Skip to main content

Proximity to forest mediates trade-offs between yields and biodiversity of birds in oil palm smallholdings

Cite this dataset

Hamer, Keith C et al. (2021). Proximity to forest mediates trade-offs between yields and biodiversity of birds in oil palm smallholdings [Dataset]. Dryad.


There is much debate about how best to mitigate the effects of agricultural expansion on biodiversity, especially in the tropics. Recent studies have emphasised that proximity to natural habitats can enhance farmland biodiversity, yet few studies have examined whether or not such proximity mediates local trade-offs between yields and biodiversity, and hence alters conclusions about the ecological benefits of alternative farming strategies. Here we examine yield-biodiversity trade-offs, focusing on birds in oil palm smallholdings at different distances from remaining areas of forest, including a large forest reserve, in Ghana. We found significantly fewer birds on higher-yielding than lower-yielding farms, in terms of both species richness and abundance. For forest specialist birds (likely to be highly vulnerable to conversion of land to agriculture) we also found a greater trade-off (i.e. lower richness and abundance for a given yield) at farms further from forest, to the extent that increasing distance to the nearest forest from 1 to 10 km had a similar effect as a 3- to 5-fold increase in fruit yield brought about by increased intensification. Our study highlights the importance of accounting for the effects of natural forest in the landscape when considering agricultural policies for biodiversity protection, underlining the importance of a landscape-scale approach to conservation.


The dataset was collected during 2017 on 31 oil palm smallholdings and nearby primary forest in the Central region of Ghana.

Birds were sampled at each of 92 sampling stations (80 in oil palm; 12 in forest) from 5:30 to 8:30 and 15:00-18:00 GMT during fine weather, with each station surveyed twice (morning and evening) on different days in July 2017. Two observers walked to the middle of the station on each occasion, waited one minute for birds to return to normal activity then recorded the identity of all birds seen or heard within 30 m during a period of 15 mins, excluding African palm-swifts Cypsiurus parvus, which were often detected above the canopy. Unfamiliar vocalizations were recorded using a Tascam DR-05V2 handheld stereo recorder (Tascam, Montebello, USA) and subsequently compared with the Xeno-Canto online bird call database ( for confirmation of identity (not to determine abundance which was always done in the field).

Non-crop vegetation structure was assessed by recording the following data at all 80 sampling stations in oil palm: number of large trees (diameter at breast height [dbh] >25cm), small trees (dbh10-25cm) and saplings (dbh< 10cm) within 30m in each of four quadrants centered at the station; dbh and distance to the two nearest large trees, small trees and saplings in each quadrant  (up to eight large trees, eight small trees and eight saplings per station); estimated vegetation cover (%) at ground level and height of non-woody vegetation. All measurements were made to the nearest 1 cm with a tape measure. Ground cover, to the nearest 5%, was recorded as the mean of at least two independent estimates; these varied among recorders by no more than 10%. These data were then used to calculate the density of large trees, small trees and saplings at each station.

Oil palm fruit yield data were obtained for each study farm from 1st May to 30th September 2017. A field assistant was appointed in each of the two focal smallholder communities and was responsible for liaising with the selected smallholders on a daily basis and measuring and recording the number of fresh fruit bunches (FFBs) obtained whenever harvesting took place, together with the combined weight of FFBs and loose fruit. All weight measurements were obtained using a 200 kg Silverline 251087 heavy-duty spring hanging scale (Silverline Tools Ltd, Yeovil, UK). We then used log10 of the sum of the combined weights over a five-month period, multiplied by 12/5 to give an annual equivalent and divided by the area (ha) of the oil palm crop on each farm, as our standardized measure of FFB yield ha-1 at each farm (palms were planted at a uniform density of ~150 stems ha-1).

Usage notes

These datasets contain all the information needed to use them, including an information sheet in each file that explains each column of data.


Government of United Kingdom, Darwin Initiative*, Award: 3108

Government of United Kingdom, Darwin Initiative, Award: 3108