Table 1.
Definitions of key terms used related to scale in conservation planning.
Fig 1.
Regional context and enlarged maps of the two study regions: (A) Micronesia and (B) Fiji. Buffers are shown around Micronesian nations to increase visibility of the numerous small coral islands and atolls.
Fig 2.
Study design showing tested factors, factor levels, and the 20 unique combinations between all levels.
Table 2.
Coding system to identify individual scenarios.
Codes are assigned to each level of each prioritisation factor. An example scenario code based on this coding system is “L1U”: large planning units (L), first level of thematic resolution (1), and uniform cost (U).
Fig 3.
Example maps of planning-unit sizes and thematic resolutions explored in the Fiji dataset.
All maps represent the same spatial extent and location; grey polygons represent Fiji terrestrial areas (islands). (a) Planning-unit sizes: ‘small’ (1 km2; blue squares) and ‘large’ (25 km2; red squares); 25 small planning units are nested within each large, non-edge planning unit. Note that both planning-unit grids were clipped to all reef areas, resulting in irregular planning units on the perimeters. (b) & (c) Examples of two of the five levels of thematic resolution: (b) level 2 (11 total reef classes in Fiji), and (c) level 4 (43 total reef classes in Fiji).
Fig 4.
Example map showing distribution of cost variability across the Fiji planning region.
Values, shown here for large planning units, are based on distance to fisher populations as a proxy for opportunity cost.
Table 3.
Summary of output comparisons and statistical analyses for each research objective.
Fig 5.
Example of Marxan output data.
For calculation of dissimilarity, we regarded Marxan individual solutions as analogous to biological sampling sites and planning units as analogous to recorded species. For single Marxan solutions, planning units were either selected (“present”) or unselected (“absent”). For selection frequencies across 100 replicate solutions in a scenario, entries for planning units were equivalent to species abundance data.
Fig 6.
The flow of analyses related to spatial nestedness.
Analyses described for spatial nestedness, defined here as extent of overlap of high-priority areas based on large planning units (left) with small planning units with high thematic resolution and both cost layers (right). Dotted arrow lines indicate repetition of the same analyses across all other coarse-scenarios performed with both of the test scenarios, S5U and S5V.
Table 4.
Method calculating the expected areas of level 5 reef classes that would be selected for reservation in each of the scenarios using large planning units.
Fig 7.
Comparisons of total reserve size and proportions of maximum possible cost.
(a) Boxplots of ranges of reserve solution sizes for each scenario based on 100 replicate runs. (b) Boxplots of ranges of total costs (expressed as proportions of maximum possible cost) for each scenario based on 100 replicate runs. Each change in shade of the same colour represents the change in thematic resolution (always presented in order from level 1–5, left to right) for each combination of planning-unit size and cost variability. Colour scheme representing all scenarios remains the same throughout all figures to facilitate interpretation.
Fig 8.
Spatial dissimilarity between all 2000 solutions for the Fiji case study.
Fig 9.
Comparison of spatial variation between all solutions produced using RDA for the Fiji case study.
Cost variability mainly explains variation along RDA1, while variation along RDA2 is mostly represented by different planning-unit sizes. Red squares are centroids of the different prioritisation-factor levels, representing the average amount of spatial variance that lines up with the plotted axes.
Table 5.
Results of the permutation test showing significance of each of the tested prioritisation factors in influencing the spatial dissimilarity of solutions.
Fig 10.
Nestedness of high-priority small planning units (test scenarios) within high-priority areas defined by large planning units.
Nestedness of S5U high-priority areas, defined at: (a) selection frequency ≥ 50, and (b) selection frequency ≥ 75. Nestedness of S5V high-priority areas, defined at: (c) selection frequency ≥ 50, and (d) and selection frequency ≥ 75.
Fig 11.
Incidental representation of level 5 reef classes by scenarios using large planning units.
(a-b) Scatter plots showing expected representation of each level 5 reef class (as a percentage of total area of reef class occurrence) for each coarse scenario with (a) uniform cost and (b) variable cost, in relation to reef class rarity (transformed to natural log). Due to spread and left-skewness of rarity values, plots are shown with x-axis breaks where no data occur to facilitate interpretation. Local regression (LOESS) curves were fitted for each coarse scenario, indicating non-linear trends in each scatter plot. Dashed horizontal lines represent the 30% objective for level 5 reef classes. (c-d) Histograms showing the distributions of expected representation of level 5 classes for coarse scenarios with (c) uniform cost and (d) variable cost, plotted with 5% bin widths. Dashed vertical lines represent the 30% objective for level 5 classes.