- Composition and ecology of vascular epiphyte communities in Mexico - 487

Journal of Vegetation Science 6: 487-498, 1995© IAVS; Opulus Press Uppsala. Printed in Sweden

Composition and ecology of vascular epiphyte communitiesalong an altitudinal gradient in central Veracruz, Mexico

Hietz, Peter1* & Hietz-Seifert, Ursula2

Instituto de Ecología, Aptdo 63, 91000 Xalapa, Veracruz, Mexico;1*Author for correspondence: present address: Botanisches Institut, Universität für Bodenkultur, Gregor Mendel-

Str. 33, A-1180 Wien, Austria; Fax +43 1 3195670; E-mail HIETZ@EDV1.BOKU.AC.AT2Present address: Institut für Pflanzenphysiologie, Universität Wien, Althanstr. 14, A-1091 Wien, Austria

Abstract. Vascular epiphytes were studied in forests at alti-tudes from 720 to 2370 m on the Atlantic slope of centralVeracruz, Mexico. The biomass of all trees of each species >10 cm diameter at breast height within plots between 625 and1500 m2 was estimated. The number of species per plot rangedbetween 22 and 53, and biomass between 9 and 249 g dryweight/m2. The highest values, both of species and biomass,were found at an intermediate altitude (1430 m). Habitatdiversity may contribute to epiphyte diversity in humid for-ests, but the importance of this effect could not be distin-guished from the influence of climate. A remarkably highnumber of bromeliads and orchids grew in relatively dryforests at low altitudes. In wet upper montane forests,bromeliads were replaced by ferns, while orchids were numer-ous at all sites, except for a pine forest. The number ofepiphytic species and their biomass on a tree of a given sitewere closely related to tree size. According to CanonicalCorrespondence Analysis, the factor determining the compo-sition of the epiphytic vegetation of a tree was altitude and tosome extent tree size, whereas tree species had practically noinfluence. The only trees which had an evidently negativeeffect on epiphytes were pines, which were particularly hostileto orchids and to a lesser degree to ferns, and Bursera simaruba,which generally had few epiphytes due to its smooth anddefoliating bark.

Keywords: Altitudinal transect; Biomass; Canopy; Diversity;Epiphyte; Host specifity.

Nomenclature: Sosa & Gómez-Pompa (1994).

Introduction

The canopy of tropical and subtropical forests hasreceived much attention within the last two decades(Perry 1984) and has, not without justification, beencalled “the last great unexplored frontier of life onEarth” (Myers 1984). As canopy access techniques weredeveloped and the enthusiasm of biologists increased,we began to understand the importance of the canopyfor the functioning of the forest, and for biodiversity.

Compared to the amount of information available forground-rooted vegetation, little is known about the com-position of epiphytic communities and the influence ofenvironmental variables on the vegetation of the canopy.No detailed studies have been published on the epi-phytic vegetation of Mexico. Moreover, most studiespublished so far are descriptive rather than analytical.Numerical methods for classification and ordination,now a standard procedure in vegetation analysis, havesuccessfully been used for epiphytic cryptogams (Kenkel& Bradfield 1981; Kantvilas & Minchin 1989; Wolf1993), but, with few exceptions (van Leerdam et al.1990), almost never for vascular epiphytes.

A number of studies carried out in tropical moun-tains (Gradstein & Frahm 1987; Frahm 1990; Wolf1993) have shown a strong stratification of cryptogamicepiphytes along altitudinal gradients. On the other hand,Sugden & Robins’ (1979) paper on epiphytes in theSierra de Santa Marta and the Serranía de Macuira,Colombia, is, to our knowledge, the only detailed ac-count of the distribution of vascular epiphytes. Theirstudy suffers to some extent from the rather small plotsizes (98 and 100 m2) and the fact that many speciescould not be identified.

For one of the sites described here, it has been shownthat there are at least two gradients which influence theepiphytic vegetation on the host trees (Hietz & Hietz-Seifert in press). One is the microclimatic gradient fromthe humid and shaded stem base to the drier and lighterouter twigs, and the other one along branches of varyingdiameters.

This study describes the differences amongst epi-phytic communities for trees at the same location andfor different locations. The importance of climatic vari-ables and variables associated with the host trees for thedistribution of species, species groups, diversity, andbiomass, is examined. In contrast to most other studies,we recorded epiphyte biomass, which is a more accuratemeasure of the importance of a species than are indi-vidual numbers or frequency.

488 Hietz, P. & Hietz-Seifert, U.

Table 1. Characterization of the six forests. According to Rzedowski’s (1986) classification, sites 1 and 2 were studied along analtitudinal transect in central Veracruz. Average temperature calculated with the regression equation from Fig. 1; annual precipitationestimated from the nearest weather station or extrapolated between the two or three closest stations. Sites 1 and 2: mixture of dry oakforest (encinar) with tropical dry forest (selva baja); site 3: pine forest (pinar); sites 4 and 5: mesophilous montane forest (bosquemesófilo de montaña); site 6: transition between mesophilous montane forest and montane pine-oak forest (bosque de pino-encino).

Site Altitude Average temperature Plot size (m2); Stem density/ha; Dominant tree species; Site descriptionAnnual precipitation Inclination (°); Basal area (m2/ha); No. of tree species/plot

Mist frequency Exposition Canopy height1;Average tree height (m)

1 720 21.9 1500 407 Quercus polymorpha Light, open forest with1200 ~20 21.1 Bursera simaruba dense undergrowth on dry

± absent SW 13.2 Brahea sp. shallow, calcareous soil9.0 ~15

2 1000 20.3 676 725 Quercus polymorpha Forest still dry and light in1300 32 34.0 Persea liebmannii appearance; on calcareous,

very rare ~N 15.9 Wimmeria concolor but deeper soil than site 111.9 ~14

3 1370 18.2 900 422 Pinus pseudostrobus Open forest on lava flow,1400 < 5 29.3 Quercus sartorii shallow soil; only one tree,

occasional NE 22.0 2 no shrub layer, Quercus17.8 only two small individuals

4 1430 17.9 625 560 Quercus salicifolia Dense forest on calcareous~1600 35 75.7 Ostrya virginiana soil; shrub layer not dense

frequent W 20.3 913.6

5 1980 14.8 625 560 Quercus salicifolia Dense forest on deep, humid,1850 37 61.5 Fagus grandifolia volcanic soil; shrub layer not

very high N 18.5 Carpinus caroliniana dense11.4 14

6 2370 12.8 900 378 Quercus laurina Forest open due to two large1500 30 59.4 Pinus patula rocks; shrub layer scarce;

very high SW 27.32 Alnus acuminata deep, humid volcanic soil;16.6 7 somewhat exposed on a ridge

1 Calculated as average height of 20 % of highest trees;2 21.8 excluding Pinus.

Study sites and Methods

The study sites are located in central Veracruz,Mexico, close to the state capital Xalapa, along analtitudinal gradient ranging from 720 to 2370 m a.s.l.(Table 1). The exact location of the sites is not revealedin order to protect plants from collectors, but is availableupon request to scientists with specific interests. In thisarea the north-south oriented Sierra Madre Oriental,which borders the central Mexican highlands, meets aneast-west oriented chain of tertiary volcanoes (Rzedowski1986), with Xalapa lying at the foothills of the 4250 mhigh Cofre de Perote. The humid air from the warmMexican Gulf rises along the Atlantic slopes and causeshigh levels of precipitation and humidity and frequentmists. While temperature shows a strong negative linearcorrelation with altitude, precipitation is more influ-enced by local climatic conditions and topography, butgenerally is highest around 2000 m (Fig. 1). Monthlymeans of the coldest (December or January) and thewarmest month (May or June) differ by ca. 6 - 7 °C.75 - 80 % of rain falls in the wet season between Juneand October.

Today, most of the original forest has been clearedor replaced by young secondary forest. Old-growth for-ests remain only on steep slopes, on soils too rocky foragriculture, and in a few poorly accessible locations.

The six sites selected were old-growth forests with-out signs of recent human disturbance, except for site 6,where some big trees were probably felled a few decadesago, and which contained two large rocks, and site 3,where goats occasionally graze. Sites 1 and 2 are situatedon rather dry and shallow soils. Consequently, they areof lower stature and have a lighter canopy than forests onsoils commonly found at this altitude, but these havebeen replaced by fields. Site 3, an almost monospecificstand of Pinus pseudostrobus, is exceptional in beingsituated on a rocky lava flow with a very shallow soil. Atall other sites, oaks were dominant trees. See Table 1 fordetails and Fig. 2 for some photographs.

Plot size was at least 625 m2, but larger if the spe-cies-area curve (Hopkins 1957) suggested so. For thepurpose of this paper some variation in plot size is notimportant; for comparison, species numbers are alsogiven for 625-m2 subplots in each case (Table 2).

All vascular epiphytes were recorded on all trees

- Composition and ecology of vascular epiphyte communities in Mexico - 489

Table 2. Species number, biomass, niche distinctiveness (see text), and diversity indices of the six epiphyte communities studied.

Site 1 2 3 4 5 6

Altitude 720 1000 1370 1430 1980 2370Species per plot 42 40 22 53 39 23Species per 625 m2 subplot 37 40 21 53 39 23Estimated epiphytic biomass (g dry weight /m2) 30 97 9 249 32 16Niche distinctiveness 0.208 0.119 0.176 0.232 0.402 0.250Shannon-Index H' 2.70 1.11 1.99 1.95 1.95 1.84Simpson-Index D 0.102 0.549 0.178 0.226 0.218 0.228

Fig. 1. Relationships betweenaltitude, temperature (circles)and precipitation (triangles) inthe study area of centralVeracruz. All climatic stationsare located on the east (Atlan-tic) slope, the highlands aremuch drier. The inset showsthe altitudinal profile along19° 30' N just south of Xalapaand north of the Cofre dePerote summit. Climatic dataare from Soto & García(1989).

with a diameter at breast height (DBH) > 10 cm. Treenumber per plot ranged from 34 to 61. Each tree wasdivided into six zones: (1) branches < 5 cm diameter; (2)branches > 5 and < 20 cm diameter; (3) branches > 20 cmdiameter; (4) upper stem; (5) lower stem; and (6) stembase to 1 m height. No tree at any site had buttresses.Most trees were climbed and observations were madeby one person in the tree and one person on the groundusing binoculars. Vegetative units (individuals, leaves,or shoots) of epiphytes were counted and (for cacti)shoot length measured. A number of vegetative unitsper species were sampled, dried, and weighed, and totaldry matter of each species within the plots was calcu-lated per m2 ground area. Only green parts of vascularepiphytes were counted, partly because the quantifica-tion of wood and roots was difficult, but also becausegreen parts, being photosynthetically active, are mostimportant for ecosystem studies, since their surface,turnover rates, and biochemical activity are greatest.

Consequently, total epiphytic biomass was underesti-mated for sites with abundant ferns (site 5), whoserhizomes may account for a large part of their biomass,and for sites 4, 5 and 6 with abundant woody hemi-epiphytes. Cryptogams were not included in this study;they contributed considerably to the total epiphyticbiomass at site 5 only. Dead organic matter, especiallytree leaves trapped among epiphytes, was conspicuousat site 2, but was not at any site more than a fraction ofthe total epiphytic biomass.

Most individuals, also juveniles, could be identifiedin the field. Only immature individuals of some bro-meliads were often difficult to distinguish; these wereassigned to groups (e.g. Tillandsia, broad-leaved). Meth-ods are explained in more detail in Hietz & Hietz-Seifert(subm.). Voucher specimens have been deposited at theherbarium of the Instituto de Ecología at Xalapa (XAL),with some duplicates, especially of bromeliads, at theInstitute of Botany, University of Vienna (WU).

490 Hietz, P. & Hietz-Seifert, U.

Fig. 2a. Site 1. Rather open forest with many orchids (flowering Encycliaradiata) and narrow-leaved bromeliads (Tillandsia juncea).

Bursera simaruba, present only at site 2) to 4 for veryrough with abundant deep fissures. The influence ofbark roughness on the number of species per tree and theestimated epiphytic biomass was tested using analysisof covariance (ANCOVA, Zar 1984), with the numberof species and log biomass, respectively, as dependentvariables, bark roughness class as factor, and log BA ascovariable. ANCOVA assumptions of normal distribu-tion and homogeneity of variances were not met for site2. Although ANCOVA is considered rather robust againstviolations of these constraints (Zar 1984), we addition-ally compared slopes of regressions for bark roughness1 (Bursera) and 4 (Quercus) for site 2.

Habitat diversity for epiphytes may be calculated asthe distinctiveness of epiphyte assemblages in different

Fig. 2b. Site 2. Similar to site 1, but the epiphyticcommunity is strongly dominated by the twobromeliads Tillandsia juncea and T. fasciculata.

Fig. 2c. Site 3. Pine forest of very homogeneous structure on a rocky lavaflow. Only bromeliads are common.

Fig. 2a-c. Representative aspects ofthe epiphytic communities in the for-ests studied: Sites 1 - 3.

Data analysis

The positive correlation between host size and thenumber of epiphytic species is most obvious (Fig 3).Whereas the relation between DBH or basal area (BA)and the number of species per tree was non-linear, logBA and species number were linearly related. Hence,linear regressions between log BA and the number ofepiphytic species per tree were calculated. Slopes werecompared by multiple linear regression (Zar 1984) totest whether trees at various sites and altitudes werehosts to a significantly different number of species,eliminating the fact that the trees were of different size.

Four categories of bark roughness were distinguished,ranging from 1 for very smooth and defoliating (only

- Composition and ecology of vascular epiphyte communities in Mexico - 491

altitude by treating it as a covariable to test whether anyof the variables associated with individual trees were ofimportance for the composition of the epiphyte commu-nity. All parameters were default values of CANOCO(ter Braak 1987).

Two commonly used diversity indices were calcu-lated (Krebs 1989). 1. Shannon-Index:

′ = ∗∑H p pi iln (1)

where pi is the proportion of a species compared to thetotal biomass, is a measure of evenness and declines iffew species are dominating. 2. Simpson-Index:

D pi= ∑ 2 (2)

which in itself is a measure of dominance.

zones on a tree. Colwell & Futuyma’s (1971) ‘collectiveheterogeneity of resource states’, calculated with thespecies biomass values of the six zones distinguished,was used as a measure of habitat diversity at each site.Theoretically, the value ranges between 1 (no resourceshared by two species) and 0 (all species taking the sameshare of the same resources).

The epiphytic communities were analysed with Ca-nonical Correspondence Analysis (CCA, ter Braak 1987)using 10log biomass values as species scores and treatingeach tree as a site or sample. Environmental variablestested were: altitude, DBH, the number of epiphytespecies, 10log of epiphyte biomass, and bark roughness.The first run resulted in altitude being the single domi-nant variable. We consequently eliminated the effect of

Fig. 2d. Site 4. Very rich in epiphytic species with epiphytic biomassdominated by a number of broad-leaved and narrow-leaved bromeliads.

Fig. 2e. Site 5. At high altitudes, Elaphoglossum glaucum and otherferns are abundant. The only frequent bromeliad is Tillandsia imperialis.

Fig. 2f. Site 6. Although ferns dominate, orchids(flowering Encyclia vitellina) may form large clus-ters on thick branches.

Fig. 2d-f. Representative aspects of theepiphytic communities in the six forestsstudied: Sites 4 - 6.

492 Hietz, P. & Hietz-Seifert, U.

Results

Floristic diversity and distributionAt the six sites between 720 and 2370 m a.s.l., 134

species of vascular epiphytes were found. The speciesnumber per site decreased from 42 at site 1 at 720 m to23 at site 6 at 2370 m (Table 2). The pine forest at 1370m, with 22 species, was much poorer than the broad-leaved forest at a comparable altitude, with 53 species.

Two days of intensive searching in an area of severalha around each plot increased the number of speciesfound by no more than 15 - 20 % and consequently theplot size is considered adequate for a description of theepiphyte communities in the forests studied. Also, spe-cies numbers and composition of the plots reported wereconsistent with what we found during many field trips tolocations of comparable altitude in the area.

Although no single forest plot below 700 m or above2400 m was studied in detail, frequent field trips and arevision of the XAL herbarium gave a fairly accuratepicture of species richness at these altitudes. 17 speciesof vascular epiphytes were recorded above 2700 m andnine species above 3000 m. In forests at altitudes be-

tween 300 and 500 m we usually found 10 - 15 species,considerably less than at site 1.

There were many orchids in most forests, contribut-ing between 19 and 45 % to the species richness (Fig. 4).However, only one orchid species was found at site 3.The number of bromeliads was high at sites 1 to 4(between 9 and 17 species) but decreased significantlyat higher altitudes. Ferns, on the contrary, increasedwith altitude from two species at site 1 to 22 species,belonging to seven families, at site 5. Cacti were richestin species in the drier forests at low elevations (sites 1and 2) and Peperomia at intermediate altitudes (site 4).Araceae, Araliaceae, Clusiaceae, Commelinaceae, Cras-sulaceae and Solanaceae were present with no morethan two species at any site, except for Araceae withfour species at site 4. Woody hemi-epiphytic trees orshrubs of the families Araliaceae, Clusiaceae and Solana-ceae were common at sites 4 to 6, but rare at loweraltitudes. Hemi-epiphytic figs, abundant and diverse inthe lowland rainforest of southern Veracruz (Ibarra &Sinaca 1987), are very rare in the montane forests of thearea and were not present in any of our plots.

Table 3 shows the number of species common to two

Fig. 3. Relationships between DBHand the number of epiphytic speciesper host tree. The most frequent hostspecies or genera are distinguished bysymbols. Lines were fitted by linearregression with log BA as independ-ent variable, but for greater clearness,x-axes of the graphs are DBH insteadof log BA. The two lines for site 1correspond to Quercus (bark rough-ness 4) and Bursera (bark roughness1).

- Composition and ecology of vascular epiphyte communities in Mexico - 493

biomass (Fig. 4). In the forests at higher elevations (sites5 and 6), biomass was dominated by ferns. Orchids,although present with many species, never contributedmore than 13 % (at site 2) to the total epiphytic biomass.Primary hemi-epiphytic trees and shrubs, which germi-nate on their host-trees and later establish contact withthe soil by aerial roots, represented about 5 % of theepiphytic biomass at sites 4 and 6, and 21 % at site 5(mainly due to one large individual of Oreopanaxliebmannii). Secondary hemi-epiphytes start as climb-ers and may later loose contact with the soil - althoughnone within our plots had reached that stage. Of this lifeform only Syngonium neglectum at site 1 was promi-nent, accounting for 3 % of the total epiphytic biomass.However, note that only leaves were counted and that ifwood were included, hemi-epiphytic trees at sites 4 to 6would probably outweigh all other epiphytes.

Sites 3 to 6 at intermediate and high altitudes hadsimilar values for evenness (H' ranging from 1.84 to1.99; Table 2) and dominance (D between 0.178 and0.228). Site 1 had the highest evenness and lowest domi-nance measures, whereas the opposite was true for thefloristically very similar site 2. In that forest twobromeliads, Tillandsia fasciculata and T. juncea, domi-nate the epiphytic vegetation and together they accountfor 88 % of the total epiphytic biomass.

Host specificityAt all sites the number of epiphytic species per tree

was positively correlated with tree size, and the correla-tion was generally stronger at sites with larger speciesnumbers (Fig. 3). Species numbers per tree differedamong sites, as tested by multiple linear regression. Site4 not only had the largest total species number, but alsothe largest number of epiphytes on a tree of comparablesize, with two oaks with a DBH of ca. 75 cm carrying 32species of vascular epiphytes each. Sites 1 and 2 hadabout the same total species number per plot as site 5,but trees of the same size carried a significantly largernumber of species at sites 1 and 2. The forest at thehighest elevation (site 6) and the pine forest (site 3) hadthe lowest number of species per tree. All differencesmentioned are significant at p < 0.001.

A significant (ANCOVA, p < 0.01) effect of barkroughness on the number of species and epiphyte bio-mass could only be shown for site 2, which was the onlysite with Bursera simaruba, the tree with the smoothestbark. Also, comparing the slopes of the regression be-tween log BA and the number of epiphytic species forbark roughness 1 (Bursera) and 4 (Quercus) resulted insignificant differences (p < 0.05). Since the forest at site2 was mainly composed of oaks with very rough barkand Bursera with very smooth bark, the comparisonbetween these two extremes resulted in a significant

sites and Sørensen’s index (SI, Krebs 1989) of similar-ity. SI ranges between 0.02 and 0.34, and similarity wasgenerally greatest between adjacent altitudes. The larg-est difference was found between sites 4 and 5, whichalso show the largest altitudinal difference. Only onespecies, the fern Phlebodium areolatum, occurred at allsites, and 75, or more than half of the species, occurredat only one site (see App. 1).

Habitat diversityBased on habitat diversity, the six zones from the

stem base to branches < 5 cm diameter recorded in eachforest differed most at site 5 (Table 2). Niche distinc-tiveness of all other sites was considerably lower, withlowest values at sites 2 (0.119) and 3 (0.176). Appar-ently there were no species adapted to the stem base inthese forests with open canopies and much light reach-ing the soil. Only one small plant was found on the stembase of one tree at site 2, whereas the stem bases of treesin the other forests with open canopies (sites 1 and 3),although without a distinct epiphyte community, wereby no means devoid of epiphytes.

BiomassEpiphyte biomass was estimated to be 250 and 100 g

dry matter/m2 at sites 4 and 2, respectively, but not morethan 35 g/m2 at the other sites (Table 2). In forests of lowand intermediate altitudes, bromeliads dominated theepiphytic community, accounting for between 70 %(site 1) and over 99 % (site 3) of the total epiphytic

Fig. 4. Number of epiphytic species per plot (upper graph) andestimated epiphytic dry matter per m2 of the six sites studied.

494 Hietz, P. & Hietz-Seifert, U.

grow at average annual temperatures below 10 °C andendure regular frosts. Forests at low altitudes receiveless rain. As mist hardly ever occurs and high tempera-tures cause a high evaporative demand, only speciesable to cope with a very tight water budget can survive.Along the transect from warm and dry to cool and humidforests, the combination of temperature and water avail-ability is an important factor determining the diversityand abundance of epiphytes. It seems to be optimal atmid-altitudes around 1500 m.

Gentry & Dodson (1987) suggest that humid forestsmay achieve a finer niche partitioning and thus a higherdiversity because a more constant environment mayfavour within-community microhabitat specializationby epiphytes. Also, as humid forests are denser, com-posed of more layers of trees and shrubs, they have morepronounced microclimatic gradients than dry forests oflow stature and with open canopies. Therefore, theremay not only be a finer niche partitioning and speciali-zation because of the constant environment, but widergradients of microclimates and substrates may also re-sult in an absolutely larger number of niches and thuscontribute to species richness in humid forests. Gentry& Dodson (1987) found 41 species of understory spe-cialists among epiphytes in a very humid forest at RioPalenque, Ecuador, but none in a semi-deciduous moistforest at Jauneche. These findings are corroborated byresults from our study. The forest with the highesthabitat diversity (site 5), which was probably the mosthumid forest in the transect, had a very distinct commu-nity of epiphytes restricted to the lower stem and stembase (Hietz & Hietz-Seifert in press). Generally, habitatdiversity was higher in the more humid forests with adense canopy cover than in dry forests with open cano-pies. This may be very important for species richness inthe forests studied, but it is not possible to separate thiseffect from the influence of climate.

Altitude is a measure which is easy to obtain, butdifficult to interpret as it stands for a complex combina-tion of climatic variables to which species may respond.From studies of a uni-dimensional gradient it is notpossible to evaluate the influence of each of these vari-ables. Consequently, for most plants found on the warmand dry but not on the cool and wet side of the transect,it is not clear whether the low temperature, the highhumidity, or other possible factors limit their distribu-tion. Only some species of Tillandsia with an atmos-pheric habit (atmospheric bromeliads have very narrowleaves and do not form water reservoirs) occur in warmand dry forests of the transect and in the cool and dryforests of the central highlands. These species appear tobe limited by too high humidity.

The species numbers reported here are considerablylower than those found in very humid forests of the

Table 3. Number of species common to two sites and theSørensen index of similarity.

Site 1 2 3 4 5 6

Species per site 42 40 22 53 39 23

2 210.34

3 11 100.26 0.24

4 13 18 140.22 0.28 0.27

5 1 2 3 70.02 0.05 0.09 0.13

6 1 1 2 4 140.03 0.03 0.08 0.10 0.31

difference. No significance was found for the other sites.The pine forest at site 3 had the lowest number of

epiphyte species per tree, but not all systematic groupswere poorly represented. Whereas only one single or-chid was found, this site had the largest number ofbromeliads (17, plus another three species found onpines outside the plot studied). Ferns were very sparsecompared to other sites, and no other epiphytes werepresent within our plot.

Ordination of single host trees by CCA showed thataltitude was the most important variable determiningthe composition of the epiphytic communities. Theeigenvalue of the first axis, which was highly correlatedwith altitude, was 0.844. When the effect of altitude waseliminated by treating it as a covariable, the resultingfirst axis had an eigenvalue of only 0.367, but was stillsignificant (p < 0.01) according to the Monte Carlopermutation test (ter Braak 1987), and was best corre-lated with the number of epiphytic species per tree.

Discussion

DiversityHumid montane forests in South and Central America

have been found to be among the richest in vascularepiphytes and are often richer than lowland forests atcomparable latitudes (Gentry & Dodson 1987).

In the study area, the largest species number andbiomass of vascular epiphytes were found at intermedi-ate altitudes, although at 2000 - 2400 m precipitation andespecially the frequency of mist (a most important watersource for epiphytes) are higher. However, low averagetemperatures at these altitudes apparently exclude manyspecies. Above 2300 m the number of species decreasesconsiderably, probably because of episodic sub-freez-ing temperatures. In the area of Xalapa, only 17 speciesof vascular epiphytes were found at altitudes above2700 m and 9 above 3000 m, where they must be able to

- Composition and ecology of vascular epiphyte communities in Mexico - 495

northern Andean region and Costa Rica. Bøgh (1992)reported 104 species of vascular epiphytes on 175 m2

and Gentry & Dodson (1987) 127 species on 1000 m2;Ingram & Nadkarni (1993) found 65 species on onesingle tree in Costa Rica. Our species numbers are wellwithin the range found in other lowland and montaneforests from Central and South America (Sugden &Robins 1979; Kelly 1985; ter Steege & Cornelissen1989). This is noteworthy, since our study area is situ-ated close to the northern limit of the neotropics, andreceives only moderate rainfall. Still more remarkable isthe large number of species in comparatively dry forestsat low altitudes. Mexico is not only an evolutionarycentre of cacti and other desert succulents, but also ofthe genus Tillandsia, and especially the subgenusTillandsia. This subgenus shows a strong tendency to-wards an atmospheric habit, low stomata/trichome ra-tios (Winkler 1986) and crassulacean acid metabolism(Medina 1974), adaptations useful for plants in environ-ments with scarce and insecure water availability. Simi-larly, the number of tree species found by Lott et al.(1989) in a dry forest in Jalisco, eastern Mexico, receiv-ing less than 800 mm of rain, was considerably higherthan that reported from dry forests elsewhere in Centraland South America. Like in the case of woody plants(Rzedowski 1986), the high diversity of epiphytesadapted to drought in Mexico seems to be the result ofthe extensive and isolated dry and semi-dry forests inMexico which caused a radiation of these groups.

Our own collections and a revision of the herbariumproduced some 200 species of flowering epiphytes froman area of ca. 4000 km2, ranging from sea level to over4000 m. Compared to these figures, the 100 species(ferns excluded) found in six plots with a total of ca. 0.5ha is a large number. It suggests that several compara-tively small but carefully selected preserved areas maycontribute much to protect the biodiversity of an area.To protect a population, preserves would, of course,have to be much larger than the 625-m2 plots of thisstudy in which we often found only one or a few indi-viduals of a species. Data on the abundance of indi-vidual species within forests and their distribution in awider area as those presented here, provide importantinformation for the design of size and location of pre-serves. In addition, more information on the demogra-phy and growth of epiphytes, which is very scarce in theliterature, is necessary as survival and reproductiondetermine the fate of populations. Such data are cur-rently being obtained in the area of this study.

Species distributionAn analysis of site 5 showed that the position within

the canopy and the trees is of importance for the compo-sition of the epiphyte community. Here we describe the

variation of epiphytes among different trees in the sameforest, and among different sites at various altitudes.

Ordination of single trees by CCA showed that alti-tude was the most important factor determining thecomposition of the epiphyte communities. Most speciesshow a rather narrow altitudinal distribution and of thesites with a vertical distance of 250 m or more, nonewere found to share more than 50 % of their species.

In contrast to the very distinct epiphytic vegetationat different sites there was little variation among trees atone site. Generally, big trees carried more biomass andmore species, and some species seemed to be restrictedto big trees. A significant effect of bark roughness on thenumber of epiphytes could only be demonstrated forBursera simaruba, whose very smooth and defoliatingbark had a negative effect on colonization by epiphytes.Bursera was generally little colonized by all groups.

Pinus spp. are unsuitable hosts for some groups ofepiphytes but not for others, as almost no orchids, butabundant bromeliads were found in the pine forest at site3. Pine bark is fissured and would easily trap seeds oforchids as well as those of other species, but phenolic orresinous substances are probably unfavourable to theirgrowth. The chemical composition of the substrate mayaffect orchids more than other groups, as orchids de-pend on mycorrhizal fungi, and pine bark and needlesare slow to decompose. Bromeliad roots, on the otherhand, serve mainly as holdfasts and the bulk of nutrientsis absorbed by specialized trichomes on the leaves(Benzing 1970). They should, therefore, be more inde-pendent of substrate chemistry.

We found no evidence of any epiphyte displaying aspecific preference for a certain host species, althoughspecies represented by only one or few individuals andat one site, only might be found on a single tree species.If a tree was a suitable host, this was due first of all to itssize and to some extent to its bark roughness. Oaks,often the biggest trees with strongly structured bark, areparticularly good hosts. We could not confirm Frei &Dodson (1972), who found Q. peduncularis to be un-suitable for colonization by orchids because of phenolicbark substances. Among the 10 or more oak species westudied in these and other Mexican forests, including acoffee plantation with Q. peduncularis as shade trees,all were suitable hosts, and densely colonized, also byorchids. However, determining Mexican oaks (with atotal of some 250 species) is difficult, their systematicrelations are insufficiently known, and it cannot be ruledout that the oak we determined as Q. peduncularisbelonged to another species than Frei & Dodson’s, orwas at least genetically sufficiently remote to showdifferent levels of phenolic substances. Besides, theproduction of bark substances may be influenced byenvironmental factors.

496 Hietz, P. & Hietz-Seifert, U.

Host specificity of vascular epiphytes has been dis-cussed by several authors (Went 1940; Benzing 1990;Daniels & Lawton 1991), but unambiguous evidence isscarce. Within a site, the size of the host tree is the mostimportant factor determining the number of epiphytesfound on it. Therefore, studies evaluating the effect ofdifferent host tree species on epiphytes by comparingthe proportion of trees colonized, remain inconclusiveas long as tree size is not included in statistical tests. Ourresults show that within a forest, describing canopyheight strata or branch size classes can explain more ofthe epiphytes’ distribution than distinguishing singletrees or tree species.

If an epiphyte appears to show some preference for acertain host species, this preference is often shared bymost other species present, indicating that the generalsuitability of a tree for epiphyte colonization rather thana particular relation between two species is responsible,which is a considerable difference. A genuine hostspecificity of an epiphyte would include that one hostspecies is frequently colonized by one species and lessoften by others and that the supposed host-specificepiphyte is less abundant on other phorophytes in thearea, which are suitable for other epiphytes. There areprobably many examples of hosts which tend to be moreor less suitable for colonization because of their size,bark structure and chemistry, persistent leaf bases, orother features. These hosts are generally suitable orunsuitable either for all species of an area or for certaingroups like bromeliads, orchids, or stranglers. A genu-ine host specificity of single species appears to be rare.

Acknowledgements. Thanks are due to M. A. Soto Arenas,UNAM, Mexico, M. Palacios-Rios, Xalapa, and W. Till,Vienna, who helped to identify orchids, ferns, and bromeliads,respectively. L. Mucina, Vienna, G. Williams-Linera, Xalapa,and an anonymous reviewer gave useful comments on themanuscript, and H. Hurtl and U. Roy-Seifert helped with thelanguage. We are grateful to the Instituto de Ecología, Xalapa,and its staff for their hospitality and co-operation. U. H.-S. wassupported by an academic exchange program of the Mexicanforeign ministry.

References

Benzing, D.H. 1970. Foliar permeability and the absorption ofminerals and organic nitrogen by certain tank bromeliads.Bot. Gaz. 131: 23-31.

Benzing, D.H. 1990. Vascular epiphytes. Cambridge Univer-sity Press, Cambridge.

Bøgh, A. 1992. Composition and distribution of the vascularepiphyte flora of an Ecuadorian montane rain forest.Selbyana 13: 25-34.

Colwell, R.K. & Futuyma, D.J. 1971. On the measurement of

niche breadth and overlap. Ecology 52: 567-576.Daniels, J.D. & Lawton, R.O. 1991. Habitat and host prefer-

ences of Ficus crassiuscula, a neotropical strangling fig ofthe lower-montane rain forest. J. Ecol. 79: 129-141.

Frei, J.K. & Dodson, C.H. 1972. The chemical effect of certainbark substances on the germination and early growth ofepiphytic orchids. Bull. Torrey Bot. Club 99: 301-307.

Frahm, J.-P. 1990. The ecology of epiphytic bryophytes onMt. Kanabalu, Sabah (Malaysia). Nova Hedwigia 51: 121-132.

Gentry, A.H. & Dodson, C.H. 1987. Diversity and biogeogra-phy of neotropical vascular epiphytes. Ann. Mo. Bot.Gard. 74: 205-233.

Gradstein, S.R. & Frahm J.-P. 1987. Die floristische Höhen-gliederung der Moose entlang des BRYOTROP-Transektesin NO Peru. Beih. Nov. Hedw. 88: 105-113.

Hietz, P. & Hietz-Seifert, U. In press. Structure and ecology ofepiphyte communities of a cloud forest in central Veracruz,México. J. Veg. Sci.

Hopkins, B. 1957. The concept of minimal areas. Ecology 45:441-449.

Ibarra, G. & Sinaca, S. 1987. Listados florísticos de México.VII Estación de Biología Tropical Los Tuxtlas, Veracruz.UNAM, México.

Ingram, S.W. & Nadkarni, N.M. 1993. Composition and dis-tribution of epiphyte organic matter in a neotropical cloudforest, Costa Rica. Biotropica 25: 370-383.

Kantvilas, G. & Michin, P.R. 1989. An analysis of epiphyticlichen communities in Tasmanian cool temperate rainfor-est. Vegetatio 84: 99-112.

Kenkel, N.C. & Bradfield, G.E. 1981. Ordination of epiphyticbryophyte communities in a wet-temperate coniferousforest, South-Coastal British Columbia. Vegetatio 45: 147-154.

Kelly, D.L. 1985. Epiphytes and climbers of a Jamaican rainforest: vertical distribution, life forms and life histories. J.Biogeogr. 12: 223-241.

Krebs, C.J. 1989. Ecological methodology. Harper & Row,New York, NY.

Lott, E.J., Bullock, S.H. & Solís-Magallanes, J.A. 1989. Flo-ristic diversity and structure of upland and arroyo forestsof coastal Jalisco. Biotropica 19: 28-235.

Medina, E. 1974. Dark CO2 fixation, habitat preference andevolution within the Bromeliaceae. Evolution 28: 677-686.

Myers N. 1984. The primary source. W.W. Norton, NewYork, NY.

Perry, D.R. 1984. The canopy of the tropical rain forest. Sci.Amer. 251: 138-147.

Rzedowski, J. 1986. Vegetación de México, 3rd ed. EditorialLimusa, Mexico, DF.

Sosa, V. & Gómez-Pompa, A. 1994. Flora de Veracruz,Listado de especies. Instituto de Ecología, A.C., Xalapa,México and University of Riverside, Riverside, CA.

Soto, M. & García, E. 1989. Atlas climatico del estado deVeracruz. Instituto de Ecología, A. C., Xalapa.

Sugden, A.M. & Robins, R.J. 1979. Aspects of the ecology ofvascular epiphytes in Colombian cloud forests, I. Thedistribution of the epiphyte flora. Biotropica 11: 173-188.

- Composition and ecology of vascular epiphyte communities in Mexico - 497

ter Steege, H. & Cornelissen, J.H.C. 1989. Distribution andecology of vascular epiphytes in lowland rain forest ofGuyana. Biotropica 21: 331-339.

ter Braak, C.J.F. 1987. CANOCO - a FORTRAN program forcanonical community ordination by [partial] [detrended][canonical] correspondence analysis, principal compo-nents analysis and redundancy analysis (version 2.1).TNO Institute of Applied Computer Science, Wageningen.

van Leerdam, A., Zagt, R.J. & Veneklaas, E.J. 1990. Thedistribution of epiphyte growth-forms in the canopy of aColombian cloud-forest. Vegetatio 87: 59-71.

Went, F.W. 1940. Soziologie der Epiphyten eines tropischenUrwaldes. Ann. Jard. Bot. Buitenz. 50: 1-98.

Winkler, S. 1986. Differenzierungen und deren Ursacheninnerhalb der Bromeliaceen. Beitr. Bid. Pflanzen 61: 283-314.

Wolf, J.A.D. 1993. Ecology of epiphytes and epiphyte commu-nities in montane rain forests, Colombia. Ph.D. Thesis,Univ. Amsterdam.

Zar, J.H. 1984. Biostatistical analysis. 2nd ed. Prentice-Hall,Englewood Cliffs, NJ.

Received 23 June 1994;Revision received 6 February 1995;

Accepted 2 March 1995.

Site Life form 1 2 3 4 5 6Number of trees per site

Pteridophyta 61 49 38 35 35 34

AspleniaceaeAsplenium cuspidatum H-co 5A. monanthes H-co 1 2

GrammitidaceaeGrammitis cf. prionodes H-co 1

HymenophyllaceaeHymenophyllum cf. crispum H-cr 1H. polyanthos H-cr 5H. thunbrigense H-cr 2Trichomanes sp. 01 H-cr 8T. reptans H-cr 21

LomariopsidaceaeElaphoglossum glaucum H-co 21E. petiolatum H-co 16

LycopodiaceaeLycopodium taxifolium H-pe 2L. dichotomum H-co 1

PolypodiaceaeCampyloneurum angustifolium H-co 2C. xalapense H-co 9Niphidium crassifolium H-co 4Pecluma spp. H-co 1 1 15Phlebodium areolatum H-cr 1 1 6 14 2 5Pleopeltis angustata H-cr 4P. crassinervata H-cr 1 22 5P. mexicana H-cr 20 27Polypodium arcanum H-cr 10 2P. eatonii H-cr 1P. fraternum H-cr 4 12P. furfuraceum H-co 27 13 2 14P. lepidotrichum H-cr 2 7P. montigenum H-cr 5P. plebeium H-cr 1 21P. plesiosorum H-cr 1 26P. polypodioides H-cr 6 1P. puberulum H-cr 19 14P. triseriale H-cr 4

Site Life form 1 2 3 4 5 6Number of trees per site

VittariaceaeAnthorphyum ensiforme H-co 1Vittaria graminifolia H-co 2

Spermatophyta

AraceaeAnthurium schlechtendalii H-co 3A. scandens H-pe 5 11Philodendron advenae He-cl 2Philodendron sp. He-cl 1Syngonium neglectum He-cl 25 1

AraliaceaeOreopanax capitatus He-tr 7O. flaccidus He-tr 2O. liebmannii He-tr 1 7 1

BromeliaceaeAechmea bracteata Ta 2Catopsis mooreniana Ta 1C. nutans Ta 6 3+ 21C. paniculata Ta 5C. sessiliflora Ta 3+ 22Tillandsia butzii At 5 7 27T. capitata Ta-at 1T. concolor Ta-at 2T. fasciculata Ta-at 19 28 19 3T. filifolia At 11T. foliosa Ta-at 6T. ghiesbreghtii Ta 10T. gymnobotrya Ta 2 1 2+T. heterophylla Ta 1+T. imperialis Ta 23T. ionantha At 43 4 4T. juncea At 22 32 28 17T. kirchhoffiana Ta-at 26T. limbata Ta-at 11 3T. lucida At 1+T. multicaulis Ta 2+ 23T. polystachya Ta-at 13T. pseudobailey At 2T. punctulata Ta-at 8 17T. recurvata At 9 2 3

Appendix 1. Occurrence of epiphytes in six forest sites near Xalapa, Mexico. Figures are number of trees occupied by species at eachsite. + indicates that probably more trees were occupied by this species but that juveniles could not be identified. Abbreviated lifeforms: H-cr = herbaceous long creeping; H-co = herbaceous compact or with very short-creeping rhizome forming dense stands; H-pe = herbaceous pendent or scandent; Sf = suffrutescent; He-tr = hemi-epiphytic tree; He-cl = hemi-epiphytic climber; Ta = tankforming rosette; At = atmospheric bromeliad; Ta-at = tank atmospheric intermediate with narrow leaves and small axillary tanks.

498 Hietz, P. & Hietz-Seifert, U.

App. 1, cont.

Site Life form 1 2 3 4 5 6Number of trees per site

T. schiedeana At 33 28 35 18T. streptophylla Ta-at 3 2T. tricolor Ta-at 1 2 2T. usneoides At 5 5 7 19T. violacea Ta 1+T. viridiflora Ta 29

CactaceaeEpiphyllum phyllanthus H-pe 7 7Hylocereus undatus H-pe 9Rhiposalis baccifera H-pe 12 5 2Selenicereus cf. coniflorus H-cr 1+S. testudo H-cr 4

ClusiaceaeClusia sp. He-tr 10

CommelinaceaeGibbasia cf.geniculata H-cr 5 1

CrassulaceaeEcheveria rosea Sf 12 2Sedum bottieri Sf 2S. dendroideum Sf 4

OrchidaceaeAcineta barkeri H-co 3Brassia verrucosa H-co 4Dichaea neglecta H-cr 9Encyclia cochleata H-co 6 4E. livida H-co 3E. ochracea H-co 2 3 2E. polybulbon H-cr 3 8E. radiata H-co 4E. vittelina H-co 1 3Epidendrum longipetalum H-co 2 3E. polyanthum H-co 3Gongora galeata H-co 1Isochilus unilaterale H-co 7 7 8Jacqueniella teretifolia H-co 12Laelia anceps H-co 8 9Lemboglossum ehrenbergii H-co 15 3Lepanthes avis H-co 11L. moorei H-co 2L. schiedei H-co 4Lycaste aromatica H-co 6Maxillaria densa H-pe 6 2M. meleagris H-co 1M. variabilis H-pe 1 4Nageliella purpurea H-co 8 6Nidema boothii H-co 1Notylia barkeri H-co 21 2Oncidium cebolleta H-co 6O. maculatum H-co 10 2O. stramineum H-co 17 1Ornithocephalus inflexus H-co 5Pleurothallis pubescens H-co 1P. schiedei H-co 1P. tribuloides H-co 1 2P. tubata H-co 4 2Restrepiella ophiocephala H-co 3Rhnycholaelia glauca H-co 6Scaphyglottis livida H-co 4 6Stanhopea oculata H-co 2Stelis sp. 01 H-co 4 1Stelis sp. 02 H-co 1Vanilla insignis He-cl 1

PiperaceaePeperomia sp. 01 H-co 1 10P. deppeana H-co 13 2P. galioides H-co 23 7P. glabella H-co 4 22P. aff. quadrifolia H-co 17P. pseudoalpina H-co 17P. quadrifolia H-co 29 16P. reflexa H-co 3 10

SolanaceaeJuanulloa mexicana He-tr 2Solandra maxima He-tr 3

Site Life form 1 2 3 4 5 6Number of trees per site