SecSel, a new software tool for conservation prioritization that is applicable to ordinal-scale data for multiple biodiversity features

boundary length or the nearest-distance to the site. If no site was selected by chance in the cen- tral part in the earlier steps, almost no site was selected in the later steps either. If we used continuous values for site quality, a very limited number of sites was repeat- edly selected because SecSel selected sites with the highest quality in order, and there was almost no flexibility when we directly used the area of each type of vegetation. For example, exactly eight sites (the same number as the conservation target) were selected for shrubs. The small flexibility for other types of vegetation was due to the flexibility in the process of resolving conflict between tourism and conservation. When there are conflicts with other conservation objectives, the priority of objectives may change at each iteration. The use of ranks as a measure of site quality increases the flexibility of prioritization by ignoring small differences in areas. Discussion Assessment of SecSel In a review of conservation planning tools, Sarkar et al. (2006) [3] have proposed six criteria for the assessment of the tools. Here, we use those criteria to evaluate SecSel. Spatial economy: Planning tools designed to select sites should either minimize the cost of sites or maximize the representation of features within cost constraints. SecSel tries to find a minimum set of sites that meet the given multiple targets. If the cost of including each site is given, suppression of the total cost is considered. The algorithm used by SecSel is a

greedy algorithm, which is heuristic and does not necessarily find the global minimum solu- tion. However, the difference between the solutions of the greedy algorithm and the global minimum was not large in most cases [17]. Because each site is added in a step-by-step fashion to find the local optimum, the tentative set of selected sites, which expands during the process of site selection, partly fulfills the given target. The process helps determination of how large the selected area should be. Computational efficiency: Planning tools should resolve data sets rapidly, particularly if multiple scenarios must be evaluated and stakeholders are involved in real-time negotia- tions. The greedy algorithm allows SecSel to find a solution rapidly. For example, selection from 3000 sites to satisfy the top-3 of 150 features is repeated 100 within 10 seconds on a desk- top PC. In the case of the Daisetsuzan National Park, it took no more than a few seconds for 100 iterations. The short time required for SecSel to find a solution allows SecSel to be used on-site for discussions in real time. Flexibility: Planning tools should allow the incorporation of a wide variety of criteria. The data used to evaluate features to be conserved/used need to be on an ordinal scale, but not necessarily on an interval or ratio scale. SecSel can handle semiquantitative data such as expert judgements. The resolution for the ordinal scale can be set arbitrarily. Furthermore, cost data do not need to be on an interval or ratio scale. The small constraints imposed by the data require-

ments provide high flexibility that is a distinct advantage of SecSel. In the example of Daisetsuzan National Park, we showed that SecSel could easily treat the course-time from the nearest trail entrance. That metric has qualitatively different units than the vegetation area metric. Transparency: It should be clear why each site is selected. The logic for site selection in SecSel is straightforward. In every step, it is clear why the newly added site is selected: it makes the largest contribution to the achievement of targets. It is clear which of the features within the site contribute to the achievement of targets. This transparency is an advantage of the heu- ristic step-by-step approach. Furthermore, the transparency is also apparent in the process of resolving conflicts of features within a site. The logic of the resolution of conflicts is simple: local units that are more dispensable to the achievement of the feature are given lower priority. PLOS ONE SecSel, a new software tool for conservation prioritization applicable to ordinal-scale data PLOS ONE | https://doi.org/10.1371/journal.pone.0247737 July 23, 2021 15 / 19