Our results show that the effects of the forest disturbance drivers on biodiversity are likely to be different from those simply expected from the baseline proportions of the forest disturbance drivers if we take into account the threatened species’ distributions. The amount of forest habitat is a primary factor for species diversity of many taxa, including mammals, amphibians, reptiles, birds, insects, and plants18. Indeed, our revealed that threatened forest species have been exposed to a disproportional decrease in their habitat amount globally (ie, lower proportions of forest with no or minor loss in all regions when species ranges were considered). Although this finding may be intuitive as population size and/or species range is part of the criteria in the IUCN assessment19, the detected pattern supports the validity of our approach of combining a forest disturbance map and species ranges for evaluating the impact of forest disturbances on threatened species. Moreover, we found that the dominant drivers differ among regions: the proportion of forestry, for example, increased in regions such as North America and Europe, whereas that of shifting agriculture increased in tropical regions when threatened species’ distributions were considered. These facts indicate although several international schemes for conservation have been implemented for regulating forestry20,21, different mechanisms aiming to directly tackle the over land use for local agriculture may increase their importance when we consider conservation in tropical regions. Our findings suggest that the social and economic drivers underlying the forest disturbance that impacts biodiversity differ among regions or nations, and it is important to establish specific conservation strategies in order to be effective.
Based on the findings, we further emphasize that the combinations of multiple interacting drivers are likely to vary among regions. For example, the frequency and extent of stand-replacing natural disturbances such as wildfires have clearly been magnified by climate change, particularly in the Northern Hemisphere (eg,22). After such natural disturbances, societal demand for timber and/or pest reduction compels forest managers to ‘salvage’ timber by logging before it deteriorates, a common practice even in locations otherwise exempt from conventional green-tree harvesting, such as national parks or wilderness areas23. Thus, salvage logging clearly mediates the interaction between disturbances by forestry and wildfires and is likely to further affect biodiversity under climate change. Especially in regions where infrastructure (eg, irrigation systems) has not been well developed, unpredictable changes in precipitation due to climate change was reported to increase forest disturbance by unregulated increases of agricultural land use24. Such regions largely overlapped with regions where shifting agriculture was identified as a dominant driver for threatened species in this study. Moreover, species themselves shift their ranges in response to climate change25, which would also shift major disturbance drivers and interactions of drivers to which species are exposed, given the region-specific driver patterns. These examples clearly suggest the necessity to understand both the region-specific interrelations among multiple drivers and species’ responses for better prediction of land-use change and thus its effects on biodiversity.
Shifting agriculture was the most dominant driver in all tropical regions corresponding to the recent estimates suggesting that the cover of regenerating secondary forest is increasing worldwide26. We demonstrated that this tendency is more drastic especially within the range of threatened species. The effect of shifting agriculture per unit area might be more limited than that of commodity-driven deforestation, which permanently alters forests into other land uses, since habitat structure might recover as the forest vegetation regenerates to a secondary state following the abandonment of the small clearings . However, ample evidence shows that many types of agricultural activities significantly degrade the conservation value of primary forest, especially in the tropics27which often recovers very slowly if ever28 with the loss of irreplaceable conservation values. Therefore, given the wide areas of dominance of shifting agriculture across all tropical regions, its effect is likely to be pervasive. Consistently, our results show that species extinction risk (ie, IUCN Red List status) is positively related to the proportional coverage of shifting agriculture (Fig. 2). In addition, as expected, a larger current proportion of shifting agriculture within a species ranges worsens the rate change in IUCN Red List status of the species (Fig. 4b). Furthermore, the effect is anticipated to be magnified for forest specialists because they are exposed to larger proportions of shifting agriculture than are forest generalist (Fig. 2), and they are also reported to recover more slowly than do forest generalists27,28.
A guideline for forest restoration suggested that appropriately sized landscapes should contain ≥40% forest cover (higher percentages are likely needed in the tropics), with about 10% in a very large forest patch and the remaining 30% in evenly dispersed smaller patches and semi-natural wooded elements (eg, vegetation corridors)29. Importantly, the guideline also suggests that the patches should be embedded in a high-quality matrix. Although younger secondary forest cannot be a substitute for pristine forest until 50 years or more after a disturbance, it can help to improve the quality of matrix in agricultural landscapes30. Indeed, we show that the negative impacts of shifting agriculture and forestry on IUCN status change have improved over time (Fig. 4b, c), presumably corresponding to the forest regenerating and recovery process. In contrast, the pattern of commodity-driven deforestation, a land use accompanied with permanent forest loss, showed a prolonged negative impact on IUCN status change (Fig. 4a). Notably, whether regenerating forests can move towards a highly diverse and structurally complex state or towards a state of low to intermediate levels of biodiversity and structural complexity depends on the amount of remaining intact mature forest in the landscape29. Therefore, a promising direction for future research would be to develop our analysis further to include spatiotemporal relationships among mature forest remnants, secondary forests, disturbance drivers, and threatened species populations.
For conserving the core patches of mature forests, the establishment of protected areas (PAs) is one of the most effective legal measures that has been widely used to regulate land use for biodiversity31. On the other hand, for improving matrix quality, balancing conservation and use of the ecosystem would be critically important; Shifting agriculture, for example, causes forest degradation, but it also contributes to food supply chains sourced from smallholder farmers and to food security of local communities8. In fact, establishing mechanisms for managing biodiversity-friendly landscapes has been intensively discussed recently, given the large potential influence of these landscapes on conservation32. These mechanisms include setting an international target on OECMs15th. Our finding of a disproportional decrease in forest proportions with minor or no loss within species ranges supports the urgency of the discussion. At the same time, our results highlight an opportunity because large portions of the disturbed forests for species are dominated by shifting agriculture at the global scale, especially in the tropics. As suggested above, if manged properly, such landscapes can still retain or improve functions as essential habitats and/or matrix for a variety of forest-dwelling species. Our analytical method provides a tool set to identify and prioritize areas where such attempts are urgently needed.
Global demands for natural resources and ecosystem services drive land use in forests33 and thus affect biodiversity. Therefore, connecting the supply chains to the five major drivers of forest disturbance and their spatial overlaps with biodiversity is essential to inform how we should regulate and design material flows from forest ecosystems to keep them sustainable by minimizing the effects on biodiversity. Existing studies examining the impacts of resource consumption on biodiversity through supply chains of various sectors have often been assessed at the country scale (eg,12), partly because the availability of statistics needed to estimate material flows in supply chains is usually limited at finer (ie, subnational) scales (but see34). We believe that our study provides the first basis for filling the resolution gap between trade statistics and local biodiversity effects by identifying patterns of the local co-occurrence of biodiversity and the forest disturbance drivers that can be directly linked to resource production at the national scale. Note, however, that downscaling a remotely sensed global data set into finer scales inevitably propagates errors and biases which include both those in the original maps and those in the processed data produced by analyses. Thus, preparation of more high-resolution data sets is essential, especially for disturbance drivers and threatened species’ distributions in our case, to keep the errors and biases at a reasonable level at focal spatial scales.
The effectiveness of area-based conservation measures to regulate land use for conservation including PAs and OECMs also strictly on social and ecosystem conditions. For example, a few studies show that the effectiveness of PAs in halting or slowing forest disturbances depends on PA characteristics such as size and history, as well as on the management entities such as subnational governments or indigenous peoples35,36,37. Moreover, there has been no attempt to elucidate whether PAs and OECMs are effective at regulating supply chains as a supply-side measure by balancing resource production, ecosystem services for local communities, and biodiversity conservation; to tackle this issue, it will be necessary to conduct extensive analyses integrating spatial and temporal patterns of biodiversity, forest loss, its drivers, and material flows in global food supply chains. Though it is challenging and beyond the scope of this paper, solving this issue is urgent and raises a promising opportunity for future research.