See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/351971320

Mapping forest condition in Europe: Methodological developments in support

to forest biodiversity assessments

Article in Ecological Indicators · September 2021

DOI: 10.1016/j.ecolind.2021.107839

CITATIONS READS

18 569

6 authors, including:

Ana Isabel Marín Dania Abdul Malak

University of Malaga University of Malaga
38 PUBLICATIONS 796 CITATIONS 65 PUBLICATIONS 2,280 CITATIONS

SEE PROFILE SEE PROFILE

Annemarie Bastrup-Birk Gherardo Chirici

European Environment Agency (EEA) University of Florence
69 PUBLICATIONS 2,859 CITATIONS 399 PUBLICATIONS 7,614 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Dania Abdul Malak on 01 June 2021.

The user has requested enhancement of the downloaded file.

 Ecological Indicators 128 (2021) 107839

Contents lists available at ScienceDirect

Ecological Indicators
journal homepage: www.elsevier.com/locate/ecolind

Mapping forest condition in Europe: Methodological developments in

support to forest biodiversity assessments
Ana Isabel Marín a, *, Dania Abdul Malak a, Annemarie Bastrup-Birk b, Gherardo Chirici c,
Anna Barbati d, Stefan Kleeschulte e
a
European Topic Centre of University of Malaga, Malaga, Spain
b
European Environmental Agency, Forests and Environment, Copenhagen, Denmark
c
University of Firenze, Firenze, Italy
d
University of Tuscia, Viterbo, Italy
e
space4environment, Niederanven, Luxembourg

A R T I C L E I N F O A B S T R A C T

Keywords: Forest condition, biodiversity, and ecosystem services are strongly interlinked. The biodiversity levels depend to
Forest condition a large extent on the integrity, health, and vitality of forests at the same time as losses of forest biodiversity lead
Forest biodiversity to decreased forest productivity and sustainability. Under this conceptual framework, this study presents a
Mapping
methodology for mapping forest condition at European scale supporting the attainment of the 2020 Aichi
Environmental policy
Europe
Biodiversity Target 5 “the rate of loss of all natural habitats, including forests, is at least halved and where feasible
brought close to zero, and degradation and fragmentation is significantly reduced” and the implementation of Sus­
tainable Development Goals (SDG), as well as the EU forest strategy since the sustainable forest management is
oriented to support the provision of forest services and to enhance the condition of biodiversity forests’ host.
The work presents the developments of an operational indicator at European scale. This spatially explicit
information on forest condition can be the baseline map with a 1 km resolution to monitor the state and changes
of condition by exposition to pressures and threats. This condition indicator considers structural, functional, and
compositional aspects of forest with relevance for health and vitality of species and habitats hosted by forest
ecosystems.
The methodology implemented used harmonized, published and open datasets. It provided confident results
for the assessment of the condition within hemiboreal, temperate and alpine forests, showing the Carpathian,
Dinaric Alps and Alps, among others, as hotspots with pre-dominantly good condition. The results were validated
with data derived from the reporting for the EU Habitat Directive and explicit dataset on known primary forests
in Europe. However, this method underestimated the forest condition in the Mediterranean and Boreal forest
types due to data gaps, regional specific characteristics, and design limitations.
This study illustrates an operational and transferable approach for addressing the assessment of ecosystem
forest condition at European scale being considered as a support tool for European countries when mapping and
assessing their national territory, as potential common approach to map forest ecosystems that allows for
consistent aggregation and comparisons across scales.

1. Introduction regulating climate and nutrient cycling, and contributing to human

health and recreation (Thompson et al., 2009). Although Sustainable
In recent years, the multiple functions and potential uses of forests Development Goal 15 (SDG 15) (life on land) includes several targets
have attracted interest. In addition to wood supply, forests also provide related to forests; the State of the World’s Forests highlights the pro­
multiple ecosystem functions and services that are vital to society and found interlinkages that exist between forests and trees and multiple
human well-being. These include providing freshwater and clean air, goals and targets of the 2030 Agenda for Sustainable Development

* Corresponding author at: European Topic Centre of University of Malaga Edificio de Investigación Ada Byron, C/Arquitecto Francisco Peñalosa, 18, Ampliación
Campus Teatinos, 29071 Malaga, Spain.
E-mail address: [email protected] (A.I. Marín).

https://doi.org/10.1016/j.ecolind.2021.107839
Received 22 March 2020; Received in revised form 24 May 2021; Accepted 24 May 2021
Available online 29 May 2021
1470-160X/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
A.I. Marín et al. Ecological Indicators 128 (2021) 107839

(FAO, 2018) going well beyond SDG 15 to contribute to achieving last decade, but further efforts are needed to integrate, harmonise and
multiple goals and targets across the 2030 Agenda. interpret this data (i.e. making data useable for non-experts) (Ruiz-
In Europe, forest covers about 180 million hectares, this means more Benito et al., 2020). The output contributes to a general forest moni­
than 40% of land, making it as one of the most forest-rich regions in the toring, at EU or regional scale, and can support the evaluation of
world (EEA, 2016). However, this does not necessarily mean that progress towards attaining forest-related SDGs.
Europe’s forests are not subject to pressures and changes that may
threaten their biodiversity and multi-functionality. This ecosystem is 2. Material and methods
under increasing pressure as result of climate change which aggravates
other key drivers of pressures such as pests, diseases, extreme weather According to the main objective above described, Fig. 1 outlines the
events and forest fires. Additionally, other pressures come from air conceptual approach followed in this work. The combination of the el­
pollution, land abandonment and encroaching human development ements, datasets and tools, allows to integrate biodiversity, health and
(infrastructure and tourism) (Moreira et al., 2020). While there has been structures forest characteristics into a composite condition indicator.
an increase in the societal and political responses to biodiversity loss The condition indicator is composed by three sub-indicators that
with ambitious targets, Europe continues to lose biodiversity at an comprise forest attributes based on diversity and abundance of forest
alarming rate, and many agreed policy targets will not be achieved species (FBI), the structural ecosystem attributes by the horizontal
(EEA, 2019a). The mid-term evaluation of progress towards attaining structure measured as forest extension and compactness (FSI), and the
the 2020 Aichi Biodiversity Targets set in the United Nations Convention biomass value expected for the forest type (FFI).
on Biological Diversity (CBD) showed that, if the world stays on its The study area was defined by the ecosystem types of Europe map
current development path, the state of biodiversity will continue to that represents the terrestrial habitat classes of the European Nature
decline. Information System (EUNIS) at level 2 (Table 1; EEA, 2019b). We used
Cardinale et al. (2012) revealed that the biodiversity levels depend the class “woodland, forest and other wooded land” to outline the forest
largely on the integrity, health and vitality of forests. Losses of forest areas from this ecosystem map. These areas are dominated by woody
biodiversity led to decreased forest productivity and sustainability. vegetation of various age or have a succession climax vegetation types
The condition of forest ecosystems depends on a multiplicity of on most areas, thus supporting many ecosystem services.
factors whereas forest health and vitality are often considered the most The main datasets used to assess the forest ecosystem condition are
significant factors to underpin its optimal function. Mapping forests with summarised in Table 1. Most of the datasets are not truly empirical data
their bio-ecological values and functional status is an urgent task for its but products derived from original empirical data by different modelling
protection and sustainable development. approaches properly described and validated in their respective meta­
It is widely recognized that biodiversity is a major driving force in data (see table 1).
ecosystem function (Hooper et al., 2005; Schulze and Mooney, 2012). The spatial resolution of the condition indicator is 1 km2. The
One of the main threats to forest biodiversity comprises the loss of coverage is the 27 EU Member States and the United Kingdom together
’naturalness’ of forest ecosystems due to an inappropriate ecosystem with the other EEA member countries Iceland, Liechtenstein, Norway,
management. Consequently, naturalness assessments are increasingly Switzerland, and cooperating countries: Albania, Bosnia Herzegovina,
being required to determine and monitor both the ecological status and Kosovo under the UN SCR 1244/99, North Macedonia, Montenegro, and
the development of forests (Chiarucci and Piovesan, 2020). However, Serbia. The Macaronesia region (Azores, Madeira and the Canary
naturalness is a complex term that could be defined from different Islands) and Turkey were excluded of the assessment due to lack of in­
perspectives (McRoberts et al., 2012). In the present study, naturalness formation in some of the data input.
is defined as ’the similarity of a current ecosystem state to its natural The target area of the study is the European forest, above mentioned,
state’ Winter (2012) and biodiversity is defined as ’the diversity of life in but the assessment was designed and implement by disaggregating the
all its forms and all its levels of organization’ (Hunter, 1990). EU forest extent in forest typologies. The European Forest Types - EFTs-
On the other hand, forest structure influences forest biodiversity classification stratifies the forest area into homogeneous ecologically
directly, by complex interactions, through the formation of microhabi­ units in a meaningful way. This facilitates the analysis, interpretation,
tats as well as the determination of larger-scale habitat characteristics and reporting of forest data, especially concerning biodiversity-related
(McElhinny et al., 2005; Mac Nally et al., 2001). Forests with more information (Barbati et al., 2011). The EFTs classification was trans­
complex structure are thought to be more resilient and potentially even ferred to spatially explicit information through the European habitat
more productive, delivering more forest functions. They provide valu­ suitability map which relation is one-to-one (Casalegno et al., 2011).
able habitats for a greater diversity of plants and animals than do forests The following sections outline the composite indicator on Forest
with less structural complexity. Forest degradation is usually associated Condition indicator (FCI) that drives the overall condition of forest
with a reduction in vegetative cover, especially trees and modification of ecosystems.
spatial patterns toward a fragmented forest. This means a decline in
forest condition that leads to forest biodiversity loss by reducing the
available habitat of forest-dependent species and indirectly through 2.1. Forest biodiversity sub-indicator
disruption of major ecological processes such as pollination, seed
dispersal and gene flow. The forest biodiversity sub-indicator (FBI) is a composition of the
natural assemblage tree species (FNI) and the tree species distribution
This work aims to develop an operational indicator to assess the
general forest condition in Europe. The methodology is based on the (TDI) variables here considered as a proxy of the biodiversity (Chirici
et al., 2012; Garo et al., 2014). The TDI shows, in percentage, the
definition of the forest condition (health and vitality), on the combined
presence of abiotic and biotic pressures and the way they affect tree number of tree species present within each pixel in respect to the
maximum number of tree species present in the respective forest type
growth and survival, the yield and quality of wood and non-wood
products, wildlife habitat, recreation, and scenic and cultural values (Eq. (1)). The data source was the relative probability of the presence of
tree species (RPP) maps available in the database of the European Atlas
(FAO, 2017). Therefore, this approach includes the role of biodiversity,
of Forest Tree Species (de Rigo et al., 2016; San-Miguel-Ayanz et al.,
forest health and structures in the ecosystem condition and, conse­
2016). A tree species was counted in a pixel if the RPP > 1%.
quently, the capacity to provide services. This indicator provides
spatially explicit information on general forest condition in Europe (∑K )
k=1 TSi
using, as inputs, available, accessible, and reliable databases at Euro­ TDI = ∙100 (1)
MaxTS
pean scale. The forest data availability across Europe is improving in the

2
A.I. Marín et al. Ecological Indicators 128 (2021) 107839

Fig. 1. Methodological approach used for the development of the condition indicator of European woodland and forest ecosystems.

species with high biodiversity in respect of their forest type.

Table 1
The second variable evolved in FBI is related to the assessment of the
Sources and reference year of the main datasets included in the development of
natural assemblage tree species (FNI). This measures is a proxy for
the forest condition indicator.
measuring the spatial congruency between the suitable habitat distri­
Name of the data set Attribute Year Data bution of tree species (based on predicted potential distribution of
supported owner
EUNIS habitat suitability (EEA, 2015a)) and current tree species distri­
European Ecosystem map v.2.1(1) General. 2012 EEA bution based on RPP of tree species maps (de Rigo et al., 2016; San-
Ecosystem
Miguel-Ayanz et al., 2016). Those tree species did the relationship be­
coverage
Forest type habitat suitability(2) General. Forest 2011 JRC tween these two datasets that the EUNIS system classifies as dominant in
type distribution each habitat (EEA 2015b). The approach follows the methodology
Corine Land Cover (CLC)(3) Structural 2012 EEA proposed for calculating forest naturalness in Bastrup-Birk et al. (2014)
attributes by the Eq. (2).
EUNIS woodland, forest and other Biodiversity 1940–2011 EEA
wooded land habitats, predicted attributes
EUNIS habitat maps represent the suitability distribution by habitat
potential distribution of habitat type. These datasets, modelized based on sample plot data and the
suitability(4) Maxent software package, indicate where conditions are favourable for
Relative probability of presence of Biodiversity 2006 JRC the habitat type. The grid values in these maps represent the probability
forest tree species (RPP) of attributes
(ranging from 0 to 1) that the cell is suitable for each habitat.
European Atlas of Forest Tree
Species(5) For the FNI calculation, the EUNIS habitat maps were converted to
Above ground biomass of forest(6) Functional 2012 JRC Boolean layers (suitable / non suitable) setting the break value as the
attributes percentile 25 of the statistical distribution of the sample points used for
EEA: European Environment Agency. the EUNIS habitats suitability modelling (EEA, 2015b).
JRC: Joint Research Centre of the European Commission. Then, naturalness is related to the presence and the coverage of those
(1)
https://www.eea.europa.eu/data-and-maps/data/ecosystem-types-of- potentially dominant species within the habitat suitable area according
europe. to the RPP maps and greater than 1% criterium.
(2)
https://forest.jrc.ec.europa.eu/en/past-activities/tree-species-and-forest-h The result of Eq. (2) expresses the percentage of natural tree species
abitat-suitability/. composition in the fuzzy values between 0 and 100. The values close to
(3)
https://land.copernicus.eu/pan-european/corine-land-cover/clc-2012. 100 mean a high percentage of native tree species (natural) whereas
(4)
https://sdi.eea.europa.eu/catalogue/srv9008075/api/records/af43952
values close to 0 approximate a low level of naturalness.
c-393b-426f-8660-01fccdf758eb.
(5) ∑n
https://forest.jrc.ec.europa.eu/en/european-atlas/atlas-data-and-metada FN I = (C∙P)i (2)
ta/. i=1
(6)
https://data.jrc.ec.europa.eu/dataset/71a38170-4a68-443c-9425-cd713 i = EUNIS habitats
c85291d. C = fractional area covered by tree species
P = suitable habitat for tree species. 0 (absence/ non-suitable) or 1
TSi = tree species present in i pixel (presence/suitable).
MaxTS = maximum number of tree species by EU Forest Type The forest biodiversity indicator (FBI) was calculated based on the
The TDI ranges between 0 and 100 where low values show forest average natural assemblage species and tree species distribution sub-
areas hosting species assemblages having low tree species richness, as indicators, as shown in the Eq. (3). The indicator values range from
proxy of biodiversity, and high values identify forests hosting tree 0 to 100 where higher values indicate high value from biodiversity

3
A.I. Marín et al. Ecological Indicators 128 (2021) 107839

perspective including high natural tree species composition and high poor forest condition.
tree biodiversity. In contrast, low values of FBI define forest with low
FFI + FSI + FBI
naturalness and low tree species diversity. FCI = (6)
3
FBI = 0.5∙TDI + 0.5∙FNI (3)
2.5. Validation and reliability assessment
2.2. Horizontal forest structure sub-indicator
The developed composite forest condition indicator was tested by
Horizontal forest structure was assessed by two morphological fea­ regression analysis against other spatial explicit information about for­
tures: extent (S1) and spatial pattern (S2) of forest. S1 identifies the est condition available in Europe (Fig. 2). The analyses aimed to
extent of forests as percentage of forest cover per 1 km2 grid cell. S2 demonstrate any congruencies and differences between the different
evaluates the extent and compactness of forest patch as percentage of sources of information, validate the produced map, and gauge the po­
core forest. The core forest is defined as the interior of forest patches, tential use of forest condition map in the context of ecosystem service
excluding the perimeters. Essentially these areas refer to the interior part assessments.
of the forest. From a purely geometric concept, the core pixels are Primary forests are considered hot spots for forest biodiversity and as
defined as those forest pixels whose distance to the non-forest is greater such represent a high overall forest condition regarding compositional,
than the given size-parameter (100 m in this study) (Soille and Vogt, structural and functional aspects. The forest condition map was tested
2009). against field observations from primary forest patches inventoried by
From a forest functionally perspective, within the same forest com­ Sabatini et al. (2018) being the most comprehensive spatially explicit
munities, core forests ensure a higher and more specific biodiversity dataset on known primary forests in Europe currently available.
than in forest boundaries. Forest core function ensures the best condition Furthermore, a sub-set of 56 Natura 2000 sites were randomly selected
for area-sensitive edge-intolerant species (Burkey, 1995; Ferraz et al., across Europe. The conditions of the forest habitats protected by Natura
2003). Both indicators were computed based on Corine Land Cover 2000 sites were derived from the information reported by the countries
(CLC) 2012 at a pixel resolution of 100 m. A forest pattern map was in compliance with the Habitats Directive (Council Directive 92/43/
derived by applying the GUIDOS Toolbox (Vogt and Riitters, 2017) on EEC). We used specifically the conservation status of habitats and on
this dataset. Eq. (4) shows the forest structure sub-indicator (FSI) as a compensation measures taken for projects having a negative impact on
composition of the two aforementioned variables. The FSI ranges from Natura 2000 sites. The proxy value was computed as the area-weighted
0 to 100 with high values indicating high interior forest habitat, whereas average of the habitats forest present in the Natura 2000 site, consid­
low values define forest with lower interior habitat quality. ering A: excellent value as 3; B: good value as 2; and C: significant value
( ) as 1.
S2
FSI = *100 (4)
S1
3. Results
S1 = % forest coverage per 1 km2
S2 = % core forest per 1 km2 The three sub-indicators that composed the European forest condi­
tion index are shown in Fig. 3. Fig. 3A shows the map of the forest
2.3. Forest function sub-indicator function sub-indicator (FFI) where higher values of forest condition
according to their functionality are in the mountainous areas, in the
The forest function sub-indicator (FFI) was derived from the forest southern part of Fennoscandia, and in some Baltic and Balkan countries.
biomass map for Europe (Barredo et al., 2012). Forest biomass is an Fig. 3B shows the results of the forest structure indicator (FSI). Central
important measure of ecosystem productivity. It is used in quantifying and Eastern Europe or mountainous regions present the higher values
the role of forests in the carbon cycle, the potential for energy produc­ than the relatively moderate to lower values in inner Boreal forests and
tion, and the carbon stock estimation for climate change modelling. lowlands, particularly the Atlantic zones and some patches in the
Eq. (5) describes the forest biomass value per grid cell normalized Mediterranean region.
among the EFTs. FFI ranges between 0 and 100 that can be interpreted as The higher values of the biodiversity indicator (FBI) are located in
a range between less favourable forest quality (low values indicate low mountainous areas (Alps, Dinaric Alps, Carpathian Mountains and
biomass value compared to the mean biomass values of its respective Pyrenees), with other small patches in the Ligurian region and southern
forest type) to 100 (where higher values indicate higher biomass value part of the Iberian Peninsula (Fig. 3C). Low values are found for forest
compared to the respective mean). plantations with exotic tree species as Eucalyptus and in the Mediter­
⎛ ⎞ ranean region.
Bi Bi The overlap of the three maps abovementioned by Eq. (6) resulted in
FF I = ⎝ ⎠*100/max( (5)
B B the forest condition map shows in Fig. 4A. This map indicates that the
areas with the highest forest condition values are located along the
Bi = Forest biomass value in pixel i. Lithuania-Belarus and Poland-Belarus borders, in mountainous areas,
B = mean of forest biomass value within EU forest type. especially the Carpathians, the eastern Alps, the Dinaric Mountains and,
⎛ ⎞
to a lesser extent, the highest parts of the Pyrenees. These are the areas
Max.is the maximum value of the function ⎝BBi ⎠. where forests show high species and habitat diversity as well as the core
forest provides a strong forest structure. On the other hand, the indicator
predicted low values of forest condition in the Atlantic regions, Britannic
2.4. Forest condition indicator Archipelago, Boreal and most of the Mediterranean biogeographical
regions (Fig. 4 and Fig. 5).
The Forest Condition indicator (FCI) was computed as the three
above-mentioned indicators by the Eq. (6). In fuzzy values ranging from 4. Validation
0 and 100, the result expresses the forest condition as a composite in­
dicator of the percentage of native species, the biodiversity, the coverage The forest condition indicator developed in this study was validated
and connectivity of forest, and the aboveground biomass. The values based on a multi-criteria evaluation and compared to other independent
close to 100 mean good condition whereas values close to 0 indicate European spatial data sets (see section 3). Concerning the statistical

4
A.I. Marín et al. Ecological Indicators 128 (2021) 107839

Fig. 2. Distribution of primary forest patches (Sabatini et al., 2018) and Natura 2000 sites used as validation sites. The map of habitat suitability of forest types in the
background (Casalegno et al., 2011).

analysis done against the primary forest patches (Fig. 5), the model mostly found within these areas, areas with low values of forest condi­
predicted coherently the condition values for most of the primary forest tion are very similar to unprotected areas.
identified as Hemiboreal forests, Alpine coniferous, Mesophytic decid­
uous and acidophilus forest, beech forest, mountainous beech forest and 5. Discussion
thermophilus deciduous forest. In the case of acidophylous oakwood,
broadleaved evergreen and coniferous Mediterranean, the number of The forest condition map presented in this work can be considered as
plots was insufficient to determine the significance of results. We cannot baseline to assess the European forest condition under the evaluation of
extract conclusions of the comparison with primary forest of those forest progress towards the attainment of sustainable and biodiversity goals.
types. We developed a method for constructing an operational indicator forest
However, Figs. 5 and 6 showed an underestimation of the forest ecosystem condition from open and public data sources at 1 km2 spatial
condition for the Mediterranean (coniferous and broadleaved) and resolution showing a general overview of the ecosystem at EU 27 and UK
Boreal forest types. This bias result is linked to the gaps in the data scale. The basic requirements for computing the condition of forest
sources but also to structural specificities of the primary forest within ecosystems include indicators for the horizontal structure, biodiversity,
these biogeographical regions. and functionality. Other forest characteristics could be added to enrich
To determine the relationship of ecosystem condition and the pro­ the condition assessment such as pressure indicators, dead wood, oc­
tection regimes, a test in/out was implemented to compare condition cupancy/dominance of invasive/introduced species, presence of pioneer
values of forest within Natura 2000 areas to the rest of forest extent in species/indicator species; soil properties, species composition; etc.
Europe (EU 27 and the UK). Fig. 7 shows the distribution of forest Many of these indicators are difficult to measure or quantify at the Eu­
condition values, categorized in five classes and separated between ropean scale due to a lack of consistent and harmonised data. Complex
Natura 2000 areas. The mean value of the condition indicator is labelled methodologies for data acquisition or index calculation derived in
as number that ranges from 0 to 100. The assessment indicates that, at challenging results with the risk to be un-operational.
EU scale, there are not significant difference between the mean values of The aim of this work is to provide an approach which is simple and
the forest indicator within (35.89) and outside (40.22) Natura 2000 sites spatially explicit, as strength for its implementation and usability, for
(t-value − 0.72 and p-value 0.47 of Student’s T Test). The highest mean producing a European map from harmonized and publicly available
values were in the Alpine coniferous and mesophytic deciduous forest datasets to measure progress/trends towards the achievement of the
types, the lowest values were in the broadleaves evergreen and conif­ CBDs and SDGs target. The data sources used in this study, after the
erous Mediterranean forest types. screening process, are derived data generated from original data after
The Natura 2000 sites host most of the areas with the highest values harmonization processes and protocols led by renowned institutions, e.
of forest condition (almost 70% of the area in the upper interval is under g., Intergovernmental Panel on Climate Change (IPCC) and JRC for
Natura 2000 protection). The percentage of forest area with an esti­ Biomass dataset or EEA for CLC dataset.
mated forest condition value lower than 20 points is almost equally The European forest condition map presented in Fig. 4 shows that the
distributed in the protected and non-protected areas across Europe. This areas with better conditions are located along the Lithuania-Belarus and
indicates that the protection policy under Natura 2000 sites seems to Poland-Belarus borders, in mountainous ranges, especially the Carpa­
have a limited impact. Although forests with the best condition are thians, the eastern Alps, the Dinaric Mountains and, to a lesser extent,

5
A.I. Marín et al. Ecological Indicators 128 (2021) 107839

Fig. 3. Maps for EU forest sub-indicators used for assessing the European forest condition.

the highest parts of the Pyrenees. This finding is also consistent with for some EU regions, e.g., the Mediterranean and Boreal regions.
previous results (Forest Europe, 2015; Kuuluvainen and Aakala, 2011; The bias in the Boreal area is linked with the specific dynamics of this
Sabatini et al., 2018). Mainly remote and rural areas host most of the forest type in natural and pristine conditions. Natural disturbances
forests in good condition since accessibility and the distance from acting over a range of spatial and temporal scales (large intense fires,
markets or other centres of demand is one of the main drivers of land-use low-intensity ground fires and insect outbreaks) are shaping the forest
allocation. Indeed, mountainous regions are key for European biological stand age structure at the landscape level (Kuuluvainen and Gauthier,
richness and its conservation (EEA, 2010). 2018).
The high values respect the functional forest elements in the sub The relationship and rate of young post-disturbance stages, along
mountainous areas like the southern part of Fennoscandia, and some with mature forest stages with old-growth characteristics is key to
Baltic and Balkan countries and are consistent with the results of Avi­ maintain an adequate share and the ecological qualities of these eco­
tabile and Camia (2018). systems. Fig. 6 shows a negative correlation between the forest condi­
In contrast, the low values of forest condition were identified in the tion value predicted in this study and the global condition reported in
Atlantic regions, Britannic Archipelago, Boreal, north-west of Iberian the assessment of the value of the Natura 2000 sites in boreal forests.
Peninsula and most of the Mediterranean biogeographical regions This means that the forest characteristic included in this assessment are
(Figs. 4 and 5). The low values of forest condition forest in Western not positively correlated with the “good” condition of the boreal forest
Europe were expected considering the accessibility and historically high as reported, which was the basic assumption in this multi-criteria
population density and high pressures from urban and transport infra­ approach. Values of the structure and function indicators close to the
structure expansion (EEA, 2018). Accessibility and population density maximum value (1 0 0) were not associated with “good” condition since
have been identifying as important spatial determinants for explaining the heterogeneous stand age distributions and landscape mosaics
the patterns of wood production and harvesting intensity in Europe resulted in intermediate values.
(Levers et al., 2014; Verkerk et al., 2015). Also, the north-western forest However, in the Mediterranean case, two aspects coexist in the
of the Iberian Peninsula is highly pressured by the Eucalyptus planta­ weakness index: gap of data information related to the tree species
tions for paper pulp production (Cerasoli et al., 2016). distribution and the structural specificities of the Mediterranean forest.
However, the final condition index presents a degree biased results Most central and northern European forests are dominated by around

6
A.I. Marín et al. Ecological Indicators 128 (2021) 107839

Fig. 4. A) European forest condition indicator. B) Zoom to Natura 2000 sites. Indicator shows as colour ramp where areas in dark blue mean high forest condition
values, whereas areas in red are poor forest condition. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of
this article.)

a dozen of tree species only. The Mediterranean forests are much more Additionally, Mediterranean forests show fundamentally different
diverse, as more than 100 different tree species can be found in this structural characteristics from temperate mesic forests, due to the high-
biogeographical region (EC, 2009). Those forests are exceptionally rich drought stress Mediterranean forests experience during the summer and
in endemic species: 201 out of the 290 woody species and subspecies due to fire disturbance (Karavani et al., 2018). These conditions may
(shrubs and trees) are exclusively or preferentially found under Medi­ hinder the development of structural features typically associated with
terranean bioclimates (Quézel & Médail, 2003). So, the biodiversity old-growth stages, as dense forest, unfragmented as was assumed in this
values calculated in this assessment underestimate the biodiversity for work (Burrascano et al., 2013; Kulakowski et al., 2017). This is a limi­
this zone. However, the biodiversity indicator can be considered tation found in other studies for modelling Mediterranean forest char­
adequately defined and formulated, although it could be further acteristics on large scale (Sabatini et al., 2018).
enhanced by adding other factors such as endemism or genetic vari­ The method presented requires readjustments in the sub-indicators
ability of individual species. The errors found for Mediterranean forests of functional and horizontal structure to integrate the aspects associ­
are linked to data limitations at the European level. The main data ated with the natural disturbances to which these forests are exposed
source for the biodiversity sub-indicator, European Atlas of Forest Tree and which form part of their condition, in a natural or non-managed
Species, provides limited datasets for the dominant tree species of regime.
Mediterranean forest type (e.g., Juniperus spp, Platanus orientalis, Quercus Additionally, it is to note that in areas dominated by forest planta­
coccifera, Quercus faginea, among others) (San-Miguel-Ayanz et al., tions with indigenous tree species, this index could show, due to a
2016). weakness of the algorithm, medium to high values as e.g., in Les Landes

7
A.I. Marín et al. Ecological Indicators 128 (2021) 107839

Fig. 5. Box plot graph of the distribution of forest condition values predicted within primary forest patches from Sabatini et al. (2018). See locations in Fig. 2. Forest
categorized by EFT (number of plots by EFT in labels).

Fig. 6. Forest condition relationships for sub-set of 56 Natura 2000 sites (see locations in Fig. 2) characterized according to the major habitat suitability of forest map
of Casalegno, et al. (2011). X-axis represents the global assessment of the site’s value for conservation of the natural habitat type concerned reported by the country as
part of the information provided in Natura 2000 standard data form (only the forest habitats were included in the assessment, http://natura2000.eea.europa.eu/). Y-
axis shows the forest condition indicator predicted in this study, extracted and averaged for the pixels corresponding to the Natura 2000 areas.

de Gascogne (southwestern of France) an indigenous species plantation the Timber Committee of the United Nations Economic Commission for
of about 1 million hectares of Pinus pinaster (Vallauri et al., 2012). This Europe (UNECE), among others, collect and collate statistics on the
limitation could be minimized by including additional information production and trade of wood through their Joint Forest Sector Ques­
related to the forest plantation and forest harvesting intensity. Eurostat, tionnaire (JFSQ). However, this information is mainly tabular, i.e., it is

8
A.I. Marín et al. Ecological Indicators 128 (2021) 107839

Fig. 7. Relative distribution of the forest condition values within and outside of the Natura 2000 sites per European forest type (in % the forests in each region). In
label the mean value of the forest condition indicator.

not spatially explicit data, being a gap for forest monitoring and analysis that, in our opinion, being based on relative indices can be transferred
as presented in this work. Recently, there are important advances by and replicated using other available datasets and providing rooms for
monitoring the harvest and management pressure by Earth observation updating ingesting emerging data sources. Currently, the combination of
techniques (Ceccherini et al., 2020). This can be an additional variable high-resolution satellite records and cloud-computing infrastructures
to include in a future enhanced version of this forest condition index, that can handle ‘big data’ provides a complementary asset for moni­
one the methodology is solid and consistent at EU scale. toring forest. In this way, the structural indicator that here was imple­
Natura 2000 sites are the focus of EU’s nature and biodiversity pol­ mented using as input CORINE land cover (updated every 6 year) could
icies. The aim of the Natura 2000 network is to secure the long-term be calculated by Pan-European High Resolution Forest Layers (HRL)
survival of Europe’s most valuable and threatened species and habi­ from Copernicus Program (updated every 3 year). Biomass estimations
tats. As part of this study, the role of the protection practices within the (data set 6) are based in a very simple approach (Barredo et al., 2012),
Natura 2000 sites was assessed by comparing the forest condition within being selected due to the limitation of available datasets for a consistent
and outside Natura 2000 areas. The assessment did not show statistically assessment year. However, there are being relevant scientific progresses
significant difference between the average values within Natura 2000 in terms of monitoring biomass at EU scale based on Earth observation
sites but almost 70% of areas with higher values of condition are found (Ceccherini, et al., 2020) that would allow a sophisticated and regular
within Natura 2000 sites. However, the percentage of forest area with updating of the Forest function sub-indicator. However, one of the key
low predicted values of forest condition is almost equal in and outside elements in forest condition monitoring is the biodiversity-related
protected areas, indicating that the protection policy that is carried out characteristic. The data sources here used (tree species distribution,
under the Natura 2000 umbrella has limited impact on the indicator. habitat suitability, table 1) are created based on forest national in­
Indeed, the latest EEA State of nature in the EU about the conservation ventories and other sources by modelling. In both cases, regular
status of species and habitats protected under the Habitat and Bird Di­ updating is not planned but they are not static variables. They are based
rectives indicated that the conservation status of forests habitats is not on probabilistic models and would be updated since the environmental
good in general (EEA, 2015c). conditions, that act as predictors, and the current species distribution
The map shows consistency, with good reliability, for a large area of (compiled by NFI) could change in the next future under the current
European forest ecosystems. The performed validation and reliability climate change framework.
assessment showed coherent pattern distributions with the relative The study demonstrates the strong demand for access to biodiversity
likelihood of primary forest occurrence map published by Sabatini, et al. status and trend data at multiple spatial scales to effectively inform
(2018). The predicted condition values are consistent with the primary policy and decision-making processes. Our results propose a way to
forests identified as Hemiboreal forest, Alpine coniferous, Mesophytic produce baseline to assess progress in the state and condition of forest
deciduous and Acidophilus forest, Beech forest, Mountainous beech biodiversity. It highlights the challenges in the updating of biodiversity
forest and Thermophilus deciduous forest. For these forest types, the indicators, e.g., when the needed data are not available as part of regular
indicator shows positive relationship with the conservation status re­ monitoring or inventory schemes (Biala et al., 2012; Han et al., 2014).
ported by the countries as part of the information provided in Natura Technical barriers such as access to data and information and large
2000 standard data form. differences in the scientific standards for monitoring and data analysis
However, the study here presented shows limitations regarding the may prevent insights on efforts towards improved and maintained
regular update and replicability of the datasets used as input for having biodiversity protection (Kühl et al., 2020; Aubin et al., 2020).
accurate and effective operational monitoring as aiming in this study.
This limitation affects to the index but not the methodological approach

9
A.I. Marín et al. Ecological Indicators 128 (2021) 107839

6. Conclusions Acknowledgments

The forest condition index (FCI) is the first European wide indicator This work is part of the support provided by the European Topic
developed so far to assess the condition of forest ecosystems at this scale Centre on Urban Land and Soil Systems (ETC/ULS) to European Envi­
considering a variety of factors including the structure and function of ronment Agency. The authors thank JI Barredo and M Erhard for the
this ecosystem and the biodiversity that depends on it. Good condition discussions and FM Sabatini and the Forest and CO project that facili­
will most likely result in an increased capacity to supply ecosystem tated the access to the datasets and provide support during the meth­
services. odology development. Finally, we want to acknowledge the anonymous
We estimated the relative condition of forest, from 0 to 100, into grid reviewers for their constructive comments, which contributed to
cell at 100 m spatial resolution. Although we recognize that the rela­ improving the manuscript.
tively coarse assumption of the main characteristics that drive the forest
condition may weaken the performance of our indicator, we focused our References
assessment on the operability of the indicator as well as the data
availability at large scale. Avitabile, V., Camia, A., 2018. An assessment of forest biomass maps in Europe using
harmonized national statistics and inventory plots. For. Ecol. Manage. 409, 489–498.
The results were validated against data derived from the EU Habitat https://doi.org/10.1016/j.foreco.2017.11.047.
Directive reporting obligations and a dataset on known primary forests Aubin, I., Cardou, F., Boisvert-Marsh, L., Garnier, E., Strukelj, M., Munson, A.D., Bello, F.,
in Europe. The map provided confident results for the assessment of the 2020. Managing data locally to answer questions globally: the role of collaborative
science in ecology. J. Veg. Sci. 31 (3), 509–517. https://doi.org/10.1111/jvs.
condition within Hemiboreal, Temperate and Alpine forests, showing v31.310.1111/jvs.12864.
the Carpathian, Dinaric Alps and Alps, among others, as good condition Barbati, A., Corona, P., Marchetti, M., 2011. Annex 1: Pilot application of the European
hotspots. However, the assessment showed weak results in the of Med­ Forest Types. Forest Europe, UNECE and FAO, pp. 259–273.
Barredo, J.C., San-Miguel-Ayanz, J., Caudullo, G., Busetto, L., 2012. A European map of
iterranean forest due to data limitations for these areas. The work pre­ living forest biomass and carbon stock - Executive report. EUR 25730 JRC77439, 16.
sented requires additional inputs for Macaronesian, Mediterranean and Biala, K., Conde, S., Delbaere, B., Jones-Walters, L., Torre-Marin, A., 2012. Streamlining
Boreal forest types representing singularities linked to species compo­ European biodiversity indicators 2020: building a future on lessons learnt from the
SEBI 2010 process. European Environment Agency, Copenhagen, Denmark.
sition, coverages, spatial pattern and ecosystem dynamics.
Burrascano, S., Keeton, W.S., Sabatini, F.M., Blasi, C., 2013. Commonality and variability
The results reported in this paper contribute to increase the knowl­ in the structural attributes of moist temperate old-growth forests: A global review.
edge and the implementation of biodiversity and environmental pol­ For. Ecol. Manage. 291, 458–479. https://doi.org/10.1016/j.foreco.2012.11.020.
icies. The approach and maps presented here, illustrate the integration Burkey, T., 1995. Extinction rates in archipelagos - implication for populations in
fragmented habitats. Conserv. Biol. 9, 527–541.
of multiple data sources, target-oriented, for constructing environmental Cardinale, B.J., Duffy, E., Gonzalez, A., Hooper, D.U., Perrings, C., Venail, P.,
information at European scale. This work provides underpinning infor­ Narwani, A., Mace, G.M., Tilman, D., Wardle, D.A., Kinzig, A.P., Daily, G.C.,
mation for the mid-term evaluation of progress towards the attainment Loreau, M., Grace, J.B., Larigauderie, A., Srivastava, D., Naeem, S., 2012.
Biodiversity loss and its impact on humanity. Nature 48, 59–67. https://doi.org/
of the Aichi Biodiversity Targets supporting the monitoring of trends 10.1038/nature11148.
and potential improvements of the forest diversity by restoration of Casalegno, S., Amatulli, G., Bastrup-Birk, A., Durrant, T.H., Pekkarinen, A., 2011.
degraded forest ecosystems. Additionally, it highlights the challenges in Modelling and mapping the suitability of European forest formations at 1-km
resolution. Eur. J. Forest Res. 130 (6), 971–981. https://doi.org/10.1007/s10342-
the updating of biodiversity indicators, e.g., when the needed data are 011-0480-x.
not available as part of regular monitoring or inventory schemes. Ceccherini, G., Duveiller, G., Grassi, G., Lemoine, G., Avitabile, V., Pilli, R., Cescatti, A.,
The maps presented in this paper are to be considered a first-tier 2020. Abrupt increase in harvested forest area over Europe after 2015. Nature 583
(7814), 72–77. https://doi.org/10.1038/s41586-020-2438-y.
approach to map forest condition that need to be defined using addi­ Cerasoli, S., Caldeira, M.C., Pereira, G., Caudullo, G., de Rigo, D., 2016. Eucalyptus
tional characteristics and factors, not only by the concept per se but also globulus and other eucalypts in Europe: distribution, habitat, usage and threats. In:
incorporating regional differences in the European forests. San-Miguel-Ayanz, J., de Rigo, D., Caudullo, G., Houston Durrant, T., Mauri, A.
(Eds.), European Atlas of Forest Tree Species. Publ. Off. EU, Luxembourg, pp. 90–91.
In summary, the presented forest condition map contributes to the
Chiarucci, A., Piovesan, G., 2020. Need for a global map of forest naturalness for a
assessment of current and coming changes that affect European forest sustainable future. Conser. Biol. 34, 368–372. https://doi.org/10.1111/cobi.13408.
ecosystems and supports the monitoring of post-2020 CBD targets as Chirici, G., McRoberts, R.E., Winter, S., Bertini, R., Brändli, U.B., Alberdi, I., Bastrup-
well as the SDGs. This information on forest condition, biodiversity and Birk, A., Rondeux, J., Barsoum, N., Marchetti, M., 2012. National Forest Inventory
Contributions to Forest Biodiversity Monitoring. For. Sci. 58 (3), 257–326. https://
the extent of forest degradation is essential for prioritizing human and doi.org/10.5849/forsci.12-003.
financial resources to prevent further degradation and restore and de Rigo, D., Caudullo, G., Houston Durrant, T., San-Miguel-Ayanz, J., 2016. The
rehabilitate degraded forest ecosystems. European Atlas of Forest Tree Species: modelling, data and information on forest tree
species. In: San-Miguel-Ayanz, J., de Rigo, D., Caudullo, G., Houston Durrant, T.,
Mauri, A. (Eds.), European Atlas of Forest Tree Species. Publ. Off. EU, Luxembourg,
CRediT authorship contribution statement pp. 40–47.
EC, European Communities, 2009. Natura 2000 in the Mediterranean Region. Office for
Official Publications of the European Communities, Luxembourg, p. 12.
Ana Isabel Marín: Conceptualization, Investigation, Formal anal­ EEA, European Environment Agency, 2019a. The European environment — state and
ysis, Methodology, Writing - original draft, Writing - review & editing. outlook 2020. Knowledge for transition to a sustainable Europe. 496 pp. https://doi.
Dania Abdul Malak: Conceptualization, Project administration, Su­ org/10.2800/96749.
EEA, European Environment Agency, 2019b. Mapping Europe’s ecosystems. Briefing no.
pervision, Writing - review & editing. Annemarie Bastrup-Birk: 19/2018. https://doi.org/10.2800/850732.
Conceptualization, Project administration, Supervision, Writing - review EEA, European Environment Agency, 2018. Forest dynamics in Europe and their
& editing. Gherardo Chirici: Conceptualization. Anna Barbati: ecological consequences. Briefing no. 16/2018, pp 4. https://doi.org/10.2800/
905921.
Conceptualization, Supervision. Stefan Kleeschulte:
EEA, European Environment Agency, 2016. Europe’s forest ecosystems – state and
Conceptualization. trends, EEA report 5/2016, p.132.
EEA, European Environment Agency, 2015a. EUNIS woodland, forest and other wooded
Declaration of Competing Interest land habitats, predicted potential distribution of habitat suitability (raster) - series.
https://sdi.eea.europa.eu/catalogue/eea/api/records/af43952c-393b-426f-8660-
01fccdf758eb.
The authors declare that they have no known competing financial EEA, European Environment Agency, 2015b. Linking in situ vegetation data to the EUNIS
interests or personal relationships that could have appeared to influence habitat classification: results for forest habitats. Technical report No 18/2015.
EEA, European Environment Agency, 2015c. State of nature in the EU. Results from
the work reported in this paper. reporting under the nature directives 2007-2012.
EEA, European Environment Agency, 2010. Europe’s ecological backbone: recognising
the true value of our mountains. European Environment Agency EEA Report N◦ 6/
2010, p. 252.

10
A.I. Marín et al. Ecological Indicators 128 (2021) 107839

FAO, Food and Agriculture Organization of the United Nations, 2018. The State of the floodplains. Biol. Conserv. 99 (2), 191–205. https://doi.org/10.1016/S0006-3207
World’s Forests, [Online] Available at: http:// www.fao.org/3/ca0189en/ca0189en. (00)00180-4.
pdf. McElhinny, C., Gibbons, P., Brack, C., Bauhus, J., 2005. Forest and woodland stand
FAO, Food and Agriculture Organization of the United Nations, 2017. Forest health. structural complexity: its definition and measurement. For. Ecol. Manage. 218 (1–3),
Accessed: 04/05/2017. Available at: http://www.fao.org/forestry/pests/en/. 1–24. https://doi.org/10.1016/j.foreco.2005.08.034.
Ferraz, G., Russell, G.J., Stouffer, P.C., Bierregaard, R.O., Pimm, S.L., Lovejoy, T.E., 2003. McRoberts, Ronald E., Winter, Susanne, Chirici, Gherardo, LaPoint, Elizabeth, 2012.
Rates of species loss from Amazonian forest fragments. Proc. Natl. Acad. Sci. 100 Assessing Forest Naturalness. For. Sci. 58 (3), 294–309. https://doi.org/10.5849/
(24), 14069–14073. forsci.10-075.
Forest Europe, 2015. State of Europe’s Forests 2015. In: Ministerial Conference on the Moreira, Francisco, Ascoli, Davide, Safford, Hugh, Adams, Mark A, Moreno, José M,
Protection of Forests in Europe, p. 312. Pereira, José M C, Catry, Filipe X, Armesto, Juan, Bond, William, González, Mauro E,
Garo, T., Hedblom, M., Emilsson, T., Nielsen, E.A., 2014. The role of forest stand Curt, Thomas, Koutsias, Nikos, McCaw, Lachlan, Price, Owen, Pausas, Juli G,
structure as biodiversity indicator. For. Ecol. Manage. 330, 82–93. https://doi.org/ Rigolot, Eric, Stephens, Scott, Tavsanoglu, Cagatay, Vallejo, V Ramon, Van
10.1016/j.foreco.2014.07.007. Wilgen, Brian W, Xanthopoulos, Gavriil, Fernandes, Paulo M, 2020. Wildfire
Han, Xuemei, Smyth, Regan L., Young, Bruce E., Brooks, Thomas M., Sánchez de management in Mediterranean-type regions: paradigm change needed. Environ.
Lozada, Alexandra, Bubb, Philip, Butchart, Stuart H.M., Larsen, Frank W., Environ. Res. Lett. 15 (1), 011001. https://doi.org/10.1088/1748-9326/ab541e.
Hamilton, Healy, Hansen, Matthew C., Turner, Will R., Jones, Julia A., 2014. Quézel, P., Médail, F., 2003. Ecologie et biogéographie des forêts du bassin
A Biodiversity Indicators Dashboard: Addressing Challenges to Monitoring Progress méditerranéen. Éditions scientifiques et médicales Elsevier SAS, Paris.
towards the Aichi Biodiversity Targets Using Disaggregated Global Data. PLoS One 9 Sabatini, F.M., Burrascano, S., Keeton, W.S., Levers, C., Lindner, M., Pötzschner, F.,
(11), e112046. https://doi.org/10.1371/journal.pone.0112046. Verkerk, P.J., Bauhus, J., Buchwald, E., Chaskovsky, O., Debaive, N., 2018. Where
Hooper, D.U., Chapin, F.S., Ewel, J.J., Hector, A., Inchausti, P., Lavorel, S., Lawton, J.H., are Europe’s last primary forests? Divers. Distrib. 24 (10), 1426–1439. https://doi.
Lodge, D.M., Loreau, M., Naeem, S., Schmid, B., 2005. Effects of biodiversity on org/10.1111/ddi.12778.
ecosystem functioning: a consensus of current knowledge. Ecol. Monogr. 75, 3–35. San-Miguel-Ayanz, J., de Rigo, D., Caudullo, G., Houston Durrant, T., Mauri, A., 2016.
Hunter Jr, M.L., 1990. Wildlife, Forests, and Forestry: Principles of Managing Forests for European Atlas of Forest Tree Species. Publication Office of the European Union,
Biological Diversity. Prentice-Hall, Englewood Cliffs, NJ, p. 370. Luxembourg, p. 198.
Karavani, A., Boer, M.M., Baudena, M., Colinas, C., Díaz-Sierra, R., Pemán, J., de Ruiz-Benito, Paloma, Vacchiano, Giorgio, Lines, Emily R., Reyer, Christopher P.O.,
Luis, M., Enríquez-de-Salamanca, Á., Resco de Dios, V., 2018. Fire-induced Ratcliffe, Sophia, Morin, Xavier, Hartig, Florian, Mäkelä, Annikki,
deforestation in drought-prone Mediterranean forests: drivers and unknowns from Yousefpour, Rasoul, Chaves, Jimena E., Palacios-Orueta, Alicia, Benito-
leaves to communities. Ecol. Monogr. 88, 141–169. https://doi.org/10.1002/ Garzón, Marta, Morales-Molino, Cesar, Julio Camarero, J., Jump, Alistair S.,
ecm.1285. Kattge, Jens, Lehtonen, Aleksi, Ibrom, Andreas, Owen, Harry J.F., Zavala, Miguel A.,
Kühl, Hjalmar S., Bowler, Diana E., Bösch, Lukas, Bruelheide, Helge, Dauber, Jens, 2020. Available and missing data to model impact of climate change on European
Eichenberg, David., Eisenhauer, Nico, Fernández, Néstor, Guerra, Carlos A., forests. Ecol. Model. 416, 108870. https://doi.org/10.1016/j.
Henle, Klaus, Herbinger, Ilka, Isaac, Nick J.B., Jansen, Florian, König-Ries, Birgitta, ecolmodel.2019.108870.
Kühn, Ingolf, Nilsen, Erlend B., Pe’er, Guy, Richter, Anett, Schulte, Ralf, Schulze, E.D., Mooney, H.A. (Eds.), 2012. Biodiversity and ecosystem function. Springer,
Settele, Josef, van Dam, Nicole M., Voigt, Maria, Wägele, Wolfgang J., Heidelberg.
Wirth, Christian, Bonn, Aletta, 2020. Effective biodiversity monitoring needs a Soille, P., Vogt, P., 2009. Morphological segmentation of binary patterns. Pattern
culture of integration. One Earth 3 (4), 462–474. https://doi.org/10.1016/j. Recogn. Lett. 30 (4), 456–459. https://doi.org/10.1016/j.patrec.2008.10.015.
oneear.2020.09.010. Thompson, I., Mackey, B., McNulty, S., Mosseler, A., 2009. Forest resilience, biodiversity,
Kulakowski, D., Seidl, R., Holeksa, J., Kuuluvainen, T., Nagel, T.A., Panayotov, M., and climate change: a synthesis of the biodiversity, resilience, stability relationship
Svoboda, M., Thorn, S., Vacchiano, G., Whitlock, C., Wohlgemuth, T., Bebi, P., 2017. in forest ecosystems, Technical Series No 43, Secretariat of the Convention on
A walk on the wild side: Disturbance dynamics and the conservation and Biological Diversity, Montreal, Canada.
management of European mountain forest ecosystems. For. Ecol. Manage. 388, Vallauri, D., Grel, A., Granier, E., Dupouey, J.L., 2012. Les forêts de Cassini. Analyse
120–131. https://doi.org/10.1016/j.foreco.2016.07.037. quantitative et comparaison avec les forêts actuelles. Rapport WWF-INRA, Open
Kuuluvainen, T., Gauthier, S., 2018. Young and old forest in the boreal: critical stages of Archive TOULOUSE Archive Ouverte, Marseille, p .69.
ecosystem dynamics and management under global change. For. Ecosyst. 5 (26), Verkerk, P.J., Levers, C., Kuemmerle, T., Lindner, M., Valbuena, R., Verburg, P.H.,
1–15. https://doi.org/10.1186/s40663-018-0142-2. Zudin, S., 2015. Mapping wood production in European forests. For. Ecol. Manage.
Kuuluvainen, T., Aakala, T., 2011. Natural forest dynamics in boreal Fennoscandia: A 357, 228–238. https://doi.org/10.1016/j.foreco.2015.08.007.
review and classification. Silva Fennica 45, 823–841. Vogt, P., Riitters, K., 2017. GuidosToolbox: universal digital image object analysis. Eur.
Levers, C., Verkerk, P.J., Müller, D., Verburg, P.H., Butsic, V., Leitão, P.J., Lindner, M., J. Remote Sens. 50 (1), 352–361. https://doi.org/10.1080/
Kuemmerle, T., 2014. Drivers of forest harvesting intensity patterns in Europe. For. 22797254.2017.1330650.
Ecol. Manage. 315, 160–172. https://doi.org/10.1016/j.foreco.2013.12.030. Winter, S., 2012. Forest naturalness assessment as a component of biodiversity
Mac Nally, Ralph, Parkinson, Amber, Horrocks, Gregory, Conole, Lawrie, monitoring and conservation management. Forestry 85 (2), 293–304. https://doi.
Tzaros, Christopher, 2001. Relationships between terrestrial vertebrate diversity, org/10.1093/forestry/cps004.
abundance and availability of coarse woody debris on south-eastern Australian

11

View publication stats