Data from: Economic costs of the invasive Yellow-legged hornet on honey bees.
This dataset result from combination of large-scale field data, niche modelling techniques and agent-based models to spatially assess the ecological and economic impacts of the Yellow-legged hornet, Vespa velutina nigrithorax, on honey bees and beekeeping in France. We estimated an overall density of 1.08 Asian hornet nest/km2 in France, based on the field record of 1260 nests over a searched area of 28,348 km2. We estimated that at least 2.6 % up to 29.2 % of bee colonies, i.e. so over 300,000 colonies, were at risk of being lost each year due to the hornet in France, corresponding to an economic cost from € 2.8 million up to € 30.8 million per year at the national scale. For beekeepers, this cost may reach 26.6 % of their honey revenues, due to livestock replacement. Our results suggest non-negligible ecological and economic impacts of the invasive Asian hornet on honey bees and beekeeping activities. Moreover, this study meets the urgent need for more numerous and accurate economic estimations, necessary to calculate the impact of biological invasions on biodiversity and human goods, with a view to enhance policies of biodiversity conservation. The data are related to the scientific paper "Requier, F., Fournier, A., Pointeau, S., Rome, Q., Courchamp, F. (2023) Economic costs of the invasive Yellow-legged hornet on honey bees. Science of the Total Environment".
Three datasets are available:
(1) We carried out a five-year fieldwork program in four French areas (districts of Pyrénées-Atlantiques, Vendée, Morbihan, and Deux-Sèvres) to detect and count the number of hornet nests, from 2013 to 2017. Field observations were performed following a standardized protocol consisting in drawing 3 km radius circles around randomly and yearly selected locations in the selected districts. Then, after the trees shed their leaves, we drove along all the public roads to cover a maximum of the circle area searching for nests located on trees or urban supports. Depending on the landscape, ranging from intensive open fields to urban areas, it represented between 600 to 800 km drive in each district. In addition to these standardized field observations, we considered supplemental nest locations provided by local citizens in order to ensure an almost exhaustive dataset inside these 90 circles (17±9 circles/district) randomly drawn and prospected. The precise GPS locations of each detected hornet nest were recorded. This dataset totalled to 1260 nests records distributed over a searched area of 28,348 km2 (the total area of the four monitored districts).
Data are available as a csv file titled "Hornet_nest_count.csv” with the following metadata:
# 'data.frame': 1260 obs. of 3 variables:
#Lat = Latitude
#Long = Longitude
#Year = Year of the observation
(2) We used the national-wide dataset of bee colony livestock from the French Ministry of agriculture (French Ministry of Agriculture, 2017) to map the density of colonies over the country. This dataset is based on mandatory beekeeper declarations and records the number of bee colonies per township (i.e. municipality area) across the whole French territory. We considered the last year available of the national database (i.e. 2015), which totalled 1,056,314 bee colonies. For each township, we used this observed nest numbers (file “Hornet_nest_count.csv”) to model and predict hornet nest density across the whole French territory. To do so, we used random forest regressions to model the relationship between a species density and a set of explanatory variables. We ran a random forest model associating hornet nest density data (nest/km2) with a number of explanatory variables, and we projected this relationship across space to predict the potential density of hornet nests over France. We then used the mechanistic BEEHAVE model (Becher et al., 2014) to assess the probability risk of bee colony mortality related to hornet predation.
Data are available as a csv file titled " Predictions_per_township.csv” with the following metadata:
#'data.frame': 36571 obs. of 13 variables:
#Zip_code = Zip code of the French township
#Township_name = Name of the French township
#X_center = X location of the French township
#Y_center = Y location of the French township
#Township_area = Area of the French township (in ha)
#No_hives = Number of bee colonies per township declared by beekeepers for the year 2015 according to the national dataset of bee colony livestock (French Ministry of Agriculture, 2017)
#No_nests_pred = Number of Yellow-legged hornet nests predicted per French towmnship
#Average_loss_high = Predicted average honey bee colony loss due to Yellow-legged hornets for the high predation scenario (i.e. large nests with 20 hornet/nest simultaneously predating in front of the colony) per French towmnship
#No_dead_pred_high = Total number of predicted honey bee colonies lost due to Yellow-legged hornets for the high predation scenario per French towmnship ( computed as #round(tab$No_hives*tab$Average_loss_high/100,0) )
#No_alive_pred_high = Total number of predicted honey bee colonies surviving to Yellow-legged hornets for the high predation scenario per French towmnship ( computed as #tab$No_hives-tab$No_dead_pred_high )
#Average_loss_low = Predicted average honey bee colony loss due to Yellow-legged hornets for the low predation scenario per French towmnship
#No_dead_pred_low = Total number of predicted honey bee colonies lost due to Yellow-legged hornets for the low predation scenario per French towmnship ( computed as #round(tab$No_hives*tab$Average_loss_low/100,0) )
#No_alive_pred_low = Total number of predicted honey bee colonies surviving to Yellow-legged hornets for the low predation scenario per French towmnship ( computed as #tab$No_hives-tab$No_dead_pred_low )
(3) To test the validation of the spatially predicted bee colony mortalities related to the hornet predation, we carried out combinations of field-work observations of (1) the hornet predation activity at colony entrance and (2) surveyed the survival of the observed bee colonies. The predation activity data consisted in counting of the number of hornets predating simultaneously in front of the colony entrances at the arrival at the apiary (around 1-minute observation per hive). The observations were repeated every 15 days, from the 1st of June to the 19th of November 2017. Each observation consists in counting the number of hornets simultaneously hovering in front of each colony. The observation was repeated consecutively five times per colony, and this was carried out on all colonies of apiaries (ranging from 4 to 24 colonies). A total of 359 observations were recorded at 51 different plots. The survival of the colony was checked during the last observation date (i.e. in November) from the outside of the hive by observing the entrance of the colony. Colonies were considered dead when no more activity (i.e. in and out walk/flight of bees at the colony entrance) could be seen. These survival observations were done with weather condition that allowed bee activities (e.g. temperature >20°C, sunshine and no wind). This independent dataset was used to cross-validate our predicted colony mortality at township scale. To do so, we compared the predicted mortality with the observed mortality at the end date of the field observations (i.e. the mortality probability of the model at the date of November 19th).
Data are available as a csv file titled "Cross_validation.csv” with the following metadata:
#'data.frame': 51 obs. of 9 variables:
#Lat = Latitude
#Long = Longitude
#Plot_ID = Identity of the plot (apiary)
#No_alive_obs = Number of colonies observed as alive
#No_dead_obs = Number of colonies observed as dead
#Mean_hornet_obs = Mean number of hornets predating simultaneously in front of the colony entrances (1-minute observation per hive)
#Max_hornet_obs = maximum number of hornets predating simultaneously in front of the colony entrances (1-minute observation per hive)
#No_alive_pred = Number of colonies predicted as alive
#No_dead_pred = Number of colonies predicted as dead
