Pacific Science (1987), vol. 41, nos . 1-4© 1988 by the University of Hawai i Press. All right s reserved

Stunted Cloud-forest in Taveuni, Fiji1

ABSTRACT: The vegetation and microclimate of a stunted ridge-top cloud­forest on Mt. Koroturanga (l210m), Taveuni, Fiji (Lat. l7 °S, Long. 180°) isdescribed. Canopy heights decreased from about 30 m at sea level to 10m at1140m altitude and to 3-7m on the ridge and upper windward slopes . Thestunted trees were of low height for their stem diameter, and had abundantepiphytic bryophytes. The upper windward slopes and ridge were usually cloudenveloped and had low temperature (c 17°C), high relative humidity (c94%) andhigh wind speed (c 5 m S-1 at 15m height) . Canopy height was closely cor­related with estimated rates ofleaftranspiration. The cloud-forest had abundantFreycinetia urvilleana in the upper canopy and included species restricted to thisenvironment on a few peaks in Fiji, e.g. Ascarina swamyana and Medinillawaterhousei.

STUNTED, OR ELFIN, RIDGE-TOP FORESTS (thick­ets, shrublands or woodlands) are recordedfrom cloudy montane environments through­out the wet tropics and sub-tropics, espe­cially where humid onshore winds rise overcoastal ranges (Howard 1968, Grubb 1977).These forests , also called "moss forests ", arerecognized by both the low height of the treecanopy and the abundance of epiphytic bryo­phytes. The canopy height varies from I mto 18m in different localities, and the abun­dance of bryophytes ranges up to a cover of10-20 ern depth, (Howard 1968, Russell andMiller 1977). Despite the lack of a precisestructural definition, these forests are inevi­tably associated with sites frequently envel­oped in clouds, and the trees are stunted incomparison with most other sites at similaraltitudes. Severely stunted forests occur abovec 600-800 m altitude, the height at whichrising maritime air masses start to formclouds, and the lower limits rise away fromthe coast, an aspect of the massenerhebungeffect (Grubb 1971). These stunted forestsare not simply the result of greater altitude,though tropical forests at c 3000-4000 m alti-

1 Manuscript accepted August 1986.2 School of Natural Resources, University of the

South Pacific, Suva, P.O. Box II, Fiji.

tude also typically have a low canopy andabundant bryophytes.

Hypotheses to explain the stunted trees andabundant bryophytes have involved most ofthe peculiar aspects of the microclimate andsoils . Wind speeds on ridges are generallyhigher than those on lower slopes (Grace1977, Bradley 1980). It has been suggestedthat this could cause a thigmomorphogeneticresponse which produces stunting (Lawton1982), that the wind damages buds and newleaves thereby reducing growth (Weaver,Byer and Bruck 1973), or that dessicatingwinds reduce growth (Beard 1944, Grace1977). There is, however, little evidence ofwind-clipping in tropical forests (Baynton1968, Grubb 1977) and the winds are typicallyvery humid (Baynton 1968, 1969, Weaveret al. 1973). The cloudy conditions result in atleast a 40-60% reduction in solar radiation(Baynton 1968, Gates 1969) reducing the rateof photosynthesis (Grubb 1977, Roth, andde Bifano 1980), lowering temperatures, andreducing growth rates (Beard 1944, Grubb1947). Through a combination of microcli­matic conditions, clouds greatly reduce tran­spiration (Beard 1944), and rates of only3-10 x 10- 7 g cm"? S-1 were calculated byGates (1969) for canopy leaves. Rates ofabout 0.4 x 10- 7 g em-2 S-1 were measuredwith potometers for lower canopy leaves

191

192

(Weaver et al. 1973) in Central Americancloud-forests. During a summer season in aNew Zealand cloud-forest the shoot waterpotentials ranged from -0.1 to -1.2 Mpa(Green and Jane 1983), indicating little waterstress. Stomata closed as the saturation de­ficit increased when the enveloping cloudcleared (Jane and Green 1985), possibly aconsequence of poor root development inwaterlogged soils. The significance of verylow transpiration rates is uncertain: Grubb(1977) contends that they are probably notlimiting growth, but Weaver et al. (1973) sug­gests that they may result in nutrient defi­ciency which causes stunting. Grubb (1971,1977) also suggests that stunting is relatedto nutrient deficiency, especialIy of nitrogenand phosphorous, because of the low rate ofmineralization and nutrient release in thesoils. A further characteristic of the soils is aprobable high rate of leaching and high soil­water content, resulting in a low soil nutrientconcentration; factors which would contri­bute to nutrient deficiency (Grubb 1977). AshalIow soil may also be the cause of stuntingin some situations (Gleason and Cook 1926,Grubb 1977, Kirkpatrick, and HassaII1985).

In the Pacific region stunted ridge-top for­ests occur on most mountainous islands, e.g.New Guinea (Paijmans 1976), Samoa (Banks1982), Kosrae (MaxwelI 1982), Cook Islands(Merlin 1985), and the Fiji islands (Smith1979, Kirkpatrick and HassalI 1985). Kirk­patrick and Hassall (1985) describe the floraof a stunted forest at about 400-420 m alti­tude near the summit of Mt. Korobaba inViti Levu, stunting which they attribute toshallow soils and windiness. Stunted cloud­forest is apparent on the high ridges in VitiLevu, Vanu a Levu, and Taveuni above about900 m altitude and is most extensive on theridges about Mt. Koroturanga (1210 m) andMt. Uluingalau (1241 m) in Taveuni. Theridge-top forests in Taveuni described in thispaper apparently have the highest rainfalIand cloud cover of any sites in Fiji and maybe considered as an extreme type of stuntedforest. Microclimatic observations were madeto determine how transpiration rates varybetween the lowland, upland, and ridge-topforests .

PACIFIC SCIENCE, Volume 41 , 1987

Matei

10 kmI

N

Io

17 S

F IGURE 1. Map of Taveuni showing localities men­tioned in the text. The altitudinal tran sect is indicated bythe dotted line.

STUDY AREA

Ta veuni is 42 km long (N.E.-S.W.) and10-14km wide (S.E.-N.W.) with a centralmountain range rising to 900-1241 movermuch of its length (Figure 1) (Lat. l6 °50'S.,Long. 180°). The landforms are volcanic inorigin , ranging from Tertiary lavas in thenorth-east to mid-Holocene eruptions in thesouth-west (Fiji Geological Survey, pers.comm.). Up to about 800 m altitude thereare relatively uniform slopes of 10-14° dis­sected by river gorges, giving way to steeperslopes of 15-35° on the pyroclasts whichcomprise the higher ranges. The study area isthe ridge 800 m north of Mt. Koroturanga(Des Veoux Peak) at an altitude of about1200m. The substrate is a clayey soil of 10­30 em depth passing into pyroclasts with acoarse sand-sized texture. The pyroclasts aredipping at about 25-35° on the north-westslope, virtually paralIel with the hill slope.Excavations, e.g. roadworks, tend to induceinstability and scree formation, but naturallandslides were not apparent in the studyarea although they were observed on otherslopes.

Stunted Cloud-forest in Fiji-v-Asa

The rainfall at the summit of Mt. Korotu­ranga is recorded daily by the Depar tment ~[Public Works and averages 9970 mm year(5 years) and 27 mm day", with no obviousseasonal trend except for higher rainfall asso­ciated with cyclones, typically between De­cember and April. There is a dai ly cycle ofincreasing cloud cover over the main rangeand the study area is reported to be enve­loped in cloud during many morn~ngs (5 :00­8: 00 am) and intermittently du~mg the res~

of most days. Pilots rou tinely flying to Matei(Figure I ) indicated that the peak was onlyvisible, from a similar altitude, abo ut onceevery 20-30 flights, and sometimes not forseveral mon ths in succession.

Rainfall on the southeast coast of Taveuniis about 5000 mm year"? (e.g. Salialevu), whilerainfall on the northwest coast is about 2800mm year" (e.g. Matei) . Evidently the prevail ­ing southeast trade winds rise over the centralranges causing orographic rainf~lI on t?e east­ern slopes and peaks, but there IS a rainshad­ow on the western slopes. On the drier westernslopes the rainfall pattern is more seasonaland monthly rain fall is only about 100- 200mm between April and October. Averagewind velocities are about 2-3 m S-1 at south­east coastal recording stations in Fiji. Windvelocity is not recorded on the west coast ofTaveuni but is evidently much less, and thesea surface is often smooth. The averagedai ly range of air temperatures at Ma tei is24-30°C in January and 21-27°C in Ju ly,and relat ive humidity at 9 :00 am is about80% in the wet season and 75% in the driermonths, declining by about 5% in the earlyafternoo n and increasing by about 5% atnight. .

The central part of the southeast coast IS

uninhabitated and forest occurs from sea levelup to the highest peaks, while on the othercoasts there are gardens and plantations upto about 350 m altitude. The flora , fauna,and water chemistry of a crater swamp 3 kmnortheast of the ridge has been described bySouthern, Ash, Brodie and Ryan (1986), andthe vegetatio n is otherwise poorly know.nexcept for notes in the floras by Brownlie(1977), Parham (1972), and Smith (1979,1981, 1985).

193

METHODS

The study area was visited on severa l occa­sions during 1982-1 983 and a vegetatio n tran­sect was examined across the island (N .E.­S.W.) via Mt. Koroturanga. Canopy heightswere noted and herbarium specimens col­lected. At several localities transects of 100canopy trees were recorded includi?g theiridentity canopy height, and stem diameter.The res~lts of transects on the summit-ridgeof Mt. Koro turange (1200 m), and at 1000 mand 800 m altitudes on the northwestern slopesare presented in this paper. Specimens wereidenti fied at the Suva herbarium and nomen­clature follows Smith (1979, 1981, 1985) andBrownlie (1977), or Parham (1972) for otherdicotyledons. Mean stem diamete r for eachspecies was calculated from mean basal area.

Meteorological observations were madefrom 8 am to 12 pm on 17 November 1983,which had the typical daily weather patternof continual high-level clouds over the moun­tains and intermittent enveloping clouds onthe upper windward slopes and ri?ge top.Similar weathe r occurre d on succeeding days,so no contrasting sunny periods were ob­served . Observations were made on a latticetower up to 15 m in height on the summit, andrepeatedly at several sites i.n the forest , f~rm­

ing a transect across the ndge. Observationswere also made at altitudes of 800 m and atsea level, combined with sea-level recordstaken at Matei Airfield. Wind velocity wasmeasured with a Wallac thermoanemometer,leaf temperatures and cloud temperatures witha Barnes Instantatherm infra red thermometer,and other temperatures were recorded withthermocouples connected to a Conmark elec­tronic thermometer. Relative humidity wasestimated from a whirling hygrometer andlight intensity was estimated with a cameraphotocell.

RESULTS

Vegetation

The composition of the vegetation is indi­cated by the transects on the ridge-top (1200m) and at 1000m and 800 m altitude (Table I ).

194

TABLE 1

PACIFIC SCIENCE, Volume 41,1 987

CANOPY AND EPIPHYTIC SPECIES RECORDED ON TRANSECTSIN FOREST ATT HREE ALTITUDES, MT . KOROTURANGA,TAVEUNI. + = present nearby

SPECIES1200 m 1000 m 800 m

(% STEMS AND MEAN STEMDIAMETER, em)

Antrophyum smithii C. Christense nArdisia bracken ridgei (A . G ray) M ezAscarina swamyana A. C. Sm.Blechnum milnei (Carruthers) C . Christe nse nCyathea medullaris (Forst. f.) Sw.Cy rtandra tempest ii C . B. ClarkeDysoxylum lenticellare G illesp ieFagraea vitiensis Gil g et Ben ed ictFaradaya ovalifolia (A. Gray) Seem.Ficus barclayana Miq .Freycinet ia urvilleana H ombr. et Jacq .Grammitis hookeri (Brackenridge) CopelandHymenophyllum flab ellatum LabiI!.Hym enophyllum javanicum Sp ren gelMa cropiper vitiense (A . C . Sm .) A . C . Sm .M aesa insularis G illespieM edinilla waterhousei Seem.Pipturus argenteus Wedd.Randia vitiensis (Seem .) F osbergScaevolafloribunda A . G rayTimonius affi nis A. G ra yTrimenia weinmanniifolia Seem.Weinmannia spAstronidium par viflorum (Gillespie) A . C . Sm .Citronella vitiensis HowardClinostigma exo rrhizum (Wend!.) M artelliCryptocarya spFicus vitiensis Seem. (?)Geissois ternata A. GrayRapanea myricifolia (A . Gray) MezSpireanthemum serratum GillespieCya thea alata Cop e!.Dysoxylu m gillespieanum A. C. Sm .Ma caranga seemannii (M uel!. Arg.) Muell. Arg.Sy zyg ium spElaeocarpus chelonimorphus Gillesp ieFlacourtia spGarcinia sessilis (Forst. f.) Seem.Hedycarya dorstenioides A . GrayLi tsea spNeuburgia corynocarpa (A . G ray) LeenhoutsSt erculia vitiensis Seem.Aglaia spCalophyllum vitiense TurrillDecaspermum vitiense (A. Gray) NiedenzuDysoxylum richii (A . Gray) C . DC.Omolanthus nutans (Forst. f.) Guill.Planchonella pyru lifera (A. G ray) van R oyenTrichospermum richii (A. Gray) Seem.Alphitonia zizyphoides (Spreng.) A . G rayAlstonia vitiensis Seem .Arytera concolor (G illespie) A . C. Sm .Bischofia javanica BlumeCalophyllum neo-ebudicum GuillauminCapparis quiniflora DC.

+1%( 8)++1%(6)1% (3)1% (15)2% (14)1% (3)3% (14)3%(3)+++1% (2)3%(4)+1%(2)5%(5)3% (11)3% (15)1% (12)1%(20)11%(10)2% (7)1%(10)1%(4)11%(10)3%(8)1%(15)19% (19)7% (8)4% (21)5%(12)4% (16)

1% (8)1%(14)8%(15)3%(13)2% (11)1%(45)2% (23)7% (26)4% (8)15%(26)10% (24)6% (22)1%(25)1%(1 7)1% (25)1% (27)2% (26)4% (22)1% (45)2% (18)15%(27)1% (10)1%(16)1% (15)4% (19)2% (12)

2%(8)9%(48)6%(22)3%(3 1)

2%(19)13%(49)1% (17)20% (45)1%(20)1%(45)3% (36)6% (36)1%(1 8)1% (40)1% (35)4%(47)1% (45)

Stunted Cloud-forest in Fiji-c-Asu

TAB LE 1 (continued)

195

SPECIES

Cupaniopsis amoena A. C. Sm.Cynometra insularis A. C. Sm.Elattostachysfalcata (A. G ray) Radlk .Endospermum macrophyllum (M ue1!. Arg.) Pax et H offmHeritiera ornithocephala Kosterm.Hernandia olivacea GillespiePalaquium hornei (Baker) DubardParasponia andersonii (Planch.) Planch.Parinari insularum A. Gray

Although a few species occur in the canopy atall three altitudes, e.g. Dysoxyfum gillespia­num , Macaranga seemannii and Syzygium sp.,most species are apparently restricted in alti­tudinal range. The ridge-top cloud-forest in­cludes trees such as Sp ireanthemum serratumand Ascarina swamyana which are rarely ob­served in other habitats. Several epiphytes aresimilarly restricted to cloud -forests , e.g. M e­dinilla waterhousei and ferns such as Antro­phyum smithii, Grammitis hookeri, Hymenophyf­fum flabella tum and H. javanicum . Epiph yticbryophytes were abundant at the ridge top.Climbing Freycinetia spp occur at all sites, andon the upper windward slopes and ridge top F.urvilleana forms about 40% of the upper leafcanopy and their stems form a dense tangl e.

Tree canopy height and trunk diameter de­cline with altitude, (Figure 2, Figure 3), andfrom Table I it is apparent that this trend alsooccurs within species. The reduction in can­opy height is greater than the reduction inmaximum stem diameters, giving the ridge­top trees a stunted appearance.

Leaf and leaflet size in the cloud-forestranged from the tiny leaflets of bryophytesand ferns to tree leaves of about 6 x 4 em to12 x 8 em, and Freycinetia leaves of about30-40 x 3-5 em. At lower altitudes therewas an increasing proportion of larger-leavedtrees, ranging from about 6 x 4 cm to 30 x 20ern at sea level.

M icrom eteorofogy

The spatial pattern of observations acro ssthe ridge top is indicated in Figure 3. Wind

1200 m 1000 m 800 m(% STEMS AND MEAN STEM DI AMETER, em)

1%( 35)1% (35)3%(3 4)6%(59)2%(42)1% (18)5%(43)1% (20)3%(36)

speeds increased with elevation above the can­opy surface, reaching about 5 m S-l at IS mheight. Wind velocity fluctuated greatly overperiods of 1-30 s and the values shown inFigure 2 indicate intermediate values betweenlulls of about half that velocity and gusts oftwice that velocity. The wind velocity aroundthe upper canopy was about 0.8-1 .5 m s" onthe windward side of the ridge and 0.5-1 .0 mS- l on the lee side. Canopy leaf surface tem­peratures were similar in different species andcontinually varied by about ± 1°C over aperiod of 1-2 minutes, decreasing during pe­riods of enveloping cloud . On average , leaftemperatures were about 17°C on the wind­ward slopes and ridge top , increasing to about19°C on lee slopes at similar altitude, and in­creasing to about 20°C at 40 m below the ridgeon the lee side. Relative humidities were allhigh, and successive estimates indicated val­ues of about 90-96% on the windward slopesand ridge top, declining to about 84-92% onthe lee slopes. A fine mist ofcloud water drop­lets formed intermittently over the upperwindward slope and ridge top but rarely ex­tended more than about 20 m down the leeslope. The infrared temperature of this cloudwas 16-17°C. In comparison with sunny sites,light intensity fluctuated with cloud coverfrom about 15% to 40% , with a typical valueof about 30% . This corresponds to about 1.5­4.0 x 107 erg. em"? S-l (12-33 mW cm-Z)

solar radiation.During the same period at Matei, wind ve­

locity was about 2-3 m S-l (10 m height) , airtemperatures reached 28°C, relative humiditywas 75-81 %, and there were intermittent

196 PACIFI C SCIENC E, Volume 41,1987

30

00 o 0 0

25 0 0 00

00 0 0 o 0 0 0

00 000 0 0

E0 o 0 000 o 0 0 0 800mo [R. 00

0 020 o 0000000 0

o 00 0 00000

(010 00 00000...- go 0

00 0

.s: o 000 00 001 0 0 0000 0

<1J15 0

°0

0 00 ° °.s: 0

° 0 °0 0 00 00000 °0

0 1000 m>. 000'600 0~ 00 0Cl. 10 00 000 0 09:5'0 oU~ ~ooooo (0 1

c: o~ ° 00 000000 0

rtJ 0000°0 00

LJ ..-. .5

o 0 0 0 }200m... ~ ::.....:. ~ 0 0..:.:~~~.. .:. e: .. ...( 0 1•':1-: : ••-::::..

00 10 20 30 40 50 60 70

Stem diameter ( em 1

FIGURE 2. Scatter diagram showing the canopy height and stem diame ter of canopy trees from transects at 800 m,1000m and the ridge top at l200m on Mt. Koroturanga, Taveuni.

1220 S.E. N.W.

12 10

E 120 0

QI1190-0

:J........« 1180

1170

10 m1160

FIGURE 3. Profile diagram across the ridge north of Mt. Koroturanga, Taveuni, showing average micro-climaticparameters at 9am-12pm, 17 Nov . 1983: winds, m s" x 0.5-2; Leaf temperature , °C ± 1°C; relative humidity,% ± 2% . The tree canopy height is shown by the dotted line, and the prevailing wind direction is indica ted (S.E. toN.W:'!.

Stunted Cloud-forest in Fiji-v-Asa 197

TABLE 2

MICROCLIMATE AND EsTIMATED TRANSPIRATION RATES AT FOUR SITES ON OR NEAR MT. KOROTURANGA, TAVEUNI,WITH LEAVES OF 5 x 5 CM, LEAF REsISTANCEOF 5 SCM -I , AND A WIND VELOCITY OF I MS-I . Calculations based on

Gates and Pap ain (1971)

ALTITUDE AIRTEMP. REL: HUMID. RADIATIONSITE (m) ("C) (%) (10- 7 erg em? S- I)

Ridge-top 1200 17 93 3Lee slope 1160 20 86 6Lee slope 800 24 80 8Matei 0 28 75 8

TRANSPIRATION(10- 7 g cm v s")

2122429

-7 _2_1Tr an spir ati on [10 g.cm .s )

FIGURE4. Scatter diagram showing canopy height andestimated rate s of transpiration at four alti tudes on Mt.Koroturanga, Tav euni . See text and Table 2 for furtherdetai ls.

ing the canopy heights of humid tropicalforests, and that the stunted cloud-forests re­present the most extreme situation. The tran­spiration rates estimated for the Taveunicloud-forest are intermediate between thoseestimated by Gates (1969) and those mea­sured for lower canopy leaves by Weaver et al(1973) at similar latitudes and altitudes in theneotropics.

Transpiration rates are not the only factorwhich can limit growth and Grubb (1977) isprobably correct in suggesting that nutrientdeficiency is the direct cause ofmuch stunting.His dismissal, however , of low transpirationrates as a causal factor is based on evidencewhich is as indirect as the evidence he presents

20 30

30

E

.... 20

..c.c:n

OJ.c:

>. tCo 100c:Itl

+U

00 10

DISCUSSION

The results of this study suggest that tran­spiration rates could be a major factor limit-

cloudy periods and no rainfall. Wind velocityabove the ridge was evidently nearly twice thatat sea level. At 800 m altitude on the westernslopes there was intermittent sunshine and airtemperatures reached 24°C. Air temperaturesapparently declined by O.soC 100m-I altitudeup to 800 m, but much more rapidly on thecloudy upper slopes. In the absence of cloudsit was estimated that the air temperature atthe altitude of the ridge would be about 22°C,indicating that the cloud caused a 4-6°C low­ering of temperature. Relative humidity in­creased with altitude, most rapidly towardsthe cloud-enveloped ridge (Table 2).

On the basis of the meteorological obser­vations, rate s of transpiration were calcu­lated at four sites for a leaf of 5 x 5 emeffective diameter (an actual diameter ofabout 7 x 7 em), a leaf resistance of 5 s cm" ,and a wind velocity of I m S-I (Table 2), usingthe equations of Gates and Papain (1971).Even though these assumptions and the ob­servations may be inexact , the trend betweenthe sites is clear. Transpiration rates declinedgradually with altitude up the sunny lee slopesto 83% at 800 m and then declined morerapidly to 40% at the cloudy site on the leeslope at 1160m and to about 7% on the c1oud­enveloped ridge. When the rates of transpira­tion are plotted against corresponding canopyheights (Figure 4), it is apparent that these twoparameters are almost directly proportional,suggesting that there may be a direct causalrelationship.

198

in favor of a limiting rate of nutrient mobi­lization in the soil. The effect of low lightintensity, typically 20-40% daylight undercloudy conditions, must also contribute to re­duced production, but it appears to be lesssignificant than the reduced rate of transpi­ration. The potential importance of sunnyperiods for both transpiration and photo­synthesis (Roth and de Bifano 1980) may besubstantially reduced by the closure of sto­mata (Jane and Green 1985). Experimentalstudies and long-term monitoring evidentlyare required to determine the causes of stunt­ing and the more general limits to tree height.

Lawton's (1982) observation that stuntedridge-top trees show a greater reduction in thelength of trunk and branches than in theirdiameter is supported by the observations inthis paper, both within and between species(Figure 2), but it is not obvious that this is theresult of a thigmomorphogenetic response(i.e. shaking). Coastal trees and palms, e.g.Cocos nucifera L., are not obviously stuntedwhen contrasted with trees further inland,where the coastal habitat is subject to muchgreater wind velocities and neither habitat issubject to the confounding effect ofenvelop­ing clouds. At high relative humidity and lowlight intensity, changes in wind velocity (e.g.10-400 em S- 1) have a relatively minor effecton leaf temperatures or transpiration rates(Gates and Papain 1971), and the associationof wind velocity with the stunting of ridge-topcloud-forest appears to be indirect.

Gusts of 30-35 m S-1 are experienced dur­ing cyclones with a recurrence interval ofabout 4.5 years but there was little sign ofuprooting or broken branches, and thoughless frequent severe cyclone gusts of 40-55 mS-1 may occur, the stem diameter-canopyheight relationship (Figure 2) indicates thatthe ridge-top forests are stunted rather thanan early stage of regrowth on a devastated site.The infrequent impact ofdevastating cyclonesshould not, however, be overlooked.

Kirkpatrick and Hassall (1985) consideredwind to be a major determinant of the stuntingofforest (8-12 m high) on Mt. Korobaba. Onthe basis ofclimatic records from the environsof Suva , the average mid-day rate of transpi­ration was estimated as 13 x 10- 7 g cm ? S-1

which is only slightly greater than the esti-

PACIFIC SCIENCE, Volume 41, 1987

mates from Taveuni for a forest of this height(Wind ~ 1 m S-1; light transmitted ~ 100%for 15% year, 50% for 75% year, 20% for10% year; relative humidity ~ 75-95%; airtemperature 21-27"C; leaf size and resistanceas in Taveuni calculations). It appears thatcloud cover, even though it is rarely at groundlevel, may be of more significance than windvelocity , though at times the winds may havea desiccating effect. Similar stunted forest oc­curs on most prominent ridges from about 350to 900 m altitude, above which stunted cloud­forest becomes predominant. The frequencyof enveloping cloud is indicated by the in­creasing abundance of epiphytic bryophytes.

The extent of cloud-forest in Fiji is verylimited, occupying only the ridges on thehighest peaks. In Taveuni cloud-forest occursalong about 10 km of the main range and inthe other islands it is restricted to 1-2 km ofridges around isolated peaks which are sep­arated by distances of 30-100 km. Althoughsome species occur in most of the cloud-for­ests, others are apparently restricted to a fewpeaks: M edinilla spectabilis A. C. Sm. occursonly in Taveuni, M. Waterhousei occurs onlyin Taveuni and on Mt. Seatura in VanuaLevu, Paphia vitiensis Seem. is known onlyfrom peaks above 870 m altitude in Viti Levu,and Ascarina swamyana occurs only in Ta­veuni and on Mt. Voma in Viti Levu. As yetmany of these cloud-forest localities have notbeen thoroughly investigated by botanists andthe distribution of most cloud-forest species isinadequately known.

There is considerable similarity in the ge­neric composition of the Fijian cloud-forestand those in Samoa (Banks 1982) and theCook islands (Merlin 1985). Further taxo­nomic studies may reveal close affinity be­tween these isolated populations, and whetherthey represent widely dispersed cloud-foresttaxa or local evolution from lower altitudeforest species to cope with the singular cloud­forest environment.

ACKNOWLEDGMENTS

I thank Wendy Southern and Saula Vodo­naivalu for assistance with field work andidentification of specimens, and Nicole Nan­kervis who typed this paper. Meteorology

Stunted Cloud-forest in Fiji-AsH

equipment was loaned by the Physics Depart­ment at U .S.P., and meteorological informa­tion was provided by the Department of Pub­lic Works at Waiyevo, the Fiji MeteorologicalService at Nadi, the air traffic controller atMatei, and pilots of Fiji Air.

LITERATURE CITED

BANKS, R. C. 1982. Wildlife and wildlifehabitat of American Samoa, U.S. Dept.Interior. Fish and Wildlife Service.

BAYNTON,H. W.1968. The ecology ofan elfinforest in Puerto Rico, 2. The micro-climateof Pico del Oeste . J. Arnold Arbor. Harv.Univ.49 :419-430.

BAYNTON, H. W. 1969. The ecology of an elfinforest in Puerto Rico, 3. Hilltop and forestinfluences on the micro -climate of Pico delOeste. J. Arnold Arbor. Harv. Univ. 50:80-92.

BEARD, J . S. 1944. Climax vegetation in trop­ical America. Ecology 25: 127-158.

BRADLEY, E. F . 1980. An experimental studyof wind speed, shearing stress and turbu­lence at the crest of a large hill. Quart.J. Roy. Met. Soc. 106: 101-123.

BROWNLIE, G. 1977. The pteridophyte flora ofFiji, Nova Hedwigia (suppl.). 55 : 1-397.

GATES, D. M . 1969. The ecology of an elfinforest in Puerto Rico , 4. Transpiration ratesand temperatures of leaves in cool humidenvironment. J. Arnold Arbor. Harv. Univ.50: 93-98.

GATES, D. M., and L. V. Papian. 197I.Atlasof energy budgets of plant leaves. Acad.Press, New York.

GLEASON, H. A., and M. J. COOK. 1926. Plantecology of Puerto Rico, Sci. Surv . PuertoRico and Virgin lsI. 7 : 1-173.

GRACE, J. 1977. Plant responses to wind .Acad. Press, New York.

GREEN, T. G. A., G. T. JANE. 1983. Diurnalpatterns of water potential in the evergreencloud forests of the Kaimai Ranges, NorthIsland, New Zealand. N.Z. J. Bot. 21 :379­389.

GRUBB, P. J . 1971. Interpretation of the"Massenerhebung" effecton tropical moun­tains. Nature. 229, 44-6.

GRUBB, P. J. 1977. Control of forest growth

199

and distribution on wet tropical mountains,with special reference to mineral nutrition.Ann. Ann. Rev. EcoI. Syst. 8: 83-107.

HOWARD, R. A. 1968. The ecology of an elfinforest in Puerto Rico , 1. Introduction andcomposition studies. J. Arnold Arbor. Harv.Univ.49 :381-418.

JANE, G. T., and T. G. A. GREEN. 1985. Pat­terns of stomatal conductance in six ever­green tree species from a New Zealandcloud forest. Bot. Gaz., 46: 413-420.

KIRKPATRICK, J. B., and D. C. HASSALL. 1985.The vegetation and flora along an altitudinaltransect through tropical forest at MountKorobaba, Fiji. N.Z. J. Bot. 23 :33-46.

LAWTON, R. O. 1980. Wind stress and elfinstature in a montane rainforest tree: anadaptive explanation. Am. J. Bot. 69:1224-1230.

MAXWELL, B. D. 1982. Floristic description ofnative upland forests on Kosrae, EasternCaroline Islands. Micronesica, 18: 109-120.

MERLIN, M. D. 1985. Woody vegetation in theupland region of Rarotonga, Cook IslandsPac . Sci., 39 :81-99.

PAIJMANS, K. 1976. Vegetation, in Paijmans,K. (ed.) New Guinea Vegetation, p . 23­105, CSIRO-ANU Press, Canberra.

PARHAM, J. W. 1972. Plants ofthe Fiji Islands,2nd ed. Govt. Printer, Suva .

ROTH, T., and M . T. DE BIFANO, 1980. Mor­phological and anatomical studies ofleavesof the plants in a Venezuelan cloud forest, 2:Stomatal density and stomatal patterns.Acta BioI. Venez., 10: 69-108.

RUSSELL, K. W., and H . A. MILLER. 1977. Theecology of an elfin forest in Puerto Rico,17. Epiphytic mossy vegetation of Pico delOeste . J . Arnold Arbor. Harv. Univ . 58: 1­24.

SMITH, A. C. 1979, 1981, 1985. Flora VitiensisNova vol 1-3, Pacific Tropical BotanicGarden, Lawai, Hawaii.

SOUTHERN, W., J. ASH, J . BRODIE, and P.RYAN. (1986). The flora , fauna and waterchemistry of Tagimaucia crater, a tropicalhighland lake and swamp in Fiji. Freshw.BioI. 16, 509-520.

WEAVER, P. L., M. D. BYER, and D. L. BRUCK.1973. Transpiration rates in the Luquillomountains of Puerto Rico . Biotropica, 5:123-133.