In this article, we propose a novel difference image (DI) creation method for unsupervised change detection in multi-temporal multi-spectral remote-sensing images based on deep learning theory. First, we apply deep belief network to learn local and high-level features from the local neighbour of a given pixel in an unsupervised manner. Second, a back propagation algorithm is improved to build a DI based on selected training samples, which can highlight the difference on changed regions and suppress the false changes on unchanged regions. Finally, we get the change trajectory map using simple clustering analysis. The proposed scheme is tested on three remote-sensing data sets. Qualitative and quantitative evaluations show its superior performance compared to the traditional pixel-level and texture-level-based approaches.

Understanding forest cover changes is especially important in highly threatened and understudied tropical dry forest landscapes. This research uses Landsat images and a Random Forest classifier (RF) to map old-growth, secondary, and plantation forests and to evaluate changes in their coverage in Ecuador. We used 46 Landsat-derived predictors from the dry and wet seasons to map these forest types and to evaluate the importance of having seasonal variables in classifications. Initial RF models grouped old-growth and secondary forest as a single class because of a lack of secondary forest training data. The model accuracy was improved slightly from 92.8% for the wet season and 94.6% for the dry season to 95% overall by including variables from both seasons. Derived land cover maps indicate that the remaining forest in the landscape occurs mostly along the coastline in a matrix of pastureland, with less than 10% of the landscape covered by plantation forests. To obtain secondary forest training data and evaluate changes in forest cover, we conducted a change analysis between the 1990 and 2015 images. The results indicated that half of the forests present in 1990 were cleared during the 25-year study period and highlighted areas of forest regrowth. We used these areas to extract secondary forest training data and then re-classified the landscape with secondary forest as a class. Classification accuracies decreased with more forest classes, but having data from both seasons resulted in higher accuracy (87.9%) compared to having data from only the wet (85.8%) or dry (82.9%) seasons. The produced cover maps classified the majority of previously identified forest areas as secondary, but these areas likely correspond to forest regrowth and to degraded forests that structurally resemble secondary forests. Among the few areas classified as old-growth forests are known reserves. This research provides evidence of the importance of using bi-seasonal Landsat data to classify forest types and contributes to understanding changes in forest cover of tropical dry forests.

Despite high fragmentation and deforestation, little is known about wildlife species richness and occurrence probabilities in tropical dry forest (TDF) landscapes. To fill this gap in knowledge, we used a Sentinel-2-derived land-cover map, Normalized Difference Vegetation Index (NDVI) data and a multi-species occupancy model to correct for detectability to assess the effect of landscape characteristics on medium and large mammal occurrence and richness in three TDF areas that differ in disturbance and seasonality in Ecuador. We recorded 15 species of medium and large mammals, distributed in 12 families; 1 species is critically Endangered, and 2 are Near-Threatened. The results indicate that species occupancy is related to low forest cover and high vegetation seasonality (i.e., high difference in NDVI between the wet and dry seasons). We believe that the apparent negative effect of forest cover is an indicator of species tolerance for disturbance. The three sampling areas varied from 98% to 40% forest cover, yet species richness and occupancy were not significantly different among them. Vegetation seasonality indicates that more seasonal forests (i.e., those where most tree species lose their leaves during the dry season) tend to have higher mammal species occupancy compared to less seasonal, semi-deciduous forests. Overall, occupancy did not vary between the dry and wet seasons, but species-specific data indicate that some species exhibit higher occupancy during the wet season. This research offers a good understanding of mammal species’ responses to habitat disturbance and fragmentation in TDFs and provides insights to promote their conservation.

Tropical dry forests (TDF) are highly threatened ecosystems that are often fragmented due to land-cover change. Using plot inventories, we analyzed tree species diversity, community composition and aboveground biomass patterns across mature (MF) and secondary forests of about 25 years since cattle ranching ceased (SF), 10–20-year-old plantations (PL), and pastures in a TDF landscape in Ecuador. Tree diversity was highest in MF followed by SF, pastures and PL, but many endemic and endangered species occurred in both MF and SF, which demonstrates the importance of SF for species conservation. Stem density was higher in PL, followed by SF, MF and pastures. Community composition differed between MF and SF due to the presence of different specialist species. Some SF specialists also occurred in pastures, and all species found in pastures were also recorded in SF indicating a resemblance between these two land-cover types even after 25 years of succession. Aboveground biomass was highest in MF, but SF and Tectona grandis PL exhibited similar numbers followed by Schizolobium parahyba PL, Ochroma pyramidale PL and pastures. These findings indicate that although species-poor, some PL equal or surpass SF in aboveground biomass, which highlights the critical importance of incorporating biodiversity, among other ecosystem services, to carbon sequestration initiatives. This research contributes to understanding biodiversity conservation across a mosaic of land-cover types in a TDF landscape.