next
previous
items

Indicator Assessment

Fragmentation of natural and semi-natural areas

Indicator Assessment
Prod-ID: IND-152-en
  Also known as: SEBI 013
Published 21 May 2010 Last modified 11 May 2021
12 min read
This is an old version, kept for reference only.

Go to latest version
This page was archived on 24 Aug 2017 with reason: A new version has been published

European ecosystems are literally cut to pieces by urban sprawl together with a rapidly expanding transport network. The increase of mixed natural landscape patterns due to the spread of artificial and agricultural areas into what used to be core natural and semi-natural landscapes is more significant in south-western Europe.

Fragmentation is in many places caused by forest harvesting and has a dynamic and cyclic nature but in south-western Europe, losses towards agricultural and artificial surfaces are more frequent. In the period 1990 - 2000 the connectivity for forest species was stable in approximately half of Europe's territory and increasing or decreasing slightly for another 40 %. The decrease was significant in about 5% of provinces spread in Denmark, France, the Iberian Peninsula, Ireland and Lithuania.

National patterns of core forest loss (%) by type of forest conversion and forest fragmentation process

Note: How to read the graph: In Netherlands, nearly 60% of core forest loss is towards artificial/agricultural cover and dominated by shrinkage (around 45%), then attrition (above 10%)

Spread of artificial and agricultural surfaces into previously core natural or semi-natural landscapes

Note: How to read the map: in south-west Spain, the spread of artificial and agricultural surfaces into previously core natural/ seminatural landscapes was significant between 1990 and 2000

Core forest fragmentation between 1990 and 2000

Note: Data from Corine Land Cover (CLC) for years 1990 and 2000, hence with same geographical coverage and forest definition as CLC; mathematical morphology based software GUIDOS (Soille and Vogt, 2009) and GIS analysis; results aggregated at provincial units (NUTS level 2/3).

Data source:

Change in forest connectivity between 1990 and 2000

Note: How to read the map: in eastern Spain, there was a high decrease in forest connectivity between 1990 and 2000 for forestdwelling species with 1 km average dispersal distance

Patterns of natural/semi-natural landscapes (1) (Figure 2)
Pattern changes can be naturally-occurring phenomena but are mostly driven at this scale by anthropogenic causes. The increase in mixed natural landscape patterns due to the spread of artificial and agricultural areas into previously core natural/seminatural landscapes was found to be more significant in south-western Europe (see Figure 2). The increase in core natural landscape patterns, when observed, is generally driven by the spread of core forest and other wooded landscape.

Core forest fragmentation in the period 1990 - 2000 (Figure 1 and Figure 3)
It is well known that forest area is currently increasing in Europe but this is not uniformly distributed. Locally, the spatial forest pattern is changing due to different dynamics: loss of forest areas, fragmentation of forest cover and therefore loss of connectivity. Those processes are likely to have ecological effects.

Fragmentation is in many places caused by forest harvesting and has a very dynamic and cyclic nature that may be beneficial to some species and highly detrimental to others (land mechanically disturbed after clear cut may be replanted or left to natural regeneration).

The term 'core forest' refers to the area of a forest patch minus a 100 m edge (2). Fragmentation processes in core forest loss potentially lead to effects on species (reduction of resource base, vulnerability to external disturbances, etc.). They can be due to different spatial pattern processes that were quantified at national level, e.g. forest patch shrinkage (forest loss at the periphery of a forest patch, with potential area effects on species), patch perforation (forest loss in the interior part of the patch introducing potential edge effects on species),patch attrition (the forest patch is totally removed, with potential sample effects on species). Countries in Figure 1 were ranked according to the proportion of total losses being converted towards artificial and agricultural cover in the period 1990 - 2000.

Forest connectivity (Figure 4)
The connectivity measure considers the inter-patch and intra-patch connectivity for forest dwelling species with a selected dispersal distance. In particular, the measures accounts for the shortest paths and potential dispersal flux between every pair of forest patches, the connected area existing within the patches themselves, and the role of forest patches as connectors or stepping stones that facilitate dispersal between other patches in the landscape. The non-forested landscape is considered as homogeneous.

Connectivity was rather stable in half of the provinces. The most significant decrease was foundin about 5 % of provinces spread in the eastern and western part of the Iberian Peninsula, the northern part of Ireland, southern Denmark and locally in France and Lithuania. All provinces in the hemi boreal countries, central Poland, south Germany, central France and parts of Portugal and Spain experienced connectivity loss.

(1) Natural/semi-natural lands include forest, transitional wooded land, grassland/shrub land, open space with little vegetation, inland and coastal wetlands. Patterns of natural/semi-natural lands are defined according to the composition in terms of natural/semi-natural, artificial/built-up and agricultural surfaces in the 50 ha surroundings of each natural/semi-natural land pixel (1 ha). Core natural landscape patterns are natural/semi-natural lands with a 100 % natural neighbourhood. Mixed natural landscape patterns are natural/semi-natural lands with at least 60 % natural neighbourhood and the rest as agricultural and/or artificial.

(2) Because edge effects are species specific, a 100m edge width was arbitrarily selected as a generic protection belt for interior forest species (100 m is for example the penetration distance of noise disturbances affecting interior forest birds). Forest class: a single forest class after dissolving boundaries between Corine classes 3.1.1 (broad-leaved forest), 3.1.2 (coniferous forest) and 3.1.3 (mixed forest); include young plantation when 500 subjects/ha, transitional woodland when canopy closure > 50 %. Non-forest class: includes transitional other wooded land, young plantations (< 500 subjects/ha), clear cuts, burned areas, forest nurseries and natural/semi-natural non-wooded vegetation (CLC classes 3.2 and 3.3), artificial (CLC class 1) and agricultural (CLC class 2) surfaces, wetlands (CLC class 4).

FURTHER INFORMATION

Supporting information

Indicator definition

The indicator shows the change in average size of patches of natural and semi natural areas, on the basis of land cover maps produced by photo-interpretation of satellite imagery.

Units

% change of area


 

Policy context and targets

Context description

The indicator is intended to address the question of ecosystem integrity by providing a measurement of 'disintegration' of the countryside across Europe.

Land use in Europe has changed substantially during the past century. The changes in land use have in turn affected the size of natural and semi-natural patches of land and have introduced fast growing fragmentation of the wider countryside. This indicator gives information on the trends in patch size of natural and semi natural areas at the pan-European level, by calculation of values derived from land cover maps.

Land cover maps are developed from satellite imagery based on the spectral properties of each pixel within a scene. For this indicator we use data from the Corine land cover database (CooRdinate Information on the Environment - Corine). The CLC data are based on 44 land cover classes of which 26 are considered as natural and semi natural for the purpose of this indicator (see Annex 1). These can be grouped into forests, pasture, agricultural mosaics, semi-natural land, inland waters and wetlands.

By calculation of size values for areas belonging to these land cover classes, we have information on the extent of fragmentation which has occurred in the natural and semi-natural areas, within the limitations of the CLC data (see Section on main disadvantages).

Relation of the indicator to the focal area

Natural and semi-natural areas represent an important integrity component of any given ecosystem, by supporting the full range of ecosystem services and the majority of species and habitats to be found in this type of ecosystem. If the size of such areas decreases, the integrity of the whole ecosystem is at risk. This in turn might affect the potential of the given ecosystem to deliver goods and services.

Targets

No targets have been specified

Related policy documents

No related policy documents have been specified

 

Methodology

Methodology for indicator calculation

Natural and semi-natural areas are represented by selected land cover categories which are forests, pasture, agricultural mosaics, semi-natural land, inland waters and wetlands. For a given region/ country, the change in average patch size of the selected land cover categories is the difference between two dates in their mean value, calculated as their quadratic mean.

The indicator is produced by using a simple mathematical calculation, the quadratic mean between the mean values of the patch size of a given area between two dates. By using the quadratic mean, the size of the individual objects matters as much as their number: in most cases, strong fragmentation of the larger areas matters more than fragmentation of small ones. At the same time, when a small patch in an area disappears completely (in time 2), the mean value for that area will be greater than at the time it was still present (time 1), unless the number of patches (n) in time 2 can not be less than in time 1. That means that patches with size = 0 have to be taken into account too.

The Quadratic Mean or Root Mean Square (RMS) is the square root of the mean square value of a variable so it is a statistical measure of the magnitude of a varying quantity. It can be calculated for a series of discrete values or for a continuously varying function, using the following formula:

Quadratic Mean or Root Mean Square = SQRT (1/n ((X1)2 + (X2)2 + (X3)2 +........+ (Xn)2 )
where X = Individual score and n = Sample size (number of scores or units)

The values are calculated from the available Corine land cover data following the selection of classes considered as natural or semi-natural areas. The classes proposed here are listed in Annex 1.

Calculation can be done by NUTS level 2 or 3, or by river basin, as well as by country and biogeographical zone. The analysis can be done separately for different classes of patch size (e.g. large, medium and small), in order to capture specific trends and avoid some bias mentioned previously. The analysis can also be performed as aggregated for all selected classes (e.g. those selected for the Green Background Index, see EEA, 2006) or separately by broad habitat types (proxy: land cover types).

Methodology for gap filling

No methodology for gap filling has been specified. Probably this info has been added together with indicator calculation.

Methodology references

 

Uncertainties

Methodology uncertainty

No uncertainty has been specified

Data sets uncertainty

No uncertainty has been specified

Rationale uncertainty

MAIN DISADVANTAGES OF THE INDICATOR

Methodology: one remaining difficulty with the use of the quadratic mean is with the mere disappearance of small areas (smaller than the arithmetic average) which pushes the indicator up. This means that in the case of smaller areas disappearing completely, which should be interpreted as a loss of diversity in the landscape, it may be expected that the larger areas have increased in size and this will be then interpreted as a positive sign for biodiversity.This can be neglected when dealing with a large number of areas but it may be a problem with a small number of units and a high standard deviation. But even in that case, the distortion is less important with the quadratic mean than with the arithmetic average. A second remark is that this highlights again the multimodal character of the distribution:
averaging large areas with small areas is to some extent arbitrary and should be kept to the purpose of a high level indicator only.

Data set resolution: the main disadvantage of using the CLC data set is that fragmentation occurring below the threshold of minimum resolution of 25 Ha is not detectable. The CLC data however are the best available at present to cover large areas of Europe in a harmonised way.

Biodiversity relevance: the indicator does not provide direct information on the impact of habitat fragmentation on the status of species populations.

ANALYSIS OF OPTIONS

The proposed indicator corresponds directly to the CBD proposed indicator for immediate testing under a similar name. The present indicator gives a broad brush picture of the integrity of ecosystems in Europe.

Complementary to this indicator, other measurements of ecosystem integrity should be proposed especially dealing with fragmentation / connectivity in relation to species. Indicators that focus on ecologically more relevant characteristics than 'mean habitat patch size have been developed and tested and are currently available. The JRC-Ispra work on change in spatial pattern of selected ecosystems (see http://forest.jrc.it/biodiversity/) produces indicators that give (per grid cell, nut level, etc.) the state and trends over the 1990-2000 time period of six pattern classes, namely of 'core habitat', 'edge', 'small forest fragments', 'perforation isolated patches', 'branches and short cuts' and/or 'corridors' for selected ecosystems (on the basis of CLC). One of these indicators may complement this indicator, as it has more potential to be linked with functional aspects that are meaningful for biodiversity/species.

Data sources

Other info

DPSIR: Pressure
Typology: Descriptive indicator (Type A - What is happening to the environment and to humans?)
Indicator codes
  • SEBI 013
Frequency of updates
Updates are scheduled every 6 years
EEA Contact Info

Permalinks

Geographic coverage

Temporal coverage

Dates

Document Actions