Seagrasses, corals, marsh plants, kelps, and mangroves support valuable coastal ecosystems but are threatened by environmental stressors. The need to manage these foundation taxa has spurred more than a decade of study on the relationship between genetic diversity and function or stability. We synthesized this literature base (129 relevant publications) and found more reported instances of neutral to positive relationships between genetic diversity and function than negative. However, much of the scientific understanding is based on the response of three genera and from unreplicated observational studies that correlate genetic diversity to measured response variables. When a disturbance was present, the studies often lacked controls or baseline data. Only 5.5% of the studies robustly tested for stability. These shortcomings preclude a rigorous evaluation of whether more genetically diverse foundation populations increase stability and hinder the use of genetics-based conservation strategies. Future studies should be focused on diverse species and ecosystem-level impacts using manipulative designs.

To survey for literature on the topic of genetic diversity and resilience of the five coastal foundation taxa, we performed a Web of Science search in December 2021 using the following search string:

(((ALL=(Diversity or Variation) and (genet* or genotyp* or allel*) and (stability or resilience or function or epifauna or epibio* or epiphyte or productivity or production or trophic) and (seagrass* or mangrove* or submerged aquatic vegetation or coral* or saltmarsh or marsh* or kelp*)))))

The initial search returned 1,533 publications from peer-reviewed journals, which were each screened for relevance and inclusion by a team of researchers. To be included in the extraction process, publications could not be a review or a meta-analysis and had to (1) focus on one of the focal foundation species, (2) contain a metric of genetic diversity (e.g., genotypic richness, allelic diversity), and (3) include a metric of function (e.g., survival, metabolism, production, or whole-ecosystem function) or the environment (e.g., temperature, salinity, irradiance). When a publication met the requirements of 1, 2, and 3, we noted whether the study design included replication at all levels (within and among diversity and disturbance treatments/observations). Furthermore, if these studies included any kind of disturbance, we noted whether the authors reported baseline information on function before the disturbance began and whether they included proper controls (i.e., plots/treatments that did not experience a disturbance). Finally, we also noted if these publications could be used to answer any of three predictions of the diversity-resilience relationship. The predictions are higher genetic diversity (1) is related to greater function, (2) helps maintain function under a disturbance, and (3) confers resilience to a disturbance. To test prediction 1, a study had to report both diversity and function. To test prediction 2, a study had to report genetic diversity and function under a disturbance; and to test prediction 3 a study had to follow a before-after-control-impact (BACI) design to truly assess resilience.

To minimize personal bias among researchers in the selection process, the team held a workshop where we evaluated the relevancy and inclusion of five publications based on the criteria described above. Once we all agreed on the inclusion or exclusion of those five publications, the 1,533 publications returned by the search were broken up into three sets of 511 publications. A pair of researchers were assigned to each set. Both researchers in each pair independently evaluated at least 140 publications in their assigned set for the inclusion criteria and prediction(s) the publication could test. The two researchers then compared their evaluations and discussed and resolved discrepancies. Upon reaching full agreement, one or both researchers would move on to finish the remainder of the set. In the case when an agreement could not be reached between the two researchers, the publication was reviewed by a larger committee and was included or not based on a majority consensus. Reviews and meta-analyses (107 of the 1,533) were not included in our dataset, but the literature cited was cross-referenced to find 20 additional relevant publications.

We extracted metadata from the 129 publications that met our inclusion criteria. Metadata on the study design and results of each publication were extracted by one of four reviewers. Information on the focal organism (taxon and species), the study type (observational or manipulative), location (field or laboratory), region (including GPS coordinates when provided), and study duration were extracted from each publication. When researchers collected from or conducted experiments at multiple sites, GPS coordinates for all sites were recorded. Publications that included metrics of coral Symbiodinium diversity were included into the “coral” category. A single coral species can be functionally different as a result of the type of symbionts, so Symbiodinium types A-I were for this purpose considered genetically different entities of coral.

We recorded the type of diversity metric used and the observed range of these values reported by each study. If the study manipulated levels of genetic diversity, we also noted the total number of levels in the study and the number of replicates used at each level. A list of all metrics used in each publication was compiled for each taxon (Table S1). When a paper reported multiple metrics, each metric was tallied separately to have a total >129 instances of diversity metrics. Values are reported as the percentage of the total number of metrics tallied rather than the total number of publications.

Measured diversity metrics were divided into eight different categories. “Genotypic Information” consisted of measurements such as calculated genotypic richness, number of genotypes, and evenness. The “Allelic Information” category included measurements such as the number of alleles, and calculations of allelic richness and allelic diversity. “Haplotype/Clade Metrics” included the number of clades, the number of ITS-2 (ribosomal internal transcribed spacer 2) types, and haplotype diversity. Other categories included “Observed Heterozygosity”, “Expected Heterozygosity”, “Inbreeding Coefficients” (F_IS and F_ST), “Nucleotide Specific Metrics”, and “Diversity Indices” (Simpson’s index, Shannon Wiener index, etc.).

We recorded any metrics of the environment reported in the study and observed ranges. Based on the results, we tallied publications that found a correlation between genetic diversity and one or more environmental metrics. The type and duration (in days) of any manipulated or natural disturbances that occurred during the study were recorded. In studies that conducted a manipulative disturbance, the number of levels of each stress treatment was recorded. In studies that experienced a natural or accidental disturbance, it was noted whether the stressor was continuous and ongoing after the study ended. 

We recorded the functional metric(s) used in the experiment and noted how many times throughout the experiment it was measured. We also noted whether there was a correlation (positive, neutral, or negative) between each diversity and function metric. In some cases, authors reported multiple function and diversity metrics in one paper and the correlation was dependent on the function or diversity metric reported. When positive and neutral outcomes were included in the same publication it was tallied as an overall “positive” outcome, because if one function metric is positive, the outcome for the population would also be positive. In publications that reported both positive and negative results, the responses were tallied in their own category. If all metrics reported in a publication were neutral, the outcome was reported as “neutral”. For example, in a study where genotypic richness is positively correlated with biomass but not correlated with leaf length, the publication would be tallied as positive. However, if genotypic richness was positively correlated with biomass, but negatively correlated with leaf length, the paper was tallied as “positive and negative”.

Extracted metrics of function were categorized into five categories: “Abundance”, “Physiology”, “Morphometric”, “Reproduction”, and “Community”. Abundance metrics consisted of measurements such as density and biomass. Physiology included metrics related to metabolism, growth rates, and productivity. Morphometric measurements included height, length, width, and surface area. Reproduction consisted of measurements such as flower and fruit production and seed count. Community metrics were those that related to any other organism beyond the focal species. For example, macrofauna abundance, levels of herbivory, and grazer and epiphyte biomass/diversity were all counted as community metrics. We also looked for metrics that measured overall ecosystem function such as nitrogen cycling or decomposition, but ultimately did not find any within the 129 relevant publications. The number of metrics in each category was tallied for each taxon. Results are reported as a percentage of instances in each category out of the total number reported for each taxon. 

We were also interested in the mechanism(s) behind the relationship between genetic diversity and function, which can be determined through some manipulative studies. These studies must have replication of genetic diversity treatments and genotypes within each treatment. Manipulative studies with proper replication were further investigated for any conclusions made about mechanisms behind the diversity-function relationship (functional redundancy, niche complementarity, etc.).