Evaluating the ecosystem-friendliness and suitability of potential future autochthonous tree species for ecosystem-based restoration

Introduction

With the adoption of Regulation (EU) 2024/1991 on nature restoration (NRR), the European Union has established a legally binding framework for restoring degraded ecosystems and halting biodiversity loss (European Parliament and Council of the European Union, 2024). A key objective is the restoration of forest ecosystems, for which the “proportion of stands dominated by native tree species” serves as an indicator of restoration success. The NRR defines a native tree species as one “that occurs within its natural past or current distribution range and its potential distribution range, i.e., the distribution range that it naturally occupies or could occupy without direct or indirect introduction or maintenance by humans”.

This definition, while seemingly straightforward, raises considerable practical and scientific challenges. First, the temporal reference for “past” distribution ranges remains unspecified. Should reference be made to pre-industrial conditions, early Holocene vegetation, or other benchmarks? This ambiguity is significant, as European vegetation has been shaped by continuous interactions between natural processes and human activities since at least the early Neolithic, making any “pristine” baseline without people elusive (Brown, 1997; Küster, 2013; Speier, 2006). Second, the concept of “potential distribution range” requires ecological projections under future climates, which are inherently uncertain and depend on assumptions about migration dynamics, dispersal capacity, and biotic interactions (Santini et al., 2021; Tsiftsis et al., 2024). Third, the regulation lacks criteria for determining which species could naturally colonize a region without human assistance, particularly given rapid climate change and landscape fragmentation.

These questions have direct implications for selecting tree species in restoration projects, funding eligibility, and the long-term composition and resilience of European forests.

The term “native” lacks a generally accepted definition. Species are often classified based on political boundaries rather than ecological criteria such as ecoregions (Lemoine and Svenning, 2022). Most definitions rely on two criteria: a species’ natural range at a specific past time and its distribution without human influence. Both are problematic. Reconstructing original ranges is methodologically difficult, and the distinction between natural and anthropogenic distribution is increasingly questioned, as nearly all ecosystems are influenced by human activities (Angermeier, 2000; Pereyra, 2020). Prehistoric human influences and global environmental changes have shaped current distribution patterns (Py-Saragaglia et al., 2025; Ren et al., 2021).

The perception of “native” species is also shaped by cultural ideas and emotional assessments (Estévez et al., 2015; Warren, 2023). Terms such as “foreign,” “exotic,” or “non-native” often have negative connotations and are sometimes accompanied by militaristic metaphors, which can unintentionally reproduce migration policy or colonial narratives (Heinrich et al., 2015; Penny and Beach, 2021; Warren, 2023; Guareschi et al., 2024; Wu et al., 2025; Tauli and Dellasta, 2026). The dichotomous native/non-native classification is increasingly criticized for failing to reflect ecological complexity and potentially hindering biodiversity restoration (Lemoine and Svenning, 2022). Introduced species can contribute to ecosystem resilience under certain conditions (Nackley et al., 2017), while native species can exhibit invasive behavior in altered environments. Consequently, more precise, neutral terminology - such as autochthonous and allochthonous - is recommended for biogeographical classification without normative evaluation (Soto et al., 2024).

Anthropogenic climate change is already profoundly impacting forest ecosystems. Since the early 20th century, Germany has experienced significant warming and increased frequency of extreme weather events (Debel et al., 2021; Rohde, 2026). Recent droughts have caused large-scale tree mortality, leading to the loss of over 900,000 hectares of forest (8.5% of the total forest area) since autumn 2017 (DLR (Deutsches Zentrum für Luft- und Raumfahrt), 2024). During the drought years of 2018–2020 alone, the loss of tree cover nationwide amounted to approximately 501,000 hectares, with some regions losing up to two-thirds of their coniferous forest (Thonfeld et al., 2022). These effects are amplified by bark beetle infestations, wildfire risk, and phenological mismatches, threatening forest integrity (Blumröder et al., 2022; Grünig et al., 2026; Thom et al., 2023).

To anticipate future forest dynamics, scientists rely on climate projections and species distribution models (SDMs) using scenarios such as RCPs and SSPs (Calvin et al., 2023). However, these projections are inherently uncertain. Current climate models struggle to capture non-linear feedback mechanisms - such as permafrost thaw, forest dieback, and ocean circulation changes - that can produce climate states with no historical analogue (Caesar et al., 2018, 2021; Goessling et al., 2025). Climate anomalies outpacing model predictions underscore these limitations (Binder et al., 2017; Hubau et al., 2020). Such “no-analog” climates will likely create novel species assemblages difficult to foresee (Gougherty et al., 2024; Williams and Jackson, 2007).

These uncertainties cascade into species distribution modeling. SDMs are typically calibrated on current climate data and have limited capacity to project into novel conditions (Chakraborty et al., 2024; Guisan et al., 2013). They often overlook critical non-climatic factors such as soil properties, microclimate, and biotic interactions (Meier et al., 2012; Weigel et al., 2023); assume homogeneous species responses, ignoring local adaptation; fail to capture extreme event impacts, which may be more decisive than mean climatic shifts (Ibisch, 2025; Stolz, 2023); and assume climatic equilibrium, ignoring demographic time-lags and biotic complexities (Wiens et al., 2009).

Migration processes themselves introduce major uncertainty. Postglacial migration rates, once estimated at 100–1,000 m/year from pollen data, are now considered significantly lower when accounting for genetic evidence and micro-refugia, with many late-successional species migrating below 100 m/year (Clark et al., 1998; Feurdean et al., 2013; Hošek et al., 2024; McLachlan et al., 2005), well below the 500–1,000 m/year required to track projected climate change (Li et al., 2006; Malcom et al., 2002). This “migration lag” is exacerbated by landscape fragmentation from agriculture, infrastructure, and urbanization, which creates artificial barriers restricting dispersal and gene flow (Fini et al., 2017; Ibisch et al., 2016; Mann et al., 2023).

Existing tools for climate-adapted species selection in Central Europe typically rely on a single dimension, either climatic suitability projected from species distribution models (Chakraborty et al., 2024; Mauri et al., 2023) or a binary native/non-native classification (European Parliament and Council of the European Union, 2014; Nehring et al., 2025). Lieven et al. (2025) published literature-based species profiles that already consider selected ecological aspects of candidate tree species, but only narratively and for a limited set of species. The framework proposed here aims to breach the gap between qualitative, single-species profiles and a systematic, comparable index across the full pool of candidate species, by combining three independent dimensions, biogeographic proximity, phylogenetic proximity, and ecosystem friendliness, into a single transparent, literature-anchored score, and by deliberately avoiding reliance on climate-projection uncertainty as the sole criterion for species selection.

This study distinguishes autochthonous species (autophytes and archaeophytes), which are considered native under the NRR, from allochthonous species originating elsewhere, a distinction that directly affects how the Regulation’s native-species indicator is assessed. Against this background, the present study proposes a novel evaluation framework integrating biogeographic, phylogenetic, and ecosystem-functional criteria to identify tree species with high potential for integration into forest ecosystems in Germany, including species not currently present in the country. While the present study takes EU legislation as baseline and focuses on a European context, the proposed methodology is designed to be applicable worldwide.

Methods

Concepts and terms

Terminology: autochtonous vs. native or autophytes, archeophytes, and “metecophytes”

In line with Barthlott et al. (1999), we define autochthonous species (autophytes) as species that have evolved within a given area or have migrated there without direct or indirect human assistance. In contrast, there are allochthonous species (allophytes), which originate from elsewhere. These are traditionally divided into archaeophytes - species spread by human activity before 1,492 AD - and neophytes, whose spread has increased significantly since around 1,500 AD as a result of global exchange. However, the frequently used cut-off date of 1,500 is historically conditioned and only applicable globally to a limited extent, which is why it seems sensible to focus more on the type of introduction.

On this basis, it is proposed that autophytes and archaeophytes be considered “native” species in the broader sense. Species from neighboring ecoregions that could potentially migrate without human assistance continue to be classified as neophytes or as potentially future species. For species from distant regions or regions separated by strong migration barriers, whose distribution is primarily due to long-distance human transport, the term neoterophytes is applied (Barthlott et al., 1999).

In addition, we suggest a new category: “metecophytes” (compare Ancient Greek: μέτοικος, métoikos: from μετά, metá, indicating change, and οἶκος, oîkos “dwelling”; metics were residents of Athens and some other cities coming from another polis). It includes species from neighboring regions that could migrate by themselves into the area under consideration in the future and become integrated into existing ecosystems. In the context of the EU Restoration Regulation, all species that are autophytes, archaeophytes or metecophytes are collectively considered to be potentially “native” under the terms of the Regulation.

The existing terminology does not adequately capture this category of species. Neophytes and neoterophytes, by definition, describe species already present in a region as a result of human introduction, and archaeophytes likewise describe species that arrived through human agency before a historical cut-off date. None of these terms identifies species that are biogeographically proximate and ecologically plausible future arrivals. The term “metecophyte” is proposed specifically to fill this conceptual and terminological gap, which is central to interpreting the “potential range” language of the EU Nature Restoration Regulation.

Ecosystem-friendliness

As “ecosystem friendliness” we understand the degree to which measures, land use practices, or species are compatible with or actively promote the functionality, resilience, and biodiversity of an ecosystem. The focus is not only on avoiding negative environmental impacts, but also on supporting key ecological processes such as nutrient cycles, trophic interactions, mycorrhizal symbiosis, habitat connectivity, or soil formation processes (Ansari et al., 2020; Hasan et al., 2021; Zhao et al., 2024). In practice to date, the term has been used primarily in technical or agricultural contexts, such as in the assessment of environmentally friendly materials, infrastructure, or production systems. These assessments are often qualitative and lack standardized ecological criteria (Agula et al., 2018, 2019; Chen et al., 2021; Hossain et al., 2015; Mallick et al., 2022; Mussadiq et al., 2023; Sha et al., 2021; Singh et al., 2023). For the restoration and development of ecosystems, a systematic assessment of ecosystem friendliness can be an integrative tool for comparing different measures, such as the selection of suitable plant species or the design of ecological structures. In this context, the concept of tree species ecosystem friendliness is introduced, which shall be underpinned by a semi-quantitative assessment. It refers to the ability of a tree species to integrate functionally into existing or restored ecosystems and support their development under an intensifying climate crisis, without causing drastic changes to ecosystem traits.

The species assessment is not primarily based on forestry criteria such as growth or habitat suitability, but rather on its integration into complex ecological relationships, for example regarding biodiversity, soil microbiome, humus formation, or water balance. Unlike traditional assessments that primarily evaluate potential invasiveness, this approach proactively identifies species’ traits beneficial for ecosystem functioning.

It is emphasized that “ecosystem friendliness” is not an intrinsic property of species, but rather an anthropocentric, factual assessment in the context of nature conservation and restoration goals, such as those pursued under the EU Nature Restoration Regulation (NRR). The starting point is the assumption that functional ecosystems and their services are highly valuable and should be specifically strengthened in view of the increasing pressures posed by climate change. Accordingly, those characteristics of species that can contribute to the stabilization of ecological processes, the promotion of resilience, and the regulation of micro, meso, and global climate are considered particularly relevant.

Framework for the identification of potential future autochthonous tree species (“metecophytes”) and the assessment of their suitability for restoration

Two key premises guide our approach: First, tree species not currently present in Central Europe will likely play a role in future forest ecosystems. Second, there is discernible interest, particularly among forestry actors, in utilizing allochthonous species. Rather than focusing on assumed “autecological” suitability or attempting to restore past ecosystem states, we prioritize the plausible effects of tree species within the ecosystem context. Considering the escalating climate crisis, it is particularly relevant to select species that (1) can plausibly integrate into existing and developing ecosystems, and (2) whose positive effects on ecosystem resistance and resilience are credible.

The framework presented here is based on three core dimensions: A. biogeographic proximity, B. phylogenetic proximity, and C. ecosystem friendliness. These dimensions are integrated into a composite assessment framework, allowing transparent and systematic evaluation of potential future native species (“metecophytes”) for inclusion in forest restoration measures under the NRR (Figure 1).

Selection criteria for the candidate species pool

The initial step in this assessment involves the identification of metecophytes suitable for inclusion in the analysis. Given current and projected climatic trends, including warming, an increased frequency and intensity of extreme events such as heatwaves and droughts, and overall greater climatic volatility, it is anticipated that tree species originating from neighboring southern and (south)eastern ecoregions adjacent to those found within Germany will, with varying degrees of probability, establish themselves in the future (Liepe et al., 2024; Pütz et al., 2024). The delineation of relevant neighboring ecoregions is based on the Ecoregions2017©Resolve dataset (Dinerstein et al., 2017), which provides a framework for ecologically and biogeographically coherent units. Within Germany, five ecoregions are present: the European Atlantic Mixed Forests (ID 664), Baltic Mixed Forests (ID 647), Central European Mixed Forests (ID 654), Western European Broadleaf Forests (ID 686), and Alpine Conifer and Mixed Forests (ID 689).

To identify potential source regions for metecophytes, the analysis was restricted to forest-dominated biomes across Europe and adjacent areas, with a primary focus on temperate broadleaf and mixed forests, as well as Mediterranean forests. To ensure that only actual forested areas were considered, the biome-based classification was spatially intersected with tree canopy cover data derived from the Hansen Global Forest Change dataset (reference year 2000; Hansen et al., 2013). A combined forest mask was subsequently created at a spatial resolution of 1 km, defining forest areas as raster cells exhibiting a tree canopy cover of 30% or greater.

The selection of potential future autochthonous tree species - referred to here as metecophytes - was guided by their natural distribution ranges and biogeographic proximity to the ecoregions of Germany. Species were considered eligible if their native ranges exhibit connectivity to German ecoregions via continuous or semi-continuous forest areas, thereby ensuring a realistic potential for natural migration within decadal timescales and ecological compatibility with existing species assemblages.

The temporal horizon for this analysis was set to 2050, aligning with the target achievement deadlines stipulated by the EU Nature Restoration Regulation. Longer-term projections were not pursued, as they would entail substantial uncertainties stemming from climate projections, estimates of migration rates, and the multitude of factors governing dispersal, including seed dispersal mechanisms, landscape structure, and biotic interactions.

It should be noted that the biogeographic proximity scores derived from this analysis are not predictions of actual immigration events for specific species, but rather proxy estimates of the relative likelihood of autonomous range expansion under the assumed conditions.

The identification of tree species started with the list provided by the European Forest Genetic Resource Programme (EUFORGEN, 2026). This initial list was subsequently supplemented with species from the European Atlas of Forest Tree Species (San-Miguel-Ayanz et al., 2016), along with additional targeted research on specific ecoregions. Biogeographic distribution data for most tree species were derived from the chorological maps for the main European woody species developed by Caudullo et al. (2017). Finally, to evaluate phylogenetic proximity and assess the ecosystem compatibility of candidate tree species, a targeted literature review was conducted for each species, focusing on their genetic and ecological characteristics.

For transparency and reproducibility, each species score is documented in the Supplementary material S1, S2 with its primary literature sources and a written justification. It is envisioned that the individual scores are updated and revised as new evidence becomes available. As the framework is designed as an evidence-anchored instrument rather than an expert-opinion assessment, formal inter-rater reliability testing was not conducted. Instead, each score is explicitly linked to its supporting literature, enabling independent verification of its evidentiary basis. This documentation provides a transparent audit trail and serves as the primary safeguard against scorer bias.

The framework is intended to be updated as new evidence becomes available, allowing individual scores to be revised accordingly. Because the framework is designed as an evidence-based instrument rather than an expert-opinion assessment, formal inter-rater reliability testing was not conducted.

Biogeographic proximity

Assessment of biogeographic proximity

Biogeographic proximity is particularly related to the plausibility that species will be able to migrate and colonize regions of the target area. As precise information on this matter is unavailable, we propose four proxies to facilitate an assessment. The first criterion concerns distance. It is assumed that species whose natural ranges are located at a greater distance from Germany are less likely to immigrate. The second criterion considers the similarity of ecoregions. It is assumed that species originating from ecoregions with climatic conditions similar to those of ecoregions in Germany will establish more successfully. It is assumed that the likelihood of natural migration to Germany is also influenced by the degree of forest fragmentation and the presence of barriers, which is assessed in criterion three. The fourth criterion addresses the dispersal potential of tree species, which affects the speed of migratory movement and the likelihood of overcoming fragmentation and barriers. The four criteria are summarized in Table 1.

Table 1. Criteria and scoring scheme for assessing the biogeographic proximity of allochthonous tree species.

Forest fragmentation within ecoregions was quantified using landscape metrics computed in R (R Core Team, 2023) following the FRAGSTATS framework (Hesselbarth et al., 2019). Derived from a 1-km forest raster, including Number of Patches (NP) and Largest Patch Index (LPI), and additional class-level indicators stored in an attribute table indexed by ECO_ID. While NP was selected for cartographic visualization. For geographic proximity analysis, forest ecoregions were converted to vector polygons and projected into an equal-area metric coordinate system (EPSG: 3035). For ecoregions consisting of multiple spatially separated polygons, the nearest polygon was identified, and a point on its surface was determined as the geographic centroid. Distances between this centroid and those of other European forest ecoregions were calculated in the metric coordinate system.

Spatial connectivity was analyzed using a Voronoi (Thiessen) polygon approach, defining a connectivity index as the ratio between the area of a contiguous forest patch and the area of its assigned Voronoi cell. Higher values indicate stronger spatial integration and lower isolation.

Climatic similarity between ecoregions was assessed using hierarchical cluster analysis based on seasonal climate profiles derived from Copernicus ERA5-Land data (Muñoz Sabater, 2019), consisting of seasonal mean air temperatures and seasonal precipitation sums. These variables were spatially averaged within each ecoregion and standardized. Pairwise distances between standardized climate vectors were calculated to create a dissimilarity matrix, visualized as a dendrogram showing relative climatic similarity patterns.

Calculation

Biogeographic proximity is calculated as the mean of the four criteria values and assigned to categories:

• Category 1 (Very low biogeographic proximity): Species originate from distant ecoregions with low climatic similarity, face severe dispersal barriers and high forest fragmentation, and possess low dispersal potential, making natural migration highly unlikely (≤1.49).
• Category 2 (Low biogeographic proximity): Species originate from moderately distant ecoregions with low climatic similarity, encounter significant barriers and marked fragmentation, and have limited dispersal ability, substantially hindering natural migration (1.50–2.49).
• Category 3 (Moderate biogeographic proximity): Species originate from ecoregions within moderate distance with similar climates, face moderate barriers and fragmentation, and possess moderate to high dispersal potential, allowing for gradual range expansion (2.50–3.49).
• Category 4 (High biogeographic proximity): Species originate from nearby ecoregions with high climatic similarity, face few barriers and low fragmentation, and possess high dispersal potential, enabling natural migration and successful establishment (≥3.50).

Phylogenetic proximity

Phylogenetic proximity is considered an indicator of potential functional and ecological similarity. Under rapid environmental change, increased genetic and ecological connectivity can contribute to adaptive capacity. Phylogenetically close taxa are expected to be ecologically more similar, although exceptions may exist (Cadotte et al., 2017).

The phylogenetic relationship between candidate allochthonous species and the autochthonous tree flora of Central Europe is evaluated using a four-point scale based on taxonomic relatedness and, where available, molecular genetic evidence:

• Category 1 (no phylogenetic relationship): The allochthonous species belongs to a genus not represented by any autochthonous species in the study area.
• Category 2 (distant relationship): The allochthonous species belongs to the same genus as autochthonous species but to a different species group.
• Category 3 (close relationship): The allochthonous species is a sister species to an autochthonous species, with frequent hybridization reported. Molecular genetic data confirm membership in a common species group or species swarm.
• Category 4 (very close relationship): The allochthonous species belongs to a species complex that also includes an autochthonous species. Hybridization is very frequent, and cryptic species are possible or likely. Molecular genetic data demonstrate very high phylogenetic proximity.

The phylogenetic proximity score contributes to the overall Tree Suitability Index (TSI) by providing an evolutionary perspective on the likelihood of functional integration into Central European forest ecosystems (Table 2).

Table 2. Criteria and ordinal scoring categories for assessing phylogenetic relatedness between allochthonous and autochthonous tree species.

Ecosystem friendliness

Justification of assessment criteria

Four main criteria were selected based on their central importance to the functionality and stability of forest ecosystems.

Ellenberg indicator values (Ellenberg et al., 1992) characterize the realized niche of individual species along abiotic site gradients (light, temperature, moisture, soil reaction, nutrients) and are designed for site classification rather than for evaluating a species’ functional contribution to ecosystem processes. Because the criteria used here assess process-level ecosystem effects rather than site preferences, class thresholds were instead derived from criterion-specific ecological literature, documented individually for each species in the Supplementary material S1, S2.

Habitat function

Reflects the capacity of a tree species to provide habitat structures, food resources, and interaction opportunities for autochthonous organisms. Structural diversity and provision of microhabitats for pollinators, fungi, arthropods, and other animal groups increase ecological relevance (Baldrian, 2017; Holzschuh et al., 2012; Kitching, 2000). The absence of toxic or growth-inhibiting substances is also significant in terms of compatibility with existing biodiversity (Burghardt and Tallamy, 2015; Slippers et al., 2005).

Soil-ecological effects

This is a central trait, particularly related to long-term site quality and nutrient cycle stability. Tree species influence humus formation, nutrient availability, and soil structure through litter quality, rooting patterns, and interactions with soil organisms. Symbiotic relationships with soil microorganisms and contributions to water balance regulation are key functions (Doty, 2011; Hofmann et al., 2023). As soil serves as a central resource for all higher trophic levels, this criterion is essential for ecosystem friendliness (Baeten et al., 2018; Baldrian et al., 2022).

Microclimatic regulation

Crucial for buffering climatic extremes and maintaining favorable physical conditions within forest stands. Tree species with pronounced structural and phenological suitability for climate buffering contribute to protecting sensitive organism groups and stabilizing biotic interactions. Shading and evapotranspiration regulation through canopy structure are particularly important (Norris et al., 2012; Pretzsch, 2014; Blumröder et al., 2021, 2022).

Promotion of resistance and adaptive resilience

Describes the capacity of a tree species to contribute to the overall system’s resistance to disturbances such as drought, storm, pest infestation, or climate anomalies. By promoting natural succession processes and increasing adaptive capacity, tree species actively support long-term ecosystem development (Loreau et al., 2001; Baeten et al., 2018; Jactel et al., 2021). This criterion directly concerns long-term functional security and is particularly relevant given increasing climatic changes (Grossiord et al., 2015; Liang et al., 2016; Matevski et al., 2023).

Together, these four criteria form an integrative, differentiated assessment system that captures both immediate and long-term effects of tree species on central ecological processes (Table 3). A more detailed justification of individual sub-criteria is provided in the Supplementary material.

Table 3. Main criteria, sub criteria, and ordinal scoring categories for evaluating the ecosystem-friendliness of tree species.

Sensitivity analysis of ecosystem friendliness (sub-)criteria

To identify the main drivers of ecosystem friendliness and to assess the reliability of individual parameters, all 16 ecosystem-friendliness sub-criteria were evaluated for each of the 95 identified tree species (archaeophytes and metecophytes), yielding 1,520 values for analysis. For each species, an overall ecosystem friendliness score was also calculated.

First, the strength of the association between each indicator and ecosystem friendliness was examined using rank-based correlation tests (Spearman, Kendall). Subsequently, the stability of these relationships was tested through 2,000-fold repeated analysis with bootstrap resampling. Finally, indicators were examined for content overlap using correlation analysis, variance inflation factor, and condition number.

Calculation of ecosystem friendliness

Ecosystem friendliness is calculated using a two-step procedure based on four main criteria, each comprising four sub-criteria.

Calculation of main criterion values:

• For each main criterion, the values of its four sub-criteria are summed
• The sum is then divided by 4 to obtain the mean value for that main criterion

Calculation of overall ecosystem friendliness score:

• The four calculated mean values of the main criteria are summed
• The total sum is divided by 4 to obtain the overall ecosystem friendliness score

Classification into ecosystem friendliness categories:

The resulting overall score is assigned to one of four categories:

• Category 1 (potentially harmful): value ≤ 1.49
• Category 2 (low “ecosystem-friendliness”): value between 1.50 and 2.49
• Category 3 (moderate “ecosystem-friendliness”): value between 2.50 and 3.49
• Category 4 (high “ecosystem-friendliness”): value ≥ 3.50

Exclusion procedure

Within the ecological suitability assessment of allochthonous tree species, an alternative exclusion procedure was developed based on a combination of qualitative minimum requirements and aggregated assessment thresholds. The objective of this procedure is to enhance ecological safety when dealing with potentially problematic species without compromising the overall discriminatory power of the differentiated assessment system.

The underlying premise of this approach is that certain negative evaluations - particularly concerning biodiversity impacts and fundamental ecosystem processes - cannot be offset by positive characteristics in other aspects. To prevent such exceedance of ecological tolerance limits, three alternative exclusion rules were formulated, each of which independently triggers the classification “not ecosystem-friendly”.

Specifically, the following conditions serve as decisive knockout criteria, whereby a tree species is classified as unsuitable for near-natural forest ecosystems regardless of its overall score:

• A rating worse than 2.0 in the criterion “Compatibility with biodiversity” (corresponding to a severely negative impact on autochthonous species diversity) constitutes an immediate exclusion criterion. Species with such a rating cannot under any circumstances be considered ecosystem-compatible.
• Four or more individual ratings with the lowest category (1) across the entire criteria matrix indicate a pattern of multiple negative effects. This accumulation of severe deficiencies is deemed ecologically unacceptable, even if individual positive characteristics are present.
• A mean rating worse than 2.0 in either of the two central main criteria - “Habitat Function” or “Soil-Ecological Effects” - also leads to disqualification. Both criteria are considered fundamental pillars of ecological system performance and must not fall below this threshold without compromising the functional integrity of the ecosystem.

This set of rules ensures that species with significant negative effects in central areas of ecosystem functionality can be excluded regardless of isolated positive characteristics or moderate overall scores.

Uncertainty assessment of ecosystem friendliness evaluations

To conclude the assessment process, the degree of uncertainty associated with each tree species ecosystem friendliness evaluation was systematically recorded (Table 4). The objective is to make transparent the robustness of the underlying evidence and to contextualize the evaluation not only in terms of content but also regarding its scientific foundation. This ensures that the ecological effects of a species are always interpreted in light of the available data.

Table 4. Criteria for assessing the uncertainty of ecosystem friendliness evaluations.

The classification is based on a four-point scale:

• Category 1 (very high uncertainty): is assigned when no scientific studies, forestry literature, or reliable practical experience exists for the species under consideration - including for closely related taxa - and the assessment is based exclusively on plausible assumptions or expert estimates.
• Category 2 (high uncertainty): characterizes evaluations that rely predominantly on analogical reasoning, general ecological principles, or indirect evidence, with only very limited direct evidence available.
• Category 3 (medium uncertainty): is assigned when individual scientific studies or documented practical experience exist, but not for all sub-criteria; analogies may be used as Supplementary material.
• Category 4 (low uncertainty): is assigned when multiple scientific studies, forestry literature sources, or consistent practical experience are available for the majority of assessment criteria.

The evaluation process fundamentally prioritizes the inclusion of robust scientific evidence. At the same time, it is designed to be dynamic: should new empirical findings or further studies become available, the assessments are to be adjusted accordingly. In this way, the ecosystem friendliness evaluation remains evidence-based, transparent, and continuously updatable.

Suitability assessment of potential future autochthonous tree species (“Metecophytes”)

The Tree Suitability Index (TSI) provides a simplified yet informative approach for evaluating the suitability of potential future tree species. The index integrates three core assessment criteria: biogeographic proximity (A), phylogenetic proximity (B), and ecosystem friendliness (C) of a tree species.

Tree Suitability Index for assessing potential future tree species (Formula 1):

A + B + C = TSI (1)

Where:

• A = Biogeographic proximity
• B = Phylogenetic proximity
• C = Ecosystem friendliness
• TSI = Tree Suitability Index (composite score)

Based on the total Tree Suitability Index score, species are classified into six suitability categories. Species achieving scores between 12 and 11 are considered very suitable, while those with scores of 10–9 are classified as highly suitable. A score of 8–7 indicates the species is suitable for restoration measures. Species scoring 6–5 are categorized as cases requiring discussion, for which context-specific evaluation is needed. Scores of 4–3 result in classification as not suitable, and species with scores of 2–1 are considered clearly unsuitable for forest restoration.

The three core dimensions are combined with equal weighting. This choice reflects a precautionary, non-substitutable design, analogous to the exclusion criteria used within the ecosystem-friendliness dimension: a species should not be able to compensate for poor performance in one dimension through strong performance in another. To test the robustness of this equal-weighting assumption, we conducted a sensitivity analysis comparing species rankings under six alternative weighting schemes and 20,000 randomly sampled weight vectors spanning the full range of possible weightings. These weight vectors were sampled from a symmetric Dirichlet (1,1,1) distribution, allowing the analysis to explore the full A/B/C weighting space without favouring any of the three TSI dimensions (Supplementary material Section A.6). The six suitability categories reported below represent an equal-interval partition of the theoretically possible score range (3 to 12 points, given three dimensions each scored 1 to 4). This arithmetic threshold placement is not itself empirically derived, but is complemented by the external validation against the list of potentially invasive species maintained by the Federal Agency for Nature Conservation (BfN), discussed below. Invasive-species risk is not scored as an independent dimension of the TSI. It is instead captured through the impact-on-autochthonous-biodiversity sub-criterion within ecosystem friendliness and through the exclusion (KO) criteria described above.

Table 5 illustrates the complete calculation of the TSI for Fagus orientalis, the highest-scoring species in the study (TSI 11). The complete set of scores, together with the corresponding literature sources, is provided in Supplementary materials S1, S2. The four sub-criteria of biogeographic proximity (distance, climatic similarity, fragmentation/barriers, dispersal potential) average to 3.75, placing the species in Category 4. Phylogenetic proximity is scored 4, reflecting that Fagus orientalis belongs to the same species complex as the autochthonous Fagus sylvatica, with frequent hybridization and molecular confirmation of very high relatedness (Denk et al., 2024). For ecosystem friendliness, each of the 16 sub-criteria is averaged within its main criterion (habitat function, soil ecological impact, microclimatic regulation, and resistance and adaptive resilience), and the four resulting main-criterion means are themselves averaged to an overall value of 3.4375, placing the species in Category 3. Summing the three dimension scores gives TSI = 4 + 4 + 3 = 11.

Table 5. Complete TSI calculation for Fagus orientalis, showing every sub-criterion score and how the three dimensions combine to the reported composite score.

Results

Sensitivity and redundancy of ecosystem-friendliness criteria

Correlation analysis revealed a distinct gradation in the strength of associations. The strongest associations with ecosystem friendliness were shown by:

• Humus and nutrient cycling (ρ ≈ 0.74).
• Interaction with soil microbiome (ρ ≈ 0.74).
• Canopy structure and microclimate (ρ ≈ 0.69).

A second group with also strong associations included soil moisture regime (ρ ≈ 0.65), biodiversity compatibility (ρ ≈ 0.65), and soil chemistry (ρ ≈ 0.64). Moderate to weak correlations were exhibited by the remaining indicators, with climate resilience (ρ ≈ 0.179) showing the weakest association.

Bootstrap analysis confirmed the robustness of these findings. Humus and nutrient cycling (98.1%), interaction with soil microbiome (96.5%), and canopy structure and microclimate (86.2%) appeared among the five most influential indicators in the vast majority of resamples. In contrast, climate resilience (0.05% in Top 5) and adaptive capacity (< 3.00% in Top 5) proved to be unreliable predictors. Consequently they were removed from the assessment.

Sensitivity of rankings to dimension weighting

Because the three TSI dimensions are equally weighted, we evaluated the sensitivity of species rankings to alternative weighting schemes. Sensitivity analyses included six predefined weighting scenarios and a Monte Carlo analysis of 20,000 randomly sampled weight vectors spanning the full weighting space. Rankings were generally robust to moderate changes in weights: Spearman’s rank correlation with the baseline ranking exceeded 0.8 for most predefined schemes that retained all three dimensions and exceeded 0.7 for 79–92% of randomly sampled weight vectors. In contrast, rankings changed substantially when the weight of one dimension approached zero, with rank correlations declining to 0.33–0.42 and individual species shifting by more than 28 positions. Full results, including all six predefined scenarios and the Monte Carlo distribution, are provided in Supplementary material Section A.6.

Biogeographic proximity of candidate metecophytes

The assessment of biogeographic proximity identified a limited subset of species with a high likelihood of natural migration to Germany under current conditions.

At the highest proximity level (rank 4), no archaeophyte species were identified. In contrast, a total of 14 metecophyte species reached this category, including Acer tataricum, Arbutus unedo, Fagus orientalis, Fraxinus angustifolia, Fraxinus ornus, Juniperus excelsa, Juniperus foetidissima, Malus trilobata, Picea omorika, Pinus nigra, Pinus pinaster, Prunus laurocerasus, Syringa vulgaris, and Tamarix ramosissima.

Expanding the threshold to include rank 3 substantially increased the number of potentially immigrating species, 7 archaeophytes and 41 metecophytes.

Overall, these results indicate that species with the highest biogeographic proximity are predominantly metecophytes, whereas archaeophytes are only marginally represented in the top category.

Suitability of archaeophytes

The archaeophytes occurring in Germany were evaluated for their suitability for forest restoration (Table 6). Among the 17 archaeophyte species assessed, none was classified as very suitable. Nine species were classified as suitable candidates, demonstrating sufficient ecological compatibility and potential for integration into near-natural forest ecosystems. Seven species were categorized as cases requiring discussion, for which use in restoration measures should be evaluated case-by-case considering local site conditions. The common walnut (Juglans regia) was the only archaeophyte classified as not suitable, due to the production of allelopathic compounds (juglone, flavonoids, phenolic acids) that inhibit the growth of other plant species (Đorđević et al., 2022; Medic et al., 2021).

Table 6. Archaeophytes of Germany, sorted by their tree suitability index value (TSI).

A: Biogeographic proximity, B: Phylogenetic proximity, C: Ecosystem friendliness. The list is arranged according to the TSI. Species marked with an x were classified as unsuitable for use in near-natural forest ecosystems as part of the exclusion process. The color coding represents TSI suitability categories: dark green = very suitable (score 11–12), green = highly suitable (9-10), light green = suitable (7–8), yellow = cases requiring discussion (5–6), red = not suitable/unsuitable (1–4, or “x” = excluded via knockout criteria).

As outlined in the previous section, none of the archaeophytes found in Germany have a natural range located in the immediate biogeographic vicinity of Germany. Of the seven species with moderate biogeographic proximity, four were classified as suitable: Aesculus hippocastanum, Prunus cerasifera, Pyrus communis, and Morus nigra. The remaining three species Ficus carica, Laurus nobilis, and Olea europaea were classified as cases for further discussion, whose use requires a site- and context-specific case-by-case assessment.

Suitability and ecoregional distribution of metecophytes

The number of suitable tree species (TSI scores 7–12) per ecoregion ranged from 0 to 30 (Figure 2; see Supplementary Table B.5).

Ecoregions with the highest diversity of metecophytes were concentrated in Euxine-Colchic broadleaf forests (ID 665), which contained 30 suitable species, followed by Aegean and Western Turkey sclerophyllous and mixed forests (ID 785) (28 species), and Balkan mixed forests (ID 646) and Caucasus mixed forests (ID 650), each contained 22 suitable species. A further 12 ecoregions exhibited high suitability potential, each hosting more than 15 metecophytes: Northern Anatolian conifer and deciduous forests (ID 703), Pindus Mountains mixed forests (ID 801), Anatolian conifer and deciduous mixed forests (ID 786), Central Anatolian steppe and woodlands (ID 652), Crimean Submediterranean forest complex (ID 658), Illyrian deciduous forests (ID 794), Eastern Mediterranean conifer-broadleaf forests (ID 791), Eastern Anatolian deciduous forests (ID 662), Northeast Spain and Southern France Mediterranean forests (ID 799), Southern Anatolian montane conifer and deciduous forests (ID 804), Tyrrhenian-Adriatic sclerophyllous and mixed forests (ID 806), and Italian sclerophyllous and semi-deciduous forests (ID 795).

Conversely, only a small portion of the study area showed limited suitability for the ranked species. Ten ecoregions were found to contain fewer than five suitable species. However, of these ten ecoregions, the four found in Germany (Baltic mixed forests (ID 647), Central European mixed forests (ID 654) European Atlantic mixed forests (ID 664) and Alps conifer and mixed forests (ID 689)) should be excluded, as the small number of suitable species is due to the fact that the majority of species found in these regions already occur naturally in Germany. The remaining six ecoregions comprise Mediterranean conifer and mixed forests (ID 701), East European forest steppe (ID 661), Pyrenees conifer and mixed forests (ID 676), Carpathian montane forests (ID 692) and Sarmatic mixed forests (ID 679). Notably, the Southeast Iberian shrubs and woodlands (ID 803) were identified as having zero tree species within the suitable TSI range (7–12) based on the current classification.

Within the group of metecophytes (n = 83), the following suitability classifications were assigned (Figure 3; Table 7):

• Very suitable: 2 species
• Highly suitable: 15 species
• Suitable: 32 species
• Cases requiring discussion: 10 species
• Not suitable: 24 species

Table 7. List of potential future autochthonous tree species in Germany, sorted by their tree suitability index value (TSI).

A = Biogeographic proximity, B = Phylogenetic proximity, C = Ecosystem friendliness. The list is arranged according to the TSI. Species marked with an x were classified as unsuitable for use in near-natural forest ecosystems as part of the exclusion process. Coniferous trees are marked with *. The color coding represents TSI suitability categories: dark green = very suitable (score 11–12), green = highly suitable (9-10), light green = suitable (7–8), yellow = cases requiring discussion (5–6), red = not suitable/unsuitable (1–4, or “x” = excluded via knockout criteria).

Of the 14 metecophytes with high biogeographic proximity, only Fagus orientalis was classified as “very well suited.” Four species received a “well suited” rating: Acer tataricum, Fraxinus angustifolia, Fraxinus ornus, and Malus trilobata. The remaining nine species - Arbutus unedo, Juniperus excelsa, Juniperus foetidissima, Picea omorika, Pinus nigra, Pinus pinaster, Prunus laurocerasus, Syringa vulgaris, and Tamarix ramosissima - were classified as unsuitable due to potential negative ecological impacts.

When species with moderate biogeographic proximity are included, the scope of consideration expands to include an additional 41 species. Of these, five species were classified as well-suited: Corylus colurna, Quercus frainetto, Quercus pyrenaica, Salix apennina, and Tilia tomentosa. 18 species received the classification “suitable”: Acer heldreichii, Acer hyrcanum, Betula medwediewii, Carpinus orientalis, Celtis australis, Cercis siliquastrum, Fagus hohenackeriana, Frangula rupestris, Ostrya carpinifolia, Pinus heldreichii, Pistacia atlantica, Pistacia terebinthus, Platanus orientalis, Prunus lusitanica, Quercus cerris, Quercus ithaburensis, Quercus trojana, and Rhamnus alaternus. The remaining 18 species were assessed as unsuitable: Arbutus andrachne, Cedrus libani, Chamaerops humilis, Erica arborea, Erica scoparia, Juniperus oxycedrus, Juniperus phoenicea, Juniperus thurifera, Phillyrea latifolia, Picea orientalis, Pinus brutia, Pinus halepensis, Pinus peuce, Pinus pinea, Quercus coccifera, Tamarix africana, Tamarix gallica, and Tamarix parviflora.

Discussion

The primary aim of this study was to develop a novel framework for assessing potential future autochthonous tree species (“metecophytes”) in Germany, responding to the operational challenges posed by the EU Nature Restoration Regulation’s definition of “native” tree species. Rather than relying on conventional modeling approaches or static historical baselines, we propose an integrative assessment based on biogeographic proximity, phylogenetic proximity, and ecosystem friendliness.

The NRR’s definition of native species requires that a species occurs within its natural or potential distribution range without direct or indirect human introduction. The biogeographic proximity component of the TSI identifies species for which autonomous immigration is biogeographically conceivable. However, as acknowledged in the methods section, the practical likelihood of autonomous immigration for most candidate species within the 2050 timeframe is constrained by the severe fragmentation of forest habitats across Central Europe and the increasing climatic volatility that reduces establishment success even when seeds arrive. At the same time, most of the potential future autochthonous species are already present in Germany, and in some cases, planted at scale. Even though it is possible to trace the migration routes of some species and thus distinguish between natural and introduced populations, this will not be possible in all cases, making the question of distinction primarily an academic one. A high biogeographic proximity score should therefore be understood as an indicator of probability rather than a prediction of autonomous immigration during a set timeframe.

Our findings highlight distinct integration trajectories for different categories of future autochthonous species. Archaeophytes have been present in Germany for at least 500 years, meaning their long-term effects on local biodiversity are generally well-documented and understood. Our assessment shows that almost all evaluated archaeophytes were considered suitable, with the notable exception of the common walnut (Juglans regia), which was classified as unsuitable due to its production of allelopathic compounds that inhibit other plant species. However, it is noteworthy that most of these archaeophytes do not currently form extensive wild populations and exist primarily in controlled or cultivated environments due to low cold resistance. While progressive warming may facilitate their future spread, exceptional late frost events currently act as a natural mechanism to reset their expansion.

Metecophytes, on the other hand, represent a large pool of species that have the potential to establish themselves successfully in the target area on their own due to shifting climate envelopes. The species with the highest probability to migrate successfully in the near future are the species with the highest biogeographic proximity (rank 4). In our assessment we identified 14 metecophytes: Acer tataricum, Arbutus unedo, Fagus orientalis, Fraxinus angustifolia, Fraxinus ornus, Juniperus excelsa, Juniperus foetidissima, Malus trilobata, Picea omorika, Pinus nigra, Pinus pinaster, Prunus laurocerasus, Syringa vulgaris, and Tamarix ramosissima.

A critical question guiding our framework is whether highly proximate species can also be considered ecosystem-friendly. To address this, we applied our Tree Suitability Index (TSI) alongside a set of exclusion criteria, which yielded a clear dichotomy among the candidate species. The following taxa were classified as ecosystem-friendly and suitable: Fagus orientalis (TSI 11), Fraxinus angustifolia (TSI 10), Fraxinus ornus (TSI 9), Malus trilobata (TSI 8), and Acer tataricum (TSI 8). In contrast, the species Juniperus excelsa, Juniperus foetidissima, Picea omorika, Pinus nigra, Pinus pinaster, Prunus laurocerasus, Tamarix ramosissima, Syringa vulgaris, and Arbutus unedo were excluded from consideration. Despite their high likelihood of natural migration, these taxa triggered our strict exclusion criteria, which include negative impacts on autochthonous biodiversity or the presence of multiple critical deficiencies.

The TSI results show a notable pattern: virtually all species classified as unsuitable are either Mediterranean conifers or Mediterranean sclerophytes. This is not a coincidence of scoring, nor an artifact of normative preference for the Central European mixed forest. It reflects a genuine and empirically well-documented functional mismatch between the ecology of Mediterranean conifers and sclerophytes and temperate forest ecosystems, operating across at least four interdependent mechanisms: 1. Soil chemistry and litter quality, 2. Mycorrhizal ecology and soil biotic networks, 3. Microclimate buffering and 4. Compatibality with the autochthonous biodiversity.

The needle litter produced by Mediterranean conifers is characterized by high C: N ratios, high lignin content, and low base cation concentrations (Bonanomi et al., 2021), driving a self-reinforcing acidification cycle, in which slow decomposition under acidic conditions reduces base cation return to the topsoil and further suppresses decomposer activity and nutrient cycling (Růžek et al., 2021). This dynamic is well-documented and detectable both at local and landscape scale (McTiernan et al., 2003). Long-term solute monitoring in Central German forests showed that conifer stands generated up to fivefold higher nitrate and sevenfold higher sulphate leaching than adjacent beech stands, with base cation return via litterfall only half as large (Růžek et al., 2019), and a 200-plot study in forest patches across German found the deciduous-to-conifer ratio to be the single most influential management variable for soil biological functioning and nutrient cycling, even more important than harvesting volume or deadwood (Klein-Raufhake et al., 2025).

Apart from litter chemistry, Mediterranean conifers and sclerophytes differ from Central European broadleaf species in their mycorrhizal ecology in ways that constrain integration into existing forest soil networks. All excluded conifers, as well as Arbutus unedo, form obligate ectomycorrhizal associations form obligate ectomycorrhizal (EcM) associations with specialist fungal partners that differ substantially from those recruited by the deciduous EcM trees dominating Central European forests (primarily Fagaceae and Betulaceae). The remaining excluded sclerophytes Prunus laurocerasus, Syringa vulgaris, and Tamarix spp. associate primarily with arbuscular mycorrhizal fungi. Common-garden experiments demonstrated that tree mycorrhizal type significantly shapes soil fungal community composition as well as soil fauna abundance, biomass, and functional diversity. A study by Peng et al. (2022) for example showed that soil fauna communities under species with low litter quality and EcM associations were consistently smaller and less diverse than those under arbuscular mycorrizhal (AM) or high-quality-litter EcM species (Peng et al., 2022). Litter decomposition experiments across nine temperate tree species confirmed that broadleaved EcM, broadleaved AM, and coniferous EcM trees each recruit fundamentally different microbial communities during decomposition, governed by different proportions of deterministic and stochastic assembly processes (Tanunchai et al., 2023). This suggests that introducing Mediterranean conifers into existing Central European forest ecosystems would not simply add nodes to an existing soil network, rather it would restructure it.

The microclimatic dimension of ecosystem friendliness captures a further axis of functional mismatch that has particular relevance under global warming. Long-term meteorological monitoring across 14 forest types in Switzerland demonstrated that broadleaved and non-pine conifer forests buffered daytime temperatures approximately twice as effectively as pine forests, which additionally showed negative effects on relative humidity at nighttime (von Arx et al., 2012). A multi-site study spanning a 1,800 km latitudinal gradient in Europe confirmed that the proportion of broadleaved trees in the canopy is a significant driver of understory microclimate buffering independently of forest density (Díaz-Calafat et al., 2023). This buffering capacity is not merely a climatic amenity, but is mechanistically coupled to soil moisture retention, which is itself determined in part by litter properties and canopy water interception. Broadleaf litter retains more moisture than needle litter, contributing to the soil moisture that in turn drives the evapotranspiration-based cooling of the understory (Greiser et al., 2024). Mediterranean conifers and by analogy sclerophytes with similarly drought-adapted canopy and litter properties, therefore impair microclimate buffering both through their canopy architecture that transmits more radiation to the soil surface (von Arx et al., 2012), and their litter that reduces the soil moisture that would otherwise sustain evaporative cooling (Greiser et al., 2024). As extreme summer temperatures become more frequent in Central Europe, this buffering function will grow in ecological importance, since forests maintaining cooler and moister understories provide thermal refugia for shade-dependent species and dampen the physiological heat stress on tree seedlings during regeneration (De Frenne et al., 2019; Schnabel et al., 2025). Species that structurally impair this function represent not only a present ecological mismatch but a reduction in ecosystem resilience capacity for a warming future.

The combined effects of soil acidification, mycorrhizal restructuring, and reduced microclimate buffering have clear consequences for understory plant diversity. A literature review by Barbier et al. (2008) demonstrated that coniferous forests consistently support less diversity of vascular plants in understories than broadleaved forests across the temperate zone, with the principal mechanistic pathways being reduced light transmittance, acidifying litter, and lower soil nutrient availability (Barbier et al., 2008). These patterns are directly confirmed by large-scale empirical surveys in Germany, where a 150-plot study across the country found that beech-dominated stands harbored significantly higher taxonomic richness across multiple taxa (plants, birds, moths, mollusks, soil fungi, bacteria, and protists) compared to conifer-dominated stands (Springer et al., 2024). Therefore, the convergent negative scores for Mediterranean conifers and sclerophytes do not represent biased judgements but coherent assessments of ecological incompatibility. It is worth noting that the same would be true if the roles were reversed. Central European broadleaf species would score poorly on ecosystem friendliness criteria calibrated to sclerophyllous Mediterranean shrublands, because they are equally mismatched to that ecosystem’s functional structure. The pattern reflects the tight co-evolution between tree species functional traits and the soil biotic communities, decomposer networks, and microclimatic regimes of their autochthonous ecosystems.

General considerations regarding the methodological approach

The observed discrepancy between migration probability and ecological suitability highlights a significant methodological advancement of the present framework. Conventional approaches to the management of non-native species have historically been dominated either by assessments of invasion risk or by binary classifications based on the “native vs. non-native” dichotomy, neither of which adequately captures ecological complexity. Current strategies for assisted migration mostly rely on species distribution models (SDMs) and focus on the respective climatic fitness and expected growth potential of the taxa to be introduced.

The present framework goes beyond these limited perspectives. Above all, it increases the transparency of species assessment and broadens the spectrum of aspects to be considered in species selection. This is, of course, a falsifiable or adaptable approach. Should certain assessments or even the criteria themselves prove to be less relevant or unsuitable in light of new findings, they can be removed. It is equally conceivable and desirable that additional criteria be identified and applied. In any case, the approach promotes scientific progress and enriches the technical foundation for discussing species selection. Naturally, certain characteristics and criteria are stronger predictors of the assessment outcome than others, and it is important to bear their respective contributions in mind. For this reason, a statistical sensitivity analysis was conducted.

Methodologically, the assessment framework is based on a four-level ordinal scale. In this context, one might critically question the extent to which marginal variations in individual parameters could systematically influence the overall results. This objection appears legitimate but fails to recognize the primary objective of the methodology. The choice of an ordinal scale does not represent a methodological shortcoming but is a deliberate response to the inherent uncertainties. Although the present approach is not free of ambiguities, the assessment is based on three core dimensions-biogeographic proximity, phylogenetic proximity, and ecosystem compatibility-which are grounded in universally quantifiable ecological and spatial metrics. In the case of biogeographic proximity, established ecological descriptors-spatial distance, climatic similarity, degree of fragmentation, and specific dispersal potential-were used, which collectively allow for a robust approximation and have been validated in the literature. In contrast, the apparent precision of climate-model-based species distribution models (SDMs) would tend to mask the actual uncertainties rather than represent them scientifically adequately.

Given the degree of forest fragmentation across Central Europe and the increasing climatic volatility that affects establishment success even where dispersal occurs, fully autonomous immigration to Germany by 2050 must be regarded as unlikely for the majority of candidate species. The biogeographic proximity component of the TSI should therefore be interpreted as an indicator of biogeographic plausibility and ecological relatedness to forest conditions in Germany, rather than as a migration forecast. In principle, one could question why specific parameters or thresholds were chosen for all criteria used. Ultimately, the degree of precision always represents a methodological compromise. For example, regarding distance calculation, it is debatable why the distance to the nearest German ecoregion was not used as a benchmark. This aggregation represents a deliberate methodological simplification. However, since this approach is applied strictly and uniformly, the inherent consistency and comparability of the assessment are preserved. A more granular spatial differentiation would significantly increase the complexity of the procedure without generating decisive scientific added value for the identification of large-scale migration patterns at the supraregional level.

Since the study evaluates potential tree species under future climate conditions, it could be argued that an assessment based on projected future climates would be more coherent. However, climate projections are subject to such high levels of uncertainty on the spatial and temporal scales relevant for tree species assessments that their use would not increase but rather decrease the robustness of the results. The period 1995–2025 was deliberately chosen because it includes the warming that has already occurred and thus reflects a climate trend that can be observed in reality. The use of current climate data is therefore not a disregard for climate change, but a conservative and well-founded decision.

These uncertainties are also the reason why the assessment does not provide precise timeframes for the potential immigration of tree species. The classification of biogeographic proximity should be understood as a heuristic approximation. The categorization of dispersal potential does not serve to forecast specific migration rates, but rather to provide a comparative ranking of species relative to one another. As one of four equally weighted subcriteria of biogeographic proximity, it systematically contributes to the differentiation of biologically very distinct dispersal strategies without dominating the overall assessment. The qualitative distinction between dispersal strategies-anemochory (wind dispersal), zoochory (animal dispersal), hydrochory (water dispersal), and autochory (self-dispersal) is ecologically grounded and well-documented in the literature. Future iterations of the assessment framework should, however, examine whether more quantitative estimates of dispersal capacity-such as those based on diaspore weight, fruit type, and available bird migration data-allow for a more nuanced assessment.

The assumption that phylogenetic proximity to autochthonous species implies greater compatibility with existing ecological networks is not unsubstantiated speculation but is supported by extensive literature on phylogenetic community structure. Relevant studies are systematically recorded in the literature database. Nevertheless, it must be noted that specific studies are not available for all species. The expert opinion therefore consistently treats phylogenetic proximity as an indicator of plausibility, not as causal evidence. In this context, however, a look at reticulate evolution can provide important insights. Particularly for tree genera important to Central Europe, such as Quercus, Fagus, Tilia, and Pinus, introgression between sister taxa is well documented, which empirically supports the assumption of functional similarity in cases of phylogenetic proximity. Category 2, which represents mere genus affiliation without closer kinship, is deliberately assigned only a moderate value and thus contributes less to the overall index.

The concept of ecosystem friendliness is a novel concept developed within the scope of this study that has not yet undergone independent validation in scientific literature. Despite its novelty, the concept is fundamentally falsifiable. The criteria are operationalized, each individual assessment is substantiated with sources from the database, and new empirical findings are intended to lead to adjustments. The sensitivity analysis of the 16 subcriteria comprising the ecosystem compatibility assessment yielded the important finding that the indicators “climate resilience” and “adaptability” were statistically unreliable predictors of overall ecosystem integration. Although these properties are central concerns for restoration under climate change, scoring them consistently across species proved unfeasible with the available literature, resulting in high variance that would have introduced noise rather than a reliable signal into the TSI. This difficulty is compounded by the limited statistical power inherent to a sample size of fewer than 100 species, at which correlations below approximately p = 0.25 cannot be reliably distinguished from zero. The removal of these subcriteria therefore reflects the current state of the evidence base rather than a judgement on their ecological importance. The inconsistency in scoring is itself indicative of a broader gap in literature. While long-term species survival under projected climate conditions is extensively modelled using species distribution models, the species-specific mechanisms of physiological resilience, including drought recovery, hydraulic failure thresholds and phenological flexibility, are not yet documented across the full range of candidate species assessed here. As the base of evidence matures, the two excluded sub-criteria should be reintegrated into future iterations of the framework.

Although climate resilience remains a central focus in many studies on assisted migration and forestry plans, our results suggest that considering these characteristics alone is insufficient for ecosystem-based restoration. A species’ ability to survive droughts does not necessarily mean that it is capable of supporting soil microbiomes, promoting stable humus formation, or regulating the microclimate at the stand level. By shifting the analytical focus to the assessed traits that more consistently support ecosystem functionality, with soil ecological effects showing the strongest correlation with ecosystem friendliness (ρ ≈ 0.74), the proposed framework offers significant added value. It can help ensure that restoration efforts are not only aimed at the successful establishment of trees, but above all at those that plausibly contribute to the resilient development of the forest ecosystem.

The classification of many Mediterranean conifers as unsuitable may give the impression that the study is based on an implicitly normative model favoring the Central European mixed deciduous forest. This must be countered by the fact that this classification does not result from a value-laden bias, but directly from the empirical evidence. While there is ample scientific evidence of positive soil ecological effects and contributions to biodiversity for the deciduous tree species considered, the scientific literature consistently documents more unfavorable effects on soil chemistry as well as lower biotic compatibility with native forest ecosystems for Mediterranean conifers. Thus, the assessment should not be interpreted as an artifact of normative preferences, but rather as a reflection of the current state of ecological knowledge.

Finally, the question of the practical discriminatory power of the classification of metecophytes must be critically examined. Since virtually all assessed species in this category are already present in the study area due to anthropogenic activities, the empirical differentiation between independent immigration and naturalization from forest plantings proves difficult in many cases. However, this circumstance does not undermine the validity of the method, as it contributes to the establishment of a systematic framework for assessing the ecological suitability of allochthonous tree species for near-natural forest ecosystems. The actual presence of many metecophytes also enhances the practical relevance of the study: it provides a sound basis for decision-making regarding the planting, tolerance, or active management of species that are already an integral part of the forest landscape but for which a consistent ecological assessment standard has been lacking until now.

Robustness, adaptability, and global transferability of the method

The methodological robustness of the framework is based on its multi-criteria foundation and integrated safety mechanisms. The exclusion procedure serves as a crucial safeguard and ensures that serious known ecological risks - particularly those related to impacts on autochthonous biodiversity - cannot be mathematically offset by positive characteristics in other criteria.

External validation

Even though the specific numerical formulation of the thresholds (such as the limit of four critical individual ratings or an overall average below 2.0) is not empirically derived, it follows a conservative, precautionary logic in the interest of ecosystem protection. The substantive validity of the procedure is confirmed by a high degree of external consistency. Despite fundamentally different methodologies, the present framework reliably identifies all tree species classified as potentially invasive by the Federal Agency for Nature Conservation (BfN) as unsuitable, which underscores the robustness of the selected KO criteria. The agreement between the framework and results obtained through an entirely independent methodology provides evidence of external, criterion-based validation.

Weighting sensitivity

The sensitivity analysis reported in the Results indicates that the three TSI dimensions make complementary contributions to the index and that the resulting prioritization cannot be adequately represented by any single dimension alone. This provides empirical support for retaining biogeographic proximity, phylogenetic proximity, and ecosystem friendliness as distinct, jointly weighted components of the index, rather than as substitutable or optional criteria.

Limitations and transferability of the framework

It should be noted that a complete prospective validation is structurally unfeasible due to the temporal dimension of the processes under investigation - in particular, long-term ecological integration under changing climatic conditions. Since the scenarios to be assessed extend beyond the current planning horizon, this limitation remains an inherent feature of all existing assessment frameworks for area shifts.

A further limitation concerns the reliance on literature-derived ecological information. Literature availability is uneven across taxa and regions, meaning well-studied genera such as Fagus and Quercus are better represented in the primary literature than several Mediterranean and (south)eastern European taxa. This asymmetry could bias apparent confidence rather than true suitability. This risk is partly mitigated by the four-point uncertainty scoring system (Table 4), which flags species with limited direct evidence, and scores for species with low certainty ratings should be treated with particular caution.

Furthermore, the framework remains highly adaptable. The systematic documentation of uncertainties ensures that assessments can be dynamically updated as soon as new empirical findings become available, without altering the underlying logical structure. Future iterations could be further strengthened by validating the present classifications using long-term empirical field data, particularly through the monitoring of underground microbial interactions and the buffering capacities of the tree canopy for the highest-ranking metecophytes.

Although the present study uses the EU legislation - specifically the Restoration Regulation - as it’s starting point and focuses on Germany, the underlying methodology is designed for straightforward regional adaptation. The three core dimensions - biogeographic proximity, phylogenetic proximity, and ecosystem compatibility - are based on universally quantifiable ecological and spatial metrics. For any country or region facing shifting ecoregions, climate uncertainty, and the need for evidence-based ecosystem restoration, this framework can be applied to systematically evaluate candidate pools and identify region-specific metecophytes. Of the four sub-criteria of biogeographic proximity, distance and dispersal potential require no regional adjustment, as they are based on geometric calculations and species-level biological properties independent of geography. Climatic similarity and landscape barriers and fragmentation require recalibration of reference points and threshold values, but the underlying analytical procedures remain unchanged. Phylogenetic proximity requires only a different autochthonous species pool as the reference set. The sub-criteria of ecosystem friendliness are ecologically general in principle, yet it has to be acknowledged that sub-criteria and classifications were developed in the context of Central European forest ecosystems and may carry different relevance or require adjusted weighting in substantially different biomes. It is therefore recommended that these criteria be reviewed by regional ecologists before the framework is applied outside of temperate broadleaf and mixed forests, to ensure that its application rests on ecologically appropriate assumptions.

Management recommendations

The overarching goal across all contexts of restoration measures should be to support the development of structurally rich and functionally diverse ecosystems with a high capacity for ecological integration. This will support both biodiversity and ecosystem resilience under uncertain future conditions. Against the backdrop of the discrepancy between natural migration rates and climate change velocity, assisted migration is increasingly discussed as an adaptation strategy. Potential advantages include maintaining productivity, biodiversity, and ecosystem services (Chakraborty et al., 2024; Mauri et al., 2023). However, risks include maladaptation, adverse biotic interactions, and introduction of novel pathogens. Successful implementation requires coordinated strategies, regionally specific provenance selection, and improved diversity of planting material (Bucharova, 2017; Twardek et al., 2023).

The cumulative effect of anthropogenic influences fundamentally alters the context within which tree species migration occurs. Natural migration operates within increasingly fragmented landscapes subject to novel climatic conditions, challenging restoration approaches based on static historical baselines. The inherent unpredictability of complex ecological systems (Benincà et al., 2015; Nielsen et al., 2019) suggests that even well-designed assisted migration interventions carry irreducible uncertainty, arguing for precautionary approaches that prioritize ecosystem function and maintain maximum genetic and species diversity.

If it is deemed necessary to introduce tree species to certain areas, to speed up the expected immigration of metecophytes or to enrich ecosystems with archaeophytes, then taxa in suitability categories 1–3 should generally be appropriate for restoration measures. Category 4 represents borderline cases requiring context-specific evaluation and flexible management. Categories 5–6 are inappropriate for forest restoration. Context-specific recommendations range from allowing spontaneous succession in near-natural forests to targeted introduction of ecosystem-friendly pioneer species on strongly degraded soils.

The classification of uncertainty described in Table 4 indicates the degree to which individual species ratings rest on direct empirical evidence or to analogical reasoning or interference from closely related taxa. For species with uncertainty scores of 1 or 2, the available literature was insufficient to support direct scoring for most if not all sub-criteria. Hence, evaluations were necessarily based on general ecological principles or evidence from phylogenetically or functionally related species. While this approach is consistent with established practice in expert-based ecological assessment frameworks, it should be acknowledged that scores for species with low certainty carry inherently greater interpretive uncertainty than those underpinned by direct empirical evidence. Practitioners using the TSI for management decisions are therefore advised to consult the uncertainty ratings alongside the composite scores. Species with low certainty scores should be treated with particular scrutiny and, where feasible, site-specific empirical validation before deployment in restoration programs.

Forest ecosystem restoration does not depend solely on tree species selection; the entire ecosystem and appropriate management practices must be considered. “No-regret” options of an ecosystem-based approach include strengthening forest ecosystems through soil protection, preservation of hydrological and microclimatic buffering capacity, and reduction of anthropogenic disturbances. Preserving microclimatic buffering through maintaining canopy cover and complexity, limiting fragmentation, and securing local water balance may be more important than any other adaptation measure (Ibisch, 2022; De Frenne et al., 2021; Greiser et al., 2024).

Where introduction is necessary, sowing is preferable to planting, as it mimics natural regeneration, preserves greater genetic diversity, causes less site disturbance, and enables working on larger areas (Aleksandrowicz-Trzcińska et al., 2017; Pemán et al., 2017). Priority should be given to autochthonous species from regional provenances, with the six German occurrence regions providing an operational framework (Barsch et al., 2012). Provenance effects are significant for functionally relevant traits (Aspalter et al., 2025), yet within-population genetic variability generally exceeds between-population variability (Kempf et al., 2020; Schmeddes et al., 2024).

Conclusion

This study introduces the Tree Suitability Index (TSI) as a transparent, multi-criteria framework for evaluating the potential of allochthonous tree species, including archaeophytes and the newly defined metecophytes, to contribute to resilient forest ecosystems under climate change. By integrating biogeographic proximity, phylogenetic proximity, and ecosystem friendliness, the TSI moves beyond a binary native or non-native classification and offers an evidence-based alternative for operationalizing the native-species criterion of the EU Nature Restoration Regulation (Regulation (EU) 2024/1991).

Applied to 100 candidate species, the framework identified a limited number of species, including Fagus orientalis, Fraxinus angustifolia, Quercus frainetto, and Tilia tomentosa, that combine high biogeographic plausibility with favorable ecosystem effects in the case study area, while excluding most Mediterranean conifers and sclerophytes on the basis of documented functional mismatches with Central European forest ecosystems. These results provide forest managers and policymakers with a prioritized, reproducible basis for species selection in ecosystem-based restoration planning, and the underlying literature justifications (Supplementary materials S1, S2) allow individual scores to be revisited as new evidence emerges.

Future research should prioritize empirical validation of the TSI against independent restoration case studies, extension of the framework to other biogeographic regions, and a formal sensitivity analysis of the weighting scheme applied to the three core dimensions.

Generative AI statement

The author(s) declared that Generative AI was used in the creation of this manuscript. The authors would like to acknowledge the use of Mermaid.ai for the generation of Figure 1 and Appendix Figures B.1, B.2. Appendix Figure A.1. was created with the help of Claude.ai.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/ffgc.2026.1895173/full#supplementary-material

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: Xiaochun Wang, Northeast Forestry University, China

Reviewed by: Muhammad Khalid, Kean University-Wenzhou, China

Ni Kadek Erosi Undaharta, National Research and Innovation Agency, Indonesia

Emanuele Giachi, University of Florence, Italy