Very low-intensity logging can be a compromise between strict conservation and income-generating land use in tropical forests. Investigating how selective logging influences the understory environment and seedling dynamics as the forest regenerates offers insights into whether logging negatively impacts forest dynamics, influencing the composition and structure of future forests. We explored how very low-intensity logging (<2 trees ha-1) influences understory factors and seedling dynamics across a logging chronosequence (unlogged forest vs. forest actively logged and forest logged four and fourteen years prior). To do this, we assessed (i) how light levels, prevalence of vegetation damage, and elephant trails differ in logged forests at different recovery stages compared to unlogged forest; (ii) how these understory factors influence seedling dynamics; (iii) how seedling dynamics differ across the logging chronosequence; and (iv) how logging impacts liana vs. tree seedlings across the chronosequence. We observed higher light availability and vegetation damage in logged forests up to four years after logging and higher elephant trail prevalence at fourteen years after logging compared to unlogged forests. Seedling survival was lower in plots with vegetation damage, elephant trails, or lower light, while seedling growth rates were higher in vegetation-damaged areas. Selectively logged forests initially had lower seedling survival and recruitment, but higher growth rates compared to unlogged forests. However, 14 years after logging, seedling dynamics were similar to patterns in unlogged forests. Lianas had a slight seedling growth advantage over trees in all logged forests compared to unlogged forests. Results from our study suggest that logging causes temporary shifts in understory dynamics rather than long-term shifts in forest recovery trajectories. These managed areas have potential as land that can contribute to OECM targets – functioning as mixed-use corridors, connecting protected areas across a landscape and contributing to biodiversity and wildlife conservation – especially in high forest cover, low deforestation countries.

Data collection methods:

Study area and seedling censuses

We sampled vegetation communities across eighty 20×20 m plots in the Société Equatoriale d'Exploitation Forestière (SEEF), a logging concession to the east of the Mbé sector of the Monts de Cristal National Park (0°42′41′′ N, 10°17′18′′ E) in northwestern Gabon (Sullivan et al. 2022). We sampled 20 plots within each of four different treatment areas, or logging “blocks” that varied in the year that they were logged (Fig. 1A). The blocks in our study were logged in 2008, 2018, 2020, and one additional block was sampled as an unlogged control. SEEF started systematic logging in this logging concession in 2000 and has a very low timber removal rate (<2 trees per ha, Medjibe et al. 2011, SEEF 2019), which is typical for the Congo basin region (Sist, 2000). 
Each 20×20 m vegetation plot was established between September 2018 and October 2019 (Sullivan et al. 2022) and contained nine 1×1 m seedling plots (Fig. 1B), spaced 10 m apart, for a total of 180 seedling plots in each of the three logging treatments and the unlogged block. In each seedling plot, we identified, tagged, and measured the height of all woody seedlings ≤1m tall. We re-censused the seedling plots between February and September 2022, recording the survival status (dead or alive) and height of each previously tagged seedling, and tagging and measuring new seedlings that had recruited into the seedling plots since the previous census. Any observations that were missing survival or height data were removed before analyses. Tree seedlings were identified to species or morphospecies level, and liana seedlings were identified to the taxonomic level that was possible in the field – any liana that lacked family-level taxonomic information was labeled “liana” in the data (see Sullivan et al. 2022 for more details).

Environmental factors and elephant trails

To quantify understory light availability, we measured canopy openness using a Spherical Crown Densiometer (Concave Model C, Forestry Suppliers, Jackson, MS, USA) (Lemmon, 1956) at a height of 1m in the center of each 1×seedling plot. In each 1×1m seedling plot, we also recorded the presence/absence of vegetation damage from falling woody debris or skid trails and the presence/absence of elephant trails. All environmental factors and elephant trails were assessed during the 2022 census period.