Preserving endangered animals from extinction is an overwhelming challenge for researchers throughout the globe. Monitoring wildlife in their natural habitat is crucial. This proposed work develops an automated system that can detect endangered animals in nature. It might be challenging to manually identify endangered animals since there are so many distinct species. For that, we constructed a custom dataset carefully consisting of 1,880 images representing 10 endangered animal species, sourced from multiple online platforms. The images performed preprocessing, including resizing, normalization, and data augmentation techniques such as random shear, zooming, and horizontal flipping to enhance model generalization. We assessed five deep learning(DL) models on this dataset: Custom Convolutional Neural Network (CNN), VGG16, InceptionV3, MobileNetV2, and DenseNet121. DenseNet121 showed the best results, achieving 99.67% accuracy on both the training and testing datasets, which means it generalizes better, learns faster, and avoids overfitting compared to the other models. Furthermore, to improve model transparency and interpretability, we used Explainable Artificial Intelligence(XAI) approaches, notably LIME, which highlighted the key image regions that contribute to the model's predictions. These findings demonstrate how well DenseNet121 can identify endangered animal species, offering a dependable tool to aid in conservation and wildlife monitoring.