A Robust Prediction Model for Species Distribution Using Bagging Ensembles with Deep Neural Networks

Multispecies deep learning using citizen science data produces more informative plant community models

In the age of big data, scientific progress is fundamentally limited by our capacity to extract critical information. Here, we map fine-grained spatiotemporal distributions for thousands of species, using deep neural networks (DNNs) and ubiquitous citizen science data. Based on 6.7 M observations, we jointly model the distributions of 2477 plant species and species aggregates across Switzerland with an ensemble of DNNs built with different cost functions. We find that, compared to commonly-used approaches, multispecies DNNs predict species distributions and especially community composition more accurately. Moreover, their design allows investigation of understudied aspects of ecology. Including seasonal variations of observation probability explicitly allows approximating flowering phenology; reweighting predictions to mirror cover-abundance allows mapping potentially canopy-dominant tree species nationwide; and projecting DNNs into the future allows assessing how distributions, phenology, and dominance may change. Given their skill and their versatility, multispecies DNNs can refine our understanding of the distribution of plants and well-sampled taxa in general.

Assessment of agricultural prospects in relation to land use change and population pressure on a spatiotemporal framework

The urbanisation process moves quickly in emerging nations like India and Bangladesh, transforming natural landscapes into unsustainable landscapes. Consequently, growing development has had a significant impact on agricultural land as a natural environment. Moreover, there is a scarcity of research on fragmentation probability modelling in the extant literature. Thus, by combining random forest (RF) and bagging with the datasets which are multi-temporal in a GIS framework, the probability of fragmentation of LULC at Jangipur subdivision in India and Bangladesh can be modelled. Parallelepiped, Mohalnobis distance, support vector machines (SVM), spectral angle mapper (SAM), and artificial neural networks (ANN) classifiers were used for LULC classification, where SVM (Kappa coefficient: 0.87) surpassed other classifiers. The LULC maps for 1990, 2000, 2010, and 2020 were created using the best classifier (SVM). During this time, the built-up area grew from 23.769 to 158.125 km². Then, using an ANN-based cellular automata model, the future LULC map for 2030 was predicted (CA-ANN). In 2030, the built-up area would be 201.58 km². Then the matrices of class and landscape were taken out of the LULC maps utilising FRAGSTAT software and included the patch number (NP), largest patch index (LPI), edge density (ED), contagion index (percentage) (CONTAG), perimeter and area (P/A), aggregation index (AI), landscape percentage (PLAND), the area of class (CA), patch density (PD), edge in total (TE), total core area (TCA), and largest shape index (LSI). The validation results revealed that bagging (0.915 = AUC) and RF (0.874 = AUC) are capable of assessing fragmentation probability, with the bagging model having the greatest precision level of the two. Almost 20% of the total LULC was in a high and very high zone of fragmentation vulnerability, necessitating the use of direct measures to safeguard it. As a result, adequate LULC management is required.

Dynamic Forecast of Desert Locust Presence Using Machine Learning with a Multivariate Time Lag Sliding Window Technique

Desert locust plagues can easily cause a regional food crisis and thus affect social stability. Preventive control of the disaster highlights the early detection of hopper gregarization before they form devastating swarms. However, the response of hopper band emergence to environmental fluctuation exhibits a time lag. To realize the dynamic forecast of band occurrence with optimal temporal predictors, we proposed an SVM-based model with a temporal sliding window technique by coupling multisource time-series imagery with historical locust ground survey observations from between 2000–2020. The sliding window method was based on a lagging variable importance ranking used to analyze the temporal organization of environmental indicators in band-forming sequences and eventually facilitate the early prediction of band emergence. Statistical results show that hopper bands are more likely to occur within 41–64 days after increased rainfall; soil moisture dynamics increasing by approximately 0.05 m³/m³ then decreasing may enhance the chance of observing bands after 73–80 days. While sparse vegetation areas with NDVI increasing from 0.18 to 0.25 tend to witness bands after 17–40 days. The forecast model combining the optimal time lags of these dynamic indicators with other static indicators allows for a 16-day extended outlook of band presence in Somalia, Ethiopia, and Kenya. Monthly predictions from February to December 2020 display an overall accuracy of 77.46%, with an average ROC-AUC of 0.767 and a mean F-score close to 0.772. The multivariate forecast framework based on the lagging effect can realize the early warning of band presence in different spatiotemporal scenarios, supporting early decisions and response strategies for desert locust preventive management.