Evaluating the protection effectiveness of natural protected areas is an important step in successful management. Adopting 330 natural protected areas on the Qinghai–Tibet Plateau as research subjects, the regional dominant ecosystem service function was selected, and various temporal and spatial analysis methods were employed to analyze the evolution characteristics and influencing factors of ecosystem service patterns from 2000 to 2020. Our results indicated that (1) the water conservation function stabilized after fluctuation and decline, the soil conservation function fluctuated upward, and the windbreak and sand fixation function exhibited an increase after a decreasing fluctuation. (2) The protection effectiveness of25 protected areas significantly improved, that of 151 protected areas improved, that of 84 protected areas stabilized, that of 56 protected areas worsened, and that of 14 protected areas significantly worsened. (3) The top three influencing factors in descending order were precipitation change > altitude > mining area density. (4) Remarkable protection results were achieved in national protected areas, established management institutions, earlier established areas (before 2000), and areas exhibiting alow built-up area density (<0.75%) and low mining density (<1%). Our study provides technical support for the construction and management of protected areas and improvement in ecosystem service functions on the Qinghai–Tibet Plateau.

The expansion of linear infrastructure presents a significant threat to biodiversity, emphasizing the urgent need for regional studies on spatial variations and comprehensive multispecies research. This study surveyed and monitored wildlife passages across eight nature reserves in Xinjiang, evaluating their construction status, utilization patterns, and key factors influencing utilization rates, as well as providing optimization recommendations. The findings revealed that dedicated wildlife passages were scarce, especially in smaller reserves, which primarily relied on small bridges and culverts originally designed for water flow. Enhancing the construction of passages in these areas is strongly recommended. A total of 32 wildlife species were recorded, comprising 13 bird species and 19 mammal species. Ungulates, including bharal (Pseudois nayaur), goitered gazelle (Gazella subgutturosa), Asian wild ass (Equus hemionus), and wild Bactrian camel (Camelus ferus), exhibited high relative abundance indices, indicating a strong preference for utilizing passages. Passage utilization rates exhibited significant seasonal and diurnal variations. Winter usage was the lowest, followed by an increase in spring, peaking in summer and autumn. Birds exhibited particularly high passage utilization rates during migratory seasons, with the peak occurring in April. Additionally, nocturnal passage utilization rates were significantly higher than during other periods, with both sunrise and sunset showing positive selection, as indicated by 86.86% and 91.30% of monitored sites recording JSI > 0. To enhance the effectiveness of passages, seasonal and diurnal variations should be fully considered in passage construction and management, particularly by minimizing human activity during nighttime, dawn, and dusk. Utilization rates were significantly constrained by the intensity of human activity and the density of linear infrastructure. Passage type, size, and proximity to water were also critical factors. Optimizing passage layout, implementing dynamic management in grassland fence areas, increasing passage density, and enhancing ecological functionality are recommended strategies to facilitate wildlife movement and support biodiversity conservation.

The reintroduction of rare and endangered species is one of the most important approaches to conservation and ecosystem restoration, but it has still proven to be an adventurous undertaking and most reintroduction programmes fail, so successful demonstrations are needed. Père David's deer (Elaphurus davidianus, Milu in Chinese) could be considered one of conservation's great success stories, as this species' path on the road to extinction has been reversed by a combination of ex-situ conservation and successful re-introduction in China. The species had been consigned to an imperial hunting ground when the last Chinese herds were exterminated during the fall of the Qing Dynasty (1644–1912). Fortunately, a few of the last remaining individuals were sent to European zoos. From these animals, a herd was bred on the 14th Duke of Bedford's estate, Woburn Abbey, and between 1985 and 1987, and 38 Milu were donated back to China for re-introduction in Beijing Milu Park (BMP), the former imperial hunting ground. An additional 39 deer were released at Dafeng National Nature Reserve (DFNNR), Jiangsu Province in 1986. In both of these safe and protected locations, the Milu thrived allowing for over 700 Milu to be sent to a further 82 sites throughout the species' original Chinese range over the last 36 years. As a result, the Milu population totaled 9136 by 2021, with 2855 individuals now living back in the wild; while another 5681 individuals inhabit the DFNNR, and 186 reside in BMP. Wild Milu, however, still face significant conservation challenges. The population lacks genetic diversity, leading to severe inbreeding depression and carrying multiple risks, such as high miscarriage rates, a reduced lifespan, and susceptibility to disease. Environmental constraints such as pollution and habitat fragmentation further result in small, fragmented wild populations. Moreover, the species currently lacks a national level conservation master plan, the associated coordinated monitoring platforms, and breeding plans for China's captive populations. Finally, there is now a lack of international cooperation in the conservation of this species. We therefore call for both a national-level conservation master plan in China and international cooperation to develop a shared database and germplasm databank covering Milu across all countries with ex-situ populations, as crucial steps for securing the long-term conservation of Milu and preventing it from ever becoming “extinct in the wild” again.

We leveraged deep-learning-enabled autonomous acoustic recorders and high-definition cameras to quantify how climate, upper-level winds, and vegetation cover influence first arrival dates (FADs) of birds across the subtropical-temperate transition zone of China. From December 2021 to June 2025, we conducted continuous monitoring at 81 wetland and forest stations spanning 22-45° N and established an end-to-end workflow covering data acquisition, quality control, modeling, and validation. In total, we extracted 5959 FADs encompassing 618 resident and migratory species along the East Asian-Australasian Flyway. We linked FADs to daily mean temperature, daily cumulative precipitation, tailwind anomaly, crosswind strength, and vegetation anomaly. Using covariates averaged over a backward 30-day window, we fitted linear mixed-effects models (LMMs) and generalized additive mixed models (GAMMs) with species and region as random effects. Results were consistent across model classes: warming is associated with delayed arrivals-each +1 °C in 30-day mean temperature corresponds to a ∼2.58-3.60-day delay in FAD. Precipitation and stronger crosswinds likewise tend to delay arrival, whereas higher vegetation anomalies and favorable tailwinds advance arrival. Collectively, our findings clarify how climate, upper-level wind fields, and vegetation jointly shape avian phenology across continental scales.

Fine-grained spatio-temporal action detection in continuous, unconstrained field videos remains a formidable challenge due to severe background clutter, high inter-class similarity, and the scarcity of domain-specific benchmarks. To address these limitations, we first introduce a large-scale Wintering-Crane Benchmark, providing dense, individual-level bounding box annotations for six complex behaviors across diverse habitat scenes. Leveraging this data, we propose AviaTAD-LGH, a real-time multi-task framework that incorporates auxiliary motion supervision into a dual-pathway 3D backbone to enhance feature discriminability. A critical bottleneck in such multi-task settings is the negative transfer caused by conflicting optimization objectives. To resolve this, we present Lightweight Gradient Harmonization (LGH), a plug-and-play optimization strategy that dynamically modulates task weights based on the cosine similarity of gradient directions. This mechanism effectively aligns optimization trajectories without introducing inference latency. Extensive experiments demonstrate that AviaTAD-LGH achieves a state-of-the-art mAP of 68.60%, surpassing strong public baselines by 7.44% and improving upon the single-task baseline by 2.80%, with significant gains observed on ambiguous dynamic classes. The proposed pipeline enables efficient, scalable ecological monitoring suitable for edge deployment.