Animal Molecular Breeding 2024, Vol.14, No.1, 130-140 http://animalscipublisher.com/index.php/amb 134 effective data sharing between genomic and remote sensing platforms (Bourlat et al., 2013; Stephenson, 2019). Additionally, the high cost and technical complexity of genomic technologies can be a barrier to their widespread adoption in wildlife monitoring (Carroll et al., 2018). Another challenge is the need for interdisciplinary collaboration and training to bridge the gap between genomics and remote sensing experts. Many researchers may lack the expertise required to use these computationally intensive methodologies, highlighting the importance of workshops and training programs to build capacity in this area (Fitak et al., 2019). Furthermore, the integration of these technologies requires addressing issues related to data quality and resolution. For example, genomic data from minimally invasive sampling methods often yield low-quality DNA, which can limit the type of molecular methods used. Developing robust protocols and error-correction techniques is essential to overcome these limitations (Carroll et al., 2018). In conclusion, while the integration of genomics and remote sensing technologies holds great promise for wildlife monitoring, addressing the methodological, technological, and collaborative challenges is crucial for realizing their full potential in conservation efforts. 4 Advances in Data Analysis and Modeling 4.1 Bioinformatics and spatial analysis techniques The integration of bioinformatics and spatial analysis techniques has significantly advanced the field of wildlife monitoring. Genomic data, which have become increasingly affordable and accessible, are now being utilized to provide precise estimates of wildlife population features such as effective population size, inbreeding, demographic history, and population structure. These estimates are critical for conservation efforts, as they help in understanding the genetic health and adaptive capacity of wildlife populations (Hohenlohe et al., 2020; Schmidt et al., 2023). Moreover, the development of standardized methods for assessing genetic variation and inbreeding, as well as identifying genetic interchange patterns between populations, has been emphasized. These methods require robust bioinformatic support to handle the complex data and analyses involved (Schmidt et al., 2023). The application of these genomic tools in conservation biology is still evolving, but the potential for these techniques to inform and improve conservation strategies is substantial (Hohenlohe et al., 2020; Schmidt et al., 2023). 4.2 Predictive modeling and its applications Predictive modeling, particularly through the use of species distribution models (SDMs), has seen significant advancements with the incorporation of remote sensing technologies. Traditional SDMs often faced limitations due to spatial biases in occurrence data and a lack of spatially explicit predictor variables. However, modern remote sensing technologies, including multispectral and hyperspectral sensors, LiDAR, and RADAR, are revolutionizing the way habitat characteristics are captured and integrated into these models (He et al., 2015). These advanced sensors, deployed on satellites, planes, and unmanned aerial vehicles, provide high-resolution data that enhance the accuracy and predictive power of SDMs. This allows for better detection and monitoring of both plant and animal species across various ecosystems (Kerr and Ostrovsky, 2003; He et al., 2015). Additionally, the integration of machine learning algorithms with remote sensing data has enabled automated species identification, habitat mapping, and population monitoring (Figure 2), further enhancing the effectiveness of predictive models in wildlife conservation (Drakshayini et al., 2023). The research results of Toro et al. (2023) evaluate the suitability of different remote sensing methods for classifying land use and land cover (LULC) in integrated crop-livestock systems (ICLS). Two study sites (SS1 and SS2) are analyzed using Sentinel-2 data, with different algorithm and time window specifications. In the SS1 map, the classification accuracy for various LULC classes such as eucalyptus, native forest, and pasture shows a high degree of precision and recall, indicating robust model performance. The confusion matrix for SS1 reveals minor misclassifications between classes, with overall high F1-scores, especially for pasture and eucalyptus. The SS2 map demonstrates the capability of Sentinel-2 data in a more heterogeneous landscape. Despite the complexity,
RkJQdWJsaXNoZXIy MjQ4ODY0NQ==