Biodiversity and Conservation 10: 1473–1495, 2001.© 2001 Kluwer Academic Publishers. Printed in the Netherlands.

Pteridophyte species richness in Andean forestsin Bolivia

MICHAEL KESSLERAlbrecht-von-Haller-Institut für Pflanzenwissenschaften, Abteilung Systematische Botanik,Untere Karspüle 2, D-37073 Göttingen, Germany (e-mail: 106606.464@compuserve.com;fax: +49-551-392329)

Received 18 May 2000; accepted in revised form 10 October 2000

Abstract. I assessed the magnitude and distribution of pteridophyte species richness on the eastern Andeanslope in Bolivia based on 676 study plots of 400 m2 each in forest habitats at 65 study sites. In total755 species were recorded, including 426 (56%) epiphytes and 598 (79%) terrestrials, with 266 species(35%) recorded under both groups. Mean number of species per plot at a given site varied from 0 to 31.1for epiphytes, 0 to 20.9 for terrestrials, and 0 to 47.9 for all species combined. The highest numbers ofspecies recorded at a given study site were 110 epiphytes, 101 terrestrials, and 167 species in total. Whileoverall there were more terrestrial than epiphytic species, at individual sites and plots the reverse was true,indicating that terrestrial species tended to be more patchily distributed than epiphytes. Despite high surveyintensity, many species went unrecorded; the minimum estimate of total species richness obtained throughextrapolation was 975 species overall, including 559 epiphytes and 880 terrestrials. A correlation analysisof species richness to 14 environmental parameters revealed a highly positive correlation to mean annualprecipitation and bryophyte cover on tree branches (a proxy for air humidity). Significant correlations toother parameters (e.g. human impact, canopy height, etc.) reflected the covariance of these factors withprecipitation and bryophyte cover. Despite a lack of data on the pteridophyte communities from much ofthe Bolivian Andes, it appears that in most of the countries, pteridophyte diversity can be protected byfocussing the most humid parts of the Andean forests.

Key words: diversity, ferns, humidity, relative species richness

Introduction

Most information on diversity and composition of tropical forests is based on studiesof woody plants (e.g. Gentry 1988, 1995), even though non-woody life forms typi-cally contribute at least half of the vascular plant species richness in these habitats(Whitmore et al. 1985; Gentry and Dodson 1987; Duivenvoorden 1994; Ibisch 1996;Galeano et al. 1998; Balslev et al. 1999). Pteridophytes, i.e. ferns and fern-allies, areone of the most abundant and diverse non-woody plant groups, including primari-ly terrestrial and epiphytic herbs, in addition to some hemiepiphytes and tree ferns.Because of their comparatively well-known taxonomy, their small size, their highspecies richness, the wide distribution of individual species, and because they repre-sent morphologically and ecophysiologically well-defined guilds (Barrington 1993),pteridophytes have been repeatedly used to study patterns of community composition

1474

and diversity of tropical forests (Young and León 1989; van der Werff 1992; Tuomistoand Ruokolainen 1994; Poulsen and Nielsen 1995; Tuomisto et al. 1995, 1998; Tuo-misto and Poulsen 1996; Øllgaard and Navarette 1997; Lwanga et al. 1998; Øllgaardet al. 1998). However, most of these studies have focussed on tropical lowland forestsand little is known about the magnitude and distribution of pteridophyte diversityin montane regions. Pteridophyte species richness shows a hump-shaped distributionin relation to elevation, with maxima between 500 m and 2000 m (Lellinger 1985;Jacobsen and Jacobsen 1989; Parris et al. 1992; Kessler et al. 1999), but it is notknown what determines this pattern or the elevational variability between the differentgeographical regions.

In the present study, I sampled the pteridophyte diversity at 65 forest sites inthe Bolivian Andes, covering an environmental range from perhumid evergreen todrought-deciduous forests and from 200 m to 4050 m elevation. The main aims wereto assess the overall species richness of pteridophytes in Bolivian Andean forestsand to determine how this richness is distributed in relation to environmental factorssuch as elevation, humidity, latitude, ecoclimatic variability, habitat heterogeneity,and human impact. While most of these factors have frequently been linked to speciesrichness (Huston 1994; Rosenzweig 1995), ecoclimatic variability, i.e. the short- andlong-term variability of climatic factors influencing plant development, has rarelybeen considered, even though diversity and composition of biotic communities arewell known to be influenced by habitat dynamics and disturbances such as droughtsor frost events (Phillips et al. 1994; Fjeldså et al. 1999b). I here use interannual vari-ability in biophysical attributes measured by remote sensing at monthly intervals overthe period 1983–1992 as a proxy for ecological variability at individual sites, usingthe data set employed by Fjeldså et al. (1999b) for a study of the relationship of eco-climatic variability and bird endemism. The used attributes were NDVI (NormalizedDifference Vegetation Index), which is highly correlated to vegetation parameterssuch as green-leaf biomass, absorbed photosynthetically active radiation, and biomassgreen vegetation (Holben et al. 1980; Tucker et al. 1985), and Ts (land surface ‘skin’brightness temperature), which is related, through the surface energy balance equa-tion, to surface moisture availability and evapotranspiration, as a function of latentheat flux (Planet 1988). This data set thus provides a unique opportunity to measureregional patterns of variations in surface conditions, as well as their climate-drivenfluctuations at the seasonal and inter-annual scales.

Methods

Study areas and field sampling

Field work was conducted at 65 sites (Figure 1, Table 1) for 1–12 days per siteduring six to eight months of the dry season (April/May to October/November) in

1475

Figure 1. Location of the study sites in the Bolivian Andes. Numbering refers to Table 1. Stippling indi-cates the distribution of drough-deciduous forests, with more humid areas stippled more densely. Dashedlines delimit the approximate lower (east) and upper (west) limits of humid montane forest.

1995, 1996, and 1997, with limited additional sampling in 1998. The study sites werechosen to include the widest possible range of environmental conditions betweenthem, while being ecologically homogeneous within. Thus, no site covered notice-ably different geological substrates or an elevational range of more than 500 m.Generally, plots at any given study site were no more than 10 km apart from eachother. Study sites in close geographical proximity to each other were considered tobe independent samples if separated by at least 500 m elevation or by at least 200 melevation plus different geological substrates and/or strikingly different humidity con-ditions. These differences generally caused a turnover in species composition by atleast 45%.

At each study site, pteridophytes were sampled in plots of 400 m2 each, mostlysquare in shape but occasionally in other shapes to minimize habitat heterogeneity.This size corresponds to the minimum area required for representative surveys in the

1476

Table 1. Study sites included in this analysis. Numbering refers to Figure 1. The number of plots refersonly to forest plots included in this analysis. Broadly physiognomically defined forest types are deciduous(D), semideciduous (SD), and evergreen (E). References (Ref.) include more detailed site descriptions aswell as species lists.

No. Study siteNo.plots

Foresttype Coordinates Elevation (m) Ref.

1 Rıo San Antonio 16 E 13◦35′ S, 68◦30′ W 4002 Rıo Machariapo 13 SD 14◦37′ S, 68◦27′ W 1050–1350 [1]3 Estacion Biologica del Beni 16 SD 14◦47′ S, 66◦19′ W 200 [2]4 Rıo Yuyo 3 E 15◦02′ S, 68◦28′ W 850–10005 Cerro Asunta Pata 11 E 15◦04′ S, 68◦28′ W 1200–15006 Rıo D’Artagnan 17 E 15◦07′ S, 67◦05′ W 400–5007 Serranıa Pilon Lajas 8 E 15◦09′ S, 67◦06′ W 700–950 [3]8 Chullina 6 E 15◦09′ S, 68◦55′ W 3300–34509 Charazani 11 E 15◦10′ S, 68◦55′ W 2150–2400

10 Rıo Camata 11 SD 15◦13′ S, 68◦43′ W 1150–135011 Camata 8 D 15◦14′ S, 68◦46′ W 1400–145012 Consata 19 D 15◦25′ S, 68◦32′ W 1000–140013 Sapecho 14 E 15◦32′ S, 67◦18′ W 450–750 [4]14 Tacacoma 2 D 15◦32′ S, 68◦40′ W 2000–240015 Quiabaya 2 E 15◦38′ S, 68◦35′ W 350016 Serranıa Bellavista 14 E 15◦40′ S, 67◦30′ W 1150–150017 Yolosillas 9 SD 16◦12′ S, 67◦45′ W 1100–1200 [5]18 Zongo 2 E 16◦12′ S, 68◦07′ W 405019 Las Mercedes 18 SD 16◦15′ S, 67◦18′ W 850–1000 [5]20 Chuspipata 9 E 16◦17′ S, 67◦42′ W 2500–2800 [6]21 Coscapa 20 E 16◦18′ S, 67◦44′ W 3000–3500 [6]22 Miguillas 13 D 16◦33′ S, 67◦22′ W 1300–1800 [5]23 Huara 10 D 16◦35′ S, 67◦26′ W 1550–1700 [5]24 Huajchillas 4 D 16◦37′ S, 68◦03′ W 3100–3150 [7]25 Lambate 2 SD 16◦38′ S, 67◦34′ W 2300–235026 Cotacajes deciduous 7 D 16◦46′ S, 66◦44′ W 1600–1750 [8]27 Pujyani 19 E 16◦47′ S, 66◦42′ W 2600–3100 [8]28 Cotacajes semideciduous 12 SD 16◦47′ S, 66◦44′ W 2000–2150 [8]29 Casay Vinto 9 D 16◦52′ S, 66◦38′ W 3350–3500 [8]30 Inquisivi 14 D 16◦54′ S, 67◦09′ W 2250–2500 [5]31 Saila Pata Podocarpus 24 E 16◦55′ S, 66◦55′ W 3000–3500 [8]32 Saila Pata deciduous 6 D 16◦55′ S, 66◦56′ W 2550–2800 [8]33 Villa Tunari 14 E 16◦58′ S, 65◦25′ W 300–450 [9]34 Quime 2 E 17◦01′ S, 67◦16′ W 335035 El Palmar 19 E 17◦06′ S, 65◦30′ W 500–1000 [9]36 Sajta 15 E 17◦07′ S, 67◦05′ W 22037 Carrasco NP 1500–2000 m 17 E 17◦08′ S, 65◦38′ W 1500–2000 [9]38 Carrasco NP 2500–3000 m 13 E 17◦14′ S, 65◦43′ W 2500–3000 [9]39 Carrasco NP 3500–3950 m 8 E 17◦19′ S, 65◦44′ W 3500–3950 [9]40 Rıo Colomelın 14 E 17◦23′ S, 64◦24′ W 350–600 [10]41 Macuñucu 15 SD 17◦44′ S, 63◦36′ W 450–50042 Santa Cruz Botanical Garden 17 D 17◦47′ S, 63◦04′ W 450 [11]43 Karahuasi 10 E 17◦49′ S, 64◦42′ W 2150–240044 Siberia 12 E 17◦50′ S, 64◦43′ W 2600–300045 Rıo Jaya Mayu 5 D 17◦53′ S, 65◦55′ W 2500–2800 [5]46 San Juan del Potrero 17 D 17◦59′ S, 64◦15′ W 1350–1600 [5]47 Rıo Caine 6 D 18◦06′ S, 65◦46′ W 2200–2300 [5]

1477

Table 1. Continued.

No. Study siteNo.plots

Foresttype Coordinates Elevation (m) Ref.

48 Los Volcanes 15 SD 18◦07′ S, 63◦35′ W 1000–130049 San Vicente 5 SD 18◦13′ S, 65◦18′ W 2550–2700 [5]50 Novillero 10 D 18◦18′ S, 65◦15′ W 2300–2450 [5]51 San Lorenzo 14 E 18◦35′ S, 63◦54′ W 1800–2300 [12]52 Loma Larga 13 E 18◦46′ S, 63◦50′ W 1200–1500 [12]53 Masicurı 13 D 18◦49′ S, 63◦41′ W 500–800 [12]54 Rıo Grande 6 D 18◦51′ S, 64◦18′ W 900–130055 Rıo La Haciendita 4 D 18◦56′ S, 64◦17′ W 1500–175056 Nuevo Mundo 14 E 18◦59′ S, 64◦18′ W 2100–245057 Sopachuy 2 D 19◦20′ S, 64◦33′ W 210058 Icla 3 D 19◦23′ S, 64◦48′ W 2250–250059 Padilla 2 D 19◦30′ S, 64◦11′ W 170060 Tarvita 2 D 19◦34′ S, 64◦27′ W 200061 Rıo Pilcomayo 3 D 19◦34′ S, 64◦51′ W 2000–205062 Rıo Azero 22 D 19◦40′ S, 64◦06′ W 1150–1500 [5]63 Villa Orias 2 E 19◦50′ S, 64◦29′ W 2150–220064 Rıo Itacua 10 D 19◦56′ S, 63◦32′ W 850–100065 Cordillera de Mandinga 3 SD 19◦56′ S, 64◦34′ W 2250–2550

References: [1]: Parker and Bailey 1991; Perry et al. 1997; Kessler and Helme 1999*; [2]: Hanagarth1993; [3]: Smith and Killeen 1998; [4]: Seidel 1995; [5]: Herzog et al. 1997; [6]: Morales 1995; [7]:Forno and Baudoin 1991*; [8]: Fjeldså et al. 1999*; Herzog et al. 1999; [9]: Kessler et al. 1999*,2000*; [10]: Kessler et al. in press*; [11]: Saldias 1991; Parker et al. 1993; [12]: Kessler et al. 2000*;Kessler in press a. References including pteridophyte species lists are marked with (*).

studied vegetation types and is small enough to keep environmental factors more orless uniform throughout the plot (Kessler and Bach 1999). Plot location was chosen soas to include ecologically homogeneous and physiognomically representative forestsamples. For the purpose of the present analysis, only those 676 vegetation plotslocated in natural to moderately disturbed forest vegetation were considered (Table 1).Strongly disturbed forests and non-forest habitats were completely excluded fromthis analysis. The number of plots at each location varied with respect to speciesrichness, with as few as 2 forest plots in very species-poor habitats and up to 24 plotsin the more diverse areas. Presence/absence of all species was registered in each plot,treating terrestrial and epiphytic plants separately. Species growing as both epiphytesand terrestrials were included in both groups, while climbing species were treatedas epiphytes. Epiphytic species were collected from fallen branches and trees (alsooutside the actual plots to obtain a better idea of the epiphyte communities in the studyarea), sampled up to heights of 10 m with trimming poles, and observed throughbinoculars at greater heights. All species encountered at each study site (but not inevery single plot) were collected at least in triplicate and have been deposited at theHerbario Nacional de Bolivia, La Paz (LPB) (including all unicates), at the HerbariumGöttingen, Germany (GOET), and with the respective specialists: J.T. Mickel (El-aphoglossum, New York Botanical Garden, NY), R.C. Moran (Pteridophyta in part,

1478

NY), B. Øllgaard (Lycopodiaceae, Herbarium Jutlandicum, Aarhus, Denmark, AAU),and A.R. Smith (most Pteridophyta, University of California at Berkeley, California,UC). Collections that could not be identified, in most cases presumably representingundescribed taxa, were included in the analysis if considered by the specialists torepresent distinct morphospecies.

Environmental variables

The following environmental variables were measured or estimated for each studysite.

Elevation: calculated as the mean elevation of all plots at any given site.Elevational range: the elevational range spanned by a study site. While not bio-

logically meaningfull, this parameter was included because more extensive samplingshould result in higher species numbers and hence could influence the analysis.

Latitude: measured with a global positioning system (Trimble Scout) and throughreference to 1:250.000 topographic maps and given for the center of each study area.

Area: size of area is well-known to influence botanical diversity (Rosenzweig1995). However, in the present case, there are few well-defined habitat islands withclearcut boundaries. This involves mainly isolated dry forest patches in northern andhumid forest patches in southern Bolivia (Figure 1). As a rough estimate of area,the continuous distribution of the respective forest types (deciduous, semideciduous,evergreen) within a belt extending to 1000 elevational meters above and below thestudy site was expressed in km2 to the closest order of magnitude (i.e., 1 km2 = 1,10 km2 = 2, etc.).

Mean annual precipitation: as there are no reliable precipitation maps or modelsfor the Bolivian Andes, precipitation values had to be interpolated in many cases.Climatic data of 131 stations within or close to the study region were kindly providedby E. Jordan (Geographical Institute, University of Düsseldorf, Germany), with ad-ditional data gleaned from Morales (1990) and Jordan (1991). Nine of the study siteswere in the immediate vicinity of climatic stations while twelve additional sites couldbe clearly linked to a climatic station at no more than 10 km linear distance. Theremaining 44 study sites were distant from any climatic station and their values wereinterpolated from nearby stations with similar vegetation types. To account for esti-mation inaccuracies, all precipitation values were grouped into classes of 200 mm be-low 1400 mm mean annual precipitation, classes of 400 mm at 1400–3000 mm meanannual precipitation and classes of 600 mm at higher precipitation levels. Since thisresulted in no less than 17 classes with data, precipitation was treated as a continuous,rather than categorical variable, during data analysis.

Mean annual temperature: data was available for the same 131 climatic stationsmentioned above. To correct for elevational differences between the stations and thestudy sites, a lapse rate of 0.54 ◦C × 100 m−1, calculated on basis of the 131 stations,was used.

1479

Number of frost days per year: calculated from elevation after Eriksen (1986)and Jordan (1991). This parameter is not linearly related to either elevation or meanannual temperature and may directly influence the distribution of plant species.

Variability of NDVI and Ts: These data sets were provided by E. Lambin and co-workers (Dept de Géographie et de Géologie, University Catolique de Louvain, Bel-gium) as indices of the interannual climatic variability (see Fjeldså et al. 1999b, fordetails). To quantify erratic and anomalous climatic events that may be biologicallycritical (e.g. droughts, cold spells, exceptional rains), coefficients of variation in NDVIand Ts values were calculated by on a month-by-month basis based on 10 years data(1983–1992) of the Global Area Coverage (GAC) remote sensing data provided by theNational Aeronautics and Space Administration (NASA) for global change research.

Distance to wetter habitats: the linear distance to the closest site with at least30% higher precipitation and not more than 1000 elevational meters apart from thestudy area was estimated as a measure of the environmental heterogeneity of a givenstudy site. Localities with high heterogeneity may be expected to have higher speciesrichness due to sink or mass effects (Shmida and Ellner 1984).

Distance to dryer habitats: as above, but for 30% lower precipitation.Canopy height: average canopy height was estimated in each study plot and av-

eraged over all plots of each study site as a proxy for ecosystem productivity (Lieth1975) and structural forest complexity (Terborgh 1977).

Bryophyte cover: cover of epiphytic bryophytes on canopy tree branches was es-timated in each study plot and averaged over all plots of each study site as an indexof air humidity (Frahm and Gradstein 1991; Wolf 1993).

Human impact: human impact was classified as: absent (0); very low (1), foreststructure basically intact; low (2), general forest structure maintained, canopy heightnot lowered relative to natural forest; medium (3), forest structure clearly modified,canopy lowered, understory composition changed; and strong (4), forest stronglymodified with a lower and more open canopy.

Of the possible additional parameters, geological substrate was not included be-cause of their homogeneity throughout the study region, corresponding mostly tored sandstones, lutites and quarcitic rocks of Devonian and Ordovician age (Parejaet al. 1978) with fairly constant nutrient availability (P. Schad, University of Münchenat Freising, Germany, pers. comm.). Only three sites differed noticeably by havingeither nutrient-poor white sands (sites 6 and 33 in Table 1) or nutrient-rich mixedalluvial soils (site 36). Furthermore because of the dependence of soil developmenton climatic parameters, especially humidity and temperatures, and because of therestriction of the study to zonal forest vegetation, soil and climatic parameters areprobably highly correlated at the scale of the present study (Sevink 1984).

Pairwise correlations between the environmental variables revealed a rather well-balanced data matrix with few spurious correlations and reflecting all the known orexpected causal relationships, e.g. between elevation and mean annual temperature(Table 2).

1480

Tabl

e2.

Pear

son

corr

elat

ion

coef

ficie

nts

desc

ribi

ngth

ebi

vari

ate

rela

tions

hips

betw

een

abio

ticat

trib

utes

ofth

e65

stud

ysi

tes.

Ele

vatio

nal

rang

eL

atitu

deA

rea

Prec

ipita

tion

Tem

pera

ture

No.

fros

tda

ysV

aria

bilit

yN

DV

IV

aria

bilit

yT

s

Dis

tanc

eto

wet

ter

habi

tats

Dis

tanc

eto

drye

rha

bita

tsC

anop

yhe

ight

Bry

ophy

teco

ver

Hum

anim

pact

Ele

vatio

n0.

130.

11−0

.54*

*−0

.42*

−0.9

6***

0.90

***

0.01

0.07

−0.0

2−0

.46*

−0.7

0**

0.35

0.40

*E

leva

tiona

lran

ge0.

02−0

.10

0.32

−0.2

2−0

.01

0.37

−0.1

40.

25−0

.25

0.12

0.56

**−0

.15

Lat

itude

−0.1

1−0

.40*

−0.0

70.

31−0

.03

−0.2

00.

14−0

.06

−0.2

4−0

.08

0.10

Are

a0.

50*

0.55

**−0

.34

0.05

0.04

0.50

**0.

41**

0.63

**−0

.03

−0.4

7*Pr

ecip

itatio

n0.

32−0

.47*

0.25

0.04

0.41

*−0

.07

0.67

**0.

43*

−0.6

6**

Tem

pera

ture

−0.8

5***

−0.1

0−0

.05

0.01

0.48

*0.

63**

−0.4

2*−0

.28

No.

fros

tday

s−0

.11

0.06

0.13

−0.3

6−0

.59*

*0.

300.

36V

aria

bilit

yN

DV

I0.

00−0

.07

−0.0

6−0

.04

0.17

−0.1

5V

aria

bilit

yT

s−0

.03

0.15

−0.1

3−0

.02

0.02

Dis

tanc

eto

wet

ter

habi

tats

0.00

0.32

0.48

*−0

.37

Dis

tanc

eto

drye

rha

bita

ts0.

13−0

.51*

*−0

.09

Can

opy

heig

ht0.

16−0

.63*

*B

ryop

hyte

cove

r−0

.40*

*P

<0.

05;*

*P

<0.

01;*

**P

<0.

001;

prob

abili

ties

subj

ecte

dto

Bon

ferr

onic

orre

ctio

n.

1481

Data analysis

Overall species richness was extrapolated from species-accumulation curves usingthe presence/absense of species at individual study sites through Chao’s (1984) Chao1 formula S1 = Sobs + (a2/2b), where S1 is the estimated number of species, Sobs theobserved species number, a the number of species registered only once (singletons),and b the number of species recorded twice (doubletons). This approach providesa rough estimate of total species numbers in areas that have not been thoroughlysampled, as well as of the completeness of the sampling (Colwell and Coddington1995). Among the estimators presented by Colwell and Coddington (1995), the Chao1 estimator was selected because of its relatively low sensitivity to varying sample in-tensity and species richness (Walther and Morand 1998), because of the simplicity ofcalculation, and because several other potential estimators were not applicable to thedata structure of the present study. Extrapolations were based on 2626 species-recordsfrom the 65 study sites. An additional 289 site-records, based on 120 collections thatcould not be identified with certainty, were excluded from the estimation.

Species richness of each study site was quantified in two different ways. Firstly, asthe point diversity (sensu Whittaker 1972), which was measured as the mean numberof species recorded per plot at a given site. Secondly, as the alpha diversity (sensuWhittaker 1972), i.e. the total species richness of each site. These two measures allowthe comparison of species richness at two different spatial scales, i.e. on areas of400 m2 (point diversity) and on areas of 1–10 km2, corresponding to individual studysites (alpha diversity). Since the number of plots varied strongly between study sites,and given the importance of controlling for variation in sampling intensity (Rahbek1997), alpha diversity was not expressed as the absolute species numbers recordedat each site, but through the index of relative species of Prendergast et al. (1993)(see also Lwanga et al. 1998). This approach compares the richness of rarefied site-specific species accumulation curves (Figure 2). For each comparison of two studysites, the ratio of species richness at the highest sampling intensity common to bothsites, is calculated. The overall richness index for any given site is expressed as thegeometric mean of all such pairwise values. This approach allows the comparison ofthe overall species richness of different sites despite varying sampling intensity.

To visualize the elevational distribution of pteridophyte species richness and it’srelationship to bryophyte cover, data from 123 zonal forest plots along a fully sampledelevational transect in Parque Nacional Carrasco (spanning sites 33 and 35–39, butalso including further sampling between those sites) were averaged in steps of 200elevational meters and plotted against elevation (Figure 5).

Bivariate correlations of species richness were calculated with all environmen-tal variables. In some cases, additional multivariate correlations, involving some orall environmental attributes, were calculated. Where necessary, values were log- orsquare root-transformed to approximate normality to fit the requirements of para-metric statistics. To control for type I error, significance levels where subjected to a

1482

Figure 2. Rarefied species-accumulation curves of pteridophytes for (A) the 65 study sites in Bolivia, and(B) three selected study sites (sites 1, 37, 46). The relative richness index is calculated as the ratio of therichness of such rarefied curves at the highest sampling intensity common to two sites (Prendergast et al.1993). For any given site, these ratios are averaged to obtain a single value for each site. In the presentexample, Rıo San Antonio would obtain values of 0.2395 relative to Carrasco (40 species in 16 plots at RıoSan Antonio/167 species in 16 plots at Carrasco) and 3.3613 relative to San Juan del Potrero (40/11.9),giving a mean total index value of 1.800.

Bonferroni correction (Sokal and Rohlf 1995). Statistics were carried out withSYSTAT version 7.0 for Windows (SYSTAT 1997).

Results

Species richness

In total, the study included 755 pteridophyte species, representing 426 (56%) epiphy-tic and 598 (79%) terrestrial taxa, i.e. including 266 species (35%) recorded as bothepiphytes and terrestrials (Table 3). The genera with the highest species counts wereElaphoglossum (120 species; 16%), Asplenium (53; 7%), Thelypteris (49; 7%), andPolypodium and Hymenophyllum (both 37; 5%). One hundred and fourty two spe-cies (19%) corresponded to unnamed morphospecies. While overall there were moreterrestrial than epiphytic species, at individual sites and plots there were generallymore epiphytic taxa (Table 3). Total species numbers were estimated with the Chao1 formula at 975 species overall, including 559 epiphytes, and 880 terrestrials, indi-cating that at least 23% of the actual total species number in the study region wentunrecorded (Table 3).

The majority of species were recorded at few study sites (Figure 3). No less than255 species (34%) were recorded at only one study site, 119 species (16%) even inonly one of the 676 study plots. Only 140 species (19%) were recorded at six ormore study sites, while a mere 18 species (2%) occurred in 66 (10%) or more plots.

1483

Table 3. Values of pteridophyte species richness based on 676 study plots in forests at 65 sitesin the Bolivian Andes. Estimated numbers were calculated with the Chao 1 formula (Chao1984).

All Epiphytic Terrestrial

Number of plot-records 9940 5354 5613Number of site-records 2626 1314 1802Recorded species number 755 426 598Estimated species number 975 559 880Mean/median number of species per site 40.4/26 20.2/8 27.7/19Maximum number of species per site 167 110 101Mean/median number of species per plot 14.7/6 12.6/5 9.4/4Maximum species number in a single plot 67 46 31Maximum mean species number per plot at a given site 47.9 31.1 20.9

Over 50% of all plot-records were contributed by only 109 species (14%). Unnamedspecies ocurred at significantly fewer sites per species (1.69) than the named taxa(3.92) (two-tailed t-test, P < 0.01).

Values of point diversity (mean number of species per plot) varied from 0 to 31.1for epiphytes, 0 to 20.9 for terrestrial taxa, and 0 to 47.9 for all species combined(Table 3). Respective values of alpha diversity (relative species richness) ranged from0 to 22 for epiphytes, 0 to 10.2 for terrestrial taxa, and 0 to 11.7 for all species com-bined. The highest numbers of species recorded at individual study sites were 110epiphytes, 101 terrestrials, and 167 species in total. All of these maximum values

Figure 3. Rank-order classification of the pteridophytes recorded at 65 study sites in Bolivia. In this plot,the species are ordered according to the number of site-records. Note that a large number of species,especially among terrestrials, were recorded at only one or two sites, while very few species were recordedat ten or more sites.

1484

were obtained at study site 37, located at 1500–2000 m in the wettest part of theBolivian Andes (Figure 1). Values of pteridophyte species richness were highly cor-related among both measures of species richness as well as between epiphytic andterrestrial taxa (Table 4).

Environmental correlates

Pteridophyte species richness was highly correlated with mean annual precipitation,epiphytic bryophyte cover, and human impact (Table 5). Weaker correlations weredetected in some cases to canopy height, elevational range spanned by the study site,and distance to the closest area with at least 30% higher precipitation. However, withcorrelation values of r = 0.85 to r = 0.90 mean annual precipitation and epiphy-tic bryophyte cover together included virtually all of the variation explained by allenvironmental factors combined, which varied from r = 0.85 to r = 0.91. Whenprecipitation and bryophyte cover were controlled for, the remaining variability didnot show significant correlations to any of the other environmental factors, the high-est value being a correlation coefficient of r = 0.33 of point richness of terrestrialpteridophytes to elevation (data not shown).

The elevational distribution of point diversity of epiphytic and terrestrial pterido-phytes, and of bryophyte cover along an elevational transect in central Bolivia showedall hump-shaped distributions (Figure 5). However, the maxima of pteridophyte rich-ness were located at around 1800 m, while bryophyte cover peaked at about 3600 m.Furthermore, the decline of epiphytic pteridophyte richness at higher elevations wasmuch more pronounced than that of terrestrial taxa.

Discussion

Species richness

The present study involved roughly 1500 person-labour-days for field sampling, lo-gistics, and specimen handling. Despite this high labour investment, the survey was

Table 4. Pearson correlation coefficients describing the bivariate relationships between the var-ious measures of pteridophyte species richness (point diversity, alpha diversity). All values aresignificant at P < 0.001 (subjected to Bonferroni correction).

All species,alpha

Epiphytes,point

Epiphytes,alpha

Terrestrials,point

Terres-trials,alpha

All species, point 0.94 0.95 0.92 0.89 0.89All species, alpha 0.93 0.96 0.86 0.95Epiphytes, point 0.96 0.81 0.84Epiphytes, alpha 0.79 0.85Terrestrials, point 0.87

1485

Table 5. Bivariate linear regression coefficients between the measures of pteridophyte diversity and the14 environmental variables, and multivariate linear regression coefficients with precipitation + bryophytecover and with all environmental variables.

All species,point

All species,alpha

Epiphytes,point

Epiphytes,alpha

Terrestrials,point

Terrestrials,alpha

Elevation −0.02 0.06 0.11 −0.16 0.07 0.03Elevational range 0.37 0.48* 0.43 0.43 0.36 0.53Latitude −0.30 0.29 0.43 0.29 −0.27 −0.25Area 0.32 0.32 0.39 0.37 0.18 0.23Precipitation 0.76*** 0.80*** 0.77*** 0.79*** 0.70*** 0.74***Temperature −0.06 −0.02 0.04 0.09 −0.14 −0.12No. frost days −0.09 −0.15 −0.21 −0.25 0.01 −0.06Variability NDVI 0.26 0.29 0.33 0.33 0.16 0.25Variability Ts −0.01 −0.06 −0.01 −0.03 0.03 −0.06Distance to wetter

habitats0.50* 0.51* 0.49* 0.47 0.43 0.49*

Distance to dryerhabitats

−0.30 −0.26 −0.17 −0.13 −0.42 0.35

Canopy height 0.41 0.46* 0.45* 0.48** 0.33 0.41Bryophyte cover 0.71*** 0.71*** 0.67*** 0.62*** 0.73*** 0.76***Human impact −0.62*** −0.61*** −0.64*** −0.63*** −0.55** −0.55**Precipitation +

bryophyte cover0.87*** 0.90*** 0.86*** 0.85*** 0.85*** 0.89***

All variables 0.88*** 0.91*** 0.87*** 0.85*** 0.88*** 0.91***

* P < 0.05; ** P < 0.01; *** P < 0.001; probabilities subjected to Bonferroni correction.

far from complete, as indicated by large number of species recorded at few sites andby the extrapolated number of 975 species. Even this figure is likely to be an underes-timate. Computer simulations using artificial communities of known species richnesshave shown that when less than 70% of the actual species number are recorded byrandom sampling, the Chao 1 estimator (and other estimators as well) yields estimatesat least 10–20% below the actual values (S.K. Herzog and M. Kessler, unpublisheddata). In the present case, the large difference between estimated and recorded speciesnumbers (77%) indicates that this extrapolation error probably applies. An exampleof this situation is presented by the genus Elaphoglossum, of which 120 species wererecorded based on about 400 gatherings. The Chao 1 formula yields an estimated totalnumber of 155 species. However, a sample of about 1000 Bolivian Elaphoglossumspecimens deposited at New York Botanical Garden (including the specimens of thepresent study as well as those of other botanists) studied in 1998 by J.T. Mickel andthe author, already contained about 145 species from Andean forest habitats (unpub-lished data). Applying the Chao 1 formula to this data set, the total species numberfor this genus in Bolivian Andean forests can be estimated to be around 185. Onecan only speculate at the number of species that will be estimated to occur once 185species have been recorded. The paucity of knowledge on Bolivian pteridophytes isfurther highlighted by the large number of recent new records from the country andthe high proportion of undescribed taxa (Smith et al. 1999).

1486

Sampling at individual study sites was just as incomplete as that of overall speciesrichness. Most study sites had species-accumulation curves that were far from ap-proaching an asymptotic value (Figure 2). As a further example, along an elevationaltransect in central Bolivia 473 species of pteridophytes were recorded in 184 plots(including non-forest vegetation), but the actual species number was extrapolated toexceed 600 (Kessler et al. 1999). On average, it may be estimated that about twothirds of the species present at each study site were actually recorded in this study.

The difficulty of assessing the total species number of pteridophytes in the Bolivi-an Andes relates to the fact that the majority of the species are rare and/or of localizedoccurrence. A similar situation has been reported on Mount Kinabalu, Borneo, wheredespite the accumulation of over 3500 collections of pteridophytes, 22% of the spe-cies remain known from just one collection or one locality (Parris et al. 1992; Parris1997). No fewer than 20% of the species have not been recorded for more than 60years. Presumably, many species are subject to metapopulation fluctuations, perhapsin response to the marked interannual climatic variability (Parris et al. 1992; Parris1997).

Much additional field work is necessary before the number of rare and localizedpteridophyte species in the Bolivian Andes is known or can even be reliably esti-mated. The high proportion of localized taxa is particularly noteworthy in view ofthe spore dispersal in this group that should ensure good dispersal (Barrington 1993).Possibly, many taxa have specific ecological requirements, which would explain theirpatchy distribution. Comparative studies of other plant groups with different dispersalmodes would be of interest.

Table 6 provides an overview of the number of pteridophytes recorded from vari-ous tropical regions. These numbers should be compared with caution, however, sincethe present study includes undescribed taxa but not species from non-forest habitats.

Table 6. Number of pteridophyte species recorded from various tropical regions.

Site Species number Reference

Worldwide 9000+ Tryon and Tryon 1982Neotropics

South America 3000 Moran 1995Bolivian montane forests 755 (975 estimated) This studyPeru 1060 Tryon and Stolze 1994Ecuador 1345 Jørgensen and Leon-Yañez 1999Venezuelan Guayana 671 Smith 1995Jamaica 609 Proctor 1985Mesoamerica 1358 Moran and Riba 1995Oaxaca, Mexico 690 Mickel and Beitel 1988

Malesia 3000+ Parris 1993Borneo 986 Parris 1993Phillipines 982 Parris 1993

Continental Africa 500 Parris 1985

1487

As a further complication, the species concepts applied in the Peruvian treatment arenarrower than those used here, especially for the genera Asplenium, Blechnum, Poly-stichum, Polypodium, and Selaginella. The greatest discrepancy, however, is certainlyposed by the different collecting intensities on which these numbers are based. Inany case, the Bolivian montane forests contain one of the richest pteridophyte florasdocumented to date on a worldwide basis.

The change of the ratio of terrestrial to epiphytic taxa at different spatial scales(Table 3) is noteworthy. This ratio declines from about 1.5 based on total species num-bers, to 1.37 at individual sites, and to 0.75 within individual plots. Clearly, epiphyticspecies are more regularly and evenly distributed than terrestrial taxa. This patternis presumably a result of the smaller-scale spatial ecological niche differentiation ofepiphytes, taking place mostly within individual tree crowns, while terrestrial taxa aremore strongly influenced by the topographic and edaphic variation at larger spatialscales (Johannson 1974; Ibisch 1996; Ibisch et al. 1996).

Environmental correlates

The relationship of pteridophyte species richness to humidity, here measured throughmean annual precipitation and epiphytic bryophyte cover, is well known and relatesto the poikilohydric nature of many pteridophytes (Schimper 1888; Gurung 1985;van der Werff 1990). However, the degree to which this relationship dominated theanalysis is surprising. The correlation values of r = 0.85 to r = 0.90, correspondingto determination values of r2 = 0.72 to r2 = 0.81, are very high for ecologicalstudies (Gotelli and Graves 1996), especially considering the unavoidable samplingbiases involved and the incomplete climatic data available for the Bolivian Andes.

This close correlation raises the question whether all trends of pteridophytespecies richness, e.g. the hump-shaped pattern along elevational gradients, can beexplained with respect to humidity. In the tropical montane regions, precipitationgenerally peaks at mid-elevations just below the prevailing cloud condensation levels(Lauer 1975). As shown by a comparative analysis of the elevational distribution ofprecipitation from mountain areas worldwide, this peak is located at lower elevationsin areas with very high precipitation as well as in small isolated mountain rangesand at higher elevations in less humid areas and in large mountain areas (Lauer1975). This variability appears to correspond to the different elevational maxima ofpteridophyte species richness observed along elevational gradients. Panama, with thelowest recorded maximum at 500–1500 m (Lellinger 1985) is characterized by highprecipitation as well as rather low overall mountain ranges, whereas eastern Africa,where richness peaks around 1500–2000 m (Jacobsen and Jacobsen 1989) is muchdryer overall. Other localities are intermediate, both with respect to precipitation lev-els, montane area, and pteridophyte species richness patterns, with maxima at 1500 min Borneo (Parris et al. 1992) and in Bolivia (Kessler et al. 1999), and at 1000–1500 min southern Africa (Jacobsen and Jacobsen 1989).

1488

Further evidence for the importance of humidity on the elevational distributionof pteridophytes is provided by the distributional patterns of numerous fern speciesthat occur throughout the humid tropical Andes at mid-elevations but that reach theAmazonian lowlands in eastern Ecuador and adjacent Peru (Figure 4). This unusualdistribution pattern can be related to the year-round high precipitation levels recordedin the western Amazon basin (Figure 4). While up to 3000 mm mean annual precip-itation has also been recorded from Bolivian lowland sites, the climate in this regionis characterized by a distinct dry season from June through September. Thus, it seemslikely that lower elevational limit of numerous pteridophyte species in the BolivianAndes is determined by a decrease of humidity from the mist-shrouded cloud forestsat mid-elevations to the lowlands.

The decrease of species richness at higher elevations, at the other end of the hump-shaped diversity curves, represents a different situation. While precipitation certainlydeclines at elevations above 1500–2500 m in the Bolivian Andes, this is coupled witha decrease of temperatures, resulting in lower evapotranspiration and hence higherenvironmental humidity. This pattern is evidenced by the increase of epiphyte coveralong elevational transects in the Andes, usually peaking somewhere between 2500m and 3500 m (Wolf 1993). This humidity peak is located well above the maxima ofpterdidophyte species richness. For example, along an elevational gradient in ParqueNacional Carrasco, depto. Cochabamba, Bolivia, pteridophyte species richness peaksat 1500 m, while epiphytic bryophyte cover is highest from about 1800 m to 3000 m(Figure 5). Several factors may be involved. Low temperatures at high elevations mayaffect the development of vascular plants to a higher degree than that of bryophytes.It is thus conceivable that pteridophytes are competitively excluded by bryophytesunder the most humid conditions at high elevations. Alternatively, the less complexstructure of high-elevation forests may reduce the number of ecological niches avail-able to pteridophytes. These hypotheses could be subjected to experimental testing.In any case, these factors are more likely to affect epiphytic than terrestrial species,as evidence by the stronger decline of epiphytic pteridophyte richness with elevation,compared to terrestrial taxa (Figure 5). It thus appears that the decline of pteridophytespecies richness at high elevations is not directly linked to a corresponding decline inhumidity but that other factors also play a role.

Even if additional environmental factors are causally involved, the close correla-tion of pteridophyte species richness to humidity may be used to predict pteridophytediversity on the basis of climatic data. Interestingly, elevational patterns of speciesrichness appear to be quite similar in Bolivia and Borneo (M. Kessler and B.S. Parris,unpublished data), raising the possibility that the relationship of humidity to pteri-dophyte species richness may be applicable to both the Neotropics and to Malesia.Africa has a depauperate pteridophyte flora (Parris 1985) and even though the patternsmay be similar, the actual richness values are undoubtedly lower.

Given the overriding importance of humidity in determining pteridophyte spe-cies richness, none of the other factors played an important role. For example, while

1489

Fig

ure

4.M

aps

ofno

rthw

este

rnSo

uth

Am

eric

a(c

onto

urlin

eat

1000

m)

show

ing

(A–E

)se

lect

edpt

erid

ophy

tesp

ecie

sof

wid

espr

ead

mon

tane

occu

rran

cew

hich

spre

adin

toth

elo

wla

nds

inth

em

osth

umid

regi

onof

the

Am

azon

basi

n.M

apF

(mod

ified

afte

rH

affe

r19

87)

show

sth

edi

stri

butio

nof

perh

umid

clim

ate,

i.e.w

ithno

mon

thbe

low

100

mm

mea

npr

ecip

itatio

nin

low

land

wes

tern

Am

azon

ia.A

sim

ilar

dist

ribu

tion

patte

rnis

shar

ed,e

.g.b

yA

sple

nium

serr

aL

angs

d.an

dFi

sch.

,Dan

aea

elli

ptic

aSm

.in

Ree

s,D

ipla

zium

ambi

guum

Rad

di,D

.exp

ansu

mW

illd.

,Hym

enop

hyll

umap

icul

atum

Met

tex

Kuh

n,an

dM

icro

gram

ma

late

vaga

ns(M

axon

)L

ellin

ger.

1490

Figure 5. Species richnes of epiphytic (open circles, closed line) and terrestrial (filled circles, closed line)pteridophytes, and epiphytic bryophyte cover (squares, dashed line) along an elevational transect in Carr-asco National Park, Bolivia. Smoothed lines were fitted by distance-weighted least-squares smoothing.

human disturbance was significantly negatively correlated to pteridophyte richness,this appears to be a result of much stronger human impact in arid areas that are sub-jected to timber extraction and grazing than in very wet regions that generally havea low human population density in the Bolivian Andes (Kessler and Beck, in press).This assumption is supported by comparisons between adjacent natural and anthro-pogenically disturbed forests, which reveal little change in species richness as longas the forest habitat as such is not destroyed (Kessler, in press b). Ecoclimatic sta-bility, while potentially an important factor, showed no significant relationship topteridophyte richness.

Conclusions and conservation implications

Despite intensive recent efforts, the total number of pteridophyte species occurringin Bolivian montane forests cannot yet be reliably estimated. The knowledge on thedistribution and ecology of the individual species is fragmentary, to say the least.Accordingly, it is difficult to assess the impact of human activities on pteridophytediversity in humid Andean forests or to design conservation strategies in order topreserve as much of the pteridophyte flora as possible.

1491

While the feasibility of estimating pteridophyte species richness at a given sitebased on climatic data is certainly of use for conservation purposes, a more importantfactor is the complementarity between sites, i.e. the degree to which different sitesinclude additional components of the overall pteridophyte diversity. Thus, even therichest site in the present study (site 37) did not include more than 167 (22%) of the755 species recorded in total, showing that the conservation of overall pteridophytespecies richness in the Bolivian Andes depends on the protection of a large numberof sites. Taken together, the ten study sites with the highest species numbers included530 species, i.e. 70% of the total. However, it is impossible at present to decide onthe conservation importance of sites that have not been sampled, many of which arelocated in areas already under legal protection. Thus, an analysis of complementarity,such as that conducted by Moreno et al. (1996) to prioritize conservation areas forpteridophytes on the Iberian Peninsula, cannot be performed until a more completecoverage of the distribution of Bolivian pteridophytes has been achieved.

On the other hand, an analysis of pteridophyte endemism based on the same 65study sites included in the present study showed that endemism is positively correlat-ed to bryophyte cover (r = 0.63 to r = 0.65) and to a lesser degree to precipitation(M. Kessler and E. Lambin, unpublished data). Thus, localities with high pterido-phyte diversity are also characterized by a high proportion of species with restrictedranges. In this sense, the conservation of the wettest forest sites in the Bolivian Andeswill include both the most diverse and most unique pteridophyte communities, eventhough other habitats naturally include further species not present in humid forests.Given that a full inventory of Bolivian pteridophytes is unlikely to be completed inthe near future and considering the rate at which forest destruction is proceeding inthe country (Kessler and Beck, in press), a conservation strategy for pteridophytesfocussing on the most humid forest areas currently appears to be the most practi-cal approach. Fortunately, much of the humid montane forest in Bolivia is alreadyincluded in several national parks, although most of these are under pressure fromencroaching agriculture, logging, and road construction.

Acknowledgements

I thank A. Acebey, K. Bach, J.A. Balderrama, J. Bolding, J. Fjeldså, J. Gonzales, A.Green, S.K. Herzog, B. Hibbits, S. Hohnwald, I. Jimenez, T. Krömer, J.-C. Ledesma,M. Olivera, A. Portugal, J. Rapp, J. Rodriguez, and M. Sonnentag for help and goodcompanionship during field work, J.T. Mickel, R.C. Moran, B. Øllgaard, A.R. Smith,and I. Valdespino for specimen identification, and K. Bach and E. Kessler for databasemanagement. This study would have been impossible without the logistic support bythe Herbario Nacional de Bolivia, La Paz, in particular by S.G. Beck, M. Cusicanqui,A. de Lima, R. de Michel, and M. Moraes. For working and collecting permits, Ithank the Dirección Nacional de Conservación de la Biodiversidad (DNCB), La Paz.

1492

E. Lambin and J. Fjeldså kindly provided the remote sensing data. For discussion anduseful comments on earlier versions of the manuscript I thank H. Balslev, A.R. Smith,and an anonymous reviewer. This study was supported by the Deutsche Forschungs-gemeinschaft, the A.F.W. Schimper-Stiftung, and the DIVA project under the DanishEnvironmental Programme.

References

Balslev H, Valencia R, Paz y Miño G, Christensen H and Nielsen I (1999) Species count of vasular plantsin one hectare of humid lowland forest in Amazonian Ecuador. In: Dallmeier F and Comiskey JA (eds)Forest Biodiversity in North, Central and South America and the Caribbean. Research and Monitoring,pp 585–594. Man and the Biosphere series 21, UNESCO, Paris and Parthenon Publishing Group, NewYork

Barrington DS (1993) Ecological and historical factors in fern biogeography. Journal of Biogeography 20:275–280

Chao A (1984) Non-parametric estimation of the number of classes in a population. Scandinavian Journalof Statistics 11: 265–270

Colwell RK and Coddington JA (1995) Estimating terrestrial biodiversity through extrapolation. In:Hawksworth DL (ed) Biodiversity Measurement and Estimation, pp 101–118. Chapman & Hall, London

Duivenvoorden JF (1994) Vascular plant species counts in the rain forests of the middle Caquetá area,Colombian Amazonia. Biodiversity and Conservation 3: 685–715

Eriksen W (1986) Frostwechsel und hygrische Bedingungen in der Punastufe Boliviens. Ein Beitrag zurÖkoklimatologie der randtropischen Anden Bolivien. In: Buchholz HJ (ed) Beiträge zur physischenGeographie eines Andenstaates, pp 1–21. Jahrbuch der Geographischen Gesellschaft Hannover 1985,Hannover

Fjeldså J, Lambin E and Mertens B (1999) Correlation between endemism and local ecoclimatic stabilitydocumented by comparing Andean bird distributions and remotely sensed land surface data. Ecography22: 63–78

Forno E and Baudoin M (1991) Historia natural de un valle de los Andes: La Paz. Instituto de Ecología,UMSA, La Paz

Frahm J-P and Gradstein SR (1991) An altitudinal zonation of tropical rain forests using bryophytes.Journal of Biogeography 18: 669–678

Galeano G, Suárez S and Balslev H (1998) Vascular plant species count in a wet forest in the Chocó areaon the Pacific coast of Colombia. Biodiversity Conservation 7: 1563–1575

Gentry AH (1988) Changes in plant community diversity and floristic composition on environmental andgeographical gradients. Annals of the Missouri Botanical Garden 75: 1–34

Gentry AH (1995) Patterns of diversity and floristic composition in neotropical montane forests. In: Chur-chill SP, Balslev H, Forero E and Luteyn JL (eds) Biodiversity and Conservation of Neotropical MontaneForests, pp 103–126. The New York Botanical Garden, New York

Gentry AH and Dodson CH (1987) Contribution of nontrees to species richness of a tropical rain forest.Biotropica 19: 149–156

Gotelli NJ and Graves GR (1996) Null Models in Ecology. Smithonian Institution Press, Washington, DCGurung VDL (1985) Ecological observations on the pteridophyte flora of Langtang National Park, central

Nepal. Fern Gaz. 13: 25–32Haffer J (1987) Quaternary history of tropical America. In: Whitmore TC and Prance GT (eds) Biogeog-

raphy and quaternary history in tropical America, pp 1–18. Oxford University Press, OxfordHanagarth W (1993) Acerca de la geoecología de las sabanas del Beni en el noreste de Bolivia. Instituto

de Ecología, UMSA, La PazHerzog SK, Kessler M, Maijer S and Hohnwald S (1997) Distributional notes on birds of Andean dry

forests in Bolivia. Bulletin of the British Ornithologists’ Club 117: 223–235

1493

Herzog SK, Fjeldså J, Kessler M and Balderrama JA (1999) Ornithological surveys of the Cordillera Coca-pata, depto. Cochabamba, Bolivia, a transition zone between humid and dry intermontane Andean hab-itats. Bulletin of the British Ornithologists’ Club 119: 162–177

Holben BN, Tucker CJ and Fan CJ (1980) Spectral assessment for soybean leaf area and leaf biomass.Photographic Engineering and Remote Sensing 46: 651–656

Huston MA (1994) Biological Diversity: The Coexistence of Species on Changing Landscapes. CambridgeUniversity Press, Cambridge

Ibisch PL (1996) Neotropische Epiphytendiversität-das Beispiel Bolivien. Archiv naturwissenschaftlicherDissertationen 1. Wiehl, Martina Galunder-Verlag, Germany

Ibisch PL, Boegner A, Nieder J and Barthlott W (1996) How diverse are neotropical epiphytes? An analysisbased on the ‘Catalogue of the flowering plants and gymnosperms of Peru’. Ecotropica 1: 13–28

Jacobsen WBG and Jacobsen NHG (1989) Comparison of the pteridophyte floras of southern and easternAfrica, with special reference to high-altitude species. Bullitin Jardin Botanique Belgique 59: 261–317

Johansson DR (1974) Ecology of vascular epiphytes in West African rain forests. Acta PhytogeographicaSuecica 59: 1–136

Jordan E (1991) Die Gletscher der bolivianischen Anden. Franz Steiner, StuttgartJørgensen PM and León-Yañez S (1999) Catalogue of the Vascular Plants of Ecuador. Missouri Botanical

Garden, St. LouisKessler M (in press a) Species richness, endemism, and altitudinal zonation of selected plant groups along

an elevational transect in the central Bolivian Andes. Plant EcologyKessler M (in press b) Maximum plant community endemism at intermediate intensities of anthropogenic

disturbance in Bolivian montane forests. Conservational BiologyKessler M and Bach K (1999) Using indicator groups for vegetation classification in species-rich Neotrop-

ical forests. Phytocoenologia 29: 485–502Kessler M and Beck SG (in press) Bolivia. In: Kappelle M and Brown AD (eds) Bosques de Montañas

Tropicales IUCN – INBio – Fundación ANA. Instituto Nacional de Biodiversidad. Santo Domingo deHeredia, Costa Rica

Kessler M and Helme N (1999) Floristic diversity and phytogeography of the central Tuichi Valley, anisolated dry forest locality in the Bolivian Andes. Candollea 54: 341–366

Kessler M, Smith AR and Gonzales J (1999) Inventario de pteridófitos en una transecta altitudinal delParque Nacional Carrasco, dpto. Cochabamba, Bolivia. Rev. Soc. Boliviana Bot. 2: 227–250

Kessler M, Krömer T and Jimenez I (2000) Inventario de grupos selectos de plantas en el Valle de Masicurí(Santa Cruz – Bolivia). Rev. Boliviana Ecol. Conserv. Ambiental 8: 3–15

Kessler M, Smith AR, Acebey A and Gonzales J (in press) Registros adicionales de pteridófitos del ParqueNacional Carrasco, dpto. Cochabamba, Bolivia. Rev. Soc. Boliviana Bot.

Lauer W (1975) Klimatische Grundzüge der Höhenstufen tropischer Gebirge. Innsbruck: 40, Deutsch,Geographentag

Lellinger DB (1985) The distribution of Panama’s pteridophytes. Monographs of Systematic Botany of theMissouri Botanical Garden 10: 43–47

Lieth H (1975) Modeling the primary productivity of the world. In: Lieth H and Whittaker RH (eds)Primary Productivity of the Biosphere, pp 237–263. Springer-Verlag, New York

Lwanga JS, Balmford A and Badaza R (1998) Assessing fern diversity: relative species richness and itsenvironmental correlates in Uganda. Biodiversity and Conservation 7: 1378–1398

Mickel JT and Beitel JM (1988) Pteridophyte flora of Oaxaca, Mexico. Bronx, The New Botanical Garden,New York

Morales Cde (1990) Bolivia: medio Ambiente y Ecología Aplicada. Instituto de Ecología, UMSA, La PazMorales Cde (1995) Caminos de Cotapata. Instituto de Ecología-FUNDECO, La PazMoran RC (1995) The importance of mountains to pteridophytes, with emphasis on Neotropical mon-

tane forests. In: Churchill SP, Balslev H, Forero E and Luteyn JL (eds) Biodiversity and Con-servation of Neotropical Montane Forests, pp 359–363. Bronx, The New York Botanical Garden,New York

Moran RC and Riba R (eds) Psilotaceae a Salviniaceae. Flora Mesoamericana, Vol. 1. Universidad Na-cional Autónoma de México, Mexico City

1494

Moreno SJC, Castro PI, Humphries CJ and Williams PH (1996) Strengthening the national and natural parksystem of Iberia to conserve pteridophytes. In: Camus JM, Gibby M and Johns RJ (eds) Pteridology inPerspective, pp 101–124. Royal Botanic Gardens, Kew, UK

Øllgaard B and Navarette H (1997) Pteridophyte species richness in the valleys of Río Oyacachi, RíoQuijos, and upper Río Aguarico. In: Oyacachi. People and Biodiversity, pp 64–68. Kalø, DIVA TechnicalReport no. 2, Denmark

Øllgaard B, Ståhl B and Navarette H (1998) Plant diversity and endemism. In: Borgtoft H, Skov F, FjeldsåJ, Schjellerup I and Øllgaard B (eds) People and Biodiversity. Two Case Studies from the AndeanFoothills of Ecuador, pp 143–158. Kalø, DIVA Technical Report no. 3, Denmark

Pareja J, Vargas C, Suárez R, Ballón R, Carrasco R and Villarroel C (1978) Mapa geológico de Bolivia.YPFB – Servicio Geológico de Bolivia, La Paz

Parker TA and Bailey B (eds) (1991) A biological assessment of the Alto Madidi region and adjacent areasof Northwest Bolivia. RAP Working Papers 1. Conservation International, Washington, DC

Parker TA, Gentry AH, Foster RB, Emmons LH and Remsen JV Jr (1993) The lowland dry forests ofSanta Cruz, Bolivia: a global conservation priority. RAP Working Papers 4. Conservation International,Washington, DC

Parris BS (1985) Ecological aspects of the distribution and speciation in old world tropical ferns.Proceedings of the Royal Society of Edinburgh 86B: 341–346

Parris BS (1993) The phytogeography of West Malesian ferns. Fragmenta Floristica et Geobotanica Suppl.2: 435–455

Parris BS (1997) The ecology and phytogeography of Mount Kinabalu pteridophytes. Sandakania 9: 89–102

Parris BS, Beaman RS and Beaman JH (1992) The Plants of Mount Kinabalu I. Ferns and Fern Allies.Royal Botanic Gardens, Kew, UK

Perry A, Kessler M and Helme N (1997) Birds of the central Río Tuichi valley, with emphasis on dryforests, Parque Nacional Madidi, depto. La Paz, Bolivia. In: Remsen JV Jr (ed) Studies in NeotropicalOrnithology Honoring Ted Parker. Ornithological Monographs 48, pp 557–576. Washington, DC

Phillips OL, Hall P, Gentry AH, Sawyer SA and Vásquez R (1994) Dynamics and species richness oftropical rain forests. Proceedings of the National Academy of Sciences of the United States of America91: 2805–2809

Planet WG (1988) Data extraction and calibration of TIROS-N-NOAA radiometers. NOAA Tech. Mem.NESS 101-Rev. 1. US Dept of Commerce – Nat. Oceanic and Atmos. Adm, Washington, DC

Poulsen AD and Nielsen IH (1995) How many ferns are there in one hectare of tropical rain forest?American Fern Journal 1: 29–35

Prendergast JR, Quinn RM, Lawton JH, Eversham BC and Gibbons DW (1993) Rare species: the coinci-dence of diversity hotspots and conservation strategies. Nature 365: 335–337

Proctor GR (1985) Ferns of Jamaica. British Museum (Natural History), LondonRahbek C (1997) The relationship among area, elevation, and regional species richness in neotropical

birds. American Naturalist 149: 875–902Rosenzweig ML (1995) Species Diversity in Space and Time. Cambridge University Press, New YorkSaldias PM (1991) Inventario de árboles en el bosque alto del Jardín Botánico de Santa Cruz, Bolivia.

Ecología en Bolivia 17: 31–46Schimper AFW (1888) Die epiphytische Vegetation Amerikas. Bot. Mitt. Tropen II. G. Fischer, JenaSeidel R (1995) Inventario de los árboles en tres parcelas de bosque primario en la Serranía de Marimonos,

Alto Beni. Ecología en Bolivia 25: 1–35Sevink J (1984) An altitudinal sequence of soils in the Sierra Nevada de Santa Marta. Studies on Tropical

Andean Ecosystems 2: 131–138Shmida A and Ellner S (1984) Coexistence of plant species with similar niches. Vegetatio 58: 29–55Smith AR (1995) Pteridophytes. In: Steyermark JA, Berry PE and Holst BK (eds) Flora of the Venezuelan

Guayana, Vol 2, pp 1–334. Missouri Botanical Garden, St. LouisSmith AR, Kessler M and Gonzales J (1999) New records of pteridophytes from Bolivia. American Fern

Journal 89: 244–266Smith DN and Killeen TJ (1998) A comparison of the structure and composition of montane and low-

land tropical forest in the Serrania Pilón Lajas, Beni, Bolivia. In: Dallmeier F and Comiskey JA (eds)

1495

Forest Biodiversity in North, Central and South America, and the Caribbean. Research and Monitoring,pp 681–700. UNESCO, Paris, and Parthenon Publishing Group, New York

Sokal RR and Rohlf FJ (1995) Biometry. Freeman, New YorkSYSTAT (1997) SYSTAT for Windows, statistics, version 7.0. SPSS Inc, ChicagoTryon RM and Stolze RG (1994) Pteridophyta of Peru. Part VI. 22: Marsileaceae-28: Isoetaceae. Fieldiana

Botany, New Series 34: 1–123Tryon RM and Tryon AF (1982) Ferns and Allied Plants, with Special Reference to Tropical America.

Springer-Verlag, New YorkTucker CGF, Townshend JRG and Goff TR (1985) African land-cover classification using satellite data.

Science 227: 369–375Tuomisto H and Poulsen AD (1996) Influence of edaphic specialization on pteridophyte distribution in

neotropical rain forests. Journal of Biogeography 23: 283–293Tuomisto H and Ruokolainen K (1994) Distribution of Pteridophyta and Melastomataceae along an eda-

phic gradient in an Amazonian rain forest. J. Veget. Sci. 4: 25–34Tuomisto H, Poulsen AD and Moran RC (1998) Edaphic distribution of some species of the fern Genus

Adiantum in Western Amazonia. Biotropica 30: 392–399Tuomisto H, Ruokolainen K, Kalliola R, Linna A, Danjoy W and Rodriguez Z (1995) Dissecting Amazo-

nian biodiversity. Science 269: 63–66van der Werff H (1990) Pteridophytes as indicators of vegetation types in the Galapagos Archipelago.

Monographs of Systematic Botany of the Missouri Botanical Garden 32: 79–92van der Werff H (1992) Substrate preference of Lauraceae and ferns in the Iquitos area, Peru. Candollea

47: 11–20Walther BA and Morand S (1998) Comparative performance of species richness estimation methods.

Parasitology 116: 395–405Whitmore TC, Peralta R and Brown K (1985) Total species count in a Costa Rican rain forest. J. Trop.

Ecol. 1: 375–378Whittaker RH (1972) Evolution and measurement of species diversity. Taxon 21: 213–251Wolf JHD (1993) Diversity patterns and biomass of epiphytic bryophytes and lichens along an altitudinal

gradient in the northern Andes. Annals of the Missouri Botanical Garden 80: 928–960Young KR and León B (1989) Pteridophyte species diversity in the Central Peruvian Amazon: importance

of edaphic specialization. Brittonia 41: 388–395