Pacific Science (1992), vol. 46, no. 2: 197-220 © 1992 by University of Hawaii Press. All rights reserved

Vegetation of the Wet Windward Slope of Haleakala, Maui, Hawaii1

KANEHIRO KITAYAMA AND DIETER MUELLER-DOMBOIS2

ABSTRACT: The vegetation on the wet windward slope of Haleakala was studied for community organization along a transect between 350 m a.s.l. and the summit (3055 m). The plant communities classified by the Braun-Blanquet synthesis table technique showed a hierarchical arrangement and were correlated with altitude. First, the forest and the treeless vegetation were differentiated by two major species groups. The boundary between the two was coincident with the trade wind inversion (ca. 1900 m a.s.l.) where the wet, low to mid-altitudinal climate changed abruptly upslope to an arid high-altitude one. These two wide-ranging vegetation types were subdivided into three units, corresponding to three broad altitudinal zones: the lowland, the montane, and the high-altitude zones. The three units were further partitioned into seven plant communities, which indicated six altitudinal subzones and one dieback belt. The floristic composition of the communities, the community structures, and their environ­ mental relationships are briefly described with a summarized differential table. The depauperate and disharmonic nature of the Hawaiian flora is reflected in such altitudinal patterns as the low species turnover and the depressed forest line.

THE HAWAIIAN ISLANDS, biogeographically the wetter Hawaiian habitats (Mueller­ the world's most isolated archipelago, sup­ Dombois 1987). port high endemism in their flora (956 flower­ The monodominance of M. polymorpha ing plant species with 89% endemic; Wagner seems to be a factor in the recurring phenome­ et al. 1990). The extreme isolation has acted non of canopy dieback (Mueller-Dombois as a sieve allowing only a limited number of 1986, 1987). Metrosideros polymorpha, being species to cross the ocean. As a result, the flora shade-intolerant in the sapling stage (Burton has become disharmonic (Hubbell 1968). For and Mueller-Dombois 1984), depends on instance plants with larger disseminules are canopy openings to maintain its regenera­ not common in the native inland forests in tion. Consequently, episodic stand-level Hawaii. The flora is also depauperate, and the regeneration in response to canopy dieback of number of potential canopy species is low M. polymorpha seems to have worked as a (Mueller-Dombois 1987). mechanism for successive generations. The The taxonomic disharmony and relative process of dieback has been discussed as biotic impoverishment resulted in widespread primarily a demographic phenomenon of monodominance of the native rainforests by cohort senescence, interacting with abiotic Metrosideros polymorpha (Mueller-Dombois stresses (Mueller-Dombois 1988a). 1981a). This is a myrtaceous tree species with We also suspect that the monodominance capsulate fruits, which produce very small is manifested in reduced interspecific competi­ wind-dispersed seeds. The species dominates tion. Therefore, some distributional attributes of plant communities that are released from high species competition may be displayed on 1 This project was supported in part by a grant from the slopes of the Hawaiian high mountains. the National Tropical Botanical Garden. Manuscript accepted for publication 2 May 1991. Investigations have been done to compare 2 Department of Botany, University of Hawaii at species turnover along mountain transects. Manoa, Honolulu, Hawaii 96822. One of the analyses relates to Haleakala, 197 198 PACIFIC SCIENCE, Volume 46, April 1992

Maui, an isolated oceanic island mountain, Conservancy; and the summit area (from 2100 the other to Mt. Kinabalu, Borneo, a species­ to 3055 m) in Haleakala National Park. Cur­ rich continental island mountain. rently, the vegetation is relatively well pro­ In this paper, we present some results ofthe tected. The summit area has been severely Haleakala transect study. The following ques­ influenced by feral ungulates, particularly by tions guided this study: (I) Does a depaupe­ goats (Stone 1985). But control efforts have rate flora become organized into altitudinally supressed their activity. Feral pigs are the definable communities? (2) If definable, are current major disturbance factor in the wetter such altitudinal communities broader and forests (Stone 1985). The vegetation below fewer in number than in floristically richer 350 m a.s.l. has largely been converted to areas? (3) How are the plant distribution plantation forests. patterns related to the altitudinal environ­ Widespread forest dieback was noted in the mental gradient? lower segment of the transect and adjacent areas early in this century by Lyon (1909). Study Area Holt (1988) reassessed the same area. The vegetation of the summit crater was mapped The northeast slope ofHaleakala (3055 m), by Whiteaker (1983). Maui, exposed to the prevailing trade winds, was selected for the study. Haleakala is a Climate shield-shaped volcano of early Pleistocene origin (0.8 million yr) and now quiescent. Itis There are great changes in climate over the third highest mountain in Hawaii (after short distances along the transect, largely due Mauna Kea, 4205 m, and Mauna Loa, 4169 to two factors: the altitudinal reduction ofair m). The summit, located at 20° 45' Nand 156° temperature and the midslope increase in 15' W, has a huge caldera (12 km long, 4 km cloudiness. The climate of the windward low­ wide) with cinder cones on its floor and land, classified as Af in Koppen's system exposed pyroclastic materials on the outer (Koppen 1936), is warm-tropical and per­ walls. The northeast slope is covered with soils humid year-round. The summit climate, derived from recent (late Pleistocene) volcanic which may be classified as Cs in Koppen's rocks of the Kula volcanic series. There are system, is cool-tropical with a dry summer also still younger (Holocene) rocks from the season. Hana volcanic series. The study area is located The mean annual air temperature at the on the Kula volcanic series. The parent rock Kailua meteorological station (213 m a.s.l.), is largely from alkalic basalt (Stearns 1985). which is near the low end of the transect, is The topography of the study area consists 21SC (Figure 2). The mean monthly tem­ of undulating gentle slopes (generally less perature is 22.9°C in the warmest month than 8°), dissected by numerous streams run­ (August) and 20.l oC in the coldest month ning parallel to one another downslope. The (February). This indicates a maritime temper­ lateral dissections become steeper and wider ature regime, characterized by a small annual near the coast, where they form deeply sliced, change (2.8°C). However, the diurnal fluctua­ V-shaped valleys. tion is nearly 10°C. Temperatures generally The study area is a belt transect, I km wide. decrease upslope in accordance with the lapse It starts at 350 m a.s.l. near Kailua and rate of 0.55°C on Mauna Loa, Hawaii (Blu­ extends upward on the interfluves of either menstock 1961). However, Haleakala's up­ side ofWaikamoi Stream to the summit (Fig­ slope temperatures may divert from this lapse ure I). The belt transect traverses three pro­ rate because ofdifferences in cloudiness. Tem­ tected areas: a watershed forest (from 350 to perature increases sharply upslope at the trade 1600 m), managed by the East Maui Irrigation wind inversion because ofdescending dry air. Company; the Waikamoi Preserve (from 1600 Results of a short-term measurement in June to 2100 m), managed by the Hawaii Nature 1988 (Figure 2) indicate the presence of the Vegetation of Windward Ha1eaka1a-KITAYAMA AND MUELLER-DoMBOIS 199

o o 160 W 155 W

•Kauai Haleakala Transect Oahu ... / ,~ 41 Maui

Hawaii

HAWAII

N o 5km '---' 1

FIGURE 1. Location of the transect established on the windward slope of Haleakala, Maui, Hawaii. inversion at 1900 m with a sharp temperature the mean monthly air temperature is ca. 6°C increase by 5°C. The inversion fluctuates, in the coldest month, and slightly exceeds but most frequently appears between 1900 lOoC in the warmest month. and 2000 m (Mendonca and Iwaoka 1969, Noguchi et al. (1987) estimated the occur­ Noguchi et al. 1987). At the summit (3055 m), rence of ground frost to be 187 days per year 200 PACIFIC SCIENCE, Volume 46, April 1992

Air Temperature (OC) Annual Rainfall (x 1,000 mm) 30,------, 7.5

...... 1"".25 25

20 5.0 temperature maxima June 1988 15 .75 temperature lapse rate 0.550 C/100m 10 / 2.5

1.25 5 mean annual rainfall o o +------r------,-----,------,r------,------~---,-J o 500 1000 1500 2000 2500 3000 Altitude (m)

KAILUA 21.5·C SUMMIT 2131R 2.985l1t11 3.055. 1.000.11I

5 ••

300 3••

'0. t••

I. I.

I. I.

o. 2. o.

2. •• 2•

J J 0 J J o

FIGURE 2. Climate along the transect. The climate diagram of Haleakala's summit is shown with monthly mean daily maximum and minimum air temperatures. The air-temperature lapse rate, 0.55°CjI00 m, is a theoretical value. The air temperature maxima measured in June 1988 indicate an inversion at 1900 m. at the summit. They also found evidence of The moisture regime ofthe northeast slope ground frost as low as ca. 2700 m a.s.l. This of Haleakala is largely controlled by the elevation coincides closely with the ground­ northeast trade winds and the trade wind frost line of Mauna Loa found by Mueller­ inversion (Lyons 1979). The orographic uplift Dombois (1967). of the trade winds results in high rainfall Vegetation ofWindward Haleakala-KITAYAMA AND MUELLER-DoMBOIS 201 below the inversion. Rainfall rapidly increases RESULTS upslope, from 4000 mm at 350 m a.s.l. to a maximum mean annual amount of 6500 mm General Description ofthe Soils at ca. 1000 m a.s.l. (Giambelluca et al. 1986; Figure 2). A dry area occurs above the inver­ The soils below the inversion are wet and sion because clouds are prevented from mov­ histic. Gleyed B horizons (dark gray) due to ing upward by the inversion; the mean annual anaerobic conditions with an overlay ofthick rainfall at 2000 m a.s.!. is 2000 mm and mors (10-30 cm) are characteristic ofsuch wet becomes less than 1000 mm at the summit histic soils. The soils are highly acidic, with pH (3055 m). In the midslope area, the total values between 3 and 4. They probably exhibit precipitation may far exceed the annual rain­ toxic levels of aluminum (unpublished data). fall because of fog interception by the forest Above the inversion, the soils are drier and canopy (Juvik and Ekern 1978). better drained. A strongly reduced subsurface horizon is found in the lowland, particularly on the flat interfluve between 450 and 600 m a.s.l. despite METHODS the fact that the midslope receives even higher The comparative transect method (van der rainfall. The reducing conditions are probably Hammen et al. 1989) was employed to enable related to the almost flat or gently sloping comparisons of the results of this study with topography, which prevents rainwater from those of other tropical high mountains. draining laterally and aggravates the anaero­ The belt transect was stratified into even bic condition. A still further degraded soil altitudinal intervals of 100 m to locate vegeta­ type with a placic horizon consisting of an tion sample plots (i.e., releves). In each inter­ iron hardpan beneath the B horizon is found val, releve analyses (Mueller-Dombois and in the lowest segment at 450 m a.s.l. Ellenberg 1974) were performed in several The montane soils have less strongly gleyed quadrats of20 x 20 m by a system ofstratified horizons and thicker peaty organic horizons sampling: one releve in the most developed than the lowland soils. Some soil profiles stand of a gentle slope and several additional between 1000 and 1600 m a.s.l. show clearly releves under various canopy conditions rang­ eluviated horizons underneath the organic ing from widely opened to closed canopies. horizons, suggesting strong leaching with low­ In these releves, species composition, species er water tables whose levels may fluctuate only cover using the Braun-Blanquet cover-abun­ below the eluviated horizons. Concave to flat dance scale (Mueller-Dombois and Ellenberg montane slopes are completely saturated. 1974), layer structure, and physical environ­ The soils gradually become better drained ment were recorded. The quadrat size of toward the inversion, where thin gley horizons 20 x 20 m, chosen for allowing direct com­ are still recognizable but waterlogging no parisons to other Hawaiian rainforest studies longer appears. done with the same quadrat size, exceeded the The soils above the inversion are less devel­ minimal area. Some low-growing stands were oped, showing fewer horizons and diffuse sampled with 10 x 10 m quadrats, which also boundaries. Associated with the lower rainfall satisfied the minimal area requirement. No­ and the better drainage, the soils are weak to menclature follows the Flora ofWagner et al. neutral in acidity (pH 5-6) and high in cation (1990) for flowering plants and an unpub­ contents (unpublished data). The soils in lished checklist (1987) ofW. H. Wagner and the subalpine zone have loamy textures and F. S. Wagner for pteridophytes. thicker A horizons (exceeding 50 cm). They A soil profile was described from a repre­ resemble grassland soils. sentative stand in every 200-m altitudinal The land above 2700 m is stony. The soils interval. Soil color, texture, mottling, root can be placed in the order of Entisols. They distribution, and other properties were noted lack pedogenic horizons, and incorporated for each horizon. organic matter contents are low. 202 PACIFIC SCIENCE, Volume 46, April 1992

General Description ofthe Vegetation al alien species appear preferentially in certain zones, have high constancy values, and are Using a total of III reieves comprising assembled in associations. 189 taxa, the vegetation along the transect was hierarchically classified by the Braun­ I. Forest Vegetation Blanquet synthesis table technique (Mueller­ Forest vegetation extends over the lower Dombois and Ellenberg 1974). However, our two-thirds (from 350 to 1950 m a.s.l.) of the aim was not the designation of abstract com­ slope. The forest line (1950 m a.s.l.) coincides munities into a hierarchical ranking, but rath­ with the level where the trade wind inversion er a local classification of communities along appears most frequently. the transect. The vegetation types classified Fourteen endemic taxa (Metrosideros poly­ here have not yet been compared with other morpha var. glaberrima, Cheirodendron tri­ comparable vegetation. Therefore, the no­ gynum, Vaccinium dentatum, Myrsine less­ menclature for the diagnostic species in the ertiana, Broussaisia arguta, Athyrium micro­ following description does not refer to the phyllum, Astelia menziesiana, Carex alligata, traditional terms: groups of differential spe­ Polypodium pellucidum, Hedyotis (Gouldia) cies may contain both character and differ­ terminalis, Athyrium sandwichianum, Sadleria ential species. pallida, Myrsine sandwicensis, and Smilax At the first level, two vegetation types, sandwicensis), five indigenous species (Elapho­ contrasting in physiognomy (forest vegeta­ glossum hirtum, /lex anomala, Pleopeltis thun­ tion versus treeless vegetation), are differenti­ bergiana, Asplenium polyodon, and Asplenium ated by mutually exclusive species: This re­ lobulatum), and one alien species (Rubus sults in two major differential species groups. argutus) are associated with each other. At the second level, the forest vegetation is These species differentiate the forest veg­ subdivided into two units, while the treeless etation from the treeless vegetation. They vegetation remains one unit. At the third level, have extremely broad altitudinal distributions these three units are further partitioned into spanning either the entire or most of the seven plant communities: one of these is split forested zone. Of the 20 differential species, into two subunits, yielding a total of eight only M. polymorpha var. glaberrima (a gla­ units (noda). The subunit is a floristic varia­ brous-leaved variety of M. polymorpha), I. tion of a community and is diagnosed by a anomala, and C. trigynum are potential cano­ group of species that are absent only in other py trees, and the rest of the species are subunits of the same community. The classi­ understory components. Therefore, in this fied plant communities are named by their forest vegetation even the undergrowth taxa leading species. They are shown in the sum­ in lower layers or those with lower cover marized differential table (see Appendix). values have broad altitudinal distributions. The floristically classified vegetation units The tree cover is monodominant in its are well correlated with altitude (Figure 3). entire range, dominated by a single species, The three broader units are identified by three M. polymorpha, in spite of various habitat species groups: the Elaphoglossum crassifo­ types encountered within the forest range. lium group, the Vaccinium calycinum group, Cheirodendron trigynum is persistent in the and the Dubautia menziesii group. These cor­ canopy but low in cover. respond to three broad ecological zones, re­ Trees of Metrosideros commonly, except at spectively the lowland, the montane, and the the upper limit ofclosed forest, show gnarled high-altitude zone. The seven plant communi­ morphologies that may be indicative ofoligo­ ties in the lower hierarchy are defined by seven trophy (Grubb 1977). Most of the trees are species groups with more restricted ampli­ perched on fallen logs or on mounds of peat tudes, which indicate finer zonal arrange­ underlain by waterlogged soils. Tree heights ments of vegetation within the broader alti­ and diameters are reduced in relation to tudinal zones. the severely anaerobic soils downslope. They Many alien species are sporadic in distribu­ show an unusual altitudinal pattern: trees are tion and remain unclassified. However, sever- most stunted (merely 2 m tall) at 450 m a.s.l. Vegetation ofWindward Haleakala-KITAYAMA AND MUELLER-DoMBOIS 203

Metrosideros polymorpha v. glaberrima group

Dubautia menziesii group (C) Elaphoglossum crassifolium group (A) !§M Vaccinium calycinum group (8)

Odontosoria chinensis group (A1)

Adenophorus pinnatifidus group (A2)

11II Nertera granadensis 9'OUP (81)

Pelea clusiifolia group (82) E3

SadIeria cyatheoides group (83) E§I

Sophora chrysophylla group (C1) Tetramolopium humile group (C2)

350 1000 Altitude 2000 3055m Lowland Montane High altitude A2 81 --82 83 cl c 2 Al

FIGURE 3. Altitudinal distributions of selected differential species along the transect.

They gradually increase in size upslope. The gaps, but they are scarce under closed maximum bole height of 17 m and the maxi­ canopies. mum diameter of nearly 70 cm was found at Effects of exotic plants are evident in the 1800 m a.s.l., near the upper limit ofthe closed lower forest zone, as they are in other parts of forest (1850 m). Above the closed forest limit, Hawaii (Vitousek et al. 1987). The mid and trees again decline in size. Saplings and seed­ upper segments are almost purely native. lings of Metrosideros are common in canopy The forest vegetation is floristically parti- 204 PACIFIC SCIENCE, Volume 46, April 1992 tioned into two units: A, Elaphoglossum cras­ and species number. In the differential species sifolium unit and B, Vaccinium calycinum unit. group, A. virginicus, P. cattleianum, and C. hirta are listed among the 12 worst weeds in A~ Elaphoglossum crassifolium unit (lowland Hawaii's national parks by Smith (1990). zone) The community lies entirely in the dieback The Elaphoglossum crassifolium unit occurs area, initially described by Lyon (1909) and below 1000 m a.s.l., corresponding to the more recently by Holt (1983, 1988). Dieback lowland zone. Four endemic species (E. cras­ is widespread on the lowland flat interfluve sifolium, Adenophorus hymenophylloides, An­ along the transect. This stunted plant com­ tidesma platyphyllum, and Tetraplasandra munity, which is underlain by saturated soils oahuensis), four indigenous species (Psi/otum with thick moss layers, is a consequence of complanatum, P. nudum, Huperzia phyllan­ forest dieback associated with soil deteriora­ thum, and Nephrolepis cordifolia), and four tion for native tree growth (Holt 1988). There alien species (Paspalum conjugatum, Cyperus is a possibility that the stand reduction due halpan, Rubus rosifolius, and Setaria palmi­ to dieback in turn further aggravates the folia) are associated with each other and soil waterlogging (Mueller-Dombois 1988b). differentiate this unit. The moss layers develop a peculiar mound­ The following two plant communities are forming microtopography like a high moor. distinguished in this unit: AI, M. polymorpha A similar process of a more recent stand­ var. glaberrima-Odontosoria (Sphenomeris) reduction dieback on Mauna Kea with soil chinensis community and A2, M. polymor­ degradation was termed bog formation die­ pha var. glaberrima-Adenophorus pinnatifidus back by Mueller-Dombois et al. (1977) and community. Mueller-Dombois (1980). AI. M. polymorpha var. glaberrima­ Odontosoria chinensis community (lowland A2. M. polymorpha var. glaberrima­ dieback belt, stunted open-canopy evergreen Adenophorus pinnatifidus community (intact scrub) (see Figure 4a) lowland zone, partially open to closed The M. polymorpha var. glaberrima­ canopy evergreen rainforest) (see Figure 4b) Odontosoria chinensis community, differenti­ The M. polymorpha var. glaberrima­ ated by two indigenous species (0. chinensis Adenophorus pinnatifidus community is differ­ and Machaerina mariscoides) and six alien entiated by six endemic species (Adenophorus species (Andropogon virginicus, Psidium cat­ pinnatifidus, Psychotria mariniana, Peperomia tleianum, Centella asiatica, Sacciolepis indica, obovatilimba, P. hirtipetiola, Syzygium sand­ Tibouchina herbacea, and Clidemia hirta), oc­ wicensis, and Labordia hedyosmifolia) and two curs between 450 and 600 m a.s.l. indigenous species (Freycinetia arborea and The canopy is stunted (4-6 m) in height Diplopterygium pinnatum). Most stands of and open (10-40%) in cover; the understory this community occur between 600 and 1000 is dense (100% cover). The stunted Met­ m a.s.l. with the exception of a few stands on rosideros trees are of vegetatively low vigor, lower ridge crests at 350 m, where lateral with dead or defoliated branches. Their roots drainage is better than on the surrounding flat are restricted to a few centimeters of surface interfluves. The lower ridge crests support the soils or directly abut on the soil surface and better preserved forest fragments in the die­ are covered only by mosses. Such trees are back territory. typically stilted by aerial roots. The same The vegetation is low to tall (7-12 m) and trees, on the other hand, show reproductive widely open to closed (30-80% canopy cov­ vigor, producing abundant flowers. However, er). The shrub layer is 5 m tall, sparse to dense regeneration is only sporadic, probably due to (30-80% cover), and dominated by Dicran­ the paucity of such substrates as fallen logs, opteris linearis, Cibotium chamissoi, and C. on which seedlings can establish themselves, glaucum. The herb layer is 1 m tall, sparse to as stated also by Holt (1988). Dicranopteris dense (20-100% cover), and, on imperfectly linearis and A. virginicus dominate the drained soils, is dominated by the same spe­ ground. Weeds are abundant both in cover cies as those of the shrub layer, or by Paspa- Vegetation ofWindward Haleakala-KITAYAMA AND MUELLER-DOMBOIS 205

c

FIGURE 4. Views of the lower transect communities: a, lowland dieback community (unit AI in Appendix) at 450 m, showing remaining dead snags of Metrosideros; b, lowland intact forest (A2) at 800 m; C, canopy layer of the lower and upper montane forests (B I and B2) viewed from 1350 m. 206 PACIFIC SCIENCE, Volume 46, April 1992 lum conjugatum, Cyperus halpan, and Juncus highly saturated air, which retards transpi­ planifolius on saturated soils. Trees become ration and in turn nutrient uptake (Leigh progressively mossy, and epiphytic ferns be­ 1975). come abundant toward the upper limit. The canopy is tall (10-15 m) and closed or partially open. The shrub layer (3-5 m tall) is B. Vaccinium calycinum unit (montane zone) generally dense (50 to 80% cover); there is no The Vaccinium calycinum unit is distributed single dominant shrub species, however Vac­ in the middle segment of the slope between cinium den tatum, V. calycinum, Broussaisia 1200 and 1950 m a.s.l., corresponding to arguta, and the forb Astelia menziesiana are the montane zone. Eight endemic species (V. abundant. The herb layer is dense (near 90% calycinum, Elaphoglossum wawrae, Coprosma cover) and dominated by Carex alligata on ochracea, Rubus hawaiensis, Dryopteris sub­ saturated soils. It becomes sparser (less than bipinnata, D. glabra, Ctenitis rubiginosa, and 70%) and more mixed with other herbaceous Adenophorus tripinnatifidus) and two indige­ species on somewhat better drained soils. nous species (Dryopteris wallichiana and Un­ Epiphytic ferns (Elaphoglossum hirtum, E. cinia uncinata) differentiate this unit. Three plant communities are distinguished wawrae, Mecodium recurvum, Sphaerocionium lanceolatum), terrestrial ferns (Dryopteris and in this unit: Bl, M. polymorpha var. glaber­ Asplenium), and Peperomia are abundant rima-Nerteragranadensiscommunity; B2, M. among low-growing shrub species in the herb polymorpha var. glaberrima-Pelea clusiifolia layer. Invasion by alien-weeds is rarely seen in community; and B3, M. polymorpha var. most stands. However, along the flume, where glaberrima-Sadleria cyatheoides community. the forests were cleared and regularly visited, These three communities correspond to three alien graminoids such as Juncus planifolius are subzones, the lower montane zone (Bl), the upper montane zone (B2), and the forest line abundant. The canopy trees of Metrosideros (B3). show even-sized boles. Canopy dieback is locally evident in this montane zone where Bl. M. polymorpha var. glaberrima-Nertera soils are saturated, but dieback stands are not granadensis community (lower montane zone, distinguishable floristically from intact ones closed canopy evergreen moss forest) (see in the differential table. Figure 4c) The M. polymorpha var. glaberrima­ B2. M. polymorpha var. glaberrima-Pelea Nertera granadensis community, differenti­ clusiifolia community (upper montane zone, ated by 11 endemic taxa (M. polymorpha closed canopy evergreen rainforest) (see var. incana, Peperomia expallescens, P. mac­ Figure 4c) raeana, Xiphopteris saffordii, Labordia venosa, The M. polymorpha var. glaberrima-Pelea Cyrtandra hashimotoi, Adenophorus mon­ clusiifolia community, differentiated by three tanus, Psychotria hawaiiensis, Cyrtandra pla­ endemic species (P. clusiifolia, Asplenium nor­ typhylla, Thelypteris sandwicensis, and Dryop­ male, and Peperomia membranacea), occurs teris acutidens) and three indigenous species between 1750 and 1950 m a.s.l. This commu­ (Nertera granadensis, Grammitis hookeri, nity marks the upper limit of closed forest and Korthalsella complanata), occurs between beyond which forest canopies become open. 1200 and 1700 m a.s.l. Clouds persistently The mosses no longer prevail, and epiphytic envelope this altitudinal zone. Consequently, mosses are largely replaced by lichens. The epiphytic mosses and ferns grow abundantly canopy layer is tall (15 m) and closed. Acacia in all strata, forming mossy forests. Met­ koa becomes codominant near the upper limit rosideros polymorpha var. incana (a pubescent of the community range (1900 m a.s.l.). The variety of M. polymorpha) intermixes with canopy trees are ground-rooted, unlike those M. polymorpha var. glaberrima with various at lower elevation. The shrub layer is sparse degrees ofcover in the canopy. Both varieties (5% cover); Coprosma ochracea, Vaccinium are sclerophyllous. Sclerophylly may be re­ calycinum, and P. clusiifolia are frequently lated primarily to the oligotrophic montane encountered. The herb layer is dense (100% soils as suggested by Grubb (1977), or to the cover), predominated by Athyrium sandwichi- Vegetation ofWindward Haleakala-KITAYAMA AND MUELLER-DoMBOIS 207

anum interspersed with other terrestrial ferns range from the forest line to the tree line at such as Dryopteris, Asplenium, and Ctenitis. 2200 m a.s.l. The dense fern layer and the occurrence of This unit contains the following two plant ground-rooted trees are related to the well­ communities, which are physiognomically drained soils. Invasion by weeds into the and floristically distinct: CI, Sophora chryso­ community is scarcely seen. phylla community and C2, Tetramolopium humile community. B3. M. polymorpha var. glaberrima­ Sadleria cyatheoides community (forest line, Cl. Sophora chrysophylla community ecotone community between the upper (subalpine zone, tropical subalpine scrub) montane forest and the subalpine scrub) (see Figure 5b) (see Figure 5a) The Sophora chrysophylla community, dif­ The M. polymorpha var. glaberrima­ ferentiated by three endemic shrubs (S. chry­ Sadleria cyatheoides community occurs at the sophylla, Coprosma montana, and Geranium forest line between 1900 and 1950 m a.s.l. The cuneatum ssp. tridens) and one endemic sedge differential species group includes four en­ (Carex wahuensis), occurs between 1950 and demic taxa: Sadleria cyatheoides, M. poly­ 2700 m a.s.l. The canopy shrub layer is low morpha var. polymorpha, Oreobolus furcatus, (1.5-3 m in height) and widely open (5-50% and Polystichum bonseyi. The forest structure cover). The herb layer is relatively dense consists ofa low (5-8 m tall) and partially to (40-90% cover). The differential shrub spe­ widely open canopy hiyer (5-60% cover) and cies and Styphelia tameiameiae prevail in the a dense herb layer (70-100% cover). M. shrub layer. In the herb layer, endemic species polymorpha var. polymorpha (a tomentose­ (Dubautia menziesii, Coprosma ernodeoides, leaved variety of M. polymorpha) intermixes Deschampsia nubigena, Vaccinium reticula­ with M. polymorpha var. glaberrima in the tum, and Luzula hawaiiensis) and alien species canopy at the lower limit of the community, (Hypochoeris radicata, Anthoxanthum odora­ but forms pure stands at the upper limit. tum, and Holcus lanatus) are prevalent. The Acacia koa is another canopy tree species that community shows some xeromorphic adapta­ becomes abundant locally. The herb layer is tions to the arid environment, as reflected in dominated by either Dryopteris wallichiana or sclero-microphylly. S. cyatheoides, with two alien grass species The community is further subdivided into (Anthoxanthum odoratum and Holcus lanatus) two subunits: Cia, Prunella vulgaris subunit as codominants. The community can be con­ and Clb, Trisetum glomeratum subunit. The sidered an ecotone between the closed upper P. vulgaris subunit is differentiated by Prunel­ montane forest and the subalpine scrub. Sev­ la vulgaris, Epilobium billardierianum ssp. cin­ eral subalpine elements such as Vaccinium ereum, and Coprosma ernodeoides. These spe­ reticulatum, Coprosma ernodeoides, and Des­ cies occur in the lower segment of the subal­ champsia nubigena consistently occur in the pine zone (1950-2400 m a.s.l.). The T. glo­ community. meratum subunit is differentiated by Trisetum glomeratum, Rumex acetosella, Pellaea terni­ II. Treeless Vegetation folia, Asplenium adiantum-nigrum, A. tricho­ manes, and Dodonaea viscosa, occurring at C. Dubautia menziesii unit (high-altitude higher elevation (2300-2700 m a.s.l.) and zone) marking the shrub line. Sparse vegetation characterizes the land­ scape above 1950 m a.s.l. Mutually exclusive C2. Tetramolopium humile community with the forest vegetation, a single endemic (alpine zone, tropical alpine desert) (see species differentiates the Dubautia menziesii Figure 5c) unit, which lies entirely in the subalpine and The Tetramolopium humile community, dif­ alpine treeless vegetation (Figure 3). ferentiated by two endemic species (T. humile Tall trees are absent from this vegetation ssp. haleakalae and Argyroxiphium sandwic­ type. The only exception are widely scattered ensis ssp. macrocephalum), occurs above 2700 M. polymorpha var. polymorpha trees, which m a.s.l. The substrates are pyroclastic ashes 208 PACIFIC SCIENCE, Volume 46, April 1992

FIGURE 5. Views ofthe upper transect communities: a, forest line (unit 83 in Appendix) at 1950 m; b, lower subalpine scrub (CIa) at 2100 m; c, alpine desert (C2) near the summit. Vegetation of Windward Haleakala-KITAYAMA AND MUELLER-DoMBOIS 209 and rocks. The community is very sparse, with The lower montane forest is relatively more less than 5% cover, low « I m in height diverse in flora than the lowland intact forest except when Argyroxiphium flowers), and (A2). The lowland dieback community (AI) rather barren in physiognomy. A. sandwic­ has a lowered species richness (21 species per ensis ssp. macrocephalum (silversword) is a stand), comparable to that of the forest line monocarpic giant rosette of the Compositae, community (B3, 23.2 species per stand). a characteristic life form of tropical alpine zones elsewhere (e.g., in the afroalpine zones Relationships between the Classified and the Andean mountains). The other con­ Communities stituents include the native species Dubautia menziesii, Trisetum glomeratum, Asplenium Figure 6 depicts community relationships adiantum-nigrum, A. trichomanes, Styphelia among the eight classified units in the form tameiameiae, and Deschampsia nubigena and of a dendrogram. The hierarchical ranking one alien species, Hypochoeris radicata. shown in the Appendix was based on the selection of diagnostic species, but the rela­ tionships depicted in Figure 6 are based on the Species Richness along the Transect species quantities of all constituents in each The numbers ofspecies in the communities unit. The calculation of community simi­ are shown in Table I. The total number of larities is based on species constancy values species per community ranges from 14 in the (%) per unit, using S0rensen's similarity alpine desert (unit C2 in the Appendix) to 103 index modified for quantitative application species in the lower montane forest (BI). The (Mueller-Dombois and Ellenberg 1974): mean species number per stand ranges from IS = 2Mwj(MA MB) 7.3 species (4-10) in the alpine desert to 43.5 + species (31-55) in the lower montane forest. where Mw = sum of the smaller constancy

TABLE I SPECIES RICHNESS OF THE CLASSIFIED PLANT COMMUNITIES Total number of plant species per community and mean species number per stand for each community are shown. The communities are indicated by codes: A I. Lowland dieback (M. polymorpha var. glaberrima-O. chinensis community) A2. Lowland intact forest (M. polymorpha var. glaberrima-A. pinnatifidus community) BI. Lower montane forest (M. polymorpha var. glaberrima-N. granadensis community) 82. Upper montane forest (M. polymorpha var. glaberrima-P. c/usiifolia community) B3. Forest line (M. polymorpha var. glaberrima-S. cyatheoides community) Cia. Lower subalpine scrub (S. chrysophy//a community, P. vulgaris subunit) Clb. Upper subalpine scrub (S. chrysophy//a community, T. glomeratum subunit) C2. Alpine desert (T. humile community)

LOWLAND MONTANE HIGH ALTITUDE

PLANT COMMUNITY Al A2 BI B2 B3 Cia Clb C2

Total number 47 100 103 52 56 28 26 14 Mean per stand 21 38.4 43.5 24.7 23.2 14.2 15.2 7.3 (Range) 19-25 14-54 31-55 15-31 13-32 11-17 11-18 4-10 210 PACIFIC SCIENCE, Volume 46, April 1992

C2. Alpine desert

40.5 C1b. Upper subalpine scrub I 68.5 I C1a. Lower subalpine scrub

83. Forest line 7.3 I 53.4 R 2. Upper montane forest

35.2 - 8 1. Lower montane forest I 52.7 A2. Lowland intact forest 24.2

A1. Lowland dieback

I I o 50 Percentage Similarity (Ofo) 100

FIGURE 6. Dendrogram analysis applied to the classified plant communities on the windward slope of Haleakala.

values of the species common to two com­ the alpine community. Several common alien munities; MA and MB = sum ofthe constan­ species, spanning both zones, are responsible cy values of all species in each of the two for this high similarity. communities. Overall, the communities share relatively high similarities despite the differences in their environments. The lower subalpine scrub DISCUSSION (unit CIa in the Appendix) and the upper Altitudinal Zonation subalpine scrub (CIb) have the highest simi­ larity since they are subunits of the same Previous workers (Egler 1939, Krajina community. The lowland intact forest (A2) 1963, Knapp 1965, Mueller-Dombois and has a higher similarity with the lower montane Spatz 1981, and Gagne and Cuddihy 1990) forest (Bl) than with the lowland dieback have variously described and characterized community (AI), although Al and A2 are altitudinal vegetation zones of the Hawaiian parts of the same unit in the hierarchy (Ap­ Islands. The results of the strictly floristically pendix 1 and Figure 3). This is because the defined study reported here show altitudinal lowland intact forest shares more species with zones comparable to those of Knapp (1965), the lower montane forest than with the low­ Mueller-Dombois and Spatz (1981), and land dieback community. The lowland die­ Gagne and Cuddihy (1990). Moreover, the back community has low similarities with any number ofaltitudinal zones agrees with those of the intact forest communities (A2, B1, B2, described for continental tropical high moun­ and B3) (joining them at the level of 24.2%), tains by Grubb (1974), who used vegetation suggesting that canopy dieback in that zone structure and physiognomy as the principal impoverished the species composition. classifying criteria. The similarity between the alpine (C2) and An important difference from the previous subalpine (CIa and CI b) communities is rela­ studies is the hierarchical arrangement of tively high despite the low species richness of vegetation in this study. In descending order Vegetation of Windward Haleakala-KITAYAMA AND MUELLER-DoMBOIS 211 of the hierarchy, two major species groups same altitudes on both mountains (i.e., the have defined the two broadest ecological upper limits of the montane and subalpine zones that split at the trade wind inversion. At zones). the second level, three floristic units define the In comparison to the high mountains of lowland, the montane, and the high-altitude Malesia, with which the Hawaiian islands (subalpine and alpine) zones. At the third have some floristic affinity (Wagner et al. level, seven plant communities and two addi­ 1990), the limits ofsubalpine and alpine zones tional subunits within one ofthe communities ofHaleakala are more than 1000 m lower than designate finer altitudinal zones (Figure 3). Of those of Malesian mountains (except on re­ these communities, the M. polymorpha var. cent Malesian volcanoes). The difference is glaberrima-Odontosoria chinensis community especially evident when forest line~ are com­ defines the lowland dieback belt. This is con­ pared: the forest line ofMt. Wilhelm (4672 m) sidered a retrogressive variation of the low­ in Papua New Guinea is at 3900 m a.s.l. land rainforest zone, because the impover­ (Wade and McVean 1969), that of Mt. Kin­ ishment of native species and invasion of abalu (4101 m) in Borneo is at 3700 m a.s.1. aliens seem to be a direct consequence of (Kitayama 1987), while that of Haleakala is canopy dieback. only at 1950 m. The forest line of Mt. Kin­ Species groups that differentiate lower units abalu coincides with the daily ground-frost in descending order of the hierarchy consist line; that of Mt. Wilhelm is found above the mostly of herb and shrub species (except daily ground-frost line. On Haleakala the Syzygium sandwicensis and M. polymorpha ground-frost line (2700 m a.s.l.) coincides with var. incana), while canopy tree species segre­ the lower limit ofthe alpine zone, which is 700 gate the forest from the nonforest vegetation. m above the forest line. Therefore, any finer altitudinal differentiation The two Malesian mountains (Wilhelm in the forest vegetation is floristically recog­ and Kinabalu) were uplifted between the late nizable only by understory species. This fact Pliocene and the early Pleistocene. They are contrasts with the situation in continental rather similar in geological age to Haleakala, tropical mountains where zonal differentia­ which originated in the late Pleistocene. The tion is often evident through changes in cano­ Malesian mountains are, however, said to py species composition (Whitmore 1975). support a more harmonic flora and are defi­ The upper limits ofthe subalpine and mon­ nitely much richer in species than is Halea­ tane zones on Haleakala are similar to those kala. One reason for the zonal depression is on Mauna Loa (Mueller-Dombois and Spatz probably related to floristic differences. For 1981) on the neighboring island of Hawaii. instance, the Hawaiian flora lacks podocar­ This is so in spite of the montane segment paceous tree species, which are dominant studied on Mauna Loa being "seasonal," and subalpine elements in circum-Pacific regions thus different from the year-round perhumid of the Southern Hemisphere (Troll 1958). In montane environment on Haleakala. More­ addition, it seems that no highland-adapted over, the montane seasonal environment on tree species, which are tolerant ofthe cold and Mauna Loa is dominated by Acacia koa, while arid environment, have evolved on the Hawai­ on Haleakala there is a narrow belt of Acacia ian mountains. This may be due to some koa where the upper montane zone changes evolutionary constraints of Hawaiian tree into the subalpine zone. In the subalpine zone, species (e.g., no highland-adapted Syzygium Metrosideros trees are more abundant on is found in Hawaii). Mauna Loa than on Haleakala. When com­ pared with that of a higher mountain of the Species Turnover same floristic region, vegetation of a lower mountain is often telescoped into depressed, On Haleakala, altitudinal species turnover, narrower zones than those of the higher which is the compositional change along an mountain. Despite the differences in the domi­ altitudinal gradient by spatial species dis­ nance type and in the summit height (ca. 1000 placements, is exceedingly low despite the m), major species turnover points occur at the diverse habitat types along the transect. The 212 PACIFIC SCIENCE, Volume 46, April 1992 low species turnover stems from the fact that may be considered a characteristic ofisolated many species are associated over broad alti­ island mountains. tudinal ranges. For instance, 20 species are associated and range from the lowland to the forest line (M. polymorpha var. glaberrima Altitudinal ConspecificfGeneric Segregation group in Figure 3). The boundary between the M. polymorpha Subdivisions of the forest vegetation were var. glaberrima group and the D. menziesii recognizable notably by the species of the group (Figure 3) shows an abrupt species understory. However, our results also show turnover. However, several species range an incipient canopy tree segregation: a con­ across the boundary formed by the two major specific segregation of M. polymorpha into species groups. These species are represented zonal varieties. Metrosideros polymorpha is by Styphelia tameiameiae, which sho~s the present with three morphological varieties most individualistic distribution and spans along the transect (Figure 7). Of these, var. from the lower montane zone to the summit glaberrima occupies the broadest range. Over­ area without associated species (see Appen­ lapping with this, var. incana is confined to dix). Consequently, such unassociated species the lower montane zone. Var. polymorpha is lower somewhat the otherwise high turnover sharply separated from the first variety, and rate between the two species groups. This low occurs above the inversion. Stemmermann species turnover along the mountain slope (1983) stated that the pubescent character,

V. reticulatum V. calycinum Vaccinium V. dentatum

C. montana C. ochracea C. pubens Coprosma C. ernodeoides

P. orbicularis Pelea P. clusiifolia

P. obovatilimba P. membranacea P. expallescens--- Peperomia P. macraeana P hirfjoefjola P. eekana

Metrosideros polymorpha v. incana v. polymorpha v. glaberrima I large coriaceous leaf small sclerophyllous leaf I I I I 390 1.000 2,000 3,055 Altitude (m)

FIGURE 7. Altitudinal distribution patterns ofselected conspecificjcongeneric taxa along the transect. Vegetation of Windward Haleakala-KITAYAMA AND MUELLER-DoMBOIS 213 which is a principal morphological distinction in altitudinal zones similar in number to those for the varieties, is genetically fixed. found on species-rich mountains. The impli­ Moreover, within the glabrous variety (gla­ cation is that factors controlling the zonal berrima) a large coriaceous leaf variant and a vegetation patterns and their limits are pre­ small sclerophyllous leaf variant are discern­ dominantly climatic since the differential spe­ able, although not recognized by taxonomists. cies are well correlated with altitude. Tem­ These morphological leafvariants are dissoci­ perature lapse rates are probably a chief ated from each other in distribution, and determinant because air temperature consis­ respectively confined to the lowland and to the tently decreases upslope independently ofoth­ montane zone. The leaf variants may be eco­ er factors on both continental and oceanic­ typic (Corn and Hiesey 1973) and are likely in island mountains. On Haleakala, moisture is the process of altitudinal segregation. another important factor, as reflected in the In addition to Metrosideros, altitudinal seg­ sharp boundary at the inversion. regation ofmutually related taxa is commonly Grubb (1977) discussed causative factors of seen also at the specific level. Examples of tropical mountain zonation. He suggested genera showing such species segregations that the upper limits ofspecies are set primari­ along the altitudinal gradient include Co­ ly by temperature and the lower limits by prosma, Vaccinium, Asplenium, Peperomia, competition. He further suggested that other Adenophorus, Dryopteris, and Pelea. Among adverse effects such as soil nutrient deficien­ these, the genera Coprosma, Vaccinium, and cies, waterlogging, and reduced light in the Asplenium have exceedingly broad ranges: cloud zone depress the potential distribution 600-2700 m, 600-2800 m, and 600-3000 m limits. a.s.l., respectively. Figure 7 depicts altitudinal The question ofwhether lower limits are set species distributions of a few representative by competition is intricate. This question can angiosperm genera. In these genera, constitu­ only be resolved by an experimental ap­ ent species are probably genetically closely proach. Distribution analysis can only sug­ related. Morphological resemblance and gest a mechanism. In our example, Styphelia mutually exclusive distributions in Pelea tameiameiae, which has the most individualis­ haleakalae and P. clusiifolia, and in Coprosma tic distribution, seems little affected by species pubens, C. ochracea, and C. montana indicate competition, while the lower limit of the their vicarious or ecotypic relationships. heliophytic subalpine species Coprosma erno­ Congeneric segregation has also been deoides appears to be controlled by competi­ reported from the Mauna Loa transect study tion with montane trees. Coprosma ernode­ (Mueller-Dombois 1981b). Such segregations oides is occasionally found farther downslope are also found on upper slopes ofcontinental in anthropogenically cleared areas. Recur­ tropical mountains (e.g., altitudinally vicari­ rence of the same endemic shrub taxa at high ousafroalpine species [Hedberg 1986] and and low elevations has also been recorded adaptive altitudinal radiation [Lee and Lowry along the Mauna Loa transect. There, the 1980)). The examples on the oceanic islands double occurrence was attributed to the natu­ relate to the invasion and subsequent adap­ ral openness associated with pioneer habitats tation to mountains volcanically emerged (Mueller-Dombois and Spatz 1981). from the ocean (Carlquist 1974), while those Whittaker (1975) suggested that the distri­ on the continental mountains appear to be bution of species along an environmental related to the upward adaptation to the post­ gradient is a function of both adaptation and glacial high-altitude environment. competition. He further suggested that the evolving patterns of species packing result Community Organization in increasingly narrower distribution ranges with random assortment ofindividual species This floristic analysis, applied to an isolated rather than associated species patterns. The oceanic-island mountain with a biologically study on Haleakala reported here has shown impoverished flora, has unexpectedly resulted that a large number of species assemble in 214 PACIFIC SCIENCE, Volume 46, April 1992 associated patterns, confirming an earlier States, No. 60-51. U.S. Department of study done on Mauna Loa during the IBP Commerce, Weather Bureau, Washington, project (Mueller-Dombois 1981 c). D.C. One of the associations is indicative of BURTON, P. J., and D. MUELLER-DoMBOIS. invading patterns ofalien species due to cano­ 1984. Response of Metrosideros polymor­ py dieback (i.e., the Odontosoria chinensis pha seedlings to experimental canopy open­ group, which largely consists of aliens). The ing. Ecology 65 :779-791. associations consisting mostly of native spe­ CARLQUIST, S. 1974. Island biology. Columbia cies are variously distributed in range. University Press, New York. Evolutionary processes of radiation may CORN, C. A., and W .. M. HIESEY. 1973. Alti­ progress spatially in two scenarios: A group tudinal variation in Hawaiian Metrosideros. of generalistic species may assemble over a Am. J. Bot. 60:991-1002. broad range limited only by environmental EGLER, F. E. 1939. Vegetation zones ofOahu, constraints. Subsequently, the individual taxa Hawaii. Emp. For. J. 18:44-57. split by evolutionary adaptation into related GAGNE, W. C., and L. CUDDIHY. 1990. Vege­ congeneric taxa with narrower amplitudes. tation. Pages 45-114 in W. L. Wagner, D. The second scenario involves species invasion R. Herbst, and S. H. Sohmer, Manual of with slow range expansion accompanied by the flowering plants of Hawai'i, vol. 1. adaptation. University of Hawaii Press, Honolulu. An initially low species packing may allow GIAMBELLUCA, T. W., M. A. NULLET, and the first process of generalist expansion to be T. A. SCHROEDER. 1986. Rainfall atlas of more prevalent. In contrast, where species Hawaii. Department of Land and Natural packing has advanced to a higher degree, Resources, Division of Water and Land invasion ofnew species is expected to proceed Development, State of Hawaii. Report less explosively, and subsequent range exten­ R 76. sions can only occur by adaptive changes or GRUBB, P. J. 1974. Factors controlling the dis­ competitive displacement of other taxa. The tribution offorest-types on tropical moun­ overall outcome is the same. However, since tains: New facts and a new perspective. this study revealed a large number of species Pages 13-46 in J. R. Henley, ed. Altitudinal with broad-ranging distribution patterns, it zonation in Malesia, Transactions of the seems likely that the first scenario is more third Aberdeen-Hull symposium on Male­ typical for island mountains, where inter­ sian ecology, Hull, 1973. Department specific competition is initially only a minor of Geography, University of Hull, Hull, constraint. England. ---. 1977. Control of forest growth and distribution on wet tropical mountains: ACKNOWLEDGMENTS With special reference to mineral nutrition. We thank Reiko Kitayama and Yutaka Annu. Rev. Ecol. Syst. 8:83-107. Kitayama for assisting our fieldwork, and HEDBERG, O. 1986. Origins of the afroalpine Donald Drake for commenting on the manu­ flora. Pages 443-468 in F. Vuilleumier and script. We are grateful to East Maui Irrigation M. Monasterio, eds. High altitude tropical Company, the Hawaii Nature Conservancy, biogeography. Oxford University Press, and Haleakala National Park for permitting New York. the fieldwork in their protected areas. HOLT, R. A. 1983. The Maui forest trouble: A literature review and proposal for research. Hawaii Botanical Science Paper42. Univer­ sity of Hawaii, Honolulu. LITERATURE CITED ---. 1988. The Maui forest trouble: Re­ assessment of a historic forest dieback. BLUMENSTOCK, D. 1. 1961. Climates of the M.S. thesis, University ofHawaii at Manoa, States, Hawaii. Climatology of the United Honolulu. Vegetation ofWindward Haleakala-KITAYAMA AND MUELLER-DoMBOIS 215

HUBBELL, T. H. 1968. The biology of islands. H. L. Carson, eds. Island ecosystems. Proc. Natl. Acad. Sci. U.S.A. 60:22-32. US/IBP Synthesis Series 15. Hutchinson JUVIK, J. 0., and P. C. EKERN. 1978. A Ross, Stroudsburg, Pennsylvania. climatology of mountain fog on Mauna ---. 1981b. Island ecosystems: What is Loa, Hawaii Island. Hawaii Univ. Water unique about their ecology? Pages 485-501 Resour. Res. Cent. Tech. Rep. 118. in D. Mueller-Dombois, K. W. Bridges, and KITAYAMA, K. 1987. Vegetation of Mount H. L. Carson, eds. Island ecosystems. US/ Kinabalu. Rev. For. Culture 8: 103-113 (in IBP Synthesis Series 15. Hutchinson Ross, Japanese). Stroudsburg, Pennsylvania. KNAPP, R. 1965. Die Vegetation von Nord­ ---. 1981c. Spatial integration of the or­ und Mittelamerika und der Hawaii Inseln. ganisms studied along the transect. Pages Fischer, Stuttgart. 181-210 in D. Mueller-Dombois, K. W. KOPPEN, W. 1936. Das geographische System Bridges, and H. L. Carson, eds. Is­ der Klimate. Handbuch der Klimatologie, land ecosystems. US/IBP Synthesis Series Band 1, Teil C. Gebriider Borntraeger, 15. Hutchinson Ross, Stroudsburg, Penn­ Berlin. sylvania. KRAJINA, V. J. 1963. Biogeoclimatic zones of ---. 1986. Perspectives for an etiology of the Hawaiian Islands. Hawaii. Bot. Soc. stand-level dieback. Annu. Rev. Ecol. Syst. News!. 2: 93-98. 17: 221-243. LEE, D. W., and J. B. LOWRY. 1980. Plant ---. 1987. Forest dynamics in Hawaii. speciation on tropical mountains: Lepto­ Trends Ecol. Evol. 2: 216-220. spermum (Myrtaceae) on Mount Kinabalu, ---. 1988a. Towards a unifying theory for Borneo. J. Linn. Soc. London Bot. 80: 223­ stand-level dieback. GeoJournal 17: 249­ 242. 251. LEIGH, E. G. 1975. Structure and climate in ---. 1988b. Vegetation dynamics and tropical rain-forest. Annu. Rev. Ecol. Syst. slope management on the mountains of 6: 67-86. the Hawaiian Islands. Environ. Conserv. LYON, H. L. 1909. The forest disease on Maui. 15:255-260. Hawaii. Plant. Rec. 1: 151-159. MUELLER-DoMBOIS, D., and H. ELLENBERG. LYONS, S. W. 1979. Summer weather on 1974. Aims and methods ofvegetation ecol­ Haleakala, Maui. Department of Meteo­ ogy. John Wiley & Sons, New York. rology, University ofHawaii, UHMET 79­ MUELLER-DoMBOIS, D., J. D. JACOBI, R. G. 09. Honolulu. COORAY, and N. BALAKRISHNAN. 1977. MENDONCA, B. G., and W. T. IWAOKA. 1969. 'Ohi'a rain forest study: Final report. Co­ The trade wind inversion at the slopes of operative National Park Resources Studies Mauna Loa, Hawaii. J. Appl. Meteorol. Unit Technical Report 20. Department of 8: 213-219. Botany, University of Hawaii at Manoa, MUELLER-DoMBOIS, D. 1967. Ecological rela­ Honolulu. tions in the alpine and subalpine vegetation MUELLER-DoMBOIS, D., and G. SPATZ. 1981. on Mauna Loa, Hawaii. J. Indian Bot. Soc. Altitudinal distribution oforganisms along 96: 403-411. an island mountain transect: Vascular ---. 1980. 'Ohi'a rain forest study: Eco­ plants. Pages 77-97 in D. Mueller­ logical investigations of the 'ohi'a dieback Dombois, K. W. Bridges, and H. L. Carson, problem in Hawaii. Hawaii Agric. Exp. Stn. eds. Island ecosystems. US/IBP Synthesis Misc. Publ. 183. College of Tropical Agri­ Series 15. Hutchinson Ross, Stroudsburg, culture and Human Resources, University Pennsylvania. of Hawaii at Manoa, Honolulu. NOGUCHI, Y, H. TABUCHI, and H. HASEGAWA. ---. 1981a. Some bioenvironmental con­ 1987. Physical factors controlling the for­ ditions and the general design of IBP mation of patterned ground on Haleakala, research in Hawaii. Pages 3-32 in D. Maui. Geogr. Ann. 69: 329-342. Mueller-Dombois, K. W. Bridges, and SMITH, C. W. 1990. Weed management in 216 PACIFIC SCIENCE, Volume 46, April 1992

Hawaii's national parks. Monogr. Syst. VITOUSEK, P. M., L. L. LOOPE, and C. P. Bot. Mo. Bot. Gard. 32: 223-234. STONE. 1987. Introduced species in Hawaii: STEARNS, H. T. 1985. Geology of the State of Biological effects and opportunities for eco­ Hawaii, 2d ed. Pacific Books, Palo Alto, logical research. Trends Eco!. Evo!. 2: 224­ California. 227. STEMMERMANN, L. 1983. Ecological studies of WADE, L. K., and D. N. MCVEAN. 1969. Mt. Hawaiian Metrosideros in a successional Wilhelm studies. 1. The alpine and sub­ context. Pac. Sci. 37: 361-373. alpine vegetation. Publication BG/l. The STONE, C. P. 1985. Alien animals in Hawaii's Australian National University Press, native ecosystems: Toward controlling the Canberra. adverse effects of introduced vertebrates. WAGNER, W. L., D. R. HERBST, and S. H. Pages 251-297 in C. P. Stone and J. M. SOHMER. 1990. Manual of the flowering Scott, eds. Hawaii's terrestrial ecosystems: plants of Hawai'i. University of Hawaii Preservation and management. Coopera­ Press, Honolulu. 2 vols. tive National Park Resources Studies Unit, WHITEAKER, L. D. 1983. The vegetation University ofHawaii at Manoa, Honolulu. and environment in the crater district of TROLL, C. 1958. Tropical mountain vegeta­ Haleakala National Park. Pac. Sci. 37: 1­ tion. Proc. 9th Pac. Sci. Congr. 20: 37-45. 24. VAN DER HAMMEN, T., D. MUELLER-DoMBOIS, WHITMORE, T. C. 1975. Tropical rain forests and M. A. LITTLE. 1989. Manual of me­ of the Far East. Oxford University Press, thods for mountain transect studies. Com­ London. parative studies of tropical mountain eco­ WHITTAKER, R. H. 1975. Communities and systems. International Union of Biological ecosystems, 2d ed. MacMillan, New York. Sciences-Decade of the Tropics. IUBS, Paris.

APPENDIX

SUMMARIZED DIFFERENTIAL TABLE OF THE VEGETATION ON THE WINDWARD SLOPE OF HALEAKALA, MAUl, HAWAII Figures are constancy values (%) per vegetation unit. **, alien species; *, indigenous species; unmarked, endemic species. Forest Vegetation A. Elaphoglossum crassifolium unit AI. M. polymorpha var. glaberrima-O. chinensis community A2. M. polymorpha var. glaberrima-A. pinnatifidus community B. Vaccinium calycinum unit BI. M. polymorpha var. glaberrima-N. granadensis community B2. M. polymorpha var. glaberrima-P. clusiifolia community B3. M. polymorpha var. glaberrima-S. cyatheoides community Treeless Vegetation C. Dubautia menziesii unit CI. Sophora chrysophylla community Cia. Prunella vulgaris subunit CIb. Trisetum glomeratum subunit C2. Tetramolopium humile community

PLANT COMMUNITY CODE AI A2 BI B2 B3 CIa Clb C2 NUMBER OF RELEvEs 8 19 23 9 12 16 13 II Differential species of the forest vegetation 100 100 100 100 75 0 0 0 Metrosideros polymorpha var. glaberrima Cheirodendron trigynum 0 95 100 100 75 0 0 0 Vaccinium dentatum 0 95 100 67 17 0 0 0 Myrsine lessertiana 0 58 96 78 50 0 0 0 Elaphoglossum hirtum* 13 S3 91 100 33 0 0 0 Broussaisia arguta 0 79 100 33 0 0 0 0 Vegetation of Windward Haleakala-KITAYAMA AND MUELLER-DOMBOIS 217

PLANT COMMUNITY CODE Al A2 BI B2 B3 CIa Clb C2 NUMBER OF RELEvEs 8 19 23 9 12 16 13 11

Athyrium microphyl/um 0 26 100 44 58 0 0 0 Hedyotis (Gouldia) terminalis 0 74 87 11 0 0 0 0 Astelia menziesiana 0 47 100 11 8 0 0 0 Sadleria pallida 0 53 87 22 17 0 0 0 Carex alligata 0 5 70 56 67 6 0 0 Polypodium pel/ucidum 0 26 87 44 17 0 0 0 Asplenium polyodon* 13 32 57 89 17 0 0 0 Myrsine sandwicensis 0 63 70 22 0 0 0 0 Athyrium sandwichianum 0 42 44 100 17 0 0 0 Ilex anomala* 13 47 52 22 25 0 0 0 Pleopeltis thunbergiana* 38 16 39 78 33 0 0 0 Smilax sandwicensis 0 58 44 11 8 0 0 0 Asplenium lobulatum* 0 37 44 33 0 0 0 0 Rubus argutus** 0 47 4 11 75 0 0 0 Differential species of C Dubautia menziesii 0 0 0 0 0 50 100 100 Differential species of A Elaphoglossum crassifolium 88 95 0 0 0 0 0 0 Paspalum conjugatum** 100 68 0 0 0 0 0 0 Adenophorus hymenophyl/oides 63 58 13 0 0 0 0 0 Psi/otum complanatum* 38 68 4 0 0 0 0 0 Antidesma platyphyl/um 63 47 0 0 0 0 0 0 Tetraplasandra oahuensis 13 58 9 0 0 0 0 0 Cyperus halpan** 50 47 0 0 0 0 0 0 Huperzia phyl/anthum* 63 42 0 0 0 0 0 0 Nephrolepis cordifolia* 25 58 0 0 0 0 0 0 Rubus rosifolius** 25 47 9 0 0 0 0 0 Psi/otum nudum* 38 32 0 0 0 0 0 0 Setaria palmifolia** 13 42 0 0 0 0 0 0 Differential species of B Vaccinium calycinum 0 0 96 100 83 0 0 0 Dryopteris wallichiana* 0 0 91 89 83 25 0 0 Elaphoglossum wawrae 0 0 100 100 50 0 0 0 Dryopteris glabra 0 0 61 89 58 0 0 0 Coprosma ochracea 0 0 39 100 83 0 0 0 Rubus hawaiensis 0 0 30 89 83 0 0 0 Uncinia uncinata* 0 0 61 78 33 0 0 0 Ctenitis rubiginosa 0 0 44 100 17 0 0 0 Dryopteris subbipinnata 0 0 4 78 42 0 0 0 Adenophorus tripinnatifidus 0 0 26 44 8 0 0 0 Differential species of Al Odontosoria (Sphenomeris) chinensis* 100 11 0 0 0 0 0 0 Andropogon virginicus** 88 16 0 0 0 0 0 0 Psidium cattleianum** 88 5 0 0 0 0 0 0 Centel/a asiatica** 75 5 0 0 0 0 0 0 Sacciolepis indica** 50 16 0 0 0 0 0 0 Tibouchina herbacea** 75 0 0 0 0 0 0 0 Clidemia hirta** 50 5 0 0 0 0 0 0 Machaerina mariscoides meyenii* 25 0 0 0 0 0 0 0 Differential species of A2 Adenophorus pinnatifidus 0 84 4 0 0 0 0 0 Freycinetia arborea* 0 63 13 0 0 0 0 0 Psychotria mariniana 0 68 0 0 0 0 0 0 Peperomia obovatilimba 0 63 4 0 0 0 0 0 Syzygium sandwicensis 0 58 0 0 0 0 0 0 Peperomia hirtipetiola 0 42 13 0 0 0 0 0 218 PACIFIC SCIENCE, Volume 46, April 1992

PLANT COMMUNITY CODE Al A2 BI B2 B3 Cia Clb C2 NUMBER OF RELEvEs 8 19 23 9 12 16 13 II

Diplopterygium pinnatum* 0 42 17 0 0 0 0 0 Labordia hedyosmifolia 0 26 0 0 0 0 0 0 Differential species of BI Nertera granadensis* 0 0 100 11 0 0 0 0 Peperomia expallescens 0 0 91 II 0 0 0 0 Metrosideros polymorpha var. incana 0 16 78 0 0 0 0 0 Peperomia macraeana 0 16 74 0 0 0 0 0 Xiphopteris saffordii 0 11 74 0 0 0 0 0 Grammitis hookeri* 0 0 78 0 0 0 0 0 Cyrtandra hashimotoi 0 11 65 0 0 0 0 0 Labordia venosa 0 0 74 0 0 0 0 0 Adenophorus montanus 0 0 65 0 0 0 0 0 Psychotria hawaiiensis 0 0 65 0 0 0 0 0 Cyrtandra platyphylla 0 11 52 0 0 0 0 0 Korthalsella complanata* 0 16 44 0 0 0 0 0 Thelypteris sandwicensis 0 0 39 0 0 0 0 0 Dryopteris acutidens 0 0 30 0 0 0 0 0 Differential species of B2 Pelea c1usiifolia 0 0 22 78 0 0 0 0 Asplenium normale* 0 0 4 56 8 0 0 0 Peperomia membranacea 0 0 4 56 8 0 0 0 Differential species of B3 Sadieria cyatheoides 0 0 0 33 100 0 0 0 Metrosideros polymorpha var. polymorpha 0 0 0 0 83 13 0 0 Oreobolusfurcatus 0 0 0 0 42 0 0 0 Polystichum bonseyi 0 0 0 0 33 0 0 0 Differential species of A2 and/or BI Cibotium chamissoi 38 100 91 0 0 0 0 0 Mecodium recurvum 75 90 65 0 0 0 0 0 Adenophorus tamariscinus 88 90 61 0 0 0 0 0 Grammitis tenella 88 79 48 0 0 0 0 0 Cibotium glaucum 75 100 96 0 0 0 0 0 Dicranopteris Iinearis* 100 79 61 0 0 0 0 0 Lycopodium cernuum* 100 68 22 0 0 0 0 0 Dryopteris tetrapinnata 0 5 22 0 0 0 0 0 Liparis hawaiensis 0 16 26 0 0 0 0 0 Dryopteris nuda 0 16 35 0 0 0 0 0 Pelea haleakalae 0 84 87 0 0 0 0 0 Sphaerocionium lanceolatum 0 84 70 0 0 0 0 0 Clermontia arborescens waihiae 0 84 65 0 0 0 0 0 Elaphoglossum alalum 0 74 44 0 0 0 0 0 Alyxia oliviformis 0 58 57 0 0 0 0 0 Pelea orbicularis 0 47 52 0 8 0 0 0 Juncus planifolius** 0 47 22 0 0 0 0 0 Stenogyne rotundifolia 0 21 22 11 8 0 0 0 Wikstroemia monticola 0 37 30 0 0 0 0 0 Vandenboschia cyrtotheca 0 21 9 0 0 0 0 0 Machaerina angustifolia* 0 26 4 0 0 0 0 0 Hedyotis (Gouldia) hillebrandii 0 21 9 II 0 0 0 0 Coprosma pubens 0 32 9 0 0 0 0 0 Differential species of CI Sophora chrysophylla 0 0 0 0 0 94 100 0 Coprosma montana 0 0 0 0 0 81 100 0 Geranium cuneatum tridens 0 0 0 0 0 88 62 0 Carex wahuensis 0 0 0 0 0 63 39 0 Vegetation of Windward Haleakala-KITAYAMA AND MUELLER-DoMBOIS 219

PLANT COMMUNITY CODE Al A2 81 82 83 Cia Clb C2 NUMBER OF RELEvEs 8 19 23 9 12 16 13 11

Differential species of C2 Tetramolopium humi/e haleakalae 0 0 0 0 0 0 0 100 Argyroxiphium sandwicensis macrocephalum 0 0 0 0 0 0 0 64 Differential species of C Ib Trisetum glomeratum 0 0 0 0 0 0 92 82 Rumex acetosella** 0 0 0 II 33 38 92 0 Pellaea ternifolia* 0 0 0 0 0 0 62 9 Asplenium adiantum-nigrum* 0 0 0 0 0 0 46 36 Asplenium trichomanes* 0 0 0 0 0 0 31 46 Dodonaea viscosa* 0 0 0 0 0 0 23 0 Differential species ofCia and/or 83 Epi/obium billardierianum cinereum** 0 0 0 0 17 50 0 0 Prunella vulgaris** 0 0 0 0 58 69 0 0 Coprosma ernodeoides 0 0 0 0 58 50 0 9 Hypochoeris radicata** 0 0 0 0 83 100 100 91 Deschampsia nubigena 0 0 9 0 67 81 100 82 Vaccinium reticulatum 0 0 0 0 83 100 92 18 Anthoxanthum odoratum** 0 0 4 22 83 100 62 0 Pteridium aquilinum var. decompositum* 0 0 0 0 42 100 100 0 Holcus lanatus** 0 0 4 0 58 100 54 0 Luzula hawaiiensis 0 0 0 0 33 44 85 18 Lycopodium venustulum 0 11 4 0 33 6 0 0 Unclassified species Styphelia tameiameiae* 0 0 100 89 92 100 100 46 Peperomia eekana 0 5 26 22 0 0 0 0 Erechtites valerianifolia** 13 32 4 0 0 0 0 0 Acacia koa 13 0 0 33 25 0 0 0 Lapsana communis** 0 0 0 11 25 0 23 0 Stenogyne kamehamehae 0 II 13 11 0 0 0 0 Dryopteris unidentata 0 11 4 22 0 0 0 0 Ageratina adenophora** 13 16 4 0 0 0 0 0 Pteris excelsa* 0 0 4 22 17 0 0 0 Sphaerocionium obtusum 13 21 0 0 0 0 0 0 Asplenium acuminatum 0 0 13 II 0 0 0 0 Commelina diffusa** 0 16 4 0 0 0 0 0 Coniogramme pi/osa 0 0 13 II 0 0 0 0 Marattia douglasii 0 0 17 0 0 0 0 0 Nephrolepis multiflora** 38 5 0 0 0 0 0 0 Cyanea kunthiana 0 0 13 0 0 0 0 0 Digitaria sp.** 0 16 0 0 0 0 0 0 Elaphoglossum pellucidum 0 16 0 0 0 0 0 0 Huperzia serratum* 0 II 4 0 0 0 0 0 Gahnia gahniiformis* 0 0 0 0 0 13 8 0 Paspalum urvillei** 13 11 0 0 0 0 0 0 Scaevola chamissoniana 0 16 0 0 0 0 0 0 Cerastiumfontanum triviale** 0 0 0 0 0 6 8 0 Clermontia kakeana 0 II 0 0 0 0 0 0 Dactylis glomerata** 0 0 0 0 0 13 0 0 Dryopteris fusco-atra 0 0 9 0 0 0 0 0 Dubautia reticulata 0 0 0 0 17 0 0 0 Fragaria chi/oensis sandwicensis 0 0 0 0 8 6 0 0 Lythrum maritimum* 0 0 0 0 17 0 0 0 Melaleuca quinquenervia** 25 0 0 0 0 0 0 0 Ophioglossum pendulum* 13 5 0 0 0 0 0 0 Pinus radiata** 0 0 0 0 0 13 0 0 Pittosporum glabrum 0 5 4 0 0 0 0 0 220 PACIFIC SCIENCE, Volume 46, April 1992

PLANT COMMUNITY CODE Al A2 BI B2 B3 Cia Clb C2 NUMBER OF RELEvEs 8 19 23 9 12 16 13 11

Plantago lanceolata** 0 0 0 0 0 6 8 0 Psidium guajava** 13 5 0 0 0 0 0 0 Pteris cretica* 0 0 0 0 17 0 0 0 Setaria gracilis** 13 5 0 0 0 0 0 0 Silene struthioloides 0 0 0 0 0 0 15 0 Agrostis sandwicensis 0 0 0 0 0 0 8 0 Aspleniumfragile* 0 0 0 II 0 0 0 0 Carex sp.** 0 5 0 0 0 0 0 0 Clermontia grandiflora grandiflora 0 0 4 0 0 0 0 0 Ctenitis latifrons 0 0 0 II 0 0 0 0 Cuphea carthagenensis** 13 0 0 0 0 0 0 0 Cyanea sp. 0 0 4 0 0 0 0 0 Kyllinga brevifolia** 13 0 0 0 0 0 0 0 Dryopteris insularis* 0 0 4 0 0 0 0 0 Dryopteris hawaiiensis 0 0 0 11 0 0 0 0 Conyza bonariensis** 0 0 0 0 0 0 0 9 Eucalyptus sp. ** 13 0 0 0 0 0 0 0 Juncus effusus** 0 5 0 0 0 0 0 0 Ludwigia octovalvis** 13 0 0 0 0 0 0 0 Paspalum scrobiculatum** 13 0 0 0 0 0 0 0 Peperomia cookiana 0 0 4 0 0 0 0 0 Perrottetia sandwicensis 0 5 0 0 0 0 0 0 Phymatosorus scolopendria** 0 5 0 0 0 0 0 0 Pittosporum confertiflorum 0 0 0 0 8 0 0 0 Pritchardia arecina 0 5 0 0 0 0 0 0 Rubus macraei 0 0 0 0 8 0 0 0 Sadleria squarrosa 0 0 4 0 0 0 0 0 Trifolium repens** 0 0 0 0 0 6 0 0 Unidentified 0 5 0 0 0 0 0 0 Unidentified 0 0 0 0 0 0 8 0