Research Methods

Site design & data collection

Sites were chosen along a farming intensity gradient ranging from low to high (see site selection criteria here). Initially population density was used to define intensity as no other data were available. During the data collection phase, we aimed to obtain various other measures of intensity such as the amount of labour used, the amount of land under fallow and also the gross yield of the main cash crops (mostly coffee and bananas) andrelate these to biodiversity measures.

Each of the 26 sites used in this project actually consists of a 1km square.Landuse, birds, bee, butterfly and woody vegetation (trees) data are recordedaccording to the design below. Ten bird points were selected randomly and invertebratepan-trapping was carried out at alternate ones as the number of individual invertebratescollected from each pan trap meant that sampling at all 10 points would not havebeen feasible. Habitat data was also recorded at each point within a 25m radiusbut this is unlikely to capture the landscape and so 5 1-km transects were walkedat fixed points across the square to capture this. Further details about eachaspect of the data collection can be found in the sections


Two approaches are being used to assess the abundance and diversity of birds using the study sites. The first is point counts, where the observer records the species present and places then into distance bands and the second is a straight ten minute count where only the presence of each species is noted. We used both approaches as analysis of pilot data showed that the more intricate point count method meant that some rare species were being overlooked.

Fieldwork took place at these sites from February 2006 to March 2007. Five visits were to each site, each at different times of year, to ensure that birtds were recorded in every season. At each site, on day one PCs were performed at five of the ten points and TMCs at the other five. The following day the survey method was switched at each point so that 10 PCs and 10 TMCs were mad for each site visit.

Bees, pollination and butterflies

Sites were visited at the same time at the bird surveys. Three main survey methods were used to quantify the bees and butterflies.

Pan traps

Pan traps (3 colours – blue, red and yellow) filled with water were set for 24 hours in suitable habitat. These attracted a great variety of invertebrates and the samples were therefore initially sorted to remove just the bees which were the main focus of the project. The remainder of the samples have been kept and if anyone can make use of them, we would be interested to hear.

Butterfly traps
These were baited with ripe bananas and left for 24 hours. The butterflies were identified, counted and released. Species which could not be identified were taken for subsequent identification with a reference collection at the Zoology Department at Makerere University.

Butterfly & bee transects.

Two km transects were walked at random around the sites and the number of bees and butterflies within a 10m radius were counted. Bee identification was especially difficult and so these were sweep-netted and kept for subsequent identification.

Land use
The land use survey is made up of two parts. First five 1 km transects will be walked and the length of each landuse type will be evaluated using a GPS handheld unit. This will give a percentage cover of important landuse classes, such as fallow, crops, woodland and homesteads. In addition, as woody vegetation is likely to be a very important component biodiversity component a separate tree survey was commissioned. This included identifying individual trees by species and their origin (exotic versus native), their size, the canopy cover and an analysis of the spatial distribution/dispersion of the trees within the landscape.

The draft report can be viewed in the document centre here


The overall purpose of the survey is to collect quantitative data to analyse how levels of biodiversity are influenced by the level of agricultural intensification, specific land management practices, and, importantly, the demographic, economic and institutional factors influencing such practices.

In response to a recent article by Rhys Green et al. (Science, Vol. 307, January 2005), the survey, combined with the surveys of birds and insects, will also be used to test whether there is a relationship between agricultural yield and wildlife density (‘yield – density function’) in the study area. We use ‘gross value of production per hectare’ as the measure for agricultural yield. We do not find it feasible to estimate ‘net value of production’ (i.e. gross value less costs) due to the great difficulties in estimating labour costs, which may account for up to 90% of all production costs in the smallholder sites.
Cropping frequency is a key indicator for land use intensity and is estimated in the land use survey. Yet given the key importance to the whole study of ‘land use intensity’, in the household survey we will attempt to estimate land use intensity in terms of the ‘inputs per hectare’. Since labour is by far the most important input in production in smallholder farming, we will use labour inputs as the key indicator. Use of agro-chemicals (fertilisers, pesticides, herbicides) will also be measured.

The data are collected at the farm household and plot levels, respectively, through the administration of a questionnaire to farmers residing within each site. The questionnaire is administered twice, once after each season, in order to reduce the recall period for the farmer (especially regarding production). Only some of the questions are asked in the second interview. The survey also includes GPS measurements of the size and location of the surveyed plots and homesteads. Five to six farmers

The survey will be complemented by the collection local level (site specific) information on variables that do not vary between households within the same site, but that are likely to vary between sites or clusters of sites. This includes, especially, data on physical market access (estimated through existing GIS data on travel time to urban centres), rainfall, land tenure conditions, population density, farmers’ perceptions and evaluation of the biodiversity and how it should be managed, etc. A focus group interview with a small group of 3-4 key informants (farmers, LC1 officials, etc.) in each cluster or site will be conducted to collect such information that is not available through secondary sources.

  Copyright © Project Partners. All rights reserved.

Birding Top 500 Counter