The reference ecosystem concept in ecological restoration is often defined based on one or a few reference sites (Gann et al., 2019), but ecosystems vary continuously over space, leading to natural differences in species composition across regions, i.e. beta diversity (Condit et al., 2002; Rosindell et al., 2012). Thus, a relevant question is how well restoration recreates beta diversity patterns at subcontinental scales. This question is timely as ecological restoration because ecological restoration is applied at increasingly large spatial scales (Aronson et al., 2020).

The representativeness of native species in ecological restoration initiatives is an issue that has been concerning the scientific community. Studies have identified that rare, endangered, and other groups of species are vulnerable in remaining forest fragments and not able to colonize areas undergoing restoration (Engert et al., 2020). In degraded and fragmented forest landscapes, planting native tree seedlings is a common and sometimes necessary restoration strategy (Rodrigues et al., 2009; Chazdon et al., 2021). These plantings introduce tree species with the objective of restoring basic ecological processes, such as the re-creation of a forest structure, elimination of invasive species, attraction of seed dispersers, and thus the promotion of natural regeneration processes (Holl, 1999). Studies have identified, however, that the arrival of new species in restoration plantings can take a long time or fail to occur (Reid et al., 2015; Shoo et al., 2015), so some species must be actively introduced. If they are not, their populations will be confined to dwindling and often degraded forest remnants (Suganuma and Durigan, 2021). However, current restoration practices are characterized by poor representation of the full set of regional species in the pool of species planted (Jalonen et al., 2018; Ladouceur et al., 2018; Almeida et al., 2024). Most planted species are abundant in forest remnants and are dispersed by wind rather than by animals (Brancalion et al., 2018; Engert et al., 2020). In addition, most planted species are pioneers with small seeds and rapid growth (Holl et al., 2017; Brancalion et al., 2018).

Biotic homogenization is the process of gradual replacement of endemic species by widespread ones over time (Olden and Rooney, 2006), and it can occur either through the increase in the abundance of invasive or generalist species (Beauvais et al., 2016). Thus, if restoration initiatives promote planting a small set of species throughout a region, and colonization by species from the surrounding native sites is limited, the landscape can become dominated by a restricted set of species, leading to biotic homogenization (McKinney and Lockwood, 1999; Olden and Rooney, 2006). Although the process of biotic homogenization is known for some regions, such as the Atlantic Forest (Zwiener et al., 2018; de Lima et al., 2020), we lack a comprehensive understanding on how restoration efforts may affect this process. This evaluation is important to define strategies and public policies to make restoration initiatives more effective in promoting regional biodiversity conservation.

The Atlantic Forest is a highly diverse but highly fragmented biome, with less than 16% of the original forest cover remaining, which is concentrated in small and isolated remnants (Ribeiro et al., 2009). To reverse this scenario, the Atlantic Forest has been undergoing large-scale restoration in recent decades (Rodrigues et al., 2009; Brancalion et al., 2013). We evaluated whether lists of species planted for restoration in the Brazilian Atlantic Forest represent the natural variation in tree species composition found in nearby forest remnants. We hypothesized that plantings are more floristically similar to each other than to forest remnants due to limitations imposed by the species pool currently available for planting (i.e., trees in nurseries; Ladouceur et al., 2018) and because some species are widely used in restoration plantings across the Atlantic Forest. Thus, we expected a lower beta diversity among and within restoration plantings compared to native forest remnants.

We found 423 species in the plantings and 1,891 in the remnants, totaling 1,954 species, with 81% of the species found in remnants absent in plantings (Table S3). In the site-to-site comparison for the whole Atlantic Forest (without splitting into forest types), the mean Jaccard dissimilarity index for remnants is 0.91 while for plantings it is 0.70 (Fig. 2A). In addition, beta-diversity is higher for planting-remnants and remnants-remnants comparisons than it is for planting-planting comparisons (Table S4). These results indicate that overall remnants are more floristically diverse than plantings.

Fig. 2.

Boxplot of beta diversity expressed by the Jaccard dissimilarity index and its turnover and nestedness values for plantings (orange) and remnants (green) for the whole Atlantic Forest (A–C) and for its three different forest types: Araucaria forest (AR, D–F), rainforest (RF, G–I), and seasonal forest (SF, J–L). In the analysis, each site (planting or remnant) was compared for beta diversity with each other site.

(0.23MB).

In site-to-site comparison within forest types, the mean Jaccard dissimilarity index is higher for remnants than for plantings in the three forest types (Fig. 2, Table S5), indicating that remnants are more floristically diverse than plantings across the Atlantic Forest biome. Turnover values are higher than nestedness (Fig. 2, Table S5), which indicates that at the scale of individual sites, dissimilarities in tree species composition are more related to species replacement from one site to another than to representation of species subsets.

At the higher level comparing the entire flora of different forest types, again remnants have higher beta diversity and turnover than plantings (Figure S1). However, at this scale, turnover is the main component of beta diversity only for remnants while for plantings it is nestedness (Figure S1). Turnover is also higher than nestedness when plantings are compared to remnants of the same forest type, except for the seasonal forest (Table S6). In addition, when we compare seasonal forest remnants to Araucaria forest and rainforest plantings, we unexpectedly also have a higher nestedness than turnover (Table S6), suggesting that a subset of the seasonal forest flora is predominating in the restoration of all forest types.

Our NMDS analysis split remnants into the three forest types: rainforest, seasonal forest, and Araucaria forest, but grouped all plantings together (Stress = 0.165, Fig. 3). In addition, we found strong evidence that forest type, land use (planting or remnant), and their interaction influence the floristic dissimilarities between groups (Table 1). Thus, plantings form a single group that was not differentiated by the regional forest type associated with their geographic location. In other words, plantings in one forest type are more floristically similar to plantings in other forest types than they are to forest remnants within their own forest type.

Fig. 3.

Nonmetric multidimensional scaling (NMDS) ordination of tree species in forest remnants (filled green symbols) and restoration plantings (hollow orange symbols) in rainforest (squares), seasonal forest (circles) and Araucaria forest (triangles) within the Brazilian Atlantic Forest.

(0.21MB).

Our results support the hypothesis that, in the Brazilian Atlantic Forest, plantings for forest restoration are floristically more similar to one another than to forest remnants of the same forest types. This trend could be even more pronounced if we had the complete list of species from remnants. Remnant species composition data come from samples and not from complete censuses, thus we likely underestimate remnant species richness and rare species presence; and rare species are frequently absent in restoration plantings (Almeida et al., 2024). Plantings are more frequently dominated by the same species than forest remannts, so there is a lower beta diversity among plantings, and they form a single floristic group across the Atlantic Forest. In other words, remnant forest communities shift strongly across geographic and environmental gradients, but the same tree species are often planted in restoration sites regardless of their geographic location. As such, tree planting-based restoration is failing to reproduce regional beta diversity patterns that characterize the Atlantic Forest biome.

Most of the difference in beta-diversity between restoration plantings and remnants was represented by turnover, indicating that the replacement of species plays a greater role than nestedness in their dissimilarity. However, when we consider only plantings, nestedness plays a greater role in their difference, which means that the less diverse groups are impoverished subgroups of the larger groups of species. Nestedness indicates that restoration plantings may be contributing to biotic homogenization, although this pattern is likely counterbalanced at some sites by natural regeneration inside the plantings, which was not accounted for in our analysis of tree planting lists.

Planting the same set of species across diverse regions could be an artifact of a framework species restoration strategy, in which a relatively small set of species with selected traits (e.g., rapid growth, fruit production) are planted to quickly recreate forest structure and the basic conditions to facilitate recolonization of restoration sites by plants and animals (Viani et al., 2015a; Elliott et al., 2022). If this recolonization process occurs, with natural regeneration of the more distinctive regional species from different forest types, planting the same set of species across diverse landscapes might not lead to biotic homogenization. However, restoration plantings are usually done in fragmented landscapes with low natural regeneration capacity, where the time needed for plantings to reach species compositions similar to reference forests is long and uncertain (Suganuma and Durigan, 2015; Beltrán et al., 2022). In such landscapes, the recolonization of restoration plantings by species from some functional profiles, such as slow-growing species and species dispersed by gravity or large mammals, will rarely fully happen (Pohlman et al., 2021; Suganuma and Durigan, 2021). In other words, some species will only reach restoration sites if they are introduced by plantings, which our analysis finds, is not occurring.

Biotic homogenization is not limited to tree plantings; it is also predicted to occur in native forest remnants due to climate change and human actions, including seed disperser decline (Nielsen et al., 2019; Fricke and Svenning, 2020). In the Atlantic Forest, biotic homogenization is driven by the expansion of small-seeded pioneer species that proliferate in forest edges and small fragments (de Lima et al., 2020), a process that can lead many species to become locally extinct (Zwiener et al., 2018). Small-seeded pioneers are also predominant in forest restoration plantings, a result of natural regeneration and the ease of collecting large quantities of small seeds for nursery propagation (Holl et al., 2017; Brancalion et al., 2018; Engert et al., 2020, Almeida et al., 2024). These observations from disturbed, regenerating, and restored forests further aggravate our concern that planted forests will remain floristically distinct from remnants.

Many tropical fragmented landscapes, including the Atlantic Forest, have experienced an erosion of biodiversity, particularly for late-successional, large-seeded, and endemic tree species in forest remnants. This pattern is exacerbated in the more fragmented regions where active restoration efforts are concentrated (de Lima et al., 2020). Regional declines of select tree types suggest that even if tree plantings and natural regeneration adequately represented the flora of local remnants, there would still be low representation in the long-term of some groups that are already missing or poorly represented in degraded landscapes. There is no easy solution to countering biotic homogenization in restoration, but improving the representation of local floras in plantings would help, as would promote landscape restoration approaches that aid the recovery of degraded fragments (Viani et al., 2015b) and the increase of forest cover to levels that can increase the probability that rare and endemic species arrive and survive in remnant and restored forests (de Lima et al., 2020; Hansen et al., 2020).

Our results indicate that forest restoration plantings require improvement to effectively contribute to biodiversity conservation in the United Nations Decade on Ecosystem Restoration and the 30 × 30 Biodiversity Target. In the world of large-scale restoration initiatives, the reference ecosystem concept needs to consider macroecological patterns, like species turnover through space (Condit et al., 2002). Failing to address this concern means that restoration is at risk of exacerbating biotic homogenization (Holl et al., 2022), and failing to contribute to the conservation of endemic, threatened and specialist species. Information on species with limited capacity to colonize plantings (Suganuma and Durigan, 2021) and lists of species with high conservation values for restoration (de Lima et al., 2020) are important, but they are only the first part of the plan to solve this issue. Additionally, it is important to increase the regional, endemic and/or threatened species in the pool of species available in nurseries and used for active restoration practices. This inevitably involves raising awareness, but also public policies and economic incentives that can help nurseries and practitioners to diversify their production and uptake in large-scale restoration initiatives.

Crislaine de Almeida: Conceptualization, Methodology, Formal analysis, Data Curation, Writing - Original Draft. J. Leighton Reid: Methodology, Writing – Reviewing and Editing. Renato A. Ferreira de Lima: Data Curation, Writing - Review & Editing. Luis Fernando Guedes Pinto: Resources, Writing- Reviewing and Editing. Ricardo Augusto Gorne Viani: Conceptualization, Methodology, Formal analysis, Writing- Reviewing and Editing, Supervision.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.