Seed dispersal and seedling recruitment Protea laurifolia

The pattern of seedling recruitment around a single isolated Protea laurifolia Thunb. individual after a fire, was determined using a chi-square goodness of fit test for uniformity and Rayleigh's test for direction. Recruitment occurred largely to the north-west of the parent plant whereas the wind after the fire came largely from the south-east, providing some evidence of the influence of wind on dispersal. The maximum distance of a seedling from the parent plant was 26,3 m, with 95% of recruitment occurring within 15 m of the parent. To speed up the rate of migration in areas where the species has been eradicated the establishment of point seed sources is suggested. S. Afr. J. Bot. 1986, 52: 421 - 424

The pattern of seedling recruitment around a single isolated Protea laurifolia Thunb. individual after a fire, was determined using a chi-square goodness of fit test for uniformity and Rayleigh 's test for direction. Recruitment occurred largely to the north-west of the parent plant whereas the wind after the fire came largely from the south-east, providing some evidence of the influence of wind on dispersal. The maximum distance of a seedling from the parent plant was 26,3 m, with 95% of recruitment occurring within 15 m of the parent. To speed up the rate of migration in areas where the species has been eradicated the establishment of point seed sources is suggested.

Introduction
Seed dispersal in members of the genus Protea is considered to be limited in space (Jordaan 1972;Bond 1980;Brits 1982;Midgley 1983;Bond et a/. 1984). There are few direct observations to support this. Bond (1980) noted poor dispersal of serotinous Proteaceae (species with canopy-stored seed) from unburnt vegetation to burnt areas in the Swmtberg. Small barriers such as a vehicle track and a thin belt of riverine vegetation were apparently sufficient to stop dispersal. Bond eta/. (1984) also provide some evidence of the short dispersal distances in serotinous Protea species.
Sclerophyllous South African Proteaceae of the fynbos biome (Mollet a/. 1984) are largely dispersed by either wind or ants (Slingsby & Bond 1982). The wind-dispersed members include all species of the genera Protea and Au/ax which have hairy fruits (Slingsby & Bond 1982), and all the species in the section Alatosperma in the genus Leucadendron, in which the species have winged fruits (Williams 1972). The efficiency of dispersal appears to vary considerably between species, with Protea repens (L.) L. seeds having the largest diameter of achene plus hairs and the highest dispersal efficiency within the genus Protea (Bond eta/. 1984). A few species in Leucadendron, such as L. argenteum (L.) R. Br. and L. rubrum Burm. f., have distinct parachute-like plumes, although the seeds are relatively heavy (Williams 1972). The remaining genera, and some Leucadendron species produce seeds with elaiosomes which are dispersed by ants (Slingsby & Bond 1982). In a third group, comprising a number of species of the genus Leucadendron, no dispersal strategy has been identified to date (Slingsby & Bond 1982).
Theoretical relationships exist between the distribution of dispersed seed and the distribution of established seedlings (Janzen 1970;Geritz et a/. 1984). Serotinous Proteaceae regenerate only after the parent plants have been killed by fire (Bond eta/. 1984), and regeneration occurs in a biotically uniform environment. It is assumed therefore that the distribution pattern of established seedlings represents the pattern of effectively dispersed viable seeds, and the maximum distance between the parent plant and a seedling indicates the maximum effective seed dispersal distance. Burrows (1973) defines the aerial path between the position of release of a seed and the position of initial contact with the ground as the primary trajectory. The 'seed' of Protea laurifolia Thunb. (a serotinous member of the genus) consists of an achene which is densely covered with trichomes, which project at right angles from the surface. The style is persistent. The seed is therefore apparently suited to wind dispersal, falling into the category of woolly or plumed seeds of Burrows (1975). However, since there is no distinct plume and since the persistent style decreases the surface area to volume ratio, the seed is probably not specialized in this respect and the primary trajectory may be limited. Bond et at. (1984) and Bond (1985) reported that the structure of Protea seed facilitates rolling over the soil surface in the absence of barriers after a fire. This suggests the existence of a second phase of dispersal after reaching the ground at the end of the primary trajectory. On the other hand, the function of the trichomes and style may be to limit further dispersal after reaching the ground. According to Jordaan (1972), the trichomes of seeds bind together to hold seeds together in clumps, and Brits (1982) observed grains of sand being trapped in the trichomes, apparently helping to weigh down and bury the seed.
Major declines in serotinous Protea population densities and even local extinctions can result from both inappropriate intervals between fires (Bond 1980;Van Wilgen 1981) and unfavourable season of burn (Bond 1985;Van Wilgen & Viviers 1985). Therefore the rate at which denuded areas can be re-colonized through migration, which depends on natural dispersal, is important in determining management procedures for conservation areas.
This paper reports on the recruitment of seedlings around a single burnt Protea taurijotia plant in the Zachariashoek experimental catchment near Paarl in the south-western Cape Province (34°49'S and 19°02'E). The presence of this isolated shrub, approximately 24 years old and 2 to 3 m high was noted before a prescribed burn of the catchment on 14 November 1977 (F.J. Kruger pers. comm.). P. taurijotia normally grows in fairly dense stands or at least within a short distance from conspecific individuals. This individual therefore presented an unusual opportunity to determine the recruitment pattern around a point seed source. A further advantage was the presence of a large number of seedlings (215), since mean seedling to adult ratios are generally less than 18: 1 (Bond et at. 1984;Van Wilgen & Viviers 1985). This investigation aimed to determine the effective distance of P. taurijotia seed dispersal, and whether dispersal is related to wind speed and direction.

Methods
The site has a slope of approximately 7° with an aspect of 120°. No other large shrub species occur and it was assumed that the site was relatively free of vegetative barriers after the fire. The ground is not stony and there were no other visible barriers to dispersal.
The area of the study site was searched and all seedlings were found to be within an area of 45 m by 25 m around the stump of the parent plant. The positions of all seedlings in this area were recorded in September 1984. From these data the distance and azimuth from true north of each seedling from the parent plant was calculated.
To examine the effects of wind direction on seed dispersal, the mean azimuth of the seedlings was calculated and the uniformity of the seedling distribution around the point of origin was tested by means of a chi-square goodness of fit test. This analysis was performed on the observed circular frequency distribution in eight direction intervals, with the first interval centred on true north. As the sample may have had a non-uniform distribution without a mean direction, the significance of the mean angle was determined with Rayleigh's test (Zar 1974).
Data on wind velocities and direction during the first 31 days after the fire were obtained from the Zachariashoek weather station situated approximately 1 km from the study site. Wind is recorded at this station by means of a Woelfle S.-Afr. Tydskr. Plantk., 1986, 52(5) type mechanical wind recorder. The data for this period were summarized in velocity and direction categories. The study site is near the top of a ridge and there are no rocky outcrops in the vicinity. Therefore,_ it was assumed that wind speeds and direction on the site were similar to those measured at the weather station.

Results
The distribution of the seedlings is summarized in Figure 1a in such a manner as to be comparable to the wind rose (Figure  1 b) summarizing the wind data.
The mean azimuth of the seedlings from the parent plant is 302°. The total chi-square value for the frequencies in the eight direction categories was 105,8. The null hypothesis, that the observed circular frequency is uniformly distributed around the origin is therefore rejected at P < < 0,001. The value of Rayleigh's R calculated was 95,68, which is significant at P < < 0,001. The mean angle is therefore significant.
The frequency of seedlings with distance from the parent plant (the population recruitment curve) is shown in Figure 2.

Discussion
The marked concentration of seedlings to the north-west of the parent plant's stump ( Figure 1a) corresponds with the high frequency and speed of the wind from the south-east ( Figure  1b). However, the pattern of seedling distribution is not a perfect inversion of the wind rose. This indicates the need to know the exact time of release of the seeds and the extent to which they move after reaching the ground. Seeds can be fully released from the plant within 24 hours of a fire (D.C. le Maitre pers. comm.). It was assumed here that there would be no significant movement of seeds on the ground after 31 days. Although the site in this study was well exposed, local differences in wind direction relative to those recorded at the weather station may have existed.
Although the results indicate dispersal by wind, dispersal distances in Protea taurijotia were relatively short, with 950Jo of the recruitment occurring within 15 m of the parent plant.
The limited dispersal of protea seed is considered to ensure the establishment of seedlings within a favourable habitat. Midgley (1983) reasoned that as seeds are released after fire, dispersal would be restricted to within the burnt area, which is more favourable for both germination and establishment than the adjacent unburnt vegetation. Brits (1982) proposed that the limitation of seed dispersal could have adaptive advantage in variable, nutrient poor habitats, such as are found in the fynbos.
Although the data in this study were obtained from one sample only, the site was well exposed to the wind. Therefore the maximum distance here may be regarded as a reasonable estimate of the order of magnitude of dispersal distances. The maximum dispersal distance of 30 m recorded implies that the average size of a homogeneous soil type or a burnt area would have to be less than a few hectares for the limited dispersal of seed to have an adaptive value.

Conclusions
Evidence has been presented to support the hypothesis that Protea taurijotia seed is wind dispersed. Questions regarding the function of the structure of the seed and whether the seed is dispersed during or after the primary trajectory, remain unanswered.
The potential rate of migration into an area which has been denuded of serotinous Proteaceae is in the order of 30 m per generation. Generation intervals are directly related to the S. Afr. J. Bot., 1986, 52(5)  interval between fires . On the basis of a prescribed bum every 15 years it would take roughly 500 years for a population to advance I krn into a previously occupied area from an upwind source. In areas where serotinous Proteaceae have been exterminated the management option of allowing gradual re-colonization by natural seed dispersal is impractical. The reestablishment of point seed sources through strategic planting could speed up the re-colonization process considerably where nature conservation is a priority. Since seed production is negatively correlated with adult density (Bond et al. 1984), such point plantings should be more effective than establishing large patches of proteas.