This publication is categorised as a CSO Frontier Series Output. Particular care must be taken when interpreting the statistics in this release as it may use new methods which are under development and/or data sources which may be incomplete, for example new administrative data sources.
The System of Environmental Economic Accounting - Ecosystem Accounting (SEEA-EA) is a spatially-based, integrated statistical framework for organizing biophysical information about ecosystems, measuring ecosystem services, tracking changes in ecosystem extent and condition, valuing ecosystem services and assets and linking this information to measures of economic and human activity. It is an integrated statistical framework adopted by the United Nations Statistical Commission.
The SEEA-EA outlines five sets of ecosystem accounts:
1. Ecosystem extent accounts
2. Ecosystem condition accounts
3. & 4. Ecosystem services flow accounts (physical and monetary)
5. Monetary ecosystem asset accounts
As part of a recent amendment to Regulation (EU) No 691/2011 on environmental economic accounts, it will be mandatory to report ecosystem extent, condition and services (physical flow) accounts in line with SEEA-EA. The monetary flow of ecosystem services and the monetary value of ecosystem assets will not need to be accounted for under the new regulation. Mandatory reporting by member states will commence in 2026.
Most data used for producing these crop pollination ecosystem service accounts were for the reference year 2018. Therefore, unless stated otherwise, data was accessed for the year 2018.
The geospatial crop data were extracted from the Land Parcel Identification System (LPIS) database. This database is used by the Department of Agriculture, Food and the Marine (DAFM) to administer payments to farmers under area-based schemes such as the Basic Income Support for Sustainability (BISS), and the Areas of Natural Constraint Scheme (ANC). The LPIS database details over a million land parcels corresponding to different land uses by farmers, including the type of crop grown. Land parcels were selected where the indicated crop type produces fruit or seed which is dependent on pollination by wild insects. For example, although the potato plant requires pollination as part of its reproductive cycle, because the potato crop output (i.e., root tubers) is not dependent on pollination, it is not included here as a wild pollinator-dependent crop. In the case of strawberries and tomatoes, which are generally grown under cover in Ireland, neither were included because their fruits tend to be a product of pollination by managed and not wild bees. See Table 6.1 for a list of the different crop types included from LPIS, which are aggregated categories as classified in the Annual Crop Statistics Handbook.
The CORINE Land Cover (CLC) Accounting Layers were used as the geospatial data source for ecosystem types. As done for the production of Ecosystem Extent Accounts for 2000-2018, a crosswalk was performed between the EU Level 1 ecosystem typology and the CLC Accounting Layers third level classes. The only exception was CLC class 243 (“Agriculture mosaics with significant natural vegetation”), which has now been reallocated from Croplands to Grasslands following feedback from a national expert group. It is worth mentioning here that the 2018 ecosystem extent map could not be used directly as its Level 1 ecosystem typology does not meet the detail required for geospatially modelling this service. Therefore, the supply of tonnes of pollinator-dependent crops from each land cover type was aggregated to the Level 1 ecosystem types after the model was run.
Landscape features specifically include linear landscape features such as rivers and treelines. As the spatial resolution of the CLC Accounting Layers is not detailed enough to delineate these, geospatial data had to be sourced from elsewhere. Data for hedgerows, treelines, rivers and streams, and lakes and ponds were extracted from the 2018 National Land Cover Map (NLCM), which is 250 times more detailed than the CLC data. Data for different types of roads, railway lines and runways were sourced from the Digital Landscape Models (DLM) Core Data from Tailte Éireann. This was the most recent version of the data, which was for 2024.
Three land management practices were accounted for in the model. These included land managed as species rich grassland, land managed as traditional hay meadow and farmland which can be considered as having a high likelihood of holding a special biodiversity value. The latter is also known as high nature value (HNV) farmland. Areas of species rich grassland and traditional hay meadow were extracted from the LPIS database, whereas the European HNV map produced by the European Environment Agency (EEA) as used as the geospatial data source of HNV farmland. This map, which is for the reference year 2012, was used instead of the more recent Irish HNV farmland map because the Irish version used hedgerows as one of its input variables. As hedgerows are already accounted for in this model, the Irish HNV data can therefore not be used to avoid double counting.
Two types of geospatial weather data were required for the calculating the Forager Activity Index (FAI). This included the average ambient temperature and the average solar irradiance. The ambient temperature data (degrees Celsius) were downloaded as a 1 × 1 km grid from Met Éireann. This included minimum and maximum temperature datasets, which were divided by two to get the average temperature. The solar irradiance data (Kilojoule m−2 day−1) was downloaded as a 25 × 25 km grid from the Agri4Cast resources portal hosted by the European Joint Resource Centre. Data for both temperature and irradiance were downloaded as daily records for the periods April to September and for the years 2017, 2018 and 2019, across which means were taken for each weather data type.
Annual wild pollinator-dependent crop yield data was downloaded from the dataset on Crop production in EU standard humidity from Eurostat. This data were originally submitted by the CSO after collection from Teagasc and DAFM. Pollinator-dependent crops for which yield data were not available were instead imputed by deriving yield factors for similar crops (e.g., loganberry yield derived using raspberry yield factor). Yield factors were derived by dividing the total yield for a given crop by the total land area which grew that crop in the same period.
The effect of different landscape features and land management practices on habitat suitability was accounted for in the model by attributing correction factor values for each in the geospatial model. In short, each landscape feature and land management practice was assigned a value greater than one if was expected to positively affect habitat suitability. By contrast, a value of less than one was assigned if it was expected to negatively affect habitat suitability. The absence of any relevant landscape feature or landscape practice intersecting was given a neutral value of one. Correction factors, which were selected following feedback from a national pollination expert group, are provided in Table 6.2.
The floral attractiveness (FA) and nesting attractiveness (NA) of each land cover and crop type were accounted for in the model by assigning scores for each (See Table 6.3). Scores could vary between zero and one, with one being most positive for FA and NA. Default scores from the Eurostat methodology were reviewed and adjusted after expert group feedback. The only changes arising from this feedback was a reduction in the NA scores for annual crops, which may be in situ for too short an amount of time to favour nest establishment.
The Forager Activity Index (FAI) expresses the relative pollinator weather-dependent activity in a given location. To calculate it, the average ambient temperature and solar irradiance values per hectare were first used to calculate Tblackglobe, which stands for the stands for the temperature in a black, spherical model, which simulates the body temperature of an insect. This is calculated using the below equation:
| Tblackglobe | = | -0.62 + 1.022 T + 0.006 R |
Where T is the mean ambient temperature and R is solar irradiance in Watt/m2 (note, this requires that the solar irradiance data be converted from Kilojoule m−2 day−1). Tblackglobe was then used to calculate the FAI (%) using the below equation:
| FAI(%) | = | -39.3 + 4.01 Tblackglobe |
The model can differentiate between two broad pollinator groups: (1) Long-distance pollinators (e.g. bumblebees and hoverflies) and (2) short-distance pollinators (e.g. solitary bees). Habitat suitability maps were modelled for both separately. The differences between the two were:
After accounting for the different biophysical parameters of each land cover and crop type, including for relevant intersecting landscape features and land management practices, values were used in conjunction with FAI values in a geospatial model to produce a map of habitat suitability. In this map, every hectare of land cover (and therefore ecosystem) was assigned a habitat suitability score ranging between zero and one. A threshold score of 0.2 per hectare was selected, as recommended by Eurostat and supported by national pollination experts. Therefore, a hectare with a suitability score of 0.2 or more was deemed to be “moderate to high suitability” for providing habitat for wild insect pollinators.
After producing the habitat suitability map, the next step was to assign wild pollinator-dependent yields to each hectare of crops before then distributing those yield values to nearby land cover types providing suitable habitat. A crosswalk between each land cover type and ecosystem type could then be performed to determine the wild pollinator-dependent yield contribution from each ecosystem type to each crop category.
All crops included here will still produce a yield in the absence of pollinators. However, the yield will be, to varying extents, lower than if pollinators were present. Therefore, pollination factors for each crop were applied to the reported crop yields before then dividing by the total number of hectares growing each crop. For example, for spring-sown oilseed rape, the total yield in 2018 was 3,520 tonnes. The pollination factor for this crop is 0.27 (i.e. a maximum of 27% of its yield can be attributed to pollinators), meaning that 680.4 tonnes was the maximum pollinator-dependent yield in 2018. This yield was then distributed equally across 1,813 hectares of cropland growing spring-sown oilseed rape in 2018, giving a maximum pollinator-dependent yield per hectare of 0.38 tonnes. Crop pollination factors ranged between 0.05 and 0.95 and are given for each wild pollinator-dependent crop type in Table 6.4. Factors for each crop were taken directly as the default values provided in the Eurostat methodology. However, where indicated in Table 6.4, factors for three crops (apples, oilseed rape and beans) were selected from values observed in studies of crops grown in Ireland. These were selected following feedback from national experts.
For each hectare unit of crop, its allocated pollinator-dependent yield was distributed to all nearby hectare units of land cover harbouring suitable pollinator habitat, as determined by the model. This yield was distributed proportional to the suitability score, meaning a crop hectare allocated a maximum pollinator-dependent yield of 0.38 tonnes would be distributed as 0.15 and 0.23 tonnes where two nearby land cover types had suitability scores of 0.2 and 0.3, respectively. After this, the crosswalk between land cover and ecosystem types could be performed to get the ecosystem yield contribution in tonnes of pollinator-dependent crops.
The distribution of pollinator-dependent yields had to take into account both long- and short-distance pollinator habitats, which were modelled separately. As recommended by the Eurostat methodology and supported by national pollinator experts, this was done by distributing yields to suitable short-distance habitats first. Then, any remaining yields which could not be distributed (i.e. due to an absence of nearby habitat for short-distance pollinators) were distributed to any nearby long-distance pollinator habitats. This means that short-distance pollinators were prioritised, with long-distance pollinators having occupied an ecological niche whereby they only visited crop fields that had not yet been visited by short-distance pollinators. This approach is known as the ‘distance priority method’ and is based on the ‘neutral theory’ of species-abundance distribution in ecology.
After distributing all pollinator-dependent yields from each crop field to land use types harbouring suitable pollinator habitat, the crosswalk to Level 1 ecosystem type was performed. Then, the attributed yield for each ecosystem type could be summed for each crop category, thereby giving the total ecosystem contribution (or ‘Supply’) to the crop pollination ecosystem service. For this service, the supply is equal to the use, which in this case is ‘intermediate consumption by industry'.
How comprehensive this is determines the extent to which all wild pollinator-dependent crops have been spatially accounted for in the crop geospatial data source. The source used here was the Land Parcel Identification System (LPIS), which is used by the Department of Agriculture, Food and the Marine (DAFM) to administer area-based payment schemes to farmers. Where farmers are not availing of these schemes, their land may not be recorded in the LPIS database. This could mean that the crop pollination service is slightly underestimated.
Any land cover types (and therefore ecosystem types) smaller than the 25-hectare minimum mapping unit (MMU) for aerial phenomena in the 2018 Corine Land Cover (CLC) accounting layers will not have been detected and are instead considered as the larger ecosystem type. For example, a 0.1-hectare urban garden will appear as the much larger urban landcover type surrounding it, meaning the garden will not be considered as a potential wild-pollinator habitat.
The model can accommodate for pollinator species distribution maps, which can be used to express the probability of pollinator occurrence in each location. This data was not available for 2018 but results from The Irish Pollinator Monitoring Scheme could be incorporated into future iterations of these accounts.
Learn about our data and confidentiality safeguards, and the steps we take to produce statistics that can be trusted by all.