PORTRAYAL OF LIKELY CLIMATE CHANGE IMPACTS ON BEEKEEPING BASED ON EARTH OBSERVATION DATA

Introduction

Bees (Apis), as one of the most important pollinators in nature, are of significant economic and ecological value, contributing a lot in both global food safety and the maintenance of rich biodiversity. According to the report from the International Union for Conservation of Nature (IUCN), 78% floral resource and 84% crops in the European Union rely partially or fully on animal pollination, of which bees account for the majority. However, many scientific researches and international associations have reported a concerning reduction in the health of bee population. For instance, over 9% of European bee species are reported by the IUCN to face the risk of extinction (Nieto, 2014), and multiple surveys in Europe (Bommarco et al., 2012; Rasmont et al., 2015) have revealed decline in distribution and population of many bee species over the course of this century (Cameron & Sadd, 2020). All these findings point to the urgency and necessity to study the reasons behind the bee crisis. Fundamental studies dedicated to exploring the causes of death and effects of human-induced environmental change, will be required to understand the reasons favoring bee population reduction.

The causes for the decline mentioned previously can be habitat loss, pesticide use, invasive species, climate change, etc. (Brown & Paxton, 2009). And among all these factors, this study focuses on evaluating the patterns of climate change as a factor that might affect bee abundance, since climate change has become a growing concern on a planetary scale. The change of climate conditions can affect the survival of bees in many ways: it can influence the behavior and physiology of bees directly, have an indirect impact by changing the phenology of floral resource, or affect the disease profiles and incidence of bees (Le Conte & Navajas, 2008).

Although there have been quite a few studies about the influence of climate change on beekeeing, e.g., Schmickl et al. (2017) analyzed how climate conditions like temperature and precipitation influence bee mortality and Flores et al. (2019) evaluated the effect of climate change on bee colonies. Most of these studies are carried out at a local scale based on conventional data collection methods, which is an enormous endeavor with inherent limitations, considering the fact that the bee crisis is happening on a global scale. In fact, the advancement of earth observation could fill this research gap effortlessly due to its nature of global coverage and massive data volume, whereas the application of earth observation data in the context of likely climate change and beekeeping support stays fairly understudied. The reason could be that many bee experts specialize in entomology or biology, and have very little knowledge of the latest advances in remote sensing science. With these problems being stated, this diploma thesis aims to provide new insights into how earth observation data can be used to benefit the current situation of bees, to give bee experts an overview of the methodologies that can be applied to analyze earth observation datasets related to bee health, and to reveal the possibilities of breaking through the limitations of traditional research in terms of spatial and temporal scale.

Reference

Methodology

Study Area

The study area for the evaluation of climate change is the entire European continent, which is part of the Eurasian continent and located to the west of Asia, with its west, north and south sides bordered by the Atlantic Ocean, the Arctic Ocean and the Mediterranean Sea respectively. Most parts of Europe are located in the mid-latitudes of the north temperate zone and belong to one the belts of westerly winds in the atmospheric circulation system. Therefore, the climate there is heavily influenced by the Atlantic Ocean, dominated by a temperate oceanic climate which is mild and humid throughout the year. And the study area for the assessment of climate change impacts on beekeeping the Czech Republic (highlighted in the map), which is located in central Europe.

Workflow

The methodology of this study can basically be explained by the workflow figure below. First and foremost, datasets with variables that have a relatively notable impact on the living environment of bees need to be sifted out from countless earth observation datasets and then fed into the web application for beekeeping support. These filtered datasets are then visualized in an interactive way in the web application, allowing the beekeepers, bee experts and enthusiasts to inspect the climate conditions which are essential to the survival of bees not only at a large spatial scale, but also for a long time period. The output of this web application together with the raw data from observations can be used in multidimensional raster analysis, e.g., trend analysis and multidimensional principal components analysis, the result of which can be further integrated with beekeeping data to assess possible impacts of likely climate change on bees.

Results

The results can be divided into two parts: the result of the web application development and the results from the assessment of climate change and its possible impacts on beekeeping. The web application part consists of the web app itself and the outputs from the web app including maps, animations and videos. The results from the analysis part are comprised of maps, figures and tables demonstrating the patterns and statistics of climate change in Europe as well as the possible association between climate change and beekeeping.

EO4BEE Web Application

Available at the URL: http://eo4bee.herokuapp.com/

Evaluation of climate change in Europe

trend analysis

multidimensional Principal components Analysis

Assessment of likely climate change impacts on beekeeping in Czech Republic

Trend Analysis

Regression Analysis

Regression analysis was conducted for 162 districts with valid bee colony data to explore how bee colony winter losses are related to the temperature trend or the precipitation trend. The results for both regression analyses didn't show statistically significant correlations.

Temporal Profile Analysis

The fluctuations in bee colony winter losses have a similar pattern to the variation in temperature anomalies with one year delay. The temperature anomalies had three peaks in 2013, 2015, and 2018, while the peaks of honey bee colony winter losses can be observed in 2014, 2016, and 2019, and for both temperature anomaly and honey bee colony winter loss, each of the three peaks are higher than the previous one.

Conclusion

The principal aim of this diploma thesis was to explore the potential of earth observation data in the context of beekeeping and likely climate change. The practical objective was to develop a web application for beekeeping support by utilizing the strong cloud computing power and huge earth observation data catalogue of GEE. And the Theoretical objectives were to extract the spatial patterns of likely climate change in Europe from multidimensional raster obtained by earth observation, to evaluate the possibility of integrating crowdsourced bee colony data with relevant earth observation data, and to explore how the analysis results can benefit beekeeping research.

To achieve the practical goal of this study, a web application named "EO4BEE" has been developed using the geemap python package and the Google Earth Engine python API in jupyter notebook. The web app was deployed to the Heroku cloud platform and is available at the url: http://eo4bee.herokuapp.com. It managed to enable the time series visualization of earth observation variables that might have an impact on bee health based on the implementation of a time slider, and allow users to generate timelapse to have an overview of the changing pattern of the selected variable. The visualization results from the web app can provide convenience for beekeepers and bee experts during the examination of local living conditions and environment for bees so that they can make more informed decisions.

In terms of theoretical objectives, several multidimensional raster analyses have been applied to the time series raster datasets of temperature and precipitation in ArcGIS Pro to investigate the patterns of likely climate change, the results of which were further integrated with bee colony winter loss data to assess the possible impacts climate change has on beekeeping. The study area is Europe, where the trend raster was calculated and visualized for both annual average temperature and annual total precipitation from 1950 to 2021. The results indicated that the whole study area demonstrated a rising trend of temperature, which provides strong evidence for the warming of climate, whereas the annual total precipitation didn't show a consistent direction of change in Europe. The statistics for temperature trend and precipitation trend have also been calculated and summarized for 44 countries cover by the study area as a result table. Due to the limited availability of bee colony data, the area was limited to the Czech Republic when analyzing the integration of climate data and bee colony winter losses. Regression analyses were performed to study the relationship between bee colony winter loss and temperature trend, and between bee colony winter loss and precipitation trend. Although no statistically significant correlations were found for both of them, the methodologies applied in the analysis can still produce new inspiration from earth observation in the research on beekeeping with concerns of climate change. Furthermore, the time profiles were plotted for bee colony winter losses and temperature anomalies calculated for every year from 2012 to 2020, and similar patterns of fluctuation were discovered from the comparison: the changing pattern of bee colony winter losses basically followed the evolution of temperature anomalies but with exactly one year delay, which indicates that high bee colony winter loss can be associated with high temperature anomalies.

In conclusion, this study provided a new perspective for beekeeping support research in the context of likely climate change, namely the application of earth observation data. The methodologies applied in this thesis can be easily transplanted to any region of interest, any other observation variables such as NDVI, or even any specific bee species, to study the influence of the changing earth's surface on bee survival.

Data source

Data source for the web app

Data source for the analysis

Copernicus Climate Change Service:  ERA5-Land monthly averaged data from 1950 to present

MOD11: Land Surface Temperature/Emissivity Daily L3 Global 1km

TerraClimate: Global, high-resolution gridded temperature, precipitation, and other water balance variables

Beekeeping data

Bee colony winter loss: Crowdsourced data from beekeepers in the Czech Republic (from the COLOSS survey)

Elements

i = 0;
										while (!deck.isInOrder()) {
											print 'Iteration ' + i;
											deck.shuffle();
											i++;
										}

										print 'It took ' + i + ' iterations to sort the deck.';