Map indicating serpentine outcrops of the Barberton Greenstone Belt in Mpumalanga, South Africa. Survey sites are identi fi ed by callout labels. Map was prepared using data provided by the Chief Directorate: Surveys and Mapping, Department of Land Affairs, Republic of South Africa. 

Contexts in source publication

Context 1

... of tree species, these trees are present in low numbers on the serpentine and individuals are often small and/or stunted. The species to genus and species to family ratios (2.33 and 7.91 respectively) for the serpentine sites combined are considerably higher than those for each of the individual serpentine outcrops (Table 2). This suggests a family tolerance of the serpentine condition and that there are different members of families and genera on different outcrops contributing to the relative uniqueness of each outcrop. Less than 1% of the taxa recorded on the serpentine outcrops are Fern and Fern allies as compared to the rest of Mpumalanga Province with 4.2% of the taxa being Ferns (Table 3). The percentage of monocotyledonous taxa is also lower (21%) than the rest of the province (29%). The ratios of the number of monocotyledons to dicotyledons (Table 4) recorded from the plots placed in serpentine vegetation were found to be statistically different (P b 0.02) from the ratio recorded from the adjacent non-serpentine vegetation. The ratios calculated from combining the species lists from all the sampled sites showed a signi fi cantly (P b 0.001) higher monocotyledon to dicotyledon ratio on serpentine (0.35) than on the adjacent non-serpentine vegetation (0.29). The serpentine condition leads to an environment with reduced moisture levels and one would expect that monocotyledons with their well developed fi brous root systems are better able to access soil moisture than plants with tap root systems. In addition plants with fi brous root systems are naturally replacing their roots at a high rate and will thus be less affected by the toxic levels of nickel in the soil (Brady et al., 2005). It is thus surprising that the number of monocotyledons relative to dicotyledons is lower than in the province as a whole. The fi ve most important plant families represented in the vegetation of the serpentine outcrops of the Barberton Greenstone Belt are presented in Table 5. The comparison to the entire Mpumalanga Province demonstrates the distinctiveness of the serpentine fl ora. The list of major families shows the decreased representation of the Apocynaceae and Cyperaceae and the increased representation of the Asteraceae on serpentine soils. The Rubiaceae and Euphorbiaceae also show an increased representation on some individual outcrops. The serpentine vegetation has a third of the Asteraceae taxa found in the entire province; however, the serpentine outcrops represent only 0.1% of the area of the province. Batianoff et al. (2000) suggest that such a high representation may be explained by a family tolerance of soil conditions such as high Mg, commonly associated with serpentine soils and that this innate tolerance facilitates prominence of these families on serpentine outcrops. This is borne out by all the nickel accumulators on the Barberton Greenstone Belt being members of the Asteraceae. Spearman rank correlation coef fi cients for all pair wise comparisons of ranking of the twenty most diverse families of the sampled serpentine sites of the Barberton Greenstone Belt are listed in Table 6. Lower, non-signi fi cant correlations were found with all comparisons with Magnesite Mine (MM) and between CoreZone (CZ) and the fl ora of Mpumalanga Province (0.539). The UPGMA dendrogram (Fig. 2) shows that Mundt's Concession (MC) and Sawmill (SM) sites are most similar to one another. The Groenvaly (GV) and Agnes Mine (AM) sites show a high level of similarity in their familial diversity. The Rosentuin (RT) and CoreZone (CZ) show higher levels of similarity with each other than with the fl ora of Mpumalanga Province. The fl ora of the Magnesite Mine (MM) site shows the least similarity to any of the other sites and also a low degree of similarity with the fl ora of the Mpumalanga Province. There is no substantial difference in the soil chemistry (Table 7) between the Magnesite Mine site and the other sites. This suggests that the difference in the vegetation could be due to the lower rainfall of the area in which the Magnesite Mine occurs (Table 1), which, possibly results in an exacerbation of the ‘ serpentine condition ’ . The large difference in the total number of taxa from combined lists and individual sample site lists suggests that the composition of taxa on each serpentine outcrop is near-unique and supports a large number of taxa found at one site only. For instance, 35% of the fl ora on the Groenvaly (GV) outcrop and 36% of that on the Magnesite Mine (MM) outcrop are found on only those sites and not on the others sampled (Table 2). The Sørenson's coef fi cient of similarity (Table 8), which measures the degree of similarity between sites indicates that less than 26% of the plant taxa recorded are shared between most sites. Two sites, namely the Groenvaly (GV) and Rosentuin (RT) sites are most similar in terms of species composition (35%), which could be explained by the relative proximity of the sites (Fig. 1). The Sawmill (SM) site and the Groenvaly site, which are also relatively close together, share 34% of their plant taxa. The three outlying outcrops i.e. Magnesite Mine (MM), Kalkloof (KK) and CoreZone (CZ) share the lowest number of taxa with the other outcrops. These three outcrops also have the lowest soil nickel concentrations measured (i.e. 0.54, 0.13, 0.51% of metal in air dried soil, respectively). A possible source of error includes the oversampling of common and widespread taxa compared with uncommon, endemic or patchily distributed species. Thus with more sampling the number of taxa shared between sites would decrease. 18 taxa have been found to be undescribed and are listed in the checklists as sp. nov. or subsp. nov. This would suggest that the area is historically under-collected and although the area is considered to be relatively high in diversity, this level may be underestimated. Many of these taxa could be found to be taxa endemic to the Barberton area or to Mpumalanga Province increasing the conservation value of the area. While the family composition of the fl ora of the Barberton Greenstone Belt shows some variation from the fl ora of the surrounding areas (Table 5), this difference is not seen in the fl ora of the greenstone outcrops of Western Australia. Here the family composition is very similar to and typical of the fl ora of the South Western Interzone (Gibson and Lyons, 1998a). However, the relative uniqueness of the fl ora of each group of outcrops of the Western Australian greenstones is similar to that of the Barberton Greenstone Belt with the greenstone of the Bremer Range and the Parker Range only sharing 32% of the recorded fl ora (Gibson and Lyons, 1998b). The fl ora of the Central Queensland serpentine has a higher representation of Fabaceae (narrowly de fi ned) (5%), Mimosaceae (2%) and Rubiaceae (1.5%) compared to the Port Curtis District fl ora (Batianoff et al., 2000). It is postulated that these families are characterised by a higher proportion of serpentine-tolerant species which fi nd the serpentine soils adequate and/or neutral for establishment, growth and reproduction, thus facilitating their relative expansion. The 20 most frequently occurring plants on the serpentine of the Barberton Greenstone Belt are listed in Table 9 and are compared to those of the adjacent non-serpentine areas sampled. The serpentine vegetation and the surrounding non-serpentine vegetation are dominated by grasses such as Themeda triandra Forssk., Heteropogon contortus (L.) Roem. & Schult. and Loudetia simplex (Nees) C.E.Hubb. However, grasses such as Cymbopogon caesius (Hook. & Arn.) Stapf, Tristachya leucothrix Trin. ex Nees and Trachypogon spicatus Kuntze appear to become more common on serpentine than on non-serpentine. Due to their widespread distributions these taxa would not be considered to be indicator species (Kruckeberg, 1986). Grasses such as Cymbopogon pospischilii (K. Schum.) C.E.Hubb. and Hyparrhenia fi lipendula (Krauss) Stapf, which are relatively common throughout South Africa, were not recorded on any of the sampled serpentine outcrops. The tree species Dichrostachys cinerea (L.) Wight & Arn. and herbaceous Oxalis obliquifolia Steud. ex A.Rich. were found to be relatively common on non-toxic soils but occasional to rare on nearby serpentine outcrops (Table 9). A further 53 species representing 47 genera and 29 families (Appendix 2) are possible ‘ excluded ’ taxa as they were not recorded on serpentine sites but were recorded from the Modi fi ed- Whittaker plots placed on adjacent non-serpentine areas. Twenty genera and two families (Kirkiaceae and Strychnaceae) are absent from the serpentine fl ora. Of the 60 dicotyledonous species excluded, 55% are woody species (shrubs, trees or woody climbers), many of which are common in the surrounding areas. All of the serpentine endemics (Table 11) are considered to be rare because of their restricted distributions on the serpentine outcrops of the Barberton Greenstone Belt. However, 17 taxa are found on only fi ve or fewer outcrops and on those outcrops have small and sparse populations, increasing their measure of rarity. Eighteen of the taxa collected from the seven serpentine outcrops of the Barberton Greenstone Belt, included in this study, have been listed on the South African National Biodiversity Institute Red Data list as ei- ther threatened with extinction or of conservation concern (SANBI, 2013). An additional seven taxa are listed on the Mpumalanga Tourism and Parks Agency's list of threatened plants for the province (Lötter pers. comm. 1 ). Seven of these taxa are listed as vulnerable, three as near threatened, four as declining and four as rare. Eighteen taxa are considered to be new and undescribed species and some of these are further thought to be endemic to serpentine soils. Further taxonomic study and occurrence data are required to con fi rm these fi ndings. Only one of the sampled outcrops ...

Context 2

... of serpentinite (henceforth referred to as ‘ serpentine ’ ) rocks are often referred to as edaphic islands due to their sharp boundaries and patchy distribution. Soils derived from serpentine rocks are considered a harsh environment for plants due to low levels of calcium relative to magnesium, low nutrient content, nickel and chromium toxicity and poor water holding capacity (Harrison et al., 2006). The extreme physical and chemical properties of serpentine soils provide conditions that allow colonisation by tolerant species and then strong diversifying selection processes may lead to ecological speciation (Kruckeberg, 1986; Rajakaruna, 2004). Species of plants tolerant to serpentine soils include species found only on serpentine soils i.e. serpentine endemics; species that are local or regional indicators but are not restricted to serpentine and species that are serpentine indifferent (Kruckeberg, 1984). Taxa that are found on adjacent non-serpentine substrates but are completely excluded from serpentine soils (Harrison et al., 2006) are also important for de fi ning the distinctiveness of serpentine fl oras. Physiological and evolutionary mechanisms hypothesised to be responsible for adaptations to serpentine soils include the tolerance of a low calcium-to-magnesium ratio, avoidance of Mg toxicity, or a high Mg requirement (Brady et al., 2005). In a fl oristic analysis of the serpentine vegetation of Central Queensland, Australia, Batianoff et al. (2000) suggested a family tolerance of soil conditions and postulated that some families are characterised by higher proportions of serpentine tolerant species. It is also thought that edaphic conditions strongly in fl uence species diversity and levels of endemism. Batianoff et al. (2000) found that species richness of the serpentines of Central Queensland in Australia decreased as soil nickel concentrations increased in lowland forests and that levels of endemism increased with increasing nickel concentrations. In the Californian serpentine vegetation, soil calcium levels were negatively correlated with the number of serpentine endemic taxa (Harrison, 1999). The fl ora of the serpentine outcrops of the Barberton Greenstone Belt in the eastern parts of South Africa have been less well documented than those of Cuba (Borhidi, 1992), New Caledonia (Jaffré, 1992), California (Kruckeberg, 1984; Callizo, 1992), Zimbabwe (Wild, 1965), Australia (Gibson and Lyons, 1998a,b, 2001; Batianoff et al., 2000) and Italy (Ferrari et al., 1992; Verger, 1992). Most of these studies have lead to the identi fi cation of many plant species endemic to serpentine soils. The lack of knowledge of the fl oras of the metalliferous sites in South Africa initiated a funded research programme entitled ‘ Metalliferous Flora ’ and focused on the study of the fl oristics, biodiversity, conservation, soils and evolution of these fl oras. This study supplements fl oristic analyses conducted previously on parts of the Barberton Greenstone Belt by Williamson (1994), Changwe and Balkwill (2003) and McCallum (2006). The Barberton Greenstone Belt is located in south-eastern Mpumalanga, South Africa (Fig. 1). This province has an estimated 4946 plant species and infraspeci fi c taxa occurring within its boundaries, yet it only comprises 6.3% of South Africa's surface area (Lötter et al., 2002). This high level of plant diversity is not evenly distributed across Mpumalanga. Two regions and three Centres of Endemism were recognised by Van Wyk and Smith (2001) and an additional one for the Lydenburg area was proposed by Lötter et al. (2002). The Barberton Greenstone Belt falls within the Barberton Centre of Endemism, which has an area of about 4000 km 2 , has about 2210 plant species and more than 80 endemics with 3.6% endemism. A large percentage ( N 29%) of this area is transformed by commercial plantations of species of Pinus and Eucalyptus (Lötter et al., 2002), threatening many of the endemics on serpentine and other ultrama fi c substrates (Williamson and Balkwill, 2006). The Barberton Greenstone Belt consists of approximately 30 large serpentine outcrops in the belt surrounded by several very small outcrops (Ward, 2000). These outcrops are located in an inverted equilateral triangle centred on Barberton and extending to Malelane in the east and to Badplaas in the south. The Barberton Greenstone Belt is surrounded by extensive granitoid plutons and gabbroid intrusions. The serpentine outcrops consist of various combinations of serpentinised dunite, amphibolite, chrysotile asbestos and peridotite (Morrey et al., 1992). The largest of these outcrops is about 19 km 2 , and there are several smaller outcrops (from 0.1 km 2 ). Some outcrops are separated from others by up to 20 km (Balkwill et al., 1997). The outcrops occur in mountainous areas and are heterogeneous in altitude, slope, soil depth and other topographic features. The serpentine vegetation falls within the Mixed Lowveld Bushveld, Sour Lowveld Bushveld and North-eastern Mountain Grassland vegetation types as described by Low and Rebelo (1996). It has more recently been reclassi fi ed as Barberton Serpentine Sourveld by Mucina and Rutherford (2006) due to the unique, stunted woody vegetation that results from the high toxicity of the soils. The landscape of the areas surrounding the serpentine outcrops of the Barberton Greenstone Belt is mostly hilly with varied terrain. The outcrops range from 350 to 1400 m above sea level. The climate of the area is characterised by summer rainfall (MAP 600 – 1150 mm) with dry winters, during which frost is infrequent. A ‘ Resolution ’ was passed by the delegates of The First International Conference on Serpentine Ecology held in 1991 supporting the conservation of the vegetation of serpentine areas worldwide (Kruckeberg, 1992). Subsequent to this resolution a few publications providing evidence for the need for the conservation of serpentine vegetation have been published (Wolf, 2001; Selvi, 2007). The ability of metallophytes to tolerate extreme metal concentrations commends them for revegetation of mines and metal-contaminated sites and can be exploited in environmental technologies, such as phytostabilisation, phytoremediation and phytomining (Whiting et al., 2004). Conservation of biodiversity is the main objective of most conservation organisa- tions (Brooks et al., 2006) and thus if diversity of serpentine vegetation is conserved then metallophytes and other rare and/or restricted range (endemic) species are also conserved. Serpentine outcrops usually support many rare and endemic species, which are often threatened by a variety of activities and are in need of conservation (Wolf, 2001). However, very few accounts of the actual conservation of serpentine vegetation exist suggesting that such sites continue to be severely under-conserved. The serpentine vegetation of the Barberton Greenstone Belt is threatened by mining activities, commercial plantations of species of Eucalyptus and Pinus and by urban development. This paper aims to characterise and describe the plant species composition of serpentine sites in the Barberton Greenstone Belt as compared to surrounding non-serpentine areas. This allows the contribution of the serpentine fl ora to the overall diversity and endemicity of the Barberton Centre of Endemism to be quanti fi ed and the conservation value of the area to be determined in terms of taxon richness at different taxonomic ranks. The plant list presented for the selected serpentine outcrops of the Barberton Greenstone Belt is intended for use in land management, conservation planning and rehabilitation of disturbed serpentine landscapes. An analysis of the rarity or commonness of taxa on serpentine outcrops compared with their occurrence in the Mpumalanga Province or the Barberton Centre of Endemism will assist in identifying areas or sites that would maximise the conservation of plant diversity on serpentine outcrops. It was predicted that each serpentine outcrop of the Barberton Greenstone Belt is unique in its species composition and a large number of serpentine sites need to be ...

Context 3

... more than 80 endemics with 3.6% endemism. A large percentage ( N 29%) of this area is transformed by commercial plantations of species of Pinus and Eucalyptus (Lötter et al., 2002), threatening many of the endemics on serpentine and other ultrama fi c substrates (Williamson and Balkwill, 2006). The Barberton Greenstone Belt consists of approximately 30 large serpentine outcrops in the belt surrounded by several very small outcrops (Ward, 2000). These outcrops are located in an inverted equilateral triangle centred on Barberton and extending to Malelane in the east and to Badplaas in the south. The Barberton Greenstone Belt is surrounded by extensive granitoid plutons and gabbroid intrusions. The serpentine outcrops consist of various combinations of serpentinised dunite, amphibolite, chrysotile asbestos and peridotite (Morrey et al., 1992). The largest of these outcrops is about 19 km 2 , and there are several smaller outcrops (from 0.1 km 2 ). Some outcrops are separated from others by up to 20 km (Balkwill et al., 1997). The outcrops occur in mountainous areas and are heterogeneous in altitude, slope, soil depth and other topographic features. The serpentine vegetation falls within the Mixed Lowveld Bushveld, Sour Lowveld Bushveld and North-eastern Mountain Grassland vegetation types as described by Low and Rebelo (1996). It has more recently been reclassi fi ed as Barberton Serpentine Sourveld by Mucina and Rutherford (2006) due to the unique, stunted woody vegetation that results from the high toxicity of the soils. The landscape of the areas surrounding the serpentine outcrops of the Barberton Greenstone Belt is mostly hilly with varied terrain. The outcrops range from 350 to 1400 m above sea level. The climate of the area is characterised by summer rainfall (MAP 600 – 1150 mm) with dry winters, during which frost is infrequent. A ‘ Resolution ’ was passed by the delegates of The First International Conference on Serpentine Ecology held in 1991 supporting the conservation of the vegetation of serpentine areas worldwide (Kruckeberg, 1992). Subsequent to this resolution a few publications providing evidence for the need for the conservation of serpentine vegetation have been published (Wolf, 2001; Selvi, 2007). The ability of metallophytes to tolerate extreme metal concentrations commends them for revegetation of mines and metal-contaminated sites and can be exploited in environmental technologies, such as phytostabilisation, phytoremediation and phytomining (Whiting et al., 2004). Conservation of biodiversity is the main objective of most conservation organisa- tions (Brooks et al., 2006) and thus if diversity of serpentine vegetation is conserved then metallophytes and other rare and/or restricted range (endemic) species are also conserved. Serpentine outcrops usually support many rare and endemic species, which are often threatened by a variety of activities and are in need of conservation (Wolf, 2001). However, very few accounts of the actual conservation of serpentine vegetation exist suggesting that such sites continue to be severely under-conserved. The serpentine vegetation of the Barberton Greenstone Belt is threatened by mining activities, commercial plantations of species of Eucalyptus and Pinus and by urban development. This paper aims to characterise and describe the plant species composition of serpentine sites in the Barberton Greenstone Belt as compared to surrounding non-serpentine areas. This allows the contribution of the serpentine fl ora to the overall diversity and endemicity of the Barberton Centre of Endemism to be quanti fi ed and the conservation value of the area to be determined in terms of taxon richness at different taxonomic ranks. The plant list presented for the selected serpentine outcrops of the Barberton Greenstone Belt is intended for use in land management, conservation planning and rehabilitation of disturbed serpentine landscapes. An analysis of the rarity or commonness of taxa on serpentine outcrops compared with their occurrence in the Mpumalanga Province or the Barberton Centre of Endemism will assist in identifying areas or sites that would maximise the conservation of plant diversity on serpentine outcrops. It was predicted that each serpentine outcrop of the Barberton Greenstone Belt is unique in its species composition and a large number of serpentine sites need to be protected in order to adequately conserve the area. The serpentine outcrops of the Barberton Greenstone Belt have been visited regularly from 1991 and during these visits extensive plant collections were made. In addition taxon data were collected using Modi fi ed-Whittaker plots positioned on each serpentine site and adjacent non-serpentine area. A detailed description of the positioning and layout of these plots is given in Williamson and Balkwill (2013). Field surveys focused on seven outcrops selected in such a way that the full range of variation is presumably accounted for (Table 1). Fig. 1 indicates the location of all the serpentine sites in the Barberton Greenstone Belt and shows the sites studied in detail. Distribution data for all taxa recorded were obtained from: PRECIS (Pretoria National Herbarium Computerised Information System) a computerised data bank managed by the South African National Biodiversity Institute (SANBI); taxonomic accounts, mono- graphs, herbarium collections and distributional data collected by scientists of the Mpumalanga Tourism and Parks Agency (MTPA). This was done to identify the taxa endemic to the serpentine of the Barberton Greenstone Belt. All specimens collected by the authors were deposited at the C.E.Moss Herbarium of the University of the Witwatersrand, Johannesburg and duplicate specimens were forwarded to various herbaria in South Africa and abroad, as annotated on individual specimen labels. The plant lists produced include sterile specimens that were positively identi fi ed, but for which no herbarium specimen was kept. These specimens were added to the plant list as they represent the only record of these taxa from these sites. A number of sterile specimens that could not be identi fi ed were noted but not added to the plant list. Taxa on the plant list that are restricted to the area within which the Barberton Greenstone Belt is found and that occur more commonly on serpentine soils than on non-serpentine soils were identi fi ed as possible local or regional serpentine indicators as described by Kruckeberg (1984). In addition, taxa that occur commonly in non- serpentine plots but do not appear on the serpentine plant lists were identi fi ed to be possible excluded taxa. This list was re fi ned by removing any plants that occur on species lists compiled for other serpentine sites of the Barberton Greenstone Belt (McCallum, 2006; Changwe and Balkwill, 2003; Balkwill and Balkwill, 1999; Kidger, 1993; Williamson, 1994). About 85% of the collected specimens have been identi fi ed to species or infraspeci fi c level; however, some specimens remain unidenti fi ed (3%) or need further consultation and/or analyses to con fi rm identi fi cations (12%). All endemic species and those with relatively restricted distributions on and around serpentine outcrops were subjected to a leaf sap test, in which sap was applied to dimethylglyoxime-impregnated fi lter paper (1% in ethanol). The test gives a distinct dark pink reaction to leaf sap containing high levels of nickel. The South African National Botanical Institute's online database (SANBI, 2013) was used to determine the number of taxa recorded for Mpumalanga Province and their familial classi fi cation system was adopted when compiling the checklists for each serpentine outcrop (Appendix 1). Proportions of Fern and Fern allies, monocotyledons and dicotyledons recorded in the plant list for each site (Appendix 1) were compared to the other sites and to the vegetation of Mpumalanga Province as a whole. The ratio of dicotyledons to monocotyledons was calculated for the taxa recorded within the Modi fi ed-Whittaker plots at each site and these data were compared to those of the adjacent non- serpentine sites by way of a Student's T-test. The relative proportions of monocotyledons and dicotyledons, calculated by combining the data from all the Modi fi ed-Whittaker plots placed on serpentine outcrops and on adjacent non-serpentine areas, were compared using a Chi square test to determine if differences between these proportions are signi fi cant. Species compositions of the serpentine sites were compared by ranking the twenty most diverse families at each site. Site-speci fi c species compositions were compared to the fl ora of Mpumalanga Province to highlight any deviation in fl oristic composition at the family level from the regional fl ora. Spearman rank correlation coef fi cient tests (Sokal and Rohlf, 1995) were performed to test for signi fi cance of correlations in family rankings between pairs of sites. The null hypothesis was that the familial ranking of one site did not covary with the familial ranking of another site. A correlation analysis was conducted on the ranks in a pairwise manner between all combinations of the sites and a correlation coef fi cient (Spearman's rho) is calculated for each pair of ranks. A value of r ( rho ) = 1 indicates a perfect positive correlation and the value of r = − 1 indicates a perfect negative correlation. The signi fi cance of this coef fi cient is calculated by determining a P-value for the correlation. Two additional sites, Agnes Mine (AM) and Kaapsehoop (KH) (Williamson, 1994) were included in this analysis to broaden the basic data. Clustering by the unweighted pair group method, arithmetic average (UPGMA: was used to create a phenogram indicating the relative similarities of the fl oras of the individual sites. The plant list (Appendix 1) provides an alphabetical listing of species by family. The number of serpentine-tolerant taxa recorded within ...