

Reef 19 in Kane’ohe Bay, Hawai’i where Gorospe and Karl conducted the research. Photograph by Kelvin Gorospe

Full size image