THE SOUTH RIFT ECOLOGICAL MONITORING PROGRAM Background In order to fully understand any ecosystem, you need to start from the grass roots upwards, literally. This particularly applies to the East African savanna rangelands, where understanding how pastoralists and wildlife together maintain the rich diversities and densities that they do is crucial to conservation. The South Rift Ecological Monitoring Program was started in 2007 in order to provide the baseline data on which key conservation management decisions could be made. The area under continuous sampling is 1,000 km 2, encompassing both Shompole and Olkiramatian group ranches. The western limit is the geographically imposed boundary of the western Rift Valley wall, with the eastern limit being Lake Magadi. The Northern limit is the northern group ranch boundary (due to permissions and logistics) and the Southern boundary in Lake Natron and the Tanzania border. This â€˜ecosystemâ€™ boundary is a product of both ecological and practical considerations, but some of the more external influences will need to be looked at scales beyond this. What it does incorporate are the main water sources (namely the Ewaso Ngiro river), and the crucial dry season habitat and refuge (the Shompole swamps) which are believed to be the key to the success of this ecosystem. Much of the basis for the sampling design and methods used come from the Amboseli Research and Conservation Program, which began over 45 years ago, and thus for the most part are tried and tested methods for monitoring large mammal savanna systems. The aim was to design a sampling system that was appropriate both in time scale and spatial extent so as to capture the components of the ecosystem in an integrated, nested-hierarchical framework. This meant that the broader patterns would be captured at the largest spatial scales, and with the least frequency and the elements of the system which change with the most frequency will be captured on a more intensive spatial scale and, in most cases, more often (see Figure 1 below).
Figure 1. An integrated nested-hierarchical monitoring system. The red shows the aerial flight lines, the purple the ground transects, and the brown dots are vegetation plots.
Description of Methods Aerial Counts Sample aerial counts (conducted either personally with an experienced crew from the ARCP, or by the Department of Remote Sensing and Resource Surveys (DRSRS)). The methodology for aerial counts is standard in East Africa, as described by Norton-Griffiths (1978). We follow the methods of the ARCP and the analysis uses the standard Jolly II method (Jolly, 1969). Ideally these should be conducted 4 times a year, to correspond with the wet and dry seasons, however this is not always practical or affordable. The aim is to use the data to determine densities and broad distributions of main species of the ecosystem. The study area is divided into 2 x 2 km grids, and the aerial transects are flown down the centre of these grids, flying
east to west so as to cut across the main habitat gradients. This results in 25 transects of varying lengths depending on the topography. In addition we are working with the DRSRS to access the historical aerial counts of the region, which date back to 1978 and use them to compare with current counts and assess the trends of densities and distributions of wildlife and livestock in the area. Ground line transects Line transects (non-strip) are established on the ground, running along the same lines as the aerial transects. They are sampled using the Distance sampling methodology (Buckland et al., 1993) and will be analysed with the freely available Distance 5.0 software program. Line transects are a generalisation of strip transects. However, where as in strip transect sampling one assumes the entire strip is censused, in line transects only a narrow strip around the centreline is assumed to be censused. In other words, other than around the centreline there is no assumption that all objects are detected (Buckland et al., 1993). The analysis takes into account the various biases which would influence the detect-ability of an animal, particularly distance from the centreline. The details of the methodology are well explained by Buckland et al (1993). The end result is density estimates and distributions of the various species encountered, which can then be easily linked to ecological parameters which can either be collected simultaneously as counting the animals, or inferred afterwards using habitat maps. The transects are spaced out at 4 km intervals. This was decided upon as to space them every 2 km apart (as in the aerial counts) would take too much time every month to complete. The lengths of the transects vary according to how practical the terrain is to drive through. In the initial stages of the program, the counts were conducted approximately every four weeks for a period of three years (2008, 2009, 2010), and are now conducted only periodically. One outcome of using both the above methodologies will be the opportunity to compare results from both aerial and ground counts. We believe that aerial transects will give more accurate population estimates than the ground transects, particularly if done four times a year. However, a study by Jachmann (2002) comparing aerial and ground counts in Zambia showed that for most large herbivore species, the estimates from the aerial counts were considerably lower than those from the ground counts. The data pointed to undercounting as a major problem of aerial surveys. The data that is generated from this study, with the simultaneous capture of data from both aerial and ground counts, will contribute significantly to this debate. Vegetation monitoring plots Vegetation monitoring plots have been established across the study area. The locations were randomly allocated and are 10 metres in radius. There are two main tasks undertaken at the plots. a. Species composition analysis, which uses 80 plots, and should ideally be conducted twice a year, a few weeks after the rains begin, so as to
maximise on the probability of seeing all the major plants in the region as they were flowering, to ease identification of the species. With this analysis, all grasses, herbs, trees and bushes are sampled using a variety of methods described below. b. The second analysis, conducted on a sub-sample of 30 plots, is conducted approximately every 4 weeks or perhaps every 6 weeks in the drier seasons and measures biomass, %green, % grazed, % cover and relative abundance of grasses and herbs are measured using the pin intercept method described below. The pin intercept (point frame) method (Sutherland 1996; Mwangi & Western 1998) is a wooden A-frame which supports ten metal pins of one metre in length, angled at 33 degrees to the vertical. The plots are 10 m in radius, and the frame is placed along a string 5 times in each direction (N,S,E,W), with a one metre gap between each one within the plot, making a total of 20 frames (200 pins) in each plot. Each pin is examined and the number of ‘hits’ (plant parts that touch the pin) per pin is recorded as well as whether the ‘hit’ is green and grazed (or not). The ‘hits’ are further classified into grasses (monocotyledons) or herbs (dicotyledons). The number of hits per pin is then the total converted to a score of ‘mean hits per pin’. This measure can be directly correlated with biomass, once calibrated by measurement of clipped plots (Mwangi & Western, 1998). This sampling method is appropriate for vegetation usually less than 1 metre in height (thus mainly grasses and herbs). For the larger vegetation types such as bushes and trees the Bitterlich Stick method is more appropriate. The Bitterlich Stick is a stick of specified dimensions with a small cross bar at right angles on one end. It is effectively an exclusion quadrat which has been used extensively in forest inventory for estimated basal area cover of timber species and with some modifications (Cooper, 1963) can be applied to woodland and bushland communities to calculate species density in stems/hectare. This information can also be used as ground truthing for satellite images during classification, and for creating habitat maps, perhaps through a process such as ordination. Photographs are taken at each plot every month as well. Point animal and dung counts Point animal counts are done from the centre of each vegetation plot on arrival. Species, number and distance to the group are recorded. This simultaneous capture of animal presence and vegetation parameters has proved an informative data set in the Amboseli data and we presume the same will be achieved here. In addition, dung piles per species are counted in every plot at the end of the vegetation monitoring to illustrate basic presence or absence and then crushed. Water Arguably one of the most critical elements of the ecosystem is water. We monitor water using the following methods:
Taking a daily measurement of the Ewaso Ngiro river height from the main Ewaso Ngiro bridge Measuring rainfall using an automated weather station based at Lale’enok.
We plan to expand our general water monitoring to include understanding the flows and use of all the rivers and streams in the area, including the Oloibortoto, Entasopia, Sampu, and Pakase rivers. We also want to take more measurements at various points along the main Ewaso Ngiro river, and develop methods of assessing the quality, quantity and extent of the Shompole swamp water over the seasons and over the longer term. Other climatic factors Our automated weather station also collects: • • • •
Temperature Solar radiation humidity wind speed and direction
Mapping Maps and mapping will be a key part of the methodology of this program. This will involve; • • • • • •
Locating basic soil, topography and basin hydrology maps. This will only need to be done once and used as baseline information. Mapping of vegetation and in particular the swamp; extent and movements/changes over time. Mapping the movement patterns (settlements and grazing) of the pastoralists to get a handle on how the people are influencing the system. Mapping the main wildlife and livestock distributions according to seasons (using aerial and ground transect data). Mapping water sources, both permanent and temporary. Potentially creating maps that are based on a variety of interacting factors in a GIS framework in the form of species suitability maps based perhaps on factors such as habitat quality, distance to water, disturbances, competition and predation risks.
Rapid Biodiversity Assessments We would like to engage a team of experts to assist us conduct as rapid biodiversity assessment. This team ideally would comprise of scientists, photographers and also the local community so that we end up with not only a scientific record of diversity but also one that incorporates local knowledge and can be portrayed graphically. Establishment of a local herbarium
We would like to build a local herbarium, based at the Lale’enok Resource Centre, which will act as a reference and store for all the plant specimins found in the region. Comparisons with other ecosystems Gathering information about other ecosystems for comparative purposes, particularly the Amboseli and Tsavo ecosystems to start with will mainly involve an extensive literature search and perhaps some interviews. Way Forward The South Rift Ecological Monitoring Program has come a long way in the last six years. It now has a base at the Lale’enok Resource centre and much of the data collection and entry is now in the hands of the local community research assistants. However, the following needs to happen in order for the program to become fully relevant. •
The backlog of past data needs to be analysed, fed back to the local community and published in scientific journals.
Methods needs to be compared critically and the monitoring plan needs to be adjusted accordingly
As the program expands both in the amount of data it collects and expands into the hands of the local community resource assessors, more of those hands are needed. We thus need to hire an additional three resource assessors under this program and train them to the level of the current team leader, Albert Kuseyo.
More scientists and partner institutions need to become associates of this program in order to grow it appropriately.