Camera traps often produce massive images, and empty images that do not contain animals are usually overwhelming. Deep learning is a machine‐learning algorithm and widely used to identify empty camera trap images automatically. Existing methods with high accuracy are based on millions of training samples (images) and require a lot of time and personnel costs to label the training samples manually. Reducing the number of training samples can save the cost of manually labeling images. However, the deep learning models based on a small dataset produce a large omission error of animal images that many animal images tend to be identified as empty images, which may lead to loss of the opportunities of discovering and observing species. Therefore, it is still a challenge to build the DCNN model with small errors on a small dataset. Using deep convolutional neural networks and a small‐size dataset, we proposed an ensemble learning approach based on conservative strategies to identify and remove empty images automatically. Furthermore, we proposed three automatic identifying schemes of empty images for users who accept different omission errors of animal images. Our experimental results showed that these three schemes automatically identified and removed 50.78%, 58.48%, and 77.51% of the empty images in the dataset when the omission errors were 0.70%, 1.13%, and 2.54%, respectively. The analysis showed that using our scheme to automatically identify empty images did not omit species information. It only slightly changed the frequency of species occurrence. When only a small dataset was available, our approach provided an alternative to users to automatically identify and remove empty images, which can significantly reduce the time and personnel costs required to manually remove empty images. The cost savings were comparable to the percentage of empty images removed by models.

The elevational range where montane species live is a key factor of spatial niche partitioning, because the limits of such ranges are influenced by interspecies interaction, abiotic stress, and dispersal barriers. At the regional scale, unimodal distributions of single species along the elevation gradient have often been reported, while discontinuous patterns, such as bimodal distributions, and potential ecological implications have been rarely discussed. Here, we used extensive camera trap records to reveal the elevation distribution of Himalaya blue sheep (Pseudois nayaur) and its co-existence with other ground animal communities along a slope of Baima Snow Mountain, southwest China. The results show that Himalaya blue sheep exhibited a distinctive bimodal distribution along the elevation gradient contrasting the unimodal distributions found for the other ungulates in Baima snow mountain. A first distributional peak was represented by a population habituating in scree habitat around 4100 m, and a second peak was found in the dry-hot valley around 2600 m. The two distinct populations co-existed with disparate animal communities and these assemblages were similar both in the dry and rainy seasons. The extremely low abundance of blue sheep observed in the densely forested belt at mid-elevation indicates that vegetation rather than temperature is responsible for such segregation. The low-elevation population relied highly on Opuntia ficus-indica, an invasive cactus species that colonized the region six hundred years ago, as food resource. Being the only animal that developed a strategy to feed on this spiky plant, we suggest invasive species may have formed new foraging niche to support blue sheep population in lower elevation hot-dry river valleys, resulting in the geographic separation from the original population and a potential morphological differentiation, as recorded. These findings emphasize the important conservation values of role of ecological functions to identify different taxa, and conservation values of apparent similar species of different ecological functions.

The human gut microbiota is closely associated with human health, influencing not only overall well-being but also the incidence and treatment outcomes of diseases. Altitudinal gradients are considered to impact gut microbial community characteristics through factors such as environmental temperature, humidity, and lifestyle. While previous studies have reported altitudinal variations in human gut microbiota in specific regions, a comprehensive exploration of these patterns at a global scale is still lacking. In this study, we analyzed 16S rRNA amplicon sequencing data from healthy human gut microbiota, spanning altitudes from 3 m to 3850 m, obtained from multiple open-access databases. The analysis focused on elucidating the altitudinal patterns of microbial diversity, community composition, and functional profiles. After screening, a total of 6702 sequences from 15 countries were obtained. The diversity of human gut microbiota decreased with increasing altitude (R = -0.047, P < 0.001), but no consistent results were acquired among continents. The relative abundances of the genera Faecalibacterium and Blautia decreased with rising altitude (R = -0.131 and R = -0.135, respectively, P < 0.001 for both), while the relative abundance of the genus Prevotella increased with altitude (R = 0.336, P < 0.001). However, taxa such as Bacilliota, Bacteroides, and Bifidobacterium exhibit no consistent trends across different continents. The abundance of genes associated with the metabolism of terpenoids and polyketides, lipid metabolism, neurodegenerative diseases, and aging increased with altitude (R = 0.146, 0.037, 0.366, and 0.317, respectively; lipid metabolism P = 0.003, others P < 0.001). Conversely, the abundance of genes related to the immune system and carbohydrate metabolism decreased with increasing altitude (R = -0.166 and R = -0.219, respectively; P < 0.001 for both). Altitude significantly influences diversity, composition, and functional attributes of the human gut microbiota.

Given the widespread conflicts between humans and birds in contexts such as agricultural production, airports, and new energy installations, the development and application of bird deterrence technologies hold significant importance for both safeguarding human interests and advancing the cause of wildlife conservation. Artificial sounds are one of the most widely used methods for bird deterrence; however, there is a lack of systematic research on the effects of sound frequency and their combination. The black-necked crane (Grus nigricollis), a first-class nationally protected wild animal of China, shares some of its habitats with human agricultural activities. This study designed sounds of various frequencies and combinations and conducted deterrence experiments on black-necked cranes at their wintering sites in northeastern Yunnan Province, exploring how sound frequency and combinations affect the behavioral responses, escape distances, and escape speeds of the cranes. Our results show that: (1) in terms of behavioral response intensity, escape distances, and escape speed, sounds with high frequency and high variability significantly outperformed other sounds (n = 479, p < 0.001); (2) there were no significant differences in the responses of black-necked cranes of different flock types and age combinations to sounds. The study recommends using high-frequency, highly variable sounds for short-term control of black-necked cranes. Additionally, this research demonstrates that using either high-frequency or high variable sound combinations can achieve efficient bird deterrence in the short term, and provides a scientific basis for developing and refining bird deterrence strategies for other bird species.

Wildlife camera trap (CT) surveys typically employ two-dimensional equal-area grid sampling, which often neglects the influence of complex mountainous terrain on species distribution, potentially yielding misleading outcomes. A watershed, incorporating diverse habitats from high to low elevations and from rivers to ridges, aligns with complex mountains. Monitoring based on watersheds might address this. In southwest China’s mountain forests, under comparable sampling intensities, we contrasted the capture rate (CR), species richness, and relative abundance index (RAI) of dominant species among watershed, 1 × 1 km² grid, and elevation gradient patterns. Also, habitat factor correlations and heterogeneities were analyzed. Results reveal higher CR, species richness, and habitat heterogeneity in the watershed pattern. The elevation gradient pattern shows more stable species and RAI than the grid pattern. In small-scale mountains, topographic factors indirectly affect CT survey results via vegetation distribution. Analysis of similarities (ANOSIM) indicates significant differences in species and community among watersheds. Using watersheds as sampling units for CTs can match the mountains’ elevation differences and complex topography well, aids in capturing wildlife diversity and understanding mountain species distribution. Therefore, we recommend that the spatial sample design in mountainous areas should be based on watersheds, taking elevation gradients and topography into consideration.

The effect of founders (the potential influence of initially colonizing species on the composition, functionality, and stability of communities) plays a crucial role in community assembly; many experimental studies on priority effects or artificially assembled species have suggested the existence of this effect, but direct experimental evidence at the community level remains limited. This study used sterilized and nonsterilized paocai soup (a traditional Chinese fermented vegetable soup) from the same source to simulate initial environments with and without founders. These were placed in beakers with varying opening sizes on an open rooftop for 15 days to explore the impact of founders on community assembly under different dispersal intensities. The 16S rRNA sequencing analysis revealed that communities with founders exhibited lower species richness (320) compared to communities without founders (645). Additionally, communities with founders showed reduced species turnover and richness variation (53.7%) compared to communities without founders (60.9%). Furthermore, the average variability degree (AVD) in communities with founders (0.446 ± 0.044) was significantly lower than in communities without founders (0.927 ± 0.466), indicating higher community stability. Finally, deterministic processes dominated communities with founders (with heterogeneous selection contributing 70%), whereas stochastic processes primarily governed communities without founders (homogeneous dispersal 10% and undominated processes 70%). These findings demonstrate that founders presence reduces dispersal impacts, decreases community diversity, enhances stability, and deterministic processes. The effect of founders fundamentally shapes the direction of community assembly. This study helps further understanding of how founders influence biodiversity maintenance and community assembly processes.