The proximate causes for woody plant encroachment are still poorly understood, but land use practices, including heavy grazing and anthropogenic reduction in fire regimes, are suspected to facilitate the process (Roques, O’conner & Watkinson, 2001). Despite the recognition of woody plant encroachment as a worldwide rangeland management problem, little is known about the rates and dynamics of the phenomenon, or its impact on fundamental ecological processes related to energy flow, nutrient cycling and its effects on biodiversity (Archer, Boutton & Hibbard, 2002). Most research has focused on the effects of woody plants on grass production (Archer, Boutton & Hibbard, 2002), instead of the underlying ecological mechanisms driving encroachment. We investigated the influence of overgrazing on bush encroachment by Tarchonanthus camphoratus (Asteraceae; L.) in a semi-arid savanna. Tarchonanthus camphoratus reproduces vegetatively and has small wind-dispersed seeds (Noad & Bernie, 1989; Coates-Palgrave, 2002); this together with its hardy characteristics and aggressive resprouting behaviour after disturbance, e.g. by fire, makes it a potential bush encroacher, particularly in disturbed habitats as other studies have shown (Kiruki & Njung’e, 2007). We predict that overgrazing by livestock facilitates net recruitment of T. camphoratus.
We compared tree recruitment rates in two different savanna habitats in Hell’s Gate National Park, Kenya and a neighbouring Maasai village, Kaparitan, during September 2006. Five replicates were placed within the Park, which was considered less disturbed, being a lightly wildlife-utilized savanna. Additionally, five replicates were located near Kaparitan village, where livestock inhabit and which by observation was considered a heavily grazed area. Each replicate contained an edge of a dense stand of T. camphoratus. We considered these visually conspicuous distinct edges of T. camphoratus as source pools for net recruitment of T. camphoratus seedlings.
To quantify the degree of Tarchonanthus bush encroachment within each replicate, we measured recruitment of T. camphoratus seedlings in three 8 × 100 m transects (running orthogonal to the conspicuous edge, marked with tape), in 10 m increments. Four height classes were used: <1 m, 1–2 m, 2–3 m and >3 m. Trees <1 m in height were considered seedlings. As T. camphoratus can reproduce with seeds or vegetatively (Noad & Bernie, 1989), trees more than 50 cm apart were considered as different individuals; however, we assumed throughout the sampling that sprouting near the base of a ‘parent’ tree indicated net recruitment, and therefore shoots were considered as separate seedlings. All data were tested for normality and analysed using the Minitab Statistical Software, Minitab Inc. (http://www.minitab.com).
There was more T. camphoratus recruitment in the livestock-grazed area, indicated by the significantly higher mean number of seedlings in comparison with the wildlife-grazed area (Mann–Whitney W = 1918.0, P < 0.0001). Figure 1 shows that the mean number of seedlings decreased with distance from the source pool. There was a mean of 30 seedlings in the first 10 m interval from the source pool in the livestock-grazed areas, but this decreased with increasing distance.
This was indicative of unidirectional spread of recruitment from source pools of T. camphoratus. Assuming height class to be an indicator of tree age (King, 1996), we found no difference between seedling heights and distance from the source pool (t-test; t = −1.61; P = 0.11). The number of trees in different height classes was not significantly different (t-test; P = 0.638), although an intriguing pattern was evident in the larger size classes (Fig. 2). The 1–2 m and 2–3 m height classes had greater net recruitment in the livestock-grazed areas, which corresponded to the seedling abundance results above. Old and established trees (above 3 m), however, were more numerous in the wildlife-grazed areas.
Our study determined that overgrazing facilitates encroachment of T. camphoratus in a semi-arid savanna. This was demonstrated by the higher seedling number and higher number of short trees (1–2 m and 2–3 m) in the heavily grazed livestock-grazed area. This area was subjected to high levels of disturbance by livestock. Net recruitment from a parent source pool was expected, but, our results indicate that this process is more apparent in areas disturbed by overgrazing, corresponding well with the fact that disturbance is a facilitator of encroachment (Myers & Bazel, 2003).
In a study to characterize population density, structure and early growth characteristics of T. camphoratus,Kiruki & Njung’e (2007) also found that regeneration is the highest in more disturbed habitats especially those experiencing exploitation through tree cutting, establishment of footpaths and charcoal burning kilns. In our study site, exploitation of T. camphoratus was less clear, but our disturbed site contained well-trodden footpaths and there was evidence of extensive trampling and overgrazing by livestock. Future studies need to address and quantify different aspects of disturbance explicitly, but the role of overgrazing in promoting T. camphoratus requirement seems probable from our results.
Tarchonanthus camphoratus has the potential to be a significant encroacher in disturbed semi-arid savanna habitats. The major caveat in our study is that we did not have remote-sensing data over large temporal scales, which would be the most adequate and explicit quantification of encroachment. We suggest that there should be a long-term monitoring of this plant, with the caveat that it remains to be seen whether it poses a significant threat as a bush encroacher. If this potential is realized, the pastoral tribes are likely to be significantly affected. In Kenya, majority of people are dependent on subsistence livestock farming in the savanna. Bush encroachment by T. camphoratus, perversely facilitated by the farmers themselves, has the potential to reduce grazing carrying capacity, with dire consequences to both wildlife and pastoralists dependent on this habitat.
We thank Clive Nuttman, Anthony Kuria and Daniel Nkalubo. This work was conducted in part while attending a field course organized and funded by the Tropical Biology Association, U.K. for which we are most grateful.