Mechanistic Artificial Intelligence Model on Effects of Climate Change on Plants Diversity and Land Cover for Sustainable Livelihood

Artificial Intelligence (AI) has significant potential to assess, predict, and mitigate the risks associated with climate change. Counties in the North Rift region of Kenya, vital for food security and pastoral livelihoods, have been severely affected by climate change. However, previous studies have not utilized AI technologies to address these challenges. The purpose of the study was therefore to develop a mechanistic AI model that assess the effects of climate change on terrestrial plant diversity and land cover for sustainable livelihoods. The study aimed to analyze climatic trends in the North Rift region over a 30-year period using logistic regression, evaluate climate change impacts on plant diversity through clustering techniques, assess land cover changes with predictive random forest techniques and develop an AI-driven mechanistic model. The study was guided by the Research Science Design, considering climatic variables such as precipitation, temperature, land cover, species diversity, and richness in developing the model. The target population included eight counties in the North Rift region, with cluster sampling based on land-use practices, and snowballing for gathering indigenous knowledge. Data was collected from satellite imagery and use of guided survey questions. Data analysis used descriptive, regression, inferential, predictive and mechanistic techniques. The instruments average reliability score was (.78) based on Cronbach’s Alpha, mean rating of (4.50) and standard deviation of (.452). This study developed an AI model to accurately predict plant diversity and richness, offering insights for policy formulation in climate change mitigation and guiding future research in the field. Findings reveal a consistent increase in both maximum (3.6°C) and minimum (2.4°C) temperatures, along with a rise in annual precipitation from 800 mm to 1260 mm, highlighting significant climatic shifts. Logistic regression analysis demonstrated a correlation between rising temperatures and precipitation variability, impacting agriculture, water availability, and biodiversity. Clustering techniques indicated wetter regions maintained higher plant diversity, while drier zones, like Turkana, faced reduced species richness due to temperature increases and altered precipitation patterns. The study also found that forested areas expanded with increased rainfall, while grasslands and shrublands deteriorated, especially in arid areas. Random forest models predicted land cover changes with 95.67% accuracy, indicating vegetation density improvements in wetter zones and degradation in drier areas. The study developed a mechanistic AI model and converted it into mathematical representation that successfully predicted biodiversity shifts. The model was tested to predict climatic variables, biodiversity and land cover changes from 2021 to 2080 at 10 year intervals. From the 2021 baseline of 25.2°C (regional average), temperatures are projected to increase steadily, reaching 26.0°C by 2030, 27.1°C by 2040 and 28.0°C by 2050. By 2080, when regional temperatures reach 29.5°C (a 4.3°C increase from 2021), the impact on livestock productivity and human habitability in these arid zones will be severe. Precipitation pattern indicates severe drought conditions returning in 2050 at 1,078mm, a 25% decrease from 2021 and again in 2070 at 955mm. These drought periods will be interspersed with intense wet periods, including 2040 at 1,403mm and culminating in 2080 with precipitation reaching 1,628mm, representing a 34% increase from the 2021 baseline. Regional biodiversity decreases steadily from 69.4% in 2021 to 58.8% by 2080. Forest cover projections show a persistent decline from 59.6% in 2021 to 50.5% by 2080. The study concludes that climate change poses significant challenges to biodiversity and ecosystem services, emphasizing the need for adaptive strategies, policy development, and ecosystem restoration efforts. It also recommends expanding predictive modeling to include socio-economic factors for more holistic planning.