What is the value of biotic seed dispersal in post‐fire forest regeneration?

World forests face many threats, including wildfires, with tremendous ecological, social, and economic implications. Mediterranean ecosystems have evolved in the presence of fire, but changes to fire regimes associated with other global changes pose new challenges to postfire community regeneration. Forest regeneration largely depends on plant adaptations to survive wildfires or on the recolonization of burnt areas through seed dispersal. However, the value of the service provided by wild animals to postfire forest regeneration remains unevaluated. Here, we estimate the economic value of the biotic seed dispersal service in postfire restoration in Portugal. We combine budgeted costs of governmental Emergency Stabilisation Reports with network analysis that estimates the dependency of the Portuguese flora on biotic seed dispersal and on alternative fire regeneration strategies. Replacing the services provided by seed dispersers during postfire regeneration of Portuguese forests would cost 23,061,621€/year, highlighting the need for integrative policies that promote resilient forests.

which are intensifying due to ongoing climate change (European Commission, 2021).In this context, optimized strategies to increase the resilience of natural forests (i.e., their capacity to recover from disturbances; McCann et al., 2000), and halt biodiversity loss of forest species are recognized societal priorities reflected in introduced environmental policies (e.g., New Green Deal and the recent European Restoration Law; European Commission, 2019Commission, , 2022)).Portugal is among the countries most affected by wildfires, with an average of 122,649 ha burnt each year (ca.1.4% of the mainland territory; ICNF, 2022; Figure 1a), resulting from changes in forest composition, land abandonment, increasing number of ignitions and climate change, particularly drought intensification (Nunes et al., 2016;Parente & Pereira, 2016;Meira Castro et al., 2020).According to the National Forestry Registry (ICNF, 2015), 36% of the Portuguese mainland territory is covered by forests, with a further 31% dominated by woody Mediterranean shrublands, hereafter collectively considered as forests.Active reforestation (i.e., human-assisted) actions through different governmental and private initiatives replant on average 19% (23,245 ha) of the yearly burnt area, with the remaining 81% (99,403 ha) regenerating naturally (ICNF, 2022(ICNF, , 2023; Table S1).Most of the actively reforested area corresponds to production forests, and only c. 1.2% of the burnt area is targeted for ecological restoration (estimated from Emergency Stabilization Reports; Table S1).
In the context of wildfires, active restoration is a very important measure, for example to prevent soil degradation, assist vegetation recovery and compensate for limited seed dispersal or fight against the spread of invasive alien plants (Holl, 2012;Meli et al., 2017).It mainly consists of direct seeding and planting saplings grown in nurseries to accelerate forest regeneration.However, the extent of actively restored areas is limited by its high costs, the reduced number of species included in reforestation programs and their limited availability in plant nurseries (ICNF, 2013).In addition, the genetic pool of nursery plants is highly homogeneous and often less adapted to the local conditions.This might hinder the success of plantations and highlights the important complementary role of natural regeneration from burnt and neighboring areas (Jordan et al., 2019).In turn, the success of natural regeneration is highly variable as it depends on fire intensity, the presence of plant adaptations to survive wildfires, the quality of the seed bank, the proximity of unburnt areas that can supply propagules, and on the activity of natural seed dispersers (Busby & Holz, 2022;Kim et al., 2022).Therefore, the relative importance of different fire-coping strategies, such as resprouting (Moreira et al., 2009) and assisted recolonization through seed dispersal, will ultimately depend on a multitude of drivers, including floristic composition, landscape configuration, and fire intensity and frequency (Keeley et al., 2005).
However, for plants that can neither resprout nor germinate from the seed bank after fires, and also for those that cannot recruit under short fire-return intervals, recolonization via seed dispersal from unburnt areas is the only way to persist locally (Nolan et al., 2021).The transport of seeds may occur by abiotic (e.g., wind) or biotic (animals) vectors.Surprisingly, despite the large body of research on animal seed dispersal in ecology (Mason et al., 2022), we still know relatively little regarding the relative importance of animals in postfire vegetation restoration (Cavallero et al., 2013;Dafni et al., 2013).By allowing new plants to germinate and recruit far away from their parent plants, seed dispersal is a key regulating ecosystem service with direct implications for the long-term vegetation dynamics (Traveset et al., 2014).Seed dispersal is particularly important for the recolonization of degraded habitats, maintaining regional biodiversity (Gaston, 2022;Traveset et al., 2014), and for plants' capacity to adapt to changing climates (González-Varo et al., 2021).However, while the ecological value of seed dispersal is indisputable, its direct economic value for the human enterprise has rarely been assessed (e.g., Hougner et al., 2006;Leverkus & Castro, 2017;Puerta-Piñero et al., 2012).
Although it entails some limitations (e.g., Pascual et al., 2023), valuing ecosystem services has been very important to inform, influence and support conservation policies (Porto et al., 2020).For example, the demonstration of the dramatic economic consequences of pollinator declines helped to recognize the critical importance of pollination conservation policies (IPBES et al., 2019).Conversely, while the relevance of seed dispersal for ecosystems is well documented (e.g., Gaston, 2022), the direct benefits of healthy seed dispersal systems for humans is still poorly appreciated (Rost et al., 2012;Schoonover, 2022).A few studies have documented such benefits by using a Replacement Cost Analysis of seed dispersal services, though they were restricted to a few focal species (Hougner et al., 2006;Leverkus & Castro, 2017;Puerta-Piñero et al., 2012).
We argue that postfire restoration provides a perfect framework to explore the importance of biotic seed dispersal, particularly for plant species without adaptations to survive wildfires and that can only persist locally through recolonization.To do so, we performed an economic valuation of seed dispersal services by combining a Replacement Cost Analysis with an ecological network approach quantifying the dependence of the Portuguese flora on animal seed dispersers.

METHODS
Our methodological approach has three main steps (Figure 1).First, we performed a Replacement Cost Analysis to estimate the annual value of natural regeneration, measured as the cost of actively restoring the entire burnt area in Portugal per year.This was based on the estimated costs included on the national Emergency Stabilization Reports (Figure 1a; Table S1), which comprise the cost of planting seedlings and sowing.Second, we employed an ecological network approach to estimate the weighted dependency of the Portuguese native flora on biotic seed dispersal relative to three alternative fire-coping strategies (Figure 1b and Section 2.2).Lastly, we weighted the estimated cost of natural regeneration by the relative dependency of the flora on biotic dispersal (Figure 1c).

Replacement cost analysis
Several methods are available to estimate the economic value of ecosystem services (reviewed in Koetse et al., 2015).One of the most commonly implemented is Replacement Cost Analysis, which estimates the price required to substitute an ecosystem service by active human intervention (Freeman et al., 2014).A typical example is the cost implied in manually pollinating fruiting trees in the absence of wild pollinators (e.g., Allsopp et al., 2008).In the case of postfire restoration, replacement costs can be calculated as the cost required to manually plant or sow trees in the absence of natural seed dispersers.This method is preferable to Stated Preference analysis that requires a very high level of awareness and detailed information by questionnaire respondents (Hougner et al., 2006).We retrieved publicly available information on the planned costs of planting and sowing activities by the Portuguese government between 2015 and 2022 (see Table S1).Although these costs concern mostly the recovery of riparian galleries to protect water basins, they provide an informed perspective on the average cost of restoring one hectare of land.We also retrieved data on the total burnt area for the same period from public reports (ICNF, 2022).Based on this information, we calculated the average cost of actively restoring burnt areas with native trees and shrubs, which was estimated at 1450.0€/ha ± 486 (mean ± standard deviation).

Dependency of the Portuguese flora on biotic seed dispersal
Not all plants are equally affected by wildfires.Plant species can have different fire-coping strategies, namely, (1) the capacity to resprout after fire; (2) the capacity to endure fire in the form of seeds that germinate from the seed bank after the fire; (3) having adaptations to recolonize from adjacent unburnt areas by wind and water currents (wings, hairs, pappus or flotation tissues); or (4) having adaptations that promote animal-assisted recolonization of burnt areas (i.e., nutritive tissues to attract frugivores, or hooked and barbed diaspores that attach to animal fur).To upscale the analysis to an entire flora, we estimated to what extent the Portuguese flora depends on biotic seed dispersal or on alternative fire-coping strategies to survive in the regional species pool.
First, we retrieved the list of native plant species for mainland Portugal (Flora-On, 2023).We then collated information on the presence of each fire-coping strategy from two publicly available datasets.Information on postfire resprouting or seeding capacity was retrieved from the BROT 2.0 database (Tavşanoğlu & Pausas, 2018), while information on the presence of specific traits to facilitate biotic or abiotic seed dispersal was retrieved from the EuDis database (Vargas et al., 2023).For seed dispersal, we considered species with anemochorous, thalassochorous, hydrochorous, or ballochorous diaspores as adapted to abiotic dispersal, whereas species with endozoochorous, epizoochorous, myrmecochorous, or vertebrate hoarding diaspores were considered adapted to biotic dispersal.Before merging the three datasets (Flora-On, BROT, and EuDis), we harmonized plant scientific names according to the World Checklist of Vascular Plants, using the package U.Taxonstand 1.1.0(Zhang & Qian, 2023) in R 4.2.1 (R Core Team, 2022).Species with incomplete information about any strategy (i.e., not present in BROT or EuDis) were discarded from the analyses as their dependence on each of the four strategies could not be accurately accessed.We ended up with a list detailing the presence or absence of each fire-coping strategies for 477 plant species, corresponding to approximately one fifth of the Portuguese flora (Table S2).Although our analysis does not include the entire flora, we are not aware of any structural biases towards particular dispersal syndromes or fire-coping strategies.
Finally, we implemented an ecological network approach to estimate the relative importance of each fire-coping strategy for the assessed flora (Figure 1b), estimating the strength of each strategy by computing "species strength" (Bascompte et al., 2006) with the R package bipartite 2.17 (Dormann et al., 2008).Species strength is a network structure descriptor that quantifies the cumulative importance of each node (here each of the four strategies) to the nodes on the other level of a bipartite network (here the plant species).This is calculated summing the dependencies across all species J on each postfire strategy i, following the formula: , .Species dependency is given by  , =  , ∕  , where a i,j denotes a fire-coping strategy i of species j and A J the number of fire-coping strategies of species j.

Valuing biotic seed dispersal
The total economic value of natural regeneration for the entire burnt area per year in Portugal was estimated by extrapolating the average cost of active restoration (€/ha) to the average burnt area between 2015 and 2022.To single out the contribution of biotic seed dispersal, this was weighted to account only for the proportion of the flora estimated to depend on biotic seed dispersal, using the formula: × (    ) .
TA B L E 1 Representation of the four wildfire-coping strategies on the fraction of the Portuguese native flora (477 species) for which information on all traits is available (see BROT 2.0 and EuDis databases).et al., 2006).

Fire-coping strategies
The relative importance of biotic dispersal was obtained by dividing its species strength by the total species strength of all fire-coping strategies.

RESULTS
A large fraction of the assessed flora can germinate from the seed bank after wildfires (78% of the species), followed by species with resprouting capacity (54%), and those with adaptations for biotic (35%) or abiotic (28%) seed dispersal (Figure 1b; Table S2).Most species have multiple strategies, including 20 species that have all the fire-coping strategies (Table S2).Although only eight species are exclusively dependent on biotic dispersal, 166 species (ca.one third of those evaluated) have traits to facilitate animal dispersal.Accordingly, the relative importance of biotic dispersal is estimated at 16% (Figure 1b; Table 1).
Considering the cost of active restoration (1450€/ha), we value the service of natural regeneration of the 99,403 ha that burnt but were not actively reforested in 144,135,129€/year.Given the relative importance of the biotic dispersal strategies, we estimate that the annual value of biotic seed dispersal for the regeneration of burnt forests in Portugal in 23,061,621€ (Figure 1c).In turn, the value of abiotic seed dispersal is estimated at 18,737, 567€.

DISCUSSION
Here, we estimate the value of the ecosystem service provided by seed-dispersing animals in the context of postfire forest regeneration in Portugal.This service amounts to more than 23 million euros per year (232€/ha).As expected, a large proportion of the Portuguese flora is characterized by the ability to survive wildfires by resprouting or germinating from the seed bank.Nevertheless, this study reveals a yet underappreciated economic role of animals in assisting the natural recolonization of burnt forests with propagules from unburnt areas.We argue, however, that this is a likely a highly conservative estimate.
Resprouting and germinating from the seed bank are clearly the most important mechanisms through which Mediterranean plants can endure wildfires (Nolan et al., 2021).However, the importance of these mechanisms may decline as fire regimes change, particularly as firereturning frequency increases as part of the ongoing global changes (Robertson & Hmielowski, 2014;Shi et al., 2022).The importance of animal seed dispersal in facilitating the recolonization of disturbed areas, for example, after landslides, wildfires, floods, or agricultural activity, has long been advocated (Estrada-Villegas et al., 2022).However, the relative importance of wild animals in assisting the recolonization of burnt areas has never been estimated.Here, we estimated the contribution of animal seed dispersal to assist the recolonization of the c. 80% of burnt forest area where no active reforestation actions are implemented and that are left to recover passively.While only a few species (c.1.7%) seem to be exclusively dependent on animal seed dispersal, animals are still essential to maintain the genetic viability of populations that have other complementary fire-coping strategies (Leverkus & Castro, 2017;Puerta-Piñero et al., 2012;Traveset et al., 2014).This importance is reflected on the high number of species classified into this category (166 species).For that reason, the weighted proportion of the flora "dependent," that is, likely benefiting from having traits that facilitate biotic dispersal is higher (relative "species strength" = 16%).
When estimating the contribution of animals to promote postfire forest resilience one can take a species-centered (i.e., the contribution to maintain each plant species in the regional species pool), or an individual-centered perspective (i.e., the contribution to restore the relative proportion of individuals of all species in the community).While the second approach is valid it is highly context dependent, as it gives higher weight to the locally more dominant species and downplays the importance of maintaining rare species and is also dependent on the effectiveness of the local dispersers and their diversity.This would underestimate the consequences of the extinction of rare and "fugitive" species of high conservation priority, and, therefore, we opted to follow a species-centered focus.
While our estimate of the economic value of biotic seed dispersal is perhaps seemingly high, we argue that it is probably very conservative for the reasons presented below.First, active restoration activities include the plantation of only 40 species (ICNF, 2013), of which only 11 are native, corresponding to a very small fraction (c.0.5%) of the Portuguese flora.This strongly contrasts with the estimated proportion of the flora (35%) with adaptations favoring the dispersal by wild animals.Therefore, even where restoration has been accelerated by active restoration of target species, natural seed dispersal still plays a very important role as a complementary mechanism to assist the recovery of all other species, as well as to recover genetic diversity and forest structural heterogeneity (Díaz-García et al., 2020;Meli et al., 2017;Rodrigo et al., 2004).Second, it is critical to note that the importance of animal dispersers is not restricted to those plant species that have endozoochorous traits, such as fleshy fruited species or hooks (Green et al., 2022;Heleno et al., 2011).Third, the contribution of biotic seed dispersal can start before fires occur, contributing to build-up a diverse seed bank, a role that was not included in our estimate.Finally, the area that is actually reforested each year is smaller than the planned 19% as not all planned reforestation actions are implemented for different reasons.As a result, a larger fraction of the burnt forests is left to regenerate naturally than the one estimated here, increasing the replacement costs required should the entire burnt forest be actively restored.

Conclusion and future prospects
The world forests provide us with multiple ecosystem services, yet they are under increased pressure from the expansion of human enterprise, including by an intensification of wildfires (IPBES, 2018).It is thus vital to protect and restore degraded forests (European Commission, 2022), for which public funds should be used wisely (Birch et al., 2010;Bullock et al., 2022).Here we show that wild animals are essential in assisting the recovery of burnt forests and that losing this key ecosystem service would disrupt natural forest resilience and largely increase the costs to recover forests by active restoration alone.By taking advantage of a well-established network descriptor to estimate the relative importance of each fire-coping strategy for postfire recovery, our methodology can be easily applied in other contexts as long as a sizeable fraction of the flora has been screened for the presence of these strategies.Extrapolations should, however, be done with care as both the relative proportion of fire-coping strategies and the costs of active restoration actions are highly context specific.Naturally, these estimates can be further refined, by incorporating empirical information on the relative importance of the alternative fire-coping strategies for each plant species, or information on the differential effectiveness of natural dispersers vs. active restoration in relation to fire-size, distance to roads, seedling survival, different habitats, etc.However, these restoration approaches should always be seen as complementary rather than alternative (Meli et al., 2017, Bullock et al., 2022).
With 13% of the European bird species and 15% of the European mammal species threatened with extinction (BirdLife International, 2021;Temple & Terry, 2007) the resilience of European forests to endure threats, such as wildfires, is also rapidly eroding.This study calls for the need of integrated conservation actions that target not only forest species themselves, but functional and resilient forests.This requires the preservation of the critical support services provided by seed dispersers.Failing to do so will result in "empty" forests (sensu Redford, 1992) that will not be able to recover from disturbances without massive economic costs.

C O N F L I C T O F I N T E R E S T S TAT E M E N T
The authors declare no conflict of interest.

D ATA AVA I L A B I L I T Y S TAT E M E N T
Database EuDiS is available on the public data repository figshare through https://doi.org/10.6084/m9.figshare.22251028.v3(Vargas et al., 2023).

F
Conceptual diagram of the methodological approach developed to estimate the economic value of biotic seed dispersal for postfire forest regeneration in Portugal.(a) Estimated value of natural seed dispersal, based on the cost needed to replace this service by active reforestation of the entire area; (b) relative dependency of the Portuguese native flora (bottom gray bars) on four fire-coping strategies (top colored bars): resprout or germination of seeds in burnt areas or recolonization by seeds from unburnt areas into the affected area by biotic or abiotic dispersal; and (c) weighted estimate of the relative value of biotic seed dispersal in facilitating the natural regeneration of the burnt area per year in mainland Portugal.
R E F E R E N C E S