Copper endemism in the Congolese fl ora : a database of copper af fi nity and conservational value of cuprophytes

1Laboratoire d’Ecologie végétale et Biogéochimie, CP 244. Université Libre de Bruxelles, Avenue F.D. Roosevelt 50, BE-1050 Bruxelles, Belgium 2Département des Sciences Agronomiques, Institut Polytechnique Lasalle Beauvais, Rue Pierre Waguet, BP 30313, FR-60026 Beauvais Cedex, France 3Faculté d’Agronomie, Université de Lubumbashi, Campus de la Kasapa, Lubumbashi, D.R.Congo 4Laboratoire d’Ecologie. Université de Liège, Gembloux Agro-Bio Tech, Passage des Déportés 2, BE-5030 Gembloux, Belgium 5Herbarium de l’Université Libre de Bruxelles, CP 169. Université Libre de Bruxelles, Avenue F.D. Roosevelt 50, BE-1050 Bruxelles, Belgium 6 Université de Picardie, Dynamiques des Système Anthropisés (JE 2532), 1 rue de Louvels, FR-80037 Amiens Cedex, France 7 Equally contributing authors *Author for correspondence: pmeerts@ulb.ac.be REVIEW

All over the world the fl ora of metalliferous soils is threatened by human activities, especially mining, and Katanga makes no exception.Actions aimed at preserving metallophyte species are imperative (Whiting et al. 2004).In Katanga, a dozen sites have already been completely destroyed and many others have been profoundly disturbed (Brooks et al. 1992).
Conservation strategies should be based upon a full appreciation of biogeographic originality and conservational value of a given fl ora.Such an appreciation must also rely on a suffi cient taxonomic knowledge (Callmander et al. 2005).For a long time, botanists have relied on the seminal work of Duvigneaud for taxonomy, ecology and distribution of Katangan metallophytes (Duvigneaud 1958, Duvigneaud 1959, Duvigneaud & Denaeyer-De Smet 1960, Duvigneaud & Denaeyer-De Smet 1963) (see Leteinturier and Malaisse (2001) for a historical account of botanical exploration of copper soils in Katanga).Duvigneaud and Denaeyer-De Smet (1963) listed c.250 species on copper sites and consi d ered that the number of copper-endemics might be close to 100 species.More recent explorations (especially by F. Malaisse and co-workers) have steadily increased the number of species recorded to occur on copper-enriched-soil, which is now close to 600 (Leteinturier 2002, F. Malaisse personal obs.).New endemic species have been described (e.g.Robyns 1989, Malaisse & Lecron 1990).However, progress in the botanical exploration of Katanga has also revealed that taxa initially regarded as endemic actually had a broader range also occurring on normal soil.At the same time, taxonomic revisions have synonymised a number of taxa originally des cribed by Duvigneaud as copper-endemics.In the last published reviews some forty species were cited as endemic to the copper habitat (Leteinturier 2002, Whiting et al. 2004).Due to a lack of synthesis of this information, the conservational status of Katangan copper species has not been assessed yet.This has precluded the inclusion of Katangan copper species in the IUCN Red List, a situation which may seriously undermine conservation efforts (for a discussion of appropriate uses of the Red List, see Possingham et al. 2002, Lamoreaux et al. 2003).
In this paper, we aim to establish an inventory of species endemic of copper-enriched soils in Katanga (D.R.Congo), and to assess their current distribution range and conservation.To that end, we critically revise the taxonomic status and ecogeographic distribution of virtually all taxa that have ever been reported to be endemic of copper-enriched soil in Katanga, based upon an extensive revision of the herbarium materials and relevant literature.We also provide a fi rst assessment of their conservation status using the IUCN typology.The percentage of endemic species is compared to other fl oras of metal-enriched soil elsewhere on the earth and the possible origin of the endemic fl ora of copper-enriched soils in Katanga is discussed briefl y.

Geographical setting and climate
The "Katangan Copper Belt" (D.R.Congo) forms a crescent c. 300 km long and 50 km wide extending from Kolwezi in the west to Lubumbashi in the southeast of Katanga province.Mineralised rocks appear as rounded hills, typically a few tenths of meters above the level of the surrounding non mineralised areas.Most copper hills are typically a few tenths of hectares in area, rarely exceeding 1 km² (personal obs.).The distance between nearby hills varies enormously across the Copper Belt from less than a few hundred meters to more than 20 km.The total number of mineralised hill- Faucon et al., Copper endemics of Katanga ocks is not known precisely, but > 200 sites are recorded in the database of the Gécamines (personal obs.).Less than one hundred hills have been explored botanically on at least one occasion (Leteinturier et al. 1999, Leteinturier 2002).
The Katangan Group is of Upper Cambrian age (i.e. over 620 my old).It comprises three large series: Upper Kundelungu, Lower Kundelungu and Roan (François 1973).Copper mineralization in Katanga province is found in the Roan Series comprising calcareous rocks with dark minerals, dolomitic schist, cellular siliceous rocks, fl aky siliceous rocks, stratifi ed dolomites and argillaceous talc.
The climate is humid subtropical (Köppen-Geiger: CWa) and tempered by the relatively high elevation (c.1300 m a.s.l.).There is one rainy season (November to March), one dry season (May to September) and two transition months (October and April).Annual rainfall is about 1300 mm of which 1200 mm fall during the rainy season.Mean annual temperature is about 20°C.Temperature is lowest at the beginning of the dry season (15-17°C).September and October are usually the warmest months with daily maxima of about 31-33°C.
Three sources of information have been used to construct a database: published literature, herbarium materials and unpublished fi eld observations by the authors (often supported by a herbarium voucher specimen).First, a working list of putative endemics taxa was compiled based on all pu blished papers on the fl ora of Katangan copper hills with explicit indications on plant affi nity with copper soil (Robyns 1932, Duvigneaud 1958, Duvigneaud 1959, Duvigneaud & Denaeyer-De Smet 1960, Duvigneaud & Denaeyer-De Smet 1963, Malaisse 1983, Brooks & Malaisse 1990, Leteinturier & Malaisse 1999, Leteinturier 2002).All species that had been reported as endemic of or having high affi nity for Curich substrate by at least one of those authorities were considered.All species classifi ed as "absolute cuprophytes", "local cuprophytes" and "cuprophiles" by Duvigneaud (loc.cit.) have been considered.The second step was to validate putative endemics.To that end, each species in the fi rst list was critically assessed both from a taxonomic and phytogeographic point of view.

Nomenclatural and taxonomic revision
The following standard fl oras have been used: Flore d'Afrique centrale (Multiple authors 1948(Multiple authors -1962;;1967-1971;1972-1996;1997-), Flora Zambesiaca (Multiple authors 1960-) and Flora of Tropical East Africa (Multiple authors 1952-).For taxa that have not been treated yet in these fl oras, the most recent taxonomic revisions and monographs for Central Africa have been used.In all cases, the protologue was consulted.The following electronic resources have also been used: Botany & Plant Science of ALUKA (http://www.aluka.org/action/showDiscipline?sa=bot), the International Plant Names Index (IPNI) (http://www.ipni.org), the African Flowering Plants Database (http://www.ville-ge.ch/musinfo/bd/cjb/africa/).The most recently accepted name has been retained.Some putative endemics were rejected because they have been synonymised with taxa with a broader distribution range of copper-enriched substrates.In a limited number of cases, different taxonomic revisions have expressed confl ic-ting views as to the taxonomic value of some taxa occurring on Cu soil.Such cases will be examined in the discussion.

Phytogeographic revision
For all taxa taxonomically validated in the fi rst list, the proportion of localities on Cu soil vs. non Cu soils was determined based on an extensive survey of published literature and herbarium collections.For all these species, virtually all herbarium materials kept in BR, BRLU, BRVU, GENT and Kipopo (Katanga) have been examined for location and ecology.More than 1900 herbarium specimens have been seen.Additional data came from our own observations (personal obs.), especially since 2005 and from the original fi eld notes of P. Duvigneaud (kept in BRLU).Particular attention was paid to records of putative endemics on non Cu substrates (especially steppic savanna on Kalahari sands of the Katangan highlands ("Hauts-Plateaux katangais"), nonmetalliferous rock outcrops, lateritic crust, woodland miombo or other metalliferous habitats: Zn, Pb, Mn).Occurrence on Cu-enriched soil was accepted if it was explicitly mentioned on the herbarium label.In very few cases, occurrence on Cu soil was inferred from locality name (e.g."Mine de l'Etoile") even if no explicit ecological data were recorded by the collector.Affi nity for Cu soil was calculated as the proportion of records on Cu soil and categorised as follows: < 75%, 75-99%, 100%.Taxa with 100% of records on Cu soil are here regarded as strict endemics of Katangan copper soil (i.e.absolute metallophytes).Our defi nition of endemism is somewhat more restrictive than in the recent Californian serpentine database (Safford 2005) which accepted as serpentine ende mics all species with > 95% locations on serpentine.A second list of broad endemics was established comprising species having > 75% of localities on Cu-soil and a few species with a total of only 3 known localities, two of which (66%) are on Cu soils.

Ecology of endemics
The habitat of each species was determined based on herbarium collections and published literature completed by our recent observations in the fi eld.Three types of habitats have been distinguished.Primary habitats are natural, undisturbed plant communities, including steppic savannah, rocky steppe and scrub comprising mostly perennial species.Secondary habitats are substrates which have been disturbed by man.Two types of secondary habitats have been distinguished depending on the source of metallic contamination: i) mine deposits (mostly mineralised rock debris) disturbed by artisanal or industrial mining activities, ii) soil contaminated by atmospheric fallout from copper smelters (Mbenza et al. 1989).

IUCN red list status
The conservational status of strict and broad endemics was assessed using IUCN criteria in April 2008.In this assessment, the current or potential threats on populations were assessed as follows.Mining maps of Katanga province ("cadastre minier") allowed us to determine the status of the sites (i.e.currently mined, mining planned, no mining planned).Personal observations in the fi eld allowed us to assess current extent of habitat destruction and fragmentation.The main mining companies operating in Katanga agreed to communicate at least basic information about their operation plan schedule.Data from the ecogeographical revision have been combined with the current and predictable impacts of disturbance by mining activities to generate the Red List status of each taxon using IUCN (2001) criteria.B criterion was used to assess IUCN status using the geographic range in the form of either B1 (extent of occurrence) or B2 (area of occupancy).B1 and B2 were assessed by subcriteria a) and b) or c) by examining decline, fl uctuations, fragmentation of populations and of occurrence and of occupancy area.

RESULTS
The inventory (tables 1-3) Strict endemics -Thirty-two taxa (26 species, three subspecies, three varieties) are confi rmed as strict Katangan Cuendemics, i.e. with 100% of records on Cu-soil in Katanga (table 1).Ten of these (30%) are known only from the type locality.Only seven are known from > 10 localities.For four Taxa were rejected for one or several of the following three reasons: 1) taxon discovered off copper and with < 75% populations on copper soil; for such taxa, distribution range is given; 2) taxon merged with a broadly distributed species; 3) name rejected (unpublished name or nomen nudum).K = Katanga; BC = Bas-Congo; ZA = Zambia; AO = Angola; TZ = Tanzania; RW = Rwanda; MI = Malawi; ZI = Zimbabwe; MZ = Mozambique; TA tropical Africa.Rejected taxa -Fifty-eight other names which had been previously used for supposed Katangan Cu-endemics by at least one literature reference are rejected here for one or several of the following reasons (table 3).First, a number of supposed endemics of low taxonomic value have been merged with taxa with a broader distribution range (e.g.Xerophyta barbarae synonymised with the widespread X. equisetoides).
A second group comprises valid taxa which have been discovered in a relatively large number of sites on non mineralized soil (e.g.Gladiolus tshombeanus).Finally, a number of names are nomina nuda or have never been published.
Regional Cu-indicators -A distinct group comprises species with < 75% of their localities on Cu-soil when considering their whole distributional range, though occurring (almost) exclusively on Cu-soil in the Katangan copperbelt.These can be referred to as "regional Cu-indicators" ("local metallophytes" in Duvigneaud & Denaeyer-De Smet (1963)).Some of the most characteristic of these include Anisopappus davyi, Microchloa altera, Crepidorhopalon tenuis, Haumaniastrum katangense.

Ecology
Of the 32 strict endemics, 25 occur solely in primary communities.Three occur in primary communities and secon dary communities on reworked substrate (Acalypha cupricola, Crotalaria cobalticola, Bulbostylis fusiformis; possibly also Commelina zigzag and Silene cobalticola) and fi ve species (Haumaniastrum robertii, Crepidorhopalon perennis, Vigna dolomitica, Gutenbergia pubescens, Faroa chalcophila) are currently known only (or with few exceptions) from secondary communities on reworked mine debris, old mine tailings or otherwise disturbed substrates (table 1).
Of the 24 broad endemics, twenty apparently occur only in primary communities when present on copper-enriched soils, two occur in both primary and secondary communities and two species apparently only in secondary communities on disturbed substrate (Bulbostylis cupricola, B. pseudoperennis) (table 2).

IUCN status
IUCN red list status has been determined for 47 of the 58 (strict + broad) endemics.Of the "strict endemics", three species are extinct (Acalypha dikuluwensis, Basananthe cupricola, Vernonia ledocteana), 21 taxa are critically endangered, three are endangered, one is vulnerable and four are data defi cient (table 1).Of the broad endemics, one is critically endangered, twelve are endangered, six are vulnerable, and fi ve are data defi cient (table 2).

Taxonomic diffi culties
Thirty-one taxa (26 species, two subspecies and three varieties) are strict endemics of Cu/Co soil in Katanga.This number is much smaller than the estimation of c. 100 endemics proposed by Duvigneaud & Denaeyer-De Smet (1963: 216).This discrepancy has two explanations.Firstly, due to improved botanical exploration of Katanga many supposed endemics have eventually been found on normal soil.Secondly, recent taxonomic revisions have often used a broader species concept than that of Duvigneaud.Many taxa described by Duvigneaud from copper soil are now considered as having low taxonomic value (Ernst 1974).Many species in the Katangan copper fl ora show extensive morphological variation, with vica riant morphotypes occurring in different hills.Some of the most remarkable local variants on Cu-soil were given formal taxonomic recognition by Duvigneaud and co-workers (e.g. in the genera Ocimum, Gladiolus, Pandiaka, Ipomoea, …).Duvigneaud himself (loc.cit.) admitted that endemism on copper soil was mostly expressed at low taxonomic rank.However, the taxonomic treatment of many genera remains controversial.The genus Ocimum (= Becium) (Lamiaceae) offers a striking example.Sebald (1989) argued that the Cu-endemics described by Duvigneaud should best be treated as varieties of a single polymorphic taxon (Ocimum grandifl orum).A few years later, however, Paton (1992Paton ( , 1995) ) expressed opposing views, arguing that species rank was more appropriate.Lebrun & Stork (1997) followed Sebald (1989).The Katangan copper fl ora offers a striking example of the infl uence of the species concept on conservation stra tegies (Agapow et al. 2004, Isaac et al. 2004).Irrespective of taxonomic implications, it is clear that the Katangan copper fl ora offers outstanding examples of microevolution at work, which should be investigated with molecular tools.

Low proportion of strict endemics
The proportion of endemic species in the fl ora of the Katangan copper belt cannot be calculated precisely, because the total number of species is not known precisely.Duvigneaud & Denaeyer-De Smet (1963) recorded c. 230 species on Cu-soils, whilst Leteinturier (2002) listed 538 species.This number is still increa sing as botanical exploration improves (personal obs.).Based upon a rough estimation of c. 600 species, the proportion of strict endemics is in the order of 5% (32/600).How does this fi gure compare with the proportion of endemics in other metallicolous fl oras (table 4)?Serpentine soil in Cuba and New Caledonia, being oceanic islands with a long history of geographical isolation are admittedly not much relevant to the comparison.Serpentine soil in California and Italy have 23% and 29% of endemics, respectively, i.e. much higher values compared to Katanga.
What therefore can explain the relatively low proportion of endemic species in Katangese Cu-enriched soils?Age of exposure is an important factor to explain proportion of endemics in geochemical anomalies (Harrison et al. 2004).
Copper mineralization in Katanga dates from late Cambrian (c.620 my (François 1973)).However, Cu-rich rocks have been exposed to plant colonization for a much shorter period.Recent data indicates 2-3 my as a likely age for Katangese Cu outcrops (S.Decrée & J. Yans personal comm.).This fi gure is similar to the age of exposure of Californian serpentine outcrops with the lowest percentages of endemics (Harrison et al. 2004).Vegetation disturbance due to palaeoclimatic changes may also be relevant.During the Holocene, tropical Africa experienced dramatic climate fl uctuations, including a dry-cool period c. 18,000 y BP (Scott 1984, Van Zinderen et al. 1988, Vincens 1991, Campbell et al. 1996).Therefore, much of the steppic fl ora may have recolonised copper hills when climate became warmer and moister, and before miombo forest eventually occupied most of upper Katanga (Scott 1984, Vincens et al. 2003, Vincens et al. 2005).Evolutionary divergence between populations isolated on the copper hills may therefore be recent, thus explaining the low percentage of endemic species.Signifi cantly, the Zimbabwean Great Dyke (serpentine), which may have similar palaeoclimatic history as Katanga also has a low percentage of endemics (7%).
Pl. Ecol. Evol. 143 (1), 2010 Another factor that may also account for the low proportion of endemism is the relatively low total area of mineralised soils in Katanga.Based on a typical site area of a few tenths of hectares the total area of the copper habitat may not exceed 100 km².This is at least one order of magnitude lower compared to the thousands of km² of serpentine soil existing in California, Italy, Cuba or New Caledonia (Brooks 1987).
Finally, ecological isolation and selective forces acting upon plant populations on Cu-enriched soil may have been overestimated.Contrary to serpentine soil, Katangan Cu-soils are relatively rich in nutrients (P, Ca, Mg) (Faucon et al. 2009b) and Cu toxicity may be mitigated by interactions with other metals in the soil (Faucon et al. 2009a).This might also explain the lower rates of evolutionary divergence between populations on Cu-soils and their counterparts on normal soil.

The case for neoendemism
Many if not all endemic species have close relatives on normal soil with a broader Zambezian distribution (Brooks & Malaisse 1985).Few if any endemic species appear to be taxonomically very isolated.A number of recent taxonomic revisions have emphasized the close morphological resemblance between a number of Cu-endemics and their more widespread counterparts (e.g.Vigna dolomitica and V. reticulata (Maxted et al. 2004); Crotalaria peschiana and C. subcaespitosa (Polhill 1982); Acalypha cupricola and A. fuscescens; A. dikuluwensis and A. clu tioides (Levin et al. 2007); Silene cobalticola and S. burchellii (Malaisse et al. 1983)).Interestingly, S. burchellii possesses a copper ecotype which is intermediate between S. burchellii (on normal soil) and the narrow endemic S. cobalticola for both morphology and copper tolerance (Baker et al. 1983, Brooks & Malaisse 1985).This situation may indicate a derivative/ progenitor relationship and a scenario of recent speciation in conditions of strong ecological isolation (Kruckeberg 1986, Macnair & Gardner 1998, Rajakaruna 2004, Rajakaruna & Whitton 2004).
Strict endemic species which occur (almost) only in secondary, man-made habitats (reworked debris and mine deposits) may conceivably have a (very) recent origin.Two of the most striking examples are Haumaniastrum robertii (related to H. katangense) and Crepidorhopalon perennis (related to C. tenuis).Molecular tools should be used to test if these species have recently diverged from their counterparts in the new habitats created by mining activity.Of note is that Mimulus cupriphilus, the only copper endemic outside SC Africa, has been shown to be a recently evolved species of copper mines in California (Macnair 1988, Macnair & Gardner 1998).Morphological vicariance among populations of the same species in different sites also suggests that active speciation processes may be at work in the Katangan copperbelt.This observation also supports the hypothesis that many, if not all, copper endemics may qualify as neoendemics (Malaisse 1983, Brooks & Malaisse 1985, Brooks & Malaisse 1990).Evolutionary divergence between endemic metallophytes and their broader distributed counterparts should be investigated with molecular tools.

Relictual endemism?
For most broad endemics, populations off copper soil are located in steppic savannahs on Kalahari sands on the Katangese highlands ("Hauts-Plateaux katangais"), i.e. often more than 100 km away from the copperbelt.This suggests that some of the endemic species of the Copperbelt are relictual endemics, i.e. species that were once more widespread when steppic savannahs occupied most of Katanga (Duvigneaud & Denaeyer-De Smet 1963, Malaisse 1983, Brooks & Malaisse 1985) and which have been fragmented by forest expansion during the Holocene.

Ecology of endemics
Katangan copper hills are ecologically complex, encompassing a broad range of habitats (Duvigneaud & Denaeyer-De Smet 1963).In particular, they comprise natural habitats that are poorly represented on non mineralised substrates in Katanga, including rocky, non fo rested habitats (Duvigneaud 1958).Those rocks typically occupy the top of copper-hills and harbour a distinctive fl ora, rich in chasmophytic species.However, as most these rocks have been strongly leached by rain, their copper content within rooting depth is most often low (Duvigneaud & Denaeyer-De Smet 1963).Therefore, a number of species res tricted to copper hills may not be highly metal tolerant and are actually not genuine metallophytes (e.g.Aeollanthus saxatilis and Dissotis derriksiana), a fact already acknowledged by Duvigneaud & Denaeyer-De Smet (1963).

Conservation
The endemic copper fl ora of Katanga is obviously at risk.Sixty-seven percent of endemic taxa qualify as critically endangered and 9% are already extinct (fi g. 1).It is clear that conservation efforts should not be focused solely on strict Faucon et al., Copper endemics of Katanga endemics.Species of interest to conservation should also include the broad endemics as defi ned in this work i.e. species with more than 75% of their populations in the copperbelt.Mining is the sole anthropogenic cause of extinction at present.Twelve copper hills have already been totally destroyed and c. forty (i.e.50% of sites botanically explored) have been profoundly altered (Leteinturier 2002, Meersseman unpublished results).The steady increase of copper price on international markets between 2000 and 2008 has prompted new investments in the mines and an increasing number of sites are being destroyed rapidly, though market versatility may provide some respite.
Different ecological groups have to be distinguished among endemic species, which may require different conservation strategies.A fi rst group comprises long-lived, slow growing species (mostly geophytes, geofrutex and chamaephytes) restricted to primary steppic savannah communities (Loudetia-Cryptosepalum steppic savannah and Xerophyta stone-packed steppe in Duvigneaud & Denaeyer-De Smet 1963) associated to stabilised substrates and with low colonization ability.Those species are not able to rapidly colonise recent mine debris and are thus highly vulnerable to habitat destruction by surface mining.Conservation of intact plant communities is arguably the best option for such species.Ex situ conservation is probably most diffi cult, requiring ecosystem reconstruction including translocation of substantial amounts of undisturbed topsoil.
A second group mostly comprises annal species occurring in pioneer open communities.These species can rapidly colonize the new substrates (mining debris and barren mineral areas, tailings) created by mining activities or soil contaminated by atmospheric fallout in the vicinity of metal smelters.A number of those species (e.g.Crepidorhopalon tenuis) are actually expanding their range on the huge areas of mine debris recently reworked by mining (Faucon et al. 2009b).These examples suggest that IUCN criteria for the establishment of conservation status of threatened species could be refi ned by taking into account the ability to colonize anthropogenic habitat.Those species may be easier to conserve in secondary habitats or in restored communities on reworked mineralized substrates.In other metallicolous ha bitats, it has been shown that specialised species may colonize secondary man-made habitats (Bizoux et al. 2004) and, even, exhibit higher fi tness on those secondary habitats (Bizoux et al. 2008, Faucon et al. 2009b).Translocation of seed should be attempted in particular for Bulbostylis fusiformis, Crepidorhopalon perennis, Crotalaria cobalticola, Faroa chalcophila, Gutenbergia pubescens, Haumaniastrum robertii and Vigna dolomitica.Why some of these early successional species have such a narrow distribution range (e.g. one site for Gutenbergia pubescens and Crepidorhopalon perennis) is unclear (restricted dispersal or niche constraints) and merits investigation.

Need for further taxonomic and phytogeographic research
The Katangan copper fl ora is still insuffi ciently known.Firstly, it is signifi cant that a number of taxa in our list are known only from the type specimen and a mention in Duvigneaud & Denaeyer-De Smet (1963).Some of them may be genuinely rare, but others may have been overlooked by recent collectors.Secondly, improved exploration has allowed describing several new endemic species in the last decades.Nine strict endemics have been described after Duvigneaud's work.Much herbarium material has been collected in the Katangan Copperbelt over the last decade especially by one of us (FM).This material may include new taxa still awaiting description.Conversely, for the same reason, several species previously thought to be strict copper ende mics have been discovered off copper.Thirdly, biosystematic studies are badly needed to clarify the complex taxonomy of many genera in the Zambezian region (e.g.Ocimum: Sebald 1989, Paton 1995).Taxonomic diffi culties are greatest for genera and families for which recent systematic revisions are lacking for D.R.Congo (e.g.Acanthaceae, Amaranthaceae, Anthericaceae, Commelinaceae, Cyperaceae, Lamiaceae, Poaceae) and with much herbarium materials unidentifi ed to date (P.Meerts personal obs.).

CONCLUSIONS
The occurrence of natural copper-enriched substrates colonised by natural plant communities over signifi cant areas is exceptional on the earth's surface.In Katanga 32 taxa are strict endemics of such habitats, and the Katangan copper hills offer striking examples of ongoing microevolutionary processes.Our phytogeographic and taxonomic revision represents a step for the conservation of Congolese copper fl ora.However, biosystematic and molecular-based studies are urgently needed to clarify phylogenetic relationships between Cu-endemics and related species, explore phylogeographic scenarios and reconsider species limits.
Most endemics qualify as critically endangered.The conservation of this biodiversity may appear to confl ict with economic development and will require concerted efforts by ecologists, mining companies and stakeholders.This inventory represents a fi rst step towards an integrated conservation plan for that fascinating fl ora.

Figure 1 -
Figure 1 -Frequency distribution of IUCN categories of plant taxa strictly endemic of Cu-rich soil in Katanga.EX: extinct, CR: critically endangered, EN: endangered, VU: vulnerable, DD: data defi cient

Table 1 -Plant taxa strictly endemic of Cu-rich soil in Katanga, D.R.Congo (absolute metallophytes), with their IUCN status, habitat and number of sites.
EX: extinct, CR: critically endangered, EN: endangered, VU: vulnerable, DD: data defi cient.P: Primary habitats.S: Secondary habitats.S1: with substrate (often mine debris) disturbed and reworked by mining industry, S2: contaminated by atmospheric fallout from metal smelter.

Table 2 -Plant taxa with very high affi nity to Cu/Co rich substrates in Katanga, D.R.Congo (broad endemics), with their IUCN status, habitat and number of sites.
Broad endemics: 75-99% of localities on Cu-soil in Katanga or a single locality on Cu-soil.CR: critically endangered, EN: endangered, VU: vulnerable, DD: data defi cient, NE: not evaluated.P: Primary habitats; S: Secondary habitats, S1: with substrate (often mine debris) disturbed and reworked by mining activities, S2: soil contaminated by atmospheric fallout from an ore-smelter.

Table 3 (continued) -Names formerly used for supposed Cu-endemics in Katanga not validated in the present study. taxa
(Cyanotis cupricola, Digitaria nitens subsp.festucoides, Loudetia kagerensis subsp.jubata, Pandiaka metallorum), little information is available apart from the protologue and we feel that their taxonomic value needs critical re-evaluation.Two of these are known only from the type specimen (Digitaria nitens subsp.festucoides, Loudetia kagerensis subsp.jubata).Ascolepis metallorum also on Zn/Pb and Mn-rich soil).Bulbostylis cupricola is actually restricted to Cu-soil, but its distribution range extends to the copperbelt of Zambia and it is thus not an endemic of D.R.Congo.The other broad endemics have a few occurrences (sometimes a single) on non-metalliferous soil.In many cases, these "odd" localities are not situated close to Cu-sites, most often occurring in the steppic vegetations of Katangan Highlands on Kalahari sands ("Hauts Plateaux katangais").Most of these broad endemics are known from 10 localities or more.For two species (Ocimum monocotyloides, Monade nium pseu- doracemosum) the taxonomic status of collections from Cusoil is unclear and requires further research.For Ocimum metallorum and Xerophyta demeesmaekeriana, recent distributional data are insuffi cient.
Faucon et al., Copper endemics of Katanga

Table 4 -Endemism of metalliferous outcrops: comparison of six regions.
For each region, factors that might infl uence percentage of endemics are reported.