Effects of grassland fragmentation and precipitation on Secretarybird (Sagittarius serpentarius) reproduction in the Serengeti ecosystem
Data files
Dec 02, 2025 version files 18.89 KB
-
GAMMs.csv
3.71 KB
-
Grassland_patches_cover_plot.csv
657 B
-
MODEL_2023_2024.csv
3.37 KB
-
MODEL_2024.csv
2.98 KB
-
MODEL_All_nests.csv
3.92 KB
-
README.md
4.24 KB
Abstract
Accurate demographic data are essential for understanding population dynamics and developing effective conservation strategies for threatened raptor species. This study provides the first quantitative assessment of how grassland fragmentation at landscape scale, driven by woody plant encroachment, and interannual variation in precipitation affect the secretarybird Sagittarius serpentarius reproduction within the Serengeti ecosystem in Tanzania. Through nest density estimation models based on nest survival and detection probabilities, and generalized additive mixed models, we estimated nesting success and productivity across three habitat types with different grassland fragmentation levels and rainfall regimes during the 2023 and 2024 breeding seasons. Nest density was highest in habitats with moderate to low fragmentation (50–<90% grassland cover: 7.6 nests/100 km²), lower in highly fragmented habitats (<50% grassland cover), and lowest in minimally fragmented ones (≥90% grassland cover). Nesting success increased significantly with grassland cover, from 18.9% in highly fragmented habitats – possibly due to greater influence of edge effects – to 71% in minimally fragmented ones. Productivity was lowest in highly fragmented habitats (0.8 successful nests/100 km²), highest in moderately to low fragmented ones (2.5), and intermediate in minimally fragmented habitats (1.7). Wet-season rainfall emerged as dominant driver of nest abundance across all habitat types. Substantially higher precipitation in 2024 than 2023 correlated with a nearly three times higher nest abundance in 2024 compared to 2023, positively influencing productivity (2.2 vs. 0.8 successful nests/100 km²). Additionally, lower long-term dry-season rainfall negatively affected nesting success. These findings highlight the secretarybird vulnerability to habitat and climate change, which may disrupt predator-prey dynamics and reduce nesting site quality and the number of breeding attempts. Conservation efforts in fragmented habitats should prioritize limiting the effects of woody plant encroachment and actively protecting nests to mitigate predation risks, thereby enhancing productivity and sustaining steady populations of this African savannah raptor.
Dataset DOI: 10.5061/dryad.p8cz8wb4r
Description of the data and file structure
Dataset Structure
The dataset consists of five distinct files. The first four are associated with the construction of the four sets of models presented in the reference article.
Files ‘MODEL_2023_2024.csv’, ‘MODEL_2024.csv’, and ‘MODEL_All_nests.csv’ contain the data used to develop the time-to-event density estimator (TNDE) models. The files presented here are named in accordance with the model sets as described in the reference article.
File ‘GAMMs.csv’ contains the dataset used to build the generalised additive mixed models.
File ‘Grassland_patches_cover_plot.csv’ contains the data used to assess the correlation between patch density and the percentage of grassland cover at the landscape scale (plot level). A detailed description of the variables and analytical procedures is provided in the manuscript.
Key Variables
The variables in these files follow the naming conventions required to run the models in the NestAbund v. 2.0 package (Péron et al. 2014), with the sole exception of the precipitation variables, which must be labelled as ‘Robel’ within each model, as specified by Péron et al. (2014). A detailed description of the precipitation variables is provided in the reference article.
- The binary variable ‘Cattle’ represents the variable ‘tree species’ in the reference article, where the value 1 corresponds to Balanites aegyptiaca and the value 0 corresponds to Vachellia tortilis.
- In File ‘MODEL_2023_2024’, the variable ‘Site’ indicates the sampling year: the value 1 corresponding to 2023 and the value 2 to 2024. In Files‘MODEL_2024’ and ‘MODEL_All_nests’ , the variable ‘Site’ denotes the specific habitat type, as described in the reference article.
- NestID (numeric or text identifying nest)
- FirstFound (week number when nest was first found)
- LastPresent (week number when nest was last located)
- Fate (fate= 0 in nest successfully produced young, 1 otherwise)
- AgeFound (age of nest in weeks when first found)
- AgeDay1 (age of nest on first week of survey)
- Max Brood Size (maximum number of incubation weeks).
- DS1021_mm (mean total precipitation during the dry season in mm). Long-term rainfall (2010-2021).
- RS1021_mm (mean total precipitation during the rainy season, including the break between short and long rains). Long-term rainfall (2010-2021).
- Pre1021_mm (mean annual precipitation). Long-term rainfall (2010-2021).
- PDS_mm (total precipitation during the dry season preceding each breeding season in mm). June-October 2022 and June-October 2023.
- PRS_mm (total precipitation during the rainy season in the two study years, in mm). November 2022–May 2023 and November 2023–May 2024, including the break between short and long rains.
- PNS_mm (total precipitation during the nesting period for each nest in mm, including the 20 days preceding laying, the incubation period, and the 20 days post-hatching).
- deltaPre (the difference between PRS and PDS, in mm).
- Number of nests (total number of identified nests)
- ID Plot (sampling plot identifier)
- Year (year of sampling)
- Habitat type TNDE models (1 = <50% grassland cover; 2 = 50–<90% grassland cover; 3 = ≥90% grassland cover)
- Habitat type GAMMs (1 = grassland coverage < 50%; 2 = grassland coverage ≥ 50%)
- Effort_km (length of transects surveyed within each plot)
- Grassland_coverage (percentage of grassland within each plot)
- Detectability_TNDE (nest detection probability estimated using Time-to-Event Density Estimator models, according to habitat type)
- latitude_roads and longitude_roads (coordinates of the centroid of the road transect surveyed within each plot)
- N_grassland_patches (number of grassland patches within each plot)
Reference
Péron, G., Walker, J., Rotella, J., Hines, J. E. and Nichols, J. D. 2014. Estimating nest abundance while accounting for time‐to‐event processes and imperfect detection. – Ecology 95: 2548–2557.