Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2026 May 18;40(5):e70329. doi: 10.1111/cobi.70329

Modeling seed germination data to meet biodiversity conservation needs in the Mediterranean

Diana María Cruz‐Tejada 1,2,✉, Efisio Mattana 2, Angelino Carta 1,3
PMCID: PMC13613545  PMID: 42151736

Abstract

Seed‐based conservation and restoration programs rely on sound knowledge in fundamental seed biology. However, the adequacy of current data on seed germination to meet biodiversity conservation needs remains largely unexplored. We assessed germination data availability and quality for 1495 Mediterranean species of conservation and restoration interest to determine how much germination information is currently available; whether existing data meet the requirements for thermal germination risk assessment; and which species or ecological features help explain current data gaps. To this end, we used phylogenetic logistic regressions and assessed the adequacy of the available data by modeling the species’ seed germination thermal responses. Germination data were available for 31% of species but adequate for thermal modeling for only 9% of them. Species with data were mainly ecologically widely distributed, produced physiologically dormant seeds, including species suitable for restoration, and had traditional uses. Thermal risk was not phylogenetically clustered, but it was linked to habitat, with species from forests and scrubland exhibiting higher thermal risk. Based on our results, we devised a framework to support conservation and restoration efforts that includes steps to improve data availability (e.g., data gap analysis) and to classify species according to their vulnerability to climate warming. These steps will help distinguish between species that are potentially suitable for restoration programs and those that should be prioritized for conservation and demonstrate how germination information can be modeled and translated into thermal risk assessments to support seed‐based conservation and restoration programs.

Keywords: conservation, data quality, germination, restoration, seed traits, thermal modeling, Características de las semillas, calidad de los datos, conservación, germinación, modelación térmica, restauración, 保护, 数据质量, 萌发, 恢复, 种子性状, 热响应建模

Workflow to assess the availability and quality of germination data and assign species' thermal risk in the Mediterranean. (a) Assessment of data availability, including the phylogenetic distribution of the number of records (i.e. available germination data); (b) data quality assessment; and (c) assignment of climate risk categories for species that pass quality criteria. The number and percentages of species with germination data available, extracted from germination data sources for the pool of characteristic species of Mediterranean habitats are shown in (a). The phylogenetic tree represents a wide phylogenetic dispersion, including the most representative clades. Numbers in (b) represent the number of species that passed the modeling criteria in turquoise squares, and the number of species that did not pass the criteria in turquoise circles, as indicated; NTestTemp, Model concavity, Tail drop (not flat), correspond to the modeling criteria applied to the data quality assessment. In (c) is visualized how the thermal risk is calculated, starting from the quadratic modeling representing the thermal seed germination and the assignment of the different risk categories.

graphic file with name COBI-40-e70329-g002.webp

INTRODUCTION

Seeds play an important role in plant adaptation to changing environments, contributing to the ecosystems’ integrity and resilience (Aurelle et al., 2022). They help plants reach suitable habitats through dispersal (Muñoz‐Gallego et al., 2023), survive unfavorable periods via soil seed banks (Gioria et al., 2020), and germinate when environmental conditions are favorable for seedling survival (Baskin & Baskin, 2014). Thus, studying plant regeneration from seeds is crucial for understanding species reproduction under current and future climate scenarios (Baskin & Baskin, 2022; Walck et al., 2011) and provides critical information for biodiversity conservation (Price et al., 2024; White et al., 2025) and ecological restoration (Dalziell et al., 2022; Ladouceur et al., 2018; Pedrini & Dixon, 2020).

Germination occurs as a response to a combination of environmental cues (Baskin & Baskin, 2014), such as temperature, moisture, light, and seasonal signals, which together constitute the seed germination niche (Donohue et al., 2010; Grubb, 1977). Within this niche, temperature plays a major role in controlling the timing and success of germination (Baskin & Baskin, 2014). The thermal germination niche, defined by the optimum temperature and the range of temperatures under which germination occurs, reflects a combination of fixed adaptive traits and plastic responses that allow seeds to cope with varying environmental conditions (Donohue et al., 2010). Thus, modeling this niche allows quantification and comparison of germination responses to temperature variation across species and populations under a warming climate (Cochrane et al., 2011).

Germination niche modeling requires adequate and robust data (Kildisheva et al., 2020; Price et al., 2024), particularly to increase species representativeness across habitats (Cruz‐Tejada, Mattana, et al., 2024). Increasing amounts of seed germination data are being archived (Cruz‐Tejada, Fernández‐Pascual, et al., 2024; Fernández‐Pascual et al., 2023; Ordóñez‐Parra et al., 2022). These data support broadscale syntheses of the germination niche across multiple species, especially for temperate biomes (Carta et al., 2022; Fernández‐Pascual et al., 2021), such as the Mediterranean (Cruz‐Tejada, Mattana, et al., 2024). However, it is not clear whether these data are also adequate for species‐level niche modeling because the outcomes of such modeling approaches are highly influenced by the total number of germination tests (data availability) and the quality of the data (e.g., number of germination tests across a thermal gradient). Moreover, the availability and quality of seed germination data are usually not evenly distributed among species. Widely distributed species occurring across multiple habitat types are more likely to be studied because their broad spatial distribution or environmental tolerance makes them more accessible for seed sampling and therefore better represented in open‐access repositories. For species of conservation interest or those relevant for restoration, seeds are also expected to be more likely studied due to conservation and restoration priorities (European Commission, 1992). As a result, much of the available germination information for these species is generated through ex situ collections and experiments conducted in conservation seed banks (e.g., Mattana et al., 2023), which represent a major source of empirical knowledge on species’ regeneration requirements that support conservation planning and comparative germination studies (Carta et al., 2022). Incorporating species‐specific germination requirements in decision‐making frameworks can substantially improve the success of conservation translocations (Godefroid et al., 2025) and seed‐based ecological restoration actions (Pedrini & Dixon, 2020).

Gaps in germination data have direct conservation implications (Bucharova et al., 2025). For species with complex reproductive cycles, including pollination and dispersal strategies and deep dormancy, seed collection can be challenging, and subsequent processing and testing may be limited (Fenner & Thompson, 2005; Joffre et al., 2007). In this context, it is also important to assess whether variation in data availability and quality is phylogenetically structured and thus reflective of shared evolutionary history rather than ecological differences because species are not evolutionarily independent (Garamszegi, 2014). Although previous studies have explored broadscale germination responses or thermal risk patterns in temperate floras (Carta et al., 2022; Fernández‐Pascual et al., 2021) and tropical regions (Sentinella et al., 2020), no one has yet examined the availability, quality, and species‐level adequacy for modeling of germination data for a large Mediterranean species pool.

The Mediterranean basin is a biodiversity hotspot due to its unique geographical and environmental characteristics (Myers et al., 2000). The same characteristics make it particularly vulnerable to rising temperatures (Giorgi & Lionello, 2008), and with summer temperatures in the northern Mediterranean increasing at up to twice the rate of global warming (RCP8.5 is projected to be ∼4°C by 2100) (IPCC, 2022), Mediterranean ecosystems are considered to be highly sensitive to climate change (Thompson et al., 2026). In this context, evidence from comparable climate‐sensitive floras shows that assessing the thermal germination niche (Hirst et al., 2025) and, when relevant, integrating additional criteria, such as genetic risk, provide practical foundations for identifying species’ vulnerability and guiding ecological restoration decisions (Mattana et al., 2026). In Mediterranean ecosystems, warming‐driven shifts in germination phenology may directly reduce recruitment opportunities, thereby challenging seed‐based conservation and restoration actions. This underscores that biodiversity conservation requires a deep understanding of species‐specific seed biology (Just et al., 2024), including their germination requirements and sensitivity to temperature. Inadequate integration of species’ reproductive and germination constraints can compromise conservation translocation success across Europe (Godefroid et al., 2025). Recent evidence also highlights that conserving Mediterranean biodiversity requires safeguarding the evolutionary history through phylogenetically informed, diverse, and better‐coordinated conservation strategies (Carta et al., 2025; Thompson et al., 2026).

Our goal was to assess the availability and adequacy of germination data for thermal modeling of characteristic species in the Mediterranean, representing an essential species pool for the design of effective policy and practices for habitat restoration and biodiversity conservation. By using phylogenetic logistic models, we tested the following hypotheses: seed germination data are limited in quantity and quality for species exhibiting particular vegetative (life form), reproductive (fruit type, pollination), and seed (mass and dormancy) traits, hampering seed collection, processing, and germination; data quantity and quality are not a constraint for species of wide ecological breadth or for species of conservation interest or those commercially available for restoration. For those species with adequate data for thermal modeling, we estimated the risk to seed germination of increasing temperatures (hereafter thermal risk) with the aim of improving understanding of species responses to future global warming scenarios and conservation management outcomes. Based on our results, we identified key challenges related to the limited availability and quality of germination data and outlined potential solutions to address these limitations. We also identified a framework to support seed‐based conservation and restoration programs in the Mediterranean.

METHODS

Species pool and germination data

We used the 1495 characteristic species of Mediterranean habitats from the EUNIS expert system (Chytrý et al., 2020) as the potential pool of species of conservation and restoration interest in the Mediterranean. This list of species was originally compiled in Cruz‐Tejada, Fernández‐Pascual, et al. (2024) and subsequently reused and taxonomically standardized in Cruz‐Tejada, Mattana, et al. (2024), where data on seed germination for species were also compiled. No species were excluded. Thus, no additional matching, duplicate filtering, or taxonomic reconciliation was required. All species names follow the World Checklist of Vascular Plants (Govaerts et al., 2021).

Data availability and quality predictors

We selected nine plant features as predictors of germination data availability and quality (Table 1). We focused on traits that are directly relevant to seed sampling, handling, curation, and germination experiments in conservation and restoration contexts. We gathered vegetative (i.e., life form), reproductive (i.e., fruit type and pollination), and seed (i.e., mass and dormancy) traits and species‐level information. This information, extracted from publicly available datasets and publications, included ecological breadth (i.e., number of habitats in which each species occurs) based on the EUNIS expert system (Chytrý et al., 2020); inclusion in European conservation policies (Habitats Directive); relevance for ecological restoration (i.e., commercially available for restoration [Ladouceur et al., 2018]); and reported plant uses (Table 1). The structure and coding of all predictors and their sources are provided in Table 1. Ecological breadth was calculated based on the number of habitats in which the species is present (i.e., three‐level hierarchical classification of European habitats [Chytrý et al., 2020]). When it was not possible to find information on reproductive and seed traits for a species, we used data from congeneric species. Species with missing trait information were assigned no information category when applicable. All predictors were analyzed in separate univariable phylogenetic logistic regression models. Traits that were conceptually categorical were encoded as binary dummy variables (0/1) and standardized prior to analysis. No multilevel categorical factors were fitted. Continuous predictors were analyzed as continuous variables and log transformed. The phylogenetic tree of the species pool and the trait distribution across lineages are in Appendix S1.

TABLE 1.

Predictors of seed germination data availability (presence or absence).

Feature Predictor Predictor type (unit) Hypothesis Source of predictor information
Vegetative

Life form

annual

perennial herbaceous

woody

 Binary (1, 0) Data availability for perennial herbs and annual plants is higher than for woody plants because perennials are easily found in natural areas and woody plants have complex reproductive cycles, often with either very large or fleshy dispersal units (Joffre et al., 2007).

www.FloraVeg.eu

Govaerts et al., 2021

Reproductive

Pollination

biotic

abiotic

 Binary (1, 0) Species with biotic pollination (i.e., animal‐mediated pollination) are less studied and have very little germination data due to low seed production (e.g., influenced by pollinator availability and specificity; compatibility and competition between individuals) (Ashman et al., 2004). Kattge et al., 2020

Fruit type

dehiscent

indehiscent

fleshy

 Binary (1, 0) Fleshy fruits often have complex morphological structures that involve additional treatments to allow germination, making them difficult to study in comparison with species with dehiscent and indehiscent fruits (Spjut, 1994). Carta et al., 2024
Seed Seed mass Continuous (mg) There is lower production of large seeds relative to small seeds; thus, smaller seeds are more likely to be studied because they are more available (Fenner & Thompson, 2005). Carta et al., 2024

Dormancy

PY(PD)

PD

M(P)D

ND

NoInfo

 Binary (1, 0) Germination data on species with deep and complex seed dormancy are not available due to the difficulty of experimentally breaking dormancy (Baskin & Baskin, 2014). Rosbakh et al., 2023
Ecological breadth Number of habitats** Continuous (n) Species restricted to one or few habitats are less studied because they are more difficult to sample. Chytrý et al., 2020
Conservation interest Species of conservation interest Binary (1, 0) Species listed in the annexes of the Habitat Directive are more studied than others because they are of conservation concern. European Commission, 1992
Restoration relevance Species relevant for ecological restoration Binary (1, 0) Species relevant for ecological restoration are more studied than others. Classification derived from the European restoration species pool proposed by Ladouceur et al. (2018)
Plant uses Species with traditional uses Binary (1, 0) Species with traditional uses are more studied than others (Pironon et al., 2024). Pironon et al., 2024
Open in a new tab

Abbreviations: PY(PD), physical and physical‐physiological dormancy (combined); PD, physiological dormancy; M(P)D, morphological and morphophysiological dormancy (combined); ND, nondormant; NoInfo, no information on seed dormancy reported. **The ecological breadth was calculated based on the number of habitats on which the species is present (i.e., three‐level hierarchical classification of European habitats [Chytrý et al., 2020])

We evaluated data availability across all species as a binary response (i.e., presence or absence of data for each species) with phylogenetic logistic regression models implemented in the r package phylolm (logistic MPLE) (Tung Ho & Ané, 2014). We fitted each predictor in a separate univariable model and extracted the alpha (α) parameter, which represents the strength of phylogenetic signal (i.e., the higher the value of α, the weaker the phylogenetic dependence). Then, we used multiple factor analysis (MFA) in the package FactoMineR (Le et al., 2009) to construct a multivariate ordination to visualize the cross‐covariance among our predictors of data availability for each species.

Using the function glmmTMB in the R package glmmTMB, we evaluated the adequacy of data for quantifying thermal germination responses at the species level with binomial mixed quadratic models (Sentinella et al., 2020) in which seed lot was a random factor (Brooks et al., 2017). To account for overdispersion, models included an observation‐level random effect and were fitted using a beta‐binomial error distribution. Zero inflation was not modeled because residual diagnostics did not indicate systematic excess zeros. Based on the outcome of these models, we were able to evaluate each species according to the thermal modeling criteria in Table 2. These criteria were based on the quadratic model parameters that reflected whether the species’ germination response followed the expected unimodal pattern with downward concavity. Uncertainty associated with derived thermal parameters (e.g., optimal temperature [T opt] and thermal breadth [T breadth]) was quantified using Sentinella et al.’s (2020) approach based on the delta method (see below). When derived thermal parameters or their uncertainty were not estimable (e.g., due to lack of a real solution at the selected germination threshold), values were denoted not applicable (NA). We then modeled data quality at the species level as a binary response (passed or failed quality assessment for each species)—that is, the data were either adequate or did not fit a reliable quadratic model (details in Table 2)—by fitting phylogenetic logistic models (logistic MPLE) (Tung Ho & Ané, 2014) in the same way we tested data quantity. Finally, we included the thermal modeling outcomes as supplementary variables in the MFA ordination to visualize the correspondence between availability and quality.

TABLE 2.

Criteria used for thermal modeling to assess seed germination data quality.

Criterion Description Decision rule

Number of tested temperatures

NTestTemp >2

 Number of different germination temperatures tested for each species; if species has been tested at more than two different temperatures, NTestTemp is assigned a value 1; if species has been tested at two or fewer temperatures, NTestTemp is assigned a value 0 Species with >2 germination temperatures tested have high‐quality data (i.e., quality = 1)

Model concavity

a < 0

 a is the leading coefficient that determines the concavity of the quadratic curve; a > 0 indicates upward concavity, precludingthe identification of an optimum Species with a > 0 do not fit models (i.e., quality = 0)

Tail drop (not flat)

TestTemp > T opt & propgerm <0.75*Pmax

TestTemp < T opt & propgerm <0.75*Pmax

 Curves adequately shaped (not flat) when germination is at least 25% below the maximum on the low‐ and high‐temperature sides of T opt Pass (quality = 1): quadratic response shows downward concavity with decline at both thermal extremes; fail (quality = 0): curve is flat or lacks tail decline
Open in a new tab

Abbreviations: TestTemp, germination temperatures tested; T opt, optimal temperature (temperature where the model predicts maximum germination); propgerm, proportion of germinated seeds; Pmax, maximum germination percentage for each species.

Although the phylogenetic regressions returned an estimate of phylogenetic signal for the response to given predictors, it was more practical to estimate the phylogenetic signal of both data availability and data quality if the signal was treated as binary response without inclusion of any predictor. We used the phylo‐D function (R package caper [Orme et al., 2023]) for this estimate.

Thermal risks

For those species passing the quality assessment (i.e., data were adequate to fit a reliable quadratic model), we estimated thermal parameters, such as the T opt and T breadth, with quadratic thermal modeling for each species (Sentinella et al., 2020). Uncertainty estimates for T opt and T breadth derived using the delta method were associated with species‐level thermal metrics and are reported as standard error (SE). We excluded species for which T opt or T breadth was not estimable, for which their associated SE could not be computed, or for which T opt fell outside the experimentally tested temperature range from the thermal risk classification. We treated T opt and T breadth as point estimates derived directly from the fitted quadratic coefficients. Uncertainty estimates were used descriptively to assess model reliability and were not incorporated in the assignment of risk classes. Models yielding biologically unreliable optima were excluded through the three thermal quality criteria (Table 2), ensuring that only biologically meaningful estimates entered the risk assessment. Then, we calculated the mean of the optimal temperature (T optM) and of the optimal thermal breadth (T breadthM) for seed germination across all species: T optM = 17°C and T breadthM = 8°C. Using these two parameters as average thresholds across all species and accounting for their thermal limits, we allocated them to four risks categories: thermally resilient, species with a warm T opt and wide T breadth (T opt > 17°C & T breadth > 8°C); warm‐narrow risk, species with warm T opt and narrow T breadth (T opt > 17°C & T breadth ≤ 8°C); cold‐wide risk, species with cold T opt and wide T breadth (T opt ≤ 17°C & T breadth > 8°C); and high thermal risk, species with cold T opt and narrow T breadth (T opt ≤ 17°C & T breadth ≤ 8°C).

To evaluate whether these thermal risk categories were nonrandomly distributed across habitats, we tested the independence between the four risk classes and habitat types with chi‐square tests and Fisher's exact tests when expected frequencies were low.

We constructed a second MFA ordination to visualize the relationship between thermal limits (i.e., T opt, optimal low temperature, optimal upper temperature, and T breadth) and seed dormancy. Finally, we assessed the phylogenetic signal of thermal risk categories with the phylo‐D function and analyzed independence between thermal risk categories and habitat types separately with chi‐square (or Fisher's exact) tests, as described above.

RESULTS

Data availability and quality assessments

Data were available for 31% of the 1495 species. Among these, 91% did not meet the criteria for data quality, leaving only 9% of species (140) with adequate data for modeling the thermal germination niche. Overall, most of our hypotheses concerning data availability and quality were supported (Figure 1). Seed germination data availability was limited for perennial species exhibiting reproductive (biotic pollination, fleshy fruits) and seed (seed mass, nondormant seeds) traits that hamper seed collection, processing, and germination. Species with large seeds were underrepresented, and, contrary to our expectations, species of conservation concern had limited data availability. Conversely, data were more abundant and adequate for thermal modeling for species with wide ecological breadths that set indehiscent fruits and had physiologically dormant (PD) seeds and for species relevant for restoration and that had traditional uses (Figure 1; Appendix S2). The multivariate ordination broadly agreed with these univariate regressions but offered the opportunity to summarize the cross‐covariance among variables (details in Appendix S3).

FIGURE 1.

FIGURE 1

Open in a new tab

Estimated effect sizes of predictors related to germination data (a) availability (presence or absence) and (b) quality (based on criteria for thermal modeling) (predictor [left] groups: green, vegetative traits; orange, reproductive traits; brown, seed traits; blue, ecological breadth; gray, conservation concern and relevance for restoration; dark red, plant uses; whiskers are standard errors, SE); NoInfo, no information on seed dormancy reported; PY(PD), physical and physical‐physiological dormancy (combined); ND, nondormant; M(P)D, morphological and morfophysiological dormancy (combined). Each predictor was tested in an independent model.

Data availability and quality showed a moderately significant phylogenetic structure (D = 0.765 and D = 0.898, respectively [p < 0.001]) (i.e., closely related species shared similar levels of data representation). Logistic models that included predictors with strong effects on data availability (e.g., perennial life form, biotic pollination, PD, ecological breadth, and plant uses) and data quality (e.g., seed mass, fruit type, dormancy, and ecological breadth) indicated a phylogenetically structured availability in germination data (α < 0.2) (Appendices S4 and S5). The full dataset of the 1495 species including trait data and quality assessment results is in Appendix S6.

Thermal risks to seed germination

The species ordination based on thermal parameters for the 140 species with adequate data showed that these species were unevenly distributed in the multidimensional space. Among the four thermal risk categories, 9% were thermally resilient, 25% were warm‐narrow, 31% were cold‐wide, and 35% were high risk (Figure 2; Appendix S7). The first ordination axis separated cooler T opt species (Figure 2, left) from warmer T opt species (Figure 2, right), which generally corresponded to those with PD and nondormant seeds. The second MFA axis separated narrow T breadth species (i.e., species with morphological and morphophysiological dormancy: M[P]D) (Figure 2, bottom) from wide T breadth species (Figure 2, top) (i.e., species with physical and physical‐physiological dormancy [combined]: PY[PD]).

FIGURE 2.

FIGURE 2

Open in a new tab

Results of a multiple factor analysis ordination summarizing the cross‐covariance among thermal parameters (i.e., optimal low temperature [T low], optimal temperature [T opt], and optimal upper temperature [T upp]); optimal thermal breadth (T breadth); and seed dormancy of plant species in the Mediterranean to visualize the potential thermal risks to seed germination (purple arrows, thermal parameters used to construct the ordination; black arrows, dormancy supplementary variables; I, thermally resilient species [36 species] not at risk; II, species [42] with limited thermal breadth but warm optimal temperatures; III, species [20] with a broad thermal breadth but low optimal temperatures; IV, species [42] with narrow thermal breadth and cold optimal temperatures; NoInfo, no information on seed dormancy reported; PY[PD], physical and physical‐physiological dormancy [combined]; ND, nondormant; M[P]D, morphological and morphophysiological dormancy [combined]). Species thermal risk categories are in Appendix S7.

The species within the four risk categories were randomly distributed across taxonomic families (Fisher's exact test with simulated p, p > 0.05) (Appendix S8) but not across habitat types (Appendix S9). Thermal risk categories did not exhibit significant phylogenetic signal (D = 0.93, 0.80, 0.75, and 1.1 for high thermal risk, cold‐wide risk, warm‐narrow risk, and thermally resilient species, respectively, p > 0.05), indicating that species with similar levels of risk are not more phylogenetically related than expected by chance. Significant deviations from independence between risk levels and habitat type were found for coastal (χ2 = 17.57, p < 0.05), Forests (χ2 = 20.07, p < 0.05), grasslands (χ2 = 13.93, p < 0.05), and scrublands (χ2 = 15.14, p < 0.05). Specifically, warm‐narrow species were overrepresented in wetlands and coastal habitats, whereas cold‐wide species were consistently underrepresented, particularly in coastal, forests, and scrublands. In contrast, species at high risk were more frequent than expected by chance in forests and scrublands, whereas thermally resilient species showed a relatively homogeneous distribution across habitats.

DISCUSSION

Overall, data were more abundant and adequate for thermal modeling in ecologically widely distributed species setting PD seeds, for species relevant for restoration, and for species with reported traditional uses. Our finding that widespread common species with economic interest were more studied stands in contrast to a number of investigations in the Mediterranean that targeted threatened and endemic species are most common (Kadis et al., 2010; Luna et al., 2012; Picciau et al., 2019; Thanos & Doussi, 1995). This apparent discrepancy is likely driven by differences in study scope and sampling strategies; we targeted species characteristic of the Mediterranean rather than narrowly distributed taxa. By focusing on these species, our results are generalizable across the region rather than limited to a specific geographic area, and our methods provide a transferable framework for improving germination data to support seed‐based conservation and restoration programs.

Germination data availability and quality

In line with our expectations, data were available and adequate for thermal modeling in ecologically widely distributed species because their broader distribution across diverse habitats likely facilitates their study due to increased accessibility. Likewise, it was not surprising that species whose seeds were relevant for restoration exhibited high data availability and were adequate for thermal modeling because understanding their germination requirements is crucial for achieving successful restoration efforts (Dalziell et al., 2022; Hirst et al., 2025; Pedrini & Dixon, 2020). However, the sustainability of harvesting seeds from wild populations to supply this demand is increasingly questioned, and guidelines based on species’ generation time have been proposed to mitigate demographic risks (Bucharova et al., 2025).

Several consistent patterns of data scarcity emerged across species. Data availability was low for perennial herbs, particularly those with biotic pollination. Their reliance on a limited number of pollinators for reproduction success may constrain seed production relative to abiotic pollination (i.e., unassisted and self‐pollination), thereby reducing opportunities for germination testing. Pollinator‐dependent species are disproportionately vulnerable to environmental change and population decline, leading to reduced seed production and recruitment, which is consistent with their underrepresentation in experimental and conservation datasets (Ashman et al., 2004; Eckert et al., 2009). Data on germination in species setting fleshy fruit were limited (consistent with our hypothesis), likely because their complex morphological and functional features limit their seed processing and germination (Baskin & Baskin, 2014) (Table 1). Plants with large seeds were also underrepresented, likely due to their low seed production or the large space needed to conduct experiments (Fenner & Thompson, 2005). Species of conservation concern had similarly limited data, probably due to difficulties in seed collection and legal restrictions, which may hinder their inclusion in conservation planning. Surprisingly, species with nondormant seeds also showed limited data availability, despite their lack of specific germination requirements. Thus, their lower ecophysiological complexity may lead researchers to prioritize dormant species (Shefferson & Tali, 2007). Because many conservation and restoration practices rely on managing dormancy to enhance seed germination success (Hirst et al., 2025; Kettenring & Tarsa, 2020; Kildisheva et al., 2020), this likely explains why species with dormant seeds were more studied (White et al., 2025).

Broadly, species selection for conservation and restoration programs often relied on habitat rarity, population trends, feasibility of seed collection, or expert knowledge, rather than germination knowledge alone (Diallo et al., 2021; Fenu et al., 2023; Pérez‐Navarro et al., 2023). Nevertheless, species whose regeneration requirements are poorly known were frequently excluded from conservation and restoration initiatives (Kildisheva et al., 2020). In line with regional evidence (Fenu et al., 2023; Pérez‐Navarro et al., 2023; Thompson et al., 2026), our results showed that germination data gaps were not randomly distributed; rather, they reflected consistent ecological and practical constraints and were concentrated in specific trait‐based groups, particularly perennial herbs, fleshy‐fruited species, and biotically pollinated taxa.

Trade‐off between data availability and quality

Overall, when germination data were available, they were also of adequate quality for thermal modeling, but this was not always the case. For example, although germination data were scarce for most perennial herbs and conservation species, when available for a given species, they were adequate for thermal modeling. As such, available data for perennial conservation species are of high research value for informing management strategies, despite sampling challenges linked to their small population sizes and permit restrictions (Bartholomew et al., 2023; Heywood & Iriondo, 2003). Conversely, annuals and species setting dehiscent fruits and PY(PD) exhibited high data availability but low data quality, limiting their adequacy for thermal modeling. These contrasting patterns highlight that species’ reproductive traits critically shape, in opposite ways, the quantity and quality of germination data to inform conservation programs.

Thermal risks to seed germination

Although only 140 species passed the quality assessment, these species provided a valuable dataset for modeling thermal limits and exploring thermal risks. These thermal risk categories represented relative, rather than absolute, vulnerability. Nevertheless, modeling thermal responses at the species level remains essential because it captures biologically meaningful variation among species (and among populations when data are available), providing a comparative framework for assessing sensitivity to warming in relation to species’ habitat context and climatic setting (Hirst et al., 2025; Thompson et al., 2026). Species potentially at high thermal risk (i.e., exhibiting a cold and narrow germination niche) often lacked dormancy information or exhibited complex dormancy types, such as M(P)D, highlighting priority targets for future germination research.

The majority of species relevant for restoration did not exhibit high thermal risk, indicating that these taxa are likely to remain functional under projected warming and can therefore be considered resilient candidates for restoration. Likewise, none of the species currently identified as of conservation interest were classified as high thermal risk, partly reflecting data limitations that still preclude germination risk assessments for many conservation priority taxa. These findings demonstrated that thermal risk could identify additional taxa that warrant conservation attention beyond existing conservation frameworks (Hirst et al., 2025) and call for stronger collaboration across stakeholders to enhance seed supply systems and ensure access to diverse, high‐quality seeds for conservation and restoration (Andres et al., 2024).

Our results suggest that risk categories are mainly influenced by habitat rather than phylogeny, meaning more attention to critical ecosystems is needed. Indeed, species at high thermal risk were most frequent in forests and scrublands, highlighting their urgent need for conservation. In contrast, warm‐narrow species were overrepresented in coastal and wetland habitats, reflecting restricted thermal tolerances. Although these species currently face a lower thermal risk, their persistence could be threatened if temperatures shift beyond their tolerance limits. Their prevalence makes them valuable candidates for future restoration programs, yet targeted conservation and monitoring in coastal areas remain essential to ensure their long‐term stability under climate change.

Challenges and perspectives

Our findings showed that limitations in germination data availability and modeling adequacy are structured across traits and habitats, currently constraining the effective use of seed‐based information in conservation and restoration planning. To address these limitations, we propose a framework (Figure 3) that integrates data generation, thermal germination modeling, and decision‐making. This framework links data gaps, thermal germination modeling, and species‐level risk assessment with outputs translated into actionable guidance for conservation and restoration, including species and habitat prioritization, climate‐informed seed sourcing, and, where possible, the integration of complementary genetic information to refine management decisions (Mattana et al., 2026). By embedding thermal germination modeling within a broader assessment of species vulnerability and habitat context, this approach supports the identification of priority taxa and habitats, guides targeted data collection efforts, and facilitates the incorporation of thermal germination risk assessments into restoration and conservation strategies. More broadly, this framework provides a transferable approach for strengthening the role of seed biology in seed‐based biodiversity conservation under ongoing climate change and highlights the urgent need to generate robust germination data for underrepresented groups and to improve experimental and modeling quality where data availability alone is the limiting factor (Godefroid et al., 2025; Hirst et al., 2025; Mattana et al., 2026; Thompson et al., 2026). We showed that germination information can be modeled and translated into thermal risk assessments to support seed‐based conservation and restoration programs. A strategic shift is needed to focus research on neglected taxa, improve data integration, and foster collaboration among scientists, practitioners, and policy makers in the Mediterranean and in other climate‐sensitive regions worldwide.

FIGURE 3.

FIGURE 3

Open in a new tab

The five steps for strengthening germination evidence to support conservation and restoration in the Mediterranean basin, linking germination ecology to applied biodiversity conservation (optimal low temperature [T low]; optimal temperature [T opt]; upper temperature [T upp]; mean optimal temperature [T optM] ; mean thermal breadth [T breadthM: range between lower and upper thermal limits]; thermally resilient species, species with seed germination not at risk from increasing temperatures; warm‐narrow risk, species with limited thermal breadth but warm optimal temperatures; cold‐wide risk, species with a broad thermal breadth but low optimal temperatures; high thermal risk, species with narrow thermal breadth and cold optimal temperatures).

Supporting information

Supplementary Materials.

COBI-40-e70329-s002.docx (2.5MB, docx)

Supplementary Materials.

COBI-40-e70329-s001.csv (391KB, csv)

Supplementary Materials.

COBI-40-e70329-s003.csv (17.5KB, csv)

ACKNOWLEDGMENTS

We thank P. Stevenson for his kind review of the manuscript and helpful comments. Our research was funded by the Ministry of University and Research (MUR) as part of the PON 2014–2020 “Research and Innovation” resources—Green/Innovation Action—DM MUR 1061/2022, DOT13GFICX‐2. This study was also supported by the Italian Ministry of Universities and Research under the Department of Excellence 2023–2027 initiative. The Royal Botanic Gardens, Kew receives grant in aid from the Department for Environment, Food and Rural Affairs (DEFRA). Funders had no role in study design, report writing, or article submission.

Open access publishing facilitated by Universita degli Studi di Pisa, as part of the Wiley ‐ CRUI‐CARE agreement.

Cruz‐Tejada, D. M. , Mattana, E. , & Carta, A. (2026). Modeling seed germination data to meet biodiversity conservation needs in the Mediterranean. Conservation Biology, 40, e70329. 10.1111/cobi.70329

Article impact statement: Revealing data limitations provides a framework to strengthen seed‐based conservation and restoration strategies in the Mediterranean.

DATA AVAILABILITY STATEMENT

The original germination data are in Cruz‐Tejada, Mattana, et al. (2024). The trait dataset generated for the analyses and the thermal modeling results are in the Supporting Information and are available in GitHub (https://github.com/DianaCruzTejada/MedGerm_DataAssessment), where the R codes used for the analyses are also available. A versioned archive of the repository is available in Zenodo (https://doi.org/10.5281/zenodo.20136467)

REFERENCES

  1. Andres, S. E. , Atkinson, J. , Coleman, D. , Brazill‐Boast, J. , Wright, I. J. , Allen, S. , & Gallagher, R. V. (2024). Constraints of commercially available seed diversity in restoration: Implications for plant functional diversity. Plants, People, Planet, 6, 1341–1357. 10.1002/ppp3.10523 [DOI] [Google Scholar]
  2. Ashman, T.‐L. , Knight, T. M. , Steets, J. A. , Amarasekare, P. , Burd, M. , Campbell, D. R. , Dudash, M. R. , Johnston, M. O. , Mazer, S. J. , Mitchell, R. J. , Morgan, M. T. , & Wilson, W. G. (2004). Pollen limitation of plant reproduction: Ecological and evolutionary causes and consequences. Ecological Society of America, 85(9), 2408–2421. http://www.mendeley.com/research/relationships‐among‐species‐traits‐separating‐levels‐response‐identifying‐linkages‐abundances/ [Google Scholar]
  3. Aurelle, D. , Thomas, S. , Albert, C. , Bally, M. , Bondeau, A. , Boudouresque, C.‐F. , Cahill, A. E. , Carlotti, F. , Chenuil, A. , Cramer, W. , Davi, H. , De Jode, A. , Ereskovsky, A. , Farnet, A.‐M. , Fernandez, C. , Gauquelin, T. , Mirleau, P. , Monnet, A.‐C. , Prévosto, B. , … Fady, B. (2022). Biodiversity, climate change, and adaptation in the Mediterranean. Ecosphere, 13(4), Article e3915. 10.1002/ecs2.3915 [DOI] [Google Scholar]
  4. Bartholomew, D. C. , Shaw, K. , Rivers, M. C. , Baraka, P. , Kigathi, R. N. , Wanja, W. , Wanjiku, C. , & Williams, H. F. (2023). Overcoming the challenges of incorporating rare and threatened flora into ecosystem restoration. Restoration Ecology, 31(4), Article e13849. 10.1111/rec.13849 [DOI] [Google Scholar]
  5. Baskin, C. C. , & Baskin, J. M. (2014). Seeds: Ecology, biogeography, and, evolution of dormancy and germination (2nd ed.). Academic Press. 10.1016/C2013-0-00597-X [DOI] [Google Scholar]
  6. Baskin, C. C. , & Baskin, J. M. (2022). Plant regeneration from seeds. Academic Press. 10.1016/c2020-0-00735-8 [DOI] [Google Scholar]
  7. Brooks, M. E. , Kristensen, K. , van Benthem, K. J. , Magnusson, A. , Berg, C. W. , Nielsen, A. , Skaug, H. J. , Mächler, M. , & Bolker, B. M. (2017). glmmTMB balances speed and flexibility among packages for zero‐inflated generalized linear mixed modeling. R Journal, 9(2), 378–400. 10.32614/rj-2017-066 [DOI] [Google Scholar]
  8. Bucharova, A. , Bossdorf, O. , Scheepens, J. F. , & Salguero‐Gómez, R. (2025). Assessing limits of sustainable seed harvest in wild plant populations. Conservation Biology, 39(5), Article e70075. 10.1111/cobi.70075 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Carta, A. , Fernández‐Pascual, E. , Gioria, M. , Müller, J. V. , Rivière, S. , Rosbakh, S. , Saatkamp, A. , Vandelook, F. , & Mattana, E. (2022). Climate shapes the seed germination niche of temperate flowering plants: A meta‐analysis of European seed conservation data. Annals of Botany, 129, 775–786. 10.1093/aob/mcac037 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Carta, A. , Mattana, E. , Ensslin, A. , Godefroid, S. , & Molina‐Venegas, R. (2025). Plant evolutionary history is largely underrepresented in European seed banks. New Phytologist, 246(2), 797–806. 10.1111/nph.70000 [DOI] [PubMed] [Google Scholar]
  11. Carta, A. , Vandelook, F. , Ramírez‐Barahona, S. , Chen, S.‐C. , Dickie, J. , Steinbrecher, T. , Thanos, C. A. , Moles, A. T. , Leubner‐Metzger, G. , & Mattana, E. (2024). The seed morphospace, a new contribution towards the multidimensional study of angiosperm sexual reproductive biology. Annals of Botany, 134, 701–710. 10.1093/aob/mcae099 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Chytrý, M. , Tichý, L. , Hennekens, S. M. , Knollová, I. , Janssen, J. A. M. , Rodwell, J. S. , Peterka, T. , Marcenò, C. , Landucci, F. , Danihelka, J. , Hájek, M. , Dengler, J. , Novák, P. , Zukal, D. , Jiménez‐Alfaro, B. , Mucina, L. , Abdulhak, S. , Aćić, S. , Agrillo, E. , … Schaminée, J. H. J. (2020). EUNIS Habitat Classification: Expert system, characteristic species combinations and distribution maps of European habitats. Applied Vegetation Science, 23(4), 648–675. 10.1111/avsc.12519 [DOI] [Google Scholar]
  13. Cochrane, A. , Daws, M. I. , & Hay, F. R. (2011). Seed‐based approach for identifying flora at risk from climate warming. Austral Ecology, 36(8), 923–935. 10.1111/j.1442-9993.2010.02211.x [DOI] [Google Scholar]
  14. Cruz‐Tejada, D. M. , Fernández‐Pascual, E. , Mo, A. , Mattana, E. , & Carta, A. (2024). MedGermDB: A seed germination database for characteristic species of Mediterranean habitats. Applied Vegetation Science, 27(2), Article e12771. 10.1111/avsc.12771 [DOI] [Google Scholar]
  15. Cruz‐Tejada, D. M. , Mattana, E. , Rosbakh, S. , Fernández‐Pascual, E. , & Carta, A. (2024). Macroclimatic convergence and habitat specialisation shape the mediterranean seed germination syndrome. Ecology and Evolution, 14(11), Article e70527. 10.1002/ece3.70527 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Dalziell, E. L. , Lewandrowski, W. , Commander, L. E. , Elliott, C. P. , Erickson, T. E. , Tudor, E. P. , Turner, S. R. , & Merritt, D. J. (2022). Seed traits inform the germination niche for biodiverse ecological restoration. Seed Science and Technology, 50(2), 103–124. 10.15258/sst.2022.50.1.s.06 [DOI] [Google Scholar]
  17. Diallo, M. , Ollier, S. , Mayeur, A. , Fernández‐Manjarrés, J. , García‐Fernández, A. , Iriondo, J. M. , Vaissière, A.‐C. , & Colas, B. (2021). Plant translocations in Europe and the Mediterranean: Geographical and climatic directions and distances from source to host sites. Journal of Ecology, 109(6), 2296–2308. 10.1111/1365-2745.13609 [DOI] [Google Scholar]
  18. Donohue, K. , Rubio de Casas, R. , Burghardt, L. , Kovach, K. , & Willis, C. G. (2010). Germination, postgermination adaptation, and species ecological ranges. Annual Review of Ecology, Evolution, and Systematics, 41(1), 293–319. 10.1146/annurev-ecolsys-102209-144715 [DOI] [Google Scholar]
  19. Eckert, C. G. , Kalisz, S. , Geber, M. A. , Sargent, R. , Elle, E. , Cheptou, P.‐O. , Goodwillie, C. , Johnston, M. O. , Kelly, J. K. , Moeller, D. A. , Porcher, E. , Ree, R. H. , Vallejo‐Marín, M. , & Winn, A. A. (2009). Plant mating systems in a changing world. Trends in Ecology & Evolution, 25(1), 35–43. 10.1016/j.tree.2009.06.013 [DOI] [PubMed] [Google Scholar]
  20. European Commission . (1992). Council Directive 92/43/ECC. Official Journal of the European Union, 94(1259), 40–52. [Google Scholar]
  21. Fenner, M. , & Thompson, K. (2005). The ecology of seeds. Cambridge University Press. 10.1017/CBO9780511614101 [DOI] [Google Scholar]
  22. Fenu, G. , Calderisi, G. , Boršić, I. , Bou Dagher Kharrat, M. , García Fernández, A. , Kahale, R. , Panitsa, M. , & Cogoni, D. (2023). Translocations of threatened plants in the Mediterranean Basin: Current status and future directions. Plant Ecology, 224(9), 765–775. 10.1007/s11258-023-01303-7 [DOI] [Google Scholar]
  23. Fernández‐Pascual, E. , Carta, A. , Mondoni, A. , Cavieres, L. A. , Rosbakh, S. , Venn, S. , Satyanti, A. , Guja, L. , Briceño, V. F. , Vandelook, F. , Mattana, E. , Saatkamp, A. , Bu, H. , Sommerville, K. , Poschlod, P. , Liu, K. , Nicotra, A. , & Jiménez‐Alfaro, B. (2021). The seed germination spectrum of alpine plants: A global meta‐analysis. New Phytologist, 229(6), 3573–3586. 10.1111/nph.17086 [DOI] [PubMed] [Google Scholar]
  24. Fernández‐Pascual, E. , Carta, A. , Rosbakh, S. , Guja, L. , Phartyal, S. S. , Silveira, F. A. O. , Chen, S.‐C. , Larson, J. E. , & Jiménez‐Alfaro, B. (2023). SeedArc, a global archive of primary seed germination data. New Phytologist, 240, 466–470. 10.1111/nph.19143 [DOI] [PubMed] [Google Scholar]
  25. Garamszegi, L. Z. (2014). Modern phylogenetic comparative methods and their application in evolutionary biology. Springer. 10.1007/978-3-662-43550-2 [DOI] [Google Scholar]
  26. Giorgi, F. , & Lionello, P. (2008). Climate change projections for the Mediterranean region. Global and Planetary Change, 63(2–3), 90–104. 10.1016/j.gloplacha.2007.09.005 [DOI] [Google Scholar]
  27. Gioria, M. , Pyšek, P. , Baskin, C. C. , & Carta, A. (2020). Phylogenetic relatedness mediates persistence and density of soil seed banks. Journal of Ecology, 108(5), 2121–2131. 10.1111/1365-2745.13437 [DOI] [Google Scholar]
  28. Godefroid, S. , Lacquaye, S. , Ensslin, A. , Dalrymple, S. , Abeli, T. , Branwood, H. , Ferrando Pardo, I. , Ferrer Gallego, P. P. , Zippel, E. , Gouveia, L. , Lobo, C. A. , Fernandes, F. , Rasp, M. , Daco, L. , Séleck, M. , Frankard, P. , Fischer, L. K. , Koutsovoulou, K. , Finger, A. , … White, F. J. (2025). Current state of plant conservation translocations across Europe: motivations, challenges and outcomes. Biodiversity and Conservation, 34(3), 769–792. 10.1007/s10531-025-03013-0 [DOI] [Google Scholar]
  29. Govaerts, R. , Nic Lughadha, E. , Black, N. , Turner, R. , & Paton, A. (2021). The World Checklist of Vascular Plants, a continuously updated resource for exploring global plant diversity. Scientific Data, 8(1), Article 215. 10.1038/s41597-021-00997-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Grubb, P. J. (1977). The maintenance of species‐richness in plant communities: The importance of the regeneration niche. Biological Reviews, 52(1), 107–145. 10.1111/j.1469-185x.1977.tb01347.x [DOI] [Google Scholar]
  31. Heywood, V. H. , & Iriondo, J. M. (2003). Plant conservation: Old problems, new perspectives. Biological Conservation, 113(3), 321–335. 10.1016/S0006-3207(03)00121-6 [DOI] [Google Scholar]
  32. Hirst, M. J. , Vázquez‐Ramírez, J. , Rendall, A. R. , & Venn, S. E. (2025). Thermal germination niche: Implications for seed‐based restoration in climate‐sensitive alpine environments. Restoration Ecology, 33(4), Article e70031. 10.1111/rec.70031 [DOI] [Google Scholar]
  33. IPCC . (2022). Climate Change 2022: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. [Google Scholar]
  34. Joffre, R. , Rambal, S. , & Damesin, C. (2007). Functional attributes in Mediterranean‐type ecosystems. In Pugnaire F. & Valladares F. (Eds.), Functional plant ecology (pp. 285–312). CRC Press. [Google Scholar]
  35. Just, M. , Albert, B. , Pedrini, S. , Turner, S. , & Dixon, K. (2024). Hot and fast: Seed ecology for restoration relevant species in the Argyle region of the east Kimberley, Australia. Restoration Ecology, 32(7), Article e14201. 10.1111/rec.14201 [DOI] [Google Scholar]
  36. Kadis, C. , Kounnamas, C. , & Georghiou, K. (2010). Seed germination and conservation of endemic, rare, and threatened aromatic plants of Cyprus. Israel Journal of Plant Sciences, 58(3–4), 251–261. 10.1560/IJPS.58.3-4.251 [DOI] [Google Scholar]
  37. Kattge, J. , Bönisch, G. , Díaz, S. , Lavorel, S. , Prentice, I. C. , Leadley, P. , Tautenhahn, S. , Werner, G. D. A. , Aakala, T. , Abedi, M. , Acosta, A. T. R. , Adamidis, G. C. , Adamson, K. , Aiba, M. , Albert, C. H. , Alcántara, J. M. , Alcázar C, C. , Aleixo, I. , Ali, H. , … Wirth, C. (2020). TRY plant trait database—Enhanced coverage and open access. Global Change Biology, 26(1), 119–188. 10.1111/gcb.14904 [DOI] [PubMed] [Google Scholar]
  38. Kettenring, K. M. , & Tarsa, E. E. (2020). Need to seed? Ecological, genetic, and evolutionary keys to seed‐based wetland restoration. Frontiers in Environmental Science, 8, Article 109. 10.3389/fenvs.2020.00109 [DOI] [Google Scholar]
  39. Kildisheva, O. A. , Dixon, K. W. , Silveira, F. A. O. , Chapman, T. , Di Sacco, A. , Mondoni, A. , Turner, S. R. , & Cross, A. T. (2020). Dormancy and germination: Making every seed count in restoration. Restoration Ecology, 28(S3), S256–S265. 10.1111/rec.13140 [DOI] [Google Scholar]
  40. Ladouceur, E. , Jiménez‐Alfaro, B. , Marin, M. , De Vitis, M. , Abbandonato, H. , Iannetta, P. P. M. , Bonomi, C. , & Pritchard, H. W. (2018). Native seed supply and the restoration species pool. Conservation Letters, 11(2), 1–9. 10.1111/conl.12381 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Le, S. , Josse, J. , & Husson, F. (2009). FactoMineR: An R package for multivariate analysis. Environment International, 35(2), 253–258. 10.1016/j.envint.2008.06.007 [DOI] [Google Scholar]
  42. Luna, B. , Pérez, B. , Torres, I. , & Moreno, J. M. (2012). Effects of incubation temperature on seed germination of mediterranean plants with different geographical distribution ranges. Folia Geobotanica, 47(1), 17–27. 10.1007/s12224-011-9110-0 [DOI] [Google Scholar]
  43. Mattana, E. , Chapman, T. , Miles, S. , Ulian, T. , & Carta, A. (2023). Regeneration from seeds in a temperate native flora: A climate‐smart and natural—Capital‐driven germination risk modelling approach. Plants People Planet, 5(6), 908–922. 10.1002/ppp3.10378 [DOI] [Google Scholar]
  44. Mattana, E. , Fouce Hernandez, E. , Andrews, L. , Watt Poddar, L. , Rowe‐Habbari, J. , Negri, I. S. , Cornwell‐Davison, F. , Viruel, J. , Cockel, C. , Seal, C. E. , & Chapman, T. (2026). Thermal risk of seed germination of UK's plant populations and implications for climate‐smart ecological restoration. Restoration Ecology, 34(1), Article e70239. 10.1111/rec.70239 [DOI] [Google Scholar]
  45. Muñoz‐Gallego, R. , Wiegand, T. , Traveset, A. , & Fedriani, J. M. (2023). From seed dispersal service to reproductive collapse: Density‐dependent outcome of a palm–mammal interaction. Oikos, 2023(10), Article e10002. 10.1111/oik.10002 [DOI] [Google Scholar]
  46. Myers, N. , Mittermeier, R. A. , Mittermeier, C. G. , Da Fonseca, G. A. B. , & Kent, J. (2000). Biodiversity hotspots for conservation priorities. Nature, 403, 853–858. 10.1080/21564574.1998.9650003 [DOI] [PubMed] [Google Scholar]
  47. Ordóñez‐Parra, C. A. , Dayrell, R. L. C. , Negreiros, D. , Andrade, A. C. S. , Andrade, L. G. , Antonini, Y. , Barreto, L. C. , Barros, F. V. , da Cruz Carvalho, V. , Corredor, B. A. D. , Davide, A. C. , Duarte, A. A. , Feitosa, S. D. S. , Fernandes, A. F. , Fernandes, G. W. , Figueiredo, M. A. , Fidelis, A. , Garcia, L. C. , Garcia, Q. S. , … Silveira, F. A. O. (2022). Rock n ’ Seeds: A database of seed functional traits and germination experiments from Brazilian rock outcrop vegetation. Ecology, 104(1), 2–5. 10.1002/ecy.3852 [DOI] [PubMed] [Google Scholar]
  48. Orme, D. , Freckleton, R. , Thomas, G. , Petzoldt, T. , Fritz, S. , Isaac, N. , & Pearse, W. (2023). The caper package: Comparative analysis of phylogenetics and evolution in R . R package version 1.0.1. https://cran.r‐project.org/package=caper
  49. Pedrini, S. , & Dixon, K. W. (2020). International principles and standards for native seeds in ecological restoration. Restoration Ecology, 28(S3), S286–S303. 10.1111/rec.13155 [DOI] [Google Scholar]
  50. Pérez‐Navarro, M. A. , Lloret, F. , Ogaya, R. , Estiarte, M. , & Peñuelas, J. (2023). Decrease in climatic disequilibrium associated with climate change and species abundance shifts in Mediterranean plant communities. Journal of Ecology, 112(2), 291–304. 10.1111/1365-2745.14233 [DOI] [Google Scholar]
  51. Picciau, R. , Serra, S. , Porceddu, M. , & Bacchetta, G. (2019). Seed traits and germination behaviour of four Sardinian populations of Helichrysum microphyllum subsp. tyrrhenicum (Asteraceae) along an altitudinal gradient. Plant Biology, 21(3), 498–506. 10.1111/plb.12903 [DOI] [PubMed] [Google Scholar]
  52. Pironon, S. , Ondo, I. , Diazgranados, M. , Allkin, R. , Baquero, A. C. , Cámara‐Leret, R. , Canteiro, C. , Dennehy‐Carr, Z. , Govaerts, R. , Hargreaves, S. , Hudson, A. J. , Lemmens, R. , Milliken, W. , Nesbitt, M. , Patmore, K. , Schmelzer, G. , Turner, R. M. , Van Andel, T. R. , Ulian, T. , … Willis, K. J. (2024). The global distribution of plants used by humans. Science, 383(6680), 293–297. 10.1126/science.adg8028 [DOI] [PubMed] [Google Scholar]
  53. Price, M. , Harrison, P. A. , Jordan, R. , Steane, D. , Vaillancourt, R. E. , & Prober, S. M. (2024). Intra‐specific variation and climate differentially shape the thermal germination niches of three co‐occurring woodland forbs. Austral Ecology, 49(2), Article e13480. 10.1111/aec.13480 [DOI] [Google Scholar]
  54. Rosbakh, S. , Carta, A. , Fernández‐Pascual, E. , Phartyal, S. S. , Dayrell, R. L. C. , Mattana, E. , Saatkamp, A. , Vandelook, F. , Baskin, J. , & Baskin, C. (2023). Global seed dormancy patterns are driven by macroclimate but not fire regime. New Phytologist, 240(2), 555–564. 10.1111/nph.19173 [DOI] [PubMed] [Google Scholar]
  55. Sentinella, A. T. , Warton, D. I. , Sherwin, W. B. , Offord, C. A. , & Moles, A. T. (2020). Tropical plants do not have narrower temperature tolerances, but are more at risk from warming because they are close to their upper thermal limits. Global Ecology and Biogeography, 29(8), 1387–1398. 10.1111/geb.13117 [DOI] [Google Scholar]
  56. Shefferson, R. P. , & Tali, K. (2007). Dormancy is associated with decreased adult survival in the burnt orchid, Neotinea ustulata . Journal of Ecology, 95(1), 217–225. 10.1111/j.1365-2745.2006.01195.x [DOI] [Google Scholar]
  57. Spjut, R. W. (1994). A systematic treatment of fruit types. Memoirs of the New York Botanical Garden, 70, 1–182. [Google Scholar]
  58. Thanos, C. A. , & Doussi, M. A. (1995). Ecophysiology of seed germination in endemic labiates of Crete. Israel Journal of Plant Sciences, 43(3), 227–237. 10.1080/07929978.1995.10676607 [DOI] [Google Scholar]
  59. Thompson, J. D. , Fauché, M. , Papuga, G. , & Gauthier, P. (2026). Plant conservation in a changing Mediterranean world. Plants People Planet, 8(1), 49–72. 10.1002/ppp3.70036 [DOI] [Google Scholar]
  60. Tung Ho, L. S. , & Ané, C. (2014). A linear‐time algorithm for gaussian and non‐gaussian trait evolution models. Systematic Biology, 63(3), 397–408. 10.1093/sysbio/syu005 [DOI] [PubMed] [Google Scholar]
  61. Walck, J. L. , Hidayati, S. N. , Dixon, K. W. , Thompson, K. , & Poschlod, P. (2011). Climate change and plant regeneration from seed. Global Change Biology, 17(6), 2145–2161. 10.1111/j.1365-2486.2010.02368.x [DOI] [Google Scholar]
  62. White, K. , Cornwell‐Davison, F. , Cockel, C. , Chapman, T. , Mattana, E. , & Viruel, J. (2025). Application of a biological trait‐based framework for plant species conservation assessments in ecological restoration Author contributions. Restoration Ecology, 33(3), Article e14391. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Materials.

COBI-40-e70329-s002.docx (2.5MB, docx)

Supplementary Materials.

COBI-40-e70329-s001.csv (391KB, csv)

Supplementary Materials.

COBI-40-e70329-s003.csv (17.5KB, csv)

Data Availability Statement

We selected nine plant features as predictors of germination data availability and quality (Table 1). We focused on traits that are directly relevant to seed sampling, handling, curation, and germination experiments in conservation and restoration contexts. We gathered vegetative (i.e., life form), reproductive (i.e., fruit type and pollination), and seed (i.e., mass and dormancy) traits and species‐level information. This information, extracted from publicly available datasets and publications, included ecological breadth (i.e., number of habitats in which each species occurs) based on the EUNIS expert system (Chytrý et al., 2020); inclusion in European conservation policies (Habitats Directive); relevance for ecological restoration (i.e., commercially available for restoration [Ladouceur et al., 2018]); and reported plant uses (Table 1). The structure and coding of all predictors and their sources are provided in Table 1. Ecological breadth was calculated based on the number of habitats in which the species is present (i.e., three‐level hierarchical classification of European habitats [Chytrý et al., 2020]). When it was not possible to find information on reproductive and seed traits for a species, we used data from congeneric species. Species with missing trait information were assigned no information category when applicable. All predictors were analyzed in separate univariable phylogenetic logistic regression models. Traits that were conceptually categorical were encoded as binary dummy variables (0/1) and standardized prior to analysis. No multilevel categorical factors were fitted. Continuous predictors were analyzed as continuous variables and log transformed. The phylogenetic tree of the species pool and the trait distribution across lineages are in Appendix S1.

TABLE 1.

Predictors of seed germination data availability (presence or absence).

Feature Predictor Predictor type (unit) Hypothesis Source of predictor information
Vegetative

Life form

annual

perennial herbaceous

woody

 Binary (1, 0) Data availability for perennial herbs and annual plants is higher than for woody plants because perennials are easily found in natural areas and woody plants have complex reproductive cycles, often with either very large or fleshy dispersal units (Joffre et al., 2007).

www.FloraVeg.eu

Govaerts et al., 2021

Reproductive

Pollination

biotic

abiotic

 Binary (1, 0) Species with biotic pollination (i.e., animal‐mediated pollination) are less studied and have very little germination data due to low seed production (e.g., influenced by pollinator availability and specificity; compatibility and competition between individuals) (Ashman et al., 2004). Kattge et al., 2020

Fruit type

dehiscent

indehiscent

fleshy

 Binary (1, 0) Fleshy fruits often have complex morphological structures that involve additional treatments to allow germination, making them difficult to study in comparison with species with dehiscent and indehiscent fruits (Spjut, 1994). Carta et al., 2024
Seed Seed mass Continuous (mg) There is lower production of large seeds relative to small seeds; thus, smaller seeds are more likely to be studied because they are more available (Fenner & Thompson, 2005). Carta et al., 2024

Dormancy

PY(PD)

PD

M(P)D

ND

NoInfo

 Binary (1, 0) Germination data on species with deep and complex seed dormancy are not available due to the difficulty of experimentally breaking dormancy (Baskin & Baskin, 2014). Rosbakh et al., 2023
Ecological breadth Number of habitats** Continuous (n) Species restricted to one or few habitats are less studied because they are more difficult to sample. Chytrý et al., 2020
Conservation interest Species of conservation interest Binary (1, 0) Species listed in the annexes of the Habitat Directive are more studied than others because they are of conservation concern. European Commission, 1992
Restoration relevance Species relevant for ecological restoration Binary (1, 0) Species relevant for ecological restoration are more studied than others. Classification derived from the European restoration species pool proposed by Ladouceur et al. (2018)
Plant uses Species with traditional uses Binary (1, 0) Species with traditional uses are more studied than others (Pironon et al., 2024). Pironon et al., 2024
Open in a new tab

Abbreviations: PY(PD), physical and physical‐physiological dormancy (combined); PD, physiological dormancy; M(P)D, morphological and morphophysiological dormancy (combined); ND, nondormant; NoInfo, no information on seed dormancy reported. **The ecological breadth was calculated based on the number of habitats on which the species is present (i.e., three‐level hierarchical classification of European habitats [Chytrý et al., 2020])

We evaluated data availability across all species as a binary response (i.e., presence or absence of data for each species) with phylogenetic logistic regression models implemented in the r package phylolm (logistic MPLE) (Tung Ho & Ané, 2014). We fitted each predictor in a separate univariable model and extracted the alpha (α) parameter, which represents the strength of phylogenetic signal (i.e., the higher the value of α, the weaker the phylogenetic dependence). Then, we used multiple factor analysis (MFA) in the package FactoMineR (Le et al., 2009) to construct a multivariate ordination to visualize the cross‐covariance among our predictors of data availability for each species.

Using the function glmmTMB in the R package glmmTMB, we evaluated the adequacy of data for quantifying thermal germination responses at the species level with binomial mixed quadratic models (Sentinella et al., 2020) in which seed lot was a random factor (Brooks et al., 2017). To account for overdispersion, models included an observation‐level random effect and were fitted using a beta‐binomial error distribution. Zero inflation was not modeled because residual diagnostics did not indicate systematic excess zeros. Based on the outcome of these models, we were able to evaluate each species according to the thermal modeling criteria in Table 2. These criteria were based on the quadratic model parameters that reflected whether the species’ germination response followed the expected unimodal pattern with downward concavity. Uncertainty associated with derived thermal parameters (e.g., optimal temperature [T opt] and thermal breadth [T breadth]) was quantified using Sentinella et al.’s (2020) approach based on the delta method (see below). When derived thermal parameters or their uncertainty were not estimable (e.g., due to lack of a real solution at the selected germination threshold), values were denoted not applicable (NA). We then modeled data quality at the species level as a binary response (passed or failed quality assessment for each species)—that is, the data were either adequate or did not fit a reliable quadratic model (details in Table 2)—by fitting phylogenetic logistic models (logistic MPLE) (Tung Ho & Ané, 2014) in the same way we tested data quantity. Finally, we included the thermal modeling outcomes as supplementary variables in the MFA ordination to visualize the correspondence between availability and quality.

TABLE 2.

Criteria used for thermal modeling to assess seed germination data quality.

Criterion Description Decision rule

Number of tested temperatures

NTestTemp >2

 Number of different germination temperatures tested for each species; if species has been tested at more than two different temperatures, NTestTemp is assigned a value 1; if species has been tested at two or fewer temperatures, NTestTemp is assigned a value 0 Species with >2 germination temperatures tested have high‐quality data (i.e., quality = 1)

Model concavity

a < 0

 a is the leading coefficient that determines the concavity of the quadratic curve; a > 0 indicates upward concavity, precludingthe identification of an optimum Species with a > 0 do not fit models (i.e., quality = 0)

Tail drop (not flat)

TestTemp > T opt & propgerm <0.75*Pmax

TestTemp < T opt & propgerm <0.75*Pmax

 Curves adequately shaped (not flat) when germination is at least 25% below the maximum on the low‐ and high‐temperature sides of T opt Pass (quality = 1): quadratic response shows downward concavity with decline at both thermal extremes; fail (quality = 0): curve is flat or lacks tail decline
Open in a new tab

Abbreviations: TestTemp, germination temperatures tested; T opt, optimal temperature (temperature where the model predicts maximum germination); propgerm, proportion of germinated seeds; Pmax, maximum germination percentage for each species.

Although the phylogenetic regressions returned an estimate of phylogenetic signal for the response to given predictors, it was more practical to estimate the phylogenetic signal of both data availability and data quality if the signal was treated as binary response without inclusion of any predictor. We used the phylo‐D function (R package caper [Orme et al., 2023]) for this estimate.

Data were available for 31% of the 1495 species. Among these, 91% did not meet the criteria for data quality, leaving only 9% of species (140) with adequate data for modeling the thermal germination niche. Overall, most of our hypotheses concerning data availability and quality were supported (Figure 1). Seed germination data availability was limited for perennial species exhibiting reproductive (biotic pollination, fleshy fruits) and seed (seed mass, nondormant seeds) traits that hamper seed collection, processing, and germination. Species with large seeds were underrepresented, and, contrary to our expectations, species of conservation concern had limited data availability. Conversely, data were more abundant and adequate for thermal modeling for species with wide ecological breadths that set indehiscent fruits and had physiologically dormant (PD) seeds and for species relevant for restoration and that had traditional uses (Figure 1; Appendix S2). The multivariate ordination broadly agreed with these univariate regressions but offered the opportunity to summarize the cross‐covariance among variables (details in Appendix S3).

FIGURE 1.

FIGURE 1

Open in a new tab

Estimated effect sizes of predictors related to germination data (a) availability (presence or absence) and (b) quality (based on criteria for thermal modeling) (predictor [left] groups: green, vegetative traits; orange, reproductive traits; brown, seed traits; blue, ecological breadth; gray, conservation concern and relevance for restoration; dark red, plant uses; whiskers are standard errors, SE); NoInfo, no information on seed dormancy reported; PY(PD), physical and physical‐physiological dormancy (combined); ND, nondormant; M(P)D, morphological and morfophysiological dormancy (combined). Each predictor was tested in an independent model.

Data availability and quality showed a moderately significant phylogenetic structure (D = 0.765 and D = 0.898, respectively [p < 0.001]) (i.e., closely related species shared similar levels of data representation). Logistic models that included predictors with strong effects on data availability (e.g., perennial life form, biotic pollination, PD, ecological breadth, and plant uses) and data quality (e.g., seed mass, fruit type, dormancy, and ecological breadth) indicated a phylogenetically structured availability in germination data (α < 0.2) (Appendices S4 and S5). The full dataset of the 1495 species including trait data and quality assessment results is in Appendix S6.

The original germination data are in Cruz‐Tejada, Mattana, et al. (2024). The trait dataset generated for the analyses and the thermal modeling results are in the Supporting Information and are available in GitHub (https://github.com/DianaCruzTejada/MedGerm_DataAssessment), where the R codes used for the analyses are also available. A versioned archive of the repository is available in Zenodo (https://doi.org/10.5281/zenodo.20136467)

Articles from Conservation Biology are provided here courtesy of Wiley

RESOURCES