Convolutional Long Short-Term Memory Neural Networks for Hierarchical Species Prediction Nithish B Moudhgalya 1 , Sharan Sundar S 1 , Siddharth Divi 1 , P Mirunalini 1 , Chandrabose Aravindan 1 , S. M. Jaisakthi 2 Department of Computer Science and Engineering, SSN College of Engineering, Kalavakkam, Chennai, India 1 School of Computer Science and Engineering, VIT University, Vellore, India 2 {nithish15066,sharansundar15096,siddarth15130}@cse.ssn.edu.in {miruna,aravindanc}@ssn.edu.in,jaisakthi.murugaiyan@vit.ac.in Abstract. Building accurate knowledge of the identity, the geographic distribution and the evolution of organisms is essential for biodiversity conservation. Automatic prediction of list of species is useful for many scenarios in biodiversity informatics. In this work, we propose a hybrid model to predict the species that are most probable to be observed at a given location, using environmental features and taxonomy of the or- ganism. These environmental features are represented as k-dimensional image patches, where each dimension represents the value of an environ- mental variable, in the neighborhood of the occurrence of the species. The hybrid model Convolutional Long Short-Term Memory Neural Net- works henceforth called as CLNN, is a combination of Convolutional Neu- ral Networks(CNNs) and Long Short-Term Memory Networks(LSTMs), where the CNN forms the spatial feature generator while the LSTM focuses on finding the taxonomy. Using the dataset provided by Geo LifeCLEF 2018, the proposed method helped achieve a Mean Reciprocal Rank (MRR) score of 0.003 during the test phase. Keywords: Niche modeling · Hierarchical embedding · Taxonomic pre- diction · CLNN 1 Introduction Environmental niche models have been used by biologists and environmentalists to understand the species distribution in geographic space. These models help reduce resources expended in data collection and analysis, thus giving space for research in analyzing impacts of global phenomenon like climate change, habitat loss, species invasion and evolutionary trends that could help in translocation of species. Considering the overwhelming uses of species prediction modeling, CLEF or- ganizers posed the Geo LifeCLEF 2018 challenge [5]. The aim of the challenge is to develop a location-based species recommendation system using image-based