AI-Driven Model Enhances Predictions of Wildlife Population Dynamics

Researchers at IIT Mandi have developed an innovative AI-based mathematical model that predicts wildlife population changes, enhancing understanding of ecosystem dynamics. This model, which analyzes historical data, can forecast trends for up to five years, focusing on the interactions between wolves and moose in Isle Royale National Park. The study highlights the importance of AI in ecological research, especially in scenarios with incomplete data. With findings that have attracted international attention, this research could significantly impact wildlife conservation strategies. Read on to learn more about this groundbreaking development.
 | 
gyanhigyan

Innovative Framework for Ecosystem Conservation


Munish Sood
MANDI:
A groundbreaking advancement in scientific research has emerged from the Indian Institute of Technology (IIT) Mandi, where scientists have created an artificial intelligence (AI)-based mathematical model designed to forecast wildlife population changes. This model aims to enhance the understanding of ecosystem dynamics and bolster wildlife conservation efforts.


This novel framework leverages decades of historical wildlife data to predict variations in predator and prey populations, even when the data is sparse, inconsistent, or uncertain. It can project population trends up to five years into the future, providing researchers with a valuable tool to analyze how shifts in one species can affect the survival and growth of another.


The study specifically examines the interactions between wolf and moose populations in Isle Royale National Park, a 544-square-kilometer island located in Lake Superior, Michigan, USA. Wolves serve as the primary natural predators of moose in this area, with moose constituting nearly 90% of the wolves' winter diet. The populations of these species have been tracked annually since 1959, making it one of the longest-running studies of predator-prey dynamics globally.


Historical data indicates significant fluctuations in both populations. For example, wolf numbers experienced a dramatic decline in the early 1980s due to a disease outbreak, while the moose population saw a substantial drop in the mid-1990s. These changes underscore how disruptions in predator-prey relationships can impact the overall balance and health of ecosystems.


The research was conducted by Dr. Nitu Kumari, a professor at IIT Mandi’s School of Mathematical and Statistical Sciences, alongside her PhD student Anurag Singh.


The team developed a self-adaptive modeling framework that evaluates historical population data to estimate moose reproduction rates, natural fluctuations in wolf populations, and the effects of predation on prey numbers. This model integrates physics-informed neural networks with transfer learning, allowing it to discern significant patterns from complex ecological datasets.


During their investigation of two different methods for understanding wolf predation on moose, the researchers discovered that the model based on the relationship between predator and prey population sizes yielded superior results compared to the alternative method. It also effectively pinpointed periods of unusual population declines in historical data.


For the period from 2020 to 2024, the model forecasts an increase in wolf numbers alongside a decrease in the moose population, which aligns with the trends observed in the national park.


Prof. Kumari noted that, to her knowledge, this study is the first to utilize a self-adaptive mathematical modeling framework that combines physics-informed neural networks and transfer learning to explore fluctuations in predator-prey populations.


The results highlight the increasing potential of AI-enhanced mathematical modeling in ecological studies, especially in scenarios where long-term environmental data is lacking or uncertain. Such predictive tools could significantly aid scientists in comprehending wildlife population dynamics and evaluating potential changes in vulnerable ecosystems.


The research has garnered international interest, with Prof. Kumari recently participating in an interview with a US-based news outlet. An article detailing the research is anticipated to be published online by the outlet soon.