Bat and bird survey data from different agricultural land use gradients in Ghana and Zambia, 2020-2023

Data were collected between December 2020 and March 2023 in several localities within Ghana (n = 3) and Zambia (n = 5). A two-tier stratified random sampling design was employed to select the study sites within each locality. Initially, a spatially referenced grid measuring 1 km × 1 km was used to overlay the selected locality, segregating it into potential study areas. Following this, each site was categorised into one of five broad classifications based on the dominant land type within the grid cell:

• Not Applicable (NA): Areas unsuitable for sampling, such as urban centres.

• Agriculture Only: Regions where over 95% of the grid cell is covered by agricultural land.

• Forest edge: Locations with 25-75% of the grid cell encompassed by forest land.

• Forest: Regions where more than 95% of the grid cell is forest land.

A subset of sites within the categories of 'Agriculture,' 'Forest edge,' and 'Forest' were randomly selected for sampling.

At each survey point, birds were surveyed using a standardised point count methodology (Gregory et al., 2004). The sampling at these locations was conducted from dawn and lasted approximately 3 hours (with some variation depending on weather conditions), a timeframe chosen to coincide with the peak activity of forest birds.

Surveying for bats was conducted at the same locations used for bird mist netting. A combination of mist-nets and harp traps were deployed at each site, with sampling taking place on separate nights. Traps were established at sunset and operated for a minimum duration of 4 hours, with regular checks conducted every 15 to 30 minutes.

Several data inputters were used to digitally input the field data sheets. Then the completed dataset was compiled, error checked and validated.