Data and code from: Population structure and regeneration dynamics of Prunus africana in high-altitude habitats of Volcanoes National Park, Rwanda

Emile, Mugabo 1 ; Thadée, Muhire1; van der Hoek, Yntze 1 ; Deogratias, Tuyisingize 1

Published Sep 22, 2026 on Dryad. https://doi.org/10.5061/dryad.wdbrv165t

Data files

Sep 22, 2026 version files 47.22 KB

Abstract

Prunus africana is an evergreen African tree valued for its medicinal uses, especially in treatments for prostate conditions. Heavy harvesting of its bark has contributed to major population declines, and the species is now considered threatened by the IUCN. Despite its ecological importance as a source of food and habitat for wildlife, little is known about its status in Volcanoes National Park (VNP), Rwanda. This study assessed the distribution, regeneration, population structure, and structural characteristics of P. africana across the Afromontane forests of VNP, covering elevations from approximately 2,400 to 3,714 m. Surveys were conducted along 16 strip transects, where mature trees were measured for diameter, height, and canopy cover, while seedlings and saplings were recorded around mature trees. The study also examined how elevation, canopy cover, neighboring-tree distance, and tree size influenced regeneration and population structure. We recorded 46 mature trees, 27 saplings, and more than a thousand seedlings. The strong dominance of seedlings suggests active recruitment but limited progression into larger size classes. P. africana was most abundant between 3,000 and 3,100 m, while regeneration varied with elevation, canopy cover, neighbouring-tree distance, and focal-tree diameter. These findings provide important baseline information and highlight the need for targeted protection and habitat management to support the species’ long-term persistence in VNP.

Dataset DOI: 10.5061/dryad.wdbrv165t

Description of the data and file structure

This dataset contains the data and code required to replicate analyses. In an effort to assess the population structure and regeneration of P. africana, we set up sixteen strip transects with a length of 1000 m and 40 m wide (20 m east and 20 m west) across the VNP, and the transects were at least 1000 m apart. We identified mature P. africana tree individuals, i.e., those with >10 cm of diameter at breast height (DBH), as reference mature trees and then measured their DBH and height. To assess regeneration, we counted seedlings and saplings within a 5 m radius of each mature tree. Subsequently, we classified seedlings as young trees with a height ≤130 cm, saplings as small trees with a height >130 cm but a diameter <10 cm, and mature trees as all trees with a diameter >10 cm. Among the 46 mature tree individuals recorded, 10 overlapped within a radius of 5 m, and we equally distributed the number of seedlings/saplings.

We measured the height and DBH of the three closest tree species to the reference tree, as well as the distance interval to those neighboring trees. We estimated canopy cover using a spherical densiometer at four points located one meter from the trunk of each reference tree in the four cardinal directions. We measured tree heights using a laser range finder and recorded the geographic coordinates of all sampled P. africana individuals using a Garmin GPS device.

Data analysis

We classified the elevation gradient into four classes of 100 m difference (2900-3000, 3000-3100, 3100-3200, and 3200-3300) to assess the distribution and abundance of different P. africana individuals' size classes along the elevation. We used a Generalized Linear Model (GLM) to test the effect of elevation change on tree diameter, tree height, and regeneration rate (the number of new regenerating individuals, i.e., seedlings and saplings recorded within the 5 m radius of each recorded mature tree of P. africana at different elevation classes), and we used the total number of seedlings and saplings at each elevation gradient. We used a chi-square (X²) test to determine the difference between the observed and expected ratio of P. africana individuals within different tree size classes (seedlings, saplings, and mature trees). We created different regeneration categories (fair, good, not, new, and poor regeneration) to evaluate the regeneration dynamics of P. africana. To assess P. africana regeneration status, we evaluated it by the population sizes of seedlings, saplings, and mature individuals, following categories of previous studies (Ao et al., 2024; Balemlay & Siraj, 2021; Bogale et al., 2017).

To assess factors associated with the regeneration patterns, we fitted negative binomial GLMs (MASS::glm.nb, Brian et al., 2026; Kharrat & Boshnakov, 2019) with multiple predictors (individual, spatial, dominance, environmental, and interaction terms) to compare model performance. We ranked models by AIC and assessed overdispersion via residual deviance-to-degrees-of-freedom ratios. The model was fitted to the response variable (n = 138) as a function of DBH, height, canopy cover, mean distance to neighbors, and DBH dominance. We fitted the model and assessed it via AIC, deviance, and dispersion parameter (θ). To evaluate model robustness, we also tested the significance of each predictor individually using separate negative binomial GLMs for each variable. We calculated structural indices based on the structural group of four individuals (according to Füldner, 1995; Pommerening, 2002) that help to generate community structure around the focal tree.

Files and variables

File: raw_data_prunus_africana.csv

Description:

Variables

transect: Unique identifier for each tree
tree_id: Transect identifier
elevation: (numeric) Elevation of the sampling location (m)
elevation_class: (categorical) Elevation class assigned to the sampling location
dbh_cm: (numeric) Diameter at breast height of Prunus africana
height_m: (numeric) Height of Prunus africana
s_sn0: (integer) Number of seedlings and saplings within a 5 m radius of the focal mature Prunus africana
canopy_cover: (numeric) Canopy cover measured using a densiometer
proximity_sp: (categorical) Species identity of the three nearest neighboring trees
distance: (numeric) Distance between the focal Prunus africana and each of its three nearest neighboring trees
height_m2: (numeric) Height of each of the three nearest neighboring trees
dbh_cm2: (numeric) Diameter at breast height of each of the three nearest neighboring trees
seedlings: (integer) Number of seedlings within a 5 m radius of the focal mature Prunus africana
saplings: (integer) Number of saplings within a 5 m radius of the focal mature Prunus africana
mingling: (numeric) Species mingling index calculated for the focal Prunus africana and its three nearest neighboring trees
mean_distance:(numeric) Mean distance between the focal Prunus africana and its three nearest neighboring trees
dbh_diff: (numeric) Degree of diameter differentiation between the focal Prunus africana and its three nearest neighboring trees
dbh_dom: (numeric) Relative DBH dominance of the focal Prunus africana compared with its three nearest neighboring trees
height_dom: (numeric) Relative height dominance of the focal Prunus africana compared with its three nearest neighboring trees
height_dif: (numeric) Degree of height differentiation between the focal Prunus africana and its three nearest neighboring trees

File: R_script_prunus_africana.Rmd

Description: R Markdown script containing the data processing, statistical analyses, model comparisons, predictions, and visualizations used to investigate Prunus africana population structure, regeneration, and associated ecological and habitat characteristics along an elevation gradient in Volcanoes National Park, Rwanda.

Code/software

Code/software

The data and analysis can be accessed using free and open-source software. The dataset is provided in CSV format and can be viewed using commonly available spreadsheet or text-editing software. All data processing, statistical analyses, and visualizations were conducted in R using R Markdown. The workflow includes data cleaning and preparation with tidyverse and janitor, followed by statistical modelling using packages including MASS, mgcv, glmmTMB, pscl, lme4, mvabund, and AER. Figures and model visualizations were produced using ggplot2, gridExtra, patchwork, gratia, and showtext. The accompanying R Markdown script brings these steps together, providing a reproducible workflow from the original dataset through data exploration, statistical analysis, model comparison, predictions, and final visualizations.

Access information

Other publicly accessible locations of the data:

The data are not publicly available at any other location. The dataset is publicly accessible through this Dryad repository.

Data were derived from the following sources:

The data were derived from field surveys conducted in Volcanoes National Park, Rwanda,

Rwanda Development Board (RDB) Research permit

The research permit for this study was issued on 9 March 2023 by the RDB, Tourism and Conservation Department, and was signed by Linda Mutesi on behalf of Ariella Kageruka, Head of the Tourism and Conservation Department. The three-month research project was conducted during my professional internship at the Dian Fossey Gorilla Fund (DFGF) under the title “Population structure and regeneration dynamics of Prunus africana in high-altitude habitats of Volcanoes National Park, Rwanda.” No seeds, seedlings, or saplings were collected or removed from the forest; the study involved only field data collection and measurements of Prunus africana individuals.