Seed Dispersal of Chromolaena odorata Reconsidered
Natal Parks Board, St Lucia Research, St Lucia Estuary, KwaZulu-Natal, 3936, South
Blackmore A.C., (1998) Seed dispersal of Chromolaena odorata reconsidered. In: Ferrer,
P., Muniappan, R. and Jayanth, K.P. (eds) Proceedings of the Fourth International
Workshop on the Biological Control and Management of Chromolaena odorata, October
1996, University of Guam, Mangilao, Bangalore, India, pp. 16 -21
The rapid spread of Chromolaena odorata since introduction in KwaZulu-Natal
(South Africa), and the failure of this species, until recently, to invade conservation
and other C .odorata free areas, has necessitated the reconsideration of primary
dispersal mechanism of this species. The study of C .odorata seed movement in the
southern region of the Greater St Lucia Wetland Park indicated that
A. seed rain occurred from August to November,
B. airborne seeds were unlikely to travel distances greater than 80 m from the
C. off-road vehicles transported a significantly higher number of seed over
greater distances than were carried as seed rain.
This evidence suggests
1. that the atmosphere does not act as a large reservoir within which C .odorata
seeds are carried great distances, and
2. that mechanical transport of C .odorata seed is the most likely reason for the
rapid spread of this species. The importance of these observations in terms of
conserved and other C. odorata free areas are discussed.
Chromolaena odorata (L.) King and Robinson is considered to be the most aggressive
invader of indigenous sub-tropical areas (Liggitt, 1983; Macdonald and Jarman, 1985;
McFadyen, 1991; Wilson, 1995). C .odorata is capable of producing vast quantities of
seed, with estimates ranging from 93,000 (Weerakoon, 1972) to 1,600,000 (Wilson,
1995) per plant. It is believed that the rapid spread of this species is directly related to
the extensive seed production and wind dispersal architecture of the seeds (Erasmus,
1985; Liggitt, 1983; Macdonald and Frame, 1988). The rapid spread of C .odorata has
prompted a number of studies on its distribution (Gautier, 1992; Liggitt, 1982),
biological control (Ambika, 1990; Archibold, 1979; Joy et al., 1993; Kluge, 1990;
Kluge and Caldwell, 1992; Lyla and Joy, 1992; Viraktamath and Muniappan, 1992)
and biology (Erasmus, 1985; Wilson, 1995). However, there are few, if any, published
studies testing whether seed rain is the principal mode for seed dispersal and whether
the spread of this species is as a direct consequence of seed rain. Since the apparent
accidental (Pickworth, 1976) or horticultural (Gautier, 1992) introduction of C
.odorata into KwaZulu-Nata; (South Africa) during the Second World War (Liggitt,
1983), the maximum rate of spread recorded was in the region of 2000% for the
period 1975 to 1980 (Liggitt, 1983). Until recently, the Greater St Lucia Wetland Park
(Figure 1) has been relatively free of C .odorata. However, invasion of the Park is
now occurring at an alarming rate (Unpublished NPB data). The rapid spread and
consequent high densities of C .odorata on the periphery (Erasmus, 1985) and the
delayed invasion of the interior of the park raises questions regarding our
understanding of the dispersal mechanisms of C .odorata.
Figure 1:Location of study area, the Greater St Lucia Wetland Park
The light weight, parachutal structure of C .odorata seeds is perceived to enable them
to become easily airborne and hence easily dispersed as "seed rain" (Burrows, 1973).
Two hypotheses describe the potential wind dispersal of airborne seeds. The first
assumes that the atmosphere acts as a large reservoir within which C .odorata seeds
are carried great distances (Figure 2a). If this hypothesis holds, then large numbers of
C .odorata seeds, that originate on the periphery, may conceivably be transported
deep into the park. Prevention of C .odorata seed rain into the park would, therefore,
necessitate the eradication of C .odorata within an extremely wide belt around the
park. Failing this, conservation managers would be forced to continually re-clear
areas within the park. Under this scenario, costs of eradication programmes would
remain, at best, constant at an elevated rate over time.
Figure 2: Two hypotheses explaining the potential spread of Chromolaena odorata. The first
diagram (a) assumes that C odarata seed are blown large distances, whereas the second (b) assumes
that the bulk of C .odorata seed are blown short distances before re-establishment (see text for
The second hypothesis (Figure 2b) argues that the bulk of the C .odorata seed is
dispersed over short distances. After establishment and flowering, the seeds are again
dispersed short distances. In so doing C .odorata effectively "leap frogs" into C
.odorata free areas. If an area is cleared and maintained as such, then eradication
programmes would require a decreasing amount of effort, and hence cost, dedicated to
The objective of this study is to
1. determine which hypothesis best describes the wind dispersal of C .odorata
2. evaluate the role that off-road vehicles may play in the transport of C .odorata
seed from areas invaded by this species. In so doing, particular reference is
made to the dependence of the conserved areas on its surrounds for the
preservation of its ecological and aesthetic integrity.
The location of the Greater St Lucia Wetland Park (Figure 1), in KwaZulu-Natal, is at
the interface between tropical and subtropical climates and hence is highly sensitive to
invasion of C .odorata (Blackmore, 1991; Henderson, 1989; Macdonald and Jarman,
1985). The landscape of the park is generally of low relief, with coastal barrier dunes
in the east and the Lebombo Mountains in the west. These two prominent features are
separated by a complex mosaic of lowland wetlands, grasslands and forests. Annual
rainfall ranges from 1330 mm in St Lucia town to 1120 mm at Charters Creek in the
south west, and 1045 mm at Sodwana Bay in the north east to 650 mm at Mkuze
Game Reserve in the North West (unpublished CCWR data). Soils of the park are
generally sandy and of marine origin, increasing significantly in age, and hence
nutrient status, from east to west. The predominant wind directions are north east and
south west in nature, although easterly and westerly winds are common (Taylor,
The study area, the southern region of the park, is divided into three sections currently
known as the Western Shores, Eastern Shores and Mapelane Island. The Western
Shores is a narrow 1.5 km strip along the western edge of the estuary and St Lucia
Lake system. A large proportion of the western boundary of the Western Shores has
been afforested with Pinus elliottii. The habit of the South African Forestry Company
Limited (SAFCOL - formally the Department of Forestry) of ploughing or rotovating
fire breaks on the periphery of the plantations has facilitated the spread of C .odorata
along the western boundary of the park (Anon, 1993; Blackmore 1991).
The Western Shores is predominately grassland, however, clumps of woody
vegetation pruned by biennial fires are common. The distribution of C .odorata within
the 1.5 km strip has been limited to the woody clumps and estuary edge (Blackmore,
1991). The Eastern Shores exhibits significantly lower densities of C .odorata than
that recorded on the Western Shores and Mapelane Island. The small isolated patches
of C .odorata that occur on the Eastern Shores are limited to the timber plantations
and are substantial distances away from the study areas. All of the patches of C
.odorata within the 1.5 km strip and along the estuary were cleared prior to the study.
The Mapelane Island is highly disturbed and is almost solely covered with Casuarina
equisetifolia and C .odorata. Isolated patches of C .odorata within a radius of 2 km
from the study area on the St Lucia township side of the estuary (to the north of the
Mapelane Island) were cleared prior to initiating this study. This was the only
available site allowing exclusive investigation of seed movement on a north south
The distribution and density of C .odorata was determined in the St Lucia region prior
to the initiation of the study, and mapped at a 1: 20,000 scale. The seed traps were
located near the highest infestations of C .odorata. The movement of seed was
determined by trapping airborne seeds on a 0.5 m x 0.5 m board smeared with
petroleum jelly. Six trapping transects were located within the 1.5 km strip of the
Western Shores. These traps were spaced at 0 m, 10 m, 20 m, 30 m; 50 m, 80 m and
100 m from the afforested boundary. Two transects (three traps each) were located
between the St Lucia estuary and the Mapelane Island at 0 m, 20 m (southern estuary
bank) and 200 m (northern estuary bank).
The 0 m traps were placed within the C .odorata infestations with the smeared board
located at the average height of the lowest flowering buds. The remainder of the traps
were set at 1 m above the ground. Trapped seeds were enumerated approximately
every two days, and the petroleum jelly was renewed. Seed trapping was initiated
prior to seed set and was terminated two months after the last seed was trapped.
An index of seed release was determined by marking 5 stems consisting of
approximately 20 florets (each arising from different plants) within 5 m of each 0 m
trap. Each floret was inspected, at two daily intervals, for rupture. The numbers of
unruptured florets were counted. Many immature buds were marked, and monitoring
of these was initiated once it became evident that they would not be aborted and
hence would contribute to C odorata seed rain.
In order to determine the potential transport of C .odorata seed by vehicles, a four-
wheel-drive light delivery vehicle was driven through a C .odorata infested area for
approximately 20 minutes on 15 occasions. At the end of the first five replicates, the
vehicle was cleaned throughout and the trapped seeds enumerated. For the remaining
10 replicates, the vehicle was driven 4 km and 15 km (five replicates each), in an
uninfested area, before being cleaned and the seeds enumerated.
Seed rain season
Two seed release events were prevalent (Figure 3). The first and most extensive
occurred from August to mid September. The second, made up entirely of the florets
that had been flowering during the latter part of this seed release event, occurred from
late September to the end of October. The pattern of seed trapped was similar to that
of the index of seed release and no significant lag between the two was noted.
Figure 3: Percentage of the numbers seed trapped and florets ruptured (-) during the study
Effective distance of seed rain
It was noted that over 99% of C odorata seed rain occurred immediately below the
plant (Figure 4). Movement of seed further than 80 m from the afforested boundary
on the Western Shores was not noted to occur. Likewise, no seed rain was observed to
have occurred on the northern side of the estuary to the south of the St. Lucia
township. Release rates and timing of seed rain occurring on the estuary traps were
similar to that of the Western Shores.
Figure 4: Average number of seed trapped along western boundary transects (non-logged
standard deviation values are presented)
Vehicle transport of C odorata seed
A large number of seeds were collected by the vehicle moving through a C .odorata
infested area (Figure 5). The number of seeds counted dropped exponentially over the
distances travelled. Although there were a large number of loose seeds collected after
4 km and 15 km, the majority collected were mature seeds still contained within the
floret. Many of the florets were lodged in joints and grooves within the vehicles body-
Figure 5: Average number of seed trapped and transported by a four wheel-drive light delivery
Given that alien plant eradication programmes are a significant proportion of
conservation budgets (Erasmus, 1985), a thorough understanding of the fundamental
concepts of alien plant seed dispersal is paramount to derive a cost-effective control
and eradication strategy. In the 20 years between 1960 and 1980, the densities of C
.odorata in the Charters Creek area increased from absent to a very heavily infested
state, in places forming monospecific stands along the afforested boundary
(Blackmore, 1991; Erasmus, 1985; Liggitt, 1983). Yet in that time, the Eastern Shores
remained relatively free of C .odorata with a few infestations occurring along the
estuary (Anon., 1993). If C .odorata seed was spread by seed rain where the
atmosphere acts as a large reservoir in which seed are transported great distances
(Figure 2a), the invasion of the Eastern Shores by this species would have followed a
similar pattern to that documented outside of the park. This inference is supported by
this study in that C .odorata seed was not observed to be transported further than 80
m from its source (Figure 4). It is conceivable, therefore, that the infestations along
the estuary and within woody patches on the Western Shores were as a result of C
.odorata being dispersed consecutive short distances from the afforested boundary
In order to maintain the densities of C .odorata to that of pre-1980, the annual NPB
alien plant eradication budget had to be increased 10 fold (early 1990) and again by
eight fold in 1995 (Anon., 1996). The timings of the needs for budget increases lag
behind, but coincide with, the clearfelling of the P. elliottii plantations within the
park. Other than the patches of C .odorata along the estuary, C .odorata occurs along
tracks within, and along rotovated firebreaks surrounding, the plantations (personal
observation) on the Eastern and Western Shores. The distribution of C .odorata and
timing of budget increases, support the notion that the recent invasion of the Greater
St Lucia Wetland Park, and in particular the Eastern Shores section, has been
significantly assisted by activities of the timber industry, and not primarily by seed
being blown in from the periphery in the form of seed rain. It is, therefore,
conceivable that the rapid spread of this species throughout KwaZulu-Natal, and
beyond, maybe as a direct result of vehicles ferrying seed large distances. Thus, the
movement of airborne seed would contribute little to the "provincial invasion", but
would facilitate establishment of the monospecific stands in the vicinity of the newly
This reasoning has two profound implications for conserved and other C .odorata free
areas. The first, by maintaining a C .odorata free belt on the periphery of the park,
seed rain encroachment of the park may easily be avoided. The second, by allowing
off-road and heavy duty vehicles that have operated in C .odorata infested areas into
the park unchecked, may have profound consequences for maintaining C .odorata
C .odorata is one of a number of alien plants that are currently threatening the
ecological integrity of the conserved areas in KwaZulu-Natal. Despite the parachutal
architecture of the seed, wind distribution of seed is significantly less than that carried
by off-road vehicles. It is unlikely that large amounts C .odorata seed would be
transported great distances as seed rain. It has been observed that the bulk of
C. odorata seed are transported only short distances in the form of seed rain. Invasion
of virgin areas by this species is, therefore, a progressive stepwise process. The
clearing of a strip of greater than 80 m around the periphery of the park, would
significantly reduce the amount of seed being blown into the protected area.
A single flowering plant within a C .odorata free area presents itself as a greater
threat to the integrity of the system than incoming seed derived from seed rain.
Likewise, vehicles moving through a C .odorata infested area transport significantly
more seed, and over further distances, than does seed rain.
Thanks are extended to Mrs. N. Blackmore, Mr.. C. Mulqueeny and Dr. E. Witkowski
for their comments on the manuscript, and Mr. H. Bentley and Mr. I. Porter for their
invaluable insight into the management and eradication of C .odorata.
Ambika, S.R.1990. The problem of Chromolaena weed. Chromolaena odorata
Newsletter 3: 1-6.
Anonymous 1993. Natal Parks Board Year Book 1992-1993. Natal Parks Board
Internal Report. Pietermaritzburg 390 pp.
Anonymous 1996. Natal Parks Board Year Book 1995-1996. Natal Parks Board
Internal Report. Pietermaritzburg 400 pp.
Archibold, O. W. 1979. Seed input as a factor in the regeneration of strip-mine wastes
in Saskatchewan. Canadian Journal of Botany 58: 1490-1495.
Blackmore, A. C. 1991. The 1991 distribution of Chromolaena odorata on the
Western Shores of Lake St Lucia. Natal Parks Board Internal Report 1-5.
Burrows, F. M. 1973. Calculation of the primary trajectories of plumed seeds in
steady winds with variable convection. New Phytology 72: 647-664.
CCWR. Computing Centre for Water Research. University of Natal,
Pieterrnaritzburg, Unpublished data.
Erasmus D. J. 1985. Achene biology and the chemical control of Chromolaena
odorata, pp. 1379. Unpublished Ph.D. Thesis. University of Natal, Pietermaritzburg.
Gautier, L.1992. Taxonomy and distribution of a tropical weed: Chromolaena
odorata (L.) R. M. King & H. Robinson. Candollea 47: 645-662.
Henderson, L.1989. Invasive alien woody plants of Natal and north-eastern Orange
Free State. Bothalia 19: 237-261.
Joy, P. J., Lyla, K. R. and Satheesan, N. V. 1993. Biological control of Chromolaena
odorata in Kerala (India). Chromolaena odorata Newsletter 7:1-4.
Kluge, R. 1990. Prospects for the biological control of triffid weed, Chromolaena
odorata, in southern Africa. South African Journal of Science 86:22-230.
Kluge, R. and Caldwell, P. M.1992. The biology and host specificity of Pareuchaetes
aurata (Butler) (Lepidoptera: Arctiidae), a "new association" biological control agent
for Chromolaena odorata (L.) R. M. King & H. Robinson (Compositae). Bulletin of
Entomological Research 83: 87-93.
Liggitt, B. 1982. Chromolaena odorata: Annotated bibliography. Institute of Natural
Liggitt, B. 1983. The invasive alien plant Chromolaena odorata, with regard to its
status and control in Natal. Institute of Natural Resources 2: 1-41.
Lyla, K. R. and Joy, P. J. 1992. Biology of Aphis spiraecola (Patch) infesting
Chromolaena odorata in Kerala, India. Chromolaena odorata Newsletter 5: 2-3.
Macdonald, I. A. W. and Frame, G. W. 1988. The invasion of introduced species into
nature reserves in tropical savannas and dry woodlands. Biol. Conserv. 44: 67-93.
Macdonald, I. A. W. and Jarman, M. L. 1985. Invasive alien plants in the terrestrial
ecosystems of Natal, South Africa. South African National Programmes Report No
118, Pretoria .88 pp.
McFadyen, R. E. 1991. The accidental introduction of Chromolaena mite Acalitus
adoratus into Southeast Asia. Chromolaena odorata Newsletter 4:10.
Pickworth, G. 1976. Triffid Weed (Eupatorium odoratum). An address to the lower
Tugela Farmers soil conservation committee. Unpublished Report.
Taylor, R.H. 1980. A land capability study for hippopotamuses at Lake St Lucia,
Zululand. Unpublished M.Sc. Thesis, University of Natal, Pietermaritzburg.
Viraktamath, C. A. and Muniappan, R. 1992. New records of insects on Chromolaena
odorata in India. Chromolaena odorata Newsletter 5: 1.
Weerakoon, L. 1972. Studies on the biology and control of Eupatorium odoratum L.
M.Sc. thesis, Vidyodaya Campus, University of Ceylon.
Wilson, M. 1995. Autecology of the invasive alien plant, Chromolaena odorata, in
the Greater St Lucia Wetland Park. M.Sc. Thesis, University of the Witwatersrand,