GoKawiilData
Data were sampled in the context of the Research Unit ‘Reassembly’, studying a chronosequence of a rainforest ecosystem in the Chocó tropical rainforest (see ref. 47 for more details).
Study area
The study area is located in the reserves Canandé and Tesoro Escondido in the province of Esmeraldas in northwestern Ecuador (0.5° N, 79.2° W). Both reserves consist of a mosaic of actively used cacao and pasture plantations, secondary forest of varying age and primary forest. In this area, logging and agricultural land use started approximately five decades ago. Forests were logged for timber industry and subsequently mainly transformed into cacao plantations or pasture. Annual rainfall at the site is 4,700–6,000 mm. Mean temperature range 21–28 °C and relative humidity 90–100% (ref. 54).
Chronosequence
The study design represents a chronosequence approach that substitutes space for time, that is several forest plots at various successional stages were studied in a short time span instead of monitoring the same patch over a long time span. This approach assumes that the plots follow similar successional trajectories and had comparable starting points. Information about previous land use and environmental conditions for each site are important and need to be representative and unbiased across the successional stages. This has been tested in detail for trees (see ref. 47 for analyses of spatial autocorrelation and environmental bias). We explicitly distinguished actively used cacao and pasture plots as well as secondary forest plots regenerating from previous use as cacao or pasture. Several plots for each category were distributed over the landscape (200 km2) to acknowledge variation in forest attributes (mean distances between plots in each category were around 5 km and did not differ across successional stages47). Our study comprised a total of 62 plots (each 0.25 ha) of which 12 plots represent active agriculture (6 cacao plantations and 6 pastures) and 33 plots secondary forest varying in age between 1 year and 38 years since start of recovery (17 recovering from use as cacao plantation and 16 from use as pasture, both with a similar time span). A total of 17 plots serve as old-growth forest reference and have neither been cut nor used. Supplementary Table 6 provides information on the ages of each plot. The topography is hilly and plots range in elevation from 159 m to 615 m. There was no significant difference in mean elevation between active, recovering or old-growth forest plots and no significant correlation between elevation and plot age47. Regeneration plots were located 58 m ± 7 m from the nearest forest edge and old-growth plots 388 m ± 72 m to the nearest forest edge47.
Pastures were grazed extensively by low densities of cattle and occasionally by horses and were dominated by aggressive pasture grasses such as Brachiaria or Axonopus scoparius. Cacao plantations were monocultures of sun-exposed Theobroma cacao trees which were spaced 2–4 m apart and grew to heights of 5–10 m. These plantations were regularly treated with herbicides. Whereas cacao plantations generally lacked shade trees, pastures had some remnant trees or palms, or shrubs along creeks. These characteristics were also typical of the pastures and cacao plantations that were used several decades ago and which now represent regenerating forests in our study area. Old-growth forests contained large, slow-growing trees with potential for timber use and showed no signs of harvesting. Before they regenerated as secondary forests, pastures and cacao plantations had a similar duration of land use (mean 11.4 years, range 1–30 years). Regenerating pastures were larger (11.3 ha, range 1.2–46.7 ha) than cacao plantations (2.0 ha, 0.3–5.7 ha). All plots were located in a relatively intact landscape with a mean forest cover of 74% ± 2.8% (24–100%) within a 1-km radius. For further details on the study site, see ref. 47. Sampling methods for each taxon along the chronosequence are summarized in Supplementary Methods.
Calculation of biodiversity metrics
For all taxa and plots we calculated the total abundance (number of individuals of each taxon) per plot and alpha-diversity Hill numbers (that is, effective number of species) for the orders q = 0, 1 and 2 (species diversity, Shannon diversity and Simpson diversity). We defined the similarity of species composition to old-growth forests as the mean of all pairwise comparisons between an agricultural or secondary forest plot to each of the 17 old-growth forest plots. Note that old-growth forests were distributed across the entire study area and represent natural spatial variation in biodiversity and composition. To define the old-growth forest reference value for species composition similarity, we thus compared each old-growth forest plot against all other 16 old-growth forest plots and calculated the mean similarity per plot. The median value of these 17 mean old-growth forest similarities was then used as the asymptotic reference for the recovery (\({\varPsi }_{{\rm{Ref}}}\) in equations (1) and (5)). Therefore, full recovery of species composition refers to the point at which the compositional similarity of an agricultural or secondary forest plot to an old-growth forest cannot be differentiated from the similarity among old-growth forests. We calculated the pairwise Bray–Curtis similarity as well as the beta-diversity Hill numbers for orders q = 0, 1 and 2. Beta-diversity Hill numbers were calculated as the fraction of gamma- and alpha-diversity Hill numbers of orders q = 0, 1 and 2 according to ref. 55:
$$\genfrac{}{}{0ex}{}{q}{}{D}_{\beta }=\genfrac{}{}{0ex}{}{q}{}{D}_{\gamma }/\genfrac{}{}{0ex}{}{q}{}{D}_{\alpha }$$ (1)
... continue reading