1
Vittoz and Engler (2007) Botanica Helvetica 117: 109-124
Seed dispersal distances: a typology based on dispersal modes and plant traits
Pascal Vittoz
1
and Robin Engler
2
1
Faculty of Geosciences and Environment, University of Lausanne, Bâtiment Biophore, CH-1015
Lausanne, Switzerland; e-mail: pascal.vittoz@unil.ch (corresponding author)
2
Department of Ecology and Evolution, University of Lausanne, Bâtiment Biophore, CH-1015
Lausanne, Switzerland; e-mail: robin.engler@unil.ch
Abstract
Vittoz P. and Engler R. 2007. Seed dispersal distances: a simplification for data analyses and
models. Bot. Helv. 1xx: xx–xx.
The ability of plants to disperse seeds may be critical for their survival under the current constraints
of landscape fragmentation and climate change. Seed dispersal distance would therefore be an
important variable to include in species distribution models. Unfortunately, data on dispersal
distances are scarce, and seed dispersion models only exist for some species with particular
dispersal modes. To overcome this lack of knowledge, we propose a simple approach to estimate
seed dispersal distances for a whole regional flora. We reviewed literature about seed dispersal in
temperate regions and compiled data for dispersal distances together with information about the
dispersal mode and plant traits. Based on this information, we identified seven "dispersal types"
with similar dispersal distances. For each type, upper limits for the distance within which 50% and
99% of a species' seeds will disperse were estimated with the 80
th
percentile of the available values.
These distances varied 5000-fold among the seven dispersal types, but generally less than 50-fold
within the types. Thus, our dispersal types represented a large part of the variation in observed
dispersal distances. The attribution of a dispersal type to a particular species only requires
information that is already available in databases for most Central European species, i.e. dispersal
vector (e.g. wind, animals), the precise mode of dispersal (e.g. dyszoochory, epizoochory), and
species traits influencing the efficiency of dispersal (e.g. plant height, typical habitats). This
typology could be extended to other regions and will make it possible to include seed dispersal in
species distribution models.
Key words: Anemochory, anthropochory, autochory, hydrochory, plant migration, zoochory.
Introduction
Plant dispersal has attracted scientists since long ago (Darwin 1859; Schmidt 1918; Ridley
1930; Müller-Schneider 1983) and is particularly relevant with relation to human-driven
environmental changes. For example, the survival of plant metapopulations in fragmented
landscapes strongly depends on their dispersal potential (Fischer et al. 1996; Couvreur et al. 2004;
Soons and Ozinga 2005), and the predicted global warming will require considerable migration
rates for plant species to remain under similar climatic conditions (Malcolm et al. 2002).
Nevertheless, most models attempting to predict future plant distributions did not include dispersal,
considering it as unlimited (Guisan and Theurillat 2000; Thuiller et al. 2005). Even without
constraints on seed dispersal, these models already predict local extinctions, e.g. for isolated
populations in mountains (Guisan and Theurillat 2000; Dirnbock et al. 2003; Thuiller et al. 2005).
The actual extinction rates might be even higher if plant species cannot keep pace with rapid
climate change due to limited dispersal. A more precise assessment of plant species extinction risk
2
Vittoz and Engler (2007) Botanica Helvetica 117: 109-124
thus calls for the incorporation of plant dispersal potential (Pitelka et al. 1997; Davis et al. 1998;
Ronce 2001).
Many studies have measured or estimated dispersal distances of plants in the field (Schneider
1935; Stöcklin and Bäumler 1996; Jongejans and Telenius 2001), and several mathematical models
have been developed to estimate these distances (Tackenberg et al. 2003; Mouissie et al. 2005a;
Nathan et al. 2005; Soons and Ozinga 2005). However, all of these studies have considered only a
limited number of species or dispersal vectors. No dispersal distance data exist for a complete
regional flora. Müller-Schneider (1986) reviewed dispersal vectors for the entire flora of
Graubünden (East of Switzerland), but his work includes only few dispersal distances, most of
which stem from anecdotal observations. Likewise, Bonn and Poschlod (1998) and Bonn (2004)
wrote important syntheses on seed dispersal in Central Europe, but dispersal distances were only
provided for a few dispersal vectors, mainly from anecdotal observations. It is thus currently
impossible to conduct an assessment of the extinction risk of plant species under landscape
fragmentation or global warming that would take dispersal into account.
The distance over which plants disperse seeds depends on plant traits as well as environmental
conditions and varies strongly in time and space. This variability can be represented by a dispersal
curve (dispersal kernel), which gives the proportion of seeds reaching a given distance (Mouissie et
al. 2005a). However, it would be highly time consuming, if not impossible, to determine dispersal
kernels for each species of a region. Thus, a simplified approach is needed to estimate dispersal
distances for a whole regional flora. For example, if dispersal curves could be classified into a
limited number of types with similar dispersal distances, and if plant species could be attributed to
these "dispersal types" based on generally available plant traits, it would be possible to estimate
dispersal kernels for all of them.
In this paper, we develop such an approach for the Swiss flora based on an extensive review of
seed dispersal literature. We propose a typology of dispersal curves that can be applied to most
Swiss and Central European plants. This typology could be extended to other regions and could be
used to account for dispersal distances in species distribution models, enabling refined extinction
risk assessments to be made for large numbers of species.
Methods
Plant dispersal is generally achieved through seeds. These can be enclosed in fruits or larger
structures (usually called "diaspores"), but for the sake of simplification, the term “seed” will be
used here as a general denomination.
Data for seed dispersal distances were compiled by reviewing a large proportion of available
literature from Switzerland and other European countries, including monographies (Müller-
Schneider 1983, 1986), reviews and research articles. Swiss species or close relatives were
considered first priority, since our aim was to develop a typology for this region. However, other
species were included when data available for Swiss species were insufficient to assess dispersal
distances for certain dispersal modes (see below). The complete data set (ca. 300 values) is
presented in Appendix 1. Species nomenclature follows Aeschimann et al. (1996) for the Swiss
species.
The data set proved to be very heterogeneous. A small proportion of the distances had been
determined through experiments, detailed field observations of seed or seedling distributions, or
mathematical models. In such cases, it is often possible to calculate a dispersal kernel. However,
most of the available data represent isolated and often anecdotal observations, from which a precise
dispersal kernel cannot be derived. Some of these isolated observations clearly represented long-
distance dispersal events (LDD), i.e. extreme values reached only by a very small minority of seeds.
We therefore classified the data into three categories: (1) mean, mode or median values, (2)
maximum values (99th percentiles of distribution kernels) and (3) values for LDD (clearly above
the potential dispersal of 99% of the seeds). LDD values were excluded from the further analysis of
the data.
3
Vittoz and Engler (2007) Botanica Helvetica 117: 109-124
Our typology of dispersal curves was based on the dispersal modes recognised by Müller-
Schneider (1983). The English translation of Müller-Schneider's German terminology generally
follows Bonn et al. (2000). Müller-Schneider's (1983) classification of dispersal modes is primarily
based on the dispersal vector (wind, water, animals, etc.), with additional subdivisions for the
differing ways in which seeds are released and transported (e.g. on the fur or after ingestion by
animals). Additional subdivisions were made for dispersal modes whose efficiency clearly depends
on supplementary factors: plant height, pappus efficiency and environing vegetation structure for
anemochory, and vector size for zoochory. Of the numerous possible subdivisions, only those
considered most relevant were retained for our classification, as explained in the next section. This
yielded a total of 21 refined dispersal modes (Tab. 1).
Tab. 1. Dispersal distances for seven dispersal types, estimated as the upper limits of the distances within which 50%
and 99% of the seeds of a plant population are dispersed. Note that actual dispersal distances will usually be lower than
those given here (cf. Fig. 1). The dispersal distances were estimated from the 80
th
percentile of the data compiled in Fig.
1 as well as additional qualitative information as explained in the text ('Dispersal modes and evaluation of published
dispersal distances'). The dispersal modes included in each dispersal type are indicated; they are based on dispersal
vectors (categories in parentheses) and plants traits that influence the efficiency of dispersal.
Type Corresponding dispersal modes
50% 99%
1 0.1 1 Blastochory (autochory)
Boleochory (anemochory) for species < 30 cm
Ombrochory (hydrochory)
2 1 5 Ballochory (autochory)
Cystometeorochory (anemochory)
Chamaechory (anemochory) for fruits in grassland
Boleochory (anemochory) for species > 30 cm
3 2 15 Pterometeorochory (anemochory) for herbs
Myrmecochory (zoochory)
Cystometeorochory (anemochory) ferns, Orchidaceae, Pyrolaceae, Orobanchaceae in forest
Trichometeorochory (anemochory) in forest or little efficient plumes
Epizoochory (zoochory) for small mammals
4 40 150 Chamaechory (anemochory) for seeds on snow or dry inflorescence
Pterometeorochory (anemochory) for trees
Dyszoochory (zoochory) for seeds not stocked and dispersed by small animals
5 10 500 Trichometeorochory (anemochory) in openland with efficient plumes
Cystometeorochory (anemochory) ferns, Orchidaceae, Pyrolaceae, Orobanchaceae in openland
6 400 1500 Dyszoochory (zoochory) for seeds stocked by large animals
Endozoochory (zoochory) for seeds eaten by birds and large vertebrates
Epizoochory (zoochory) by large mammals
7 500 5000 Agochory (anthropochory)
Dispersal distances [m]
Each dispersal distance in our data set was attributed to a dispersal mode, which was either the
mode for which the distance had been determined (if mentioned in the original study) or the main
dispersal mode of the species according to Müller-Schneider (1986). For species with more than
one dispersal mode, distances that could not be clearly related to one of the modes were excluded
from further data analysis. Dispersal types were then defined by grouping together dispersal modes
with similar dispersal distances. This was done graphically by plotting the mean and maximal
distances for each dispersal mode and identifying modes for which distances were in the same order
of magnitude (Fig. 1).
Finally, we estimated upper limits of the distances, within which 50% and 99% of the seeds
would disperse, by using the 80
th
percentiles of the available mean, mode or median values and of
the maximum values. Results were rounded to one significant digit to reflect their approximate
nature. Our aim was to provide a conservative estimate of the dispersal constraint experienced by
most species belonging to a dispersal type. Therefore we did not take the average of the published
values (Fig. 1), but rather the 80th percentile of the distribution, as this allowed us to exclude the
most extreme values. In some cases, a comparison of the results with qualitative information from
the literature or with the authors' experience indicated that the available data were not quite
4
Vittoz and Engler (2007) Botanica Helvetica 117: 109-124
representative for a certain dispersal type; values were then adjusted to obtain more realistic
estimates. Such decisions are explained in the next section of the text for the individual dispersal
modes.
Fig. 1. Distribution of the dispersal distances found in literature (Appendix 1) for each dispersal mode, and subdivision
of the data set into seven dispersal types. Diamonds are for mean, median or mode values, and crosses for 99% or
maximum values (without long-distance dispersal). Four retained maximum values of type 5 are outside of the graph:
1714 m, 2112 m, 2194 m and 3673 m. See Table 1 for definitions of the dispersal modes.
Dispersal modes and evaluation of published dispersal distances
Autochory
Autochorous plants disperse seeds without the help of an external vector. As a result, dispersal is
limited to very short distances.
In blastochory, the stem of the plant grows or crawls on the ground to deposit the seeds as far as
possible from the mother plant (e.g. Cymbalaria muralis, Polygonum aviculare, Veronica
hederifolia; Müller-Schneider 1983). No data were found in the literature, but since the dispersal
5
Vittoz and Engler (2007) Botanica Helvetica 117: 109-124
distance corresponds to the length of the stem, although species-specific, it is mostly very short and
blastochory can hence be classified as type 1 (Tab. 1). This dispersal mode is, however, frequently
completed by another one (Müller-Schneider 1986).
In ballochory, the explosion of the fruit ejects the seeds (ballistichory, ballistic dispersal). This
explosion may be due to the turgescence of tissues (Impatiens sp., Cardamine sp.) or the tension
between cells or different cell layers when the fruit is drying (Viola sp., Vicia sp., Lotus sp.).
Published values are scattered and very variable (maximum 0.89-6.2 m; Fig. 1). Ballochory is
classified in dispersal type 2 (Tab. 1).
Two further dispersal modes are barochory (seeds fall from the plant) and herpochory (seeds
creep on the soil by the movement of organs in a succession of dry and wet conditions). However,
since these strategies are not very efficient and always combined with other dispersal modes
(Müller-Schneider 1986), they were not retained here.
Anemochory
Anemochorous seeds are dispersed by wind, often with the help of specific organs. This
dispersal vector is the most studied as it is easily observable and measurable, at least over short
distances (e.g. Bullock and Clarke 2000; Jongejans and Telenius 2001). Moreover, it relies on
physical processes that can be translated into models (Tackenberg et al. 2003; Nathan et al. 2005;
Soons and Ozinga 2005). Anemochory is subdivided according to the organs used to slow down the
falling of seeds.
An air filled structure lightens small seeds in cystometeorochory (balloon-like). This dispersal
mode is little studied. Maximum calculated distances are below 2 m (Soons and Ozinga 2005), but
extreme values were measured up to 80 m for Calluna vulgaris (Bullock and Clarke 2000). This
mode is certainly less efficient in forests, as wind is weaker, but it seemed useless to subdivide
these already low values and thus cystometeorochory as a whole was attributed to type 2.
The tiny seeds of Orchidaceae, Pyrolaceae and Orobanchaceae also have a low falling velocity
(0.2-0.31 m/s for Orchidaceae; Müller-Schneider 1986). But only a calculated dispersion distance is
available (median 0.95 m and 99-percentile 14.7 m for Cephalanthera damasonium, Soons and
Ozinga 2005). However, because it is thought that very light seeds (<0.05 mg), even without
corresponding adaptation for anemochory, are as efficient in wind as plumed seeds (Bonn and
Poschlod 1998; Greene and Calogeropoulos 2002), we decided to classify these plant families with
trichometeorochory in type 5 in openland but decreased to type 3 for forest species (Tab. 1). Fern
spores can be included in cystometeorochory as well, but no data exist on their dispersal capacity
except a calculated distance of 330 km for Lycopodium sp. based on its very low falling velocity
(1.8 cm/s; Schmidt 1918). This value seems exaggerated and in the absence of a more precise value,
we attributed the ferns to the same types as orchids.
Plumed seeds are more efficient for wind dispersal. In trichometeorochory, seeds are
completed with a hairy structure (e.g. pappus) to reduce falling velocity. These organs have very
variable efficiency, however, with falling velocity varying from 8 cm/s for Epilobium angustifolium
to 165 cm/s for Pulsatilla alpina (Müller-Schneider 1986). With an arbitrary separation at 30 cm/s,
on the basis of our own observations, we distinguished species with less efficient plumes from those
with efficient plumes (long plumes for small seeds). The first group has maximum distances
between 1-15.7 (36) m, corresponding to type 3, and the second mainly between 20 and 179 m (Fig.
1). However, because some species have much higher calculated potentiality (e.g. up to 3600 m for
Typha latifolia; Soons and Ozinga 2005), we retained intermediate values and assigned
trichometeorochory to type 5. Forest species were classified with trichometeorochory for less
efficient plumes (type 3).
In pterometeorochory (or pterochory), seed dispersal is improved through wings. Trees are
frequent in this category, but herbs are present as well, with a generally higher falling velocity.
Because tree seeds are often large and easy to find, many available maximum dispersal distances
are to be classified as LDD (e.g. Müller-Schneider 1983). Reviewed maximum distances ranged
mainly between 80-314 m for trees and 1-12 m for herbs (Fig. 1). Pterometeorochory was thus
classified as dispersal type 3 for herbs and type 4 for trees (Tab. 1).
