Diversity and Distributions, (Diversity Distrib.

) (2005) 11, 57–72

The ones we left behind: Comparing plot

Blackwell Publishing, Ltd.

BIODIVERSITY
RESEARCH
sampling and floristic habitat sampling
for estimating bryophyte diversity
Steven G. Newmaster1, René J. Belland2, André Arsenault3, Dale H. Vitt4 and
Tara R. Stephens1

1
Botany Department, University of Guelph,
Guelph Ontario N1G 2W1, Canada, 2Devonian
Botanic Garden, University of Alberta,
Edmonton, Alberta T6G 2E1, Canada,
3
Southern Interior Forest Region, BC Forest
Service, 515 Columbia Street, Kamloops, British
Columbia V2C 2T7, Canada and 4Southern
Illinois University, Department of Plant Biology,
Mailcode: 6509, Carbondale IL 62901, USA
*Correspondence: Steven G. Newmaster, Botany
Department, University of Guelph, Guelph
Ontario, Canada N1G 2W1, E-mail:
snewmast@uoguelph.ca.
ABSTRACT
An efficient method for estimating bryophyte diversity in forest stands must con-
sider more than just the dominant forest mesohabitat. We compared two methodo-
logies commonly used for estimating diversity in forest ecosystems. Floristic habitat
sampling (FHS) utilizes stratification of all forest mesohabitats, which includes the
natural diversity of microhabitats found within and stratifies a mosaic of mesohabitats
(e.g. forest, streams, seeps, and cliffs) and microhabitats (e.g. rocks logs, etc.) that are
often not considered in forest research projects that use plot sampling to estimate
species diversity. In Canadian cedar hemlock forest, FHS methodology recorded
more than twice as many bryophyte species as plot sampling (PS). A comparison of
the dominant forest mesohabitat concluded that plot sampling was not as efficient as
FHS in estimating bryophyte diversity and that plot sampling can result in different
interpretations of species diversity. Rare species ordination of stands sampled using
FHS showed strong clustering of sites with respect to biogeoclimatic zones and age since
the last major disturbance (fire or logging) as compared with rare species ordinations
from PS data, which showed no delineation of stands along temporal gradients. Plot
sampling has many useful applications in ecology, but floristic habitat sampling is
more efficient for quantifying overall bryophyte diversity. FHS provides an excellent
way to record a comprehensive list of species.
Keywords
Biodiversity, Canada, cryptogams, sampling protocols, species richness, temperate
rainforest.

INTRODUCTION
An appropriate sampling methodology is crucial to understand-
ing patterns of community and taxon diversity on the landscape.
In the early 20th century, Clements (1905) described methods
for collecting plant species data using plots. Since that time many
variations of quantitative measurements using plots have been
used (e.g. Cain & Castro, 1959; Cochran, 1977; Bonham, 1989;
Krebs, 1989). The bounded nature of plots in relation to a spe-
cific sample area allows for quantitative sampling of species
abundance and frequency, and later statistical analysis. The use-
fulness of these two measurements have made plot sampling
a successful method for studying population and community
dynamics in bryophytes and many other groups of plants (e.g.
Slack, 1977; Söderström, 1988a, 1988b; Bonham, 1989; Lesica
et al., 1991; McCune, 1993, 1997; Frego & Carleton, 1995; Gignac
et al., 1998; Morneault et al., 1998; Rambo & Muir, 1998b).
Traditionally, randomly placed bounded plots have been used
to sample communities for species diversity. Clements (1905)
described the methods for collecting data using plots and quad-
rats. These methods have been used very successfully in many
population and community dynamic studies (Bonham, 1989).
Researchers that use this sampling method have acknowledged
its limitations in acquiring a comprehensive sampling of habitats
and species within an ecosystem (e.g. Shafti & Yarranton, 1972;
Pike et al., 1975; Gustafsson & Hallingbäck, 1988; Økland et al.,
1990; Johnston & Elliot, 1996; Bell & Newmaster, 1998; Rambo &
Muir, 1998a; Newmaster et al., 2003). Slack (1984) recognized
that a completely random plot sampling method is likely to miss
important types of variation within the sampling area unless the
intensity of the sampling (i.e. number of plots) is very high.
However, it is not clear if simply increasing the number of plots
(sampling area) can efficiently sample species diversity (Økland
et al., 1990; McCune & Lesica, 1992). Other studies have shown
that intensive plot sampling in mature forest and areas disturbed
by silvicultural operations do not record uncommon or distinctive
habitats that offer considerable bryophyte diversity (Newmaster
& Bell, 2002). Furthermore, intensive sampling requires a

© 2005 Blackwell Publishing Ltd www.blackwellpublishing.com/ddi 57

S. G. Newmaster et al.
substantial time commitment that is often not possible given the
reality of most research budgets. A popular solution to this
problem is to use ‘stratified random sampling’, which samples the
variation within a plot using ‘strata’ (Cochran, 1977; Krebs, 1989).
Although some researchers have investigated the influence of
stratifying at different sampling scales on phanerogams (Zamfir
et al., 1999) there has been little research on this topic regarding
bryophytes (Økland, 1994).
One alternative to random quadrats is the relevé sampling
method (Braun-Blanquet, 1932) used for many years in Europe
to classify vegetation. Henry Conrad (1935) introduced this
method to North American ecologists. The relevé method never
became popular in North America owing to its subjective,
semiquantitative nature (Barbour et al., 1987). However, many
researchers have used this to successfully study phytosociology
and community structure along environmental gradients in
vascular (Cain & Castro, 1959; Wolf, 1994) and nonvascular
plants (Wolf, 1993a; 1993b, 1993c; Lücking, 1999b). This
method has been used in cryptogamic research to study biogeo-
graphy (Kürschner & Parolly, 1998a; Lücking, 1999a), biodiversity
(Lücking, 1997; Komposch & Hafellner, 1998), ecomorphology
(Kürschner & Wolfgang, 1999) and life form strategies (Kürschner
& Parolly, 1998b; Kürschner, 2003). Researchers studying epi-
phytic diversity in tropical rain forests have focused this method
at finer scales on tree microhabitats (i.e. leaves, fine branches,
etc.) (Gradstein et al., 1996).
Several researchers have shown that bryophyte diversity
is influenced by a multitude of ecological variables including
microhabitat, stand age, disturbance, and available moisture
(Vitt et al., 1975; Pøcs, 1980; Söderström, 1989; Herben et al.,
1991; Gradstein, 1992; Belland & Schofield, 1994). It is difficult
to separate how each of these factors influence or contribute to
patterns of diversity (Kimmerer & Allen, 1982; Leonard et al.,
1985; During & ter Horst, 1987; During et al., 1987; Carleton,
1990; Herben & Söderstrom, 1992; Økland, 1994). Ultimately, it
is desirable to quantify environmental variables in a statistical
model to predict species richness in undisturbed and disturbed
ecosystems. This requires a sampling methodology that includes
the majority of habitats in the study area and their respective
environmental parameters.
Recently Vitt & Belland (1997) presented a framework for
understanding the patterning of bryophyte habitats on the land-
scape. Habitats can be defined as either mesohabitats or micro-
habitats. Mesohabitats are localized physiographic (e.g. streams,
seeps, cliffs) or physiognomic (e.g. forests) features. In a forested
landscape mesohabitats are arranged into a mosaic of dominant
mesohabitats (e.g. forests) in which restricted mesohabitats (e.g.
streams, seeps, cliffs) exist (Vitt & Belland, 1997). Microhabitats
(e.g. logs, rocks, stumps) are the smallest landscape units and
may be unique to one type of mesohabitat (e.g. crevices on cliffs).
As bryophytes are closely linked to their habitats (Schuster,
1949; Slack, 1977; Söderström, 1988a), it is essential to consider
the patterning of these habitats on the landscape. The import-
ance of habitats in the patterning of bryophyte diversity has been
documented in several studies (Slack, 1976, 1984; Söderström,
1988b; Gignac & Vitt, 1994; Belland & Vitt, 1995; Vitt et al., 1995;
58
Newmaster et al., 2003). Vitt et al. (1995) used stratified plot
sampling to demonstrate that bryophyte diversity in wetlands is
most strongly correlated with habitat heterogeneity. Vitt & Belland
(1997) have shown that rare species diversity is in part a function
of the types and distributions of all mesohabitats and the
number of microhabitats. They proposed that microhabitat
heterogeneity might define the quality of the individual meso-
habitats (Vitt & Belland, 1997). This has been shown quantitat-
ively for peatlands (Belland & Vitt, 1995; Vitt et al., 1995) and
temperate/boreal forests in Canada (Newmaster et al., 2003). A
quantitative analyses of 287 forests (Newmaster et al., 2003)
revealed that patterns of bryophyte diversity are clearly defined
by both mesohabitat quality and quantity. Mesohabitat quality
was not only dependent on microhabitat quantity, but also the
quality or types of microhabitats found within a particular meso-
habitat (Newmaster et al., 2003). Therefore, it is appropriate to
use a sampling methodology that focuses on microhabitats as the
sampling unit.
Sampling methods aimed at assessing total bryophyte diversity
studies should include all of the potential habitats in an ecosystem,
and incorporate the elements of a floristic inventory (Newmaster
& Bell, 2002; Newmaster et al., 2003). Belland’s (Belland & Brassard,
1988; Belland & Schofield, 1994) method of sampling and
analysis is, in part, derived from Bouchard et al. (1978) and uses
habitats as the basic sampling units. The method has been used
successfully to document bryophyte diversity in National Parks
of eastern Canada (Belland & Brassard, 1988; Belland, 1989,
1995; Belland & Schofield, 1994; Belland & Vitt, 1995). Vitt
(1991) used a similar methodology in a study of south Pacific
mosses.
This method is referred to as floristic habitat sampling (FHS),
a term reflecting its roots in the floristic tradition and the use of
habitats as sampling units or strata (Newmaster, 2000). FHS has
been used by Newmaster et al. (2003) to reveal patterning of
bryophyte diversity in both temperate and boreal forests. This
study used FHS to identify bio-indicators for conservation of
bryophyte diversity in Canada’s old Growth Rain Forests.
Recently, Heinlen & Vitt (2003) have also used FHS to develop an
emerging coarse filter approach to rare moss conservation. These
researchers chose the FHS method because they believe it is effi-
cient at sampling bryophyte diversity and captures higher species
richness per sampling unit. Although, FHS and PS are both
accepted methodologies for studying forest diversity, there are no
published reports that compare the efficiency of FHS with that of
the more commonly used PS method.
The purpose of our research is a simple comparison of two
sample protocols (FHS and PS) that are used to estimate divers-
ity. The two protocols take inherently different approaches to
accomplishing the same goal. This paper analyses patterns of
diversity to compare these two protocols. More specifically, this
paper addresses the questions: (1) Which is the more efficient
sampling method for bryophyte diversity in forest stands, PS or
FHS? (2) If PS is intensified (more sampling area), is it as efficient
as FHS for sampling bryophyte diversity? and (3) Are patterns of
bryophyte diversity in stands similar using PS or FHS sampling
methodology?
Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd

Figure 1 Temperate rain forest research site
locations within the Coastal Western Hemlock
(CWH) and Interior Cedar Hemlock (ICH)
biogeoclimatic zones in British Columbia,
Canada (
= CWH sites, 49 –50° N and
123 –126° W;  = ICH sites, 50–53° N and
199 –120° W).
STUDY AREA
Sampling was conducted in British Columbia, Canada, within
two distinct biogeoclimatic zones; the Coastal Western Hemlock
zone (CWH) and Interior Cedar Hemlock zone (ICH — Mei-
dinger & Pojar, 1991). The CWH is located on the westerly edge
of the Coast Mountains and is also known as Canada’s coastal
temperate rainforest (Fig. 1). The ICH is located on lower slopes
of the Columbia Mountains, windward side of the continental
divide along the Rocky Mountains, including much of the
Shuswap and Quesnel highlands in BC’s interior and just east of
the Coast Mountains in Northern BC (Fig. 1). The wetter por-
tions of the ICH (wk1 and vk1 variants) are inland rainforests, a
phenomenon unique to Canada (Arsenault & Goward, 2000).
Detailed descriptions of glacial history, climate and floristics can
be found in Schofield (1988), Arsenault (1995), Hebda (1995),
Schoonmaker et al. (1997), Newmaster (2000) and Newmaster
et al. (2003).
METHODS
In this study, a stand follows the definition of Barbour et al.
(1987) and Kimmins (1987) as a standing growth of trees with
similar physiognomy. Stands defined this way show distinctive
dominant tree species, age, structure, elevation, slope position,
and aspect. Although they vary in size, most consist of a dominant
mesohabitat (the forest), which encloses numerous restricted
mesohabitats (e.g. cliffs, streams, seeps). Within each mesohab-
itat there are a number of microhabitats (e.g. tree base, stumps,
acidic rocks), some which may be specific to one type of meso-
habitat (e.g. wet cliff crevices, submerged rocks in streams).
Floristic habitat sampling (FHS) and plot sampling (PS) were
used to assess patterns of diversity in cedar-hemlock stands over
two field seasons. Comparisons of these sampling methods are
fair because they sample the same stands and the exact same
sample size. In 1996, 102 stands were sampled in the interior
cedar-hemlock (ICH) biogeoclimatic zone located within the
Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd
Estimating bryophyte diversity
Wells Gray, upper Adams River, and Seymour watersheds. Within
these watersheds, sampling was evenly distributed between stands
that were burned approximately 80–160 years ago, and old growth
stands of 250+ years in age. In 1997, 185 stands were sampled in
the coastal western hemlock biogeoclimatic zone (CWHvm1)
within the Capilano and Seymour watersheds along the mainland
coast, and in the Sidney, Clayoquot, Tofino, and Walbran water-
sheds along the western coast of Vancouver Island. Extensive log-
ging activities in the Capilano and Seymour watersheds allowed
balanced sampling among stands that were logged 80 years ago,
and old growth stands (> 250 years). Sampling on Vancouver
Island was limited to older stands owing to the relatively recent
logging activity and lack of fire disturbance.
Stands were randomly chosen for species and environmental
data sampling and stratification was applied at two scales:
between stands and within stands. Biogeoclimatic zone and dis-
turbance (stand age) were used to stratify between stands. Slope
position (toe, low, mid, upper) was used to stratify within stands.
Random points (RP) were located within each stand and were
used as starting points for both PS and FHS sampling. Environ-
mental variables (Newmaster et al., 2003) were collected at each
RP and reflect environmental processes at the stand scale.
Voucher collections (15,000 in total) were made at each watershed
for common species and where possible, within each stand for
rare species (occurring in less than 15% of stands). Species
nomenclature follows Anderson et al. (1990) for mosses and
Stotler & Crandall-Stotler (1977) for hepatics. Voucher specimens
are deposited in the University of Alberta Cryptogamic Herbarium
(ALTA), Southern Interior Forest Region Herbarium, and Uni-
versity of British Columbia Herbarium (UBC).
Floristic habitat sampling
This study expands on the sampling methodology used by
Belland & Brassard (1988), Belland (1989, 1995), Vitt (1991),
Belland & Schofield (1994) and Belland & Vitt (1995). It incorp-
orates floristic sampling with a rigorous method that ensures
59
S. G. Newmaster et al.
that a forest stand and its representative meso/microhabitats are
efficiently and fully sampled. FHS is similar to a floristic survey as
it can provide a method that records the presence of all species
within a study area without relevés or quadrats. Sampling is
stratified within forest stands by mesohabitat units, each with a
suite of microhabitats that are the main sampling unit.
FHS differs from plot sampling in several respects: (1) PS usu-
ally restricts sampling to dominant mesohabitats; FHS samples
the full diversity of mesohabitats and microhabitats; (2) PS is
bounded by a relatively small, predefined, sampling area; FHS is
bounded by the limits of the stand or sampling time; (3) PS prov-
ides a list of species within a fixed, random bounded space with
quantified abundance measures useful for experimental designs
that test variation in treatments; and (4) FHS results in a nearly
complete list of species and habitat characteristics since it is flo-
ristically based and emphasizes species presence/absence.
In FHS, the sampling area is systematically divided into domin-
ant and each type of restricted mesohabitats. Typically, in FHS,
sampling within any type of randomly sampled mesohabitat is
accomplished by systematically walking a randomly placed grid
of transects and sampling the different types of microhabitats
until no new species are found. However, in this study, mesohab-
itat sampling was conducted in successively larger areas (quad-
rats) to compare the efficiency of variously sized bounded plots
nested within the FHS methodology. These plots started at one
square metre, and increased in size over nine steps (i.e. 1 m2,
5 m2, 25 m2, 100 m2, 250 m2, 500 m2, 1000 m2, 2500 m2, 5000 m2).
The maximum area sampled was 5000 m2, which was determined
to be adequate in all 287 stands as assessed by species area curves.
Plot size was quantitatively related to species richness and a
proportional frequency index (Brillouin): 0 –20%, 20 – 50% and
50–100%.
Abundance was recorded for each species and microhabitat.
The abundance categories follow that of Vitt et al. (1995): 1 = one
to few occurrences, 0–20% cover; 2 = several occurrences to
frequent in one or some areas of the microhabitat, 20–50%
cover; 3 = frequent throughout the microhabitat, 50–100%
cover. Abundance was averaged for each species, within respect-
ive types of mesohabitats and for each stand.
The FHS sampling methodology used in this study can be
defined as follows:
Dominant Forest Mesohabitat — (1) A list of all microhabitats
was generated for the first quadrat (located at the RP); (2) within
the first quadrat, each species and its respective abundance category
was recorded for each type of microhabitat; (3) Microhabitat
sampling continued in the following quadrats (9 in total) of
increasing size (maximum area 5000 m2), or until maximum
richness is achieved (no new species are found).
Restricted Mesohabitat — Each type of restricted mesohabitats
(i.e. stream, cliff, seep), was systematically sampled following
the three steps listed above (see dominant forest mesohabitat).
All restricted mesohabitats were sampled within a 1-km
radius of the RP. Three types of restricted mesohabitats were
sampled:
1 Streams — Streams are defined as a stream gully containing
the stream itself and 5 m of bank (2.5 m on either side). They are
60
a complex mesohabitat and contain microhabitats that are also
common to seeps, cliffs, and the dominant forest mesohabitat
(Table 1). The stream banks were included because they offer a
complex mix of microhabitats that contain considerable diversity
not found elsewhere in cedar hemlock forests. Sampling started
within the actual streambed (i.e. 1–5 m) and continued to
include 5 m of bank. Sampling continued along the stream
(including the 5 m of bank) for 1000 m for a minimum sampling
area of 5000 m2.
2 Cliffs — Cliffs are defined as large (> 100 m2) rock faces or
outcrops. Like streams they are a complex mesohabitats and
may include trees, logs, stumps, and seepages, but may also
contain many microhabitats found in the dominant forest
mesohabitat (Table 1). Sampling was limited to a maximum area
of 5000 m2.
3 Seeps — Seeps are swampy areas of cedar hemlock forest with
poor drainage. Seeps have many of the same microhabitats as the
dominant forest mesohabitat. Only seeps larger than 100 m2 were
considered for sampling. Sampling was limited to a maximum
area of 5000 m2.
Plot sampling
The plot sampling method used follows standard forestry
protocols suggested for sampling species richness and abundance
for use in community ecology studies including those used in
biodiversity research (Krebs, 1989). PS was conducted within a
20 m-diameter circular sampling plot (314 m2), located at the RP
within each stand. All species were recorded with their abund-
ance (ocular estimate of percentage cover within quarters of the
20 m-diameter circular plot). Species area curves were generated
to confirm the number of plots needed to properly sample a
watershed or biogeoclimatic zone. These sampling curves lev-
elled off with 10–15 sampling plots per watershed or 30– 40 sam-
pling plots per biogeoclimatic zone. Therefore, we sampled 20
plots per watershed and more than 100 plots per biogeoclimatic
zone. We should emphasize that this methodology is a standard
protocol for sampling forest community diversity. In addition to
this standard sampling protocol, we sampled within all types of
mesohabitats using randomly placed plots. In order to allow a
fair comparison to FHS we increased the sampling intensity until
the area (5000 m2) sampled per mesohabitat type equalled the
area sampled using the FHS methodology.
Analysis
Diversity analyses
Bryophyte diversity was analysed at several scales following the
structure and terminology proposed by Whittaker (1972, 1977)
(Appendix s1).  (2 Factor Model III — stand age — random
and sampling method — fixed) was used to assess species richness
at different stand ages for each sampling method. Epsilon divers-
ity is the total species richness for cedar hemlock forest sampled
in this study. Gamma diversity is a measure of species richness
in watersheds, biogeoclimatic zones and variants. Alpha diversity
Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd
 Estimating bryophyte diversity

Table 1 A list of microhabitats for each type of meso-habitat within cedar hemlock stands (DMH = dominant mesohabitats; RMH = restricted
mesohabitat)

DMH RMH

Microhabitat Forest Cliff Stream Seep

Coniferous tree species x x x x

Deciduous tree species x x x x
Size of tree (10–25 cm d.b.h., 30 – 60 cm d.b.h., > 70 cm d.b.h.) x x x x
Position on tree (trunk or base — 50 cm above tapered bowl) x x x x
Snag (dead coniferous or deciduous trees) x x x x
Twig (CWD < 10 cm diam.) x x x x
Log size (10–30 cm, 30– 60 cm d.b.h., > 70 cm d.b.h.) x x x x
Log decay class (D1, D3 or D5 — CWD codes) x x x x
Organic soils (LFH) x x x x
Mineral soil (sand, silt, loam, clay) x x x x
Moist depression (small isolated pools of water/mud) x x x
Intermittent stream (narrow and ephemeral) x x x
Rock (sample type, pH) x x x x
Tree stump x x x x
Upturned tree roots (‘tip-up’) x x x x
Adjacent bank (sand, silt, clay, loam, gravel, cobble, rock) x x x x
Submerged habitat (rocks or logs) x x
Shallow bars (sand, silt, clay, loam, gravel or cobble; dry/wet) x
Waterfall (< 1 m, 1–2 m, 2– 5 m, 5 –10 m, > 10 m) x
Depth (< 10 cm, 10– 30 cm, > 30 cm) x
Rapid (flow rate (ms−1)) x
Crevice (horizontal/vertical; < 5 cm, 0.5 –1 m, > 1 m; wet/dry; seepage, soil cover, sand, silt, clay, loam) x x
Ledges (size, wet /dry, seepage, soil covered) x x
Caves (size, wet/dry, seepage) x x
Vertical rock face (size, wet/ dry, seepage) x x
Talus x x
Rock surface (rough/ smooth) x x

is the number of species occurring in our sampling units (i.e.
stands, mesohabitats or microhabitats). Diversity (inventory) was
calculated for each of the following: British Columbia’s cedar
hemlock forest (epsilon diversity), biogeoclimatic zones, variants,
watersheds and mesohabitats (gamma diversity), and stands
(alpha diversity). All statistics were carried out using SPSS for
Windows 10 (SPSS Inc., Chicago, IL, USA). An Index based on
species and proportional frequencies (Brillouin index [eqn 1],
Pielou, 1966; Peet, 1974; Clifford & Stephenson, 1975) were
calculated for biogeoclimatic zones, variants, and watersheds
for stands stratified by disturbance (logging, fire, old growth).
Indices were calculated using Krebs/WIN (Krebs, 1997).
ln N ! − ∑ ln ni !
HB =
N
where, HB = Brillouin index, n i = Number of individuals
in each species, N = Number of individuals in entire
collection.
Accumulation (rarefaction) curves were generated to assess
species richness were calculated from 287 stands for both FHS
and PS sampling methods (Colwell, 1997). Estimates of the total
number of species within the bryophyte community for both
FHS and PS were made using the Chao-2 index (Chao, 1984;
Colwell & Coddington, 1994). These estimate were also calculated
for the rare (occurring in less than 15% of stands) and common
flora for both FHS and PS data. Estimates of richness means
calculated based on 100 randomizations of sample pooling order
including the corresponding standard deviations (error bars in
figures). Both rarefaction and Chao-2 were calculated in EstimateS
(Colwell, 1997).
The analyses of the patterns of diversity are presented at a
regional scale (both ICH and CWH) using relationships between
stands, species, and environmental variables. The data sets are
based on similar sample sizes within all types of mesohabitats at
the stand scale, using both PS and FHS sampling methodologies.
(1) Canonical Correspondence Analysis (CCA; ter Braak, 1998) was
used to produce ordinations of 287 stands sampled in this study,
which were constrained by 22 environmental variables (Newmaster
et al., 2003). The following four ordinations were constructed
based on common and rare species floras (rare occurence in less
than 15% of all stands): (1) PS common species, (2) PS rare spe-
cies, (3) FHS common species and (4) FHS rare species. Details
of the environmental variable measurements can be found in
Newmaster (2003).

Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd 61

S. G. Newmaster et al.

diversity dramatically from 196 species to 417 species. Mesohabitats

RESULTS
What is the most efficient sampling method,
FHS or PS?
Epsilon and gamma diversity
Epsilon and gamma diversity estimates from plot sampling
(20 m-diameter circular plots) were compared with those from
the dominant forest mesohabitat (using FHS) to allow compar-
isons of bryophyte diversity within only the forest mesohabitat
(one type of mesohabitat). There was a significant difference
(P = 0.002; n = 287) in these diversity estimates when using the
two different types of sampling units. Epsilon was 196 species
using PS and 296 species using the FHS unit (i.e. the entire dom-
inant forest mesohabitat as a sampling unit). Gamma diversity
estimated using FHS (ICH–forest 222, stream 262, seep 111,
cliff 18; CWH–forest 172, stream 293, seep 151, cliff 163) was
between 33% and 44% higher than gamma diversity estimated
from the PS data (ICH–forest 141; CWH forest 114).
The estimates for epsilon and gamma diversity are more accur-
ate when all the mesohabitats are considered in FHS. Epsilon
diversity for cedar hemlock forests using all mesohabitats increases
offer considerable diversity (forest 262 spp.; stream 359 spp.; seep
207 spp.; cliff 237 spp.) and contain unique assemblages of
species (forest 13 spp.; stream 70 spp.; seep 2 spp.; cliff 2 spp.).
Floristic habitat sampling gamma diversity in biogeoclimatic
zones and variants is higher (40–65%) than estimates from PS
for either young or old forest (Table 2). Old forests are richer
than younger forest using either sampling method. Using FHS,
the CWH is slightly richer than the ICH and wetter variants (wk1
and vk1) are richer than dryer variants (mw3). Conversely, PS
reveals that the ICH is slightly richer than the CWH, and variants
have similar richness. Watersheds show a range of diversity using
FHS. The highest diversity is in the Seymour (ICH), Walbran
(CWH) or Sidney watersheds (CWH). Richness estimates from
PS are very similar for all the watersheds (Table 2).
Alpha diversity
Alpha diversity (mean species richness of stands) is considerably
higher using FHS than PS. Considering only the dominant
forest mesohabitat, FHS is significantly (0.01 > P > 0.001; n = 103
ICH; n = 184 CWH) higher when compared to PS. In the ICH,
mean alpha diversity in the dominant forest mesohabitat is 32

Table 2 Landscape gamma diversity for stands with different disturbance histories using both FHS and PS sampling [abundance values are
relative, i.e. sum of abundance categories/number of plots; Plan = Plantation/logging disturbance; Brillouin index is based on species and
proportional frequencies]

Spp. richness Abundance Brillouin index

Landscape Elements Hist. FHS PS FHS PS FHS PS

Biogeo-climatic ICH Fire 188 123 269 203 6.4 5.8

Old 300 141 374 251 6.9 6.1
Zones CWH Plan. 114 99 120 103 5.5 5.2
Old 317 134 621 286 6.9 5.6
Biogeo-climatic ICHmw3 Fire 171 95 229 149 6.2 5.4
ICHmw3 Old 235 93 309 146 6.6 5.4
Variants ICHwk1 Fire 162 96 218 162 6.1 5.5
ICHwk1 Old 276 109 348 196 6.8 5.7
ICHvk1 Fire 120 62 162 102 5.7 4.8
ICHvk1 Old 266 94 342 167 6.8 5.5
ICH watersheds WG Fire 125 53 171 85 5.8 4.6
Old 218 77 291 111 6.5 5.1
Azure Fire 138 75 184 124 5.9 5.2
Old 242 80 314 141 6.6 5.3
Adams Fire 151 89 200 139 5.9 5.3
Old 265 81 318 137 6.7 5.3
Seymour Fire 165 94 225 155 6.1 5.5
Old 276 116 335 204 6.6 5.8
CWH watersheds Capilano Plan. 108 89 118 97 5.4 5.1
Old 230 108 151 130 6.5 5.5
Seymour Plan. 87 42 113 91 5.4 5.1
Old 111 87 219 117 6.3 5.3
Tofino Old 213 104 265 146 6.3 5.5
Claquot Old 231 107 281 148 6.5 5.5
Sidney Old 286 106 338 147 6.8 5.5
Walbran Old 288 113 340 155 6.8 5.6

62 Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd


Figure 2 Alpha diversity of stands assessed using floristic habitat
sampling (FHS including all mesohabitats) and plot sampling (PS)
sampling. Cedar hemlock forests are divided into inland (ICH),
coastal mainland (CWH-ML), coastal oceanic (CWH-ISL), and by
age classes (class 4, young = 80 years and class 9, old > 250 years).
Error bars represent two standard errors on either side of the mean.
(± 5) species using PS and 65 (± 4) species using FHS. In the CWH,
mean alpha diversity in the dominant forest mesohabitat is 41 (± 6)
species using PS and 98 (± 8) species using FHS. When all mesohabi-
tats are considered, alpha diversity of stands in FHS is significantly
higher (0.01 > P > 0.001; n can be found in Fig. 2) than diversity
estimates from PS using data from all mesohabitats (Fig. 2).
Inherent differences in PS and FHS sampling methodologies
are apparent when estimating species richness from all stands
and mesohabitats using rarefaction or the nonparametric Chao-
2 index (a species estimation index; Chao, 1984). It is important
to note that we are now comparing similar sample sizes of all
mesohabitat types at the stand scale for each sampling meth-
odology and that an ‘accurate estimate’ (i.e. sampled most of
the bryophyte community) is identified by the plateau of the
Figure 3 Mesohabitat alpha diversity (species
richness) and Brillouin indices (based on
species and proportional frequencies) within
increasing sample size areas for 287 temperate
rainforest stands (SP = seep, CF = cliff,
FS = forest, ST = stream).
Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd
Estimating bryophyte diversity
accumulation/rarefaction curve. Surprisingly, both methodo-
logies attain accurate estimates for total richness (415 spp.) at
exactly 84 samples for Chao, and 214 samples for rarefaction.
The Chao index is the least biased estimate of species richness
at low sample sizes. In FHS, estimates of richness are significantly
higher (FHS = 417 spp.; ICH = 197 spp.; P < 0.0001; n = 287)
and much more accurate (less biased) than those in PS, which
underestimated alpha diversity. PS sampling effort required
268 days to estimate species richness at 197 species. This estimate
is largely underestimated and the FHS method achieved such an
estimate after only 13 days of sampling effort. However, FHS
sampling required also required a substantial sampling effort
(268 days) to achieve an accurate estimate of species richness
(417 species). Use of the Chao index with FHS can further reduce
sampling effort by more than two and half times (Chao 100 days;
rarefaction 268 days).
Diversity indices and abundance
Differences in diversity indices and species abundance support
the differences in gamma and alpha diversity for the different
sampling methods (Table 2). High indices are associated with
high species richness and frequency. Brillouin indices using FHS
data are greater than those calculated with PS data (Table 2). Spe-
cies abundance estimates are also substantially higher for FHS
data than PS data (Table 2).
Increasing plot size of bounded plots
Species richness and diversity indices for each sampling unit
increased with sample area (Fig. 3). The 20 m-diameter plot used
in the PS method sampled 314 m2 of forest mesohabitat resulting
in a mean species richness of 35 (± 5) species. Expanding sam-
pling area to 1000 m2 increased mean species richness by only 18
species. Furthermore, species richness steadily increases even
after 5000 m2 has been sampled, increasing mean species rich-
ness in the dominant forest mesohabitat to just over 80 (± 6)
species (Fig. 3). Using FHS, the mean species richness within the
dominant forest mesohabitat was 106 (± 9) species.
63
S. G. Newmaster et al.

Streams contain the most dramatic increase of richness and
higher diversity indices with increasing sampling area because of
the complex composition of microhabitats on the stream banks.
In this mesohabitat the first two areas (1–5 m2) contained low
diversity (< 20 species). However, the inclusion of the stream
bank in the sampling area (> 5 m2) increases mean species rich-
ness from 20 (± 6) to 50 (± 8) species after 100 m2 of the stream
and bank have been sampled (i.e. 20 m of stream including 5 m
of shore with each linear metre of the stream). Another large
jump in mean species richness (> 20 species) occurs as the sam-
pling area increases to 250 m2. Species richness increased steadily
until 5000 m2 of habitat was sampled (1000 m of stream gully —
Fig. 3). Plot sampling usually does not include sample areas of
this magnitude within one type of mesohabitat.
Seeps and forest mesohabitats appear to have a continuous
increase in richness with increasing sampling area. In cliffs, mean
species richness and diversity indices rise quickly until 250 m2
has been sampled; diversity tends to level off with increasing area
greater than 250 m2 (Fig. 3).
Patterns in diversity
The following analyses of the patterns of diversity are presented
at a regional scale (both ICH and CWH) using relationships
between stands, species, and environmental variables. The data
sets are based on similar sample sizes within all types of meso-
habitats at the stand scale, using both PS and FHS sampling
methodologies.
Patterns of alpha diversity following disturbance
Richness patterns are different when FHS and PS data are
stratified by stand disturbance. Old growth stands (> 250 years)
sampled using PS have mean species richness values that are sig-
nificantly (P = 0.041; n = 124) higher than mean species richness
in younger stands (80 years) in the coastal mainland rainforest —
CWH (Fig. 3). However, with PS old growth species richness in
the ICH is not significantly (P > 0.05; n = 103) different from
richness in young ICH stands. Floristic habitat sampling data
indicate that species richness in old growth stands is significantly
higher (0.01 > P > 0.001; n = 287) than in young stands in both
ICH and CWH forest (Fig. 3).
Patterns of rarity and commonality
FHS and PS result in different patterns of bryophyte commonal-
ity and rarity (defined as species occurring in less than 15% of
stands). In PS, 70 (34%) of the species are rare as compared with
270 (65%) rare species sampled using FSH. Consequently, the
percentage of common species sampled using PS (66% — 126
species) is higher as compared FHS (35% — 147 species). The rel-
ative number of common species is higher (66% > 35%) in PS,
but the absolute number of species is higher in FHS (147 > 126)
because of the higher numbers of species sampled using FHS.
Estimates of species richness for common and rare species
illuminate an attribute of commonality that may give a false
sense of complete species capture. The levelling off of a species
area curve represents complete species capture for a given area.
Regardless of the sampling methodology, both the rarefaction
and Chao estimates (FHS 147 spp.; PS 127 spp.) of the common
flora level off at small sample sizes (25 samples). Estimates of
richness for rare species never levels off for either sampling
method. FHS richness estimates are always higher than PS estim-
ates, particularly for rare species.
Patterns from gradient analysis
Species patterns along environmental gradients were compared
according to common or rare floras for each sampling meth-
odology. Canonical Correspondence Analysis (CCA) of the rare
species PS data and the common species FHS data (287 stands,
22 environmental variables) resulted in ordinations with over-
lapping stand groups representing old growth, and stands dis-
turbed by fire (ICH) or logging. Interset correlations and t-values
were not significant (0.01 > P > 0.001; n = 287) for any of the
environmental variables. It was not possible to define any groups
(i.e. young, old, ICH or CWH) from the ordination of stands
although the first three-axis account for some variation in the
species data, the interpretation of the ordination was not possible
(Table 3). The low species-environment correlations indicate

Table 3 Summary of canonical

Axis 1 2 3 4 correspondence analysis (CCA) of 287 stands
in British Columbia’s cedar hemlock forests
Eigenvalue R-PS 0.03 0.02 0.01 0.01 and 22 environmental variables, using plot
C-PS 0.33 0.06 0.04 0.02 sampling (PS) or floristic habitat sampling
R-FHS 0.71 0.46 0.23 0.19 (FHS) of rare (R) or common (C) species.
C-FHS 0.05 0.01 0.01 0.01 Shaded areas indicate important gradients on
Species/environment correlation R-PS 0.17 0.13 0.12 0.10 the 1st or 2nd axes
C-PS 0.98 0.93 0.85 0.77
R-FHS 0.99 0.92 0.88 0.84
C-FHS 0.70 0.35 0.46 0.43
Cumulative percentage variance of species data explained R-PS 11.9 12.6 14.6 18.1
C-PS 58.4 66.8 73.6 77.2
R-FHS 71.5 77.0 81.1 84.3
C-FHS 33.8 53.1 62.6 69.7

64 Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd


Figure 4 CCA ordination of common species
from the plot sampling (PS) data to explore the
relationships between 287 stands, 127 species
and 22 environmental variables. Abbreviations
are listed in Table 4.
that the environmental variables inadequately explain variation
in vegetation along an axis, which results in distortion of the
ordination.
Relationships between species, stands, and environmental
variables were interpretable in the ordination (CCA) of common
species PS data (Fig. 4; Table 3). High species/environment
correlations indicate a close correspondence between the species-
only and environmentally constrained ordination and that no
important environmental variables were omitted from the
analysis (Table 3). The first axis of the ordination explained
58.4% of the variation in the species data set (Table 3). Significant
(0.01 > P > 0.001; n = 287) interset correlations and t-values
were used to identify important environmental variables for axis
one (climatic variables) (Table 4). The two distinct groups on the
ordination indicate large differences in the composition of
species between the ICH (right side of axis 1) and CWH (left side
— Fig. 4). A Monte Carlo permutation test confirmed that the
first axis is statistically significant (0.01 > P > 0.001). The second
axis was not interpretable as supported by the very low eigenvalue
or a small portion of secondary gradient explained. Axes two
interset correlations and t-values were not significant (0.01 > P
> 0.001; n = 287) for any of the environmental variables.
Ordination of the rare species FHS data indicated that rare
species are crucial for the interpretation of environmental gradi-
ents (Fig. 5). The two distinct groups on the ordination indicate
large differences in the composition of species between the ICH
(right side of axis 1) and CWH (left side — Fig. 5). Furthermore,
young (bottom of ordination) and old (top) stands are separated
on the second axis (Fig. 5, Table 4). Species/environment corre-
lations were high for the rare species ordination (i.e. 0.99 axis 1;
0.92 axis 2) (Table 3). The first two axes of the ordination
explained 77.0% of the variation in the species data set (Table 3).
Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd
Estimating bryophyte diversity
Significant (0.01 > P > 0.001; n = 287) interset correlations and
t-values were used to identify important environmental variables
for axis one (climatic variables) and axis two (time since distur-
bance) (Table 5). A Monte Carlo permutation test confirmed
that the first and second axes are statistically significant (0.01 >
P > 0.001).
DISCUSSION
The species we leave behind or exclude because of sampling
technique are crucial to understanding patterning of bryophyte
diversity. The type of sampling used for estimating diversity
depends on the organism being studied, how closely that organ-
ism is associated with its substrate, and the nature of the
ecological question (Krebs, 1989). Bryophytes occur in close
association with their substrate or habitat (Vitt & Belland, 1997)
and their usefulness as indicators of habitat and environmental
change or sensitivity is well documented (Gignac, 1986;
Söderström, 1988a; Gignac & Vitt, 1994; Bell & Newmaster,
1998; Newmaster & Bell, 2002). In terrestrial ecosystems, moss
habitat limitations are often associated with substrate availability
and type (Shaw, 1981; Horton, 1988; Söderström, 1988a). In
peatlands, bryophyte species richness is closely related to micro-
habitat diversity (Vitt & Belland, 1995; Vitt et al., 1995). Vitt &
Belland (1997) proposed that rare species occurrence and divers-
ity depends on the quality and quantity of mesohabitats found
on the landscape. The patterning of bryophyte diversity is
intimately linked with habitat heterogeneity. To understand pat-
terning of bryophyte diversity we must incorporate sampling
techniques that focus on habitats as the sampling units.
This study shows that FHS is much more efficient at capturing
bryophyte richness than PS. Bryophyte diversity estimates
65
S. G. Newmaster et al.

Table 4 Plot sampling (common species only — see CCA ordination Fig. 4) statistics for variables used in canonical correspondence analysis
(CCA) of 287 stands in BC’s cedar hemlock forest. Asterisks indicate significance at P < 0.01. Absolute t-value > 2.1 are used to indicate
important canonical coefficients (ter Braak, 1998). Bold values indicate variables with significant correlation and canonical coefficients

Interset correlation Canonical coefficient t-value

Variable Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2

Site series (SS) − 0.74 − 0.19 − 0.04* − 0.01 − 2.55 − 0.46

Elevation (Elv) 0.55 0.60 − 0.08* − 0.02 − 3.09 − 0.64
Slope position (SP) 0.02 0.25 − 0.01 0.01 − 0.11 − 0.39
Aspect (As) 0.20 0.32 − 0.01 0.01 − 0.54 − 1.44
Moisture regime (Hyg) − 0.37 − 0.24 0.02 0.01 1.70 − 0.41
Rock cover (RC) 0.18 − 0.40 − 0.01 0.01 − 0.77 − 0.76
Rock acidity (RA) 0.75 − 0.09 − 0.02 0.03* − 1.83 2.26
Soil texture (ST) 0.01 0.28 0.01 0.01 0.65 0.04
Canopy height (CH) − 0.36 − 0.58 0.01 − 0.02 1.03 − 1.19
Tree density (DT) 0.39 0.41 0.01 0.01 0.99 0.75
Tree basal area (BT) − 0.13 − 0.42 0.01 0.01 1.07 0.51
Snag density (DS) 0.16 0.12 − 0.01 0.01 − 0.53 0.35
Snag basal area (BS) − 0.35 − 0.50 0.06 − 0.05 0.58 − 0.38
Log density (DL) − 0.38 − 0.68 0.02 0.02 0.95 0.64
Log basal area (BL) − 0.40 − 0.61 − 0.01 0.01 − 0.44 0.03
Shrub cover (SC) − 0.41 − 0.49 − 0.04* − 0.01 − 3.00 − 0.99
Herb cover (HC) 0.01 − 0.47 − 0.01 − 0.01 − 0.11 − 0.36
Disturbance (Ds) − 0.29 − 0.31 − 0.03 − 0.08 − 1.73 − 1.42
Mean annual temperature (AT) − 0.96 0.11 1.71* 0.71* − 9.27 2.96
Rainfall (Rn) − 0.94 0.09 1.77* 0.74 7.36 − 1.84
Degree days > 0 °C (Dd) − 0.89 0.16 1.61* 0.57 − 8.84 0.99
6 month mean temperature (6T) − 0.88 0.14 1.70* 0.69 − 3.21 0.68

Figure 5 CCA ordination of rare species

using floristic habitat sampling (FHS,
including all mesohabitats) data to explore the
relationships between 287 stands, 270 species
and 22 environmental variables. Abbreviations
are listed in Table 5.

66 Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd

 Estimating bryophyte diversity

Table 5 Floristic habitat sampling (rare species only — see CCA ordination Fig. 5) statistics for variables used in canonical correspondence
analysis (CCA) of 287 stands in BC’s cedar hemlock forest. Asterisks indicate significance at P < 0.01. Absolute t-value > 2.1 are used to indicate
important canonical coefficients (ter Braak, 1998). Bold values indicate variables with significant correlation and canonical coefficients

Interset correlation Canonical coefficient t-value

Variable Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2

Site series (SS) − 0.83 − 0.07 0.01 − 0.08 0.62 − 1.01

Elevation (Elv) 0.65 0.48 0.03 0.21* 1.09 2.32
Slope position (SP) − 0.01 0.08 − 0.01 − 0.06 − 0.59 − 1.16
Aspect (As) 0.28 0.18 0.02 − 0.04 1.85 − 1.14
Moisture regime (Hyg) − 0.46 − 0.08 0.01 0.01* 0.21 2.34
Rock cover (RC) 0.03 − 0.15 0.02 0.05* 1.21 2.44
Rock acidity (RA) 0.32 − 0.22 0.03* − 0.15* 3.11 − 3.79
Soil texture (ST) − 0.02 0.27 0.01 0.06 1.67 1.83
Canopy height (CH) − 0.35 − 0.58 − 0.02 − 0.08 − 1.79 − 1.58
Tree density (DT) 0.37 0.42 0.04* 0.09* 2.77 2.02
Tree basal area (BT) − 0.09 − 0.42 − 0.01 0.06 − 0.75 − 1.64
Snag density (DS) 0.22 0.09 0.02 0.11* 1.72 3.10
Snag basal area (BS) − 0.44 − 0.44 − 0.01 − 0.03 − 0.34 − 0.74
Log density (DL) − 0.46 − 0.61 − 0.04 − 0.19 − 1.65 − 1.89
Log basal area (BL) − 0.49 − 0.57 − 0.05 − 0.24* − 1.99 − 2.91
Shrub cover (SC) − 0.45 − 0.52 0.02 − 0.09 1.09 − 1.74
Herb cover (HC) 0.08 − 0.33 0.01 0.11* 0.66 2.29
Disturbance (Ds) − 0.21 − 0.67 − 0.03* − 0.41* − 2.35 − 7.84
Mean annual temperature (AT) − 0.98 0.10 − 2.29* − 1.06 − 16.34 1.98
Rainfall (Rn) − 0.96 0.10 1.06* − 0.76 4.73 − 0.89
Degree days > 0 °C (Dd) − 0.97 0.10 1.55* − 0.37 − 2.46 0.94
6 month mean temperature (6T) − 0.94 0.11 0.33* 0.26 − 2.57 0.55

compared within the dominant forest mesohabitat are much
greater (i.e. species richness is 50% higher) when using FHS as
compared to PS. We have also shown that PS within a mesohab-
itat will exclude important microhabitats and their respective
bryophyte communities even after sampling unconventionally
large sample areas. Intensifying PS or simply sampling large areas
using randomly placed plots will not necessarily include the nat-
ural variety in microhabitats. Even PS plots as large as 5000 m2
are not as efficient sampling units as FHS. Plot sampling was
designed to give a quantitative ‘snap shot’ of a plant community
in time and space, not a comprehensive sample of the species
within the community. PS focuses on bounded plots as sampling
units but these plots are too small and restrictive to consider the
total variety of microhabitats in a mesohabitat.
The forest landscape consists of a complex of mesohabitats
and microhabitats, and efficient sampling of species diversity
must reflect the natural variation and hierarchy of these habitat
types. In cedar hemlock forests, bryophyte diversity can be parti-
tioned within a hierarchy of stands, which contain mesohabitats,
and these latter consist of microhabitats. PS is usually confined to
the dominant forest mesohabitat in studies concerning biodivers-
ity and the impacts of silvicultural disturbance (Bell & Newmaster,
1998). Our study has shown that species diversity will be under-
estimated if the sampling method does not sample the diversity
of mesohabitats on the landscape and their component micro-
habitats. Comparisons of diversity estimates using PS and FHS
(including all mesohabitats) show significant differences. Total
species richness is 110% higher when using FHS methodology, a
difference that results because FHS focuses on the entire meso-
habitat as a sampling unit, including the variety of microhabitats
within each type of mesohabitat, and includes also a complete
floristic survey of the species within each mesohabitat. PS and
FHS require the same amount of sampling time to achieve an
estimate of richness, albeit PS underestimates richness by 50%.
The Chao index in combination with FHS produces accurate
estimates of species richness in less than half the sampling time.
The most efficient sampling method that captures greatest
diversity in stands should consider all the different types of meso-
habitats within a stand, and yield the highest species richness
values. PS and FHS were compared in their ability to capture bryo-
phyte diversity in stands. Typically, PS samples only the domin-
ant forest mesohabitat and other mesohabitats such as streams,
seeps and cliffs are excluded resulting in decreased diversity estim-
ates. FHS included sampling in the dominant forest mesohabitat
and restricted mesohabitats (i.e. streams, cliffs and seeps must be
included); all of these offer considerable diversity. Streams and
cliffs contain unique microhabitats that support communities of
species that are only found in those specific microhabitats. Streams
and cliffs have more unique species (70 and 26 species, respectively)
than the dominant forest mesohabitat (13 species). Floristic
habitat sampling captures the species diversity associated with the
variability of mesohabitat types in the cedar-hemlock forests.

Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd 67

S. G. Newmaster et al.
Estimating diversity for conservation purposes is difficult for
taxonomically challenging groups such as bryophytes (Gotelli &
Cowell, 2001; Vanderpoorten & Engels, 2002). Our estimates of
species richness indicate that PS underestimates richness by
50% when considering equal sample sizes for each methodology
including all mesohabitat types. PS wastes an incredible amount
of time randomly sampling space that adds no new species to the
pool. PS sampling effort was 20 times greater than FHS and
resulted in an estimate less than half of the species pool. Further-
more, the Chao index accurately estimated richness for FHS
using 30% of the total area sampled. FHS is a more accurate and
efficient sampling methodology for diversity, and when com-
bined with the Chao estimate of richness it becomes an expedited
method for estimating richness for conservation purposes.
Increasing sampling intensity of PS did not capture any more
rare species and certainly under sampled the rare flora. In fact the
only time our species area curves levelled off was when we used
only the common flora or the entire PS data file. We examined
the literature including several basic ecology texts (Gilbertson
et al., 1985; Barbour et al., 1987), and some of our own data files
from other projects and discovered that PS data typically produces
species area curves that level off, whereas FHS or Relevé (Gradstein
et al., 1996) data produce curves that never level off. Species area
curves are commonly used to determine the proper sample size.
Our results suggest that this may not be appropriate because FHS
species area curves do not level off and PS curves level prema-
turely, underestimating richness. Further research is needed to
investigate alternative methods to determine sampling size.
FHS is ideal for biodiversity research because it focuses on the
entire mesohabitat as the basic sampling unit and is flexible
enough to include the diversity of microhabitats in a mesohab-
itat. Stream mesohabitats provide a particularly good example.
The concept of a stream mesohabitat must include the stream
gully, which includes the stream itself and its banks. The banks
include microhabitats common to other mesohabitats, but they
also contain microhabitats unique to the stream. Rich communities
of bryophytes can be found on these unique bank microhabitats
because of the moist humid stream environment, particularly in
splash zones. It is common practice to sample only within the
stream itself and not the stream gully. A study of the bryophyte
flora of Bridal Veil Falls in the CWH recorded stream species
richness (35 species) from within the stream itself (Djan-Chekar,
1993). In our study of the CWH, mean species richness within
the stream itself (40 species) was comparable to Djan-Chekar’s
study, but much higher (103 species) when we considered the
entire stream gully. Intensive sampling that focuses only on
the stream itself misses the unique microhabitats that exist on
the immediate stream bank and underestimates biodiversity in
stream mesohabitats. Excluding these microhabitats from divers-
ity studies could result in a misinterpretation of the value of
streams in an ecosystem management plan.
The species we leave behind in plot sampling are predomin-
ately the rare ones because PS misses habitats containing
substantial bryophyte diversity. While a community analysis that
needs species abundance within fixed areas only considers the
common species, diversity studies should always include rare
68
species as these always comprise a large proportion of the flora
being considered. Recent bryophyte studies underscore this
point (Vitt, 1991; Vitt & Belland, 1997; Newmaster et al., 1999).
Stratification of habitats, as in FHS, provides a complete suite of
rare and common species. Species richness is the most appropri-
ate diversity measure, because indices based on higher moments
(Brillouin, Shannon, Simpson) include frequency values and
therefore ‘weight’ common species more heavily. A complete
sample of species requires only a simple measure of the number
of species (richness) to reveal the patterning of diversity in rela-
tion to habitats at fine scales.
The importance of using a technique that samples rare species
in diversity studies is highlighted in the analyses presented in this
study. At large scales (Regional/Provincial), PS ordinations of
common species accurately captures course floristic differences
based on strong climatic gradients. At a finer scale young and old
cedar hemlock forests stands were not distinguished in PS ordi-
nations because the sampling method records only the species
that are common and present to both young and old stands.
Ordinations of the rare PS data are not interpretable because they
reflect a fragmented or incomplete sample of the rare flora. In the
rare FHS ordinations, young and old cedar hemlock stands are
separated in ordination space because the presence of rare
species in old stands is recorded by FHS. Thus, rare species are
shown to contribute significantly to the patterning of diversity in
cedar hemlock forests. They are crucial in distinguishing differ-
ences between young and old stands, and to understanding the
patterning of diversity within them. Conversely, an ordination of
common species FHS flora is not interpretable because all of
these species are present in young and old stands, and different
biogeoclimatic zones. FHS failed to distinguish the biogeo-
climatic gradients at the large scale using the common species
because it does not provide an accurate estimate of species abund-
ance. The isobryalian mosses found in both the ICH and CWH
are much more common in the CWH, which was accurately dis-
played in the PS ordination of the common flora. McCune &
Lesica (1992) identified the trade off between species capture and
accuracy of cover estimates and concluded that sampling large
areas (similar to FHS) resulted in high species capture, but low
accuracy of cover estimates; sampling many small plots (PS)
results in low species capture and higher abundance estimates for
the common species. Further research is needed to see if modify-
ing FHS can provide an accurate measure of abundance at the
stand scale.
All sampling techniques have their disadvantages and ad-
vantages (Tilman, 1996). The sampling methodology must be
appropriate for the question being asked. The use of plot sam-
pling to address diversity questions is inappropriate for the rea-
sons previously discussed. Plot sampling method, on the other
hand, is appropriate for answering questions that require a fixed
reference to the area being sampled (i.e. for biomass or abund-
ance measures), or if you are focusing your question on a spe-
cific habitat type. These questions are generally investigated at
smaller landscape scales and include aspects of community
dynamics, plant sociology, physiological ecology, and population
ecology.
Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd

Floristic habitat sampling may be limited to questions con-
cerning biodiversity at large scales on the landscape. It would be
inappropriate to use this methodology to answer questions
regarding treatment impacts on diversity within small research
areas. The unbounded nature of this methodology limits its use
to research blocks that are large enough to capture the natural
variation in habitats. Vitt et al. (1995) successfully used a similar
habitat sampling method on bounded research blocks in peat-
lands where the microhabitat diversity is relatively homogene-
ous. Floristic habitat sampling methodology can be modified to
measure species diversity for different silvicultural treatments
within large bounded treatment blocks (experimental disturbances
in forests) with moderate habitat heterogeneity (Newmaster &
Bell, 2002), although a disadvantage of this approach is that
many large treatment areas are needed for a balanced biometric
analysis, and often this is not feasible or economical. Floristic
habitat sampling is an effective method for answering questions
about bryophyte diversity within and between stands, especially
those with different disturbance histories. Gilbertson et al. (1985)
stated that a detailed appreciation of plant-environmental rela-
tionships and the effects of management of vegetation by man can
often only be fully understood by a floristic sampling method.
Floristic habitat sampling provides results that provide insight
into the preparation of ecosystem management plans that
attempt to preserve ‘rare species and habitats’. As FHS records
both rare and common species and their frequency in microhab-
itats and mesohabitats, the method can aid in identification of
crucial habitats for rare species that need protection and man-
agement. FHS can also be used to identify larger areas such as
watersheds that have significantly high diversity and that may
warrant special preservation.
ACKNOWLEDGEMENTS
This research was supported by Forest Renewal British Columbia
and the British Columbia Forest Service. A special thank you for
the guidance and wisdom of Norm Kenkel, Denis Gignac and
Ellen MacDonald. We thank Joan Riemer, Aron Fazekas, Kevin
Bedford, Candice Newmaster and Annabel Newmaster for their
patience and diligence in helping to collect and process thousands
of specimens. A large portion of the stand environmental data was
collected by André Arsenault, Patrick Williston, Kim Johnston and
George Yearsley, this was a large task and we are extremely grateful
for their support. We would also like to thank Peter Minchin for
his constructive consultation regarding the multivariate analysis.
Finally, we thank Heather Cole for her editorial review.
SUPPLEMENTARY MATERIAL
The following material is available from
http://www.blackwellpublishing.com/products/journals/
suppmat/DDI/DDI123/DDI123sm.htm
Appendix S1 Spatial diversity at different scales for temperate
rainforest in British Columbia, Canada (terminology in Whit-
taker, 1965, 1972, 1977).
Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd
Estimating bryophyte diversity
REFERENCES
Anderson, L.E., Crum, H.A. & Buck, W.R. (1990) List of the
mosses of North America North of Mexico. The Bryologist, 93,
448–499.
Arsenault, A. (1995) Pattern and process in old-growth temperate
rainforest of southern British Columbia. PhD Dissertation.
Department of Botany, University of British Columbia,
Vancouver, B.C.
Arsenault, A. & Goward, T. (2000) Ecological characteristics of
inland rainforests. In: Proceedings of a conference on the biology
and management of species and habitats at risk (ed. by L. Darling),
pp. 437–439. B.C. Forest Service, Kamloops, BC.
Barbour, M.G., Burk, J.H. & Pitts, W.D. (1987) Terrestrial plant
ecology. Benjamin/Cummings Publishing, New York,
NY.
Bell, F.W. & Newmaster, S.G. (1998) Falling snow Ecosystem
Project: floral richness, abundance and diversity. In: Third
international forest vegetation management conference: forest
vegetation management and ecosystem sustainability-popular
summaries (ed. by R.G. Wagner and D.G. Thompson). Forest
Research Information Paper no. 141, pp. 45 –47. Queen’s
Printer, Toronto, ON.
Belland, R.J. (1989) Floristic boundaries in the Gulf of St. Law-
rence region: a numerical approach based on the moss flora.
Canadian Journal of Botany, 67, 1633–1644.
Belland, R.J. (1995) The bryophytes of Prince Edward Island
National Park, Canada: Affinities, habitats, and diversity. Frag-
menta Floristica et Geobotanica, 40, 349–364.
Belland, R.J. & Brassard, G.R. (1988) The bryophytes of Gros
Morne National Park, Newfoundland Canada: Ecology and
phytogeography. Lindbergia, 14, 97–118.
Belland, R.J. & Schofield, W.B. (1994) The ecology and phyto-
geography of the bryophytes of Cape Breton Highlands Park,
Canada. Nova Hedwigia, 59, 275–309.
Belland, R.J. & Vitt, D.H. (1995) Bryophyte vegetation patterns
along environmental gradients in continental bogs. Ècoscience,
2, 395–407.
Bonham, C.D. (1989) Measurements for terrestrial vegetation.
John Wiley & Sons, New York, NY.
Bouchard, A., Hay, S. & Rouleau, E. (1978) The vascular flora of
St. Barbe south district Newfoundland: An interpretation
based on biophysiographic areas. Rhodora, 80, 228–308.
ter Braak, C.J.F. (1998)  4. Centre for Biometry, Wage-
ningen, The Netherlands.
Braun-Blanquet, J. (1932) Plant Sociology. The study of plant
communities. McGraw-Hill, Toronto, ON.
Cain, S.A. & Castro, G.M. (1959) Manual of vegetation anaysis,
p. 321. Harpour & Brothers, New York.
Carleton, T.J. (1990) Variation in terricolous bryophytes
and macrolichen vegetation along primary gradients in
Canadian boreal forest. Journal of Vegetation Science, 3,
585 –594.
Chao, A. (1984) Non-parametric estimation of the number of
classes in a population. Scandinavian Journal of Statistics, 11,
265–270.
69
S. G. Newmaster et al.
Clements, F.E. (1905) Research methods in ecology. Woodruff-
Collins, Lincoln, NB.
Clifford, H.T. & Stephenson, W. (1975) An introduction to
numerical classification. Academic Press, London, UK.
Cochran, W.G. (1977) Sampling techniques, p. 654. Wiley, New
York, NY.
Colwell, R.K. (1997) Estimates: statistical estimation of species
richness and shared species from samples, Version 6.0, http://
viceroy.eeb.uconn.edu/estimates, [April 2004].
Colwell, R.K. & Coddington, J. (1994) Estimating terrestrial
biodiversity through extrapolation. Philosophical transactions
of the Royal Society London, Series B: Biology Sciences, 345,
101–118.
Conard, H.S. (1935) The plant associations on central Long
Island: a study in descriptive sociology. American Midland
Naturalist, 16, 433 – 516.
Djan-Chekar, N. (1993) The bryophyte flora of bridal veil falls,
British Columbia: an analysis of its composition and diversity.
Masters Thesis. Department of Botany, University of British
Columbia, Vancouver BC.
During, H.J. & ter Horst, B. (1987) Diversity and dynamics in
bryophyte communities on earth banks in a Dutch forest.
Symposia Biologica Hungarica, 35, 447– 455.
During, H.J., van Tooren, B.F. & Bart, F. (1987) Recent develop-
ments in bryophyte population ecology. Trends in Ecology and
Evolution, 2, 89 – 93.
Frego, K.A. & Carleton, T.J. (1995) Microsite conditions and
spatial pattern in a boreal bryophyte community. Canadian
Journal of Botany, 73, 544 – 551.
Gignac, D.L. (1986) Ecological tolerance and niche structure of
Sphagnum along a pollution gradient near sudbury, Ontario,
Canada. Canadian Journal of Botany, 65, 1138 –1147.
Gignac, D.L., Nicholson, B.J. & Bayley, S.A. (1998) The utiliza-
tion of bryophytes in bioclimatic modeling: Present distribu-
tion of peatlands in the Mackenzie River Basin, Canada. The
Bryologist, 101, 560 – 572.
Gignac, D.L. & Vitt, D.H. (1994) Habitat limitations of Sphag-
num along climatic, chemical, and physical gradients in mires
of western Canada. The Bryologist, 93, 7–22.
Gilbertson, D.D., Kent, M. & Pyatt, F.B. (1985) Practical ecology.
Hutchinson Ltd, New Hampshire, USA.
Gotelli, N.J. & Cowell, R.K. (2001) Quantifying biodiversity: pro-
cedures and pitfalls in the measurement and comparison of
species richness. Ecology Letters, 4, 379 – 391.
Gradstein, R.S. (1992) The vanishing tropical rainforest as an
environment for bryophytes and lichens. In: Bryophytes and
lichens in a changing environment (ed. by J.W. Bates and
A.M. Farmer), pp. 234 –253. Clarendon Press, Oxford, UK.
Gradstein, S.R., Hietz, P., Lücking, R., Lücking, A., Sipman, H.J.M.,
Vester, H.F.M., Wolf, J. & Gardette, E. (1996) How to sample
the epiphytic diversity of tropicalrainforests. Ecotropica, 2,
59 –72.
Gustafsson, L. & Hallingbäck, T. (1988) Bryophyte flora and
vegetation of managed and virgin coniferous forest in south-
west Sweden. Biological Conservation, 44, 283 – 300.
Hebda, R.J. (1995) British Columbia vegetation and climatic
70
history with focus on 6 KA BP. Géographie Physique et Quater-
naire, 49, 55–79.
Heinlen, E.R. & Vitt, D.H. (2003) Patterns of rarity in mosses of
the Okanogen Highlands of Washington State: an emerging
coarse filter approach to rare moss conservation. The Bryo-
logist, 106, 34 –52.
Herben, T., Rydin, H. & Söderstrom, L. (1991) Spore establish-
ment probability and the persistence of the fugitive invading
moss, Orthodontium lineare: a spatial simulation model. Oikos,
60, 215–221.
Herben, T. & Söderstrom, L. (1992) Which habitat parameters
are most important for the persistence of a bryophyte species
on patchy, temporary substrates? Biological Conservation, 59,
121–126.
Horton, D.G. (1988) Microhabitats of New World Encalyptaceae
(Bryopsida): distribution along edaphic gradients. Beifefte Zur
Nova Hedwigia, 90, 261–282.
Johnston, M.H. & Elliot, J.A. (1996) Impacts of logging and wildfire
on an upland black spruce community in northwestern Ontario.
Environmental Monitoring and Assessment, 39, 283–297.
Kimmerer, R.W. & Allen, T.F.H. (1982) The role of disturbance
in the pattern of a riparian bryophyte community. American
Midland Naturalist, 107, 370–383.
Kimmins, J.P. (1987) Forest ecology. MacMillan Publishing Co,
City.
Komposch, H. & Hafellner, J. (1998) Biodiversity and ecology of
lichens in an Amazonian tropical lowland rain forest. Selbyana,
19, 284.
Krebs, C.J. (1989) Ecology. Harper & Row, Dawn Mills, ON.
Krebs, C.J. (1997) Ecological methodology. Harper Collins, Dawn
Mills, ON.
Kürschner, H. (2003) Epiphytic bryophyte communities of
southwestern Arabia phytosociology, ecology and life strat-
egies. Nova Hedwigia, 77, 55–71.
Kürschner, H. & Parolly, G. (1998a) Syntaxonomy of trunk-
epiphytic bryophyte communities of tropical rain forests — A
first pantropical approach. Phytocoenologia, 28, 357–425.
Kürschner, H. & Parolly, G. (1998b) Life strategies of epiphytic
bryophytes in the rainforests along the eastern Andean slopes
and the Amazonian lowlands of northern Peru. Nova Hedwi-
gia, 67, 1–22.
Kürschner, H. & Wolfgang, F. (1999) Patterns and adaptive
trends of life forms, life strategies and ecomorphological struc-
tures in tropical epiphytic bryophytes: a pantropical synopsis.
Nova Hedwigia, 69, 73 –99.
Leonard, R.E., Conkling, P.W. & McMahon, J.L. (1985) Recovery
of bryophyte community on Hurricane Island, Maine. USDA
Forest Service Research Note, 325, 1– 4.
Lesica, P., McCune, B., Cooper, S.V. & Hong, W.S. (1991)
Differences in lichen and bryophyte communities between
old-growth and managed second-growth forests in the Swan
Valley, Montana. Canadian Journal of Botany, 69, 1745–1755.
Lücking, A. (1997) Diversity and distribution of epiphyllous
bryophytes in a tropicalrainforest in Costa Rica. Abstracta
Botanica, 21, 79 –87.
Lücking, R. (1999a) Ecology of foliicolous lichens at the
Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd

Botarrama trail (Costa Rica), a neotropical rain forest. I.
Species composition and its ecogeographical implications.
Biotropica, 31, 553 – 564.
Lücking, R. (1999b) Ecology of foliicolous lichens at the Botar-
rama trail (Costa Rica), a neotropical rain forest. IV. Species
associations, their salient features, and their dependence on
environmental variables. Lichenologist, 31, 269 –289.
McCune, B. (1993) Gradients in epiphytic biomass in three
Pseudotsuga-Tsuga forest of different ages in western Oregon
and Washington. The Bryologist, 96, 405 – 411.
McCune, B. (1997) Repeatability of common data: Species
richness versus gradient scores in large-scale lichen studies.
The Bryologist, 100, 40 –46.
McCune, B. & Lesica, P. (1992) The trade-off between species
capture and quantitative accuracy in ecological inventory of
lichens and bryophytes in forest in Montana. The Bryologist,
95, 296 –304.
Meidinger, D. & Pojar, J. (1991) Special report series 6: ecosystems
of British Columbia. BC Ministry of Forest, Research Branch,
Victoria, BC.
Morneault, A.E., Parker, W.C., Naylor, B.J., Noland, T.L., Bell, F.W.,
Newmaster, S.G., Othermer, D.C., Duchesne, L.C., Woods, M.E.,
Burgess, D.M., Wetzel, S.S. & Pinto, F.N.L. (1998) Ecological
effects of site preparation in white pine stands managed under
the shelterwood system. In: Third international forest vegeta-
tion management conference: forest vegetation management and
ecosystem sustainability-popular summaries (ed. by R.G. Wagner
and D.G. Thompson). Research Information Paper no. 141,
pp. 454 – 456. Queen’s Printer, Toronto, ON.
Newmaster, S.G. (2000) Patterning of bryophyte diversity in
temperate rainforests. PhD Dissertation. Department of
Botany, University of Alberta, Edmonton, Alberta.
Newmaster, S.G. & Bell, F.W. (2002) The effects of silvicultural
disturbances on cryptogam diversity in the boreal-mixed wood
forests. Canadian Journal of Forest Research, 32, 38–51.
Newmaster, S.G., Belland, R.J., Arsenault, A. & Vitt, D.H. (2003)
Patterns of bryophyte diversity in humid coastal and inland
cedar-hemlock forests of British Columbia. Environmental
Review, 11, S159 –S185.
Newmaster, S.G., Vitt, D.H. & Bell, F.W. (1999) The effects of
triclopyr and glyphosate on common bryophytes and lichens
in northwestern Ontario. Canadian Journal of Forest Research,
29, 1101–1111.
Økland, R.H. (1994) Patterns of bryophyte associations at different
scales in a Norwegian boreal spruce forest. Journal of Vegeta-
tion Science, 5, 127–138.
Økland, R.H., Økland, T. & Eilertsen, O. (1990) On the relation-
ship between sample plot size and beta diversity in boreal
coniferous forest. Vegetatio, 87, 187–192.
Peet, K.P. (1974) The measurement of species diversity. Annual
Review of Ecology Systematics, 5, 285 – 307.
Pielou, E.C. (1966) Species-diversity and pattern-diversity in the
study of ecological succession. Journal of Theoretical Biology,
10, 370 –383.
Pike, L.E., Denison, W.C., Tracey, D.M., Sherwood, M.A. &
Rhoades, F.M. (1975) Floristic survey of epiphytic lichens and
Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd
Estimating bryophyte diversity
bryophytes growing in old-growth conifers in western Oregon.
The Bryologist, 78, 1– 39.
Pøcs, T. (1980) The epiphytic biomass and its effects on the water
balance of two rain forest types in the Ilguru Mountains,
Tanzania, East Africa. Acta Botanica Hungarica, 26, 143–167.
Rambo, T.R. & Muir, P.S. (1998a) Forest floor bryophytes of
Pseudotsuga menziesii-Tsuga hetrophylla stands in Oregon:
influences of substrate and overstory. The Bryologist, 101, 116–
130.
Rambo, T.R. & Muir, P.S. (1998b) Bryophyte species associations
with coarse woody debris and stand ages in Oregon. The
Bryologist, 101, 366–377.
Schofield, W.B. (1988) Bryogeography and bryophytic character-
ization of biogeoclimatic zones of British Columbia, Canada.
Canadian Journal of Botany, 66, 2673–2686.
Schoonmaker, P.K., von Hagen, B. & Wolf, E.C. (1997) The rain
forests of home. Island Press, Washington, DC.
Schuster, R. (1949) The ecology and distribution of Hepaticae in
central and western New York. American Midland Naturalist,
42, 513–712.
Shafti, M.I. & Yarranton, G.A. (1972) Diversity, floristic richness
and species evenness during a secondary (post fire) succession.
Ecology, 54, 897–902.
Shaw, A.J. (1981) Ecological diversification among nine species
of Pohlia (Musci) in western North America. Canadian Journal
of Botany, 59, 2359–2378.
Slack, N.G. (1976) Host specificity of bryophyte epiphytes in
eastern North America. Journal of Hattori Botanical Labor-
atory, 41, 107–132.
Slack, N.G. (1977) Species diversity and community structure in
bryophytes: New York State Studies. Bulletin of the New York
State Museum, 428, 1–70.
Slack, N.G. (1984) A new look at bryophyte community analysis:
field and statistical methods. Journal of Hattori Botanical
Laboratory, 55, 113–132.
Söderström, L. (1988a) Sequence of bryophytes and lichens
in relation to substrate variables of decaying coniferous wood
in Northern Sweden. Nordic Journal of Botany, 8, 89 –97.
Söderström, L. (1988b) The occurrence of epixylic bryophyte
and lichen species in an old natural and managed forest
stand in northwest Sweden. Biological Conservation, 45,
169–178.
Söderström, L. (1989) Regional distribution patterns of bryophyte
species on spruce logs in northern Sweden. The Bryologist, 92,
349–355.
SPSS version 10. — Anonymous (2003) Professional base system
software for statistical analysis. Chicago, IL.
Stotler, R. & Crandall-Stotler, B. (1977) A checklist of the liver-
worts and hornworts of North America. The Bryologist, 80,
405–428.
Tilman, D. (1996) Biodiversity: Population versus ecosystem
stability. Ecology, 77, 350–363.
Vanderpoorten, A. & Engels, P. (2002) The effects of environ-
mental variation on bryophytes at a regional scale. Ecography,
25, 513–522.
Vitt, D.H. (1991) Distribution patterns, adaptive strategies, and
71
S. G. Newmaster et al.
morphological changes of mosses along elevational and latitu-
dinal gradients on South Pacific Islands. In: Quantitative
approaches to phytogeography (ed. by P.L. Nimis and T.J. Crovello),
pp. 205–236. Kluwer Academic Publishers, The Netherlands.
Vitt, D.H., Achuff, P. & Andrus, R.E. (1975) The vegetation and
chemical properties of patterned fens in the Swan Hills, north
central Alberta. Canadian Journal of Botany, 53, 2776 –2795.
Vitt, D.H. & Belland, R.J. (1995) The bryophytes of peatlands in
continental western Canada. Fragmenta Floristica et Geobo-
tanica, 40, 339 –348.
Vitt, D.H. & Belland, R.J. (1997) Attributes of rarity among
Alberta Mosses: Patterns and prediction of species diversity.
The Bryologist, 100, 1–12.
Vitt, D.H., Li, Y. & Belland, R.J. (1995) Patterns of bryophyte
diversity in peatlands of continental western Canada. The Bry-
ologist, 98, 218 –227.
Whittaker, R.H. (1965) Dominance and diversity in land plant
communities. Science, 147, 250 –260.
Whittaker, R.H. (1972) Evolution and measurement of species
diversity. Taxon, 21, 213 –251.
72
Whittaker, R.H. (1977) Evolution of species diversity in land
communities. In: Evolutionary biology, Vol. 10 (ed. by M.K. Hecht,
W.C. Steere and B. Wallace), pp. 1–67. Plenum, New York.
Wolf, J.H.D. (1993a) Epiphyte communities of tropical montane
rain forest in the northern Andes. I. Lower montane commu-
nities. Phytocoenologia, 22, 1– 53.
Wolf, J.H.D. (1993b) Epiphyte communities of tropical montane
rain forest in thenorthern Andes. II. Upper montane commu-
nities. Phytocoenologia, 22, 53–103.
Wolf, J.H.D. (1993c) Diversity patterns and biomass of epiphytic
bryophytes and lichens along an altitudinal gradient in the
northern Andes. Annals of the Missouri Botanical Gardens, 80,
928–960.
Wolf, J.H.D. (1994) Factors controlling the distribution of
vascular and non-vascular epiphytes in the northern Andes.
Vegetatio, 112, 15–28.
Zamfir, M., Dia, X. & van der Maarel, E. (1999) Bryophytes,
lichens and phanerogams in an alvar grassland: relationships at
different scales and contributions to plant community pattern.
Ecography, 22, 40–52.
Diversity and Distributions, 11, 57–72, © 2005 Blackwell Publishing Ltd