Mountains of the Mist: A first plant checklist for the Bvumba Mountains, Manica Highlands (Zimbabwe-Mozambique).

Abstract The first comprehensive plant checklist for the Bvumba massif, situated in the Manica Highlands along the Zimbabwe-Mozambique border, is presented. Although covering only 276 km2, the flora is rich with 1250 taxa (1127 native taxa and 123 naturalised introductions). There is a high proportion of Orchidaceae and Pteridophyta, with both groups showing a higher richness than for adjacent montane areas, which may be due to the massif’s relatively high moisture levels as a result of frequent cloud cover. However, in contrast to other mesic montane regions in southern Africa, there are relatively few near-endemic or range-restricted taxa: there is only one local endemic, Aeranthes africana, an epiphytic forest orchid. This is likely to be an effect of the massif having limited natural grassland compared to forest, the former being the most endemic-rich habitat in southern African mountains outside of the Fynbos Biome. Six other near-endemic taxa with limited distribution in this portion of the Manica Highlands are highlighted. The high number of invasive species is probably a result of diverse human activities in the area. The main species of concern are Acacia melanoxylon, a tree that is invading grassland and previously cultivated land, the forest herb Hedychium gardnerianum which in places is transforming forest understorey with an adverse effect on some forest birds, and the woody herb Vernonanthura polyanthes which invades cleared forest areas after fire. Future botanical work in the massif should focus on a more detailed exploration of the poorly known Serra Vumba on the Mozambican side and on the drier western slopes. This will allow for a more detailed analysis of patterns of endemism across the Manica Highlands.


Introduction
Southern African mountains continue to fascinate biologists, ecologists and conservationists with their high endemism, high species diversity, and as a haven for taxonomically complex and cryptic evolutionary lineages (White 1978;Taylor et al. 2013;Uys and Cron 2013;Conradie 2014;Mynhardt et al. 2015;Padayachee and Procheş 2016;Phiri and Daniels 2016;Conradie et al. 2018;Branch et al. 2019). From a floristic perspective, there has been a steady output of comprehensive data from the region over the past 25 years, for example the Nyika Plateau (Burrows and Willis 2005) and Mount Mulanje (Strugnell 2006) in Malawi; Mounts Gorongosa, Mabu and Namuli (Müller et al. 2008;Timberlake et al. 2009;Bayliss et al. 2014;Timberlake, in prep.) in Mozambique;Chirinda Forest (Drummond and Mapaure 1994) in Zimbabwe; the Angolan Highlands (Goyder and Gonçalves 2019); the heterogeneous southern African Great Escarpment (Clark et al. 2011(Clark et al. , 2014Roth et al. 2014;Darbyshire et al. 2018;Carbutt 2019). This has greatly improved our regional understanding of montane floristics, patterns of endemism, biogeography and conservation needs. In addition, an account of all the endemic and near-endemic plants from Mozambique has recently been published (Darbyshire et al. 2019b), some of which occur in these border areas. Despite these advances, ongoing biodiversity research in southern African mountains remains a key regional need (Clark et al. 2011(Clark et al. , 2019CEPF 2012), and there remains a substantial lag in the production of fundamental biodiversity and taxonomic data compared to other mountains in Africa.
The Manica Highlands (Clark et al. 2017), which lie on the border between Zimbabwe and Mozambique and are mostly synonymous with Van Wyk and Smith's (2001) Chimanimani-Nyanga Centre of Floristic Endemism, comprise an area that has been well-botanised over the last 100 years, yet with few publications. Over the past decade, attention has been focused on improving our knowledge of plant diversity and endemism for this ecologically complex 8,000 km 2 mountain system. For instance, the first comprehensive floristic treatment of the Nyanga massif was published only recently (Clark et al. 2017), as was the first substantial revision in 50 years of the Chimanimani flora (Wursten et al. 2017).
The central parts of the Manica Highlands (from north to south: Stapleford, Penhalonga, Bvumba, Banti, Himalaya, Tsetserra) are now the outstanding areas that require synthesis of available data and further fieldwork, with the Bvumba being prob-ably the most thoroughly botanised component of the Manica Highlands. Here we present the first comprehensive plant checklist for the Bvumba massif, with some notes on the massif and its flora.

The Bvumba area
Defining the study area Clark et al. (2017) defined the Bvumba as the entire central component of the Manica Highlands, which includes the Bvumba as well as the Penhalonga and Stapleford uplands that occur immediately to the north, i.e. between the Bvumba and Nyanga. However, due to lack of adequate floristic data for Penhalonga and Stapleford, we here restrict ourselves to the Bvumba massif sensu stricto and not to the broader Bvumba area as shown in Clark et al. (2017). The checklist area is defined as that part lying primarily above 1200 m altitude with significant vegetation cover, an extent of around 276 km 2 , and differs slightly from the more rigid use of the 1200 m contour used by Childes and Mundy (2001).

Location and topography
Centred on 19°06'S, 32°47'E, the Bvumba lies 20 km south-east of the border city of Mutare and straddles the Zimbabwe-Mozambique border ( Figure 1). The largest extension lies within Mutare District in eastern Zimbabwe, but a significant area of around 30 km 2 lies over the border in Manica District in Mozambique's Manica Province. This north-eastern extension terminates at the peak of Serra Vumba (1646 m) at 18°58'35"S, 32°53'25"E, just 5 km south of Manica town (previously Vila da Manica) on the main Mutare-Beira highway. The area as a whole is bounded in the north by the Muneni valley (in which the Forbes/Machipanda border post is situated) and in the south by the Burma valley (Nyamataka River), which separates it from the Banti-Himalaya-Tsetserra massif. The Mozambique midlands/lowlands and Chicamba Real dam form the eastern limits, while the Odzi River valley forms the western boundary. Outlying ridges and inselbergs to the west − including Cecil Kop -have been excluded. The lower elevational cut-off of approximately 1000-1200 m used here roughly follows the base of the Bvumba massif where it emerges from the surrounding plains.
Consisting of an upland massif, the highest points in the study area are Castle Beacon (1911 m) and Chinyakwaremba (1714 m), while the main rivers are the Nyamataka in the south, which drains into the Rio Vanduzi via the Chicamba Real dam and then on into the Rio Búzi, the Zonwe River in the centre and the Ndonwe River in the north, both also draining into the Vanduzi. To the west the main river is the Nyachowa, which drains into the Odzi and eventually into the Save River.

Geology
The principal rocks underlying the Bvumba area are gneissic granites of Proterozoic age, perhaps 2560 million years old, interlaced with intrusions of finer-grained darker dolerite rock perhaps 1800 million years old (Bartholomew 1990;Martin 2000). Millions of years of erosion have left the granite domes exposed, standing above the surrounding younger landscape. Some peri-glacial features date from the Pleistocene period (25,000 years BP). No minerals of economic significance are noted, although gold has been mined for hundreds of years from areas just north of Mutare. The soils are often deep and well-weathered, but are not considered particularly fertile owing to their age and weathering.
Climate Surprisingly, there does not appear to be a long-term rainfall recording station in the Bvumba area (Agritex 1989), the nearest being at Mutare, which is significantly lower in elevation and with a lower annual rainfall. On isohyet maps, rainfall is indicated as being around 1800 mm/year (Agritex 1989). The warm rainy or growing season extends from November to mid-March, with a colder dry season from June through to August. Frosts are scarce. A major feature, and one which gives the mountains their name of Mountains of the Mist, are the frequent mists and low moist cloud during the dry season, sometimes called guti (Martin 2000). It is these mists that reduce the physiological stress of the dry season for many plant species and allow forest to thrive.

History and land use
The area has been settled, on the Mozambique side in particular, for perhaps 1000 years, with the first written record of people living there dating from Portuguese explorers in the first half of the 17 th century (Bannerman 2010;Martin 2000). It was possibly more heavily settled in the past than it is now, as seen from the many grinding stones found inside the forests. When European occupation first started, much of the area was settled by people of the Chirara dynasty (Martin 2000), particularly on the Mozambique side and along the Nzombe valley. The first European settlement on the Zimbabwe side started around 1890, and the first formal concession was granted by the British South African Company for the farm Cloudlands in 1898, now a private nature reserve. A road was cut in 1917 to the Bunga Forest from Mutare (then Umtali), and by 1921 the road continued over and into the Burma Valley (Martin 2000).
Within Zimbabwe, the Bvumba area is now nearly all on privately held land, much of it in small agricultural holdings, plantations or managed for conservation or ecotourism, including the famous Leopard Rock Hotel and golf course. There is a good road network and the area is well-settled with many properties, a number of which are used as holiday homes with many owners making strong efforts to conserve both the flora and bird life. Commercial farming in the Bvumba area includes dairy, Protea flowers, coffee, and with some wattle and Eucalyptus plantations. The only formal conservation areas are the Bvumba National Park and Botanic Gardens (201 ha) and the now much-expanded Bunga Forest Botanical Reserve (1,560 ha) managed by the Zimbabwe Parks and Wildlife Management Authority. There are no formally conserved areas on the Mozambique side, although from recent Google Earth imagery much of Serra Vumba's vegetation cover appears to be relatively intact and it may be protected by local traditions.

Vegetation
There are four broad vegetation types occurring in the Bvumba area -montane grassland, Afromontane evergreen forest, high-rainfall miombo woodland, and secondary scrub savanna (Martin 2000). The edges of the forests are surrounded by a dense scrub of Pteridium, Smilax, Buddleja, Vangueria and Vernonia (Childes and Mundy 2001). In addition, there are plantations of exotic timber trees and cultivated or fallow agricultural land. It is possible that montane forest originally covered much of the Bvumba, principally owing to dry season precipitation in the form of low wet clouds (guti). But there is little montane grassland, a vegetation type of great botanical significance in the Nyanga and Chimanimani areas (Timberlake et al. 2016a). It is not clear why this is so, but it might be due to the somewhat lower elevations of the Bvumba compared to Nyanga and/or the greater amount of winter precipitation that allows forest to sustain itself. The high-rainfall woodlands contain many epiphytic ferns and orchids, which also have a high frequency in montane forest. Burrows (1990) suggests that the Bvumba has perhaps the richest fern flora in southern Africa owing to the pervasive mists. Secondary scrub savanna is generally found on infertile or gravelly soils or in degraded forests or reverted agricultural land, and it is here that a number of invasive tree and shrub species are found (S. Childes, pers. obs.).

Previous studies
The Bvumba has seen much botanical activity and many collections over the last 50 years, but almost entirely from the Zimbabwe side. Two notable collectors were Norman Chase, who assembled over 8,000 specimens with meticulous notes, mostly from the Manica Highlands but with a large number from the Bvumba, and John Ball, for whom the Bvumba and Chimanimani provided the inspiration for his book on epiphytic orchids (Ball 1978). Other significant collectors include Hiram Wild, Darrel Plowes, Tom Müller and Hamish Gilliland. Three of the current authors (Mark Hyde, Petra Ballings and Bart Wursten) have collected extensively there over the last 20 years, and another (Susan Childes) runs a small forest and bird conservation area. Although no comprehensive overall checklist was available, detailed lists of the orchids and ferns of the broader area are found on the Vumba Nature website (Ballings and Wursten 2019) and a preliminary Bvumba plant list was compiled by Mark Hyde from a manuscript list by Norman Chase (Chase, no date) and Tree Society and Orchid Society records (Hyde 1999).
The Bvumba Highlands are considered to be an Important Bird Area (IBA ZW004, Childes and Mundy 2001), known for the richness of its montane avifauna. The IBA is taken to be the area above 1200 m elevation, considered to be the lower limit for montane bird species (Harwin et al. 1994). A total of 242 bird species have been recorded, including three of global conservation concern.
The only detailed plant ecological work done so far specifically in the Bvumba area is that by Plowes (2002), who looked at the impacts of the devastating Cyclone Eline on the 40-ha Bunga Forest in February 2000. He noted nearly 200 fallen trees that had created 46 patches totalling 1.57 ha, equivalent to a loss of 13% of forest cover. Gilliland (1938), in his study of the vegetation of Zimbabwe's Manicaland, surprisingly does not mention areas this far south.
During his major study on Eastern Zimbabwe's moist forests (Müller 1999(Müller , 2006, Tom Müller recorded 37 georeferenced 50 × 50 m forest plots in the Bvumba area. These fell into seven of his 12 described forest types, with most being Type 5 (Syzygium guineense subsp. afromontanum montane forest, 10 plots), Type 7 (Mixed sub-montane forest, 8 plots) and Type 11 (Medium elevation forest, 8 plots). In addition, there were a few plots each of Type 6 (Regenerating montane forest), Type 8 (Craibia brevicaudata forest), Type 9 (Albizia-dominated regenerating forest) and Type 10 (Albizia schimperiana forest). Hence most of the Bvumba forest plots recorded fell into what Muller calls sub-montane forest, with only a few from the montane or medium elevation forest zones. An interesting finding is that of the 37 plots marked on aerial photographs from the early 1970s, only three have been obviously lost or damaged from what can be seen of their canopy cover using recent Google Earth imagery (most dating from June 2019).
Although not formally documented, it does appear that disturbance over the last 100 years has led to a drying out of some of the forests and an invasion of alien plants (T. Müller, pers. comm. 2017). The impression (S. Childes, pers. obs.) is that the fern flora is moving towards the more generalist species and that some drought-sensitive species of angiosperms such as Streptocarpus umtaliensis and Cryptostephanus vansonii are reducing.

Materials and methods
As the Bvumba is well-documented botanically, with approximately a century of botanical collecting, a checklist approach to documenting plant diversity and endemism was decided upon in order to be comparable to floristic lists recently compiled for Nyanga massif (Clark et al. 2017) and the Chimanimani mountains (Wursten et al. 2017). The two main sources were (i) an extract of the Harare Herbarium (SRGH) database of all records containing the word 'Vumba' in the locality field, with any records obviously from below 1200 m elevation removed, and (ii) records from the Bvumba area above 1200 m elevation cited on the Flora of Zimbabwe website (https://www. zimbabweflora.co.zw/, Hyde et al. 2020). In addition to these there were (iii) records from published volumes of Flora Zambesiaca, (iv) orchid and pteridophyte records listed on the Nature Vumba website (Ballings and Wursten 2019), (v) confirmed records from an unpublished Bvumba checklist (Hyde 1999) including those cited as being from Chase's list (Chase, no date), (vi) any additional records from forest plot studies undertaken by Müller in the 1970s (Müller 2006), and finally (vii) personal records from the authors. A herbarium specimen or record citation (i.e. from the Flora of Zimbabwe website) is given for each taxon, or a confirmed sighting (s.r.) indicated. If there was any uncertainty over an occurrence, the record was omitted.
Families and species are listed alphabetically under pteridophytes, gymnosperms, monocotyledons and dicotyledons. Nomenclature and family arrangement follow those used on the Flora of Zimbabwe website (Hyde et al. 2020, accessed 1 February 2020). Pteridophyte families follow that used in the Pteridophyte Phylogeny Group (2016). Species authorities are abbreviated following Brummitt and Powell (1992). Synonyms are given only for significant recent changes or for taxa that have been known or recorded locally under a different name (e.g. in Mapaura and Timberlake 2004) or where confusion may occur. Where a taxon is believed to be endemic or near-endemic, this is indicated with an E or NE, respectively. Species that are said to be naturalised or introduced on the Flora of Zimbabwe website are indicated with an asterisk (*).

Results and discussion
The checklist contains 1250 taxa, comprising 137 pteridophytes, 2 gymnosperms and 1111 flowering plants (Table 1). Of these, 1127 are native species and 123 (9.8%) are naturalised or semi-naturalised introductions, most, unsurprisingly, being cosmopolitan weeds in the Asteraceae (25 species), Poaceae (8 species) and Amaranthaceae (7 species). The largest families represented in the checklist are shown in Table 2.
In terms of species, there are three findings of particular note. First, there is a particularly high number of orchids (125 taxa across 276 km 2 ), higher than might have been expected and significantly more than the number found in the more extensive Nyanga area above 1200 m elevation (92 taxa across 2181 km 2 ; Clark et al. 2017) and in the Chimanimani mountains (97 taxa across 530 km 2 ; Wursten et al. 2017). There are also a large number − 137 taxa − of pteridophytes (particularly ferns), compared to 136 taxa in the Nyanga area and just 105 in the Chimanimani. This is probably due to the greater moisture levels and frequent clouds in the Bvumba; the area is said to be possibly the richest locality for pteridophytes in southern Africa (Burrows 1990).
The third point of interest, again possibly linked to the high precipitation levels, is the lack of the montane conifer Widdringtonia nodiflora on the Bvumba, a species  that is locally common both in the Nyanga and Chimanimani areas as well as on Mt Mulanje in southern Malawi (where it also occurs with W. whytei) and Mt Gorongosa in Mozambique (Müller et al. 2008). Widdringtonia is generally found on the drier rain-shadow side of these large mountains.

Endemics and taxa of restricted distribution
There is only one taxon known to be endemic to the Bvumba, the epiphytic orchid Aeranthes africana. Noted just twice, it is apparently now not found in its first-recorded location in the forests near Castle Beacon (S. Childes, pers. comm.). This species was recently assessed using IUCN Red List criteria (IUCN 2001) as Critically Endangered (Timberlake 2020, in press) and could be almost extinct, but it is cryptic among the leaves of Podocarpus milanjianus and difficult to see except on fallen branches so may have been overlooked. Six other near-endemic taxa -here defined as taxa found only on nearby montane massifs such as Nyanga, Serra Choa, Stapleford, Banti/Himalaya and Tsetserra, but not including those also found on the Chimanimani or further afield (including Chirinda and Mt Gorongosa) -are found in the Bvumba area (Table 3). Two of them -Aloe cameronii var. bondana and Aloe inyangensis var. kimberleyana − are varieties of more widespread species and thus of lesser taxonomic significance; both have recently been assessed as Least Concern using IUCN criteria (Timberlake, in press). Of the remaining four taxa, Barleria fissimuroides is restricted to just the Bvumba area and a farm just over the border in Mozambique north of Mutare, but three are more widespread, being found from Serra Choa in Mozambique or over the border in Nyanga south to the Himalaya/Tsetserra area or, in the case of Anthospermum zimbabwense, to Mt Pene  (Mapaura and Timberlake 2002), but a small colony has been found on the Bvumba in recent years (Wursten 2007). This species has also been found in Malawi (Viphya, Mt Mulanje) and northern South Africa (Wolkberg Mountains) but is nowhere common. It is surprising that the Bvumba massif has so few endemic or near-endemic taxa, especially when compared to the Chimanimani Mountains (71 endemic taxa, Wursten et al. 2017) or the Nyanga area (19 endemic taxa, revised from Clark et al. 2017). This is obviously not due to under-collection, but possibly the result of the relatively small size of the Bvumba and the preponderance of forest vegetation compared to montane grassland and scrub habitats. Most Nyanga and Chimanimani endemics, for example, are found in montane grassland or montane scrub (Clark et al. 2017;Wursten et al. 2017), habitats poorly represented in the Bvumba and which are also substantially disturbed there.

Introduced and invasive species
There are 123 introduced species present in the Bvumba, a result of the diverse human activities there. Although most of these non-native species are benign, a handful are causing major problems. In contrast to other parts of the Manica Highlands, the Bvumba has a high proportion of introduced species that are garden escapes while the rest of the Manica Highlands is affected more by those from commercial forestry.
Invasive woody invasive species on the Bvumba are typically those also found in other parts of the Manica Highlands, and include the Australian Acacia mearnsii (wattle) and A. melanoxylon (Australian Blackwood) and Pinus patula. Australian Blackwood is a particular problem, taking over tracts of open grassland and areas previously under Protea cultivation (S. Childes, pers. obs.). Other woody but non-commercial species that have naturalised and have invasive potential (based on evidence elsewhere in the region) are Bauhinia variegata var. variegata, Cinnamomum camphora, C. verum, Homalanthus populifolius, Jacaranda mimosifolia, Psidium cattleianum, P. guajava, Sambucus canadensis and Syzygium jambos. Classic invasive shrubs are Lantana camara and Solanum mauritianum, for which there are no easy management solutions, while Cestrum aurantiacum is clearly a local problem. A more recent challenge is Vernonanthura polyanthes -locally called 'Beebush' -which has become rampant since it spread into the Bvumba and Chimanimani areas after its introduction from Brazil to Mozambique (Timberlake et al. 2016b). It spreads rapidly into disturbed and burnt areas that were under wattle, blackwood or eucalyptus (S. Childes. pers. obs.). Clark et al. (2019) postulate that Cyclone Idai (March 2019) might have encouraged the spread of this wind-dispersed species even further afield, although it had earlier also been encountered on the Ribáuè mountains in northern Mozambique (I. Darbyshire, pers. comm. 2017). Indigenous montane forests are being invaded by the garden escap-ees Tradescantia fluminensis and T. zebrina, while perhaps the worst forest invader is Hedychium gardnerianum -a species that can transform the forest understory and has adverse effects on ground-foraging birds such as Orange Thrush (Geokichla gurneyi), Buff-spotted Flufftail (Sarothrura elegans) and Cinnamon Dove (Columba larvata) that need an open understory with leaf litter (S. Childes, pers. obs.).

Limitations and future work
Despite the detailed collections, this list is a compilation. As a result, some taxa may have been accidentally omitted whereas others that are found only below 1200 m elevation may have inadvertently been included. In particular, it should be recognised that there are virtually no records from Serra Vumba on the Mozambique side, an area that needs a more detailed collection. Similarly, the drier western slopes of the Bvumba in Zimbabwe have also been undercollected (J. Burrows, pers. comm.). However, with these provisos we estimate that the list is over 90% complete, suggesting a total indigenous flora of around 1250 taxa, of which approximately 1100 would be native flowering plants.
The remaining data gaps for the botanical inventory of the Manica Highlands are centred on the areas immediately to the north and south of the Bvumba, namely Penhalonga-Stapleford and Banti-Himalaya-Tsetserra, respectively. However, some recent survey work has been conducted on Tsetserra and a list of the endemic and range-restricted species has been compiled for that area (J. Osborne, pers. comm.). In contrast to Nyanga, Chimanimani and Bvumba, which had comprehensive available data with which to work, both areas require detailed botanical collecting before reliable lists can be compiled.