Skip to main content
Dryad logo

Pinus kesiya invasions in Tapia woodland Madagascar

Citation

Rajaonarivelo, Herimino Manoa et al. (2022), Pinus kesiya invasions in Tapia woodland Madagascar, Dryad, Dataset, https://doi.org/10.5061/dryad.xsj3tx9g7

Abstract

Pinus species are among the highly invasive species which have spread outside their plantation area after their introduction in the Southern Hemisphere. The case of Pinus kesiya invasion is observed in the high plateau of Madagascar, inside the sclerophyll Tapia woodland which is dominated by the endemic Uapaca bojeri tree species. The analysis of this invasion was carried out using 375 plots of 100 m2 each in Tapia woodland. Data on the vegetation structure, the plot characteristics and the propagule pressure were collected. We recorded a total of 740 pines distributed in 29.8% of the plots. The generalized linear model revealed that the diminution in frequency of the dominant species Uapaca bojeri with the increasing degree of disturbance of the fragment led to the vulnerability of the Tapia woodland to the abundance of pine. The factors explaining pine occurrence varied according to the pine life-stage. In the seedling stage, the distance of the plot from the propagule source combined with the longitudinal position of the plot explained 18% of the pine invasion success. In the sapling and adult stages, the vegetation structure was the main important factor (22% and 11% of variation explained). The frequency of U. bojeri and the degree of disturbance were the most important factors characterizing this vegetation structure. Based on these results, a strategy to control pine invasion in the Tapia woodland may focus on enrichment with U. bojeri and limitation of the plantation of P. kesiya in proximity.

Methods

Data were collected throught inventory. In total, we have surveyed 375 plots of 100m2 each where woody species have been inventoried (counted, specified). Pinus kesiya was defined according to three different life-stages: seedlings (0 < DBH ≤ 1 cm), saplings (1 < DBH < 5 cm) and adults (DBH ≥ 5 cm)

Usage Notes

To quantify the local dominance of  U. bojeri, we calculated the density (‘D’ in trees. ha−1),  the frequency (‘f’ in % trees. ha-1), and the basal area (‘G’ in m2. ha-1) of the species considering all trees above 1 cm DBH. We characterized the local community diversity by its species richness ‘S’, the Shannon-Weaver diversity index ‘H’, and Pielou’s evenness index ‘R’ (R = H / Hʹmax where Hʹmax = Log(S)). Two classes of plot disturbance degree (‘disturb’) were defined following  Rakotondrasoa et al. (2013) by hierarchical clustering based on the following stand variables (DBH > 1 cm) calculating with all surveyed species at plot level:  tree mean diameter (d in cm), tree mean height (h in m), tree density (D tot), tree basal area (G tot), species richness (S). The cover of vegetation strata – litter, herbaceous, shrubs – was estimated using a categorical ordered scale (Calviño-Cancela and Van Etten 2018) from 1 to 3: 1: [0–40%]; 2: ]40–70%]; 3: ]70–100%]. Topographic variables were recorded using a GPS unit for elevation, a clinometer for slope and visual observation for topographic position. 

The topography was classified using two ordered factors following Randriambanona et al. (2019) (i) topographic position (Top-1: valley bottom, Top-2: middle slope, Top-3: upper slope), (ii) slope values (“low” < 8°, 8° ≤ “middle” < 20° and, “high” ≥ 20°). The position of the plot along the transect (pl_1: 10 m, pl_2: 30m, pl_3: 50 m) was used as an additional explanatory variable to account for edge effects. The geographical coordinates (‘latitude’ and ‘longitude’) were also included in order to account for a potential spatial structure in the response variables.

To estimate the propagule pressure effect, two variables were defined. We first extensively mapped the majority of Pinus seed sources in the area. Firstly, we have calculated the distance from each plot to all the propagule sources. We have selected the nearest distance which is supposed to be the potential source and named the distance between the plot and the source ‘source distance. Secondly, the number of seed sources or ‘Source density’ in the surroundings, i.e. 500 meters around each plot, was considered.

Funding

Centre de Coopération Internationale en Recherche Agronomique pour le Développement