Search

Camera traps have revolutionized wildlife resource surveys by enabling the acquisition of comprehensive ecosystem information. Camera traps usually produce massive images. Snapshot datasets captured in nature reserves proximate to human habitation often comprise a significant volume of human activity images containing humans or livestock. Manually identifying and labeling wildlife images from large-scale datasets is a labor-intensive task that necessitates a substantial number of professionals and incurs expensive personnel costs. By harnessing the capabilities of deep learning technology to automatically differentiate between wildlife and human activity images, ecologists can solely focus on manually labeling wildlife images that account for a minor fraction of the dataset. This strategic shift can significantly reduce personnel expenses and enhance work efficiency. Existing research usually treats human activity images as ordinary categories and utilizes species recognition methods to automatically identify and filter them. However, when human activity images overwhelmingly predominate the dataset, established species recognition methods are susceptible to misclassifying a substantial proportion of wildlife images as human activity images. This misclassification can potentially result in ecologists missing opportunities to discover or observe wildlife. To tackle this challenge, we proposed an ensemble learning method based on a conservative strategy and current mainstream deep learning frameworks to automatically identify wildlife and human activity images. We validated our method on a camera trap dataset from Lasha Mountain (LSM) in Yunnan, China. The experimental results demonstrated that our method automatically identified wildlife images from the dataset with accuracy, recall, and precision of 95.75%, 94.07%, and 83.89%, respectively. This led to an approximately 80% reduction in personnel costs. • A conservative ensemble strategy used to identify wildlife images is proposed. • An ensemble learning method based on the conservative strategy is proposed. • The method automatically separates wildlife images from human activity images. • The method saves 80% of the personnel costs with 94.05% of recall. • Convolutional neural networks are still the preferred species recognition method. [ABSTRACT FROM AUTHOR]

• Here we provide a detailed description of male birth attendance during a diurnal birth event in nonhuman primate. • The 2nd observed diurnal birth event in Rhinopithecus bieti. • We suggest that infant attraction or 'Aunt behavior' offers the strongest explanation of cases of birth assistance and attendance in nonhuman primates. Birth attendance, or midwifery service, is an important characteristic in human evolution, and has been argued to separate our lineage from other taxa in the animal kingdom. Recent studies, however, indicate that similar or analogous behaviors also may occur in a small number of nonhuman primate species. Here, we report the first case of both male and female attendance and female birth assistance in a wild species of nonhuman primate, the black-and-white snub-nosed monkey (Rhinopithecus bieti). At our field site in Yunnan, China we observed a diurnal birth event in which the leader male of a one-male unit (OMU) remained in close- proximity (0–2 m) to the parturient, groomed her, and remained vigilant over a five hour pre- and postpartum period. In addition, a multiparous female member of the OMU also remained in close proximity to the soon-to-be mother, helped to pull the neonate from the birth canal, took the neonate from the new mother within 15 s of the birth, held the infant for 20 min, and then severed the umbilical cord. For the next several days the leader male traveled in close-proximity to the new mother and four days after the birth event, we observed him to share food with her. Given that diurnal births are extremely rare in this primate species; it remains unclear the degree to which the events we observed commonly occur during nighttime births. We argue that adult male and female black-and-white snub-nosed monkeys are highly attracted to young infants, and birth attendance and birth assistance in this, and certain other primate species, may play a critical role in maternal and infant survivorship. [ABSTRACT FROM AUTHOR]

The Anle Zn–Pb deposit, hosted by Upper Cambrian dolostone, is located in the southern Songpan–Ganzi Block in southwest China. In this deposit, ore bodies occur as stratiform lenses and consist of galena, sphalerite and pyrite as ore minerals, and quartz, dolomite and calcite as gangue minerals. The mineralization shows mainly vein, banded and brecciated structures. Four ore bodies have been found in the Anle deposit, with a combined 2.0 million tonnes (Mt) of sulfide ores at average grades of 1.64 wt.% Pb, 6.64 wt.% Zn and 45 g/t Ag. Brown, brownish-yellow and yellow sphalerite samples have δ 66 Zn values ranging from + 0.08 to + 0.10‰ (average + 0.09‰, n = 3), + 0.12 to + 0.38‰ (average + 0.24‰, n = 8) and + 0.40 to + 0.50‰ (average + 0.46‰, n = 3), respectively. We interpret the progressively heavier Zn isotopes from brown to yellow sphalerite as being led by kinetic Raleigh fractional crystallization. Calcite samples have δ 13 C PDB and δ 18 O SMOW values ranging from − 4.8 to − 0.2‰ (average − 1.7‰, n = 7) and + 17.9 to + 21.4‰ (average + 19.6‰, n = 7), respectively. Whole-rock δ 13 C PDB and δ 18 O SMOW values of the Cambrian ore-hosting dolostone range from + 0.1 to + 1.1‰ (average + 0.6‰, n = 3) and + 23.2 to + 24.1‰ (average + 23.6‰, n = 3), respectively. This suggests that carbon in the ore-forming fluids was provided by the host dolostone through carbonate dissolution. δ 34 S CDT values of sulfide samples range between − 1.3‰ and + 17.8‰ with an average value of + 6.3‰ ( n = 25), lower than evaporites (such as barite + 19.8‰) in the overlaying Lower Ordovician sedimentary strata. The data suggest that sulfur in the hydrothermal fluids were derived from evaporites by thermo-chemical sulfate reduction (TSR). 206 Pb/ 204 Pb, 207 Pb/ 204 Pb and 208 Pb/ 204 Pb ratios for sulfide minerals range from 17.63 to 17.86, 15.58 to 15.69 and 37.62 to 37.95, respectively. The data are similar to those of the age-corrected Cambrian ore-hosting dolostone ( 206 Pb/ 204 Pb = 17.70–17.98, 207 Pb/ 204 Pb = 15.58–15.65 and 208 Pb/ 204 Pb = 37.67–38.06), but lower than those of age-corrected Ordovician sandstone and slate ( 206 Pb/ 204 Pb = 18.54–19.58, 207 Pb/ 204 Pb = 15.73–15.81 and 208 Pb/ 204 Pb = 38.44–39.60). This indicates that ore Pb was most likely to be derived from the Cambrian ore-hosting dolostone. Therefore, our new geological and isotopic evidence suggests that the Anle Zn–Pb deposit is best classified to be an epigenetic carbonate-hosted Mississippi Valley-type (MVT) deposit. [ABSTRACT FROM AUTHOR]

• A framework for integrating multi-source wildlife data for habitat mapping. • A case study of Rhinopithecus bieti habitat mapping was conducted in Yunnan, China. • Sightings from volunteer villagers and official patrol records were integrated. • Data integration improved habitat mapping accuracy. • The integration framework can support biodiversity monitoring and conservation. Wildlife habitat mapping is a widely used tool for supporting decision making in conservation. It requires data indicating wildlife habitat use to model and map habitat suitability. Collecting wildlife data, however, requires much effort, especially for species in remote mountainous regions of limited accessibility. Such circumstances often necessitate the integration of limited amounts of data available from multiple sources for habitat mapping. To that end, this study presents a framework for integrating multi-source wildlife data for habitat mapping. For evaluating the integration framework, a case study of mapping habitat suitability of the black-and-white snub-nosed monkey (Rhinopithecus bieti) by integrating sightings elicited from local volunteer villagers and obtained from official patrol records was conducted in Yunnan, China. The integration was explored at three levels: data-, knowledge- and model-level following disparate principles. The predicted habitat suitability maps were validated against monkey occurrence data independently collected though field-tracking. Results show the suitability maps predicted based on data integration were more accurate compared to maps predicted based on individual data sources. Data- and model-level integration achieved higher accuracy compared to knowledge-level integration. Further, data- and model-level integration following a conservative principle, i.e., the 'minimum' operator, led to higher mapping accuracy. The integration framework is generally applicable for integrating data from multiple sources for habitat mapping. It is also easy to implement and thus can be conveniently adopted by practitioners. Habitat suitability maps generated based on integrated data from multiple sources could better supporting decision making in biodiversity monitoring and conservation. [ABSTRACT FROM AUTHOR]