Reef Adapt: A tool to inform climate-smart marine restoration and management decisions

Abstract

A critical component of ecosystem restoration projects involves using genetic data to select source material that will enhance success under current and future climates. However, the complexity and expense of applying genetic data is a barrier to its use outside of specialised scientific contexts. To help overcome this barrier, we developed Reef Adapt ( www.reefadapt.org ), an innovative, globally applicable and expandable web platform that incorporates genetic, biophysical and environmental prediction data into marine restoration and assisted gene flow planning. The Reef Adapt tool provides maps that identify areas with populations suited to user-specified restoration/recipient sites under current and future climate scenarios. We demonstrate its versatility and practicality with four case studies of ecologically and evolutionarily diverse taxa: the habitat-forming corals Pocillopora damicornis and Acropora kenti, and macroalgae Phyllospora comosa and Ecklonia radiata. Reef Adapt is a management-ready tool to aid restoration and conservation efforts amidst ongoing habitat degradation and climate change.

Fig. 1

Conceptual overview of the Reef Adapt platform.

Fig. 2. Map showing the scale of data in the Reef Adapt webtool using existing publicly available data.

Blue points represent 420 sampling locations for 27 species of macroalgae and coral. Species highlighted in our case studies are shown clockwise from left (image credits in brackets): Pocillopora damicornis (Karen Filbee-Dexter), Acropora kenti (John Edmondson), Phyllospora comosa (Leah Wood) and Ecklonia radiata (John Turnbull).

Fig. 3. GDM and Reef Adapt output for P. damicornis restoration scenario.

A Turnover of genetic (neutral and adaptive) variation as a function of environmental and geographic variables. The shape of each function indicates how the rate of change in allele frequencies varies along the gradient. Points are site pairs, the line is the predicted relationship between predicted and observed genetic and ecological (or climatic) distance. Panels on the left represent model-fitted I-splines for each GDM model, showing predicted genetic distance/change against each of the biophysical variables included in the final model dataset. The distribution of raw data points from each covariate is indicated via rugplot on the x-axis, with the exception of the oceanographic distance unit, which is calculated internally from the NMDS coordinates during model construction. ±Standard error (in green) generated from 999 bootstrap iterations with 10% of the populations removed. B Reef Adapt webtool predictions showing suitable areas to source seedstock for restoration for P. damicornis. Species distribution highlighted in transparent white; ‘local’ areas with predicted genetic differentiation of up to F_ST 0.05 from the restoration site (red) are highlighted in green.

Fig. 4. GDM and Reef Adapt output for A. kenti restoration scenario.

A Turnover of genetic (neutral and adaptive) variation as a function of environmental and geographic variables. B Reef Adapt webtool predictions showing suitable areas to source seedstock for restoration for A. kenti using a genetic differentiation threshold value of up to F_ST 0.01 (green). Further details are presented in the Fig. 3 caption for brevity.

Fig. 5. GDM and Reef Adapt output for P. comosa restoration scenario.

A Turnover of genetic (neutral and adaptive) variation as a function of environmental and geographic variables. B Reef Adapt webtool predictions showing suitable areas to source seedstock for restoration for P. comosa (green). Further details are presented in the Fig. 3 caption for brevity.

Fig. 6. GDM and Reef Adapt output for E. radiata assisted gene flow provenancing scenario.

A Turnover of genome-wide genetic variation as a function of environmental variables, as predicted by a generalised dissimilarity model. B Reef Adapt webtool predictions showing suitable areas to source material for assisted gene flow strategies for E. radiata. Species distribution is highlighted in transparent white; areas with predicted genetic composition suitable for 2050 conditions in the focal site (red) are highlighted in red. Further details are presented in the Fig. 3 caption for brevity.

Fig. 7. Survey gap analysis output for P. damicornis.

Sample sites that contributed to the generalised dissimilarity model (GDM) are shown in red. Areas with high similarity to sampled areas are more transparent, whilst darker cell areas indicate survey gaps and uncertainty in the GDM model’s predictive power.