Improving the prediction of wildfire susceptibility on Hawai'i Island, Hawai'i, using explainable hybrid machine learning models.

This study presents a comparative analysis of four Machine Learning (ML) models used to map wildfire susceptibility on Hawai'i Island, Hawai'i. Extreme Gradient Boosting (XGBoost) combined with three meta-heuristic algorithms - Whale Optimization (WOA), Black Widow Optimization (BWO), and Butterfly Optimization (BOA) - were employed to map areas susceptible to wildfire. To generate a wildfire inventory, 1408 wildfire points were identified within the study area from 2004 to 2022. The four ML models (XGBoost, WOA-XGBoost, BWO-XGBoost, and BOA-XGBoost) were run using 14 wildfire-conditioning factors categorized into four main groups: topographical, meteorological, vegetation, and anthropogenic. Six performance metrics - sensitivity, specificity, positive predictive values, negative predictive values, the Area Under the receiver operating characteristic Curve (AUC), and the average precision (AP) of Precision-Recall Curves (PRCs) - were used to compare the predictive performance of the ML models. The SHapley Additive exPlanations (SHAP) framework was also used to interpret the importance values of the 14 influential variables for the modeling of wildfire on Hawai'i Island using the four models. The results of the wildfire modeling indicated that all four models performed well, with the BWO-XGBoost model exhibiting a slightly higher prediction performance (AUC = 0.9269), followed by WOA-XGBoost (AUC = 0.9253), BOA-XGBoost (AUC = 0.9232), and XGBoost (AUC = 0.9164). SHAP analysis revealed that the distance from a road, annual temperature, and elevation were the most influential factors. The wildfire susceptibility maps generated in this study can be used by local authorities for wildfire management and fire suppression activity.

Population collapse of a Gondwanan conifer follows the loss of Indigenous fire regimes in a northern Australian savanna.

Colonialism has disrupted Indigenous socioecological systems around the globe, including those supported by intentional landscape burning. Because most disruptions happened centuries ago, our understanding of Indigenous fire management is largely inferential and open to debate. Here, we investigate the ecological consequences of the loss of traditional Aboriginal fire management on fire-exposed savannas on the Arnhem Plateau, northern Australia, using the fire-sensitive conifer Callitris intratropica as a bio-indicator. We contrast Kakadu National Park, where traditional Aboriginal fire management was severely disrupted during the early twentieth century following Aboriginal relocation to surrounding settlements, and an adjacent Aboriginal estate where traditional Aboriginal fire management endures. Since 2006, traditional Aboriginal fire management at this site has been overlaid by a program of broad-scale institutionalized burning in the early dry season, designed to reduce greenhouse emissions. Using remote sensing, field survey, and dendrochronology, we show that on the Aboriginal estate, C. intratropica populations depend on the creation of a shifting patch mosaic of long unburned areas necessary for the recruitment of C. intratropica. However, the imposition of broad-scale fire management is disrupting this population patch dynamic. In Kakadu, there have been extreme declines of C. intratropica associated with widespread fires since the mid twentieth century and consequent proliferation of grass fuels. Fire management in Kakadu since 2007, designed to increase the size and abundance of patches of unburned vegetation, has not been able to reverse the population collapse of C. intratropica. Our study demonstrates that colonial processes including relocation of Indigenous people and institutional fire management can have deleterious consequences that are nearly irreversible because of hysteresis in C. intratropica population dynamics.

Traditional lifestyles, transition, and implications for healthy aging: An Example from the remote island of Pohnpei, Micronesia.

Lifestyle-related, non-communicable diseases, such as diabetes, hypertension, and obesity have become critical concerns in the Pacific islands of Micronesia. We investigated the relationship between the diminution of traditional lifestyle practices and the decline in the health of the population in the State of Pohnpei, Federated States of Micronesia. To assess this, our interdisciplinary team developed two scales, one to rank individuals on how traditional their lifestyles were and one to rank individuals on the healthiness of their lifestyles. Participants' locations were categorized as living on a remote atoll, living on the main island, or as a transitional population. Pohnpeians living in transitional communities (e.g. recently moved from a remote atoll to the main island, or the reverse) ranked lowest on both the tradition and health scales, rather than ranking intermediate between the remote and main island groups as we had hypothesized. As predicted, individuals residing on the remote atolls were living the most traditional lifestyles and also had the healthiest lifestyles, based on our rating system. The higher an individual scored on the tradition scale, e.g. the more traditional life they lived, the higher they scored on the health scale, suggesting the importance of traditional lifestyle practices for maintaining health. These findings have significant implications for promoting health and longevity of Micronesians and other Pacific Island peoples. We suggest the process of transition be recognized as a significant lifestyle and health risk and be given the attention we give to other risk factors that negatively influence our health. Based on our findings, we discuss and recommend the revitalization of particular traditional lifestyle practices, which may advance healthy aging among Pohnpeians.

Vegetation-Rainfall interactions reveal how climate variability and climate change alter spatial patterns of wildland fire probability on Big Island, Hawaii.

The area burned annually by wildland fire in Hawaii has increased fourfold in recent decades. The archipelago's novel fuel types and climatic heterogeneity pose significant challenges for fire risk assessment and fire management. Probability-based fire occurrence models using historical wildfire records provide a means to assess and attribute fire risk in regions of the world like Hawaii where investment in fire science is limited. This research used generalized additive models to 1) assess the relative contribution of vegetation, climate, and human-caused ignitions to the probability of fire in the northwest quadrant of Hawaii Island and 2) compare how landscape flammability varies due to interannual rainfall variability versus projected changes in mean annual rainfall (MAR) and temperature. Annual fire probability was highest for grasslands and peaked at drier conditions (0.04 at 450 mm MAR) when compared with shrublands (0.03 at 650 mm MAR) and forest (0.015 at 600 mm MAR). Excess rainfall the year prior to fire occurrence increased fire risk across grasslands, and thus overall fire probability, more so than drought the year that fire occurred. Drying and warming trends for the region under projected climate change increased maximum values of fire probability by as much as 375% and shifted areas of peak landscape flammability to higher elevation. Model predictions under future climate also indicate the largest changes in landscape flammability will happen by mid-Century. The influence of antecedent wet years on fire risk can improve near-term predictions of fire risk in Hawaii while climate projections indicate that fire management will need to be prioritized at upper elevations where high value natural resources are concentrated.

Local and global pyrogeographic evidence that indigenous fire management creates pyrodiversity.

Despite the challenges wildland fire poses to contemporary resource management, many fire-prone ecosystems have adapted over centuries to millennia to intentional landscape burning by people to maintain resources. We combine fieldwork, modeling, and a literature survey to examine the extent and mechanism by which anthropogenic burning alters the spatial grain of habitat mosaics in fire-prone ecosystems. We survey the distribution of Callitris intratropica, a conifer requiring long fire-free intervals for establishment, as an indicator of long-unburned habitat availability under Aboriginal burning in the savannas of Arnhem Land. We then use cellular automata to simulate the effects of burning identical proportions of the landscape under different fire sizes on the emergent patterns of habitat heterogeneity. Finally, we examine the global extent of intentional burning and diversity of objectives using the scientific literature. The current distribution of Callitris across multiple field sites suggested long-unburnt patches are common and occur at fine scales (<0.5 ha), while modeling revealed smaller, patchy disturbances maximize patch age diversity, creating a favorable habitat matrix for Callitris. The literature search provided evidence for intentional landscape burning across multiple ecosystems on six continents, with the number of identified objectives ranging from two to thirteen per study. The fieldwork and modeling results imply that the occurrence of long-unburnt habitat in fire-prone ecosystems may be an emergent property of patch scaling under fire regimes dominated by smaller fires. These findings provide a model for understanding how anthropogenic burning alters spatial and temporal aspects of habitat heterogeneity, which, as the literature survey strongly suggests, warrant consideration across a diversity of geographies and cultures. Our results clarify how traditional fire management shapes fire-prone ecosystems, which despite diverse objectives, has allowed human societies to cope with fire as a recurrent disturbance.

Cultural legacies, fire ecology, and environmental change in the Stone Country of Arnhem Land and Kakadu National Park, Australia.

We use the fire ecology and biogeographical patterns of Callitris intratropica, a fire-sensitive conifer, and the Asian water buffalo (Bubalus bubalis), an introduced mega-herbivore, to examine the hypothesis that the continuation of Aboriginal burning and cultural integration of buffalo contribute to greater savanna heterogeneity and diversity in central Arnhem Land (CAL) than Kakadu National Park (KNP). The 'Stone Country' of the Arnhem Plateau, extending from KNP to CAL, is a globally renowned social-ecological system, managed for millennia by Bininj-Kunwok Aboriginal clans. Regional species declines have been attributed to the cessation of patchy burning by Aborigines. Whereas the KNP Stone Country is a modern wilderness, managed through prescribed burning and buffalo eradication, CAL remains a stronghold for Aboriginal management where buffalo have been culturally integrated. We surveyed the plant community and the presence of buffalo tracks among intact and fire-damaged C. intratropica groves and the savanna matrix in KNP and CAL. Aerial surveys of C. intratropica grove condition were used to examine the composition of savanna vegetation across the Stone Country. The plant community in intact C. intratropica groves had higher stem counts of shrubs and small trees and higher proportions of fire-sensitive plant species than degraded groves and the savanna matrix. A higher proportion of intact C. intratropica groves in CAL therefore indicated greater gamma diversity and habitat heterogeneity than the KNP Stone Country. Interactions among buffalo, fire, and C. intratropica suggested that buffalo also contributed to these patterns. Our results suggest linkages between ecological and cultural integrity at broad spatial scales across a complex landscape. Buffalo may provide a tool for mitigating destructive fires; however, their interactions require further study. Sustainability in the Stone Country depends upon adaptive management that rehabilitates the coupling of indigenous culture, disturbance, and natural resources.