Oecologia (2009) 161:569–579
DOI 10.1007/s00442-009-1402-1
C O M M U N I T Y E C O L O G Y - O RI G I N A L P A P E R
Dispersal and life history strategies in epiphyte metacommunities:
alternative solutions to survival in patchy, dynamic landscapes
Swantje Löbel · Håkan Rydin
Received: 16 July 2008 / Accepted: 15 June 2009 / Published online: 5 July 2009
© Springer-Verlag 2009
Abstract Host trees for obligate epiphytes are dynamic
patches that emerge, grow and fall, and metacommunity
diversity critically depends on eYcient dispersal. Even
though species that disperse by large asexual diaspores are
strongly dispersal limited, asexual dispersal is common.
The stronger dispersal limitation of asexually reproducing
species compared to species reproducing sexually via small
spores may be compensated by higher growth rates, lower
sensitivity to habitat conditions, higher competitive ability
or younger reproductive age. We compared growth and
reproduction of diVerent groups of epiphytic bryophytes
with contrasting dispersal (asexual vs. sexual) and life history strategies (colonists, short- and long-lived shuttle species, perennial stayers) in an old-growth forest stand in the
boreo-nemoral region in eastern Sweden. No diVerences
were seen in relative growth rates between asexual and sexual species. Long-lived shuttles had lower growth rates than
colonists and perennial stayers. Most groups grew best at
intermediate bark pH. Interactions with other epiphytes had
a small, often positive eVect on growth. Neither diVerences
in sensitivity of growth to habitat conditions nor diVerences
in competitive abilities among species groups were found.
Habitat conditions, however, inXuenced the production of
sporophytes, but not of asexual diaspores. Presence of
Communicated by Brian Beckage.
Electronic supplementary material The online version of this
article (doi:10.1007/s00442-009-1402-1) contains supplementary
material, which is available to authorized users.
S. Löbel (&) · H. Rydin
Department of Plant Ecology,
Evolutionary Biology Centre, Uppsala University,
Norbyvägen 18D, 752 36 Uppsala, Sweden
e-mail: [email protected]
sporophytes negatively aVected growth, whereas presence
of asexual diaspores did not. Sexual species had to reach a
certain colony size before starting to reproduce, whereas no
such threshold existed for asexual reproduction. The results
indicate that the epiphyte metacommunity is structured by
two main trade-oVs: dispersal distance vs. reproductive age,
and dispersal distance vs. sensitivity to habitat quality.
There seems to be a trade-oV between growth and sexual
reproduction, but not asexual. Trade-oVs in species traits
may be shaped by conXicting selection pressures imposed
by habitat turnover and connectivity rather than by species
interactions.
Keywords Growth · Local processes · Metapopulation ·
Reproduction · Trade-oVs
Introduction
InterspeciWc trade-oVs have long been recognized in community ecology, especially resource partitioning and associated species traits. It is commonly assumed that long-term
coexistence of species requires some trade-oV among
important biological traits (Chase et al. 2005). Metacommunity theory draws attention to traits that operate at
regional scales, especially migration and associated life
history trade-oVs (Leibold and Miller 2004). Most work has
focused on competition-colonization trade-oVs (Amarasekare
2003; Leibold and Miller 2004; Chase et al. 2005). In
plants, seed number vs. seed size has often been used as a
surrogate for this trade-oV; nonetheless the species traits
that determine competitive and dispersal abilities are complex (Kneitel and Chase 2004) and very few empirical studies have adequately assessed both the colonization and
competition process (Cadotte et al. 2006). Neutral lottery
123
570
models and theories of unstable coexistence further challenge classic theories of species coexistence: colonization
diVerences among species can theoretically be part of an
equalizing trade-oV where species end up having equal
Wtness (Chesson 2000).
Classic metapopulation and metacommunity theory
assume system equilibrium and static landscapes. Patchtracking metacommunities, characterized by a short lifetime of the habitat patches in relation to the slow population
dynamics of the inhabiting species (Snäll et al. 2003, 2005),
may never reach the assumed colonization-extinction equilibria. Such metacommunities may be mainly structured by
dispersal and reproductive patterns rather than by competition-colonization trade-oVs (cf. Zartman and Shaw 2006),
and alternative dispersal strategies oVer several trade-oVs to
be explored.
Successful dispersal involves three main processes: the
production, transport and establishment of diaspores
(Amarasekare 2003). Most studies focus on the process of
diaspore transport, and mean dispersal distance has been
shown to be an important species trait in dynamic landscapes (Johst et al. 2002). However, if habitat turnover is
fast, patch longevity can be the most critical parameter for
species persistence (Holyoak et al. 2005; Miller and Kneitel
2005; Bossuyt and Honnay 2006). In such situations, age at
Wrst reproduction and dispersal frequency may play an
important role. Further, high establishment probabilities
and initial growth rates are important for species survival,
even in non-competitive metacommunities: Johst et al.
(2002) showed theoretically that long-range dispersal loses
its advantage in dynamic landscapes if numbers of potential
emigrants are low due to low local population growth rates.
Epiphytic bryophytes conWned to deciduous trees interspersed among dominant conifers provide an excellent
model system for exploring dispersal and metacommunity
processes. Host trees within forests are dynamic patches
that emerge, grow and fall. Epiphytes need to track these
patches for metapopulation persistence (Snäll et al. 2005).
In the boreo-nemoral and boreal regions, the typical successional trend is from deciduous trees, which establish in
great numbers following disturbances such as forest Wres, to
conifers. Thus, even deciduous forest stands are dynamic
patches, and long-term metacommunity persistence
depends on dispersal across landscapes. The dynamics of
single trees are rapid compared to the long life span of most
epiphytes, and local extinctions on trees are mainly governed by tree fall. Such extinctions can be much more common than extinctions of populations from standing trees
(Snäll et al. 2005).
Bryophytes have evolved sexual and asexual diaspores
diVering widely in size. The sexual spores are generated
by meiosis in the diploid sporophyte (spore capsule),
which remains attached to and dependent on its maternal
123
Oecologia (2009) 161:569–579
gametophyte throughout its full lifetime. Asexual reproduction is possible via production of specialized asexual diaspores (gemmae, gemmae-like branchlets), via clonal growth
or by fragmentation (Laaka-Lindberg et al. 2003; During
2007). Gemmae are tissues of one or several cells formed
on leaf surfaces, in leaf axils or on rhizoids. As most asexual diaspores are distinctly larger than spores they generally
have shorter dispersal distances but higher establishment
rates than spores (Kimmerer 1994; Laaka-Lindberg et al.
2003, 2006). Furthermore, studies suggest lower energy
costs and less sensitivity to weather and habitat conditions
of asexual compared to sexual reproduction (Pohjamo et al.
2006) and younger age at Wrst asexual reproduction (During
1979; Pohjamo and Laaka-Lindberg 2004). Production of
spores is commonly seasonal, whereas asexual reproduction often occurs continuously (Pohjamo et al. 2006). During
(1979, 1992) proposed a classiWcation of bryophyte life
history strategies based on trade-oVs between gametophyte
longevity and reproductive eVort, and between diaspore
size and diaspore number. Epiphytes can be classiWed as
colonists, short- and long-lived shuttle species or perennial
stayers with a decreasing reproductive eVort and increasing
gametophyte longevity, and possibly an increasing competitive ability and/or stress tolerance.
Dispersal limitation and metapopulation dynamics of
epiphytes are indicated by dispersal experiments (Dettki
and Esseen 2003), recorded colonizations of trees (Snäll
et al. 2005), spatial genetic structuring (Snäll et al. 2004a),
and spatially aggregated species distributions (Snäll et al.
2003, 2004b; Löbel et al. 2006b). Earlier studies on a larger
scale (forest stands in the landscape) have indicated dispersal distances of distinctly less than 100 m for large asexual diaspores, whereas a considerable fraction of small
bryophyte spores are transported >500 m (Snäll et al. 2003,
2004b; Löbel et al. 2006b, 2009). Still, asexual reproduction is common among epiphytes, suggesting that disadvantages of short dispersal distances are compensated by some
other life history traits.
Several trade-oVs could be involved in epiphyte dispersal and life history strategies. One possibility is a tradeoV between sensitivity to habitat conditions and dispersal
ability: we observed a strong impact of habitat conditions
on species richness of sexually but not of asexually reproducing epiphytes (Löbel et al. 2006a, 2009). Species richness of sexually reproducing epiphytes was explained by
the landscape structure in existence several decades ago,
whereas that of asexually dispersed species was aVected by
the present landscape only (Löbel et al. 2006b, 2009). This
could indicate a trade-oV between dispersal distance and
dispersal frequency. We have earlier suggested that competition among epiphytes is relatively unimportant (Löbel
et al. 2006a, b), but it is not yet clear whether epiphytes
form ‘interactive’ metacommunities or not. If there is a
Oecologia (2009) 161:569–579
competition-colonization trade-oV among epiphytes, we
would expect asexually reproducing species to have higher
competitive abilities and growth rates than sexually reproducing species, and similarly for perennial stayers compared to colonists and shuttle species.
Our study explores trade-oVs in dispersal and life history
strategies in epiphyte metacommunities. We hypothesize
that the stronger dispersal limitation of species reproducing
via large asexual diaspores, in comparison to species reproducing sexually via small spores, is compensated by a
higher growth rate, lower sensitivity to habitat conditions,
higher competitive ability, more frequent reproduction or
younger age at Wrst reproduction. The life history classiWcation of During (1979, 1992) further predicts an increasing
age at Wrst reproduction and decreasing reproduction
frequency from colonists over short- and long-lived shuttles
to perennial stayers. This, we hypothesize, may be compensated by increasing growth rates, competitive ability or
decreasing sensitivity to habitat conditions.
Materials and methods
Field work and study site
We studied colony growth and reproduction of obligate epiphytic bryophytes in a forest stand in the boreo-nemoral
region (transition zone between the deciduous nemoral and
the coniferous boreal zones) in eastern Sweden (Valkrör:
2.4 ha, 60°03⬘N, 18°26⬘E). The surrounding landscape consists of a mosaic of managed and unmanaged forests. Forestry was extensive until the 1970s. Thereafter, many old
deciduous forests have been cut and replaced by forests
dominated by Picea abies (Eriksson 1997). The study forest lies in a depression and is surrounded by spruce-dominated forests. The most common host tree species is
Fraxinus excelsior, followed by Ulmus glabra and Tilia
cordata. Host trees are intermixed with Picea abies, Alnus
glutinosa and Betula pubescens. The forest shows some
traces of tree cuttings, but is not aVected by recent forestry.
Host trees are expected to decline in the future due to competition from Picea abies: during a 6-year period, Snäll
et al. (2005) observed a recruitment of six new host trees
only, whereas 16 of 489 host trees fell. Spatial species richness patterns, as well as colony growth, reproduction
(Wiklund and Rydin 2004a) and colonization-extinction
dynamics (Snäll et al. 2005) of one epiphyte (Neckera
pennata) were studied within the same and nearby forest
stands.
We included seven pleurocarpous mosses, two acrocarpous mosses and three liverworts, i.e. all main obligate
epiphytes growing on broadleaved deciduous trees within the
forest stand. Pleurocarpous mosses are usually monopodially
571
branched and tend to form spreading carpets; sporophytes
are borne on short, lateral branches. Acrocarpous mosses
show only little or no branching and typically grow in erect
tufts; sporophytes are borne at the tips of the stems or
branches. Species reproduce either sexually through spores
of diVerent sizes or by asexually produced gemmae or gemmae-like branchlets. We grouped species according to their
reproduction mode (sexual, asexual) and life history strategy (During 1979, 1992). Colonists reproduce by numerous
light spores (<20 m in diameter); the reproductive eVort is
high, and the potential life span not more than a few years.
Short-lived shuttle species reproduce by fewer, larger
spores (>20 m); they are rather short-lived and show
medium reproductive eVort. Long-lived shuttle species
diVer from short-lived shuttles in their longer life span, and
even lower reproductive eVort. Perennial stayers reproduce
only rarely by small spores (<20 m); their potential life
spans are many years, or even decades. They are either
stress tolerants or competitors. Species of the same group
were analysed jointly. Reproduction modes and life history
strategies (During 1979, 1992; Dierßen 2001; Hill et al.
2007) of the study species are presented in Fig. 1, and
together with diaspore size and other attributes as Electronic Supplementary Material (S1).
Between 40 and 48 colonies of each species were
marked and their peripheries drawn on plastic sheets in
September 2004 and September 2005. Most colonies grew
on Fraxinus excelsior and very few on Ulmus glabra.
Growth height varied between 6 and 207 cm above ground.
Colonies were distributed among 136 trees, and not more
than one colony per species and tree was sampled. We
chose colonies with the criteria that they should be smaller
than 100 cm2 (larger colonies often fuse and are diYcult to
separate and measure) but variable in size, healthy and
separated from other colonies of the same species by at
least 10 cm.
We noted the number of sporophytic shoots and sporophytes as well as the presence of asexual diaspores. We
digitized the colony drawings using a scanner (HP ScanJet
5300c; resolution 300 dpi). Colony area was calculated by
counting pixels inside the perimeter (using Image; Karlsson
2007).
For each colony we recorded tree species, height above
ground, stem diameter at 1.3 m (diameter at breast height;
DBH), depth of bark crevices and stem inclination. We
measured the bark surface pH in the Weld using a pHelectrode for surface measurements (Hamilton single-pore
Xat). Distilled water was sprayed on the bark surface and
the pH was measured after 5–10 min. Light conditions were
accessed by measuring canopy openness (%) using digital
hemispherical photography (Nikon Coolpix 4500, Nikon
FC-E8 Wsheye converter). The camera was set to a Wxed
aperture (f = 3.3) and the shutter speed was chosen
123
572
Oecologia (2009) 161:569–579
(a) Anomodon longifolius
Sexual species
Sporophytes 2004
Sporophytes 2005
Sporophytes 2004 + 2005
Non-fertile
−1
Growth (mm year )
3000
2
2000
(c) Homalia trichomanoides
(b) Isothecium alopecuroides
Sexual, perennial stayer
Sexual, perennial stayer
Sexual, perennial stayer
3000
3000
2000
2000
RGR = 0.39 (0.17)
n = 47
RGR = 0.28 (0.22)
n = 42
RGR = 0.26(0.28)
n = 40
1000
1000
1000
0
0
0
(e) Frullania dilatata
(d) Neckera pennata
Sexual, long-lived shuttle
(f) Radula complanata
Sexual, long-lived shuttle
1500
Sexual, long-lived shuttle
1500
−1
Growth (mm year )
3000
1000
2
2000
RGR = 0.33 (0.23)
n = 42
1000
RGR = 0.16 (0.20)
n = 48
RGR = 0.18 (0.17)
n = 41
1000
500
500
00
0
0
(g)Orthotrichum speciosum
Sexual, short-lived shuttle
100
(i) Leucodon sciuroides
(h) Pylaisia polyantha
Asexual, long-lived shuttle
Sexual, colonist
1500
Asexual species
Gemmae 2004
Gemmae 2005
Gemmae 2004 + 2005
Non-fertile
3000
−1
Growth (mm year )
0
1000
RGR= 0.42 (0.32)
n = 46
2000
2
-100
-200
500
1000
RGR = 0.22 (0.17)
n = 43
RGR = -0.20 (0.31)
n = 44
-300
0
0
-400
(j)Metzgeria furcata
Asexual, long-lived shuttle
(k) Platygyrium repens
Asexual, colonist
1500
(l)Bryum flaccidum
Asexual, colonist
100
3000
−1
Growth (mm year )
0
1000
RGR = 0.38 (0.27)
n = 43
-100
2
2000
RGR = 0.27 (0.23)
n = 46
-200
500
1000
RGR = -0.62 (0.79)
n = 28
-300
0
0
-400
10
123
50
500
5000
2
Colony area 2004 (mm )
10
50
500
5000
2
Colony area 2004 (mm )
10
50
500
5000
Colony area 2004 (mm2)
Oecologia (2009) 161:569–579
䉳 Fig. 1 Comparison of growth of epiphyte colonies (mm2 year¡1) in
relation to original colony area (mm2) for study species in 2004; data
are ln-transformed. The mean relative growth rates (RGR) with SD
(in parentheses) and the number of studied colonies of each species,
predominating reproduction modes and life history strategies are given
for each species. Symbols indicate whether the sexual species had
sporophytes or if the asexual species had gemmae/gemmae-like
branchlets in 2004 and 2005
automatically by the camera. We photographed at a height
of 1.3 m at the northern and southern side of the tree, 20 cm
distance from the trunk, and we used the mean in calculations. All photographs were taken in July 2004, always on
overcast days. Images were analysed using the free
software Gap Light Analyzer (Frazer et al. 1999). We
estimated the soil moisture on a four-level ordinal scale
(Anon. 1997), where 1 was dry (ground water level >2 m
below soil surface), 2 was mesic (ground water level 1–2 m
below soil surface), 3 was mesic to moist (ground water
level <1 m below soil surface, Xat ground), and 4 was moist
(ground water level <1 m below soil surface, visible in hollows). In the statistical analyses we used the contrast
between ‘dry to mesic’ (classes 1 and 2) and ‘moist’ (classes
3 and 4) soils. To analyse the eVects of species interactions,
we noted the cover of all epiphytes (both bryophytes and
lichens) in deciles (0–10%, 10–20%, etc.) within 2 cm
periphery of each colony.
Data analyses
Relative growth rate (RGR) was calculated for each colony
as RGR = (lnAt ¡ lnA0)/t, where A0 is the initial colony
area, At is the colony area at time t, and t = 1 year.
We used linear mixed-eVect models to test whether species groups diVered in RGR. Species group was treated as a
Wxed factor, whereas species constituted a random factor.
Models were Wtted by maximum likelihood, and model
selection was based on Akaike’s information criteria (AIC)
(Akaike 1974). By deWnition, the best model has AIC = 0,
and models with AIC < 2 are generally worthy of consideration (Burnham and Anderson 2002). The level of empirical support for models AIC > 4 is considerably less, and
for models with AIC > 10, it is essentially absent (Burnham
and Anderson 2002).
We tested the eVects of environmental variables and
cover of adjacent epiphytes on RGRs for all epiphytes and
for the diVerent species groups using multiple linear mixedeVect models. Here, environmental and cover variables
were treated as Wxed factors and species as a random factor.
We Wrst tested the eVects of individual environmental variables one by one (both simple and squared terms) on RGRs
of the diVerent species groups, and selected those which did
not result in an increase in AIC. We built multiple starting
models using the selected variables, and included biologically
573
reasonable squared and interaction terms. Starting models
were simpliWed using stepwise variable selection minimizing the AIC; the stepwise search was performed in both
directions. To analyse the eVect of species interactions on
RGRs, we then extended the models and tested whether
including the cover of surrounding epiphytes improved the
models; we even tested squared terms and interaction
eVects with the selected environmental variables. We tested
a number of diVerent predictors: total cover, bryophyte
cover, lichen cover and cover of pleurocarpous mosses; the
cover variable which resulted in the highest decrease in
AIC was included in the Wnal models. We considered all
models with an equal or lower number of parameters than
the best model, and AIC < AICnull and AIC < 10.
We tested whether the presence of sporophytes or asexual diaspores aVected RGRs of sexually and asexually
reproducing species, respectively. Again, we used linear
mixed-eVect models. The presence of sporophytes or asexual diaspores was treated as a Wxed factor, and species as a
random factor.
We used generalized linear mixed-eVect models with a
binomial error distribution and a logit link function to test,
Wrst, whether there is a threshold size in colony area for the
onset of sexual and asexual reproduction, and second,
whether sexual and asexual reproduction is aVected by
environmental variables and cover of other epiphytes. Original colony area in 2004 and habitat variables, respectively,
were treated as Wxed factors, species as a random factor,
and the presence of sporophytes and asexual diaspores
(0, 1), in either 2004 or 2005, respectively, as a dependent
variable. Colony area was ln transformed prior to the analysis.
As before, when analysing the eVects of habitat variables,
we Wrst tested the eVects of individual environmental and
cover variables on the presence of sporophytes or asexual
diaspores. Bark pH, however, was the only variable that did
not result in an increase in AIC. To predict when diVerent
species start to reproduce, we Wtted generalized linear
models with a binomial error distribution and a logit link
function with the presence of sporophytes or asexual diaspores in either 2004 or 2005 as the response variable and ln
colony area as predictor variable.
We used the free software R 2.8.1 (R Development
Core Team 2008) for the statistical analyses, with the addon libraries lme4 version 0.999375-28 (Bates et al. 2008),
nlme version 3.1-90 (Pinheiro et al. 2008), and MASS version
7.2-45 (Venables and Ripley 1999).
Results
The mean RGRs of pleurocarous mosses and liverworts
varied between 0.16 and 0.42 (Fig. 1). The two acrocarpous
species did not show any net growth, probably because they
123
574
Oecologia (2009) 161:569–579
reach a size at which branching and growth is balanced by
losses of colony components. As we could not distinguish
between colony growth and loss of colony components,
these species were excluded from further analyses.
Comparing RGRs of the diVerent species groups, the
RGRs did not diVer between asexually and sexually reproducing species (Table 1), but did diVer among life history
strategy groups: the linear mixed-eVect model predicted a
decrease in the RGRs from colonists (0.40) over perennial
stayers (0.31) to long-lived shuttle species (0.23) (Table 1).
Bark pH was the most important predictor of RGR for
all species groups (Table 2, S2). We often observed a quadratic relationship between RGR and bark pH, but this was
sometimes obscured by interaction eVects with other environmental or cover variables (S2). RGR was positively
aVected by habitat factors that inXuence water availability,
e.g. soil moisture and growth height. Either the cover of all
epiphytes or all pleurocarps aVected RGRs of all species
groups except colonists; positive eVects were most common, but even quadratic relationships and interaction
eVects with bark pH occurred (Table 2, S2). Neither sexually and asexually reproducing species nor perennial stayers and long-lived shuttles diVered in sensitivity of RGRs to
habitat conditions. The RGRs of colonists were aVected by
neither environmental factors nor species interactions
(Table 2, S2).
The presence of sporophytes had a negative eVect on
RGRs in sexually reproducing species, and the mixed-eVect
model predicted a decrease in RGRs by 0.09. In contrast,
the presence of gemmae did not aVect RGRs in asexually
reproducing species (Table 3).
The probability of the presence of both sporophytes and
asexual diaspores increased with increasing colony size, but
this was much more pronounced in sexually than in asexually reproducing species (Table 4). Bark pH aVected the
presence of sporophytes (quadratic eVect, AIC = ¡5.4),
but not the presence of gemmae or gemmae-like branchlets
(quadratic eVect, AIC = 3). In most sexually reproducing
species, generalized linear models of sporophyte production in relation to colony area had a AIC < ¡7 (S3). The
reproductive size predicted by the Wtted models increased
from colonists (Pylaisia polyantha, no eVect), to short-lived
shuttles (Orthotrichum speciosum, 26 mm2, AIC =
¡15.3) to long-lived shuttles (Radula complanata, 145 mm2,
AIC = ¡17.3; F. dilatata, 1,813 mm2, AIC = ¡7.1;
N. pennata, 4,091 mm2, AIC = ¡17.2). Perennial stayers
did not reproduce at all in our study. Among asexually
reproducing species, we observed an eVect of colony area
on gemmae production in Leucodon sciuroides only, and
the predicted reproductive size of this species (68 mm2,
AIC = ¡4.0) was smaller than that of the sexually reproducing pleurocarps and liverworts (S3).
Discussion
Growth rates of species with diVerent dispersal
and life history strategies
Even though we only studied colony growth and reproduction for 1 year, our results are likely to be representative: the mean RGR of N. pennata (Fig. 1) was the same
as the mean annual growth rate, 18% year¡1, suggested for
the species under long-term average precipitation
(544 mm year¡1) within the region (Wiklund and Rydin
2004a).
Table 1 DiVerences in relative growth rate (RGR) between asexually and sexually reproducing species and colonists, long-lived shuttles and
perennial stayers tested by linear mixed-eVect models
RGRs of sexually compared to asexually reproducing species
RGRs of long-lived shuttles and perennial stayers compared
to colonists
EVect
EVect
CoeYcient (SE)
df
Fixed
CoeYcient (SE)
df
Fixed
Intercept
Sexual
0.2912 (0.0485)
¡0.0052 (0.0580)
Random
425
Intercept
8
Long-lived shuttles
¡0.1651 (0.0456)
8
Perennial stayers
¡0.0892 (0.0499)
425
8
Random
speciesa
0.0761
speciesa
AIC
1.99
AICb
b
0.3971 (0.0385)
0.0417
¡4.62
Species reproduction modes and life history strategies were treated as Wxed factors and species as a random factor. CoeYcients and SEs for the
parameter estimates of the Wxed factors are given. Number of observations = 435, number of species = 10
a
The SD species reXects diVerences in the intercept of the random factor
b
Akaike's information criteria (AIC) is the diVerence in AIC between the full model (including the group variable) and the null model (including the random factor only). For AIC > ¡2 the level of empirical support for the full model is not higher than for the null model (Burnham and
Anderson 2002)
123
Oecologia (2009) 161:569–579
575
Table 2 Multiple linear mixed-eVect models for RGRs of diVerent
species groups with environmental and cover variables as predictors
Model
k
AIC
Table 2 continued
Model
AIC
k
AIC
5
¡68.00
4.61
7
¡72.61
0.00
3
40.80
1.22
5
39.58
0.00
AIC
Environment
All species, n = 435, 10 species
Fixed: pH(¡), soil moisture(+)
Null model
Random: species = 0.0485
Random: species = 0.0762
3
¡17.23
18.31
7
¡32.27
3.27
Environment
Fixed: pH(squared), soil moisture(+),
pH £ soil moisture(¡)
Colonists, n = 89, 2 species
Environment and epiphyte cover
Null model
8
¡35.54
0.00
Fixed: pH(squared)
Random: species = 0.0006
Sexual species, n = 305, 7 species
Null model
3
¡4.39
8.16
5
¡12.55
1.76
6
¡14.31
0.00
3
¡7.26
3.52
5
¡8.09
2.69
6
¡10.78
0.00
3
¡10.78
25.91
8
¡25.37
11.32
10
¡36.69
0.00
3
¡65.18
7.43
Environment
Fixed: pH(squared)
Random: species = 0.0879
Environment and epiphyte cover
Fixed: pH(squared), cover epiphytes(+)
Random: species = 0.0905
Asexual species, n = 130, 3 species
Null model
Random: species = 0.0598
Random: species = 0.0005
Environment
Random: species = 0.0772
Random: species = 0.0821
Fixed: pH(¡), soil moisture(+),
cover pleu(+), pH £ cover pleu(¡)
Random: species = 0.0491
Random: species = 0.0763
Fixed: pH(+), soil moisture(+),
cover pleu(+), pH £ soil moisture(¡),
pH £ cover pleu(¡)
Environment and epiphyte cover
Environmental and cover variables were treated as Wxed factors and
species as a random factor. Signs of the eVects of the Wxed factors are
given in parentheses. The SD species reXects diVerences in the intercept
of the random factor. By deWnition, the best model has AIC = 0, and
models with AIC < 2 are generally worthy of consideration. The
level of empirical support for models AIC > 4 is considerably less,
and essentially absent for models with AIC > 10 (Burnham and
Anderson 2002). Null models (including the random factor only) and
models with the lowest AIC (including environmental and cover
variables) are presented. A complete list of all models with an equal or
lower number of parameters than the best model, and AIC < 10 and
AICi < AIC0, is given in Appendix S2. k Number of parameters; for
other abbreviations, see Table 1
Environment
Fixed: pH(squared)
Random: species = 0.0675
Environment and epiphyte cover
Fixed: pH(+), cover pleu(+),
pH £ cover pleu(¡)
Random: species = 0.0877
Perennial stayers, n = 127, 3 species
Null model
Random: species = 0.0465
Environment
Fixed: pH(¡), growth height(¡),
DBH(+), pH £ growth height(+),
pH £ DBH(¡)
Random: species = 0.0535
Environment and epiphyte cover
Fixed: pH(+), growth height(+),
DBH(+), cover epiphytes(squared),
pH £ DBH(¡),
pH £ cover epiphytes(¡)
Random: species = 0.0721
Long-lived shuttles, n = 219, 5 species
Null model
Random: species = 0.0542
There were no indications of either lower growth rates
for sexually reproducing species compared with those
reproducing asexually, or that species with frequent sexual
reproduction should have low growth rates: contrary to our
hypothesis, colonists actually had higher RGRs than shuttles and perennial stayers. It is most likely their short life
span that prevents colonists from covering larger areas on
trees.
High growth rates are important for local population persistence, since small colonies run a higher extinction risk
(Snäll et al. 2005), but generally, the importance of growth
rates may decrease with increasing habitat dynamics. For
tropical epiphylls inhabiting leaves with a duration time of
typically less than 18 months, Zartman and Shaw (2006)
found that local growth rates were not important for metapopulation persistence. Our study suggests that an important eVect of high growth rates is to reduce the time to reach
reproductive size in colonists, whereas in perennial stayers,
which also had rather high RGRs, the most important eVect
is to lower the local extinction risk. Overall, it is questionable whether species diVerences in growth rate are a major
force in structuring the epiphyte metacommunity.
123
576
Oecologia (2009) 161:569–579
Table 3 Linear mixed-eVect models of the eVects of sporophyte and gemmae (or gemmae-like branchlet) production on RGRs
Sexual speciesa: eVects of sporophyte production on RGR
EVect
Asexual speciesa: eVects of gemmae production on RGR
CoeYcient (SE)
DF
EVect
Fixed
DF
Fixed
Intercept
Sporophytes
0.3105 (0.0379)
297
Intercept
0.2780 (0.0888)
126
¡0.0913 (0.0391)
297
Gemmae
0.0141 (0.0845)
126
Random
speciesb
AIC
CoeYcient (SE)
c
Random
0.0894
¡3.35
speciesb
0.0597
AIC
1.97
c
CoeYcients and SE for the parameter estimates of the Wxed factors are given. For sexual species number of observations = 305, number of
species = 7; for asexual species number of observations = 130, number of species = 3. For abbreviations, see Table 1
The presence of sporophytes (0, 1) and gemmae (0, 1), respectively, was treated as a Wxed factor
a
Species was treated as a random factor
The SD species reXects diVerences in the intercept of the random factor
AIC is the diVerence in AIC between the full model (including the presence of sporophytes or gemmae) and the null model (including the
random factor only)
b
c
Table 4 Generalized linear mixed-eVect models with a binomial error
distribution and a logit link function of presence of sporophytes (number of observations = 360, number of species = 8) and gemmae or
Presence of sporophytes
EVect
Presence of gemmae or gemmae-like branchlets
CoeYcient (SE)
df
Fixed
EVect
CoeYcient (SE)
df
Fixed
Intercept
Colony area (ln)
¡7.968 (1.6318)
0.9082 (0.1680)
351
Intercept
351
Colony area (ln)
Random
species
AIC
gemmae-like branchlets (number of observations = 161, number of
species = 4) in either 2004 or 2005 as predicted by colony area
a
b
¡0.5435 (2.2968)
156
0.8358 (0.3897)
156
Random
3.2997
¡33.98
speciesa
AICb
1.9607
¡3.63
CoeYcients and SEs for the parameter estimates of the Wxed factors are given
Original colony area in 2004 (mm2, ln transformed) was treated as a Wxed factor and species as a random factor
a
b
The SD species reXects diVerences in the intercept of the diVerent species
AIC is the diVerence in AIC between the full model (including colony area) and the null model (including the random factor only)
Sensitivity of growth and reproduction to habitat conditions
Our results did not support the hypothesis that growth
of sexually reproducing species should be particularly
sensitive to habitat conditions. Previously documented
eVects of habitat conditions on species richness, occupancy and abundance in sexual species (Löbel et al.
2006a, b, 2009) are thus more likely to be caused by
stronger ecophysiological sensitivity of spore germination than of gemmae establishment. Further, as in other
systems (e.g. Pohjamo et al. 2006), sexual reproduction
in epiphytes critically depends on habitat conditions,
whereas asexual reproduction does not. Comparing life
strategies, there were no indications that long-lived
shuttles should be more habitat-sensitive than perennial
stayers, and contrary to our hypothesis, the growth of
123
colonists was not aVected by environmental variables at
all.
EVects of pH on RGRs and sporophyte production are in
line with earlier studies on species richness, which showed
either positive or quadratic relationships to bark pH (Löbel
et al. 2006a). One possible reason why Wiklund and Rydin
(2004a) did not observe any eVects of pH on RGRs in
N. pennata was the inclusion of several host tree species,
thus obscuring any pH eVect due to other diVerences among
trees. EVects of soil moisture and light on RGRs and reproduction of the diVerent species groups were surprisingly
low, considering that growth in bryophytes critically hinges
on humidity (Wiklund and Rydin 2004a; Ingerpuu et al.
2007). A likely explanation is that the range of moisture
and light conditions in our study was rather small: all colonies originated from the same old-growth forest.
Oecologia (2009) 161:569–579
EVects of interspeciWc interactions on growth
and reproduction
Our study suggests that colony growth of epiphytes is
aVected by interaction with other epiphytes, especially
pleurocarpous mosses, but there are no indications for
diVerences in competitive abilities among species groups.
Interaction eVects were rather weak and do not support a
competition-colonization trade-oV: the growth of colonists was not aVected by neighbouring epiphytes, and the
growth of shuttle species was not more aVected than that
of perennial stayers. In fact, the eVects of neighbours were
mostly positive. In bryophytes, contact with neighbours
can provide protection and a humid, shady microclimate
(Mulder et al. 2001; Rixen and Mulder 2005; Rydin
2009), and a general decrease in epiphyte cover could lead
to a decrease in species richness. Further, production of
neither sporophytes nor gemmae were aVected by neighbours. Hence, it is unlikely that competition is an important force for driving epiphyte metacommunity dynamics.
This may be true for many sessile communities, where
competition is a neighbourhood phenomenon, rather than
a Lotka-Volterra type of global interaction (Bengtsson
et al. 1994). In general, host trees often have large areas of
unoccupied bark (Snäll et al. 2003; Löbel et al. 2006a, b),
and mats of pleurocarpous mosses commonly fall oV and
create gaps for new colonizations on old trees. Thus, new
colonization of host trees may be limited more by the ecophysiological constraints of establishment (Wiklund and
Rydin 2004b), and thus substrate quality, than by the lack
of empty space.
Trade-oVs between growth and reproduction,
age at Wrst reproduction
The negative eVect of sporophyte production on RGR indicates a trade-oV between colony growth and sexual reproduction. This conforms with studies indicating high costs of
sexual reproduction in mosses (Ehrlén et al. 2000; Bisang
and Ehrlén 2002; Rydgren and Økland 2002), which, however, obviously do not translate to a generally lower RGR
in sexual species compared to asexual.
Our results suggest that there is a threshold in colony
size and age before a species starts to reproduce sexually,
whereas there is no such threshold for asexual reproduction.
Similar patterns have been observed within species: in the
epixylic liverwort Anastrophyllum hellerianum, asexual
reproduction occurred in smaller and younger shoots than
sexual reproduction (Pohjamo and Laaka-Lindberg 2004).
This conWrms our hypothesis that delayed sexual, but not
asexual reproduction is the reason for the strong eVect of
the historical but not present landscape structure on species
richness of sexually dispersed epiphytes at a regional scale
577
(Löbel et al. 2009). Trade-oVs between dispersal distance
and age at Wrst reproduction could thus be an important factor for the evolution of these alternative strategies.
Among species reproducing sexually, colony size at Wrst
reproduction increased from colonists to short- and longlived shuttles to perennial stayers. There is a clear link
between the colony size at Wrst sexual reproduction and the
observed impact of the historical landscape structure on
regional species occupancy and local abundance pattern
(Löbel et al. 2006b): both increased from P. polyantha via
O. speciosum and F. dilatata to the most sensitive N. pennata. The size of N. pennata colonies starting to reproduce
for the Wrst time was within the same range (13–60 cm2) as
observed in an earlier study (12–79 cm2), corresponding to
an age of 19–29 years (Wiklund and Rydin 2004a). The
sexual species Homalia trichomanoides, Anomodon longifolius and Isothecium alopecuroides did not produce any
sporophytes in our study. The few fertile colonies found on
host trees within the region (cf. Löbel et al. 2006a, b, 2009)
are very large and grow on very old trees, and consequently
for these species, the historical landscape and forest age
have a strong eVect on their occupancy and local abundance
(Löbel et al. 2006b).
Evolution of dispersal strategies in patch-tracking epiphyte
metacommunities
To link back to our hypotheses, we found that the dispersal
limitation of large asexual diaspores is compensated by an
early and frequent reproduction, and insensitivity of reproduction, to habitat conditions. However, growth rates and
sensitivity of growth to habitat conditions and competitive
abilities did not diVerentiate between species with diVerent
dispersal modes. Age at Wrst sexual reproduction did
increase from colonists, to short- and long-lived shuttles to
perennial stayers; however, this was not coupled to diVerences in growth rate, competitive ability or sensitivity to
habitat conditions.
Host trees and deciduous forest stands in the coniferous
landscape are patchy, temporal and undergo change in habitat quality during succession. The conceptual model in
Fig. 2 suggests how epiphyte dispersal and life history
strategies have evolved in response to these constraints.
Trade-oVs in species traits are possibly shaped by conXicting selection pressures imposed by the habitat rather than
by species interactions.
Asexual reproduction is regarded as an adaptation to the
temporality of patches and successional changes. First, it
allows for early reproduction, second, production of asexual diaspores is rather insensitive to habitat quality, and
third, establishment rates from large, asexual diaspores are
higher and less habitat sensitive than from spores. Hence,
the time under which a host tree can be colonized, and
123
578
Spores
20 µm
Asexual diaspores
50 µm
Metzgeria furcata
250 µm
Gemma shuttles
Reproduction mode
Platygyrium repens
Diaspore size
Increasing impact of habitat connectivity
Leucodon sciuroides
Increasing impact of habitat quality
Fig. 2 Conceptual model
illustrating possible relationships between dispersal strategies (sexual via spores, asexual
via gemmae or gemmae-like
branchlets), life history strategies based on During (1979,
1992) and Siebel and During
(2006) and species responses to
habitat turnover, connectivity
and quality
Oecologia (2009) 161:569–579
Bryum flaccidum
Gemma colonists
Life-history strategy
Asexual
Frullania dilatata
Sexual
Radula complanata
Long-lived shuttles
Short-lived shuttle
Orthotrichum speciosum
Neckera pennata
Perennial stayers
Sexual colonist
Pylaisia polyantha
Homalia trichomanoides
Isothecium alopecuroides
Anomodon longifolius
Anomodon longifolius
Colony age at first reproduction
Increasing impact of habitat turnover
under which reproduction is possible, is longer for asexually than for sexually reproducing species. But asexual
diaspores are larger and fewer and have distinctly shorter
dispersal distances than spores. This indicates two main
trade-oVs: dispersal distance vs. early and frequent reproduction, and dispersal distance vs. sensitivity to habitat
quality. Rather than a simple trade-oV between growth and
reproduction, selection for high growth rates decreases the
time to reach reproductive size (colonists) and the local
population extinction risk (perennial stayers).
We encourage complementing the classic bryophyte life
history strategies by taking asexual reproduction into
account (Siebel and During 2006). In the habitat template
(Fig. 2), sexual colonists are least vulnerable to both habitat
patchiness and dynamics, but their habitat sensitivity can
prevent reproduction or establishment: in central Europe,
sporophytes in P. polyantha have become very rare, possibly due to high air pollution (Frahm and Frey 2004).
Gemma shuttles are a group with strong dispersal limitation, whereas gemma colonists have a better potential to
colonize distant habitat—in the rare cases of sexual reproduction they have rather small spores. Perennial stayers
with small spores are insensitive to habitat connectivity, but
their delayed reproduction is problematic if habitat turnover
is fast. Critical species are long-lived shuttles, such as
N. pennata, with both large spores and late onset of sexual
reproduction. It is interesting that long-lived shuttles with
extremely large spores have developed combined dispersal
strategies with both sexual and asexual dispersal.
Loehle (2000) showed how trees with alternative life
history sets can coexist when the disturbance patterns
123
Xuctuate in space and time, and the same may well apply to
the epiphytes. Changes in disturbance regimes as imposed
by human land-use could disturb such equalizing mechanisms of species coexistence and thus lead to distinct
changes in metacommunity structure. Modern forestry has
led to a reduction of the overall amount of deciduous trees
in the landscape, which causes most harm to asexually dispersed species, whereas cutting old-growth forests of high
age and quality, as well as the suppression of the development of new large, deciduous stands of high quality, is most
harmful to sexually dispersed species.
Acknowledgements We thank Irene Bisang, Heinjo During, Tord
Snäll, Sebastian Sundberg and Lars Söderström for useful comments
on the manuscript, and Scott Spellerberg for revising the English.
Financial support was received from FORMAS. The work conforms to
the legal requirements of the countries in which it was carried out,
including those relating to conservation and welfare.
References
Akaike H (1974) A new look at statistical model identiWcation. IEEE
Trans Autom Contr AC 19:716–722
Amarasekare P (2003) Competitive coexistence in spatially structured
environments: a synthesis. Ecol Lett 6:1109–1122
Anonymous (1997) Fältinstruktion för ståndortskarteringen. Institutionen för Skoglig Marklära, SLU, Uppsala
Bates D, Maechler M, Dai B (2008) lme4 version 0.999375-28. http://
www.lme4.r-forge.r-project.org/
Bengtsson J, Fagerström T, Rydin H (1994) Competition and coexistence in plant communities. Trees 9:246–250
Bisang I, Ehrlén J (2002) Reproductive eVort and cost of sexual reproduction in female Dicranum polysetum. Bryologist 105:384–397
Oecologia (2009) 161:569–579
Bossuyt B, Honnay O (2006) Interactions between plant life span, seed
dispersal capacity and fecundity determine metapopulation viability in a dynamic landscape. Landscape Ecol 21:1195–1205
Burnham KP, Anderson DR (2002) Model selection and multimodel
inference, 2nd edn. Springer, New York
Cadotte MW, Mai DV, Jantz S, Collins MD, Keele M, Drake JA
(2006) On testing the competition-colonization trade-oV in a multispecies assemblage. Am Nat 168:704–709
Chase JM, Amarasekare P, Cottenie K, Gonzales A, Holt RD, Holyoak
M, Hoopes MF, Leibold MA, Loreau M, Mouquet N, Shurin JB,
Tilman D (2005) Competing theories for competitive metacommunities. In: Holyoak M, Leibold MA, Holt RD (eds) Metacommunities. The University of Chicago Press, Chicago, pp 335–354
Chesson P (2000) Mechanisms of maintenance of species diversity.
Annu Rev Ecol Syst 31:343–366
Dettki H, Esseen P-A (2003) Modelling long-term eVects of forest
management on epiphytic lichens in northern Sweden. For Ecol
Manage 175:223–238
Dierßen K (2001) Distribution, ecological amplitude and phytosociological characterization of European bryophytes. Cramer, Berlin
During HJ (1979) Life strategies of bryophytes: a preliminary review.
Lindbergia 5:2–18
During HJ (1992) Ecological classiWcation of bryophytes and lichens.
In: Bates JW, Farmer AM (eds) Bryophytes and lichens in a
changing environment. Clarendon Press, Oxford, pp 1–31
During HJ (2007) Relations between clonal growth, reproduction and
breeding system in the bryophytes of Belgium and The Netherlands. Nova Hedw Beih 131:133–145
Ehrlén J, Bisang I, Hedenäs L (2000) Costs of sporophyte production
in the moss, Dicranum polysetum. Plant Ecol 149:207–217
Eriksson P (1997) Ekologisk landskapsplanering i Vällenområdet,
Rapport 5. Upplandsstiftelsen, Uppsala
Frahm J-P, Frey W (2004) MoosXora, 2nd edn. Ulmer, Stuttgart
Frazer GW, Canham CD, Lertzman KP (1999) Gap Light Analyzer
(GLA), Version 2.0: Imaging software to extract canopy structure
and gap light transmission indices from true-colour Wsheye photographs, users manual and program documentation. Simon Fraser
University, Burnaby, Insitute of Ecosystem Studies, Millbrook
Hill MO, Preston CD, Bosanquet SDS, Roy DB (2007) Bryoatt—attributes of British and Irish mosses, liverworts and hornworts. CEH,
NERC, Norwich
Holyoak M, Leibold MA, Mouquet NM, Holt RD, Hoopes MF (2005)
Metacommunities: a framework for large-scale community ecology. In: Holyoak M, Leibold MA, Holt RD (eds) Metacommunities. The University of Chicago press, Chicago, pp 1–31
Ingerpuu N, Vellak K, Möls T (2007) Growth of Neckera pennata, an
epiphytic moss of old-growth forests. Bryologist 110:309–318
Johst K, Brandl R, Eber S (2002) Metapopulation persistence in dynamic landscapes: the role of dispersal distance. Oikos 98:263–270
Karlsson S (2007) Image. http://www.sk-biometri.se
Kimmerer RW (1994) Ecological consequences of sexual versus asexual reproduction in Dicranum Xagellare and Tetraphis pellucida.
Bryologist 97:20–25
Kneitel JM, Chase JM (2004) Trade-oVs in community ecology: linking spatial scales and species coexistence. Ecol Lett 7:69–80
Laaka-Lindberg S, Korpelainen H, Pohjama M (2003) Dispersal of
asexual propagules in bryophytes. J Hattori Bot Lab 93:319–330
Laaka-Lindberg S, Korpelainen H, Pohjama M (2006) Spatial distribution of epixylic hepatics in relation to substrate in a boreal
old-growth forest. J Hattori Bot Lab 100:311–323
Leibold MA, Miller TE (2004) From metapopulations to metacommunities. In: Hanski I, Gaggiotti OE (eds) Ecology, genetics, and
evolution of metapopulations. Elsevier, Amsterdam, pp 133–150
579
Löbel S, Snäll T, Rydin H (2006a) Species richness patterns and metapopulation processes—evidence from epiphyte communities in
boreo-nemoral forests. Ecography 26:169–182
Löbel S, Snäll T, Rydin H (2006b) Metapopulation processes in epiphytes inferred from patterns of regional distribution and local
abundance in fragmented forest landscapes. J Ecol 94:856–868
Löbel S, Snäll T, Rydin H (2009) Mating system, reproduction mode
and diaspore size aVect metacommunity diversity. J Ecol 97:176–
185
Loehle C (2000) Strategy space and the disturbance spectrum: a lifehistory model for tree species coexistence. Am Nat 156:14–33
Miller TE, Kneitel JM (2005) Inquiline communities in pitcher plants
as a prototypical metacommunity. In: Holyoak M, Leibold MA,
Holt RD (eds) Metacommunities. The University of Chicago
press, Chicago, pp 122–145
Mulder CPH, Uliassi DD, Doak DF (2001) Physical stress and diversity-productivity relationships: the role of positive interactions.
PNAS 98:6704–6708
Pinheiro J, Bates D, DebRoy S, Sarkar D (2008) nlme version 3.1-90.
R Foundation for Statistical Computing, Vienna
Pohjamo M, Laaka-Lindberg S (2004) Demographic structure of a
leafy epixylic hepatic Anastrophyllum hellerianum (Nees ex Lindenb.) R.M. Schust. Plant Ecol 173:3–81
Pohjamo M, Laaka-Lindberg S, Ovaskainen O, Korpelainen H (2006)
Dispersal potential of spores and asexual diaspores in the epixylic
hepatic Anastrophyllum hellerianum. Evol Ecol 20:415–430
R Development Core Team (2008) R: a language and environment for
statistical computing. R Foundation for Statistical Computing
Rixen C, Mulder CPH (2005) Improved water retention links high
species richness with increased productivity in artic tundra moss
communities. Oecologia 146:287–299
Rydgren K, Økland RH (2002) Ultimate costs of sporophyte production in the clonal moss Hylocomium splendens. Ecology 83:1573–
1579
Rydin H (2009) Population and community ecology of bryophytes. In:
Shaw AJ, GoYnet B (eds) Bryophyte biology, 2nd edn. Cambridge University Press, Cambridge, pp 393–444
Siebel HN, During HJ (2006) Beknopte mosXora van Nederland en
België. K.N.N.V., Utrecht
Snäll T, Ribeiro PJ Jr, Rydin H (2003) Spatial occurrence and colonizations in patch-tracking metapopulations: local conditions
versus dispersal. Oikos 103:566–578
Snäll T, Fogelqvist J, Ribeiro PJ Jr, Lascoux M (2004a) Spatial genetic
structure in two congeneric epiphytes with diVerent dispersal
strategies analysed by three diVerent methods. Mol Ecol
13:2109–2119
Snäll T, Hagström A, Rudolphi J, Rydin H (2004b) Distribution pattern
of the epiphyte Neckera pennata on three spatial scales—
importance of past landscape structure, connectivity, and local
conditions. Ecography 27:757–766
Snäll T, Ehrlén J, Rydin H (2005) Colonization and extinction
dynamics of an epiphyte metapopulation in a dynamic landscape.
Ecology 86:106–115
Venables WN, Ripley BD (1999) Modern applied statistics with
S-PLUS. Springer, Berlin
Wiklund K, Rydin H (2004a) Colony expansion of Neckera pennata:
modelled growth rate and eVect of microhabitat, competition, and
precipitation. Bryologist 107:293–301
Wiklund K, Rydin H (2004b) Ecophysiological constraints on spore
establishment in bryophytes. Funct Ecol 18:907–913
Zartman CE, Shaw AJ (2006) Metapopulation extinction thresholds in
rain forest remnants. Am Nat 167:177–189
123