Tatenda Gotore·Hilton G.T.Ndagurwa·Shakkie Kativu·Denis Gautier·Laurent Gazull

Abstract Miombo woodlands near human settlements are under significant pressures from human activities,with negative consequences on their structure and composition.As studies are limited,we assessed the structure and species composition of a portion of miombo woodland along an increasing disturbance gradient from a national park,through a buffer zone to communal lands in northeast Zimbabwe.Five concentric plots of 2 and 11 m radii were established in each area to record woody species composition,diameter,height,basal area,density and volume as well as evidence of disturbance.Effects of site,growth stage and their interaction on vegetation diversity and structural parameters were tested using a general linear model (GLM).Principal component analysis (PCA) tested the association between species and site and ANOVA the differences in the level of disturbance across strata.Species diversity did not differ between sites except for evenness,which increased with disturbance.Evenness and richness were greater in seedlings and saplings than mature trees across sites,respectively.Sapling and mature tree diameters differed significantly between sites.Volume and density of mature trees declined with increasing disturbance while seedling densities peaked at intermediate disturbance levels (buffer zone).Tree harvesting,was more evident in the buffer zone and in the communal area relative to the national park.In contrast,f ire frequency was greater in the national park and in the buffer zone relative to the communal area.The results of this study identify a true miombo woodland dominated by Brachystegia boehmii with a stable population,as illustrated by an inverse-J shape in diameter class distribution on all sites,and that these woodlands are generally resilient to disturbances,maintaining similar species composition and structure at various levels of disturbance.However,continued monitoring of disturbance levels and miombo woodland response is recommended to ensure sustainable utilisation of these resources.

Keywords Miombo woodlands·cComposition ·Diversity·Structure·Anthropogenic disturbances

Introduction

Miombo woodlands are floristically distinct from other African woodlands owing to the dominance of the generaBrachystegia,JulbernardiaandIsorbelinia(Dewees et al.2010;Timberlake et al.2010;Moura et al.2017).The biome has an estimated 8500 species of higher plants,over 54% of which are endemic (Frost 1996;Moura et al.2017).Their composition and structure appear relatively uniform with a partly closed canopy 10—20 m high,a discontinuous understory of broad-leaved shrubs and an often sparse but continuous herbaceous layer,while the density of woody plants varies between 1500 and 4100 stems ha−1(Frost 1996).These woodlands are important for human livelihoods in southern Africa (Geldenhuys 2010;Kalaba et al.2013;Djoudi et al.2015).However,they are subject to anthropogenic disturbances with consequences for their structure,composition and function.

Anthropogenic disturbances through agricultural expansion,fuel wood extraction and fire are negatively impacting these woodlands (Banda et al.2006;Geldenhuys 2010;Jew et al.2016).Several studies have noted that shifting cultivation and population growth drive the demand for agricultural land,which has caused the disturbance of large miombo areas (Jew et al.2016;Moura et al.2017).Consequently,this has led to forest degradation,biodiversity loss,and the decline in their ecological functions,including carbon storage,soil conservation and the provision of food and materials for local communities (Chirwa et al.2014;Jew et al.2016;Gumbo et al.2018).Although cleared areas and abandoned agricultural lands regenerate over time (Syampungani et al.2015;Zulu et al.2019),important species of the woodlands are lost and often replaced with invasive species which also impacts woodland structure and composition (Foxcroftet al.2010;Masocha et al.2011).Research has reported the response to disturbance in these woodlands (Geldenhuys 2010;Syampungani et al.2015,2017;Jew et al.2016;Muposhi et al.2016;Ryan et al.2016).However,similar studies are lacking in Zimbabwe,where miombo woodlands are extensive and important contributors to livelihoods and biodiversity conservation (Forestry Commission 2011).

In Zimbabwe,miombo woodlands are found in almost every land use category,national parks,communal and resettlement areas,and commercial farmlands (Forestry Commission 2011).In national park areas,the Parks and Wildlife Act of 1975 restrict access to forest resources,hence,due to the absence or limited human disturbance,national parks are considered to be the most appropriate areas for conserving plant diversity and maintaining intact forest composition and structure (Banda et al.2006).Conversely,in communal areas,restrictions are relaxed and communities are allowed to access forest resources according to the Communal Lands Forest Produce Act,1988.As a result,where communal areas are adjacent to national parks and their buffer areas,disturbance increases with distance from the protected area to communal settlements,thereby creating a disturbance gradient (Muposhi et al.2016).Vegetation characteristics are likely to change along the gradient (Banda et al.2 006;Chinuwo et al.2010;Muposhi et al.2016).However,studies on the effects of disturbance gradients on the characteristics of miombo woodlands remain limited.

The structure and composition of miombo woodland was studied along a disturbance gradient in Chundu communal lands,northeast Zimbabwe,at the interface with Mana Pools National Park.The objectives were to determine:(a) woodland composition;and,(b) structure along a disturbance gradient by surveying the vegetation from a protected area(the national park),through the buffer zone to a communal area.It was hypothesised that miombo woodlands do not vary significantly in composition and structure along the disturbance gradient.

Methods and materials

Study area

The research was carried out at the interface of the Chundu communal area,Ward 8 of the Hurungwe District,and Mana Pools National Park,a protected wildlife area about 260 km west of Harare (Fig.1).This area represents a gradient from the national park where there is no access to forest resources,to the communal area with complete but controlled resource access.

Fig.1 Map of Zimbabwe showing the study area

The area has a mean annual rainfall from 750 to 1000 mm between mid-November and the end of March (Anderson et al.1993).Soils vary from clay loam to black clay soils under typical miombo woodland vegetation dominated byBrachystegiaandJulbernardiaspecies (Chivuraise et al.2016).Subsistence agriculture is the primary source of livelihood,with maize,groundnuts,cotton and tobacco the major crops,and cattle,goats and sheep the major livestock(Ncube 2011).

Most of the inhabitants were resettled by the Government of Rhodesia from the Zambezi valley for the creation of Mana Pools National Park.The first inhabitants were resettled around the Chitindiva area in the 1970s,some 20 km from the park boundary (Dzingirai and Mangwanya 2015).Due to population growth and immigration from the southern parts of Zimbabwe,settlements have increased and expanded beyond the 5 km park buffer zone to about a km from the park boundary.Residents of the buffer zone are considered illegal settlers.The buffer zone in the past has been used for hunting concessions granted to the Rural District Council (RDC) through the Communal Areas Management Programme for Indigenous Resources (CAMPFIRE),and there are plans to revive these activities.The population of Chundu has grown from over 9053 people and more than 1882 households in 2002 to almost 15,388 people in nearly 3293 households in 2012,with an average household of 4.7 people (ZimStat 2012).

Field layout and data collection

Stratified random sampling was used to assess the structure and composition of the woodlands.In Zimbabwe,the general land use demarcations require a buffer zone between protected areas and communal lands.This is a peripheral area inside or outside a protected area aimed at enhancing the positive impacts of conservation and reducing the negative impacts on neighbouring communities (Ebregt and Greve 2000).Thus,a buffer zone,usually 3-km wide,separates the protected area from the surrounding communal lands.Given that anthropogenic disturbances and plant community characteristics differ markedly between these zones (Muposhi et al.2016),the study area was stratified into three strata:the national park,9802 ha;the buffer zone (2.5 km from the park boundary,11,741 ha);and,the communal area (5 km from the park boundary,9799 ha) (Fig.1).

Five concentric plots with radii of 2-m and 11-m were randomly established in each zone to assess disturbance and vegetation.The inner 2-m plot was used for assessing species and densities of tree seedlings,while the outer 11-m plot was used for assessing trees greater than 2-cm diameter.Data on species,height,diameters at breast height(DBH),and evidence of disturbance was collected in each plot.Height was measured using a clinometer and a diameter tape for measuring diameter.

Disturbance was assessed in each plot in each zone.This included the number of tree stumps at the time of measurement;the number of fires detected and forest loss over 15 and 17 years.MODIS standard fire location data from 2002 to 2017 were obtained from NASA’s Fire Information for Resource Management System (FIRMS) to determine the number of fires detected across sites over the 15-year period.Forest loss was determined from forest cover change maps developed from multi-temporal Landsat mosaic satellite images covering the period 2000 to 2017.Landsat has large historical datasets spanning more than 30 years.Supervised‘forest loss’ classification was carried out in Google Earth Engine and the classified forest loss map then exported to QGIS to generate deforestation statistics across study sites.Forest loss was assessed over the 17-year period,not yearly,and therefore annual loss was excluded in the statistical analysis.

Statistical analysis

Abundance was calculated in R version 3.4.3 as a mean of individual trees across sampling plots per site.Diversity indices were computed using PAST statistical software version 3.19 (Hammer et al.2001).Indices included Simpson’s dominance (Simpson 1949),species richness (number of species),and Shannon’s diversity index(Shannon and Weaver 1949),Buzas and Gibson (1969)evenness,equitability (Pielou 1966) and Fisher’s alpha(Fisher et al.1943).Species importance value index (IVI)was calculated using the following formulae (Curtis and McIntosh 1 951):

where,RD Ois relative dominance;BA sis basal area of a species;S BA/sum of basal area for all species.

where,R Dis relative density;D sis density of a species;S Dis sum of all species densities.

where,RFis relative frequency;Fsis frequency of a species;Sfis sum of all frequencies.

Basal area was calculated according to Loetsch et al.(1973):

where,Gis the basal area in m2ha−1,dbh is diameter at breast height in cm;volume (m3) was calculated using equations adopted from Abbot et al.(1997):

where,Vis volume (m3),anddbhdiameter at breast height(cm).Diameter size class distribution was constructed for all woody plants at 5 cm intervals.Height categorised species into seedlings (˂1 m),saplings (1—5 m) and mature trees(more than 5 m).A one-way ANOVA procedure in R version 3.4.3 tested for differences in disturbance,vegetation diversity and structural parameters between sites.A principal component analysis (PCA) in CANOCO version 5.1.1 was used to establish the association between species and sites.The effects of site,growth stage and their interactions on vegetation diversity and structural parameters were tested using a general linear model (GLM) in R version 3.4.3 (R core team 2017).

Results

Species composition

Twenty-three woody species of 19 genera and 10 families were recorded (Table 1).Brachystegia boehmiiandJulbernadia globif lorawere the most abundant on all sites,followed byDalbergiella nyasae,Dichrostachys cinereaandDiospyros mespiliformis.Ochna pulchra,Parinari curatellifoliaandPterocarpus angolensiswere only recorded in the national park whileBoscia salicifolia,Dalbergia nitidula,Monotes engleri,Pterocarpus rotundifoliusandUapaca nitidawere only found in the buffer zone.Acacia nigrescenswas recorded only in the communal area.

Brachystegia boehmiiwas the dominant species at all sites,with importance value indices of 44.7 in the national park,45.8 in the buffer zone and 43.1 in the communal area (Table 2).Co-dominant species wereJulbernadia globif lora,Brachystegia spiciformisandTerminalia stenostachyain the national park,Diplorhynchus condylocarpon,Diospyros mespiliformisandJulbernadia globiflorain the buffer zone,andPseudolachnostylis maprouneifolia,Julbernadia globif loraandDiplorhynchus condylocarponin the communal area.

PCA biplot of principal components 1 and 2 explain a cumulative variation of 84.9% in species site associations across the three study strata (Fig.2).Results of the PCA show thatB.boehmii,J.globif lora,O.pulchra,P.curatellifolia,Pericopsis angolensis,B.spiciformis,L.discolour,T.stenostachya,D.condylocarpon,D.cinereaandP.angolensiswere highly associated with the national park.P.maprouneifoliawas highly associated with the national park and the communal area.D.nyasaewas associated with the buffer zone whileD.mespiliformis,D.nitudula,F.indica,M.engleri,B.carthica,A.nigrescense,F.saligna,P.rotundifolius,U.nitidaandB.salicifoliawere associated with the buffer zone and communal area (Fig.2).

Fig.2 Principal component analysis biplot of species by site association along the disturbance gradient (taxa codes are given in Table 1)

Diversity indices did not differ (p>0.05) between sites except for evenness (p=0.034).There was a noticeable increase in species evenness with increase in disturbance(Table 3).The effects of growth stage,site and their interaction on species diversity are shown in Table 4.Species richness and evenness significantly differed with growth stage with saplings more species-rich than the other growth stages.However,species evenness was greater in seedlings than for the other growth stages across sites.The other diversity indices were neither significantly influenced by site or their interaction with growth stage(p>0.05;Table 4).In fact,diversity indices for seedlings,saplings and mature trees did not differ (p>0.05) between sites (Table 5).

Table 1 Mean ± standard error of species abundance in sampling plots in the national park (NP),buffer zone (BZ) and communal Area (CA)

Vegetation structure

Diameter class distribution showed a stable population dominated by young trees on all sites,as illustrated by an inverse J-shape (Fig.3).Tree density was significantly influenced by site,growth stage and their interaction(Table 4).Density was significantly higher in the national park than in the other areas (Table 3) and declined with increasing disturbance for seedlings,peaking at intermediate disturbance (buffer zone) (Table 6).Density of mature trees also declined with increasing disturbance (Table 6).

Fig.3 Diameter class distribution of miombo woody species in the national park,buffer zone and communal area

Diameter,height and basal area were significantly influenced (p<0.05) by growth stage but not by site or their interaction (p>0.05;Table 4).Diameter was significantly different among sites for saplings and mature trees(Table 5).Sapling diameters were larger in the buffer zone than in the other areas,while diameters of mature trees increased with increase disturbance (Table 6).Height and basal area did not differ between sites (p>0.05) for saplings and mature trees (Table 6).Volume was significantly influenced by site and growth stage but not by their interaction (Table 4).It was significantly (p=0.001) higher in the national park than in the other areas (Table 3).

Anthropogenic disturbances

The number of stumps was significantly greater in the buffer zone,followed by the communal area,and none were recorded in the national park (Table 7).Similarly,total forest area lost was greater in the buffer zone and in the communal area than in the national park (Table 7).In contrast,f ire frequency was lowest in the communal area than in the other two areas (p>0.05;Table 7).

Table 2 Species importance value index of selected sites(RF is relative frequency,RDen relative density,RDom relative dominance;IVI index is species importance value index)

Table 3 Mean diversity indices and structural variables in the national park,buffer zone and communal area (n =5 and ±=standard error)

Table 4 F-ratio and signif ciance level of GLM testing on the influence of site (national park,buffer zone and communal area),growth stage,and their interaction on diversity indices and structural variables

Table 5 Mean diversity indices by growth stage in the national park,buffer zone and communal area (n =5 and ±=standard error)

Table 6 Mean structural variables for seedlings,saplings and mature trees in the national park,buffer zone and communal area (n =5 and ±=standard error)

Table 7 Mean number of stumps and f rie frequency in the national park,buffer zone and communal area (n =5 and ±=standard error)

Discussion

Anthropogenic disturbances and characteristics of miombo woodland differ along a disturbance gradient.However,typical miombo woodlands species diversity was similar between sites of varying disturbance levels.Nevertheless,this study shows how miombo woodlands respond to disturbances,which is important for woodland management and conservation.

Species diversity did not differ between sites,suggesting that it is not adversely affected by disturbances;this is similar to other research (Williams et al.2007;Chidumayo2013;Zimudzi and Chapano 2016).Miombo species such asBrachystegia boehmii,which is dominant across sampling sites,are known to be resilient to disturbances due to their re-sprouting abilities (Chidumayo 2004;Williams et al.2007).Although species diversity was not different between sites,evenness was higher in the communal area than in the other sites,contrary to Chidumayo (2013).Evenness demonstrates equal representation of species,and less even plant communities,as demonstrated in the national park and buffer zone and in other growth formcategories,indicating the presence of dominant species.As typical of miombo woodlands,B.boehmiiwas dominant and had the highest importance value indices on all sites.Seedlings and saplings had a higher evenness and species-richness,suggesting that few species are recruited into mature trees.This is related to the suppression of seedlings and saplings by the canopy from further development.Selective harvesting also affects mature trees,where certain species such asPterocarpus angolensisare preferred for timber and their medicinal properties.For example,in this study,tree density and volume decreased with increasing disturbance for mature trees,possibly due to harvesting of o large diameter trees used in tobacco and brick curing and in construction.As a result,species diversity declines as trees grow from seedlings to mature trees (Shirima et al.2015).

Diameter size class distribution followed an inverse J-shape on all sites,with a greater abundance of individuals in the smaller classes and a decline in the number of trees from one class to the next.An inverse J-shaped size-class distribution is typical for natural miombo woodlands with active regeneration,recruitment and population maintenance(Botha et al.2004;Banda et al.2006;Syampungani et al.2017).These results in this study suggest that the regeneration pool (root suckers,sprouts,seedlings) is sufficient to maintain these stands and is typical of disturbed miombo woodlands (Chidumayo 2004;Syampungani et al.2017).Miombo species show a remarkable capacity for regrowth,even with continued disturbances (Chidumayo and Frost 1996).As a result,the regenerative capacity of miombo species through coppicing and seedlings is the main reason for their stability and resilience to disturbance.

The diameter of saplings and density of seedlings was greater in the buffer zone than in the other sites,suggesting the existence of favourable growing conditions.Miombo dominant species such asB.boehmiirequire high light intensities to establish and grow and thus require large gaps(Chirwa et al.2014).The intensive tree cutting in the buffer zone creates large canopy gaps,increasing light intensity and temperature which stimulates seed germination and rapid growth (Van Wyk et al.1996;Syampungani et al.2017).In addition,the removal of large trees reduces competition for nutrients and water as well as reduces any allelopathy effects from the interaction between species,enabling seedling establishment and growth (Syampungani et al.2017).

Tree removal,f ire frequency and general loss of forest cover were the most common disturbances in the study site.This is consistent with studies by Muposhi et al.(2016) and Syampungani et al.(2017).Of these,tree cutting and canopy loss were greater in the buffer zone and in the communal area,indicating varying accessibility to resources in the different sites.Not only does the evidence of tree removal and canopy loss in the buffer zone suggest encroachment into the buffer zone (Muposhi et al.2016),but is also consistentwith the extraction of products such as fuelwood,poles and timber for furniture by communities around the woodlands.In addition,tree removal is related to land clearing due to agriculture and settlement expansion (Chidumayo 2002).In contrast,f ire frequency was higher in the national park and buffer zone than in the communal area,possibly due to a high fuel load in the national park.Communities in miombo woodlands burn rangelands in early dry season to reduce fuel load,thereby preventing more intense and damaging fires later in the season (Butz 2009;Ryan and Williams 2011).In addition to uncontrolled fires as people prepare land for cultivation and livestock grazing,prepare charcoal and create firebreaks around homesteads,the high fire frequency in national parks could be related to use of fire to hunt for bushmeat and to smoke out beehives (Kikula 1 986;Sileshi and Mafongoya 2006).These findings suggest that management and conservation of miombo woodlands should consider fire and tree removal when developing woodland management plans.

Conclusion

Although woody species had higher evenness in the communal area compared to the other sites,there were no significant differences in the structure and composition of the woodlands along the disturbance gradient.However,tree density and volume decreased with increasing disturbance with tree harvesting,the most common disturbance in the communal area and in the buffer zone.Miombo woodlands on all sites were stable with good regeneration and recruitment,indicating the resilience of these ecosystems to disturbance.There is,however,the need for continued monitoring of anthropogenic disturbances and woodland response to ensure sustainable utilisation of these resources.

AcknowledgementsThis work was conducted within the framework of the Research Platform“Production and Conservation in Partnership”(www.rp-pcp.org),with the financial assistance of the European Union.The contents of this document are the sole responsibility of the authors and can under no circumstances be regarded as ref lecting the position of the European Union.We are also grateful to the Forestry Commission for allowing us to carry out this study.Special thanks also go to the anonymous reviewers for their time and eff ort that allowed us to increase the clarity and quality of our work.