Open Collections

Description

<b>Abstract</b><br/>

Biodiversity serves as a proxy for numerous ecosystem services that can be realized through forest restoration, benefitting both people and the environment. We investigated the magnitude of biodiversity recovery incompleteness (i.e., the recovery gap) in forest restoration within the Brazilian Atlantic Forest, hereafter referred to as the Atlantic Forest. We conducted meta-analysis to analyze how species richness and species abundance of soil microorganisms, invertebrates and vascular plants, as well as the vegetation structure, recover across major gradients in environmental conditions and human-caused disturbances. Our study shows that forest restoration in the Atlantic Forest faces a notable biodiversity gap in species richness across both passive and active restoration areas. However, the vegetation structure could potentially reach reference levels within 25 to 50 years. Forest type influenced the recovery of species abundance in active restoration areas, with dense forests displaying the largest gaps. Likewise, taxonomic group influenced species richness gaps in passive restoration areas, with invertebrates showing the largest gap. Reference forest age and past land use did not significantly affect biodiversity outcomes in either restoration approach. However, biodiversity levels were lower than those of the reference forest at various levels of the moderating factors analyzed. 4. Synthesis and applications: The study shows that after 25–50 years, restoration sites develop a vegetation structure similar to that of reference forests, regardless of the restoration approach. Species richness also tends to recover over time, but the rate and pattern of recovery differ between approaches. Passive restoration follows a gradual, long-term decline in the recovery gap, while active restoration exhibits a less clear trajectory. Past land use is the strongest predictor of biodiversity recovery, particularly for vegetation structure. The restoration age, forest type, and taxonomic group play more moderate roles but explain significant variation within particular categories of each variable. These findings highlight the importance of targeted interventions to enhance restoration outcomes and the need to prioritize efforts based on specific restoration objectives. Our results emphasize the importance of setting realistic, taxon-specific goals and provide metrics to guide resource allocation based on recovery gaps and timelines.</p>; <b>Methods</b><br />

<strong>Literature search and data gathering</strong></p>

We searched the peer-reviewed literature available on Web of Science (SCI-E, SSCI, and ESCI), Scopus, CAB Direct, and SciELO. We also used Google Scholar as a search engine. Data was first collected on August 26th, 2020 (Supplementary Material S1). We employed the following search string to gather relevant literature: (restor* or recreat* or rehabilitat* or reforest* or afforest* or recover* or regenerat* or remediat* or revege*) AND (forest*) AND (Brazil* or Brasil*) AND (biodiversity or diversity) (see Supplementary Material S1). The searching and screening procedures closely resemble those employed by Romanelli et al. (2022).</p>

Only primary research meeting the following inclusion criteria was considered for full-text analysis. Eligible studies had to be conducted in the Brazilian Atlantic Forest, with interventions classified as either active (e.g., tree planting or seeding) or passive (minimal human intervention in forest regeneration) restoration, excluding highly degraded sites like mining areas. Comparators involved quantitative biodiversity outcomes, comparing restoration sites with a reference forest within the same assessment. Acknowledging variations in species assemblages and spatial arrangements, data from multiple reference forests were included (Suganuma &amp; Durigan, 2015). Outcomes focused on statistical information related to species richness, species abundance, and vegetation structure (e.g. diameter, height, and basal area). Inclusion criteria also required direct data provision or indirect estimation, along with reporting restoration characteristics and environmental conditions (see Supplementary Material S1).</p>

We used the package metaDigitise in R (Pick, Nakagawa, &amp; Noble, 2019) to extract data from figures, and used the Revtools package in R to remove article duplicates (Westgate, 2019).</p>