Hive behaviour assessment through vector autoregressive model by a smart apiculture system in the Mediterranean area

Abstract

Precision beekeeping is defined as an apiary management strategy based on monitoring individual bee colonies to minimize resource consumption and maximize bee productivity. This subject has met with a growing interest from researchers in recent years because of its environmental implications. Today, the use of new monitoring technologies and management systems are facilitating the beekeeper's task by reducing operating costs and increasing animal welfare. Few studies in the literature apply forecasting models that could be useful as decision support to help beekeepers effectively monitor their hives. The Vector Autoregressive Regression (VAR) models are widely used in economics, but little applications have been performed in precision beekeeping data. The aim of this study was to apply a Vector Autoregressive Model to study the interrelations among internal factors (weight, internal temperature, internal relative humidity, sound pressure level) and between internal and external environmental parameters (external temperature and relative humidity, rain, wind speed, UV index) of some hives located in three different sites in Sicily (south Italy), monitored by a proper designed smart system. Time series were studied over the period April - August 2023. The significance recorded in the relationships between weight of the hive and its internal temperature and weight of the hive and its internal relative humidity, and the good predictive capacity of the models with respect to internal temperature and internal relative humidity, allowed to build a predictive model to understand when possibly intervene on the hives. Effect and duration of a system shock on the variables of interest were effectively monitored by the impulse response function in order to understand the level of the system response.