Mukuvisi Woodland, located at latitude 17˚50'35S and longitude 31˚5'42E, covers approximately 274 hectors [15] and is situated in the highveld of Zimbabwe with a distance of approximately 5 km south east of Harare city centre (Figure 1). The land is owned by Harare City Council, but leased to a private operator, Mukuvisi Woodland Association [15]. Surrounding Mukuvisi Woodland is low to medium density suburbs and light industrial areas. Mukuvisi Woodland is found between the Mukuvisi and Chiraura rivers [16]. The 274 hectares are zoned into segments which comprise approximately 106 ha of game area, 156 ha of the public walking area and 12 ha of the interpretive and administration area. Mukuvisi Woodland has a diverse vertebrate fauna that includes mammals, birds, reptiles and fish. Mammals include herbivores such as zebra (Equus quagga), giraffe (Giraffa camelopardalis), wildebeest (Connocaetus taurinus), impala (Aepyceros melampus) and eland (Taurotragus oryx). The presence of mammal carnivores is not documented, although existence of small carnivorous animals cannot be ruled out. Mammal community is dominated by impala which approximately constitutes 70% of the total biomass [16]. There is a prolific bird life which includes the spotted creeper (Salpornis spilonotus), Violet backed sunbird (Anthrepes longuemarei), miombo blue eared starling (Lamprotornis elisabeth), and white eared barbet (Stactolaena leucotis). The dominant tree species include Julbernadia globiflora, Brachystegia speciformis, Brachystegia boehmii and Parinari curatellifolia, while Heteropogon contortus, Hyparrehenia filipendula, Sporobolus pyramidalis and Hyparrehenia dissoluta represent dominant grass species [16,17].

Annual rainfall ranges between 650 - 850 mm recorded between November and March and summer temperatures range between 16˚C and 30˚C [16]. Frost in May and June is common on the vlei and along the Chiraura and Mukuvisi river banks. The soils of the Mukuvisi protected area are nutrient poor, a characteristic of most of the miombo soils [18]. Coarse-grained granite [19] underlay Mukuvisi Woodland and gives rise to coarsetextured sandy soils.

The study area was stratified into two study strata, namely the central area and the park boundary. Sample plots were randomly placed based on Mukuvisi Woodland topographical map using grid intercept method across the defined two study strata. A total of 12 sample plots were sampled between April and May 2012. Four plots were 100 m from the boundary and the remaining 8 were located 5 km from the boundary and were in the central area of Mukuvisi Woodland. Sampling plots measuring 30 × 20 m were delineated on each of the strata. Woody species occurring within a sample plot were identified using field identification guides [20]. Further, the following variables were assessed in each plot following methods outlined by Gandiwa and Kativu [21]: tree height, basal circumference, plant status (dead or alive), tree stumps as a result of poaching, fire evidence, number of stems and saplings.

Data were tested for normality using KolmogorovSmirnov test in STATISTICA version 6 for Windows [22] and were found to be normal. Differences across the study strata were determined using t-tests for independent sample groups. Species diversity across the strata was determined using the Shannon-Weiner diversity index [23]. A principal component analysis (PCA) was used to analyze woody vegetation composition and structure across the strata, following Davies and Fearn [24]. In addition, hierarchical cluster analysis (HCA), using the Ward’s method, was used to analyze woody species

cluster composition across the strata [25].

A total of 117 woody stumps were recorded, mostly close to the boundary, translating to about 12.1% of poached woody trees across the strata. Figure 2 shows a tree stump left after cutting. One would notice that a type of a tree-cutting saw, probably a chainsaw basing on the tree trunk diameter and the precision displayed, was used as opposed to the common hand-held axe. This may explain mechanised techniques used by wood poachers to evade detection.

As indicated by the prevalence of recorded stumps, J. globiflora (69%) and B. speciformis (23%) had high tree stumps, showing that they were the most preferred and harvested while species like Vangueria infausta, Protea roupelliae roupelliae, Combretum molle and Newtonia hildebrandtii, were not only less common, but also least preferred with each accounting for less than two percent of stumps recorded.

A total of 968 woody plants, representing 47 woody species were recorded in the 12 sample plots. The dominant woody species recorded include J. globiflora, B. speciformis, B. boehmii and Parinari curatellifolia. All the measured or recorded woody vegetation variables indicated no significance difference (P > 0.05) between the strata (Table 1).

The Principal Component Analysis of the woody vegetation variables showed a variance of 34% (eigenvalue = 3.4) on Factor 1 and 23% (eigenvalue = 2.3) on Factor 2 in the woody species composition and structure (Figure 3). Sample plots C2, C3, C5, and C6 were drawn from the central area of Mukuvisi Woodland. These have high tree density, species diversity, shrub height and dead plant density. Those near the boundary, sample plots B1, B2, B3 and B4 show high shrub density, sapling density and a larger basal area compared to the plots in the central area. Fire damage was high in plot C4 and tree height is higher in plot C7 as shown by its outlying characteristic. Positively correlated were tree height and basal area, similarly to tree density and woody species diversity. Tree height and shrub density were negatively correlated to fire damage, whereas, sapling density was also negatively correlated to dead plants.

The Hierarchical cluster analysis (HCA) using the Ward’s method with Euclidian distances showed that there are 47 species in abundance data in the study sites. We noted two sub clusters A and B. Sub cluster A, sample plot B4 and C8 show a slight variability in species

composition compared to the sample plots B1, B2 and B3. In sub cluster B, sample plots C1, C2, C3, C5 and C7 show little variability in terms of species composition (Figure 4).

Our results showed that woody vegetation composition of the conserved Mukuvisi Woodland area was dominated by J. globiflora, B. spiciformis and B. boehmii, specifically resembling the Miombo woodlands of southern African savanna [26-29]. Regardless of the high representation of J. globiflora and B. speciformis and many other endemic woody plants at local or regional scale, these Miombo woody species are threatened by anthropogenic activities especially wood poaching as recorded in this study and this could be attributed to their hardwood type preferred for fuel wood, poles, construction and carpentry [5]. Our results are consistent with those of others like Pote et al. [5] and Furukawa et al. [30] who observed that these species besides being dominant in miombo woodlands are also preferred because of their basal circumference, as evidenced in this case by the stump size. Mukuvisi Woodland, being almost surrounded by urban residential areas is subjected to anthropogenic pressure where human livelihoods like energy needs are heavily dependent on wood derived from the study area. High wood poaching impacts were close to the boundary compared to the central parts as wood poachers want easy and quick exit from the park to reduce chances of being caught [31] and increasing distance from human settlements has been observed to correlate positively to wood biomass [32]. However, wood poaching may allow regrowth on under-represented species and woody biomass may accumulate rapidly after disturbance [31] as evidenced by the presence of saplings in the study area.

Our study results showed that fire damage was negatively correlated to tree size (tree height and basal area) and sapling density, whereas, fire damage was positively correlated to dead plant density (Figure 3). Anthropogenic fires in Africa form the ancient disturbance regimes that have probably shaped the miombo vegetation [33,34]. Effects of repeated burning include topkill of woody plants [35,21]. Repeated topkill of small trees and saplings makes them susceptible to “fire trap”, which suppress recruitment to large tree size [36]. It has been recorded that savanna woodlands respond variably to long term burning [37]. Our study results suggests that fire had little effect on tree density and woody species diversity as recorded in earlier studies elsewhere in the African savanna [34].

There were no significant differences among all assessed variables across the strata except sapling density. This indicates that wood poaching between the strata are not yet pronounced. Significant difference in sapling density

between the central and boundary strata may explain the higher impact of fire damage rather than wood poaching in the central area. The most observed trend is that human disturbed vegetation is characterised by high sapling density.

For woody species of slow growth rate, recruitment is interpreted as viable when woody species have a larger proportion of saplings than larger trees, and recruitment is considered poor when there are larger proportions of trees than saplings [38-40]. On the basis of this indicator, our study results of high proportion of tree density as compared to saplings could suggest less vibrant recruitment of woody species in Mukuvisi Woodland. There was evidence of large tree extraction on Mukuvisi Woodland as represented by relatively low tree densities, high number of stumps and high dead plant density. Wood poaching is known to affect micro-sites for plant establishment and the resultant woody plant diversity and density [41,42], this could be a contributing factor to relatively low woody species diversity and high dead wood density in Mukuvisi Woodland. Consistent with observations by Fadiman [43], species diversity was recorded to be decreasing in the central area as preferred species are getting scarce along the park boundary hence harvesting is now taking place in the interior. That herbivory is an ongoing activity in the studied woodland, together with fire, could have confounded the results of our study and, together with the effect of time, masked wood poaching impacts on the regeneration potential of miombo woodlands. Mukuvisi Woodland dominated by the miombo tree species, like the Brachystegia species of the Arabuko-Sokoke Forest, Kenya, exhibit a low diversity of canopy tree species [44]. This scenario, if left unchecked, is disastrous to biodiversity conservation and therefore, shows the importance of policy and law enforcement, particularly in the conservation and protection of micro-sites conserving remnant indigenous vegetation in areas where these are threatened by expanding urban settlements. Although we noted that there are some law enforcement activities in the study area and a perimeter game fence to try and ensure protection, there is need to increase surveillance and monitoring to cope with increasing pressure for wood fuel and forest products.