Purpose of the Project
On the morning of May 16, 2021, it was reported that a forest fire was caused near the Dujyuan campground at the Batongguan historic trail in Yushan National Park due to the careless use of fire by mountain climbers, which lasted for 12 days and was extinguished on May 27, 2021. The damaged area is about 22.1 hectares, and the area of extended burn is about 79.7 hectares. The damaged area was about 22.1 hectares, and the extended burn was about 79.7 hectares.
In order to understand the hazard of this forest fire, this project assesses the severity of the fire and the impact of the fire on the overall ecosystem by focusing on the fire sites and the surrounding vegetation. In addition, this project offers management strategies and long-term monitoring models for forests in the fire sites.

Research Methods
In this study, the fire-affected area of the Dujyuan campground was separated into the fire core, ecotone, and undisturbed areas(control area). The monitoring plots were established to conduct ground cover plant surveys, woody plant health surveys, automatic camera animal monitoring, bird sound recording monitoring, soil property analysis, and soil eDNA animal analysis. The historical fire events within the Yushan National Park were also compiled and discussed compared to related studies. The standardized vegetation indicators(NDVI) and standardized precipitation evapotranspiration indicators(SPEI) from satellite images were also used to examine the correlation between long-term vegetation changes and changes in water balance and fires.

Result and discussion
After compiling the reports and databases on fires in the Yushan National Park, the spatial distribution of fires in the past has shown that most of the fires occurred in the vicinity of people's activities, and the main cause of fires was manufactured.
The NDVI values were used to determine the areas affected by the fire within three months after the fire. Comparing the NDVI values of different years, we found that some areas with good vegetation cover had changed to sparse shrubs or grasses near no vegetation cover after the fire. The fire was mainly concentrated on the south-facing slope and spread from the fire site in the form of surface fire, while more serious damage was formed on the uphill side of the fire site, some steep terrain, or near the ridge. In the core area, where the fire damage was most severe, most trees were nearly blackened, and only a few taller Pinus taiwanensis survived. The tree crown was burned for an average of 83.5% in the major crown layer, while it was mostly up to 100% in the secondary crown layer, and ground plants were almost burned out. In the ecotone, most of the canopies were not directly affected by fire, and most of the Pinus taiwanensis had green leaves at the first survey in February. Although there were no apparent signs of crown fire, some branches and leaves in the canopy were killed by the high heat of the fire. At the same time, almost no herbaceous plants survived in the ground cover, and nearly all shrubs and small trees in the sub-canopy were dead due to surface fires. During the monitoring, we also found that the ground cover plant was gradually increasing, and the main species that were expanding were the Yushania niitakayamensis. In addition, a large number of woody plant seedlings were found in the ground cover, with a density of 12.51 individuals/m2, most of which were Pinus taiwanensis, and a certain number of Abies kawakamii and Tsuga chinensis var. formosana seedlings were also found. The number of naturally restored seedlings in the fire-affected areas was sufficient.
Soil analysis showed that only EC, pH, and effective phosphorus M3 differed significantly between the three type areas. The significant decrease in electrical conductivity in the core area and the lapse zone may be due to the lack of surface vegetation cover, resulting in soil salinity being affected by rainfall washout. Soil pH was strongly acidic in the undisturbed zone, slightly more weakly acidic in the core zone, and intermediate in the ecotone.
The animal survey was recorded for 11 months, 13 days, and 5 hours, totaling 8,237 hours. A total of 4,772 photographs were taken during the monitoring period, and 1,409 valid pictures were taken, including 406 individuals. A total of 16 species in 17 families and five orders were identified, including 12 species in 12 genera, nine families and four orders of mammals, and four species in five genera, three families, and one order of birds. The undisturbed area recorded the most species; the ecotone recorded was the second. The overall activity frequency of Taiwan water deer in both seasons was higher than other species in all environmental sample areas.
The taxonomic information identified from the e-DNA sequences of this project is too limited to allow for the standard categorization of ecological functional groups or functional traits. However, it is still helpful for understanding the differences in the composition of animals in the fire area and the surrounding environment. Concerning the taxa identified, the presence of Collembola, Trombidiformes, and Mesostigmata in the core area suggests signs of natural recovery in the core area after the fire. This project used soil eDNA to detect mammals in three types of areas. Although the number of species was small, three species with widely varying characteristics were detected: bats, rats, and water deer. It shows that soil eDNA analysis provides a relatively simple and generalized method for detecting mammalian groups that require diverse survey methods to be thoroughly investigated.
In the bird survey, a total of 44 species from 42 genera and 21 families were recorded in three surveys. Among them, 30 species were endemic, 14 species were native, and no exotic species were investigated. Eight species were classified as Near Thread (NT) or higher in the Red list.

Suggestion
This study showed that some of the plants in the fire-affected area are slowly recovering, and a large number of seedlings were found to be regenerating in the sample area. Still, some of the surviving trees were also found to have health problems and fungal invasion. Because of the large number of seedlings, it is recommended to continue the long-term monitoring of the forest regeneration status in the monitoring plots and observe the effect of fire on the surviving trees. Fire is not a single factor but requires a dry environment, fuel, and a fire source. We suggested that the concept of recreation and fire education be strengthened and that fire areas be well planned to facilitate the management of fire areas for the public. We also advised regularly monitoring and inspecting the fuel accumulation in the vicinity of people's activities and removing it as appropriate.