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

This pattern of selection likely resulted from the fact that there were many cells where shrubs grew, but the distribution of snow meadow, fellfield, and wilderness was more limited (Fig 3 and Table 2). Although the target vegetation types overlapped between tourism use and biodiversity conservation, SecSel selected separate sites for these objectives in accordance with the settings when there were conflicts between them. Sites where the area of each type of vege- tation was large tended to be selected for conservation, and sites where the total area of the three types of vegetation was large tended to be selected for tourism in accord with the defini- tion of the values of each site for each feature. When consideration was given to cost (cost-on in SecSel), the cost function worked as expected, and the selected sites concentrated close to the trail entrances (Fig 5B). When the boundary length was taken into consideration (boundary-on in SecSel), the frequency of selec- tion of sites in the central part of the Taisetsu Mountains decreased (Fig 5C) relative to the fre- quency with the cost-off and boundary-off settings (Fig 5A). The highest-quality sites for snow fields and fellfields are in the northern part, and the highest-quality wilderness sites are in the southern part. The northern and southern sites were thus selected in the early steps of the pri- oritization, and the sites subsequently selected were chosen because of their spatial cohesion around the previously selected sites. The resultant length of the boundary of the selected con-

servation area averaged 66.8 after 100 runs during which consideration was given to a bound- ary length penalty. If no consideration was given to the boundary penalty, the lengths of the boundary averaged 110.7 and 97.7 for the cost-off (Fig 5A) and cost-on (Fig 5B) settings, respectively. The boundary length of the selected sites was thus substantially decreased when consideration was given to the boundary-length penalty, as expected. When consideration was given to the nearest-distance penalty (distance-on), there was an increase in the frequency with which sites in the central part of the Taisetsu Mountains were selected (Fig 5D), in contrast to the results with the boundary-on setting. Selected sites must adjoin each other to reduce the boundary length, and thus the boundary penalty hindered selection of sites in the central area, which is separate from the northern and southern parts, where there are larger numbers of high-quality sites. The nearest-distance penalty required that a newly selected site be close to already-selected sites, but it was not necessary that the site be adjoint. Furthermore, even larger numbers of sites were selected in the central part to avoid isolation of them and thereby decrease the nearest distance to other sites. The resultant dis- tances to the nearest site averaged 1.99 and 12.90 with and without consideration of the near- est-distance penalty, respectively. Examination of the site-selection pattern for each run revealed that a small number of sites

in the central part were consistently selected on all the example runs when the SecSel settings were cost-off, boundary-off, and distance-off (the first column on the left in S1 Fig). In contrast, the number of sites selected in the central part varied greatly between runs when the settings were boundary-on or distance-on (the third and fourth columns from the left in S1 Fig). This difference was caused by the case-dependency of the boundary length and nearest-distance pen- alties. If a site was selected by chance in the central part in an early step of the site-selection pro- cess, sites close to the site were selected to avoid isolation of the site and thereby decrease the Table 2. Number of cells in each rank for each type of vegetation and ecosystem service. Rank Snow meadow Fellfield Wilderness Shrub Tourism resource 0 201 239 256 44 213 1 83 36 35 97 53 2 17 22 13 92 35 3 17 21 14 85 17 https://doi.org/10.1371/journal.pone.0247737.t002 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 13 / 19