TOWARDS A PROFILE OF AN ENDEMIC MOUNTAIN FYNBOS FLORA: IMPLICATIONS FOR CONSERVATION

D. J. McDonald

R. M. Cowling

(Rmsiwd 20 Novembcr 1993; revised version rccdvtd 27 May 1994; accept& 8 June 1994)

Keywork Cape Flotistic Region, mnservation, en- demism, fynbos, Langeberg mountains.

INTRODUCTION High costs in terms of money and manpower determine that many conservation decisions have to be bawd on limited data from often very large areas (Margtda & Stein. 1989; Kuoin & Gaston, 1993). This apjdies to mountain environments in the Cape Floristic Region (CFR) where the chattas of extensive biological sw veys in the future are slim. Investigation of patterns in endemic floras. like research on rare species (Kuoin & Gaston, 1993), pmvides one option for obtaining data for more infomwl, rational, conservation management decisions (van Wilgen d al., 1992). For exampk, spy& with narrow distribution ranges (i.e. endemic to relatively small areas) are used to daine cmtrer of endemism or core areas for the delimitation of reseIws

C.Terbor~ & Winter, 1983; Gentry, 1986; Rebelo & Siedried. 1990: Saetersdal et a.!. 19931. It is therefore ., important to know whether or not a” endem% flora constitutes B random arrrmblage with respect to taxon- omy, habitat preference and biological attributes (Ktuckeberg Br Rabinowitz, 1985; Major, 1988). If not, then the pauliar charactetistii of the emiemk flora should be considwed when formulating mattagement policies for thzii maintenance in nature Gservcs.~

The CFR has extmmiiiarily high kvels of endemism (Goldblatt, 1978; Cowling et ol., 1992). This is mostly associated with fynbas, a tire-prone, sckropbyllous evergn~~~ shrubland (Muir, lm Kruger, 1979). Studies of lowland floras of the Cape Flo&ic Region (Cowling & Holmes. 1992: Cowline PI al.. 1992) have show0 (I) that certain famiiies havepmpo;liotu& tttotz end& that others: (2) that emiemico are not rattdom am- blsges with respect to biological attributes, i.e. they are most likely to be dwarf to low non-sprouting shrubs with .mtdiipmed, soil-stand seed andh mkrosym- bioat-msdited nutrient uptake; and (3) most et&mica are edaphk speciausts.

The mountains of the CFR are topogmphkally and climatically complex and this stody focuses speci&ally cm a montane iynbos flora. The main aim is to identi& correktes of endemism useful for conscnation. The role of habitat is investigated (are endemics more abun- Ant than extrated by cbattce is wrticular habitats?) and the g&h form, rcgeiwratio~ strategy and & dispersal correlates of endemism analysed. Tbcse latter thm factors are important for developing policies and guide-lines for the tire twtagemettt of fj’ttbos, the most bttportant tool availabk to managers for maiotaittlng biodivenity in this extraordittarily s&es-rich region (van wilgcn et al., 1992).

The southern Langeberg (Fig. I) is one of the west-east tmding mountain ranges. in the eastern me of the Cape Fold Belt. Although gradiatts are impressively

Fig. 1. The location of the so-them Langeberg mountaias (D in relation to the A&has Plain and other mountains in the Cap ). Cape ce%tm ofendemism @Ft.% Weimarck, 1941) are shown in the inxt: BR, BredasdorpRiversdale (from

Cowling et a/. (1992)); L. Lang&erg; KM, Karoo Mountain: KR, Knysna; NW, North Western; SE, South Eawm: SW. South Westem.

stew. altitudes are moderate: the highest waks reach a&d 1700 m. The Langeberg c&i& mainly of ortboouanzites and shales of the Table Mountain Groud(Fig. 2). By global standards, soil3 in the Lru@erg are extremely nutrient-poor (Groves e* al., 1983).

The study area lies in the non-seasonal rainfall zune of the CFR. Rainfall is associated with circumpolar westerly fronts (highest frequency in winter) and post- frontal conditions associated with advection of cool moist air over the warm Indian Ocean (highest fre- quency in spring and autumn). Topography and aspect exert a strong influence on (climate. The north slopes (rainshadow) of the sauthem Lang&erg rec.& between 200 and 400 mm per amuun (Fig. 2); the lower south slopes receive 600-800 mm per amuun and those at higher altitude (loo0 m) between 700 and 800 mm per anmuu. The high peaks (> 1200 m) receive an estimated mean annual precipitation of 120+1400 mm (Dent et o/.. 1987). much of which is accounted for bv ,. high-altitude mist.

The southern Langeberg lies at mid-latitudes, result- ina in a distinct diRerewe in insolation between the s&h- and north-facing slopes, particularly in winter. Temperatures are generally mild. On the south slopes, highest mean daily maximum temperature occurs in January (29.4’C) and lowest mean daily minimum in July (6.6’0 (Fig. 2). Highest mean daily maximum on the lower north slopes is in February (3O@X!) with a winter minimum of 2YC. With au environmental law rate of O&‘C tax 109 m (Cowling, 1984) a d&ease in temparatu~e of approximately 9.45Dc on south-facing slopes and 7.8’C an north-facing slopes at an altitude of 1700 m a.s.1. is expected.

The predominant fynbos vegetation on the southern Lang&erg is typical of the coastal mountains of the southern Cape both in structure and levels of species

richness (Bond, 1981: Campbell. 198% This is man- aged prir&ily for water p&uction a& nature conw- vation (van Wilaen er I% 1992). Until receutlv mountain catchme& were bumt as ‘block bums’ on H I>20 year rotation: the catchments were burnt in late summer and autumn to promote regeneration of the fynbos and prevent extensive wildfires (van Wilgan et al., 1992). The current approach is implementation of ‘adaptive interference tie management’, a flexible sys ten-which ranges from an al&t natural bu&g regime with some interference to a more rigid can- trolled compartment burning system (Seydack, 1992).

METHODS Data collection A check-list of the flora of the southern Langebwg comprising 1228 species and subspwiiic taxa was corn- piled from two main sources: the catalogue of plant species of the Cape Floristuc R&n (Bond & Gold- blatt, 1984) and the plant coUections and identifications from 304 plots sampled by the iirst author during phytwociological studies of the southern Langeberg fMcDonald. 1993n.6.cl. Additional unoubli~dxd infor- mation was obtained on tutdescribed endemic taxa from taxonomists specializing in the systamatics of the Cape flora. Endemics refer to spa&s con&ted to tha southern Langeberg mount&s (Fig. I; Appendix I), whiih are part of the Langeb+rg Euden&C!entre as originally delimited by Weimarck (1941).

&ta bn the growth form, regeneration strategy, dis- persal q ude and preferred habitat for each spa&s wen obtained from a variety of sounxa inchtdiig taxonomic revisions, herbarium sp&mens, collecting rag&m and field notes. Save” growth-form categories were recog- “ized: forbs (FO), geophytes (G), gmrdnoids (HO),

3

4 D. J. McDonald, R h4. Cowling

low shrubs 5 I.0 m (LSH), mid-high shrubs 14X?.O m (MSH). tall shrubs >2 m (TSH) and trees (T - recorded only among noneodemics). In thC analysis of growth-form associated with endemism, ferns (28

not sharp and could overlap. For the purposes of cat* gorical analysis, each species was placed in only one habitat category. where it was most cotmoo”, although it could occur in more than one.

species), vines, liaoes and epiphytes (19 species) were excluded because of low numbers in the flora, Four dis- persal modes were recognized: wind ON), vertebrate (V), ant (A), and passive~unknown (P). Seeds dispersed by ballistic means are not moved long distances from the parent plants. Thus, depending on whether or not the seeds had elaiosomes for secondary dispersal by ants, the plant species were assigned to ant-dispersed or pessive/ynknowo dispersal categories. Regeneration strategies were reduced to two: whether plants resproot or not after fire. Tbe categories used are consistent with those used by Cowling era/. (1992).

TmomemIc aspee& of endemism To determine whether Lang&erg endemics are a taxo- nomically heterogeneous group or whcthcr certain taxe have a higher than expected probability of being endemic, local endemism in the plant families of the southern Lang&erg flora was analysed using contin- gency tables. Chi-square analysis was used to test the hypothesis that the frequency of eodemics in a ‘test family’ would not be significantly diRerent from the frequency for the remaining flora (icdependent sample). Nine f&lies were c”nsid&ed (out “E e total of iO4) where the expected frequency of endemics in the chi- square analysis was 25 (Table I).

Habitat aspects We recognized four broad habitat categories. based on climate and topographic position (see Study Area): (i) Mesic South Slopes (MSS), 500-900 m a.s.1.. (ii) High-altitude Wet St&es (HAW). 901-1710 m a.s.1. on south-facing aspects. (iii) High-altitude Dry Slopes (HAD), 901-1710 to a.s.1. on north-facing slopes: and (iv) Arid Slopes (AR), 50&9UO m a.s.1. on north-facing slows adioinins the Little Karoo lFie I).

a crude e&ate of area of the f&different habi- tats was obtained by measuring the areas delimited on I: 50 000 topographic maps, using a LI-COR optical area measu&~ apparat& (LI-C!OR Inc., Lit&In, Nebraska. USA). The error in determininc surface area on sloping, mountainous terrain using &is method is within acceptable limits (H. RUther. pers. comm.). Althcugb the habitat ‘zones’ were treated as distinctly separate, it should be stressed that the boundaries were

Given the areas of the broad habitats measured as proportional to the total study area, the number of en- demics expected for each area was calcolated es a pro- portion of the total “umber of eodemics. The expected frequency was then compared with the observed fre- quency of endemics by the chi-squared test.

A quantitative index of endemicity (13 wes calculated for the southern Langeber8, for comparison with other areas in the CFR. -Byk&‘s index-of endemicity is I- = E/E.. where E, is the factual oerceotaze endemism 6.e. I i&% for thi southern Lao&erg) &la E. is the normal percentage endemism read from the notnogram, a log-Ion olot of area aaainst uercentage endemism (Fig- 51-i; Major, 1988)-deriveh from iykov’s date (Bykov. I979 in Major, 1988). A value 01 1. = I indi- cates that an area has the normal expected degree of endemidity whereas areas with I.>1 have higher than “onnal endemicity and areas with 1~1 have lower than normal endemicity (Major, 1988). -

BlaleglfPl aspects Biological aspects (growth form, regeneration strategy. dispersal mode) of endemism were also analysed using simple two-way contingency tables of endemic category by biological trait. Chi-square analysis was used to test for significant differences in the frequency of each trait with respect to endemic category.

RESULTS

Levels of endemism From the speciesarea relationships for the CFR (Cowling Ed a/.. 1992). the predicted number of species for the Langeberg would be 2542 if the mountain reoge was located in the south-west aad 1080 if located in the south-east. It lies in a more-or-less intermediate pai- tion, so the “umber of species and subspecies recorded (1228) is about what could lx expected. Of these, I60 (I 3.02%) are endemic, which is a ouch hiahn mooor- - . . tion than the 1.5% obtained from the endemicity nomogram given by Major (1988). The Bykov index of endemicity (lc) for the southern Lang&erg Aore is 8.68, which also indicates a much higher degree of endemic- ity than would be expected for the area (Table I). Bykov indices for other areas in the CFR are given for comparison (Table I).

The total flora of the southern Langeberg is spread over 366 genera and I04 families, of which one family, the monotypic Geissolomataceae (Dahlgreo & van Wyk, 1988) and one monotypic genus, Lungebe@ (Asteraceae), are endemic (And&erg, 1991). Twenty- five families (24.03%) have at least one endemic species (Table 2: Appndiix 1) and nine families (8.65%) each have live or more. These nioe families account for I32 endemics (72% of the total). Erica has 48 endemic species, the highest “umber in any ooe geous.

Family Nomendemic Local endemic % ’ J Simikina

61

ii 6

18 48

1: 19 9

19 49

155

:‘: 17

::

2z 1068

1:

2 5

14 I

16 I I 0

: I60

(Kn.0)

g:;; (35.3)

g:::; (30.5, (25.0)

1:::;; (15.4) (14.3)

$3 (12.5) (11.8)

11:3

$3 (8.3) (6.7) W) g:; (0) (0)

(19.1) 113.0)

- - 7794 l * *

- -

- 4.85 7

0.09 ii - -

0.0s -

NS - -

0.40 NS 3.42 l

- I.17 ii

- -

7.79 - l *

5.38 * - -

The Ericaceae and Restionaceae are the only families in the southern Lang&erg flora in whiih endemics were signilicantly over-represented relative to the remaining flora in an independent sample. Levels of endemkm in the Fabaceae, Iridaaae, Orchidace&? and Fmteaceae relkcted those of the entire southern Lang&erg flora (independent sample) whereas the numbers of mdemics were significantly low in the Asteracex, Cyperarrae and Poaceae (Table 2).

tie proponions of each habitat to the t&l siudy area g = 41.76, p c oaol). The m&c south slopes Gccupy the largest area and have the highat number of speies foll&d by the arid, hi.+altih& wet and high-altitude drv habitats. The high-altitude wet habitats are oromi- t&t in having thehigbest conantration of e&&c species while only r&dig third in species richness. Much lower numbers of emkmii spsies (c. half as

Habitat Area(b) %Study Total Endemics J Significant area spds

OhxNcd % ’ EXpeti

M&c south slo,xs 76 115 43.5 674 67 (42) 0.10 NS High-altitude wet slopes 29380 168 211 51 032) :i 21.64 l * *

High-altitude dry slopes 21230 12.1 113 0.98 NS Arid north slops 48 132 27.5 230 :: $;I E 6.57 l

Total I74857 100 I228 I60 (loo) 160 -

6 D. 1. McDonald, R. M. Cowling

TaMc4.~~6~~~~tolugei~(secTa~Z)~~~~t~~~~~~t~~ MSS. M&c south slopes; HAW. High altitude wet slopes; HAD, High-altitude dry slopes, AR, Arid north slopes. Cbilquare goodnessof-lit tests the hypothesis that endemics are distributed the &me proponion as the habitat areas. NS, not significzwt, -*.p< OWI.

MSS

Habitats

HAW HAD AR - Number of endemics

Family Observed Expected Observed Expecled Observed Expected Observed Enrwed 2 Significance

AStCiCC3e 6 hirame 2i 21 2: i Fabaceae 4 3 lridaceae 3 : : 1 Orchidmae 0 3 5 I PrOteaKXte I 4 2 I Restionaceae 4 7 5 3

many) were recorded for the high-altitude dry and arid north slope habitats combined (Table 3).

Endemics were not distributed in the same proportion as non-endemics in each of the four habitat categories (Table 4). Mare than twice the number of endemin were found at high-altitude wet sites than was expected for the area of this habitat type. Almost three times the number of endemic ericas and five t imes tho number of orchids occur in high-altitude wet habitats than excected. On mesic south-k&c slams and hi&-altitude d& slopes, the number of endemicsbid not d&r signifi- cantly from the expected number. Only endemic orchids were si@iticantiy fewer than expected on mesic sourh- facing slopes. On arid north slopes, however, endemics were under-represented by more than a third, with Eri- caceae beiig most markedly affected (Table 3).

End&s were not a random assemblage with regard TV biological attributes (Fig. 3). In the growth form classes, low shrubs were over-represented as endemics. Endemism in mid-high shrubs does not diflar signifi- cantly from expected frequencies and is in equal pmpor- lions to non-endemism in this class. All other growth forms were under-represented as endemics (Fig. 3(a)). Endemics were significantly over-represented as non- sprouters (Fig. 3(b)) and over-represented as passive/ unknown and ant-dispersed spgies but under-represented as wind and vertebrate-disparsed species (Fig. 3(c)).

DISCUSSION

From the Bykov index of endemicity (Id and the nomo- gram derived from Bykov’s data (Major, 1988), it is clear that by global standards the southern Langeberg has a high level of endemism (13.02%) for the relatively small &a (174 857 ha) occupied. Compared with othhr areas in the CFR (Table I), the southern Lang&erg has an index of endemicity about one-third higher than

0 2 3 4 3.416 5 6 2 14 >&Hx) ::

z 2 I 0 7 4 I 6.869 3.500 NS NS 0 0 I : 19.500 l -=

I 1 5.25ll 4 2 : 5 4.819 E

the lowland areas (Agulhas Plain and Humansdorp) but only a little ow half the endemicity of the Cape Penin&. Any one I&X of endemism iilikely to h&e shortcomings but the Bykov index displays well the contrast between endemism in the montane fynbos of the Lang&erg and that of the Agulhas Plain, and be- tween both these areas and the Cap Peninsula.

Taxommdc asped of sot&em Langebarg endamism Endemics are aenerallv concentrated in a limited num- ber of taxa and taxo&niczdly are not a random assent- blane (M&r. 1988). Tbis is true for other floras in the CFk. ‘na&ly the kgulhas Plain, Cape Peninsula and Humansdorp region where certain taxa such as the Eri- caceae have a higher than average level of endemism (Cowling d d, 1992). A similar pattern is observed ott the southern Langeberg where the majority of endemics are restricted to seven families. The signilirant over- representation of endemic Ericaceae was expected to be higher than other areas because large tracts of erica- ceous fynbos cover the mesic south-facing slopes of the southern coastal mountains (Campbell, 1985). How- ever, over-representation of endemic Restionaceae on the southern Lang&erg was surprising. In this respect the southern Lan&arg diRers d&.tinctly from floras of the Armlhas Plain, Humansdom region and Cape Pen&la (Cowling et al., 1992). -

Habitat aspacts Many of the southern Langeberg endemic species are cmflned to one or a few local populations, for example in the gems Cideoemo (Williarm, 1981) and other Rutawae, or located on isolated mountain peaks, e.g. Erica species (E.G.H. Oliver, pen. comm.). High-altitude wet and me& south-facing habitats together harbottr 74% of the southem Langeberg endemic flora, which indicates that in thii flora endemism is more s&on& associated with cooler, moister habitats than with d&, hotter slopes.

Although species richness is highly correlated with habitat area, ievels of endemism do not rdlect this

pattern. Spsies richoess and endemism ate not coupled and endemism is clearly not explained by the size of a given habitat on the southern Lang&erg.

logiea attributes 10 th& of the Agtdhas Plain and Hmnusdom floras K!owline & Hoti. 1992: Cowline et 111.. 1992): They arS mainI; non-sprouting, I& sb& with sbortdiitance seed diqmal; seed is eitbcr passively dispersed or disposal by ants. This reinforces tbe con- clusion that endemism is aswziated with these biological oaks in particular. This may in turn be associated with spsiation (Cowling & Holmes, 1992; Ccwlbtg et ol., 1992) but can also be conducive to local extbxtion when these specim are subject to catastmphii events such as too fteqwm tires (van Wilgen et I& 1990).

CoasatiadsoDtbaa~- Management objectives in the mountaio catcbmmts of the southern Lang&erg are no d&rent from those in otha Cape mountain catchment areas. The prime goals are production of potable water and nature cotuerva- tion, with tire hazard reductiott. wildtlowet harvesting, recreation and grazing of lesser impottaoce. In turn. the main aims of nature com.ervation are the maittte- natwe of biotic diversity and fostering ecosystem pr& cases (Seydack, 1992; van Wilgm et al., in press).

The concemration of m&mics on the mesic somb- faciog slopes and in the high-altitude wet habitats of the southern Lang&erg indkatcs that these habitats deserve careful attention. Much of the high-altitude nxu is lccated in declared nature re3erves 01 on protectal state land and is thus well-protected. However, with many endemica isolated on mountain peaks and the diioro- portionately high frequency of a&mica in tbe high-& tude wet habits% those parts of the high-altitude zone that ate not well-protect&l should be &m&d and if possible given a higher conservetiott status.

Fire plays a pivotal role in fyttbos ecosystems by governing the life-cycles of fynbos plants. For this and other reasons, e.g. wst-etTectiveness, burning is the most pmctial management option. Conscquetttly the ecology of fire in fyobos has teceiwd much attention (van Wilgat et IJ.. 1992). However, a number of aspects of the elKas of Iire 01) the oatural biota of fyo- bos still require investigation.

Much -h has beat dimcted at the response of shrubby Pro- to fire wan w&en et al., 1992). resulting in management policies wbicb have inevitably favoured this plant family. The shortcomings of the approach that ‘what is good for one is good for all’, and that biotic diversity is bound to diminish mttkt a tixed tite t&me, are rsogniad, but no satisfactory &ma- tiws have been offered (van wilgm e, al., in pm%).

Some consideration has be&n giveo to we and emianged taxa in management pokks (van Wilgen et d, in press), but ettdemim need not be rare (Gentry, 1986) or endangered. Where do fyobos endemic spsics fit in? Evaluation of the cffwts of fire ott all fyttbos species would be mt impossible task (van Wilgen et d. in press) but evaluation of tbe e&ts on ettdemks, for example those of the southern Laqelwrg, nad not be so onerow and could provide useful iosigJtts into

8 D. J. b4cDonald. R. M. Cowling

ecosystem processes. Many endemic6 have short- distance dispersal and require specific habitats with the chances of dispersal and survival beyond these limited ranges being low. Endemic species are also not con- fined to one plant family and their biological attributes are not uniform. With tome detailed information on their distribution, biological attributes, preferred habi- tats and their response to fire, management policies can be tailored where necessary to maintain endemic species. This would augment the holistic approach to catchment management (in the case of the southern Langeberg, ‘adaptive interference tire management’) in the move away from the ‘sinele-family-dominated management policies which have been implemented until now (Seydack. 1992: van Wilgen et a/.. in press). Indications are that more attention should be directed to the low, passively dispersed, non-sprouting shrubs in the Ericaceae, at least on the southern Langeberg where there is a significant over-representation of endemics in this family. ‘Adaptive interference fire management’ holds most promise for sound management for species diversity on the southern Langeberg. From what is currently known about the biological attributes of Langeberg endemic% this type of management should promote their survival.

At present. invasion by alien plan1 species in lhe southern Lang&erg is relatively limited (personal oh- servations). However, the Langeberg mountain range is potentially invasible by Pinus species from commercial timber plantations, Hakea species and exotic Acacia spxies from disturbed agricultural lands along the foothills. The mesic south-facing slopes (and to a lesser extent the high-altitude wet habitats), where most endemic spaies occur. are the most vulnerable to invasion by alien plants (Richardson e, al.. 1992: Richardson & Cowling, 1992). Low non-sprouting shrub endemic species would he most adversely aflected by alien infestation due to canopy closure etTe@s (Richardson & van Wilgen, 1986; Richardson et al., 1989).

The threat of invading alien invertebrates. such as the Argentine ant Iridonpwnex humilis ousting indige- nous ant species which dispew the seed of mymteco- chorous southern Langeberg species including endemics in the families Penaeaceae. Proteaceae anb Rutaceae (Appendix I). must not be underestimated (Bond & Slingsby, 1983, 1984a.6: de Kock er a/.. 1992). Recre- ational hiking and other tourist activities on the soutb- em Lang&& are increasing as more hiking trails and camp-sites are developed (C. Martens. pen. comm.). This greater human usage of mountain ireas increases the chances of dispersal of Argentine ants into these ecosystems (Bond & Slingsby, 1984a), which in turn could increase their negative impact on the local biota.

The wildflower trade concentratc8 mainly on floral material from three plant families in the iynbos, the Ast;raceae, Ericaceae and Proteaceae (Davis, 1984; Grey!ing & Davis, 1989). In the flora of the Lang&erg these families together contain 44% or the endemics. Flower harvesting is becoming increasingly widespread

on the Langeberg, particularly on the more accessible lower south slopes where the highest number of ett- demic species occur. The danger exists that inditimi- nate Rower picking could alI& populations of endemic species (van Wilg& & Lamb, 1986; van Wilgen et a/., 1992). This need not be only from removal of flowering material but could also &It from mechanical damage due to trampling or from the spread of detrimental pathogens @on Brocmbsen, 1979: van Broembsen & Kruger, 1985).

The high number of endemics on the mesic to wet south-facing slopes of the southern Langeberg. coupled with the possible negative impacts of alien plant (and animal) invasions, uncontrolled flower picking and inappropriate fire regimes, indicate that it is towards these habitats that monitoring should be directed to prevent the loss of endemic species.

ACKNOWLEDGEMENTS

Numerous taxonomists xvotking on the systematics of the Cape flora, in particular Anne Bean (Rutaceae), Peter doldblatt (Iridaccae), Ted Oliver (Ericacae: Eri- c&dew?). John Rourke (Proteaceae) and Charles Stir- too (F~beceae), provided inform&on on endemism; their assistance is much appreciated. Chris Martens gave useful practical insights into catchment manage- ment policy bn the southem Langeberg. Tony R&lo and Craic Hilton-Tavlor kindlv read earlv drafts of the manuscript and suggested valuable improvements.

REFERENCES

And&erg, A. A. (19991). Taxonomy and phylogeny of the tribe Gnaphalieae (Asteraceae). Op. &I., 104, I-195.

Arnold. T. H. & deWet, B. C. (eds)(l993). Plants of Southern Africa: names and distribution. Mem. bar. SUN. S. Afr. 6t. 825.

Bond. W. J. (L98L). Vegetation gradients in southern Cap mountains. Mk thesis. University of Cap Town.

Bond, P. & Goldblatt, P. (1984). Plants of the Cap Flora. A descriptive catalogue. 1. S Afi. Bar. Suppl.. 13, 1455.

Bond, W. 1. 4 Slingsby, P. (1983). seed dispersal by ants in shrublands of the Cape Province and its evolutionary implications. S. Afr. J. Sai., 79,231-3.

Bond, W. J. 4 Slingsby, P. (19&t@. F?oteas. ants and imaderx disruption or a delicate dependence. S. Afr. J. Sci. 8% 201.

Bond. W. J. 4 Slingsby. P. (1984). Collepx of an ant-plant mutmlism: The Argentine ant Jridomyrmpx IwmiliF and mymwcochorous Proteaceae. Ecology, 65. 1031-7.

Brummitt, R. K. 4 Powell, C. E. (aIs) (1992). Aur6ors oj p/m nmm, Royal Botanic Gardens. Kew.

Bykov, B. A. (1979). On a quantitative estimate of endemism. BOf. Mlmr. Gwb. Inrr. Roton. Akd Nouk. Korokk. sm.. 11 LS

Campbell, B. M. (1985). A clarsiftcation of the mountain veg- elation of the fynbas biome. Mem. lot. sm. S. Afi. No. so, pp. 121.

Cowling, R. M. (1984). A syntaxcmomic and synecnlogkal study in the Humansdorp region of the Fynbos Biomc. Bolhalio, 15, 175-227.

Cowling, R. M. 4 Holmes. P. M. (1992). Endemism and sp- ciation in a lowland flora from the Cap Floristic Region. Bid. J. Linn. Sm.. 47, 367-83.

Cowling. R. M., HoImp. P. M. & R&lo,. A. G. (1992). Plant diwsify and endemism. In The ecology o//&w: nu- wienrs. fire and dive,si,y, cd. R. M. Cowling. Oxford Uni- versity Pres. Cape Town, pp. 62-112.

Dahlg,a, R. & “a,, Wyk. A. E. (1988). Stmctu,er and rela- tionships of families endemic to o, cent& in southern Africa. h~anogr. Sys,. Ror. h,i.w. Lb,. Cord. 2.5, 1-94.

Davis, G. W. (1984). Flowers from fynbos - the need for a policy of resurst mnagemmt far the wildflower industry. Vrldd Flora. lo, II&IS.

de Kock. A. E., Gil iomee. 1. H.. P,ingls K. L. & Ma&,, J. D. (1992). The inAucnte of fire, vegcletion age and argen- tine antr I, idomymu* bmdbs on ant communirks in Swartborklwf. In .=i,e in Sourh “,,icrm nmemin jynbos: ecosystem. comnurnby md spwies ,esponse (I, Swwt- hosk/ooj: Em/. Smd.. 93. cd, B. W. van Wilgeo, D. R. Richardson. F. J. Kruger 8. H. J. van Henstwgen. Springer-Verlag, Berlin, pp. 203-15.

Dent, M. C.. Lynch, S. D. & Schulz. R. E. (1987, Mapping mean annual and other rainfall statistics we, southern Africa. AgricurrWal Carrlunenls Rmmch Unit Repon, No. 27: Worer Remmh Cmirsion Rqm,t. No. 10911189.

Gentry. A. H. (1986). Endemism in tropical versus tempaate plant communil ier. In Cmr.wvoriar biology. The s&me of scarcity and diwrriry, ed. M. E. Soul& Sinauer Atiatsr. Sunderland, Mass., pp. 153-81.

Goldblau. P. (1978). An analysis of the flora of so&em Africa: ils chsracleriaticr. rektionships and origins. Amt. M.W. Lb,. card., 65.3694%.

Greyling, T. & Davis. 0. W. (1989) (eck). The wiM6ower

Kruckeberg. A. R. dr Rabinmviwitr, D. (1985). Biological as- pects of endemism in higher plants. Ann. Rev .%I. Sy,,., 16.447-19.

Kruger. F. J. ,197’q. South African heathlands. In Eeosys- Ie.?7.5 qf the war/d, Vd 9A. Hec?lh/an& and rekmd shrub. h,s. Desc,ip,ive srudtes, cd. R. L. Spccht. ELwier. Amsterdam, pp. 19-80.

Kunin. W. E. dr Gaston. K. I. (1993). The biokw of raritv: patterns, causes and conxqu&as. ‘Tmds in I%/. Ewl., & CIS-301.

McDonald. D. 1. (1993~). The vegetation of the smnbem Langeberg, Cape Province, I. The plant mmmunit ig of the Bcamansba Wilderness Ama. BDlholio, 23, 129-51.

McDonald, D. I. (19936). The vegetation of lbc soutb%m Lang&erg. Cape Province, 2. The plant communit ies of the Marlotb Natwe Reserve. Borbatia, 2% 151-74.

McDonald. D. I. (1993~). The vegetation of the southem Lmpeberg, Cap Province. 3. The plant communids of the Bergfontein. Rooiwatmspruit and F%eantefontein areas. Roth&, 23. 23943.

Major. I. (1988). Endemism: a botanical pnpectiw. In Ancr Iyrical biogecgmphy: M inregmed app,oach m dte smdy of onbm, and plant dislri6uriow. cd. A. A. M,m e. P. S. Giller. Chapman and Hall, New York, pp. ll7-l6.

Marguks, C. R. & Stein. J. L. (1989). Patterns in the diswi- b+ms of q&es and #be sekzdon of nature -es: an exampk from Eucalyprvr fwcsts in south-easkm New south wales. Rid. conrerr. 59.219-38.

Muir. J. (1929). The vegetation of dte River&k area. Cap Provina. Mm. Bat Sum S. Afr.. No. 13. Depanmmr of Agrkulture. Pretoria.

Rebslo. A. G. & Siegfr*d. W. R. (1996). Prowlion 01 fynbor vegetation: idcal and nal-world options. Biol. Consew., S6, 15-31.

R+zhardson. D. M. & Cowling. R. M. (1992). Wby is moue tain fynbos invasiblc and which specks invade? In F7 in Swrh “,iim m~vnu,b, fynbm: emsys,,,,,. c~~udliry md specie-z response (II Swtboskloo~ Ecol Sud. 93, cd. 8. W. van Wilgen. D. R. Rkhardson, F. J. Kmga % H. J. van Hensbergen. Springer-V&g Bedin, pp. 161-81.

Richardson. D. M.. Macdomdd, 1. A. W. & Forsyth, G. G. (1989). Reductions in plant specks rkhmrr vndn stands of alien trees and shrubs in the Fynbm Biome. S. Afi. For. J.. 149, I-8.

Ricbanlson. D. M.. Macdonaki. 1. A. W., Holmes. P. M. B Cowling. R. M. (1992). Plant and animal invuionr. In 77te ecology offynbos: nurriems. jre md direMy, ed. R. M. :;;“g: Oxford Univmity PM, Cap Town. pp.

Rkbanlson. D. M. & van Wilgen. B. W. (1986). The eRcctr of Ihiriy-five year0 of aRo~t.ation with Pinus mdkm on the composit ion of mesic mountain fynbos axa, Stelkn- bosch. S. A,?. J. Bar., 52, X%15.

Saetendal. M.. Line. J. M. & Birlr. H. J. B. (1993). How 10 marimizs hiologieal diversity in natwe restwe sekcdon: vawdar plants and bt-ceding birds in deciduous wmwJ- lands. western Nomay. Biol. Comem.. 66. 131-8.

Seydack, A. H. W. (1992). Firr mawgmmu options in fyn- her mountain cat&mm arms. S. Afr. Fo,. 1.. 141. 53-S.

Terborgh. J. 6r Winter. B. (1983). A method for siting parks and reserves with qecial rsfxence to Columbia and Ecuador. Biol. Gmwv.., 27.45-58.

van Wilgen. B. W.. Bond. W. J. 6r Richarduro. D. M. (1992). Emryrtem mall.¶gemcnt. In m ecdogy of fjdvs: nutri. ems. fuc md diwsity. cd. R. M. Cowling. onford Uniter- sity PIUS, Cape Town. pp. 345-71,

van Wilgm. 8. W.. Evcrson, C. S. & Trollop, W. S. W. (19X0. Fit’e managenat in soutbem N,icx smne exam- pks of cun-ent obj;niws, practices, and pmbkms. In Fin in rha rropical b&a. cd. 1. G, Goldammer. Ecof. Stud. M. Spriagcr-Verlag. Berlin, pp.

vma Wilgen. B. W. & Lamb, A. 1. (1986). Tbc Bower picking indua-y in relation to mountain catchmcnt management in the fynbos. Acfa Ho,,., 185. 181-7.

van Wilgen. B. W.. Richardsea, D. M. & Seydack. A. H. W. (in press). Managing fynl8.x for biodiwsity: consb-aints and odious in P firearom envimnmmt. S. Air. J. Sri.

vo” B&mbsm. S. I.. 11979). Phy,o@r,kwa ai&mmni - a thrrat to the Western Cape &n-a? Veld dr Rora, 6% 53-S.

vm Bmcmbsm. S. L & Kmp. F. J. (1985). Phyropbrkm rim-i r~%&tul with mortality of native vegetation in South Aftica. Plan, Dir, 69.71%17.

Wcimmk, H. (1941). Phytogeogmpbkal groups. cmtlcs and intmak witbin tbc Cap Flora. La& Univ. Am NF Adv. 2 &I. 37 Numkr 5: l-143.

Will iams. I. (1981). Studks in the gsmra of the Dionnmr (Rumwe). 9. A revision of the genus Coiemwrm. J. S AJr. Bm.. 47.63402.

Endemic species GF -~

LSH LSH LSH LSH LSH PO Lbii LSH

En LSH FO

LSH

MS” LSH LSH LSH LSH LSH

LSH LSH LSH LSH TSH MSH TSH LSH LSH LSH LS” LS” MS” LSH LSH LSH LSH LSH LSH

Ei MS” LSH LSH LSH LSH ls” MS”

E::

E:: LSH

RS

iii NS NS NS NS NS NS NS

Ei

FZ NS

S

: S NS NS NS

NS NS

NS NS NS NS NS NS NS

Ei NS NS

ii: NS

:: NS NS NS

ii:

ii: NS NS NS NS NS NS NS

;: NS

DM HAB

w P w

LJ w

F W W W W W w

P

F P P P P

P P

F P P P P

; P P P

F P P P

F P P P P

;

5

;

;

: P P P

AR MSS AR HAW

F HAW MSS HAW MSS MSS MSS MSS MSS

AR

Mss

ii: MSS MSS MSS

HAD MSS

MSS MSS HAW HAW AR MSS MSS MSS HA0 HAW HAW HAW MSS HAW MSS MSS MSS MSS MSS HAD

Endemic species GF RS DM HAB

18247)

L&eliaccac - Laurenrio longitti E. Wimm. L&G dasyphylla E. Wm. Idelia hypsibaa E. Wimm. Lobelia rrlvvoidps cha!n.

Mexmbrvan-

LSH

kz

EI

Ei MSH LSH

2:: LSH

z LSH

$2 MSH LSH

MSH

m

G

FO FO

E

FO FO

MSH MSH MSH

LSH LSH

NS NS NS

NS NS

E NS NS NS NS S NS s NS NS S NS S

S

NS

S

NS NS

ii;

NS

;

s” S

2 NS

:: P P

; P P P P

P

:

P P P

B

5

:: P P P P

F P

P

P

: w w P

P

P P

;

P P

P w

r

z

t P

A A

HAD

i% MSS

2: HAW

ii% HAW

HAW HAW Mss

HAW AR Mss

E AR HAW

%s HAW AR MSS MSS

ZW HAW

MSS

Mss

Mss HAW Mss MSS HAW

Mss

Ms.3 MSS

FEV

AR AR

HAW

ii; HAW AR HAW

!E Mss

HAW HAD

D. J. McDonald, R M. Cowling

Aweodin I - mmd

Endemic s&es GF US

carpacoce grganrt-” Puff R”tXCX

1%3)

MSH MSH TSH LSH LSH MSH

LSH

LSH LSH

2; LSH LSH

LSH

LSH LSH LSH LSH LSH LSH LSH LSH MSH LSH LSH

LSH

LSH

LSH

LSH LSH

K NS NS NS S NS NS

S S s S

: S S

z S

: S S S

NS NS

NS NS NS NS

NS

;: NS NS

ii: NS NS

El NS

NS

NS

NS

S S

DM

W

1 A

:: A A

P P P A

p’ P P P P P P

p’ P P

F

W P P P

P

:: A

2 A A A

:: A

P

P

P

P P

HAB

HAW MSS AR HAD

A”:: AR HAW

HAW AR

HAD HAW HAD MSS HAW HAW MS.3 HAD HAD HAW

2: AR

MSS MS.3

AR HAW MSS HAW

MSS

HAW HAD AR MSS MSS HAW HAD HAD MSS HAW MSS

MSS

AR

Mss

MSS MSS