The Quantitative Representation of Seed and Pollen Dispersal

Author(s): David R. Peart

Source: Ecology, Vol. 66, No. 3 (Jun., 1985), pp. 1081-1083
Published by: Ecological Society of America
Stable URL: http://www.jstor.org/stable/1940567
Accessed: 19/08/2010 22:44
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=esa.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology.

http://www.jstor.org

Notes

and

Comments

66(3),1985,pp.1081-1083
Ecology,
ofAmerica
? 1985bytheEcological
Society

direction from the source. Define the probability func?

tion:
P{r) =

THE

QUANTITATIVE
A TION
REPRESENT
POLLEN

OF

SEED

David R. Peart2 3
The purpose of this note is to compare two types of
distance-dispersal functions (density and probability
functions) describing dispersal of propagules from a
localized source. The results are intended to apply
mainly to wind or ballistic dispersal, but also have
some relevance to animal-dispersed plant species. I will
refer only to seeds in the quantitative treatment, but
the treatment applies equally well to pollen.
First, let D'(r,B) be the density of seeds (number per
unit area per seed dispersed), falling at distance r and
direction 6 from the source. Define the density func?
tion:
(1)

Then D(r) is the seedfall density per seed dispersed, at

distance r, averaged over all directions. (Note that this
is quite distinct from the statistical usage of the term
density in the expression "probability density func?
tion" [p.d.f.]). D(r) rather than D'(r,B) will be used in
the following analysis, as the focus will be on distance
rather than directional effects. Note, however, that in
Eq. 1, D(r) can be defined for any sector of size a (0 <
a < 2ir) for a more detailed analysis of asymmetric
dispersal functions. The results that follow would then
hold for any such sector.
D(r) satisfies the conditions:
i)

r
I

D(r) dA = 1, where dA is an increment of

area at radius r;
ii) the proportion of seeds falling between r{ and r2
is
D(r) dA.
J'A(r2)
l=A(rl)
Then

i:

D(r)2irr dr,

(3)

which corresponds to the density function in Eq. 1

above. Then P{r) is the probability that a seed released
from the source will arrive in an annular area of unit
width at distance r from the source. P{r) satisfies the
conditions:

AND

DISPERSAL1

?>(/?)= ?
D'(r,6)dd.
Zir J
Z7T
*Jo=0
0=0

F(r,B) dd,
J 0=0

(2)

since dA = 2wrdr is the annular area element.

Second, let P'(r, 6) be the p.d.f. describing the prob?
ability of arrival of seeds as a function of distance and

i)

ii)

P{r)dr=
J r=0
f

J
??

1;

P{r)dr = Pl2.

(4)

(5)

The quantitative differencebetween D{r) and P{r) arises simply because of the geometrical relation between
annular area and radial distance. From Eqs. 2 and 5,
P{r) = 2irr D{r).

(6)

It can be seen that the density function declines more

rapidly with distance than does the probability func?
tion. Further, P{r) must satisfy the conditions ofa p.d.f.
(Mendenhall and Scheaffer 1973). In particular, Eq. 4
must hold, which restricts the class of admissible P{r)
curves. In contrast, D{r) is not a p.d.f., and its integral
need not converge. Although several workers have
measured dispersal in the field (e.g., for seeds: Cremer
1966, Roe 1967, Yocom 1968, Werner 1975, Watkinson 1978, Rabinowitz and Rapp 1981, Peart 1982,
Augspurger 1983; for pollen: Colwell 1951, Raynor et
al. 1970, 1971; for fungal spores: Wilson and Baker
1946), there has not been adequate statistical analysis
of dispersal data to specify the expected forms of D{r)
or P{r).
However, to illustrate the potential differences be?
tween P{r) and D{r), and the constraints upon them,
some hypothetical dispersal curves are depicted in Fig.
1. It is clear from this figure that probability curves
may differgreatly in shape from their corresponding
density curves. The linear and exponential density
functions of Fig. la, c are monotonically declining, but
are associated with peaked probability functions (Fig.
lb, d). The truncated gamma function of Fig. le (Men?
denhall and Scheaffer 1973) is given as an example of
a peaked density function; the peak in the associated
probabilitycurve (Fig. If) occurs furtherfrom the source.
The reciprocal function of Fig. lg is seen to be inappropriate for D{r), as it implies a constant probability
function (Fig. lh). In neither Fig. lh nor Fig. lj do the
P{r) curves have convergent integrals, so the last two
pairs of curves (Fig. 1g-j) cannot describe real dispersal
processes. In the firstthree pairs of curves (Fig. la-f),

NOTES AND COMMENTS

1082
DENSITY

D(r)

PROBABILITY

P (r)

Fig. 1. Several pairs of hypotheticaldensity dispersal

functionsD(r) and probabilitydispersal functionsP(r). In
each pair of curves,the probabilityfunction(on the right)is
equal to the density functionmultiplied by the factor 2wr.
The horizontalscale (radial distance r fromthe source) is the
same forboth D(r) and P(r). Parametersa, b, and c are con?
stants,but do not necessarily take the same values in all
functions.

the requirement that P(r) be a p.d.f. places constraints

on the parameters, further restricting the class of ac?
ceptable dispersal functions, but also reducing the
number of parameters to be estimated. For example,
in the D(r) of Fig. lc, a = b2/(2iv).

For a finite number of dispersed seeds, D(r) must

have a finitevalue at r = 0, while P(r) must be zero at
r = 0. Because of the latter condition, the probability
curve must be peaked. Both of these conditions are
satisfied in the firstthree examples (Fig. la-f) but violated in the last two (Fig. lg-j). The linear density
function is the simplest case, but in each of the empirical studies cited above the rate of decline of density
eventually decreases with distance. Obviously the ex-

Ecology, Vol. 66, No. 3

amples of Fig. 1 are not exhaustive, and the choice of

functions for use in modelling dispersal or fittingdis?
persal data should reflectthe shapes of known dispersal
curves and/or mechanistic models ofthe dispersal process, as well as the constraints noted here.
To obtain P{r) one can estimate the proportions of
seeds in concentric rings (annuli) around the parent. In
the field under natural conditions, it is usually difficult
to estimate the total number (7V) of seeds dispersed
(and hence the proportion in each annulus) because of
the large areas to be searched in the more distant an?
nuli. Watkinson (1978) obtained such counts fora dune
annual with very limited dispersal, although he did not
express the results as probabilities. The absolute den?
sity of seeds at distance r is D{r)N. Absolute densities
can be obtained simply by measuring the number of
dispersed seeds per unit area at various radial dis?
tances, but, as for P{r), N must be known to calculate
D{r). Where possible, N is best measured by counting
the number of seeds on the parent plant at the beginning and at the end of the period of dispersal mea?
surement. Eq. 6 can be used on dispersal data to obtain
P{r) from D{r) or vice versa, but a consequence of Eq.
6 is that a substantial proportion of seeds may fall
beyond a radial distance where the density is already
quite low. Depending on the purpose of the measure?
ments, it may be necessary to sample at greater dis?
tances (involving larger areas) for P{r) than for D{r).
The ecological and evolutionary implications of seed
dispersal patterns depend on whether D{r), P{r), or
both are important to the parent plant and the indi?
viduals (of the same or other species) with which its
progeny interact. P{r) is relevant when the number of
successful offspringis sensitive to the probabilities of
seeds travelling to (or beyond) certain distances. D{r)
is useful for the evaluation of density-dependent in?
teractions among progeny as a function of distance.
Where such density dependence is important, the total
number of successful offspringwill depend on both P{r)
and D{r).
Wide dispersal of seeds may reduce density-depen?
dent mortality due to seed predation, herbivory, or
seedling competition (Janzen 1970). The density func?
tion is relevant to this hypothesis. Wilson and Janzen
(1972) found that seed predation by bruchid beetles
increased with the local seed density of palm seeds,
but was independent of distance from the parent. Where
the probability of escape to some distance from the
point of release is more important, the relevant curve
is P{r). This may occur if the parent plant attracts the
natural enemies of seeds and seedlings (Janzen 1970,
Connell 1971) or if the parent inhibits seedling estab?
lishment in its neighborhood (Peart 1982). O'Dowd
and Hay (1980) showed seed predation by rodents to
depend on distance from the parent, rather than seed

June 19S5

NOTES AND COMMENTS

1083

persal.Annual Review of Ecology and Systematics13:201228.

Janzen,D. 1970. Herbivoresand the numberof treespecies
in tropicalforests.American Naturalist 104:501-528.
Levin, D. A. 1981. Dispersal versus gene flow in plants.
Annals ofthe Missouri Botanical Garden 68:233-253.
-.
1984. Immigrationin plants:an exercisein the subjunctive. Pages 242-260 in R. Dirzo and J. Sarukkan,ed?
itors. Perspectives on plant population ecology. Sinauer,
Sunderland,Massachusetts,USA.
Levin, D. A., and H. W. Kerster. 1974. Gene flowin seed
plants.Pages 139-220 in T. Dobzhansky,M. K. Hecht,and
W. C. Steere,editors.Evolutionarybiology,Volume 7. Plenum, New York, New York, USA.
Mendenhall, W., and R. L. Scheaffer. 1973. Mathematical
statisticswith applications. Duxbury, Boston, Massachu?
setts,USA.
O'Dowd, D. J., and M. Hay. 1980. Mutualism between
harvesterants and a desert ephemeral: seed escape from
rodents.Ecology 61:531-540.
Peart,D. R. 1982. Experimentalanalysis of succession in a
grasslandat Sea Ranch, California.Dissertation.University
of California,Davis, California,USA.
Rabinowitz, D., and J. K. Rapp. 1981. Dispersal abilities
of seven sparse and common grassesfroma Missouri prai?
rie. American JournalofBotany 68:616-624.
Acknowledgments: I thank C. L. Folt, D. E. Goldberg,
K. R. Hopper, T. E. Miller, and two anonymous re? Raynor, G. S., E. C. Ogden, and J. V. Hayes. 1970. Dis?
persionand deposition of ragweedpollen fromexperimen?
viewers for comments on the manuscript, and D. S.
tal sources. Journalof Applied Meteorology9:885-895.
Wilson forthe use of facilities. W. K. Kellogg Biological
Raynor, G. S., E. C. Ogden, and J. V. Hayes. 1971. Dis?
Station contribution number 527.
persionand deposition of timothypollen fromexperimen?
tal sources. AgriculturalMeterology9:347-366.
Roe, A. L. 1967. Seed dispersal in a bumper spruce year.
United States Forest Service Research Paper 39:1-10.
Literature Cited
Watkinson,A. R. 1978. The demographyofa sand dune
around
K.
1983.
recruitment
C.
trop?
Offspring
Augspurger,
annual: Vulpiafasiculata. III. The dispersal of seeds. Jour?
ical trees:changes in cohort distance with time. Oikos 40:
nal of Ecology 66:483-498.
189-196.
Werner,P. A. 1975. A seed trap for determiningpatterns
Augspurger,C. K., and C. K. Kelly. 1984. Pathogen mor?
of seed deposition in terrestrialplants. Canadian Journal
talityof tropicaltree seedlings:experimentalstudies of the
ofBotany 53:810-813.
effectsof dispersaldistance,seedlingdensityand lightcon? Wilson, D. E., and D. H. Janzen. 1972. Predationon Scheeditions. Oecologia (Berlin) 61:211-217.
lea palm seeds by bruchidbeetles:seed densityand distance
Colwell, R. 1951. The use of radioactive isotopes in deterfromthe parentpalm. Ecology 53:954-959.
of
American
Journal
distribution
patterns.
mining spore
Wilson, E. E., and G. A. Baker. 1946. Some aspects ofthe
Botany 38:511-523.
aerial dissemination of spores, with special referenceto
Connell, J. H. 1971. On the role of naturalenemies in preconidia of Sclerotinia laxa. Journal of AgriculturalRe?
ventingcompetitiveexclusion in some marineanimals and
search 72:301-327.
in rain foresttrees. Pages 298-312 in P. J. Den Boer and
Yocom, H. A. 1968. Shortleaf pine seed dispersal.Journal
G. Gradwell,editors.Dynamics of populations. Centrefor
of Forestry66:422.
AgriculturalPublishingand Documentation, Wageningen,
1Manuscriptreceived28
The Netherlands.
January1983;
21
March 1982; accepted 24 March 1984.
revised
from
of
seed
Dissemination
K.
W.
1966.
EucalypCremer,
Final versionreceived26 November1984.
tus regnans.AustralianForestry30:33-37.
2 W. K. Kellogg Biological Station,
Michigan State
Handel, S. N. 1982. Dynamics of gene flow in an experi?
mentalgardenofCucumis melo (Cucurbitaceae). American
HickoryCorners,Michigan 49060 USA.
University,
3 Presentaddress: Departmentof
Journalof Botany 69:1538-1546.
Biological Sciences,
DartmouthCollege, Hanover, New Hampshire 03755 USA.
Howe, H. F., and J. Smallwood. 1982. Ecology of seed dis-

density. Recently Augspurger and Kelly (1984) were

able to demonstrate the effects of both density and
distance on pathogen-induced mortality in seedlings of
a tropical tree. Dispersed seeds may be clumped due
to the movements of animal dispersers (see review in
Howe and Smallwood 1982), physical obstructions to
wind dispersal, or deposition by receding waters. In
these cases, the density of seedfall may not be simply
related to distance, and local clumping may be more
important for seed and seedling success, although the
relation between the density and probability functions
(Eq. 6) still holds.
The probability function for pollen dispersal is relevant to estimation of rates of outcrossing and gene
flow, while the density function may influence the like?
lihood of pollen competition and/or mating between
half-siblings. Animal vectors can produce clumping in
pollen as they do with seeds (e.g., Handel 1982). Sev?
eral aspects of pollen dispersal have been reviewed by
Levin and Kerster (1974) and Levin (1981, 1984).