Contents lists available at ScienceDirect

Forest Ecology and Management

journal homepage: www.elsevier.com/locate/foreco

Ecological restoration increases conservation of taxonomic and functionalbeta diversity of woody plants in a tropical fragmented landscape

Débora Cristina Rothera,b,⁎, Ana Paula Libonia,b, Luiz Fernando Silva Magnagoc, Anne Chaod,Robin L. Chazdone, Ricardo Ribeiro Rodriguesb

aUniversidade Estadual de Campinas, Instituto de Biologia, Departamento de Biologia Vegetal, Campinas, São Paulo, BrazilbUniversidade de São Paulo, Escola Superior de Agricultura “Luiz de Queiroz”, Laboratório de Ecologia e Restauração Florestal, Piracicaba, São Paulo, BrazilcUniversidade Federal do Sul da Bahia, Itabuna, Bahia, BrazildNational Tsing Hua University, Institute of Statistics, Hsin-Chu 30043, TaiwaneUniversity of Connecticut, Department of Ecology and Evolutionary Biology, Storrs, CT, USA

A R T I C L E I N F O

Keywords:Diversity recoveryDiversity metricsLarge-scale restorationSeedling recruitmentTaxonomic and functional beta diversityTropical restorationTropical forests

A B S T R A C T

Ecological restoration can re-establish plant species populations, enhance forested habitats extension, improvelandscape connectivity, and enable biodiversity persistence within a landscape. However, the potential benefitsof ecological restoration on beta diversity have never been explored. Here we use field data to investigate, for thefirst time, if restoration plantings enhance the taxonomic and functional plant beta diversity in a fragmentedlandscape of the threatened Atlantic forest. Woody species were evaluated for 320 plots established in 18 forestfragments and 14 restoration plantings within a sugarcane production landscape with low forest cover, insoutheastern Brazil. Diversity metrics were assessed using the multiple incidence-data version of Hill numbersand were compared among three sets of study sites: fragments, restoration plantings and the two combined.Fragments showed higher levels of alpha diversity and proportional abundance of non-pioneer and animal-dispersed species than restoration plantings. Exotic, pioneer and non-zoochoric species were more abundant inrestoration plantings, an expected result considering sites still be in the early or mid-successional stages ofdevelopment. Taxonomic and functional beta diversity of trees was greatest when both areas were combined. Forregenerating plants, however, beta diversity results varied according to species incidence-based frequencies.Although restoration plantings do not result in full recovery of alpha diversity, they can all together complementdiversity of forest fragments at the landscape level. The findings indicate two key ecological implications forbiodiversity conservation: the critical importance of forest fragments as biodiversity repositories and the positiveeffect of restoration efforts on landscape-scale diversity in degraded regions. These novel results highlight theimportance of species selection for restoration initiatives toward species and functional attributes recognized assignificantly reduced or locally rare. Overall, forest fragments and restoration plantings can act synergistically topromote recovery of plant diversity in heavily deforested agricultural landscapes.

1. Introduction

Forest restoration plans are an important tool to conserve biodi-versity (e.g., endangered species), and to enhance ecosystem services(Benayas et al., 2009). Restored forests play an essential role in con-servation at the landscape scale, as they provide additional forest coverfor threatened species, reconnect previously isolated biological popu-lations of plants or animals (Brancalion et al., 2013; Rother et al., 2018)and provide important functional properties (e.g., food for fauna orwood production) (Garcia et al., 2015; Silva et al., 2015; Montoya-Pfeiffer et al., 2018). Understanding the role of restored forests in

mitigating the effects of fragmentation and habitat loss is crucial, as alarge proportion of tropical forest biodiversity is restricted to smallfragments embedded within an agricultural or pasture landscape matrix(Arroyo-Rodríguez et al., 2013).

Forest habitat reduction and isolation can negatively affect taxo-nomic and functional dimensions of biodiversity (Arroyo-Rodríguezet al., 2013; Solar et al., 2015). Also, changes in forest cover and foreststructure may lead to taxonomic and functional simplification and thehomogenization of biota within regions (Arroyo-Rodríguez et al.,2013). Small and degraded forest fragments may hold missing im-portant groups of species, such as large-seeded species (Oliveira et al.,

https://doi.org/10.1016/j.foreco.2019.117538Received 6 June 2019; Received in revised form 9 August 2019; Accepted 10 August 2019

⁎ Corresponding author at: Universidade Estadual de Campinas, Instituto de Biologia, Departamento de Biologia Vegetal, Campinas, São Paulo, Brazil.E-mail address: deborarother@gmail.com (D.C. Rother).

Forest Ecology and Management 451 (2019) 117538

0378-1127/ © 2019 Elsevier B.V. All rights reserved.

T

2008), species pollinated and/or dispersed by vertebrates (Lopes et al.,2009) and understory and shade tolerant species (Tabarelli et al.,1999). In contrast, a number of disturbance-tolerant native species canbecome abundant in fragmented forests (Santos et al., 2012), con-tributing to increased levels of floristic homogenization at a range ofspatial scales (Arroyo-Rodríguez et al., 2013). Within this scenario,even forest fragments and restored forests emerge as critically im-portant sources of biodiversity, for supporting ecosystem functions andecosystem services (Arroyo-Rodríguez et al., 2017), these two eco-system types may differ in important ways about their biodiversity andthe ecosystem functions they support. While forest fragments retain theremaining species after disturbances, restored forests (active restora-tion) emerge from active human management, such as mixed-speciesplantations from nursery-grown seedling (Rodrigues et al., 2009;Crouzeilles et al., 2017).

In the Brazilian Atlantic forest, decision making on restorationstrategies and their outcomes are particularly important due to the needto an effective compliance with the Native Vegetation Protection Law(Garcia et al., 2013). The compliance with the law is also essential since90% of the native vegetation remaining in the biome is in private ruralproperties, rather than in protected areas (Soares-Filho et al., 2014;Brancalion et al., 2016). Agricultural companies and landowners,especially those devoted to sugarcane production, have initiated activerestoration plantings in riparian corridors on their lands to comply withregulations and to obtain environmental certification for their products(Rodrigues et al., 2011; Brancalion et al., 2013, 2016). Most ecologicalrestoration initiatives in such private properties are concentrated inriparian areas illegally cleared for agricultural use (Rodrigues et al.,2011).

Tropical restoration efforts can have different aims and approaches.Increasing the likelihood of native species persistence in fragmentedlandscapes, targeting plant species exploited that have become rare inthe region, and restoring habitats and ecosystem services are someapproaches of how restoration can be used (Brancalion et al., 2010,2013). In general, active restoration includes, among other methods,tree plantings with high native species diversity (e.g., about 80 treespecies) composed of different functional groups (Rodrigues et al.,2009). It is important to highlight that the number of species used in theactive restoration can vary according to environmental laws of eachBrazilian state. Also, the restoration method to be applied depends onthe aim of the project, local characteristics and landscape context.

Some recent studies have examined restoration sites from differentperspectives and for different biological taxa including invertebrates(Audino et al., 2017; Crouzeilles et al., 2017; Shuey et al., 2017), birds,mammals, amphibians, and plants (Crouzeilles et al., 2017). Althoughnatural regeneration and active restoration have been evaluated atlandscapes scales (Crouzeilles et al., 2017; Farah et al., 2017), severalimportant questions remain poorly understood, mainly when treecomposition and functional attributes used in active restoration projectswere not known (the initial planted species is unknown). Here, weaddress two specific questions: (i) Can restoration plantings enhancebeta diversity for species and functions at landscape scale compared toforest fragments? and (ii) How are functional attributes (e.g., wooddensity, seed length, dispersal syndromes and ecological strategy) dis-tributed among forest fragments and restoration plantings? To answerthese questions, we evaluated taxonomic and functional diversity atdifferent levels, using metrics based on species richness and abundancedata, in the Atlantic Forest of southeastern Brazil. This study presentsand analyses the findings on diversity metrics at different scales anddiscuss the critical importance of remaining forests for plant diversityconservation and the effects of restoration efforts on landscape-scalediversity in tropical degraded regions.

2. Material and methods

2.1. Study region

The study region is located in an ecotonal region between theAtlantic Forest and Cerrado domains (savanna) in northern Sao PauloState, Brazil (Fig. 1). It is considered as the most threatened region inthe country, as it presents high biodiversity, endemism and anthropicthreats (Klink and Machado, 2005; Ribeiro et al., 2009). Since ecolo-gical restoration often has been implemented on private lands incompliance with the Native Vegetation Protection Law (Rodrigueset al., 2011; Sparovek et al., 2012), the region also comprises manyrestoration plantings (a total of 670.49 ha planted in 97 areas). Most ofthese projects are part of the “Environmental Planning Program ofFarms” carried out from 2000 to 2010 by the Laboratório de Ecologia eRestauração Florestal, University of Sao Paulo (Rodrigues et al., 2011).

The region is covered by about 10% of native vegetation (forest andwooded savanna), restricted to small fragments of secondary vegetationin different stages of succession (Kronka et al., 2005) (Fig. 1). The

Fig. 1. Location of the woody plantcommunities studied in 18 forest frag-ments (black squares) and 14 restora-tion plantings (red circles) located in anAtlantic Forest-Cerrado ecotone insoutheastern Brazil. A description ofindividual forest fragments and re-storation plantings can be found inTable S1. (For interpretation of the re-ferences to colour in this figure legend,the reader is referred to the web versionof this article.)

D.C. Rother, et al. Forest Ecology and Management 451 (2019) 117538

2

remaining patches are interspersed within a matrix dominated by su-garcane plantations (> 90%), with less extensive areas of commercialforestry (Eucalyptus spp. and Pinus spp.), fruit crops (Citrus spp.) andcattle pastures (SigRH, 2016). Climate in the region is a transition be-tween Aw (tropical savanna climate, influenced by the altitude) andCwa (humid subtropical climate, with wet summers and dry winters).The annual mean temperature is around 21 °C, January being thewarmest (23.5 °C) and July the coldest month (17.5 °C) (Alvares et al.,2013).

We conducted our study in 14 riparian restoration plantings from 3to 7 years old and varying in size from 0.4 to 34.63 ha, covering a totalof 146.14 ha. All restoration plantings present a closed canopy, receivedgrass control treatments during the first two years and are in privatesugarcane farms. Species selection for restoration plantings were basedon their eco-physiological attributes such as fast- (called as cover spe-cies) and slow-growing species (called as diversity species). Seedlingswere planted following a 3× 2m scheme resulting in 1666 in-dividuals/ha. Species richness in the restoration plantings varied from16 to 50 plant species. For the forest fragments we included only thosefragments of semideciduous forests from the map of remaining vege-tation of Sao Paulo State. That map included 526 of semideciduousforests, 32 lowland forests and 15 of wooded savanna (Kronka et al.,2005). Our data set resulted in 18 forest fragments from 2 ha to 120 hain area, totaling 1240.68 ha. Species richness varied from 24 to 65 plantspecies per fragment. Both forest fragments and restoration plantingshave elevation from 635 to 940m. Details about each restorationplanting and forest fragment are available in Appendix S1, S2, and S3:Table S3. Fragments included upland while restoration plantings in-cluded both, riparian (nine) and upland areas (five). The list of speciesproduced in the nurseries and used in these restoration plantings wasprovided by floristic surveys made in both upland and riparian areas ofthe region. Seeds to produce seedlings in the nurseries were collected inthose areas.

2.2. Woody plant sampling

Within each forest fragment and restoration planting, we located 104× 25m plots (total of 0.1 ha per site), oriented in north–south di-rection, at least 20m distant from each other and excluding a 10m edgebuffer. We recorded all trees, palms and shrubs with diameter at breastheight (DBH) > 3.18 cm in the plots (hereafter established plants).Within each plot, we established one 1× 25m subplot (total of0.025 ha per site) to sample natural regeneration of woody plantshigher than 0.5m and with DBH≤ 3.18 cm (we called regeneratingplants). The total area sampled was 3.2 ha. Species not identified in thefield were collected for subsequent identification (APG III, 2009) in theESA Herbarium (University of Sao Paulo).

2.3. Functional traits

We measured functional traits related to morphological and phy-sical adaptations of woody species in their role as food resources,carbon storage and forest structure (Tabarelli and Peres, 2002; Bongerset al., 2009; Tabarelli et al., 2010; Magnago et al., 2014): seed dispersaltype and seed size (reproductive traits), ecological strategy, and wooddensity. The dispersal type was categorized as zoochorous and non-zoochorous following Van der Pijl (1982). Species that could not beclassified in these two groups were labelled “unclassified”. A zoochoricspecies produces diaspores surrounded by fleshy pulp, a fleshy appen-dage of the seed (aril) or has other features that are typically associatedwith dispersal by frugivorous animals, and a non-zoochoric species hascharacteristics that indicate dispersal by abiotic means, such as wingedseeds, feathers, or a lack of features that indicate dispersal via methodsother than gravity or explosive indehiscence (Magnago et al., 2014).Data for seed size was obtained through literature review.

Species were classified into three ecological strategies (pioneer,

non-pioneer canopy, and non-pioneer understory) according to theirlight requirements and occurrence in forest strata (Barbosa et al.,2015), species growth form (Carvalho et al., 2010) and expert knowl-edge (Drs. Natália Ivanauskas, Vinícius Castro Souza and Ricardo Ri-beiro Rodrigues). We considered as pioneers those species that developin conditions of high light levels and usually do not regenerate in theunderstory; non-pioneer canopy species develop in intermediateshading conditions but do not remain in the understory (early sec-ondary and emergent species); while non-pioneer understory speciesthose develop exclusively in the shaded understory. Species withoutsufficient knowledge to classify them otherwise were called unclassified(6.7% of the established plant species and 8.2% of the regeneratingplant species). We classified those species living outside their nativedistribution range and those species that occur exclusively in savannaformations as “exotic” (Oliveira-Filho, 2014). Due to the importance tounderstand how this factor is representative in the communities, exoticspecies was considered as a trait in the analysis to distinguish thecontribution of each functional attribute.

Data for wood density in dry weight (g/cm3) were obtained fromThe Global Wood Density (GWD) database in the subsection TropicalSouth America (Chave et al., 2009). For morphospecies only identifiedto the family or genus level, we used the average wood density of thetaxonomic group; and for species not in the GWD database, we used theaverage wood density for the genus (following Flores and Coomes,2011; Hawes et al., 2012; and Magnago et al., 2014).

Species identified at morphospecies and genus levels represented24% of the total species richness and 2.24% of the total abundance.These species were not included in the floristic diversity analysis butwere included in the species richness analysis. Specimens identified togenus level were included in the functional diversity analysis. Analysesfor taxonomic and functional beta diversities included only regionalnative species.

2.4. Data analysis

To describe the plant community in forest fragments and restorationplantings we constructed separate rank abundance curves for eachstratum and plotted the abundance of each plant species in a bi-di-mensional plane from the most to the least abundant (Magurran, 2004).We used the library “sads”, the “fitsads” function and the “poilog”distribution (Prado and Miranda, 2014). Analyses were performed in Rsoftware version 3.4.2 (R Development Core team, 2017).

To distinguish the contribution of each functional attribute of eachfunctional trait per forest type (i.e., forest fragments, restorationplantings or the two combined) and per plant strata we followed twosteps. First, we calculated the Community-Level Weighted Means ofeach trait (CWM). Second, we evaluated generalized linear models(GLM) or generalized linear mixed models (GLMM), using CWM valuesas response variables and forest type (forest fragments and restorationplantings) as explanatory variables. We applied GLM with Gaussianfamily for CWM values of wood density and seed length. However, forthe others CWM values of proportional abundance (seed dispersal typeand ecological strategy), we used linear mixed model with binomialfamily. For these models we used “sites” as the random effect. Thisstatistical approach was used to control the individual-level varianceand this way, correcting the overdispersion problem (Harrison, 2015).The GLM models were built using the “glm” function from R base. Tothe GLMM we used the “glmer” function from the “lme4” package. Allanalyses were performed in R software version 3.4.2 (R DevelopmentCore team, 2017).

For a given area type (fragment, restoration planting, or the twocombined) and plant strata (established or regenerating), we can formtwo kinds of data: (1) species abundance data, and (2) multiple in-cidence data (the count of occurrences of each species among sites). Wefocused our analysis on multiple incidence data because such data aremore robust to clustering/aggregation of individuals than abundance

D.C. Rother, et al. Forest Ecology and Management 451 (2019) 117538

3

data (Colwell et al., 2012). We treat each site as a sampling unit, recordonly species incidence (presence/absence) in each site, and pool speciesincidences to obtain the count of occurrences of each species amongsites (“species incidence-based frequency/abundance”). We also treatthe number of occurrences among multiple sites of each species as aproxy of its abundance (Chao et al., 2014).

Taxonomic and functional metrics were used to: (i) identify betadiversity patterns at species level and (ii) indirectly understand theecological processes and ecosystem functioning (e.g., plant-animal in-teractions; Galetti et al. (2013) and carbon stock; Brancalion et al.(2018)). We separately analyzed these metrics for the two vegetationstrata. For restoration plantings, established woody plants reflect theinitial planting, whereas woody regenerating strata reflect successionalprocess as affected by species immigration or recruitment from fruitingplanted trees. For forest fragments, established woody plants show theoutcomes of initial colonization, while woody seedlings and saplingsindicate the regenerating potential along the successional process inagricultural landscapes.

(a) Rarefaction/extrapolation for within-site alpha diversity

Hill numbers (effective number of species) are increasingly used forquantifying species diversity based on species relative abundances(Hsieh et al., 2016). Chao et al. (2014) formulated multiple incidence-data version of Hill numbers based on species relative incidence prob-abilities/rates; see Supporting Information for model formulation andmathematical formulas. The incidence-based Hill numbers are alsoparameterized by a diversity order q, which determines the measures’sensitivity to species relative incidence probabilities. These incidence-based Hill numbers include three species diversity measures as specialcases: Species richness (q=0), Shannon diversity (q=1) and Simpsondiversity (q=2); the latter two measures are computed from speciesrelative incidence probabilities. In what follows, Hill numbers refer tothe version based on multiple incidence data. Since each site is treatedas a sampling unit, sample size refers to the number of sampling units(or sites).

In assessing the species diversity patterns of the three types of areas(fragment, restoration planting, or all together), we first compared foreach type of area Hill numbers (within-site alpha diversity) based onmultiple incidence data. Chao et al. (2014) extended the conventionalrarefaction/extrapolation methods for species richness to Hill numbers.We mainly focus on comparing equally large samples (i.e., standar-dizing the number of sites) because the two standardization methodsyield generally similar patterns. For each type of area, we used theiNEXT software (Hsieh et al., 2016) separately for established and re-generating plants to obtain the sample-size-based rarefaction/extra-polation sampling curves along with 95% confidence intervals forq=0, 1 and 2.

(b) Rarefaction for among-site taxonomic beta diversity

Like alpha diversity, beta diversity also depends on sample size.Beta diversity measures the extent of difference in species compositionamong sites. Thus, beta diversity depends on how many sites are in-volved in computing the differences. Using the same idea of rarefactionfor alpha diversity, here we also rarefied the data (by standardizing the

number of sites involved in computing beta diversity) in order toevaluate the hypothesis that taxonomic and functional beta diversitywill increase when all sites are analyzed together. A statistical methodfor extrapolation of taxonomic and functional beta diversity is still notavailable.

There are two major approaches to evaluate beta diversity: (i) thevariance framework, derived from the total variance of species abun-dances (Legendre and De Cáceres, 2013), and (ii) diversity decom-position based on partitioning gamma diversity into alpha and betacomponents. Chao and Chiu (2016) recently bridged the above twomajor beta approaches by showing that both converge to the sameclasses of (dis)similarity measures. Based on species relative incidenceprobabilities, we considered the rarefaction for the two convergentdissimilarity measures, which are monotonic functions of beta diver-sities: (i) a class of Sørensen-type species-dissimilarity measures: 1-CqN

(average proportion of non-shared species in a site), and (ii) a class ofJaccard-type species-dissimilarity measures: 1-UqN 1-UqN (effective pro-portion of non-shared species in the pooled site).

(c) Rarefaction for functional beta diversity

Based on species relative incidence probabilities and species pair-wise Gower distance between calculated from species traits, Chiu andChao (2014), and Chao et al. (2014) extended the above two classes ofspecies dissimilarity measures to their functional version. These func-tional dissimilarity measures can be adapted to our multiple incidencedata. As with taxonomic beta diversity, we consider rarefaction of thefunctional version of the Sørensen-type and Jaccard-type functional-dis-similarity measures based on four species traits described earlier. BothJaccard and Sorensen-type dissimilarity measures for taxonomic andfunctional data revealed consistent patterns, so we present the Jaccardindex in the main text and the Sørensen in Supporting Information(Appendix S5: Fig. S1 and Appendix S6: Fig. S2).

3. Results

3.1. Overview on plant species community

Considering the 32 areas we recorded 5990 established woodyplants belonging to 342 species (gamma diversity) across 3.2 ha (320plots). Forest fragments and restoration plantings shared 83 species; 66occurred exclusively in restoration plantings and 193 were found ex-clusively in fragments. For regenerating plants, we recorded a total of14,468 individuals belonging to 389 species (gamma diversity) re-corded in 0.8 ha (320 subplots). For the regenerating strata, forestfragments and restoration plantings shared 104 species, with 55 ex-clusives for restoration plantings and 230 for fragments. Alpha diversityfor established plants varied from 24 to 65 species for fragments, andfrom 16 to 50 for restoration plantings. For regenerating plants, alphadiversity varied from 33 to 114 species for fragments and from 14 to 43species for restoration plantings. Separate data for each condition andstratum is presented in Table 1 and diversity data for each area andstratum are available in Appendix S2: Table S2 and Appendix S3: TableS3.

Table 1Descriptive data for woody plants from forest fragments and restoration plantings of an Atlantic Forest-Cerrado ecotone in southeastern Brazil.

Area Size-class Total abundance Total incidence Observed species richness Estimated species richness (Chao2)

Fragments Established 3953 928 276 388.891Regenerating 10,685 1306 334 457.836

Restoration plantings Established 2037 476 149 212.390Regenerating 3783 411 159 244.224

D.C. Rother, et al. Forest Ecology and Management 451 (2019) 117538

4

3.2. Species dominance, richness, and alpha diversity

Contrasting patterns of species richness are clearly displayed for thewoody plant communities in the dominance-diversity curves for con-ditions and strata (Fig. 2). The dominance-diversity curve that re-presents the regenerating plants in the fragments covers a greaterwidth, indicating a greater richness than the regenerating plants inrestoration plantings and than the established plants both in fragmentsand restoration plantings. That difference is also noted by the differencein the length of the tails of each curve. The shape of the curves high-lights differences in evenness amongst communities. Steep curves forestablished and regenerating plants in restoration plantings signifycommunities with high dominance of a few species (Fig. 2).

The rarefaction/extrapolation sampling curves for multiple in-cidence-based frequency data for established plants and regeneratingplants for species diversity for order q=0 (Species richness), q=1(Shannon), and q=2 (Simpson) are shown in Fig. 3. Standardizing forthe number of sites, fragments and all areas combined show similarpatterns of species diversity and significantly higher species diversitythan the restoration plantings (Fig. 3a and 3b, respectively). Thesetrends were consistent across all three Hill numbers.

3.3. Functional traits

The CWM revealed that all functional trait values varied sig-nificantly between fragments and restoration plantings for both estab-lished (Fig. 4a) and regenerating plants (Fig. 4b). For established plantswe found that restoration plantings, when compared to fragments,showed a lower proportional abundance of species with higher wooddensity values (t=−2.146; p=0.0401), seed length (t=−2.269;p=0.0306), non-pioneer canopy (z=−5.725; p=0.0001), unders-tory non-pioneer (z=−7.926; p=0.0001) and zoochoric species(z=−4.019; p= 0.0001). Restoration plantings showed higher meanvalues of exotic (z= 2.0868; p=0.0001), pioneer (z= 10.07;p=0.0001), and non-zoochoric (z= 4.008; p= 0.0001) speciesabundance (Fig. 4a).

The regenerating strata show patterns similar to established species(Fig. 4b). Restoration plantings have lower wood density mean values(t=−3.267; p=0.003), lower seed length (t=−3.386; p=0.002)and lower incidence of zoochory (z=−3.610; p=0.0001) comparedto fragments. Also, the exotic (z= 2.218; p= 0.03), pioneer(z= 6.275; p=0.0001) and non-zoochoric (z= 3.610; p= 0.0001)species increased their proportional abundance in restoration plantingscompared to fragments.

Comparing strata of restoration plantings, regenerating plants have

higher proportional abundance of understory non-pioneer species(z=−4.306; p=0.0001) compared to established plants. However,non-pioneer canopy species reduced their proportional abundance inregenerating stratum (z=−5.508; p= 0.0001) (Fig. 4b).

3.4. Taxonomic beta diversity across sites and regional scales

As we initially hypothesized, taxonomic beta diversity (and dis-similarity) was higher for all sites combined (forest fragments and re-storation plantings) in comparison to only forest fragments or only re-storation plantings (Fig. 5). For established plants, beta diversity valuesshowed the following ordering: All > Fragments > Restorationplantings (Fig. 5a). For regenerating plants, if species incidence-basedfrequencies are considered (q=1 and q=2), beta diversity followedthe order All > Restoration plantings > Fragments. For q=0 (in-cidence-based frequencies are not considered) beta diversity was:All ~ Restoration plantings > Fragments (Fig. 5b). Species composi-tion of seedlings and saplings varied more among restoration plantingsthan among fragments.

3.5. Functional beta diversity change

All functional dissimilarity patterns among sites were similar tothose for species dissimilarity, with some differences in magnitude(Fig. 6). For established plants, functional beta diversity showed thefollowing ordering: All > Fragments > Restoration plantings(Fig. 6a). For regenerating plants, if species incidence-based frequenciesare considered (q=1 and q=2), functional beta diversity followed theorder All > Restoration plantings > Fragments (Fig. 6b). However,for q=0 (rare species are emphasized), we have ‘All sites combined’showing similar functional beta diversity to Restoration plantings andthese two conditions with higher functional beta diversity than theFragments: All ~ Restoration plantings > Fragments (Fig. 6a and 6b).Functional traits of seedlings and saplings were more different amongrestoration plantings than among fragments.

4. Discussion

Considering the current concerns about ecosystem resilience tolarge-scale anthropogenic impacts (Nimmo et al., 2015; Arroyo-Rodríguez et al., 2017), our study shows for the first time that re-storation plantings can contribute to increase the conservation oftaxonomic and functional beta diversity in highly fragmented land-scapes. Results demonstrate the importance of establishing restorationplans that focus on planting a diversity of species selected from theregion. For functional composition, some groups were significantlyunderrepresented among trees and shrub species used to restore areas(e.g., large-seeded, zoochoric and understory species) and this wasconsistent for the regenerating stratum. Although functional groups andspecies deficits are expected to be compensated throughout the processof natural regeneration, our sites are still at early stages of ecologicalsuccession. Small-seeded, non-animal dispersed and pioneer specieswere the most abundant groups occurring in restoration plantings, butnon-pioneer species typically found in the understory had a consider-able increment in the regenerating strata (Fig. 4a and b), maybe as aresult of the seeds dispersed from both the surrounding forests and thelocal species already in reproductive age in the restoration plantings.Our results show the clear importance of restoration plantings for in-creasing plant beta diversity in fragmented landscapes via reintroduc-tion of native species, but these plantings still lacked functional groupsof threatened species (e.g., high wood density, large-seeded, animaldispersed and non-pioneer species), which can compromise the provi-sion of ecosystem services (Brancalion et al., 2018). These findingsreinforce, for example, the strong impact of species selection on geneticdiversity (Zucchi et al., 2018), carbon stock (Brancalion et al., 2018)and plant-animal interactions (Galetti et al., 2013).

Fig. 2. Dominance-diversity curve of woody species community sampled in 32sites (18 fragments - green line, 14 restoration plantings – blue line, and 32combined areas – magenta line) and two strata (established and regeneratingplantings) in southeastern Brazil. The X-axis ranks each plant species in orderfrom most to least abundant. (For interpretation of the references to colour inthis figure legend, the reader is referred to the web version of this article.)

D.C. Rother, et al. Forest Ecology and Management 451 (2019) 117538

5

Despite the legal instruments and efforts to ensure higher diversityin restoration plantings in Brazil, specifically in Sao Paulo state(Brancalion et al., 2010), species alpha diversity was consistently lowerin restoration plantings than in forest fragments. Some studies haveshown that alpha diversity of young restoration plantings (5 to 7 years-old) is still much lower than that found in natural forests (Costa et al.,2016). This limitation may be due to the restricted availability of nativespecies in seedling nurseries. The low potential of nurseries to provide ahigher fraction of local tree flora can be attributed to the nurseries’clustered distribution on the territory, underused production capacity,

and lack of seed supply and qualified labor (da Silva et al., 2017).Planning and executing ecological restoration actions in agriculturallandscapes is a challenging task that requires improved policies andprotocols to attend high-diversity forest ecosystem patterns (Rodrigueset al., 2009; Chaves et al., 2015; Costa et al., 2016; da Silva et al.,2017).

For established plants, taxonomic and functional beta diversity werehigher in fragments than in restoration plantings, but the opposite wasobserved for regenerating plants. This result reflects that establishmentof restoration plantings using the same techniques and similar woody

Fig. 3. Sample-size-based rarefaction (solid line) and extrapolation (dotted line) for established plants (A) and regenerating plants (B) based on multiple incidencefrequency data (the count of occurrences of each species among sites) for species diversity of q=0 (species richness), q=1 (Shannon diversity) and q=2 (Simpsondiversity) with 95% confidence intervals (shaded areas); Each X-axis denotes the number of sampling units (sites). All extrapolations are extended to double samplesize.

D.C. Rother, et al. Forest Ecology and Management 451 (2019) 117538

6

species composition, whereas natural regeneration showed more vari-able patterns in plant colonization. Forest fragments, in turn, are olderand structurally more complex than restoration plantings, and stillpreserve some old long-lived species that became rare in the currentfragmented agricultural landscapes, due to selective logging, defor-estation and fires (Melo et al., 2013). Such species are expected to ex-hibit a time-lag in their biological responses to landscape changes dueto their longevity (Metzger et al., 2009), and thus, rare large-seeded andnon-pioneer individuals that persisted in the remaining forests after thedisturbances may increase taxonomic and functional beta diversity

among fragments for the established stratum.Although age and local conditions (e.g., soil and light character-

istics) have not been evaluated, the results for regenerating plants inrestoration plantings suggest that those variables are important driversof the compositional differentiation of woody plant assemblages, aspreviously observed for plant communities in grasslands (Stuble et al.,2017), and tropical forests (Meli et al., 2017; Suganuma et al., 2017).Variation in conditions across sites is reported as an important de-terminant of community assembly, producing a marked spatial andtemporal variation in restoration outcomes (Stuble et al., 2017;

Fig. 4. Community-Level Weighted Means (CWM) of each functional trait for stratum (established (A) and regenerating plants (B)) and forest type (fragments andrestoration plantings).

D.C. Rother, et al. Forest Ecology and Management 451 (2019) 117538

7

Suganuma et al., 2017). The observed divergence in regenerationcommunity of restoration plantings reinforce that direct species re-sponses to local conditions can influence germination, emergence, es-tablishment and survival (Classen et al., 2010). Moreover, indirect re-sponses mediated by interspecific interactions can also be importantdrivers of site and age effects (Stuble et al., 2017). Effects of recentenvironmental changes on woody species are expected to be moreevident in the community of the regenerating stratum (Rigueira et al.,2013; Benchimol et al., 2017), as seedlings and saplings are moresensitive to recent disturbances than established long-lived plants. Thehigher taxonomic and functional beta diversity for regenerating plantsamong restoration sites compared to forest fragments when rare speciesare emphasized could be driven by greater heterogeneity in understoryenvironmental conditions among the restoration plantings and to dif-ferences in proximity to forest fragments. This result highlights theimportance of each restoration planting of the surrounding landscapefor the conservation and maintenance of the regional species pool andthe need to include species that are missing from the landscape in therestoration plans.

Restoration efforts may substantially contribute to improve thebiodiversity at different scales and represents a great opportunity toreintroduce locally extinct species, and threatened populations withparticular functional traits in agricultural regions, where forest coverhas been severely reduced. However, the effective restoration in such

landscapes in tropical regions is a challenge and will require a balancebetween conservation and production to encourage landowners. Thisencouragement could be possible by strengthening rural technical as-sistance agencies to assist landowners to comply with the local en-vironmental laws, and by mechanisms to develop the productive chainof seeds and seedlings for the recovery of native vegetation (Brancalionet al., 2016).

5. Conclusions

We found higher beta diversity when restoration plantings wereincluded in the landscape analysis if compared with diversity of forestfragments alone. This outcome highlights the positive effect of re-storation initiatives on landscape woody plant diversity and functionsin heavily agricultural regions with low native vegetation cover. Atlocal scale, however, is not expected that restoration plantings willrecover completely in terms of local diversity, but they can complementthe diversity of forest fragments if (1) there is sufficient time for eco-logical succession, and (2) species selection for restoration plantings iswell-planned in the beginning of the project. Based on our findings,there is an urgent need to maintain the complex network of biodiversityrefuges, including the remaining forests, the regenerating forests, andthe restoration plantings to conserve plant communities in human-modified landscapes. As the compliance with Native Vegetation

Fig. 5. Rarefaction curves of the Jaccard-type species dissimilarity measure for a standardized number of sites based on multiple incidence frequencies data (thecount of occurrences of each species among sites) for established (A) and regenerating plants (B). The X-axis denotes the number of sites.

D.C. Rother, et al. Forest Ecology and Management 451 (2019) 117538

8

Protection Law is the major driver of large-scale ecosystem restorationand conservation on privately owned lands in Brazil (Garcia et al.,2013), our study emphasizes the important role of agricultural land-owners in maintaining and protecting natural vegetation, due to theirirreplaceable value to conservation, in their lands and restoring areasillegally cleared in order to obtain concrete benefits for conservation indegraded landscapes. The characterization of diversity promoted byrestoration plans implemented at different scales has important im-plications for the conservation in deeply degraded landscapes. The re-covery of floristic and functional diversity is an essential indicator ofrestoration success and can be of great utility for future managementplans aiming to conserve species populations, community, ecologicalprocesses and ecosystem functioning.

Acknowledgements

We are grateful to Ismael Ferreira Rodrigues and Marcos Trigo fordata collection permission and logistical support. We also thankEvandro H. de Oliveira, Nathan R. Guedes, Reniê Michel Dutra and alllaboratory colleagues for field assistance over the years. The authorsreceived research grants and fellowships from the São Paulo ResearchFoundation (FAPESP) (#2012/24118-8) to DCR, from the NationalCouncil for Scientific and Technological Development to APL and fromthe Coordination for the Improvement of Higher Education Personnel of

Brazil (CAPES) (#88881.064976/2014-01) to RLC and(#88882.305844/2018-01) to DCR. This project was part of a multi-disciplinary study on tropical restoration coordinated by RRR (FAPESP#2013/50718-5).

Appendix A. Supplementary material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.foreco.2019.117538.

References

Alvares, C.A., Stape, J.L., Sentelhas, P.C., Gonçalves, J.L.M., Sparovek, G., 2013. Koppen’sclimate classification map for Brazil. Meteorologische Zeitschrift 22, 711–728.https://doi.org/10.1127/0941-2948/2013/0507.

APG III, 2009. An update of the Angiosperm Phylogeny Group classification for the ordersand families of flowering plants: APG III. Bot. J. Linn. Soc. 161, 105–121.

Arroyo-Rodríguez, V., Melo, F.P.L., Martínez-Ramos, M., Bongers, F., Chazdon, R.L.,Meave, J.A., Norden, N., Santos, B.A., Leal, I.R., Tabarelli, M., 2017. Multiple suc-cessional pathways in human-modified tropical landscapes: new insights from forestsuccession, forest fragmentation and landscape ecology research. Biol. Rev. 92,326–340. https://doi.org/10.1111/brv.12231/full.

Arroyo-Rodríguez, V., Rös, M., Escobar, F., Melo, F.P.L., Santos, B.A., Tabarelli, M.,Chazdon, R., 2013. Plant β-diversity in fragmented rain forests: testing floristichomogenization and differentiation hypotheses. J. Ecol. 101, 1449–1458. https://doi.org/10.1111/1365-2745.12153.

Audino, L.D., Murphy, S., Zambaldi, L., Louzada, J., Comita, L., 2017. Drivers of com-munity assembly in tropical forest restoration sites: role of local environment,

Fig. 6. Rarefaction curves of the Jaccard-type functional dissimilarity measure for a standardized number of sites based on multiple incidence frequency data (thecount of occurrences of each species among sites) for established (A) and regenerating plants (B). The X-axis denotes the number of sites.

D.C. Rother, et al. Forest Ecology and Management 451 (2019) 117538

9

landscape and space. Ecol. Appl. 27, 1731–1745. https://doi.org/10.1002/eap.1562.Barbosa, L.M., Shirasuna, R.T., Lima, F.C., Ortiz, P.R.T., 2015. Lista de espécies indicadas

para restauração ecológica para diversas regiões do estado de São Paulo. In I Anais doVI Simpósio de Restauração Ecológica, São Paulo, SP.

Benayas, J.M.R., Newton, A.C., Diaz, A., Bullock, J.M., 2009. Enhancement of biodi-versity and ecosystem services by ecological restoration: a meta-analysis. Science325, 1121–1124. https://doi.org/10.1126/science.1172460.

Benchimol, M., Mariano-Neto, E., Faria, D., Rocha-Santos, L., Pessoa, M.S., Gomes, F.S.,Talora, D.C., Cazetta, E., 2017. Translating plant community responses to habitat lossinto conservation practices: forest cover matters. Biol. Cons. 209, 499–507. https://doi.org/10.1016/j.biocon.2017.03.024.

Bongers, F., Poorter, L., Hawthorne, W.D., Sheil, D., 2009. The intermediate disturbancehypothesis applies to tropical forests, but disturbance contributes little to tree di-versity. Ecol. Lett. 12, 798–805. https://doi.org/10.1111/j.1461-0248.2009.01329.x.

Brancalion, P.H.S., Rodrigues, R.R., Gandolfi, S., Kageyama, P.Y., Nave, A.G., Gandara,F.B., Barbosa, L.M., Tabarelli, M., 2010. Instrumentos legais podem contribuir para arestauração de florestas tropicais biodiversas. Rev. Árv. 34, 455–470.

Brancalion, P.H.S., Melo, F.P.L., Tabarelli, M., Rodrigues, R.R., 2013. Biodiversity per-sistence in highly human-modified tropical landscapes depends on ecological re-storation. T. Cons. Sci. 6, 705–710.

Brancalion, P.H.S., Garcia, L.C., Loyola, R., Rodrigues, R.R., Pillar, V., Lewinsohn, T.M.,2016. Análise crítica da Lei de Proteção da Vegetação Nativa (2012), que substituiu oantigo Código Florestal: atualizações e ações em curso. Natureza & Conservação 14,e1–e16. https://doi.org/10.1016/j.ncon.2016.03.004.

Brancalion, P.H.S., Bello, C., Chazdon, R.L., Galetti, M., Jordano, P., Lima, R.A.F.,Medina, A., Pizo, M.A., Reid, J.L., 2018. Maximizing biodiversity conservation andcarbon stocking in restored tropical forests. Cons. Lett. 11, e12454 org.ez67.per-iodicos.capes.gov.br/10.1111/conl.12454.

Carvalho, G.H., Cianciaruso, M.V., Batalha, M.A., 2010. Plantminer: a web tool forchecking and gathering plant species taxonomic information. Environ. Modell. Softw.25, 815–816. https://doi.org/10.1016/j.envsoft.2009.11.014.

Chao, A., Chiu, C.H., 2016. Nonparametric Estimation and Comparison of SpeciesRichness. In: eLS. John Wiley & Sons Ltd:, Chichester, pp. 1–11.

Chao, A., Chiu, C.H., Jost, L., 2014a. Unifying species diversity, phylogenetic diversity,functional diversity, and related similarity and differentiation measures through Hillnumbers. Annu. Rev. Ecol. Evo. Syst. 45, 297–324. https://doi.org/10.1146/annurev-ecolsys-120213-091540.

Chao, A., Gotelli, N.G., Hsieh, T.C., Sander, E.L., Ma, K.H., Colwell, R.K., Ellison, A.M.,2014b. Rarefaction and extrapolation with Hill numbers: a framework for samplingand estimation in species biodiversity studies. Ecol. Monogr. 84, 45–67. https://doi.org/10.1890/13-0133.1.

Chave, J., Coomes, D., Jansen, S., Lewis, S.L., Swenson, N.G., Zanne, A.E., 2009. Towardsa worldwide wood economics spectrum. Ecol. Lett. 12, 351–366. https://doi.org/10.1111/j.1461-0248.2009.01285.x.

Chaves, R.B., Durigan, G., Brancalion, P.H., Aronson, J., 2015. On the need of legal fra-meworks for assessing restoration projects success: new perspectives from São Paulostate (Brazil). Restor. Ecol. 23, 754–759 org.ez67.periodicos.capes.gov.br/10.1111/rec.12267.

Chiu, C.H., Chao, A., 2014. Distance-based functional diversity measures and their de-composition: A framework based on hill numbers. PLoS One 9, e100014. https://doi.org/10.1371/journal.pone.0100014.

Classen, A.T., Norby, R.J., Campany, C.E., Sides, K.E., Weltzin, J.F., 2010. Climate changealters seedling emergence and establishment in an old-field ecosystem. PLoS One 5,e13476. https://doi.org/10.1371/journal.pone.0013476.

Colwell, R.K., Chao, A., Gotelli, N.J., Lin, S.Y., Mao, C.X., Chazdon, R.L., Longino, J.T.,2012. Models and estimators linking individual-based and sample-based rarefaction,extrapolation and comparison of assemblages. J. Plant Ecol. 5, 3–21.

Costa, A.S., Malhado, A.C.M., Bragagnolo, C., Correia, R.A., Ladle, R.J., 2016. Ecologicaloutcomes of Atlantic Forest restoration initiatives by sugar cane producers. Landpolicy 52, 345–352. https://doi.org/10.1016/j.landusepol.2015.12.025.

Crouzeilles, R., Ferreira, M.S., Chazdon, R.L., Lindenmayer, D.B., Sansevero, J.B.B.,Monteiro, L., Iribarrem, A., Latawiec, A.E., Strassburg, B.B.N., 2017. Ecological re-storation success is higher for natural regeneration than for active restoration intropical forests. Sci. Adv. 3, e1701345. https://doi.org/10.1126/sciadv.1701345.

da Silva, A.P.M., Schweizer, D., Marques, H.R., Teixeira, A.M.C., Santos, T.V.M.N.,Sambuichi, R.H.R., Badari, C.G., Gaudare, U., Brancalion, P.H.S., 2017. Can currentnative tree seedling production and infrastructure meet an increasing forest re-storation demand in Brazil? Restor. Ecol. 25, 509–515. https://doi.org/10.1111/rec.12470.

Farah, F.T., Muylaert, R.L., Ribeiro, M.C., Ribeiro, J.W., Mangueira, J.R.S.A., Souza, V.C.,Rodrigues, R.R., 2017. Integrating plant richness in forest patches can rescue overallbiodiversity in human-modified landscapes. Forest Ecol. Manag. 397, 78–88. https://doi.org/10.1016/j.foreco.2017.03.038.

Flores, O., Coomes, D.A., 2011. Estimating the wood density of species for carbon stockassessments. Methods Ecol. Evol. 2, 214–220.

Galetti, M., Guevara, R., Cortes, M.C., Fadini, R., Matter, S.V., Leite, A.B., Labecca, F.,Ribeiro, T., Carvalho, C.S., Collevatti, R.G., Pires, M.M., Guimarães Jr., P.R.,Brancalion, P.H.S., Ribeiro, M.C., Jordano, P., 2013. Functional extinction of birdsdrives rapid evolutionary changes in seed size. Science 340, 1086–1090. https://doi.org/10.1126/science.1233774.

Garcia, L.C., Santos, J.S., Matsumoto, M., Silva, T.S.F., Padovezi, A., Sparovek, G., Hobbs,R.J., 2013. Restoration challenges and opportunities for increasing landscape con-nectivity under the New Brazilian Forest Act. Nat. Conservação 11, 181–185.

Garcia, L.C., Cianciaruso, M.V., Ribeiro, D.B., Santos, F.A.M., Rodrigues, R.R., 2015.Flower functional trait responses to restoration time. Appl. Veg. Sci. 18, 402–412.https://doi.org/10.1111/avsc.12163.

Harrison, X.A., 2015. A comparison of observation-level random effect and Beta-Binomialmodels for modelling overdispersion in Binomial data in ecology & evolution. PeerJ3, e1114. https://doi.org/10.7717/peerj.1114.

Hawes, J.E., Peres, C.A., Riley, L.B., Hess, L.L., 2012. Landscape-scale variation instructure and biomass of Amazonian seasonally flooded and unflooded forests. ForestEcol. Manage. 281, 163–176. https://doi.org/10.1016/j.foreco.2012.06.023.

Hsieh, T.C., Ma, K.H., Chao, A., 2016. iNEXT: An R package for rarefaction and extra-polation of species diversity (Hill numbers). Methods Ecol. Evol. 7 (12), 1451–1456.https://doi.org/10.1111/2041-210X.12613.

Klink, C.A., Machado, R.B., 2005. Conservation of the Brazilian Cerrado. Conserv. Biol.19, 707–713. https://doi.org/10.1111/j.1523-1739.2005.00702.x.

Kronka, F.J.N., Nalon, M.A., Matsukuma, C.K., Kanashiro, M.M., Ywane, M.S.S., Pavão,M., Durigan, G., Lima, L.M.P.R., Guillaumon, J.R., Baitello, J.B., Borgo, S.C., Manetti,L.A., Barradas, A.M.F., Fukuda, J.C., Shida, C.N., Monteiro, C.H.B., Pontinha, A.A.S.,Andrade, G.G., Barbosa, O., Soares, A.P., 2005. Inventário florestal da vegetacãonatural do estado de São Paulo. São Paulo, Secretaria do Meio Ambiente, InstitutoFlorestal, Imprensa Oficial, pp. 200p.

Legendre, P., De Cáceres, M., 2013. Beta diversity as the variance of community data:dissimilarity coefficients and partitioning. Ecol. Lett. 16 (8), 951–963 org.ez67.per-iodicos.capes.gov.br/10.1111/ele.12141.

Lopes, A.V., Girão, L.C., Santos, B.A., Peres, C.A., Tabarelli, M., 2009. Long-term erosionof tree reproductive trait diversity in edge-dominated Atlantic forest fragments. Biol.Conserv. 142, 1154–1165. https://doi.org/10.1016/j.biocon.2009.01.007.

Magnago, L.F.S., Edwards, D.P., Edwards, F.A., Magrach, A., Martins, S.V., Laurance,W.F., 2014. Functional attributes change but functional richness is unchanged afterfragmentation of Brazilian Atlantic forests. J. Ecol. 102, 475–485. https://doi.org/10.1111/1365-2745.12206.

Magurran, A.E., 2004. Measuring Biological Diversity. Blackwell Science, Oxford, pp.256p.

Meli, P., Holl, K.D., Benayas, J.M.R., Jones, H.P., Jones, P.C., Montoya, D., Mateos, D.M.,2017. A global review of past land use, climate, and active vs. passive restorationeffects on forest recovery. PLoS One 12, e0171368. https://doi.org/10.1371/journal.pone.0171368.

Melo, F.P.L., Arroyo-Rodríguez, V., Fahrig, L., Martínez-Ramos, M., Tabarelli, M., 2013.On the hope for biodiversity-friendly tropical landscapes. Trends Ecol. Evol. 28,461–468. https://doi.org/10.1016/j.tree.2013.01.001.

Metzger, J.P., Martensen, A.C., Dixo, M., Bernacci, L.C., Ribeiro, M.C., Teixeira, A.M.G.,Pardini, R., 2009. Time-lag in biological responses to landscape changes in a highlydynamic Atlantic Forest region. Biol. Cons. 142, 1166–1177. https://doi.org/10.1016/j.biocon.2009.01.033.

Montoya-Pfeiffer, P.M., Rodrigues, R.R., Metzger, J.P., Silva, C.I., Baquero, O.S., Santos,I.A., 2018. Are the assemblages of tree pollination modes being recovered by tropicalforest restoration? J. Veg. Sci. 21, 156–163 org.ez67.periodicos.capes.gov.br/10.1111/avsc.12335.

Nimmo, D.G., Mac Nally, R., Cunningham, S.C., Haslem, A., Bennett, A.F., 2015. Vive larésistance: reviving resistance for 21st century conservation. Trends Ecol. Evol. 30,516–523. https://doi.org/10.1016/j.tree.2015.07.008.

Oliveira, M.A., Santos, A.M.M., Tabarelli, M., 2008. Profound impoverishment of thelarge-tree stand in a hyper-fragmented landscape of the Atlantic forest. Forest Ecol.Manage. 256, 1910–1917. https://doi.org/10.1016/j.foreco.2008.07.014.

Oliveira-Filho, A.T., 2014. NeoTropTree, Flora arbórea da Região Neotropical: Um bancode dados envolvendo biogeografia, diversidade e conservação. Universidade Federalde Minas Gerais.

Prado, P.I., Miranda, M. D., 2014. sads: Maximum Likelihood Models for SpeciesAbundance Distributions. R package version 0.1.10.

R Development Core Team, 2017. R: A Language and Environment for StatisticalComputing. R Foundation for Statistical Computing, Vienna, Austria http://www.R-project.org/.

Ribeiro, M.C., Metzger, J.P., Martensen, A.C., Ponzoni, F.J., Hirota, M.M., 2009. TheBrazilian Atlantic Forest: How much is left, and how is the remaining forest dis-tributed? Implications for conservation. Biol. Conserv. 142, 1141–1153. https://doi.org/10.1016/j.biocon.2009.02.021.

Rigueira, D.M.G., Rocha, P.L.B., Mariano-Neto, E., 2013. Forest cover, extinctionthresholds and time lags in woody plants (Myrtaceae) in the Brazilian Atlantic Forest:resources for conservation. Biodivers. Conserv. 22, 3141–3163. https://doi.org/10.1007/s10531-013-0575-4.

Rodrigues, R.R., Brancalion, P.H.S., Isernhagen, I., 2009a. Pacto Pela Restauração daMata Atlântica: Referencial dos Conceitos e Ações de Restauração Florestal. SãoPaulo, Brazil, Instituto BioAtlântica.

Rodrigues, R.R., Gandolfi, S., Nave, A.G., Aronson, J., Barreto, T.E., Vidal, C.Y.,Brancalion, P.H.S., 2011. Large-scale ecological restoration of high-diversity tropicalforests in SE Brazil. Forest Ecol. Manage. 261, 1605–1613. https://doi.org/10.1016/j.foreco.2010.07.005.

Rodrigues, R.R., Lima, R.A.F., Gandolfi, S., Nave, A.G., 2009b. On the restoration of highdiversity forests: 30 years of experience in the Brazilian Atlantic Forest. Biol. Conserv.142, 1242–1251. https://doi.org/10.1016/j.biocon.2008.12.008.

Rother, D.C., Vidal, C.Y., Fagundes, I.C., Silva, M.M., Gandolfi, S., Rodrigues, R.R., Viani,R.A.G., Brancalion, P.H.S., 2018. How legal-oriented restoration programs enhancelandscape connectivity? Insights from the Brazilian Atlantic Forest. Trop. Conserv.Sci. 11, 1–9. https://doi.org/10.1177/1940082918785076.

Santos, G.G.A., Santos, B.A., Nascimento, H.E.M., Tabarelli, M., 2012. Contrasting de-mographic structure of short- and long-lived pioneer tree species on Amazonian forestedges. Biotropica 44, 771–778. https://doi.org/10.1111/j.1744-7429.2012.00882.x.

Shuey, J., Labus, P., Carneiro, E., Dias, F.M.S., Leite, L.A.R., Mielke, O.H.H., 2017.Butterfly communities respond to structural changes in forest restorations and re-generation in lowland Atlantic Forest, Paraná. Brazil. J. Insect Conserv. 21, 545–557.

D.C. Rother, et al. Forest Ecology and Management 451 (2019) 117538

10

https://doi.org/10.1007/s10841-017-9994-y.SigRH, 2016. Sistema Integrado de Gerenciamento dos Recursos Hídricos do. Estado de

São Paulo.Silva, F.R., Montoya, D., Furtado, R., Memmott, J., Pizo, M.A., Rodrigues, R.R., 2015. The

restoration of tropical seed dispersal networks. Rest. Ecol. 23, 852–860. https://doi.org/10.1111/rec.12244.

Soares-Filho, B., Rajão, R., Macedo, M., Carneiro, A., Costa, W., Coe, M., Rodrigues, H.,Alencar, A., 2014. Cracking Brazil’s forest code. Science 344, 363–364. https://doi.org/10.1126/science.1246663.

Solar, R.R.C., Barlow, J., Ferreira, J., De Berenguer Cesar, E., Lees, A.C., Louzada, J.N.,Maués, M., Moura, N.G., Oliveira, V.H.F., Chaul, J.C.M., Schoereder, J.H., Vieira,I.C.G., Nally, R.M., Gardner, T., 2015. How pervasive is biotic homogenization inhuman-modified tropical forest landscapes? Ecol. Lett. 18, 1108–1118. https://doi.org/10.1111/ele.12494.

Sparovek, G., Berndes, G., Barretto, A.G.O.P., Klug, I.L.F., 2012. The revision of theBrazilian Forest Act: increased deforestation or a historic step towards balancingagricultural development and nature conservation? Environ. Sci. Policy 16, 65–72.https://doi.org/10.1016/j.envsci.2011.10.008.

Stuble, K.L., Fick, S.E., Young, T.P., 2017. Every restoration is unique: testing year effectsand site effects as drivers of initial restoration trajectories. J. Appl. Ecol. 54,

1051–1057. https://doi.org/10.1111/1365-2664.12861.Suganuma, M.S., Torezan, J.M.D., Durigan, G., 2017. Environment and landscape rather

than planting design are the drivers of success in long term restoration of riparianAtlantic forest. Appl. Veg. Sci. 21, 76–84. https://doi.org/10.1111/avsc.12341.

Tabarelli, M., Peres, C.A., 2002. Abiotic and vertebrate seed dispersal in the BrazilianAtlantic forest: implications for forest regeneration. Biol. Cons. 106, 165–176.https://doi.org/10.1016/S0006-3207(01)00243-9.

Tabarelli, M., Mantovani, W., Peres, C.A., 1999. Effects of habitat fragmentation on plantguild structure in the montane Atlantic forest of southeastern Brazil. Biol. Conserv.91, 119–127. https://doi.org/10.1016/S0006-3207(99)00085-3.

Tabarelli, M., Aguiar, A.V., Girão, I.C., Peres, C.A., Lopes, A.V., 2010. Effects of pioneertree species hyperabundance on forest fragments in northeastern Brazil. Conserv.Biol. 24 (6), 1654–1663. https://doi.org/10.1111/j.1523-1739.2010.01529.x.

Van der Pijl, L., 1982. Principles of Dispersal in Higher Plants, third ed. Springer-Verlag,Berlin.

Zucchi, M.I., Sujii, P.S., Mori, G.M., Viana, J.P.G., Grando, C., Silvestre, E.A., Schwarcz,K.D., Macrini, C.M., Bjay, M.M., Araújo, F.L., Siqueira, M.V.B.M., Pereira, A.A., deSouza, A.P., Pinheiro, J.B., Rodrigues, R.R., Brancalion, P.H.S., 2018. Genetic di-versity of reintroduced tree populations in restoration plantations of the BrazilianAtlantic Forest. Rest. Ecol. 26, 694–770. https://doi.org/10.1111/rec.12620.

D.C. Rother, et al. Forest Ecology and Management 451 (2019) 117538

11