REPORTS & DOCS

  • Here, we will link to reports published by EFSA journal associated to ENETWILD
  • We also include links to other reports associated to African Swine Fever
  • Users will be able to access to outputs (predictions) from modelling in the map section

Download ENETWILD reports​

Abstract
The ENETWILD consortium (www.enetwild.com) has implemented an EFSA‐funded project whose main objective is to collect information and model the geographical distribution and abundance of wild boar throughout Europe. This is of particular concern owing to the spread of African swine fever from Eastern areas. In January 2018, ENETWILD organised discussion workshops for 70 experts in the field of the ecology, management and epidemiology of wild boar. Three workshops addressed the following questions: (1) what kind of data is needed to develop wild boar abundance maps?; (2) how can estimates of boar abundance be harmonised between regions?; and (3) how can the collection of wild boar distribution and abundance data be improved? In order to collect data on the presence/absence and abundance of wild boar obtained from different sources (administrations, hunters, naturalists and researchers), it is necessary to work on the generation, collection and processing of data in a harmonised manner, thus enabling the information to be comparable and used at a European level. The use of information on hunting statistics (number of animals hunted and hunting effort per surface unit) is particularly essential. The strategy is based on, firstly, collecting existing non‐harmonised wild boar data in the short‐term (occurrence and hunting statistics) by collecting the more accessible data. As a second step, ENETWILD distributed a questionnaire on how and where the data concerning hunting statistics are collected throughout the different Countries or regions in Europe. The objective of this questionnaire was to identify those places in which hunting statistics are still disaggregated (at the highest spatial resolution), with the purpose of standardising the means employed to collect hunting data in Europe. The following step consisted of the appropriate collection of data, using a data model and supported by a data‐sharing agreement.

DOWNLOAD

Abstract
The ENETWILD consortium (www.enetwild.com) has implemented an EFSA‐funded project whose main objective is to collect information and model the geographical distribution and abundance of wild boar throughout Europe. This is of particular concern owing to the spread of African swine fever from Eastern areas. In January 2018, ENETWILD organised discussion workshops for 70 experts in the field of the ecology, management and epidemiology of wild boar. Three workshops addressed the following questions: (1) what kind of data is needed to develop wild boar abundance maps?; (2) how can estimates of boar abundance be harmonised between regions?; and (3) how can the collection of wild boar distribution and abundance data be improved? In order to collect data on the presence/absence and abundance of wild boar obtained from different sources (administrations, hunters, naturalists and researchers), it is necessary to work on the generation, collection and processing of data in a harmonised manner, thus enabling the information to be comparable and used at a European level. The use of information on hunting statistics (number of animals hunted and hunting effort per surface unit) is particularly essential. The strategy is based on, firstly, collecting existing non‐harmonised wild boar data in the short‐term (occurrence and hunting statistics) by collecting the more accessible data. As a second step, ENETWILD distributed a questionnaire on how and where the data concerning hunting statistics are collected throughout the different Countries or regions in Europe. The objective of this questionnaire was to identify those places in which hunting statistics are still disaggregated (at the highest spatial resolution), with the purpose of standardising the means employed to collect hunting data in Europe. The following step consisted of the appropriate collection of data, using a data model and supported by a data‐sharing agreement.

DOWNLOAD

Abstract
This report provides a review of existing models for predicting the spatial distribution and abundance of wild boar at various scales (global, continental, national and regional) in order to inform the development of a new model to produce estimates of wild boar abundance at European level. The review identifies and discusses a range of models based on a wide variety of data types, corresponding to those targeted by the data collection model set by ENETwild, such as occurrence data (presence‐only, presence‐background and presence‐absence), hunting bag data and density data. The reviewed models are categorised in two main groups, the first based on occurrence data to predict a distribution of wild boar, and the second based on hunting bag, census and/or density data to directly model abundance. Owing to the diversity of methodologies, an ensemble modelling approach is here proposed for combining the outputs from a range of complimentary models and generating density estimates of wild boar at European scale. This would retain the flexibility necessary to utilise all available data whilst maintaining a robust output. An initial model has been outlined which uses occurrence data to generate wild boar distribution across Europe. The resulting suitability scores are related to available density estimates to establish a relationship, so the suitability map can be converted into a map of absolute density. In order to further utilize other types of data in this framework, the produced outputs of prediction of habitat suitability or presence/absence are used to underpin models based on available abundance data (hunting bag, census or density).

DOWNLOAD

Abstract
Heterogeneities in the wild boar data collection frameworks across Europe were analysed using questionnaires to explore comparability of hunting data in the short term and propose a common framework for future collection. Fifty‐seven respondents representing 32 countries covering more than 95% of European territory participated to the questionnaire. The most frequently recorded information in the official statistics included the quantity of animals shot per hunting ground and season (24 countries) and the size of the hunting (management) ground (21 countries). Georeferenced maps for the hunting grounds were collected (total or partial) for 20 countries. The least frequently recorded information was at the level of hunting events. We conclude that (i) sources of hunting statistics providing quantitative information on wild boar (and by extension, for other big game species) are lacking or are not harmonised across Europe, as well as incomplete, dispersed and difficult to compare; (ii) a feasible effort is needed to achieve harmonisation of data in a short time for the most basic statistics at the hunting ground level, and (iii) the coordination of the collection of hunting statistics must be achieved first at national and then at European level. The following is recommended: (i) countries should collect data at hunting ground level; (ii) efforts should be focused on data‐poor countries (e.g. Eastern Europe), and (iii) the data should be collected at the finest spatial and temporal resolution, i.e. at hunting event level. ENETWILD proposes the development of a robust and well‐informed data collection model as the basis for a common data collection framework. The present report identified some countries where, though the potential to share good quality data is present, the data collection promoted by ENETWILD has not succeeded so far (i. e. Eastern Europe). This highlights the need of further strategies to be developed so to encourage and support these countries to share hunting data.

DOWNLOAD

Abstract
By reviewing the different types of data targeted by the ENETWILD Wild Boar Data Collection Model (occurrence, hunting bag, abundance data) that have become available, an initial model could be built with occurrence data. A preliminary model analysis was performed to estimate the likely distribution of wild boar comparing the performance of a presence‐only model (bioclim) and presence‐background model (MaxEnt). Based on the results of this modelling, locations were identified, notably in Eastern Europe, where more data are required in order to produce more robust model projections of occurrence. This report also outlines the current state of available data collected by ENETWILD (occurrence, hunting bag and density) on wild boar and the development of a model framework that can be used to produce outputs on the yearly density distributions at high resolution of wild boar at a European scale.

DOWNLOAD

Abstract

After presenting preliminary models to estimate the habitat suitability for wild boar in MSs and neighbouring countries as a proxy for its relative abundance (i.e. the relative representation of a species in a particular ecosystem, a kind of proxy of the density) the ENETWILD consortium has developed further models for the estimation of wild boar abundance across this extent based on hunting yields (HY). This report therefore presents: i) updated maps of habitat suitability at 10×10 km resolution based on newly available data of wild boar occurrence together with new analysis to test the feasibility of performing such analysis at higher resolution (2×2 km); and, ii) a new model for predicting wild boar relative abundance, also at 10×10 km resolution, using hunting yields. The results of the occurrence model show that more occurrence data are required for specific locations in Eastern Europe in order to ensure robust model prediction of habitat suitability and consequently wild boar distribution. We used the hunting yields model to identify the environmental drivers of species abundance at European scale and fitted separate models for three regions (Southern, Western and Eastern Europe) to predict the distribution of wild boar at 10×10 km resolution. Our initial results highlighted some methodological issues relating to the statistical downscaling that should be taken into account to improve the reliability of the predictions. Whilst the spatial pattern in some areas was similar when comparing the predictions from both the occurrence and abundance models, in other regions there were marked discrepancies. To improve the models it is recommended to i) collect more occurrence data in the North‐Eastern region of Europe, in particular on survey effort; ii) combine regional and local hunting records to validate hunting yield predictions to higher spatial resolutions; and, iii) incorporate new environmental variables, especially those closely associated with wild boar abundance and distribution.

Abstract

In October 2018 the ENETWILD consortium created suitability maps based on available data on wild boar occurrence at 10 km square resolution and initial version of abundance models based on hunting statistics at NUTS3 and NUTS2 resolution, that were statistically downscaled for MSs to 10×10 km grid squares. This report presents updated suitability map for wild boar presence based on additional occurrence data and new algorithms, and new models based on high‐resolution hunting yield data for MSs and neighbouring countries. New environmental variables closely associated with wild boar abundance and distribution were also included. Our results showed no consensus for a single best occurrence model: out of those tested, both Maxent and random forest could be considered the best options depending on the choice of assessment metric. Predictions from these models notably disagreed in eastern Europe where data on wild boar occurrence are limited. Despite agreement among models, predictions in the south appeared over‐predicted, most likely due to a lack of contrasting absence data. Whilst there remain some methodological adjustments which could be tested, substantial improvement in the prediction from occurrence models relies on further collection of wild boar occurrence data in the east and complimentary data on survey effort in the south.The predictive performance of the hunting yield model was high. Although the incorporation of new data at higher spatial resolution markedly improved predictions, such data is still needed in some regions, ideally coupled with hunting effort, which would allow such estimates to be transformed into reliable densities. Comparison between predictions from the occurrence and hunting yield models showed they were statistically associated, but the strength of that relationship was dependent on the type of occurrence model and the bioregion. These findings are compatible with previous interpretations of the occurrence model, and highlight the relevance of obtaining more accurate data, especially from northern and eastern bioregions in Europe.

In October 2018 the ENETWILD consortium created suitability maps based on available data on wild boar occurrence at 10 km square resolution and initial version of abundance models based on hunting statistics at NUTS3 and NUTS2 resolution, that were statistically downscaled for MSs to 10×10 km grid squares. This report presents updated suitability map for wild boar presence based on additional occurrence data and new algorithms, and new models based on high‐resolution hunting yield data for MSs and neighbouring countries. New environmental variables closely associated with wild boar abundance and distribution were also included. Our results showed no consensus for a single best occurrence model: out of those tested, both Maxent and random forest could be considered the best options depending on the choice of assessment metric. Predictions from these models notably disagreed in eastern Europe where data on wild boar occurrence are limited. Despite agreement among models, predictions in the south appeared over‐predicted, most likely due to a lack of contrasting absence data. Whilst there remain some methodological adjustments which could be tested, substantial improvement in the prediction from occurrence models relies on further collection of wild boar occurrence data in the east and complimentary data on survey effort in the south.The predictive performance of the hunting yield model was high. Although the incorporation of new data at higher spatial resolution markedly improved predictions, such data is still needed in some regions, ideally coupled with hunting effort, which would allow such estimates to be transformed into reliable densities. Comparison between predictions from the occurrence and hunting yield models showed they were statistically associated, but the strength of that relationship was dependent on the type of occurrence model and the bioregion. These findings are compatible with previous interpretations of the occurrence model, and highlight the relevance of obtaining more accurate data, especially from northern and eastern bioregions in Europe.

DOWNLOAD

Abstract

Hunting statistics can be suitable to determine wild boar density estimates if a calibration with an accepted rigorous method is performed. Here, densities calculated from drive counts during collective drive hunting activities are compared against density values calculated by camera trapping using the random encounter method. For this purpose, we selected 10 study sites in Spain, from North to South representing a diversity of habitats, management and hunting traditions without artificial feeding, plus one study site in Czech Republic where artificial feeding was practiced. Density values estimated from both drive counts and camera trapping were strongly positively correlated (R2=0.84 and 0.87 for linear and non‐linear models, respectively) and showed a good agreement. Drive counts data might be therefore used as a density estimate to calibrate models for estimating density in large areas and potentially, to compare densities among areas. For these purposes, there is still the need to harmonise hunting data collection across Europe to make them usable at a large scale. Our results need to be confirmed across a wider number of European populations to provide valid geographical wild boar density predictions across Europe.

DOWNLOAD

The ENETWILD consortium implementedtheEFSA‐funded project “Wildlife: collecting and sharing data on wildlife populations, transmitting animal diseases agents”,whose main objective is to collect wild boardensity, hunting and occurrence dataand model species geographical distribution and abundance throughout Europe.This subject is of particular concern due to the continued advance of African swine fever (ASF). In May 2019,the ENETWILD consortiumorganised aworkshop for 30game biologists, animal health professionals, and experts from national huntingand forest authorities from 14 countries form North East Europe.The overall objectives of the workshop were to present milestones and achievements of the ENETWILD project,review different country frameworks forwild boar data collection and harmonization (hunting, density and occurrence data), as well as to review scientificmethods for determiningwild boar abundance and density, and train oncamera trapping and the random encounter method (REM).It was agreed thatwild boar abundance and densityestimates available in NorthEastern Europe are unreliable because most of them are not based on scientific methods. Hence, there is a need to implement a novel method for determining wild boar abundance and densitythat uses hunting bag statistics including measures of hunting effort and efficiency during collective drive hunts, compared against density values calculated using camera trapping and the random encounter method (REM). Several collaboratorsfrom Poland, Finland, Belarus, Russia, Lithuania have declared their willingness to participate in such pilot studies, and all agreed in improving data collection, including by means of citizen science

The ENETWILD consortium implementedtheEFSA‐funded project “Wildlife: collecting and sharing data on wildlife populations, transmitting animal diseases agents”, whose main objective is to collect wild boardensity, hunting and occurrence dataand model species geographical distribution and abundance throughout Europe.This subject is of particular concern due to the vastspread of African swine fever (ASF). In September 2019,the ENETWILD consortiumorganised aworkshop in Croatia for 27game biologists, animal health professionals, and experts from national huntingand forest authorities from 14 countriesfrom South East Europe.The overall objectives of the workshop were to present milestones and achievements of the ENETWILD project,to reviewthe framework forwild boar data collection and harmonization (hunting, density and occurrence data) of thedifferent countries,as well as to review scientificmethods for determiningwild boar abundance and density, and to train oncamera trapping and the random encounter method (REM).It was agreed thathunting bag data are currently the main source of information, although not always collected within a harmonized framework and rarely accompanied by a record of the hunting effort. Instead, wild boar abundance and densityestimates available in SouthEast Europe are unreliable because most of them are not based on scientific methods. Hence, there is a need to implement a novel method for determining wild boar abundance and densitythat uses hunting bag statistics including measures of hunting effort and efficiency during collective drive hunts, compared against density values calculated using camera trapping and the random encounter method (REM). Several collaborators have declared their willingness to participate in such pilot studies, and all agreed in improving data collection, including by means of citizen science.

 

Abstract

The ENETWILD consortium (www.enetwild.com) aims at defining the spatial interface between wild boar and domestic pigs in Europe, which is essential to evaluate the risk for ASF spread between wild and domestic. This report describes the different sources of data for domestic pigs in Europe and develops a preliminary risk map of possible spatial interaction between both groups. Specific cases from Romania and Spain where reliable data were available were assessed. This modelof the interface was based on the data from Gridded Livestock of the World(GLW) database of FAO, which provides predictions on a 1×1 km scale globally and the wild boar abundance distribution model recently elaborated by ENETWILD consortium (2020).The present available census data of livestock at the European Union level (Eurostat) is restricted to a maximum spatial resolution of NUTS2, remarking the need of developing a framework to collect harmonised data with higher resolution. This will ensure that data can be comparable, validated and used. There is need also to resolve definition issues regarding the pig production systems. Our prediction model of the interface between pig and wild boar at European level indicated that the maximum risk is scattered over Central Europe, large parts of Spain, north‐east France and Romania. Hungary, so as Serbia and Croatia in the Balkans are at the highest risk in that area. In the specific case of Romania, no statistically significant association between census data of pigs collected fromnational authorities and predicted values was found when assessing the GLW model, evidencing that GLW predictions do not reliably represents the pig abundance distribution within countries. When assessing the interface model in Spain, certain areas of interaction were lacked, e.g. where extensive farming is relevant. The current discrimination of extensive vs. intensive farms of predictive models (GLW) is not reliable to perform analyses separately for each production system.The outputs this model of interface between wild boar and domestic pigs will guide future steps in both data collection and modelling approach.

DOWNLOAD

Abstract

Enetwild consortium aims at aggregating data on occurrence, abundance and hunting bag of wildlife in Europe, either as raw data or as results of statistical estimation. These data come from a large community of researchers, hunters and wildlife managers. A flexible and robust data standard is therefore necessary to present the large diversity of data and collection method. We evaluated the possibilities offered by the Darwin Core Standard. The Event core, the occurrence extension and the extended measurement or fact extension proved their utility for our purpose. However, these were not able to record statistical estimation values. We proposed to extend the measurement or fact extension to allow them to be nested among themselves. Any confidence interval or precision measure is indeed a measurement about the punctual estimate, another measurement. We proposed controlled vocabularies adapted to wildlife survey in data and metadata. This will be aligned with the EFSA data model harmonisation under the SIGMA project.

 

DOWNLOAD

Abstract

The ENETWILD consortium updated in August 2019 suitability maps of wild boar occurrence and relative abundance based on hunting statistics, providing predictions at 10×10 km. External validation of this relative abundance model and a new downscaling on a 2×2 km grid was addressed in the ENETWILD report in January 2020. In this report, we update and incorporate additional data to provide new maps of wild boar suitability with a resolution of 2×2 km, obtained from a presence‐only algorithm, and a new model based on hunting yield data for MS and neighbouring countries. Hunting yield‐based modelling provides further novelties in two ways: new predictor variables and new (two) modelling approaches, i.e. smooth bioregion modelling and two‐step independent bioregion modelling). Internal validations and comparisons among previous and new suitability and hunting yield models were also addressed, as well as external evaluation of the best new approach at European scale (at country level). The suitability map showed a good agreement with the expert‐derived species range published by the International Union for Conservation of Nature. New models of relative abundance performed in general better than the previous one according to internal validations, concretely the two‐step independent bioregion approach gives the best validation scores. This approach solved the abrupt changes in predictions between bioregion boundaries. As with the previous model, the external evaluation of the new model based on hunting yield presented certain over‐prediction of the total hunting bags reported per country, although a high linearity between observed and predicted values was achieved. Previous and new hunting yield model predictions showed disagreements particularly in North and East regions, and other scattered areas in South and West, being areas in which the new model provides more reliable predictions. Hunting yield model outputs showed a relevant improvement in smoothing transitions between bioregions due to the flexibility provided by the new approach. Our analysis showed only partial agreement between suitability and relative abundance models, and reasons for these differences are discussed in this report.

 

DOWNLOAD

Abstract

The methods for estimating relative abundance and density in wild ruminant species are reviewed and insights on how to obtain reliable estimations by using those methods are provided. Eighteen methods used in nineteen wild ruminant species widely distributed across Europe are reviewed. In accordance with the ENETWILD consortium objectives, we evaluate if different types of data can be used to generate harmonisedand comparable database at large scale and for calibration of hunting data into abundanceindices or population density. In addition, recommendations to select the methods to estimate the abundance or density and its implementations for ungulate populations are provided. How to increase the output quality provided by certain methods recognised as reliable (good accuracy and precision)and with the potential to be used for the validation and calibration ofother direct (i.e. based on observation of animals) or indirect (i.e. based on signs of animal activity) methods was recommended. Largely, the “counting” of large herbivores on a regional scale is often unfeasible, it can only be possible to accurately assess population status at local scale. We show that the habitat type plays a key role in the selection of the best method to determine density or relative abundance and that this is partially irrespective to species characteristics. A method that gives a density estimate rather than relative abundance, if possible, should be used. High‐quality hunting data statistics (collected at fine spatial resolution) have the highest availability and comparability potential across Europe, to give long‐term and large‐scale trends and should be used in predictive spatial modelling of wild ruminant relative abundance and density. Therefore, their standardized and harmonised collection is strongly recommended. On a local scale (e.g. management units), camera trapping is a method that can be conducted in different environmental conditions and at any time to collect robust data. In open areas, where camera trapping may require an excessive effort, we suggest using methods involving the direct detection of animals (vantage points, linear transects, block counts, random points). This should be carried out by correctly defining the study areas (for instance by means of distance sampling) and by estimating the repeatability of the results.

Download

Reports by EFSA about African Swine Fever