1As humanity faces an increasing number of ecological crises, our societies must organise themselves in such a way as to mitigate these crises, by adapting or modifying our activities to make them sustainable (CBD 2022). This dual objective of mitigation and adaptation is now regularly promoted in the public debate on the fight against climate change. However, it must also be considered in the light of the challenges of preserving biodiversity, given that the scientific community has highlighted the unprecedented scale and speed of the crisis facing living organisms in recent decades (IPBES 2019). There is now a strong scientific consensus on identifying the causes of the collapse of biodiversity, and this consensus is being taken forward by the Intergovernmental Platform on Biodiversity and Ecosystem Services (IPBES 2019). Among these causes, land use change, leading to the loss and fragmentation of natural habitats, is often considered to be one of the main sources of pressure on biodiversity (Foley, DeFries et al. 2005, Newbold, Hudson et al. 2015, Newbold, Hudson et al. 2016). The occupation of space by more intensive and homogeneous practices (urbanisation, agriculture or forestry) is detrimental to the proper functioning of ecosystems, which is based, in part, on the diversity of relationships between each component of biodiversity that forms the ecosystem in question. In France, this phenomenon is reflected in particular in the loss of more than 265,000 hectares of natural environments between 1990 and 2018, which have become urbanised areas, agricultural land or bodies of water (indicator from the French national biodiversity observatory, based on Corine Land Cover data).
2In an attempt to combat this phenomenon, France has been legislating for over 40 years to propose instruments for public action aimed at neutralising the environmental impact of human activities on its territory. In particular, the 1976 law on the protection of nature introduced a sequence aimed at avoiding, reducing and then compensating for the changes that works or development projects are likely to cause to the environment.
3This mitigation hierarchy requires all project, plan or programme developers to identify the environmental impacts of their activity, and in particular the impacts on species, habitats and ecological functions, then to implement the necessary measures to minimise these impacts by avoiding them, then reducing them, and finally to propose offset those which could neither be avoided nor sufficiently reduced. In practice, project developers rely on compensation operators, consultancies or associations to support them in this process.
4The application of the mitigation hierarchy subsequently evolved in line with reforms to impact studies, the establishment of a national doctrine and guidelines published by the Ministry of Ecology in the early 2010s. These largely reflected French regulations and stated that offset measures must demonstrate an ecological added value, i.e. a substantial improvement in the ecological functions performed on the site and, consequently, in its ecological status. Then, when the public authorities adopted the law for the reconquest of biodiversity, nature and landscapes in 2016, they confirmed that the implementation of the sequence must aim for "no net loss of biodiversity", establishing it as a major tool for both mitigating the impacts and adapting to that changes that our use of space implies.
5To justify "no net loss of biodiversity", the mitigation hierarchy must lead to the implementation of offset measures generating biodiversity gains equivalent to the losses caused by the projects despite the avoidance or reduction measures. These measures can take the form of creating natural habitats (e.g. planting hedges and along riverbanks, creating ponds, reforestation, etc.), restoring environments under pressure (for example, removing invasive exotic species, lowering the soil or removing embankments to reshape the topography of a plot) or changing management practices (abandoning management to set up senescence islands, which consist of leaving a forestry plot to evolve freely and without any exploitation objective despite the death of certain individuals; but also changing cultivation practices, converting to fallow or grassland, etc.). They must be carried out in functional proximity to the impacts caused by the project, i.e. they must be able to benefit the same populations of species or fulfil the same functions in the same ecological entity as the one impacted (for example, within the same network of natural habitats, or within the same water catchment area). It is then up to the decision-making authority (the departmental or regional prefect) to assess, on the basis of the appraisal work carried out by the decentralised governmental departments if these measures can be considered satisfactory. While the first two stages of the sequence should ensure that the impact of activities on the environment is minimised, some land use planning still takes place in natural areas, and helps to undermine the integrity of ecosystems by causing the loss and fragmentation of natural habitats. Offset measures must therefore help to compensate these impacts by following a gain strategy that is consistent with the objectives of preserving biodiversity.
- 1 In this article, biophysical integrity is understood as a measure of the degree of anthropisation o (...)
- 2 Context incorporating both the functioning of ecosystems and ecological connectivity, enabling the (...)
6In this article, we consider that this strategy must meet two conditions in order to achieve the objective of no net loss of biodiversity. Firstly, offset measures must be implemented in areas with a high potential for ecological gain, represented by the probability of being able to generate an ecological gain through compensatory actions. To this end, numerous authors have shown that sites where offset measures are implemented must be in a degraded state or at least have an unfavourable trend for biodiversity (Maron, Hobbs et al. 2012, Weissgerber, Roturier et al. 2019, Pope, Morrison- Saunders et al. 2021). Indeed, offset measures must make it possible to improve the biophysical integrity1 of the site by removing sources of pressure or restoring habitats and ecological functions, so that the site contributes more to local biodiversity after the measure has been implemented. Given this premise, offset measures should preferably be implemented in areas of low biophysical integrity, where the implemented actions can generate significant gains. Secondly, the measures must be part of a favourable ecological context around the site itself. A number of studies have highlighted the importance of the landscape context2 in the colonisation of sites by the species targeted by ecological actions and, in the case of compensation, the scale of the gains that can be achieved (Hodgson, Moilanen et al. 2011, Berges, Avon et al. 2020). Offset measures should therefore preferably be carried out within a landscape context that offers a high level of biophysical integrity and contributes to the connectivity of ecological networks to maximise the chances of success of the implemented actions.
7In recent years, the scientific literature in France has extensively documented the process of implementing offset measures at site level, highlighting the discrepancy between these assumptions and actual practice (Jacob, Pioch et al. 2016, Vaissière, Bierry et al. 2016, Bigard, Regnery et al. 2018, Devictor 2018, Weissgerber, Roturier et al. 2019). Most of this documentation work was based on local case studies, on the basis of which hypotheses about national trends were formulated. To complement this work, in this paper we draw on data on offset measures geolocated in France from 2017 onwards, enabling us to study more than 1,000 measures located throughout mainland France. We will therefore assess whether these are carried out on sites chosen to maximise the ecological gain, following a strategy that makes it possible to claim that the mitigation hierarchy effectively aims for a trajectory towards no net loss of biodiversity.
8The GéoMCE database, produced by the French General Commission for Sustainable Development (CGDD) and the Center for studies and expertise on risks, the environment, mobility and development (Cerema), is intended for government departments and aims to map, monitor and control the avoidance, reduction, offset and support measures prescribed for projects subject to various procedures under the Environmental Code. Measures are geolocated in several formats: single or multiple polygons, single or multiple lines, single or multiple points, or at centroid level, the municipality in which the measure is implemented. At present, only biodiversity-related offset measures are available to the public on Géoportail3, a web portal set up by the "Institut National de Géographie"(IGN), providing access to services for searching, downloading and viewing geographical data in France4. As it is the decentralised investigating departments (within the regions or departments) that enter the offset measures in GéoMCE on the basis of the information obtained from the examination of development project dossiers, the data is entered in a heterogeneous manner from one region to another (Gelot and Bigard 2021). This heterogeneity depends mainly on the time devoted to filling in the database and data entry practices (Gelot and Bigard, 2021, Table 1), but probably also on the size of the region and the economic activities taking place there. In the database’s current format, the link between the compensatory measure and the project it is intended to offset cannot be made. Consequently, this study does not assess the functional proximity of the offset measures in relation to the source of the impacts.
Table 1 - Regional breakdown of the number of ecological offset sites studied and their total surface area in hectares (data: GéoMCE database)
Region
|
Number of offset sites
|
Compensation surface area (in ha)
|
Auvergne-Rhône-Alpes
|
1172
|
2788.6
|
Nouvelle-Aquitaine
|
348
|
1177.6
|
Hauts-de-France
|
341
|
974.6
|
Grand-Est
|
304
|
2556
|
Occitanie
|
161
|
1226.25
|
Ile-de-France
|
138
|
193.74
|
Pays-de-la-Loire
|
136
|
163
|
Brittany
|
69
|
319.6
|
Centre-Val de Loire
|
62
|
671
|
Normandy
|
44
|
84.1
|
Provence-Alpes-Côte d'Azur
|
35
|
111.7
|
Bourgogne-Franche-Comté
|
29
|
94.2
|
Corsica
|
1
|
1.61
|
9Geolocalised offset measures are sometimes implemented on several distinct geographical entities. Each of these entities is then referred to as an offset site. The analyses carried out relate to the offset sites entered in single or multiple polygon format in the terrestrial domain of mainland France, over the period from 1 January 2017 to 15 March 2021, the start date of the study. The analysis therefore covers 2,840 offset sites for a total of 1,153 offset measures. Aberrant data or data that was clearly incorrectly entered were removed from the sample: for example, a measure concerning a multitude of compensatory actions carried out as part of a linear infrastructure project, but which had been entered at the scale of the entire project study area, which represented more than ten thousand hectares and did not therefore correspond to the actual size of the offset sites.
- 5 CARTNAT is an IUCN-supported (International Union for Conservation of Nature) project that has deve (...)
10In order to test the two conditions of (i) potential ecological gain depending on the site where the compensatory measure is implemented and (ii) proximity to sites connected to a network of sites with good biophysical integrity, we used the national mapping of "potential naturalness" in mainland France, CARTNAT (Guetté, Carruthers-Jones et al. 2018)5. This mapping draws on three aspects to qualify the naturalness of landscapes: "the biophysical integrity of land cover, which reflects how far a landscape is from what it would be without human intervention; the spontaneity of processes, which assesses the degree of human intervention in natural dynamics; and spatial continuity, which corresponds to the degree of landscape connectivity" (Guetté, Carruthers-Jones et al. 2021).
11For this study, we used two spatial layers of CARTNAT, that of biophysical integrity, and a combination of this same layer with that of spatial continuity, which we consider here as proxy variables for the two conditions necessary for the successful implementation of ecological compensation.
12The first map illustrates biophysical integrity (Figure 1). This is based on the concept of hemeroby, which provides a measure of the anthropogenic impact on landscapes and habitats using a scale in which the highest values (ahemerob) correspond to 'natural' or undisturbed landscapes, while artificial landscapes obtain the lowest values (metahemerob) (Paracchini and Capitani 2011). The authors used a set of data relating to land use to produce this map. The spatial resolution of the biophysical integrity layer is 20 metres.
- 6 For a detailed presentation of the data and methods used to develop the two maps, see the IUCN Fren (...)
13The second map is a model of ecological continuity based on the biophysical integrity of the land cover (Figure 1). The landscape spatial continuity map was produced using the "Omniscape" model (McRae, Popper et al. 2016). This model is based on the omnidirectional dispersion of ecological flows, which is derived from circuit theory. This tool uses an approach which does not require the definition of "core areas" and which has been developed specifically for the purpose of modelling landscape integrity on a large scale. As such, it is in fact generalisable and focuses on modelling connectivity on the basis of ecological flows and landscape integrity (Theobald 2010, Dickson, Albano et al. 2017). Connectivity modelling methods based on electrical circuit theory provide the possibility of modelling the connectivity of the entire landscape, not necessarily linked to "core areas" or the least-cost paths between core areas. This approach therefore appears relevant for identifying corridors of interest for several species, particularly those that disperse over greater distances (Breckheimer, Haddad et al. 2014, Krosby, Breckheimer et al. 2015). The more intact, large and well-connected the spaces are, the higher the index of continuity is. The two maps were produced at a scale of 1/20000th with a scale of 255 values6.
Figure 1 (a; b).
a. On the left, the biophysical integrity gradient map cross-referenced with offset sites in Savoie, and on the right, biophysical integrity data for mainland France. b. On the left, the map of spatial continuity cross-referenced with the same offset sites in Savoie, and on the right, spatial continuity data for mainland France.
Data: CARTNAT project and GeoMCE database.
- 7 The intersection between the two layers is generated using the raster, sf and exactextractr package (...)
14In order to assess the potential ecological gain of the offset sites, we cross-reference the CARTNAT biophysical integrity raster layer with the vector layer of offset sites downloaded from the Géoportail7 website. For each pixel intersecting an offset site, the corresponding surface area is weighted by the proportion of the pixel covered by the offset site. For each offset site, we therefore have the cumulative surface areas associated with each biophysical integrity value.
- 8 If the p-value is less than 0.05, we reject the hypothesis of equality between the medians. In this (...)
15To test whether ecological compensation is carried out in areas of weaker biophysical integrity, we compare the medians of the integrity scores of the offset sites with that of scores of the regional and national areas. On the basis of 2,000 pixel samples containing more than 300,000 biophysical integrity scores drawn randomly and without replacement at national level, we apply a non-parametric Wilcoxon-Mann-Whitney test. This method, known as Monte Carlo simulation, was chosen because of the large size of the groups to be tested (mainland France contains more than 1.3 billion pixels). The rank test allows us to compare the medians of biophysical integrity scores between two groups of samples8.
16To assess whether at least some of the offset sites have relatively low biophysical integrity, we have chosen to identify the score corresponding to the first decile on each site. To do this, we rank the scores for each site in ascending order and calculate the cumulative surface area for each biophysical integrity value. We then select the value for which 10% of the surface area of each site has a biophysical integrity score lower than or equal to this value. Having identified the score corresponding to the first decile of each offset site, we observe their distribution by considering the whole sample of our offset sites. Using the first decile, we have chosen to observe whether at least 10% of the surface area of the offset sites is located in areas of relatively low biophysical integrity compared with the median score observed in mainland France.
- 9 In order to obtain biophysical integrity data by region, we split the raster using the mask functio (...)
17Due to the heterogeneous way in which data is entered into the database by the regions, we also tested the statistical significance of differences between offset sites and regional territory9. A Wilcoxon-Mann-Whitney test was performed.
18To obtain an index for ecological quality at a landscape level around an offset site, we merge the biophysical integrity and spatial continuity layers. We then use a 1,000 meter buffer around each site, from which we remove the access to the offset sites. This buffer size is often used in ecology to analyse landscape effects because of its ability to include a range of different habitats (McGarigal and Marks 1995, Qi, Fan et al. 2017). If two buffers cross, their edges are merged to prevent them from being counted twice.
19In order to test whether the landscape of the offset sites has a higher ecological quality than the metropolitan median, we compare the ecological quality scores of the landscapes with those of the national territory using the same Monte Carlo simulation and by applying a non-parametric Wilcoxon-Mann-Whitney test again.
20To assess whether at least a section of the landscape of the offset sites have relatively high ecological quality, we try to obtain the score corresponding to the ninth decile of the landscape on each site. To do this, we rank the weighted scores for each buffer around each site in ascending order and calculate the cumulative proportion of the surface area covered by each new pixel. We then select the value for which 10% of the surface area of the landscape surrounding each site has an ecological quality score that is higher or equal to this value. Having identified the score corresponding to the ninth decile of each landscape around an offset site, we observe their distribution by considering the whole sample of our offset sites. Using the ninth decile, we have chosen to observe whether at least 10% of the surface area of landscapes around these sites is located in areas of relatively high ecological quality compared with the median score observed across mainland France.
21All spatial analyses are carried out on Rstudio version 2022.7.1.554.
22In our sample, covering the period from 2017 to March 2021, there is an average of 2.46 offset sites per offset measure, ranging from 1 site to 59 sites per measure. The standard deviation is 3.9. The total surface area of the offset sites is 10,337.18 ha. It ranges from less than a square metre to 1016 ha. The first quartile is 0.06 ha, the median 0.41 ha and the third quartile 1.9 ha. These results show that most of the offset sites in our sample represent very small areas. In many cases, these small sites are dedicated to the implementation of localised measures, such as the creation of ponds or the installation of hibernacula for reptiles and amphibians. Conversely, the sites with the largest surface area are dedicated to conservation management or the creation of senescence islands.
23By observing and comparing the distribution of biophysical integrity scores within the offset sites with those for the whole of mainland France, we can see that the offset measures are carried out on sites with relatively better biophysical integrity (Figure 2). Indeed, the median biophysical integrity score of the areas used for compensation is 145, while that of mainland France is 71 (Figure 2). Thus, 64% of the surface area of the offset measures is located in areas where the biophysical integrity score is higher than the French median. We even note that 40% of the surface area of the offset measures has a biophysical integrity greater than or equal to the "top 25%" in France, i.e. the third quartile on a national scale, corresponding to a score of 169.
24The results of the Wilcoxon-Mann-Whitney tests performed 2000 times give a p-value of less than 2.2e-16 for 100% of them, enabling us to reject the hypothesis of equality between the medians with solid evidence. Finally, we note that the median biophysical integrity score of the offset sites is significantly higher than that of mainland France.
Figure 2: Box plots of biophysical integrity scores for mainland France (left) and the offset sites, and a violin plot of their distribution (right).
In the figure on the left, the median is equal to the first quartile ( = 71). In the figure on the right, the first quartile is 71 and the median is 145.
Data: CARTNAT project and GeoMCE database.
- 10 The differences are significant for each test. The p-value is less than 2.2e-16 for 12 of the 13 re (...)
25For 9 of the regions tested, we obtained results similar to those for France as a whole, i.e. the offset sites are located in areas of greater biophysical integrity. Nevertheless, for 4 regions (Corsica, Provence-Alpes-Côte d'Azur, Normandy and Centre-Val de Loire), the differences show that the offset sites are in areas where biophysical integrity is significantly lower than that of the regional territory. However, these are the 4 regions with the lowest number of offset sites (1 single site for Corsica to 44 sites for Normandy)10
Figure 3: Box plots of biophysical integrity scores region by region.
The diagrams on the left (in orange) show the distribution of scores within the region, while those on the right (in green) show the distribution of scores within the offset measures present in the region.
Data: CARTNAT project and GeoMCE database.
26Our results show that 71.5% of offset sites have a proportion of their surface area in areas of low biophysical integrity - scores between 1 and 71 - which is less than or equal to the French median, and that 50% of offset sites have one tenth of their surface area in an area with a biophysical integrity score less than or equal to 16 (Figure 4). This helps to explain the high density of pixels with a similar score within the offset sites observed in Figure 2. Conversely, 28.5% of offset sites are almost entirely (at least 90% of their surface area) on sites of relatively better biophysical integrity than the median score for mainland France (Figure 4).
Figure 4: Box plot and violin plot of the distribution of the first deciles of the biophysical integrity scores for each of the offset sites.
The mean score of the first deciles for all the sites is 59 (green dot on the figure) and the median is 16.
Data: CARTNAT project and GeoMCE database.
27By observing and comparing the distribution of scores for the ecological quality of the landscapes around the compensations sites with those for the whole of mainland France, we can see that the offset measures are carried out in landscapes where ecological quality is lower (the median is 87) than that of the mainland overall (median is 94).
28The results of the Wilcoxon-Mann-Whitney tests performed 2000 times give a p-value of less than 2.2e-16 for 100% of them, enabling us to reject the hypothesis of equality between the medians with solid evidence. Thus, we note that the median score for the ecological quality of landscapes around the offset sites is significantly lower than that of mainland France (Figure 5).
Figure 5: Box plots of ecological quality scores at the scale of mainland France and the surrounding landscape of offset sites.
The median for France as a whole is 94, while the median for the landscapes surrounding the offset sites is 87.
Data: CARTNAT project and GeoMCE database.
29Our results show that a large majority of the offset sites have part of their landscape in areas of better ecological quality than the median score for metropolitan France (= 94). In 50% of cases, 10% of the landscapes around the offset sites are located in areas with ecological quality scores above 144 (Figure 6). However, there is a reverse trend, with 15% of offset sites located within a landscape, almost all of which is located in areas of relatively low ecological quality compared with the median score for metropolitan France.
Figure 6: Box plot and violin plot of the distribution of the ninth deciles of the ecological quality scores of the landscapes around each of the offset sites.
The average score for the ninth deciles of all the sites is 141 (green dot on the figure) and the median is 144.
Data: CARTNAT project and GeoMCE database.
30These results allow us to better understand whether the offset measures implemented in France since the adoption of the Biodiversity Act in 2016 follow a strategy aimed at maximising ecological gains.
31At a time when habitat loss and fragmentation have been clearly identified as major determinants in the biodiversity crisis, offset measures must be used as tools to move towards the ecological neutrality of developments. Irrespective of the administrative procedures to be fulfilled by the compensatory action, its design must aim to achieve a gain in ecological function that will benefit the species, protected or otherwise, that use the area in question, as well as the state of conservation of the natural habitats that characterise it. However, our analysis shows that most of the sites dedicated to compensation (64% of the surface area of the measures) are located in areas of greater biophysical integrity in France, and even (for 40% of the surface area of the measures) in the best quartile of the mainland. At regional level, the results are similar for 9 of the 13 regions most represented in our sample from the GéoMCE database. This choice must be questioned both in the light of the opportunity and the constraints involved in implementing offset measures for operators, as well as with regard to the ecological gain strategy imposed on them.
- 11 In around 60% of cases, for "protected species" exemptions and in 30% of cases for "water law" doss (...)
32The implementation of an offset measure is the result of a multi-factorial process that integrates the ecological objectives being pursued, the feasibility of the actions, and also the land opportunities in the area and the economic objectives of the project owner. In the majority of cases, offset measures are taken as part of an exemption for the protection of species or as part of a "water law" dossier. This is the case for more than 90% of the measures in our sample11. However, articles L163-1 to L163-5 of the French Environment Code, which define the approach for compensating for damage to biodiversity, state that these measures are compulsory in order to offset "planned or foreseeable damage to biodiversity caused by the implementation of a works project or by the carrying out of activities or the implementation of a plan, scheme, programme or other planning document", without ever limiting the approach to protected species, wetlands or aquatic environments. In practice, therefore, only a small proportion of biodiversity damage is targeted by the implementation of offset measures, and the ecological objectives pursued may have to compete with the specific requirements of each of the parties involved in land-use planning and development. For the project owner whose project requires the implementation of one or more offset measures, the choice is made in such a way as to satisfy the expectations of the decision-making authority (the prefect of the département or the region), while limiting the costs involved, even if this means putting pressure on the intermediary players (consultancy firm, environmental protection association) who will assist in implementing the ecological actions (Lucas 2009, Berté 2022). These observations may explain why the measures entered into GeoMCE largely consist of simple, tried and tested and relatively inexpensive actions (digging ponds, installing shelters, reopening overgrown areas), the ecological added value of which, however, seems to be out of step with the destruction and destruction and fragmentation of natural areas. The other parties involved in offset measures, from the State's investigating departments to the compensation operators, all have to take into account the difficulty of mobilising a unit of land, the heterogeneity of skills within the structures involved and local sensitivities, emphasizing the compensatory effort based on criteria that vary in space and time (Guimont, Petitimbert et al. 2018, Petitimbert 2018).
33These trade-offs help to explain, in part, why the offseting strategy is not focused solely on the objective of ecological added value. Should we therefore consider that project owners are abandoning the objectives of ecological gain in order to secure their projects by providing a primarily administrative response to the requirements of the decision-making authority?
34The fact that the main objectives of compensation are to maintain the populations of protected species and to preserve wetlands and watercourses reinforces the assumption that offset actions predominantly target specific components of an ecosystem that is likely to be altered. These actions therefore make it possible to improve, for example, certain parameters of a species' ecological niche to which the biophysical integrity assessment is not sensitive (such as the diversity of vegetation within the same stratum, the density of micro-habitats, etc.), without necessarily contributing significantly to no net loss of biodiversity. In practice, although it is possible to obtain an ecological gain in areas with good biophysical integrity, a large proportion of the measures taken target a specific parameter in the life cycle of a species or the presence of a wetland, without seeking to generate a gain in all the functions and habitats that make up the site.
35So while there may be occasional ecological gains in areas of good biophysical integrity, these findings are likely to call into question the overall strategy of environmental public action aimed at achieving no net loss of biodiversity. If we look at this objective at the scale of a project, should we not try to implement, where necessary, compensation actions that aim to maximise the gains in biodiversity over a large number of functions performed within natural habitats? Is it not on the basis of this condition that we can maintain biodiversity resilience in the face of disturbance, rather than simply maintaining a species or an area of wetland?
36By demonstrating that offset measures are mainly carried out in areas with the best biophysical integrity, we are highlighting the lack of ambition behind these very measures. This can be interpreted as a symptom of a process centred on the administrative objectives imposed on project owners in order to obtain authorisation for a project (Vandevelde 2013). Geolocalised data on offset measures in mainland France show that procedures for exemptions to species protection and, to a lesser extent, for the protection of wetlands and aquatic environments dominate the application of the mitigation hierarchy as a whole (Gelot and Bigard 2021). As part of the environmental assessment of their projects, project owners make a commitment to the decision-making authority to fulfil the conditions required by the regulations. In order to derogate from the protection of a particular species, for example, it is necessary to demonstrate that the conservation status of the population of this species will be maintained despite the implementation of the project. Justification of this kind, upstream of development operations, is in fact based on assumptions about population trends, backed up by proposals for specific measures that are often aimed at facilitating certain parts of the species' life cycle. In the decrees authorising the projects, the requirement to achieve results then relates to the implementation of these measures, rather than the demonstration of a functional ecological gain for the habitats of the targeted species.
37Our analyses also show that a significant proportion of offset sites include areas of lesser biophysical integrity over a small part of their surface area. Some areas even have very low biophysical integrity, and are, in most cases, very small. In these small areas, offset measures are necessarily localised, and generally consist of installing artificial features that promote the reproduction of certain species or creating or restoring specific landscape features, such as nesting boxes or ponds. By exploring the description and categories of measures in the GéoMCE database, we can see that many offset measures are in fact aimed at localised actions. This observation is corroborated by the median size of the offset sites, which is 0.41 ha in our sample, even though offset measures are often implemented over of several fragmented sites, for larger total surface areas.
38These factors allow us to refine our hypotheses concerning the project owners' offset strategy: could it be aimed at the creation or localized restoration of a habitat, targeting a specific component of biodiversity, and then supplementing this action with management actions with little ecological added value, depending on negotiations with the decision-making authority's investigating departments?
39If this is the case, there is a risk that this strategy will turn offseting into an action that is too localised, makes little contribution to biodiversity and is aimed primarily at meeting precise regulatory criteria. It may then be supplemented by actions with minor benefits to meet negotiated surface requirements. Stakeholders are then able to present a balance sheet of the surface area compensated, often compared to the surface area impacted by the project, but without demonstrating that the gain will be equivalent in quality and quantity to the loss. However, this equivalence must be assessed during the compensation sizing process, which aims to specify the characteristics of the measures to be taken to compensate for a loss estimated on the basis of an analysis of the initial state of both the impacted sites and the offset sites. In practice, however, only a minority of project owners propose offset measures based on a sizing method described in the project file (Truchon, Billy et al. 2020). Thus, despite these well-documented pitfalls, projects that do not demonstrate compliance with the regulatory objectives of ecological neutrality are authorised, in the absence of a movement in case law that could strengthen the preventive dimension of these measures (Lucas 2018).
40However, numerous studies evaluating the effectiveness of ecological actions for biodiversity emphasise that the gains are all the greater when the actions aim to restore a variety of ecosystem functions in degraded areas on a large scale (Benayas, Newton et al. 2009, Bullock, Aronson et al. 2011, Moreno-Mateos, Power et al. 2012). Conversely, observers of the implementation of offset measures in France have been documenting for years, on the basis of selected examples, the discrepancy between the technical foundations of the definition of measures and the ecological realities they are intended to compensate for (Bigard, Regnery et al. 2018, Devictor 2018). Our analysis, carried out on a large sample covering the whole of mainland France, reinforces this observation and clarifies its scope.
41In addition to the potential gain in biophysical integrity on the sites concerned, the compensation strategy should aim to develop measures in a degraded area within a favourable ecological landscape, i.e. one that has good biophysical integrity and a good contribution to ecological connectivity.
42However, analysis of our results shows that the majority of offset measures are carried out in landscapes with the least favourable relative ecological status in the region, with the median ecological status of compensatory measure landscapes being significantly lower than that of the country as a whole. However, in many cases, a small proportion of this landscape still presents a satisfactory ecological context, this time with a much higher median. These ambivalent results can be interpreted from two angles. Firstly, the relatively poor ecological state of the surrounding landscapes can be explained by the principle of proximity of offset measures, which requires the project owner to implement them in functional proximity to the impacts. In cases where the measurements are carried out for an infrastructure project, it is likely that many sources of pressure exist in the vicinity of this project, and therefore around the offset measures. However, this finding also strengthens the case for a compensation strategy that is ill-suited to achieving ecological gains, even though it emphasises that there are areas of lower ecological quality around the measures implemented. This result must be interpreted with caution. Indeed, to date, the GeoMCE database has not been populated in a uniform manner across the whole of France (Gelot and Bigard 2021). The Auvergne-Rhône-Alpes region accounts for the largest proportion of the measures recorded (41% of the sites in the database). Furthermore, according to our models, the biophysical integrity and ecological quality of this region is relatively better than the national level. Therefore, on the basis of these analyses alone, it is difficult to determine whether the presence of a proportion of the landscape with good ecological quality around the offset measures is the result of a strategy to promote gains within the sites or the effect of a favourable or particular ecological context within a region that is over-represented in the GeoMCE database. However, given the economic, land and technical constraints we have already discussed, we can assume that the landscape context is not actually taken into account in the strategy for implementing offset measures in France.
43In 2018, the French government formulated a new objective to move towards ecological neutrality: to achieve "Zero Net Artificialization" (ZAN in French, hereinafter ZNA in English) of land, by proposing that local authorities commit to a process of reducing the consumption of natural spaces and promote land restoration actions. This objective was enshrined in the 2021 "Climate and Resilience" Act, and is now binding on future land-use planning documents, in line with temporal objectives aimed primarily at reducing land use. Although its implementation is distinct, in its current provisions, from the mitigation hierarchy, this law includes in the town planning code a definition of artificialization that takes into account both the ecological functions of soil and the fact that even a partial alteration of these functions plays a role in the artificialization process. The combination of these two instruments is therefore likely to promote future land restoration initiatives aimed at improving the ecological functions of soils with poor biophysical integrity. To satisfy this hope, this new objective will have to be applied in such a way as to avoid the pitfalls observed in the implementation of the Mitigation hierarchy and call for all activities likely to contribute to one of the main causes of the collapse of biodiversity to be taken into account and then corrected, namely the intensification of human use of natural areas.
44In this article, we studied a large sample of offset measures carried out since 2017 in France and showed that these were carried out on sites with relatively good, or even very good, biophysical integrity compared with the national territory overall. Our results suggest that the compensation strategy is not primarily aimed at generating a significant ecological gain. Indeed, they seem to indicate that the strategy adopted responds to constraints that are probably linked to the administrative procedures that justify these measures, primarily to derogate from the protection of species or wetlands. To this end, offset measures are generally implemented on sites where a small surface area offers significant potential for improvement, and which are complemented by larger areas with little ecological value, which are likely to be the subject of management measures. However, the landscape surrounding the offset measures is itself often in a poorer ecological state than the rest of the area, which suggests that sites with greater potential for gains could be chosen close to the impacts. In the majority of cases, however, a small proportion of the landscape is in a relatively good ecological state, which can be conducive to the success of ecological actions, particularly if they are aimed at restoring ecosystems. However, our results do not allow us to state definitively that this finding is the result of a strategic choice on the part of project owners. In any case, the location of the offset measures is the result of various constraints, and in particular economic and land-related constraints, which seem to outweigh the ecological benefit of the actions, which must aim for no net loss of biodiversity.
45The issues raised in this study are echoed in other countries. Along with Germany, Australia, the Netherlands and the United States, France is one of the few countries to make data on offset measures available to the public. This initial material is essential to enable observers, whether they be researchers, members of civil society or government departments, to assess the objectives of no net loss of biodiversity. However, each of these systems suffers from a lack of completeness and accessibility of data, which makes this assessment complex (Kujala, Maron et al. 2022). Despite their limitations, they are nonetheless an essential tool for making a start, after years of international controversy over offset measures (Maron, Ives et al. 2016), on the large-scale documentation of the effects of public action to halt the collapse of biodiversity.