This article was published in an Elsevier journal. The attached copy is furnished to the author for non-commercial research and education use, including for instruction at the author’s institution, sharing with colleagues and providing to institution administration. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright Forest Ecology and Management 247 (2007) 48–60 www.elsevier.com/locate/foreco co p y Colonial logging and recent subsistence harvesting affect the composition and physiognomy of a podocarp dominated Afrotemperate forest Michael J. Lawes *, Megan E. Griffiths, Stéphane Boudreau 1 School of Biological and Conservation Sciences, Forest Biodiversity Research Unit, University of KwaZulu-Natal, Private Bag X01, Scottsville 3209, South Africa al Received 1 November 2006; received in revised form 16 March 2007; accepted 10 April 2007 Abstract or 's pe rs on Indigenous forests are declining throughout southern Africa, partly because of exogenous disturbances such as harvesting of forest resources. We evaluate the influence of colonial logging and present-day harvesting of pole-size trees (2–15 cm dbh), used for construction and fuelwood, on the structure and composition of iGxalingenwa forest in KwaZulu-Natal, South Africa. This forest has a distinctive physiognomy, with a thicketlike understorey and an emergent layer of primarily coniferous Podocarpus spp. These emergent trees represent individuals that were too small to be harvested during the colonial period. We argue that the dense angiosperm-dominated understorey arose in gaps created by logging, which allowed greater light levels at the forest floor. The understorey has since been intensively harvested (36% of available stems) for pole-sized stems, including canopy species, effectively suppressing the advanced regeneration as well as the replacement of the senescing canopy and emergent tree populations. The harvesting intensity of pole-sized stems is related to the species, size, and availability of trees. Few straight pole-sized stems are available among the understorey species and typically straight-stemmed canopy species are opportunistically harvested. Examination of community composition further suggests that, without management intervention, this forest will have a reduced stature, a thicket-like physiognomy, and be less diverse and less dominated by conifers. To rescue iGxalingenwa forest from this successional path requires a reduction in harvesting pressure on seedlings and saplings of species that normally dominate the canopy layer, many of which (Podocarpus falcatus, Ptaeroxylon obliquum and Calodendrum capense) appear to have declining populations. The effects of commercial logging a century ago still persist and are being compounded by subsistence harvesting with significant changes to forest physiognomy and composition and the possibility of arrested succession. # 2007 Elsevier B.V. All rights reserved. Keywords: Disturbance; Logging; Forest dynamics; Podocarpus; Regeneration; South Africa 1. Introduction Au th It is essential to understand the history of natural and anthropogenic disturbances to forests to interpret their current ecology and ensure sustainable management (Bowman, 2001; Coomes et al., 2003; Echeverria et al., 2006; Lawes et al., 2006). Natural disturbances and the use of forests by people have dramatically altered southern African forests, resulting in fragmentation (Lawes et al., 2004a), changes in important * Corresponding author. Tel.: +27 33 260 5443; fax: +27 33 260 5105. E-mail address: Lawes@ukzn.ac.za (M.J. Lawes). 1 Present address: Département de biologie, Local 3047B, Pavillon Alexandre-Vachon, Université Laval, Sainte-Foy, Québec G1K 7P4, Canada. 0378-1127/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.foreco.2007.04.012 ecosystem functions (Kotze and Lawes, 2007), and modification in forest structure and composition (Lawes et al., 2006). Here we examine the effects of extractive use of timber species over 150 years on an Afrotemperate forest. Research on the threats to indigenous African forests primarily focuses on the present-day harvesting of large trees by local inhabitants. Even though the collection of pole-sized stems constitutes the most immediate risk to wooded areas (Hall and Rodgers, 1986; Vermeulen, 1996; Luoga et al., 2000, 2002), their use by subsistence harvesters has received little attention because these poles are not commercially viable. However, pole-sized stems (2–15 cm dbh) are a critical source of timber and non-timber products that support the livelihoods of rural communities (Obiri et al., 2002; Boudreau et al., 2005). The understorey from which these poles are collected is an 49 M.J. Lawes et al. / Forest Ecology and Management 247 (2007) 48–60 al co p y biodiversity value and is a critical habitat for the endangered Cape parrot (Poicephalus robustus) (Wirminghaus et al., 1999; Downs, 2005). Reconciling these potentially conflicting needs is central to ensuring the survival of this forest and requires a comprehensive understanding of past exogenous disturbance to the forest, the natural structure and regeneration dynamics of the forest, and the factors controlling present-day harvesting. In this study we determine the regeneration patterns of the common understorey, canopy and emergent species at iGxalingenwa. We evaluate the level of use of pole-sized trees by local harvesters and identify the factors influencing selection of particular stems for harvest. Our main objective is to contrast the impacts of colonial logging and recent subsistence harvesting of trees on forest composition and physiognomy. In doing so, we demonstrate the relative importance of these events to the long-term structure and survival of this representative Afrotemperate forest and other forests that have endured colonial logging followed by subsistence harvesting in recent times. on 2. Methods 2.1. Study site Surveys were conducted in iGxalingenwa forest (30.018S, 29.638E), approximately 25 km from the town of Creighton in southern KwaZulu-Natal, South Africa (Fig. 1). This Afrotemperate mistbelt forest (673 ha) occupies a south-facing slope between 1300 and 1600 m above sea level. The region receives a mean annual rainfall of 1060 mm. Winters (June through August) are dry (median monthly rainfall = 3 mm, with short periods of snowfall every 3–4 years) and summers (December through February) are wet (median monthly rainfall = 169 mm). The forest has a mean annual temperature of 15 8C (mean monthly range: 10–19 8C) (Schulze et al., 1997). Au th or 's pe rs integral part of the forest ecosystem (Newbery et al., 1999; LaFrankie et al., 2006). While some studies have found that small understorey gaps created by harvesting of pole-sized stems do not have adverse effects on tree diversity in indigenous forest (Boudreau and Lawes, 2005), others suggest that selective removal of pole-sized stems could lead to the local extinction of certain species (Obiri et al., 2002; Boudreau et al., 2005). Forests throughout Africa also have a history of logging by colonial settlers (King, 1941; McCracken, 1986, 1987; Struhsaker, 1997). The long-term effects of the removal of large trees on forest community composition are not fully understood, although it is generally acknowledged that few, if any, forests have not been affected by anthropogenic disturbances. A primary ecological impact of logging is a reduction in canopy closure and subsequent increase in the amount of light reaching the forest floor (Sekercioglu, 2002). This can ultimately lead to a change in the species composition of the regenerating forest (Brokaw and Scheiner, 1989). We survey the structure and composition of the tree community of the iGxalingenwa forest in South Africa and identify the influence of logging of large trees (>40 cm dbh) during the colonial era (1850–1910) and the present-day harvest of mostly pole-sized stems by local communities, on contemporary forest community structure. Although remotely located, this montane Afrotemperate forest has a long history of anthropogenic disturbance. During the 1880s to the early 1900s, Podocarpus spp. were the primary trees logged from the forest, although there was also selected harvesting of Vepris lanceolata, Ocotea bullata, Calodendrum capense, and Ptaeroxylon obliquum (Fourcade, 1889; Hutchins, 1905; King, 1941). In recent decades, there has been heavy use of forest products by local inhabitants, particularly of fuelwood, building and fencing material, and plant materials used in traditional medicines (muthi) (Nomtshongwana, 1999; Robertson and Lawes, 2005). This Afrotemperate forest also has high Fig. 1. Map of iGxalingenwa forest showing the study plots, transects, and the surrounding households comprising the user community. 50 M.J. Lawes et al. / Forest Ecology and Management 247 (2007) 48–60 at two points in each Whittaker plot. Mean foliage density for each stratum was calculated from the weighted sum (Walker, 1976): P ni c i FD ¼ N on al co p y where ni is the number of sampled points where the height class has rank i, ci the class midpoint of rank i, and N is the total number of sample points. We used the coefficient of variation of the FD from all strata in a plot as a summary statistic of the vertical heterogeneity for the plot. The horizontal spatial pattern of the vegetation was estimated using Roth’s (1976) heterogeneity index. This index uses the point-center-quarter (PCQ) method, where distances are measured from a central point (i.e., a focal canopy or understorey tree) to the nearest similarly sized tree in each quadrant of a circle. These distances give information about dispersion and density of trees in the sample. The coefficient of variation of distances was used as a measure of the horizontal heterogeneity (Roth, 1976). Importance values (IV) were determined for each species in the plots (Mueller-Dombois and Ellenberg, 1974): IV ¼ ðRD þ RDo þ RFÞ where RD is the relative density, the number of individuals of one species as a percentage of the total number of individuals of all species, RDo the relative dominance, the total basal cover (at breast height) of one species as a percentage of the total basal cover of all species, and RF is the relative frequency, the percentage of all surveyed plots occupied by a particular species. We used dbh size-class frequency distributions (SCDs) to describe the population structure of the 20 most important species found in the sampling plots. Although this type of static and short-term data cannot reliably predict long-term forest dynamics (Condit et al., 1998), they can provide important insight into population stability and regeneration potential over time (Lykke, 1998; Obiri et al., 2002; Boudreau et al., 2005). SCDs were analysed using the methods of Condit et al. (1998) and Lykke (1998). Each size class had an interval of 5 cm (i.e., 0–5, 6–10, . . ., 180–185). For each of the 20 most important species, the slope of the SCD was calculated using least squares linear regression, with the midpoint of the size-class as the independent variable and the number of individuals in a sizeclass (Ni) as the dependent variable. We derived straight-line plots of the size-class distributions by transforming the number of individuals in a size-class by ln(Ni + 1) because some classes had zero individuals (Condit et al., 1998). The interpretation of each SCD was based on the shape of the regression slope. Type I SCDs have a negative slope that results from a higher number of individuals in smaller sizeclasses than in larger ones (Condit et al., 1998). When the data are not transformed, this type of SCD is characterised by an inverse J-shaped distribution, indicating normal recruitment and growth (Everard et al., 1994). Type II SCDs have flat distributions with a slope not significantly different from zero, indicating equal numbers of regenerating trees and mature rs Conifers such as Podocarpus falcatus, Podocarpus henkelii, and Podocarpus latifolius are typically emergent or canopy trees, and more rarely broadleaved canopy trees such as Celtis africana and V. lanceolata are also present as emergents (Lawes et al., 2006). The emergent trees form an upper stratum up to 38 m tall, above a canopy up to about 24 m. Along with Podocarpus spp., broadleaved taxa such as Celtis africana, C. capense, Kiggelaria africana, and V. lanceolata form the canopy; Cussonia spicata, Rothmannia capensis, and Xymalos monospora form a subcanopy; the woody understorey is dominated by Clausena anisata, Diospyros whyteana, and Eugenia zuluensis. iGxalingenwa forest was highly prized for its timber resources (Fourcade, 1889) and Podocarpus spp. were removed mainly during 1905 and 1906, and again in smaller quantities (dead and windthrown trees only) during 1928–1929 and 1936– 1937 (Hutchins, 1905; King, 1941). Numerous saw-pits and skid trails in the forest are relicts of this period of exploitation. The forest has been managed by Ezemvelo KwaZulu-Natal Wildlife (EKZNW) since the 1980s but uncontrolled harvesting of natural resources by local residents is ongoing. Two rural settlements, Qaqeni and Ngxola, are located adjacent to the southern and western borders of the forest. Mean household income is approximately US$60 per month and subsistence livelihoods are dependent in part on natural resources from the forest (Robertson and Lawes, 2005). 2.2. Forest structure Au th or 's pe Using a modified Whittaker plot sampling method (Stohlgren et al., 1995; Obiri et al., 2002) we surveyed trees in 30 plots of 0.1 ha (50 m  20 m). The cumulative species curve asymptote was 24 plots (F 2,28 = 126.57, r2 = 0.90, P < 0.001), indicating sufficient sampling intensity to capture the species in the forest. Plots were systematically located in the southern and western parts of the forest and sited near the forest edge (n = 10), at approximately 200 m from the forest edge (n = 10) or in the forest interior at least 400 m from the edge (n = 10), to cover gradients of stand characteristics with increasing distance from the forest edge (Fig. 1). Plots had similar aspect, were located on shallow slopes, and there were no notable differences in superficial soil layers among sites. In each plot we identified the species and measured the height (using range finders) and diameter at breast height (dbh) of all trees with dbh > 10 cm. We measured the diameter of all stumps and identified them to species by examining the wood grain and colour, bark, and coppices. The height and diameter of saplings (2 cm < dbh < 10 cm) were recorded in two 10 m2 (5 m  2 m) subplots (Obiri et al., 2002). In 10 1-m2 (2 m  0.5 m) subplots arranged around the perimeter, we measured the height and diameter of all seedlings (<2 cm dbh , >50 cm tall), and ranked percentage cover of herbs and grasses on a Walker scale of 0–7 (Walker, 1976). Foliage height diversity was evaluated by estimating percent cover in ten height strata (0–0.5, 0.5–1, 1–3, 3–5, 5–10, 10–15, 15–20, 20–25, 25–30, and >30 m; Krüger and Lawes, 1997). Percent cover was ranked on a Walker scale of 0–7 and scored 51 M.J. Lawes et al. / Forest Ecology and Management 247 (2007) 48–60 least two AIC units from other models) was accepted as the best fit to the data. In addition, we estimated the relative importance of the predictor variables by the method of summing the Akaike weights (wi ) across all models in which the variable occurred (Burnham and Anderson, 2002). The larger the wi (range = 0– 1), the more important a variable is relative to the others. individuals. SCDs with positive slopes result from the presence of few (Type III) or no (Type IV) regenerating individuals. In many cases a SCD with a positive slope indicates that a species requires a large-scale disturbance for recruitment to occur, but in South Africa positive SCDs sometimes indicate that forest regeneration has been disrupted by harvesting or other disturbances (Everard et al., 1994). 2.4. Statistical analysis y 2.3. Harvest intensity on al co p The number and species of mature trees, saplings, seedlings, and vines, as well as the percent grass and herb cover, number and species of tree and sapling stumps, canopy height, vertical heterogeneity, and understorey and canopy horizontal heterogeneity were compared among plots at different distances from the forest edge using MANOVA (SPSS, 2002). Multi-response permutation procedures (MRPP) were performed in PC-ORD (McCune and Mefford, 1999) to test for differences in tree, sapling, and seedling community composition among plots. The composition of the understorey and canopy strata were compared to one another using Mantel tests in PC-ORD. For transects, a MANOVA was also used to compare the number and species of living and harvested polesize stems, the number and species of harvested large stems, and the vertical heterogeneity among transects. A chi-square goodness-of-fit (GOF) test was used to examine the age of harvested stems as a function of stem diameter. Au th or 's pe rs We estimated harvest intensity of trees from vegetation surveys along 10 transects. The cumulative species curve asymptote was at seven transects (F 2,8 = 23.96, r2 = 0.86, P < 0.001). Transects were 300 m  5 m and divided into 10 m  5 m quadrats. The start of each transect was 10 m from the forest edge (Fig. 1). Transects were oriented perpendicular to the forest edge and into the interior of the forest. As for the plots, all transects were sited in the southern and western portions of the forest. Within each transect, we measured the species identity, dbh and height of all living pole-sized stems (2 cm < dbh < 15 cm). The diameter of all stumps (representing harvested trees) was also measured and, where possible, stumps were identified to species level by examining the wood grain and colour, bark characteristics, and coppices. The age of stumps was ranked according to the degree of decay of the stump following the criteria described by Boudreau et al. (2005). In each quadrat, foliage height diversity was evaluated using the method described above. The presence of paths, natural canopy gaps and saw-pits or skid-trails was also noted. We modeled the harvest intensity of pole-sized stems at iGxalingenwa using a generalised linear model (GLM) based on a Poisson distribution with a logarithmic link function (McCullagh and Nelder, 1989). The response variable was the number of harvested stems per quadrat from each of the eight most commonly harvested species (Celtis africana, Clausena anisata, D. whyteana, E. zuluensis, P. henkelii, R. capensis, X. monospora, and Zanthoxylum capense) in each of three diameter size classes (2–5, 5–10, and 10–15 cm). Harvesters are known to select pole-sized stems based on species, size, availability and accessibility (Obiri et al., 2002; Boudreau et al., 2005), therefore our model included the following variables: species, diameter size class, stem availability (total number of available stems = stems = stumps of a given diameter for each species), distance from nearest household, household density within a 1 km radius, and the number of paths passing through the quadrat. All models were fitted using GENSTAT version 8.1 and included transect as an absorbing factor (Lawes Agricultural Trust, 2005). In addition to the null model, seven candidate models were formulated a priori to avoid data dredging (Burnham and Anderson, 2002). Predictor variables were added into the models in a hierarchical fashion, starting with the variable that was expected to have the most influence on harvesting. The most parsimonious model was selected based on Akaike’s information criterion (AIC) (McCullagh and Nelder, 1989). The model with the lowest AIC value (and a difference of at 3. Results 3.1. Forest community structure We measured 802 trees in the Whittaker plots, representing 38 species. The most common emergent, canopy and subcanopy tree species were P. henkelii (16%), Celtis africana (13%), and X. monospora (11%), respectively, while the most common understorey tree species was E. zuluensis (16%). At the sapling level, 389 individuals were measured from 34 species. The most common sapling species were Clausena anisata (14%), E. zuluensis (11%), D. whyteana (9%), and Celtis africana (9%). For seedlings, we measured 903 individuals from 40 species, of which the most common were Clausena anisata (21%), Gymnosporia harveyana (7%), Andrachne ovalis (6%), and R. capensis (5%). There was evidence of recent pole-size stem harvesting and historic canopy tree logging in the plots. In total, 136 tree stumps were measured. The most commonly harvested tree species in the plots included X. monospora, E. zuluensis and P. henkelii, which represented a harvest rate of 19%, 15%, and 12%, respectively, of all standing individuals of these species. The physiognomy of iGxalingenwa was characterised by dense cover in the understorey, with decreasing cover in the higher strata (Fig. 2A). Although the cover in the highest strata (>20 m) was relatively low, the individual trees that occupied this emergent layer had a large basal area (Fig. 2B). The emergent layer comprised only six species: P. henkelii (53%), P. falcatus (19%), V. lanceolata (13%), Celtis africana (12%), C. capense (1%), and K. africana (1%). 52 co p y M.J. Lawes et al. / Forest Ecology and Management 247 (2007) 48–60 on On a forest-wide basis, the community composition differed significantly between the seedling and sapling strata (Mantel test; r = 0.20, P = 0.013), but there were no differences between the seedling and tree strata (r = 0.06, P = 0.275) or between the sapling and tree strata (r = 0.09, P = 0.199). or 's pe rs Plot location relative to the forest edge had no significant influence on the number and species richness of trees, saplings, seedlings, and vines, as well as the percent grass and herb cover, number and species of tree and sapling stumps, canopy height, vertical heterogeneity, and horizontal heterogeneity at the understorey and canopy levels (MANOVA; Wilks’ l = 0.08, F 36,20 = 1.47, P = 0.182). Only vertical heterogeneity was significantly different among locations, with greater heterogeneity at the forest edge (Table 1). Community composition differed from edge to interior for both trees (MRPP; T = 3.06, A = 0.042, P = 0.009) and seedlings (T = 2.08, A = 0.027, P = 0.040). In particular, intermediate and forest interior plots contained a higher average number of P. falcatus and P. henkelii compared with forest edge plots. The sapling community composition showed no difference among plots (T = 0.69, A = 0.006, P = 0.743). al Fig. 2. Vertical profile of iGxalingenwa forest at different height classes, showing (A) the mean percent cover of vegetation in each height class averaged from 30 sample plots and (B) the mean basal area of individuals occupying the height class. 3.2. Forest species population structure Of the tree species, P. henkelli, Celtis africana, X. monospora, and E. zuluensis had the highest importance values (Table 2). Based on SCD slope values, the 20 most important species were classified into four groups (Fig. 3) of which three were variants of the Type I distribution (see Section 2). The first group (Type IA) included P. henkelii, X. monospora, V. lanceolata, and Ptaeroxylon obliquum, all of which had slopes between 0.06 and 0.2. These species had Table 1 Number of species and abundances of woody species (1 S.E.) comparing plots from the forest edge, at an intermediate distance from the edge, and in the forest interior Au th Number of trees Number of tree species Number of saplings Number of sapling species Number of seedlings Number of seedling species Number of vines Number of vine species Mean % cover of grass Mean % cover of herbs Number of harvested trees Number of harvested tree species Number of harvested saplings Number of harvested sapling species Mean canopy tree height (m) Vertical heterogeneity (CV) Horizontal heterogeneity (canopy) (CV) Horizontal heterogeneity (undestorey) (CV) Edge Intermediate Interior F P 22.0  2.99 9.8  1.12 12.6  1.63 8.2  0.74 42.4  5.99 11.6  1.10 2.1  0.57 1.0  0.33 0.9  0.17 2.0  0.30 4.4  1.10 2.2  0.44 1.5  0.69 0.9  0.48 13.2  0.95 73.6  6.19 54.3  5.59 79.9  10.23 29.2  3.03 9.6  0.50 13.0  2.26 6.3  0.96 25.3  4.17 8.4  1.20 2.9  0.62 1.0  0.26 0.6  0.10 2.0  0.25 4.2  0.98 1.7  0.33 1.1  0.41 0.6  0.31 11.9  0.64 59.4  4.36 54.3  4.68 59.7  5.31 29.0  2.91 8.0  0.60 13.3  1.24 7.6  0.48 22.6  3.90 10.2  0.95 2.4  0.85 1.1  0.31 1.2  0.21 2.2  0.23 5.0  1.56 2.0  0.49 0.5  0.27 0.3  0.15 14.5  0.88 53.1  4.49 53.5  3.44 53.8  6.31 1.90 1.56 0.04 1.67 5.05 2.20 0.34 0.04 2.41 0.27 0.11 0.34 1.07 0.77 2.40 4.30 0.01 3.27 0.169 0.228 0.961 0.206 0.014 0.131 0.712 0.965 0.109 0.769 0.894 0.712 0.357 0.471 0.110 0.024 0.991 0.054 Data were analysed using MANOVA (Wilks’ l = 0.08, F36,20 = 1.47, P = 0.182). 53 M.J. Lawes et al. / Forest Ecology and Management 247 (2007) 48–60 Table 2 The total number of individuals sampled, the mean density and basal area (BA), and the importance value (IV) of the top twenty most important species in iGxalingenwa forest Total individuals Mean density (stems/ha) Mean BA (m2 ha1) IV Podocarpus henkelii Celtis africana Xymalos monospora Eugenia zuluensis Vepris lanceolata Kiggelaria africana Podocarpus falcatus Cussonia spicata Clausena anisata Gymnosporia harveyana Diospyros whyteana Halleria lucida Zanthoxylum davyi Zanthoxylum capense Scolopia mundii Ptaeroxylon obliquum Rothmannia capensis Cassinopsis ilicifolia Calodendum capense Pavetta lanceolata 129 105 87 132 47 35 45 45 18 15 17 13 14 9 6 5 9 7 4 3 5.38 3.62 3.78 5.74 2.35 1.75 2.81 2.81 2.00 1.67 2.13 1.63 2.00 1.50 1.20 1.00 2.25 1.75 1.33 1.00 23.20 4.07 4.23 1.13 2.87 1.01 6.75 0.52 0.09 0.12 0.10 0.13 0.15 0.12 0.51 0.36 0.20 0.04 0.28 0.34 144.02 118.2 96.26 95.46 78.45 73.12 72.89 60.02 32.42 32.13 28.99 28.55 25.39 21.37 18.47 18.03 14.87 14.28 11.09 11.07 co p al on stems were straight. The majority of straight stems came from canopy and subcanopy trees such as Celtis africana (21%), P. henkelii (11%), R. capensis (11%), Cussonia spicata (11%), and Z. capense (8%). The understorey species D. whyteana and E. zuluensis had few straight stems. Evidence of harvesting of stems was found in all transects. We recorded 458 harvested pole-size stems from 28 species, which constitutes a harvesting rate of 36% (305 stems ha1) of all standing stems. Most harvested stems could be identified to species, but 19% of the pole-sized stems were unidentified. For the identified harvested pole-size stems, 83% were from eight species: E. zuluensis (19%), X. monospora (17%), Clausena anisata (13%), D. whyteana (11%), P. henkelii (10%), Celtis africana (6%), R. capensis (3%), and Z. capense (3%) (Table 3). We also recorded 50 harvested stems with diameters larger than 15 cm. These were identified as P. henkelii (32%), X. monospora (28%), Celtis africana (15%), E. zuluensis (9%), K. africana (4%) and V. lanceolata (4%), with 8% unidentified. Overall, 8.5% of all harvested stems were coppiced, indicating that not all harvested individuals die immediately (Table 4). There were differences in harvesting intensity according to stem size, with pole-sized stems between 2 and 8 cm harvested most intensely (Fig. 4). Among the eight most commonly harvested species, harvesting of pole-sized stems also varied according to species and size (Fig. 5). Overall, for these eight species, 37% of all available stems were harvested. In the case of P. henkelii, X. monospora, and D. whyteana, all available stems in the lower size classes were harvested. For all eight species in this group, the SCD resembled a humped-shaped Weibull distribution with higher mortality among the smaller size classes caused by harvesting for poles. Based on the decay status of stumps, stems of different sizes varied in their harvest history (x2 = 66.34, d.f. = 9, P < 0.001). Harvested small stems (2–5 cm) had an even number of stumps Au th or 's pe rs seedlings and saplings at low densities and discontinuous population structures, indicating instability in their populations. For P. henkelii there was a gap in the age structure that is possibly related to the logging of relatively large trees, but this shape of SCD is also thought to be characteristic of gap opportunist species (Seydack, 1995). A second group (Type IB) consisted of Celtis africana, K. africana, Halleria lucida, Cussonia spicata, Z. capense, R. capensis, Zanthoxylum davyi and E. zuluensis. The eight species in this group all had slopes between 0.2 and 0.8 and had inverse J-shape distributions. The last group (Type IC) of species included only typical understorey species, Pavetta lanceolata, G. harveyana, D. whyteana, Cassinopsis illicifolia, and Clausena anisata, which had slopes greater than 0.1. These species all had extremely high numbers of individuals in small size classes and, because they are all understorey species, a complete absence of individuals in larger size classes. For Pavetta lanceolata there also appeared to be a gap in the age structure where intermediate-sized individuals would be expected. The remaining group (Type II) comprised Scolopia mundii, P. falcatus, and C. capense, all of which had slopes approaching zero. For these three species, seedlings were absent and saplings were very rare. y Species 3.3. Harvest intensity Across 10 transects we measured 818 pole-size trees from 36 species. The pole-sized tree community was dominated by eight species, which together accounted for 65% of the polesized trees: E. zuluensis (32%), X. monospora (17%), Clausena anisata (15%), Celtis africana (12%), D. whyteana (11%), R. capensis (10%), G. harveyana (9%), and P. henkelii (9%). Of the pole-sized stems at iGxalingenwa, only 23% of standing 54 M.J. Lawes et al. / Forest Ecology and Management 247 (2007) 48–60 Diameter size class (cm) Stems ha1 Stumps ha1 Index Celtis africana 2–5 5–10 10–15 3 27 8 11 3 2 0.79 0.10 0.20 Clausena anisata 2–5 5–10 10–15 2 37 7 22 10 1 0.92 0.21 0.13 2–5 5–10 10–15 0 29 5 9 13 5 1.00 0.31 0.50 2–5 5–10 10–15 5 72 21 13 28 6 0.72 0.28 0.22 2–5 5–10 10–15 1 20 5 7 13 5 0.88 0.39 0.50 2–5 5–10 10–15 9 23 0 5 3 1 0.36 0.12 1.00 Xymalos monospora 2–5 5–10 10–15 0 39 11 11 25 5 1.00 0.39 0.31 Zanthoxylum capense 2–5 5–10 10–15 1 19 3 2 5 1 0.67 0.21 0.25 Diospyros whyteana Eugenia zuluensis al Podocarpus henkelii co p Species y Table 3 The harvesting indices for pole-size diameter classes from the eight most commonly harvested species at iGxalingenwa forest The index is expressed as a ratio of the number of stumps to the total number of available stems (stumps plus standing stems). in all decay classes, indicating that this stem size has been harvested in the past and continues to be harvested (Fig. 6). On the other hand, most stumps in the larger size classes had advanced decay, suggesting that larger stems were harvested in the past but are no longer harvested frequently. The best-fit GLM of harvest intensity included species, diameter size class, stem availability, and distance from nearest household (Table 3). Across all possible models, the most important variable affecting the number of harvested stems was stem availability (wi ¼ 1:00), followed closely by Au th or 's pe rs on Rothmania capensis Fig. 3. Diameter size-class distributions of the 20 most important species at iGxaligenwa forest arranged in order of increasing SCD slope values. The y-axis represents ln(individuals ha1 + 1) and the x-axis is the diameter size-class in 5 cm intervals from 0 to 185 cm. Fig. 4. Size-class distribution of all harvested stems recorded in 10 transects in the iGxalingenwa forest. 55 M.J. Lawes et al. / Forest Ecology and Management 247 (2007) 48–60 Table 4 Summary of fitted generalised linear models testing the relationship between harvesting intensity of pole-sized stems and ecological and social correlates Model formula Residual deviance d.f. Parameters AIC DAIC Null 1 2 3 4 5 6 7 Constant Species Diameter Species + Diameter Species + Diameter + Availability Species + Diameter + Availability + Distance Species + Diameter + Availability + Distance + Density Species + Diameter + Availability + Distance + Density + Paths 2156 2063 2079 1985 589.9 537.8 537.8 537.8 7199 7192 7197 7190 5269 5268 5267 5266 1 8 3 10 11 12 13 14 2158 2079 2085 2005 611.9 561.8 563.8 565.8 1596 1517 1523 1443 50.1 0 2 4 co p y Model Six factors were considered in the model: species identity (Species), diameter size class (Diameter), number of available stems of a particular species and size class (Availability), distance from nearest household (Distance), density of households within a 1 km radius (Density), and number of paths (Paths). Model selection used Akaike’s Information Criterion (AIC) and DAIC = difference in AIC units from the best-fit model (shown in bold type). species (wi ¼ 0:998), distance from nearest household (wi ¼ 0:997), and size class (wi ¼ 0:992). The density of households within a 1 km radius (wi ¼ 0:271) and the number of paths (wi ¼ 0:269) were less important predictors of harvest intensity. 4. Discussion Au th or 's pe rs on al The structure and community composition of iGxalingenwa forest has been significantly affected by logging activity a century ago and by the harvesting of pole-sized stems from the Fig. 5. Size-class distribution of the pole-size trees of the eight most commonly used species at iGxalingenwa forest. Living stems are represented by the black bars and stumps of harvested stems are represented by white bars. 56 M.J. Lawes et al. / Forest Ecology and Management 247 (2007) 48–60 South Africa (0.7 m3 ha1 y1; A.H.W. Seydack, personal communication; Lawes et al., 2004b). 4.2. Ongoing disturbance by local communities co p y As occurs elsewhere in Africa (Hall and Rodgers, 1986; Vermeulen et al., 1996; Luoga et al., 2002), there was a strong preference among present-day subsistence harvesters for polesized stems with a diameter range of 2–10 cm. In fact, the harvesting intensity of 36% (305 stems ha1) of available polesized stems is among the highest harvesting rates recorded for African forests (cf. Table 3 in Boudreau et al., 2005). Understorey species are typically not straight-stemmed, while canopy species are. Our data show that, in the past, canopy species were preferentially harvested from the understorey (e.g., Podocarpus spp., O. bullata and Ptaeroxylon obliquum). The lack of straight stems in the understorey at present has caused a shift in use toward many less preferred understorey species (e.g., E. zuluensis and D. whyteana). Canopy species with straight stems (e.g., P. henkelii, Celtis africana, R. capensis, and Z. capense) continue to be opportunistically harvested in proportion to their availability. Similar to other studies in Zimbabwe (Vermeulen et al., 1996) and South Africa (Shackleton et al., 1994; Boudreau et al., 2005), we found that wood harvesting was more intense closer to the boundaries of the forests but it was not affected by the density of households. We also detected a difference in the tree and seedling community structure between edge and interior plots, indicating that the forest may have been affected by current harvest practices. In a survey of local inhabitants, 63% of interviewed households (n = 60 households, 43% of all households in the region) acknowledged a change in species composition at iGxalingenwa due to forest use (Robertson and Lawes, 2005). on pe rs understorey in recent times. The legacy of colonial logging persists through the occupation of the emergent stratum by a cohort of Podocarpus spp. trees and by the presence of a dense angiosperm-dominated understorey stratum. Together, the dense understorey and the harvest of pole-sized stems have suppressed the establishment of canopy tree species, leading to few individuals in the mid-canopy stratum. However, the most important threats to iGxalingenwa forest are the ongoing intensive and relatively indiscriminate harvest of pole-sized stems and the harvest and/or senescence of trees in the canopy and emergent strata. al Fig. 6. Relationship between harvested stem size and decay stage at iGxalingenwa forest. Decay classes were defined as follows: new = wood and bark fresh and very hard; moderately old = wood darkened and first signs of decay present; old = wood dark and obvious signs of decay; very old = wood rotten and stump easily broken. 4.1. Past colonial disturbance Au th or 's In 1889, Fourcade (p. 7) described iGxalingenwa as ‘‘a truly magnificent forest; indeed for its size, the finest that I have seen in South Africa, and the nearest approach to a pure forest of yellowwood [Podocarpus spp.]. . . the quantity of timber felled in it has been small.’’ Although logging of mainly Podocarpus spp. took place during a limited time (1905–1906; King, 1941), several lines of evidence suggest that it was intensive. Numerous saw-pits and skid trails occur throughout iGxalingenwa, demonstrating that the effects of the logging operation extended beyond the immediate vicinity of logged trees. In addition, frequent canopy gaps (250–350 m2) still exist where large trees were extracted and are dominated by thicket understorey. Podocarpus spp. dominate the emergent stratum of iGxalingenwa, potentially indicating a high-intensity disturbance in the past that opened up the canopy and allowed smaller trees in the advanced regeneration stage to grow into the gaps (Ogden and Stewart, 1995; Urlich et al., 2005). Such a situation is plausible because during 1905 and 1906 a total yield of 1274 m3 or >3.78 m3 ha1 y1 (King, 1941) of Podocarpus spp. timber was harvested from the southern and western regions of the forest (i.e., the regions we sampled). This yield is greater than both the recommended sustained yield of 1.44 m3 ha1 y1 at the time (Hutchins, 1905) and the currently implemented yield in the southern Afrotemperate forests of 4.3. Forest response to disturbance The physiognomy of iGxalingenwa forest is characterised by its almost continuous, thicket-like understorey, an emergent stratum dominated by conifers (Podocarpus spp.), and a sparse canopy (Lawes et al., 2006). This structure can be attributed, in part, to logging disturbance during the colonial era. Other Afromontane forests subjected to intensive logging in the past also have a discontinuous canopy with dense understorey forming in gaps (Babaasa et al., 2004; Hitimana et al., 2004). In mixed conifer-angiosperm forest in New Zealand, which is structurally very similar to the Afrotemperate forests of South Africa, conifer recruitment is episodic and matched to large disturbances (Wells et al., 2001). As with iGxalingenwa, this causes even-aged patches (>1000 m2) where large emergent conifers are dominant (Ogden and Stewart, 1995; Urlich et al., 2005). In addition to logging or disturbance effects, the internal dynamics and the environmental conditions of the forest are important in determining species composition and forest vertical structure. Podocarpus spp. dominate the canopy and emergent strata even in unlogged forests of the Western Cape of 57 M.J. Lawes et al. / Forest Ecology and Management 247 (2007) 48–60 on al co p y traditional medicines (Nomtshongwana, 1999). In this species, there were many seedlings and saplings but only scattered large individuals. As Ptaeroxylon obliquum does not have a history of logging during the colonial era, this discontinuous population structure must be due to current harvesting pressure. Most stems are gathered at the pole-size stem stage and therefore few individuals make it into larger size classes. While the dense angiosperm-dominated stratum may have suppressed the regeneration of some canopy tree species, with the combined effects of selective harvesting of straight-stemmed individuals of canopy species, few individuals have entered the mid-strata, causing the cathedral-like physiognomy of the forest (Lawes et al., 2006). Perhaps one of the most striking trends from our survey of iGxalingenwa is the complete absence of O. bullata in the forest. This species is highly valued and has been heavily exploited for the muthi trade (Cunningham, 1988; Mander, 1998) and for lumber (Moll, 1972). There are records of O. bullata being quite common in iGxalingenwa in the past (Fourcade, 1889). In the mid-1980s, it was suggested that the forest needed to be protected from illegal harvesting because this species was declining (Cooper, 1985). There is anecdotal evidence that O. bullata might have disappeared from forest by the end of the 1990s (Nomtshongwana, 1999). The implication from this study is that the species has, indeed, been harvested to local extinction. Au th or 's pe rs South Africa (Seydack et al., 1995; Midgley et al., 2002). Conifer physiology, particularly their hydraulics, predisposes them to dominate the canopy and emergent strata in resourcepoor environments, such as in high altitude and cold regions, because under these conditions, established conifers have a competitive advantage over broadleaved angiosperms (Bond, 1989; Coomes et al., 2005). Furthermore, conifer-dominated mixed forests allow more light to the forest floor (5% light transmission) than angiosperm-dominated forests (1%) (McDonald and Norton, 1992; Coomes et al., 2005), and subcanopy conditions in conifer-dominated forests are more favourable for recruits of faster-growing angiosperm than gymnosperm species (Bond, 1989). Podocarpus spp. are shadetolerant, gap opportunist species (Midgley et al., 1995; Fetene and Feleke, 2001) that establish where forest openings are small or otherwise unsuitable for rapid filling by fast-growing angiosperm trees (Bond, 1989). Thus, shade-tolerant conifers are likely to dominate temperate forests with small gaps sizes and where large disturbances are infrequent (Bond, 1989; Midgley et al., 1995). Given Fourcade’s (1889) description of a Podocarpus spp. dominated forest at iGxalingenwa, there is little doubt that the subsequent logging of large trees facilitated the establishment of the dense angiosperm-dominated understorey (cf. Hitimana et al., 2004; Coomes et al., 2005). The understorey at iGxalingenwa currently comprises shade-tolerant understorey specialists with few individuals from canopy species. In logged forests it is often the case that both early and late secondary species colonise logging gaps, although light-demanding species tend to predominate in the early stages (Nagaraja et al., 2005), giving way to climax shade-tolerant species later (Dekker and de Graff, 2003). The latter implies that at iGxalingenwa the dense understorey has persisted since the logging event. The infrequent occurrence or complete absence of some canopy species in the sapling layer may, in part, be explained by their exclusion by understorey species. Indeed, several canopy tree species had a discontinuous population structure, suggesting that they may have declining populations. For both C. capense and P. falcatus neither had any individuals in the smallest size classes. This could be because these species depend on the formation of small gaps for establishment of seedlings to take place (Bond, 1989) or because P. falcatus seeds have a high rate of parasitism by insects (personal observations). More likely, the unstable population structure reflects a disruption in their reproductive cycle caused by selective harvesting. For instance, P. falcatus, P. henkelii, Ptaeroxylon obliquum, and V. lanceolata are the most popular firewood and pole species at iGxalingenwa (Nomtshongwana, 1999; J. Robertson, unpublished data). We found that individuals of pole-size classes were completely absent for P. falcatus, suggesting that harvesters have destroyed the regeneration phase of this species. In addition, C. capense has been exploited in this forest and debarking is especially common (Nomtshongwana, 1999), which could explain the absence of mid- to large size classes. Another species of concern is Ptaeroxylon obliquum, which is highly sought after both for poles and as a source for 4.4. Future changes in the forest structure Many of the canopy species at iGxalingenwa have stable or expanding population structures, including Celtis africana, K. africana, Z. capense, R. capensis, and Z. davyi. These species all have a large number of small stems relative to larger stems, and therefore do not seem to be under a threat of local extinction. In fact, Celtis africana has many individuals in lower size classes and is a common species in all strata of the forest. Because of these factors and the fact that this deciduous pioneer species responds positively to disturbance, Celtis africana is well placed to eventually become a dominant species of the forest. Some understorey species, particularly E. zuluensis, G. harveyana, D. whyteana, C. ilicifolia, and Clausena anisata, also appear to have stable population structures. However, a closer analysis of the eight most commonly harvested species revealed that there were few stems available with diameter between 2–5 cm and, for many species, all available stems were harvested in these size classes. This gap in the population structure could lead to regeneration failure or the promotion of an alternative successional pathway in the future. A sign that regeneration failure is occurring at iGxalingenwa is the difference in community composition among the forest strata. There was no difference in community composition between the seedling and tree layers, which indicates that mature trees are producing propagules that successfully establish on the forest floor. Similarly, there was no difference between the sapling and tree layers, indicating that in the past, immature trees from canopy species have successfully been 58 M.J. Lawes et al. / Forest Ecology and Management 247 (2007) 48–60 co p y Lawes, 2005) to assist in controlling the use of forest resources; (2) increase the value of the forest to local communities through ecotourism (it is a favoured birdwatching site); (3) upgrade the several nearby woodlots (Acacia mearnsii) that are in poor condition (Fig. 1) to reduce harvesting pressure on the forest; (3) provision of unusable commercial timber (fire-damaged, fungal attack) by forestry companies in the region for building and fuelwood (a pilot study has already been conducted); (4) thin out the vine drape in the understorey thickets to release canopy species; (5) use enrichment plantings (Ashton et al., 2001) to restore canopy species most affected by harvesting, particularly the Podocarpus species; (6) promote an ongoing program of ecological monitoring and research to support this adaptive management framework. Acknowledgements on al We thank Stefanie Pollock, John Robertson and Harriet Eeley for assistance in the field, and Pete and Bev Everett for providing accommodation. This study was made possible by financial support from the National Research Foundation of South Africa (Conservation and Management of Ecosystems and Biodiversity) under grant number 2069339. Any opinion, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the NRF. Postdoctoral funding for SB was provided by FQRNT (Québec Government). We are grateful to Ezemvelo KwaZulu-Natal Wildlife for allowing research in this forest and the Mazda Wildlife Fund for logistic support. or 's pe rs recruited into the understorey. However, there was a difference in community composition between the seedling and sapling layers, suggesting that, although seedlings are establishing they are not entering the sapling layer. The failure of these recruits to grow into the understorey is most likely due to harvesting pressure. Further concerns about the perpetuation of the regeneration cycle at iGxalingenwa relate to the emergent Podocarpus spp. This stratum is showing signs of senescence, which in time may have significant consequences for forest physiognomy and species composition (Mueller-Dombois, 1986). Though not strictly even-aged, the emergent conifers form an ecological cohort and are at least 150 years old (unpublished dendrochronology data). These large Podocarpus spp. are susceptible to mortality effects such as windthrow and crown-snapping (unpublished data), even if they are not of senescent age. In the absence of recruitment into the subcanopy and canopy strata, coupled with the intensive harvesting of the smaller size classes of P. henkelii and other canopy species, even protracted emergent collapse will promote a denser understorey, which in turn will further suppress advanced regeneration of canopy tree species. We found a higher vertical heterogeneity at the forest edge and detected differences in the community composition associated with location relative to the forest edge, suggesting that there is already disruption to the forest margins as predicted above. Thus, unlike the situation prior to logging when this forest was likely a self-perpetuating podocarp forest (sensu Ogden and Stewart, 1995), it now appears to be set on an alternative successional pathway that is determined by the combined actions of a large-scale, albeit brief, period of logging and subsistence harvesting of canopy species for poles. The ultimate outcome, probably in 100 to 200 years from now, will likely be a forest of low stature, thicket-like in vertical structure, and dominated by shade-tolerant understorey and subcanopy species 5. Conservation implications and management recommendations Au th In the absence of advanced regeneration to replace them, removal of canopy and emergent trees will perpetuate the thicket-like understorey and significantly alter the composition and vertical structure of the tree community (Lawes et al., 2006). The careful conservation of iGxalingenwa forest depends on our understanding of the consequences of disturbance in the distant past to forest structure and development, the dynamics of species responses to disturbance, and on the management of current pole harvesting practices to ensure the release of canopy species and forest recovery. Our data suggest that this forest cannot sustain much more harvesting pressure on the pole size-class. We recommend that management interventions focus on reducing and even prohibiting the harvest of understorey trees and pole-sized stems in general. Important management interventions are: (1) active enforcement by Ezemvelo KwaZulu-Natal Wildlife of current harvesting laws and the development of a community based forest resources management institution (Robertson and References Ashton, M.S., Gunatilleke, C.V.S., Singhakumara, B.M.P., Gunatilleke, I.A.U.N., 2001. Restoration pathways for rain forest in southwest Sri Lanka: a review of concepts and models. For. Ecol. Manage. 154, 409–430. Babaasa, D., Eilu, G., Kasangaki, A., Bitariho, R., McNeilage, A., 2004. Gap characteristics and regeneration in Bwindi Impenetrable National Park, Uganda. Afric. J. Ecol. 42, 217–224. Bond, W.J., 1989. The tortoise and the hare: ecology of angiosperm dominance and gymnosperm persistence. Biol. J. Linn. Soc. 36, 227–249. Boudreau, S., Lawes, M.J., 2005. Small understorey gaps created by subsistence harvesters do not adversely affect the maintenance of tree diversity in a subtropical forest. Biol. Conserv. 126, 279–286. Boudreau, S., Lawes, M.J., Piper, S.E., Phadima, L.J., 2005. Subsistence harvesting of pole-size understorey species from Ongoye Forest Reserve, South Africa: species preference, harvest intensity, and social correlates. For. Ecol. Manage. 216, 149–165. Bowman, D.M.J.S., 2001. Future eating and country keeping: what role has environmental history in the managment of biodiversity? J. Biogeogr. 28, 549–564. Brokaw, N.V.L., Scheiner, S.M., 1989. Species composition in gaps and structure of a tropical forest. Ecology 70, 538–541. Burnham, K.P., Anderson, D.R., 2002. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach. Springer-Verlag, New York. Condit, R., Sukumar, R., Hubbell, S.P., Foster, R.B., 1998. Predicting population trends from size distributions: a direct test in a tropical tree community. Am. Nat. 152, 495–509. Coomes, D.A., Duncan, R.P., Allen, R.B., Truscott, J., 2003. Disturbances prevent stem size-density distributions in natural forests from following scaling relationships. Ecol. Lett. 6, 980–989. 59 M.J. Lawes et al. / Forest Ecology and Management 247 (2007) 48–60 on al co p y Mander, M., 1998. The Marketing of Indigenous Medicinal Plants in South Africa: A Case Study in KwaZulu-Natal. FAO, Rome. McCracken, D.P., 1986. The indigenous forests of colonial Natal and Zululand. Natalia 16, 19–38. McCracken, D.P., 1987. Qudeni: the early commercial exploitation of an indigenous Zululand forest. S. Afric. For. J. 142, 71–80. McCullagh, P., Nelder, J.A., 1989. Generalized Linear Models. Chapman and Hall, London. McCune, B., Mefford, M.J., 1999. PC-ORD. Multivariate Analysis of Ecological Data. Version 4.35. MjM Software Design, Gleneden Beach, Oregon. McDonald, D., Norton, D.A., 1992. Light environments in temperate NewZealand podocarp rain-forests. N. Zeal. J. Ecol. 16, 15–22. Midgley, J.J., Bond, W.J., Geldenhuys, C.J., 1995. The ecology of southern African conifers. In: Enright, N.J., Hill, R.S. (Eds.), Ecology of the Southern Conifers. Melbourne University Press, Carlton, pp. 64–80. Midgley, J.J., Parker, R., Laurie, H., Seydack, A., 2002. Competition among canopy trees in indigenous forests: an analysis of the ‘additive basal area’ phenomenon. Aust. Ecol. 27, 269–272. Moll, E.J., 1972. The current status of mistbelt mixed Podocarpus forest in Natal. Bothalia 10, 595–598. Mueller-Dombois, D., Ellenberg, H., 1974. Aims and Methods of Vegetation Ecology. John Wiley and Sons, New York. Mueller-Dombois, D., 1986. Perspectives for an etiology of stand-level dieback. Annu. Rev. Ecol. Syst. 17, 221–243. Nagaraja, B.C., Somashekar, R., Raj, M.B., 2005. Tree species diversity and composition in logged and unlogged rainforest of Kudremukh National Park, South India. J. Environ. Biol. 26, 627–634. Newbery, D.M., Kennedy, D.N., Petol, G.H., Madani, L., Ridsdale, C.E., 1999. Primary forest dynamics in lowland dipterocarp forest at Danum Valley, Sabah, Malaysia, and the role of the understorey. Philos. Trans. R. Soc. Lond. B: Biol. Sci. 354, 1763–1782. Nomtshongwana, N.N., 1999. Indigenous plant use in Gxalingenwa and KwaYili forests in the southern Drakensburg, KwaZulu-Natal. MSc Thesis. University of Natal, Pietermaritzburg. Obiri, J., Lawes, M., Mukolwe, M., 2002. The dynamics and sustainable use of high-value tree species of the coastal Pondoland forests of the Eastern Cape Province, South Africa. For. Ecol. Manage. 166 1331–1148. Ogden, J., Stewart, G.H., 1995. Community dynamics of New Zealand conifers. In: Enright, N.J., Hill, R.S. (Eds.), Ecology of the Southern Conifers. Melbourne University Press, Carlton, pp. 81–119. Robertson, J., Lawes, M.J., 2005. User perceptions of conservation and participatory management of iGxalingenwa forest, South Africa. Environ. Conserv. 32, 64–75. Roth, R.R., 1976. Spatial heterogeneity and bird species diversity. Ecology 57, 773–782. Schulze, R.E., Maharaj, M., Lynch, S.D., Howe, B.J., Melvil-Thomson, B., South African Atlas of Agrohydrology and Climatology, Water Research Commission, Pretoria, 1997. Sekercioglu, C.H., 2002. Effects of forestry practices on vegetation structure and bird community of Kibale National Park, Uganda. Biol. Conserv. 107, 229–240. Seydack, A.H.W., 1995. An unconventional approach to timber yield regulation for multi-aged, multispecies forests. I. Fundamental considerations. For. Ecol. Manage. 77, 139–153. Seydack, A.H.W., Vermeulen, W.J., Heyns, H.E., Durrheim, G.P., Vermeulen, C., Willems, D., Ferguson, M.A., Huisamen, J., Roth, J., 1995. An unconventional approach to timber yield regulation for multi-aged, multispecies forests. II. Application to a South African forest. For. Ecol. Manage. 77, 155–168. Shackleton, C.M., Griffin, N.J., Banks, D.I., Mavrandonis, J.M., Shackleton, S.E., 1994. Community structure and species composition along a disturbance gradient in a communally managed South African savanna. Vegetatio 115, 157–167. SPSS, 2002. SPSS for Windows, Version 11.5.0. SPSS Inc., Chicago. Stohlgren, T.J., Falkner, M.B., Schell, L.D., 1995. A modified-Whittaker nested vegetation sampling method. Vegetatio 117, 113–121. Struhsaker, T.T., 1997. Ecology of an African Rain Forest: Logging in Kibale and the Conflict Between Conservation and Exploitation. University of Florida Press, Gainesville, FL. Au th or 's pe rs Coomes, D.A., Allen, R.B., Bentley, W.A., Burrows, L.E., Canham, C.D., Fagan, L., Forsyth, D.M., Gaxiola-Alcantar, A., Parfitt, R.L., Ruscoe, W.A., Wardle, D.A., Wilson, D.J., Wright, E.F., 2005. The hare, the tortoise and the crocodile: the ecology of angiosperm dominance, conifer persistence and fern filtering. J. Ecol. 93, 918–935. Cooper, K.H., 1985. The Conservation Status of Indigenous Forests in Transvaal, Natal and O.F.S. South Africa. Wildlife Society of South Africa, Durban. Cunningham, A.B., 1988. Over-exploitation of medicinal plants in Natal/ KwaZulu: root causes. Veld Flora 74, 85–87. Dekker, M., de Graff, N.R., 2003. Pioneer and climax tree regeneration following selective logging with silviculture in Suriname. For. Ecol. Manage. 172, 183–190. Downs, C.T., 2005. Abundance of the endangered Cape parrot, Poicephalus robustus, in South Africa: implications for its survival. Afric. Zool. 40, 15– 24. Echeverria, C., Coomes, D.A., Salas, J., Rey-Benayas, J.M., Lara, A., Newton, A., 2006. Rapid deforestation and fragmentation of Chilean temperate forests. Biol. Conserv. 130, 481–494. Everard, D.A., van Wyk, G.F., Midgley, J.J., 1994. Disturbance and the diversity of forests in Natal South Africa: lessons for their utilization. In: Huntley, B.J. (Ed.), Strelitzia 1. Botanical Diversity in Southern Africa. National Botanical Institute, Pretoria, pp. 275–285. Fetene, M., Feleke, Y., 2001. Growth and photosynthesis of seedlings of four tree species from a dry tropical Afromontane forest. J. Trop. Ecol. 17, 269– 283. Fourcade, H.G., 1889. Report on the Natal forests. Watson, Pietermaritzburg, South Africa. Hall, J.B., Rodgers, W.A., 1986. Pole cutting pressure in Tanzanian forest. For. Ecol. Manage. 14, 133–140. Hitimana, J., Kiyiapi, J.L., Njunge, J.T., 2004. Forest structure characteristics in disturbed and undisturbed sites of Mt. Elgon Moist Lower Montane Forest, western Kenya. For. Ecol. Manage. 194, 269–291. Hutchins, D.E., 1905. The indigenous forests of South Africa. Rep. Br. S. Afric. Assoc. Adv. Sci. 1, 319–334. King, N.L., 1941. The exploitation of the indigenous forests of South Africa. J. S. Afric. For. Assoc. 6, 26–48. Kotze, D.J., Lawes, M.J., 2007. Viability of ecological processes in small Afromontane forest patches in South Africa. Aust. Ecol. 32, 294–304. Krüger, S.C., Lawes, M.J., 1997. Edge effects at an induced forest-grassland boundary: forest birds in the Ongoye Forest Reserve, KwaZulu-Natal. S. Afric. J. Zool. 32, 82–91. LaFrankie, J.V., Ashton, P.S., Chuyong, G.B., Co, L., Condit, R., Davies, S.J., Foster, R., Hubbell, S.P., Kenfack, D., Lagunzad, D., Losos, E.C., Nor, N.S.M., Tan, S., Thomas, D.W., Valencia, R., Villa, G., 2006. Contrasting structure and composition of the understory in species-rich tropical rain forests. Ecology 87, 2298–2305. Lawes Agricultural Trust, 2005. GenStat 8.1. VSN International Ltd., Hemel Hempstead. Lawes, M.J., Macfarlane, D.M., Eeley, H.A.C., 2004a. Forest landscape pattern in the KwaZulu-Natal midlands South Africa: 50 years of change or stasis? Aust. Ecol. 29, 613–623. Lawes, M.J., Midgley, J.J., Chapman, C.A., 2004b. South Africa’s forests: the ecology and sustainable use of indigenous timber resources. In: Lawes, M.J., Eeley, H.A.C., Shackleton, C., Geach, B. (Eds.), Indigenous Forests and Woodlands in South Africa: Policy, People and Practice. University of KwaZulu-Natal Press, Pietermaritzburg, pp. 31–75. Lawes, M.J., Midgley, J.J., Boudreau, S., Griffiths, M.E., 2006. Lack of disturbance as an explanation for the additive basal area phenomenon in a stratified Afrotemperate forest. Aust. Ecol. 31, 471–477. Luoga, E.J., Witkowski, E.T.F., Balkwill, K., 2000. Subistence use of wood products and shifting cultivation within a miombo woodland of eastern Tanzania, with some notes on commercial uses. S. Afric. J. Bot. 66, 72–85. Luoga, E.J., Witkowski, E.T.F., Balkwill, K., 2002. Harvested and standing wood stocks in protected and communal miombo woodlots of eastern Tanzania. For. Ecol. Manage. 164, 15–30. Lykke, A.M., 1998. Assessment of species composition change in savanna vegetation by means of woody plants’ size class distributions and local information. Biodivers. Conserv. 7, 1261–1275. 60 M.J. Lawes et al. / Forest Ecology and Management 247 (2007) 48–60 Walker, B.H., 1976. An approach to the monitoring of changes in the composition and utilisation of woodland and savanna vegetation. S. Afric. J. Wildl. Res. 6, 1–32. Wells, A., Duncan, R.P., Stewart, G.H., 2001. Forest dynamics in Westland, New Zealand: the importance of large, infrequent earthquake-induced disturbance. J. Ecol. 89, 1006–1018. Wirminghaus, J.O., Downs, C.T., Symes, C.T., Perrin, M.R., 1999. Conservation of the Cape parrot in southern Africa. S. Afric. J. Wildl. Res. 29, 118– 129. Au th or 's pe rs on al co p y Urlich, S.C., Stewart, G.H., Duncan, R.P., Almond, P.C., 2005. Tree regeneration in a New Zealand rain forest influenced by disturbance and drainage interactions. J. Veg. Sci. 16, 423–432. Vermeulen, S.J., 1996. Cutting of trees by local residents in a communal area and an adjacent state forest in Zimbabwe. For. Ecol. Manage. 81, 101– 111. Vermeulen, S.J., Campbell, B.M., Matzke, G.E., 1996. The consumption of wood by rural households in Gokwe Communal Area, Zimbabwe. Hum. Ecol. 24, 479–491.