Mpala Research Centre is hosting a large-scale long-term exclosure experiment in which different combinations of large (and small) mammalian herbivores are allowed access to plots of semiraid rangeland (wooded grassland). These exclosures are located in the south-western corner of the property at approximately 36o52' E, 0o17' N and an elevation of 1800 m. The UTM coordinates are (37M) 262-265 E and 030-033 N (see Figure 1).

Physical characteristics

Average yearly rainfall is 500-550 mm. The rainfall is usually low December- February, and has three small peaks, in April, August, and November. Shade high temperatures range from 25 to 30oC and lows from 12 to 17oC. July and August are often the coldest (and windiest) months. Highs are lower, and lows are higher on the days after rainfall. Daily rainfall and temperature data have been kept at sites 5 km east of the exclosures since July 1992. At the Isbell/Young camp along the Ewaso Nyiro River until September 1994, and at the Mpala Research Centre thereafter.

The exclosures are located on a level area of black cotton soil, approximately 1 km west of the main Mpala escarpment, and 4 km due west of the Mpala Research Centre. There is little or no topographic or soil diversity (but see 'Glades', below), and the soil shows little vertical zonation (at least in the upper 80 cm). Black cotton soils are characteristic of impeded drainage sites. These soils have very high clay content, and high levels (even nodules) of calcium carbonate. They are prone to expansion when wet and contraction when dry. When wet, they become exceedingly sticky and exceedingly slippery, making travel either by vehicle or by foot very difficult. Because of impeded infiltration during rains, increased evaporation (through deep cracks) during dry periods, and strong hydrophily, these soils are effectively drier than other soils for the same rainfall, from the perspective of plants. Approximately 43% of the Laikipia ecosystem (calculated from Taiti 1992) is underlain with 'black cotton' vertisols characteristic of impeded drainage, and supporting a characteristic flora.

Vegetation

The vegetation on these soils in Laikipia is grassland or bushed grassland with varying densities of two species of swollen thorn acacias (A. drepanolobium and A. seyal ssp. seyal). The dominant grasses are Lintonia nutans Stapf., Brachiaria lachnata (Hochst.) Stapf., Themeda triandra Forssk., Pennisetum mezianum Leeke, and P. stramineum Peter. The dominant herbs are Aerva lanata (L.) Juss., Rhinacanthus ndorensis Schweinf., Dischoriste radicans Nees, and Commelina sp. Occasional additional woody species include Cadaba farinosa Forssk., Balanites aegyptica (L.) Del., Rhus natalensis Krauss, and Acacia mellifera (Vahl) Benth. See Appendix 1. Vegetative keys are available for all plants known to be in the plots, with reference specimens in the Mpala Research Centre Herbarium.

Most of the study area is Acacia drepanolobium wooded grassland. This is most common plant community in Laikipia, covering 28% of the ecosystem (Taiti 1992). Acacia drepanolobium is locally restricted to black cotton soils. All individuals produce swellings and the bases of some of their stipular thorns that are red when young and black when mature. The thorns themselves are white or pale grey. At nearby locations, there are also smaller populations of A. seyal var. fistula. A. seyal occurs in two forms: var. fistula occurs on black cotton soils and produces swollen stipular thorns that are white or grey when mature; var. seyal occurs on other soil types and does not produce swollen thorns. It is not known whether this difference is genetic or facultative.

Treeless glades are scattered throughout the A. drepanolobium wooded grassland and the exclosure plots. These glades vary in size, with those in the plots being 30-70m in diameter. They are dominated by Pennisetum stramineum, with some glades having patches of Digitaria milanjiana (c.f. Young et al. 1995). All have large (excavated) termite mounds at or near their centers. The soils here appear to be different from the surrounding black cotton, being less sticky when wet (lower clay content?). At some sites in Laikipia, particularly to the west and south, similar sites cover large areas, forming plains of Digitaria, and hosting in addition to cattle large numbers of zebras, Thomson's gazelles, Grant's gazelles, and Jackson's hartebeasts.

There also appear to small patches of vegetation that are often greener after rain, dominated by Pennisetum stramineum, and with termite mounds at their centers. However, they are not treeless, have other grasses, and are smaller (10-20m in diameter) than the glades described above. Whether these are 'incipient glades' is not known.

This overall community has low species diversity. Over large areas, A. drepanolobium accounts for virtually the entire overstory (>98%). Five grasses and two forbs account for 95% of the relative cover and 86% of the relative frequencies of plants (See Appendix 1). Bare ground averages 56%. This biological simplicity makes for an workable system in which to study complex interactions among herbivores.

Animals

Similarly, there is a (relatively) low diversity of smaller, but not larger vertebrate animals (see Appendix 2). The bird and herp lists are undoubtedly incomplete, and additions are welcomed. The most common large herbivores are cattle, zebras, hartebeasts, and Grant's gazelles.

There have been no surveys of the invertebrates, other than the ants inhabiting the swollen thorns of A. drepanolobium and A. seyal var. seyal , of which there are at least nine species (Young et al., in revision). No harvester ant nests have been found. There is at least one species of termite common in the plots.

Exclosure plots

The mammalian herbivores at Mpala can be divided into four classes: megaherbivores ('M': elephants and giraffes), other wild large mammals ('W'), cattle ('C', although there are other livestock at Mpala, none use black cotton), and small mammals (rodents and hares). We are attempting to tease apart the effects of each of these on vegetation and on each other. There are four kinds of barriers used at the exclosure plots (see Thouless & Sakwa 1995):

1. Standard game fence. Electrified to 7,500 volts, this eleven-strand fence is designed to keep out all animals over five kg. Gates in these fences can allow the admission of cattle.
2. Tall 'dingle'dangle' (mega-herbivore) fence. Also electrified, but with no wire below 1.8 m. Designed to exclude only elephants and giraffes.
3. Red marker posts. These post demarcate the areas were cattle are allowed and not allowed. Each post is painted red on the side of the line where cattle are not allowed, and unpainted on the other side. Since all cattle herds on Mpala are continuously accompanied by herders, these markers should effectively exclude only cattle. Within the game fence, there is an electric wire running along the tops of these short red posts.
4. Rodent exclosures. These 3 x 7 m fences are 0.5 m tall, and located in the total exclusion plots. To minimize edge effects, only the central 1m x 5m will be surveyed.

The first three types of barriers can be used to create six different combinations of large mammal herbivores:

1. All large herbivores excluded ('0').
2. Only cattle allowed ('C').
3. Only wildlife allowed ('MW').
4. All wildlife except elephants and giraffes allowed ('W').
5. Cattle and all wildlife except elephants and giraffes allowed ('WC').
6. All large herbivores allowed ('MWC').

Note that there are eight theoretical combinations of the three large mammal classes. Two are not possible ('MC' and 'M'), because any fence that excludes wildlife also excludes megaherbivores.

Each of these treatments is represented by three plots, each 200m x 200m (four hectares). This is large enough to allow for multiple uses and surveys (see below), and large enough to be seen readily by satellite (Landsat or Spot). Six plots, representing one of each of the treatments, are grouped together to form a 'block'. There are three blocks, each 400m x 600m (North, Central, and South, also known as Nilotica, Cisticola, and Saccostomys). The locations and orientation of these blocks, and the plots within them, are shown in Figure 2. Care was taken to mix up the relative position of treatments within the blocks. Blocks themselves were located in such a way as to a) minimize vegetation heterogeneity among blocks and treatments, and b) subject natural glades to as many interesting treatments as possible.

Each 200m x 200m plot is further divided and into sixteen 50m x 50m sub-quadrats marked with numbered and tagged short posts, to allow for efficient location of individual features (seedlings, termite mounds, cages). Each of these posts also has a tall flag mounted on top. See Figure 3 for details.

Each plot is also a set of (unmarked) nested sub-quadrats, each of which delineates a different experimental use. Areas near the plot edges are available for the most manipulative uses, while the innermost area is set aside for minimal use. See Figure 4 for details.

These exclosures were built between 17 July and x August 1995, during which time animal use may have been generally restricted by a group of up to ten workers building fences and up to eight researchers collecting baseline data. The electricity was turned on x August 1995. Records are being kept of each visit that herded cattle make to each of the quadrats.

Baseline data include:

1. Surveys of the density and size distribution of trees.
2. Mapping of non-Acacia drepanolobium trees and shrubs.
3. Tagging of ~100 Acacia drepanolobium trees per plot, along with data on stem diameter, height, and ant occupant; and marked measured twigs.
4. Surveys of percent cover by each understory plant species.
5. Surveys of the frequency of each understory plant species.
6. More detailed surveys of nine rodent exclosures and nine control plots.

Details of these surveys are available upon request.

Access to the plots

These plots were designed by Truman Young and Alan Smith, and funded by the Smithsonian Institution and the National Geographic Society. They are available to any researcher with a compelling reason to visit them. This does not necessarily include site visits for interest only or short-term student projects, although both will be considered. One of the main purposes of these plots is to create controlled differences in herbivore presence. These differences will be compromised by human visitation rates that are too high and serve to discourage wildlife from plots in which their presence is desirable. In addition, we wish to minimize any trampling effects that accompany high visitation rates by humans.

Therefore, visits to the plots will a) be minimized, and b) be coordinated to coincide with as many projects as possible. In other words, visits to a particular block should be timed so that different research teams will visit the block on the same days. This coordination takes precedence over personal preferences, schedules, and other research. If you are planning a visit to the plots, make sure you leave yourself plenty of time leeway to fit your schedule into the schedules of other researchers. A timetable will be established several weeks in advance, and will be made available to interested parties.

All visits to the plots, whether for research or not, will be cleared through Truman Young. Because he is often in the United States, this requires prior planning and advanced communication. No one else has the authority to allow access, unless prior arrangements have been made.

All research carried out in the plots will require that the restrictions described in these pages be honored. A detailed description will be required of exactly what will be surveyed, measured, collected, or manipulated. This description will be sent to Truman Young at least three months before the proposed research. Enclose sample data sheets. Any and all data collected in the plots will be photocopied before leaving Mpala, along with a description of the methods sufficient to understand the data, and will be deposited in an data archive. Such data will remain the property of the researcher for a period of five years, after which time it becomes the property of the project. No one will be given access to the archival data without the permission of Truman Young and notification of the person who collected the data. Any publication arising from the use of this data (after the five year period) by any other than the original researcher will at the least fully acknowledge its source. Regardless of author, any publication using data from the plots will acknowledge Truman Young and Alan Smith, Mpala Research Centre, George Small, the James Smithson Fund of the Smithsonian Insitution, and the National Geographic Society. Depending on the level on involvement and the use of long-term records, co-authorship may be appropriate. For additional details, see 'Guidelines for exclosure researchers'.

References

• Taiti, S.W. 1992. The vegetation of Laikipia District, Kenya. Laikipia-Mt. Kenya Papers. B-2. Universities of Nairobi and Bern.
• Thouless, C.R. & J. Sakwa. 1995. Shocking elephants: fences and crop raiders in Laikipia District, Kenya. Biological Conservation 72: 99-107.
• Young, T.P., C. Stubblefield & L.A. Isbell. Coexistence among obligate acacia ants. Oecologia, in revision.
• Young, T.P., N. Patridge & A. Macrae. 1995. Glades in acacia bushland and their edge effects in Laikipia, Kenya. Ecological Applications 5:97-108.