428 S.-Afr.Tydskr. Plantk. , 1990, 56(4): 428--434 Dispersal and predation of alien seeds: effects of season and invading stand density

P.M. Holmes Department of Botany, University of Cape Town, Rondebosch, 7700 Republic of South

Accepted 3 May 1990

Acacia saligna and A. cyclops are invasive alien in lowland ecosystems of the biome, . Both species typically form dense thickets and accumulate large soil-stored seed banks. Seed removal rates, relative to seed availability in the litter layer of Acacia-infested vegetation, were studied to determine the importance of indigenous and vertebrates as dispersors and predators, respectively. There was a rela­ tive loss to the annual seed-fall of 50% and 80-95% in dense A. saligna and A. cyclops stands, respectively. Ants and rodents removed significant quantities of seeds, but a large proportion may have rotted. In a dense A. saligna stand, ants usually removed all seeds presented to them after 7 days presentation, compared to only 36% of seeds in a dense A. cyclops stand. In dense stands of both species, rodents were relatively slow to take seeds, removing 31 % within 7 days. Removal from trays was least during and following seed-fall (January-March) and greatest prior to seed-fall (September-November) and decreased as A. cyclops canopy cover increased. In a lOW-density A. cyclops stand, ants and rodents competed for seeds. The results indicate that indigenous ants have played a critical role in maintaining and accumulating Acacia seed banks and thus may have facilitated the development of dense Acacia stands.

Acacia saligna en A. cyclops is uitheemse indringerstruike wat in die laagland-ekosisteme van die fynbos­ bioom, Suid-Afrika, voorkom. Dit is kenmerkend van albei spesies dat hulle digte ruigtes vorm en dat groot saadvoorrade in die grond opgegaar word. Die tempo van saadverwydering in verhouding tot beskikbaarheid in die afvallaag van plantegroei waarin Acacia vervuil is, is bestudeer. Daar was 'n relatiewe verlies van onderskeidelik 50% en 80-95% in die jaarlikse saadval in die digte stan de van A. saligna en A. cyclops. Miere en knaagdiere het betekenisvolle hoeveelhede saad verwyder, maar 'n groot dee I mag moontlik verrot het. In 'n ruie stand van A. saligna het miere gewoonlik al die saad wat beskikbaar was binne 7 dae verwyder, teenoor slegs 36% van die saad in 'n ruie stand van A. cyclops. Knaagdiere was betreklik traag om saad in ruie stande van albei spesies te verwyder, en slegs 31 % is binne 7 dae verwyder. Verwydering uit bakke was die laagste gedurende en na saadval (Januarie-Maart) en die hoogste voor saadval (September-Novem­ ber), en het verminder soos A. cyclops se blaredak digter geword het. In 'n minder-digte stand van A. cyclops was miere en knaagdiere in kompetisie om die saad te bekom. Die resultate toon dat inheemse miere 'n kritieke rol in die handhawing en opberging van Acacia-saadvoorrade gespeel het en sodoende moontlik die ontwikkeling van ruie Acacia-stande aangehelp het.

Keywords: , Acacia saligna, ants, myrmecochory, rodents

Introduction adaptability to different substrata (Taylor et al. 1985), Many species have been translocated by man, but rapid establishment after fire (Taylor 1983) aided by in order to establish, reproduce and become invasive in nitrogen fixation (Milton 1980), efficient seed dispersal the area of introduction, they must be adapted to the (Glyph is et al. 1981; Taylor et al. 1985; Knight 1988) and new environment, survive interactions with resident hard-coated seeds which accumulate in large soil-stored competitors and predators, and re-establish any neces­ seed banks (Dean et al. 1986). As seed production in A. sary mutualistic relationships (Kruger et al. 1986; cyclops is similar in South Africa and (Gill & Crawley 1987). Aliens may have the advantage over Neser 1984), the much larger seed banks measured in indigenes of having escaped from most of their South Africa may primarily relate to a low incidence of coevolved pests and predators (Milton 1980; Mooney & pre- and post-dispersal seed predation (Milton 1980; Drake 1987), yet in order to invade new habitats, seed Milton & Hall 1981; Holmes & Rebelo 1988). dispersal is essential (Salisbury 1961, quoted in Sorensen The Acacia dispersal unit is composed of the mature 1985). seed plus an expanded and elaborated portion of the The Australian shrubs, Acacia saligna (Labill.) funicle (O'Dowd & Gill 1986). In A . saligna, the funicle Wend!. and A. cyclops A. Cunn. ex G. Don, have is a small, hard, white elaiosome. Seeds ripen in early overcome habitat and biotic barriers to become serious summer, are shed soon after dehiscence in December, invaders in lowland ecosystems of the fynbos biome, and, in Australia, are dispersed by ants (O'Dowd & Gill South Africa (Macdonald & Richardson 1986). They 1986). By contrast, in A . cyclops the funicle is a large, typically form dense thickets, suppressing the indigenous soft, red aril encircling the seed (O'Dowd & Gill 1986). vegetation and thus reducing the species richness of the Ripening is more protracted and precedes a 5-month community (Richardson et al. 1989). Their success in display period during which bird-dispersal occurs South Africa has been variously attributed to their (Knight 1986). In Australia, seeds are dispersed by both S.Afr.J. Bot. , 1990,56(4) 429

birds and ants (O'Dowd & Gill 1986). At Silvermine a single I-ha plot was demarcated in a In South Africa, A. cyclops seeds are dispersed by dense stand [projected canopy cover (PCC) of 100%] of birds (Middlemiss 1963; Glyphis et al. 1981; Knight Acacia saligna. At CGHNR, three I-ha plots in 10% , 1988) and Chacma baboons (Papio ursinus Kerr) 72% and 100% PCC stands of Acacia cyclops were (Middlemiss 1963). Efficient dispersal by rodents is demarcated, all within an area of 0.5 square km in unlikely as there are no seed-caching species in fynbos similar terrain. PCC was used as a measure of stand (Slingsby & Bond 1985), and dispersal by ants is thought density, and in A. cyclops plots, was determined using to be unimportant (Milton & Hall 1981) , despite the high the line intercept method. The dense stands differed, in incidence of myrmecochory (diaspore dispersal by ants) that A. saligna had an open understorey with some in the Cape flora (Bond & Slingsby 1983). indigenous species persisting, whereas A . cyclops formed In Australia, predispersal seed predation by several a thicker canopy precluding undergrowth. A dense A. species of Hemiptera, Coleoptera and Lepidoptera cyclops stand at Goukamma was used only for collecting results in large, though variable, losses to the annual gross annual seed-fall and seed-loss data. seed production (van den Berg 1980a, b, c) . Some Gross annual seed-fall and seed-loss were assessed Hemiptera may also feed on seeds beneath (van, using five pairs of randomly located traps in the dense den Berg 1980c). Despite their role as dispersers of A. ' stand of each species and a further 10 pairs of traps in the cyclops seeds, birds may be major predators of immature dense Goukamma stand. Traps consisted of shadescreen seeds (Gill 1985). However, there is no record of any fabric attached to 0.5-m diameter metal rings. A shallow Australian vertebrate being an important post-dispersal tray open to seed removal in the litter layer was made by seed predator of either species. pulling fabric tightly across the ring, and in the other half In South Africa, potential Acacia seed predators on of the pair, fabric was attached to the ring to form a deep the ground are granivorous birds such as the doves bag, which was then suspended from a to minimize Streptopelia senegalensis (L.) and S. capicola (Sundevall) seed removal. Traps were positioned in November 1985, (Winterbottom 1970; Glyphis et al. 1981), and rodents, before seed-fall, and were lifted after 9 months (A . particularly the striped fieldmouse [Rhabdomys pumilia saligna) and 12 months (A. cyclops). Net loss after seed­ (Sparrman)] and, in sandy soils, the gerbil [Tatera alra fall was taken to be the difference in seed content (Gray)] (David 1980). Hemipterans (Zulubius acacia­ between the two types of trap. phagus Schaffner) and larvae of one species of Seed removal rates by ants and vertebrates were Lepidoptera are the only invertebrates recorded as assessed by presenting replicates of 20 seeds of the destroying A. cyclops seeds in South Africa (Milton respective species on plywood trays (10 x 10 cm) in the 1980; Holmes et al. 1987; Holmes & Rebelo 1988), none litter layer of each plot, and giving one of four being known for A. saligna. treatments, as follows : In this study, Acacia seed removal rates in the litter 1. no exclosure (open to ants and vertebrates); layer of Acacia-infested vegetation were investigated, 2. chicken wire (0.5 cm) cage pegged over tray (to with particular reference to the following questions: exclude vertebrates); 1. Is seed dispersal by indigenous ants significant and 3. tray rim smeared with formex gel (to exclude crawling does it depend upon presence of the aril in A. cyclops? invertebrates, predominantly ants); 4. treatments (2) and (3) combined (control , to exclude 2. Do vertebrates eat significant numbers of Acacia both ants and vertebrates). seeds? Seeds for use in trials were gathered following dehis­ 3. Are seed removal rates dependent upon seed availa­ cence, in December 1985, from trees within study areas, bility in the litter layer (i.e. Acacia stand density and and stored in the laboratory in open bags. Seeds of A. seasonal seed supply)? saligna were presented intact, whereas half the A . cyclops seeds (i.e. 10 per tray) were hand-stripped of Study sites and Methods their arils. Five presentations (replicates) of the four Experimental sites were establishbd III the Silvermine treatments were placed at random locations within each (34°05'S 18°41'E), Cape of Good Hope (CGHNR: plot and these locations tagged for future trials. Trays 34°20'S 18°28'E) and Goukamma (34°03'S 22°57'E) were placed during the morning and checked 24 hours nature reserves. Mesic mountain fynbos vegetation on and 7 days later, after which they were removed. This shallow, nutrient-poor soils, derived from Table trial was repeated on a bimonthly basis in order to assess Mountain Sandstone (Moll et al. 1984) , occurs at both seasonality of removal rates. Silvermine and CGHNR. Both have a mediterranean­ Acacia seed banks were estimated by removing soil type climate, with warm dry summers and cool wet samples (A. cyclops: 20 cores of 5-cm diameter x 10 cm winters (Fuggle 1981). A mosaic of fynbos and deep; A. saligna: 25 samples of 25 x 25 cm to bedrock) kaffrarian thicket on deep recent sands (Moll et al. 1984) beneath the A cacia canopies in each plot. Samples were occurs at Goukamma, which experiences all-year rainfall sieved and seeds extracted. In addition, an A. cyclops (Fuggle 1981). The mean annual rainfall is 750 mm at plot of 42% PCC was sampled within the 0.5 km2 area of both CGHNR and Goukamma and 900 mm at Silver­ the other CGHNR plots. As seed banks were concentra­ mine (South Africa 1:250 000 rainfall maps, 1965). Sites ted under the Acacia canopies, seed bank estimates for had been subject to invasion by for at least 25 the entire plot were corrected for Acacia PCC and years, as determined from aerial photographs. plotted together with the seed input estimates. 430 S.-Afr.Tydskr. Plantk. , 1990,56(4)

Differences in seed contents between trays and bags Table 1 Mean seed density (seeds m- 2 ± SO) in bags were tested by the randomization test for matched pairs (removal minimized) and trays (open to removal (Siegel 1956). In the removal rate experiment, counts of agents) 9 months (A. saligna) and 12 months (A. seeds removed were square-root transformed to equili­ cyclops) after seed-fall brate the variance. The effects of ants, vertebrates and season on removal rates were analysed for each plot % Loss in separately, and for the two dense stands together, using Site Species Bags Trays litter two-way ANOV A with repeated measures (Programme BMDP2V, Dixon 1985). The effect of aril on A. cyclops Silvermine A. sa ligna 2102±2020 1087± 1465* 48 .3 removal rates was tested using the same ANOV A intact seed (n=5) (n=4) programme which included a split-plot design. Cape of Good A. cyclops 780±220 162 ± 77* 79.2 It must be emphasized that the seed removal experi­ Hope intact seed (11=5) (n=5) ment is pseudoreplicated (sensu Hurlbert 1984) and Goukamma A. cyclops 695 ± 618 30±54* 95.7 possible inter-site variation was not allowed for in the intact seed (11=8) (11=9) design. However, the plots of different A. cyclops deformed or 316±176 21±20 density were close by in the same habitat, thus minimi­ decayed seed zing the effects of factors other than density; and species comparisons are not generally possible at one site. seed remnants 296±334 146 ± 174 Ideally more sites should have been sampled across a produced by wider range of habitats. The results should be treated as rodents preliminary, and may be less applicable to areas outside *Difference between bags and trays significant P < 0.05 (randomi­ the Cape Peninsula. zation test for matched pairs)

Results Relative seed loss in the litter layer e .g. litter debris blown onto formex giving access to ants Annual seed-fall in dense A. cyclops stands was esti­ and heavy rain dislodging seeds from trays. mated to be 800 and 1 300 seeds m-2 , at CGHNR and Goukamma respectively; and in the dense A. saligna Effect of seed availability on removal stand 2 100 seeds m- 2 (Table 1). Differences in seed Seed removal from trays varied significantly with season contents between bags and trays (i.e. closed vs. open to in all stands: removal was lowest during and following seed removal agents) were significant at all three sites seed-fall (January- March) and greatest prior to seed­ (P < 0.05, randomization test for matched pairs) and fall (September-November) (Figure 1). In dense indicated a relative loss to annual production of 80-95% stands, seed removal by ants after 24 hours showed (A. cyclops) and 50% (A. saligna) in the litter layer seasonal variation (A. cyclops: F = 7.4, P < 0.01; A. (Table 1). Analysis of seed remains indicated that preda­ saligna: F = 9.4, P < 0.005), whereas removal by tion was reduced, but not excluded, by the bags: at vertebrates varied seasonally only in A. cyclops (A. Goukamma, 22% of seeds were destroyed in the bags, cyclops: F = 3.1, P < 0.05; A. saligna: F = 0.5, P> 0.5). compared to 74% in the trays, by rodent predation. A Among A. cyclops stands, the low-density stand gave portion of the seeds was also deformed or decayed , least seasonal variation in seed removal after 24 hours although none were germinating: these amounted to (F = 3.41, P < 0.05, compared to F = 10.6, P < 0.005 26% in the bags and 11 % in the trays (Table 1). and F = 22.1, P < 0.001, for 72% and 100% PCC stands, respectively). Seed removal in dense Acacia stands Seed removal decreased as A. cyclops canopy cover In dense stands, seed removal by ants exceeded that by increased (Figure 1). After 24 hours, ants and verte­ vertebrates (Figure 1): in A. saligna averaging (over all brates had removed similar portions of seeds in 10% and seasons) 78% and 94% of presented seeds after 24 hours 72% PCC stands, whereas in the dense stand, ants were and 7 days respectively, vs. 10% and 31 % ; and in A. more efficient than vertebrates (ants: F = 34.7, cyclops 12% and 35% vs. 5% and 31 % . The pattern of P < 0.001; rodents: F = 10.8, P < 0.01). There was an seed removal differed between the two species interaction between ants and vertebrates in the low­ (ANOV A, F = 228.3, P < 0.0001): under A. saligna it density stand only (F = 25.0, P < 0.001). After 7 days, was generally rapid and complete, primarily as a result of ants and vertebrates were highly significant seed activity, whereas it was generally , slow and removers in all three stands, with a significant inter­ incomplete under A. cyclops (Figure 1). Ground-feeding action in all except the dense stand (10% PCC: vertebrates removed a significant portion of A. cyclops F = 76.2, P < 0.001; 72% PCC: F = 16.8, P < 0.001; seeds within 24 hours (A. cyclops: F = 10.8, P < 0.01; A. 100% PCC: F = 0.4, P> 0.2). saligna: F = 2.2, P > 0.2) , but only after 7 days in A. saligna. There was an interaction effect only in the A. saligna stand after 7 days (ants x vertebrates, F = 17.1, Effect of aril on A. cyclops seed removal P < 0.005). Small losses from control trays were indic­ There was no consistent preference by ants or verte­ ative of occasional disruption, especially after 7 days: brates for A. cyclops seeds with or without arils (Figure S.Afr.l. Bot. , 1990,56(4) 431

1; ANOVA, P > 0.2 after 24 hours and 7 days In all dense Acacia stands. More seeds are lost via dispersal stands). than predation: however, decay rates in buried seed populations average 45% and 97% in A. saligna and A . Seed banks in relation to Acacia canopy cover cyclops, respectively, during the first year (Holmes Acacia cyclops seed banks were estimated to be 76 ± 41, 1989), indicating that rotting might be the major process 1579 ± 530, 1095 ± 357 and 5 170 ± 688 (X ± S.D.) occurring in dense thickets. seeds m- 2 under the Acacia canopy in 10% , 42%,72% and 100% PCC stands, respectively, compared to 7 920 Dispersal by ants ± 560 seeds m- 2 under dense A. saligna. In the low­ Ants were observed removing seeds of both species from density stand, annual seed-fall exceeded soil-seed trays , sometimes within 10 min of placement. The density, whereas in the two intermediate stands, the soil­ species involved probably belong to the widespread stored component parallelled seed input (Figure 2). fynbos genera Anoplolepis and Pheidole, which rapidly However, in the dense stand, soil-seed density exceeded remove and bury myrmecochorous seeds in subterra­ annual seed-fall by a factor of six (Figure 2). nean nests (Bond & Slingsby 1983). Seeds of another invasive alien species, Acacia longifolia (Andr.) Willd., Discussion are also removed by ants (Pieterse & Cairns in press) . Relative seed loss in the litter layer That alien acacias have established mutualistic relations A large proportion of seeds are lost in the litter layer of with indigenous ants ensures both 'escape' and 'directed'

20 A

10

o B IX: B IX: 20 UJ 20 UJ t­ t- LL LL >o NR o :E :E o UJ UJ IX: IX: C/) C/) Cl c Cl 20 UJ UJ 20 UJ UJ C/) C/) LL LL o 10 o 10 IX: IX: UJ UJ NR CO CO :E o :E 0-'---- :J :J Z Z z D Z D

10 -.0 10 NR

O~------L-~--~------O~----~~~--.w~----- 1 234 1 234 1234 123412341234 1234 1234123412341234 JAN MAR MAY JUL SEP NOV JAN MAY JUL SEP NOV Figure 1 Seed removal after (a) 24 hours and (b) 7 days (mean number out of 20 seeds ± SD, n = 5) at six times of the year from trays with different exclosures (1 = open, 2 = excluding vertebrates, 3 = excluding ants, 4 = excluding all) in a dense A. saligna plot (A) and three A. cyclops plots of different pee (B = 104, e = 724, 0 = 100%). Acacia sa ligna seeds + elaiosomes (striped); A. cyclops: seeds + arils (shaded), seeds - arils (stippled); NR = not recorded. 432 S.-Afr.Tydskr. Plantk. , 1990, 56(4)

6 been shown to drastically reduce the number of ant colonies in northern temperate regions (Seeley & Heinrich 1981). The lack of an overall preference by ants for seeds 5 with arils suggests that the aril does not contain a coevolved attractant for the indigenous ants (Bond & Slingsby 1983) and is removed as a food item only if encountered. However, after hand-stripping, a trace of t- 4 aril may remain attached to the seed, and this may have g been a sufficient attractant to the ants (cf. Bond & a. Breytenbach 1985). N ~ a: 3 Predation by vertebrates w a. Vertebrates removed a significant quantity of Acacia (/) seeds, although rate of removal was low in dense stands: C w when ants were excluded, an average of 10% and 5% w 2 were removed in A. saligna and A. cyclops stands, (/) respectively, after 24 hours. The striped field mouse u. o (Rhabdomys pumilio) was encountered in the study areas and seemed to be the principal vertebrate seed predator. (/) c All seed remnants in or adjacent to plywood trays were z ~ _____ l typical of those produced by small mammals. There was « (/) ~.. 1 no evidence of gerbils (Tatera afra) being present: soils L .... .,...",.' ... o:::> in the study plots are probably too shallow and rocky for J: ------j its burrowing activities (Smithers 1983). Avian grani­ t- o i i i i i i vores are rare in mountain fynbos vegetation (Siegfried o 20 40 60 80 100 1983) and although more common in Acacia thickets (Winterbottom 1970) , none were encountered during ACACIA CANOPY COVER ( X ) trials. Studies on Rhabdomys pumilio in a mixed A . saligna Figure 2 The relationship between soil- seed density (_ seeds and A. cyclops stand (45% total Acacia PCC) on the m-2 ± SD, n = 20 cores) and annual seed-fall (e seeds m-2 ± Cape Flats indicate that, on average, seeds form 50% of SD, n = 10 traps (R.S. Knight & M.W. Fraser, unpublished) , its diet, with a daily consumption of about 5 g (± 125 A. except in 100% PCC plot where n = 5 traps) in 1-ha plots of cyclops or ± 250 A . saligna) of seeds (David 1980). The different A. cyclops canopy cover at the CGHNR. minimum population size calculated from a continuous 5-year census was 33-256 individuals ha- 1 (David & dispersal (Howe & Smallwood 1982), and highlights the Jarvis 1985). Based on these figures, R. pumilio 1 convergent evolution of diaspore dispersal in Australia consumes a minimum of 151-1 168 seeds m-2 i (A. and South Africa (Milewski & Bond 1982). It also cyclops) or 301-2336 seeds m- 2 i 1 (A. saligna). If R. indicates that diffuse coevolution (Herrera 1985) pumilio populations are unaffected by altered vegetation between and invertebrate disperser species may structure in dense Acacia stands, they have the potential occur. However, as these ants seldom carry seeds a to consume the entire Acacia seed crop. Clearly this does distance exceeding 2-3 m (Bond & Slingsby 1983), they not occur, as large seed banks have accumulated in the are more important in maintaining Acacia soil-stored dense stands. However, the relatively small seed banks seed banks than in extending their invasive front. Whilst measured in A. cyclops stands up to 72% PCC suggests ants disperse Acacia seeds over short distances and birds that R. pumilio may consume a large proportion of the may disperse them over a few hundred metres (Glyphis seeds. That there is such a large disparity in soil-stored et al. 1981; Gill 1985), mankind is responsible for the seed density between 72% and 100% PCC stands indi­ widespread dispersal of these aliens in South Africa cates not only that predator satiation is occurring, but (Shaughnessy 1986). also that habitat alteration , possibly a reduction in Seeds of A. saligna are six times more likely to be suitable nesting habitat (grass tussocks, Smithers 1983), taken by ants in the first 24 hours than those of A. is reducing the potential R . pumilio density. cyclops. This may simply reflect the dispersal unit characteristics: A. saligna is an 'ant syndrome' and A. Effect of seed availability on removal cylops a 'bird syndrome' species (O'Dowd & Gill 1986). Both seasonal abundance of seeds and increasing seed­ Alternatively, it is possible that the thicker canopy in the rain at higher stand densities results in decreased dense A. cyclops stand may have significantly reduced removal. Although seeds are available in the litter layer ant activity relative to that in the dense A. saligna stand. of dense stands throughout the year, it appears that the Shading by an encroaching spruce/pine plantation has laboratory-stored seeds are preferred in the latter half of S.Afr.J. Bot. , 1990,56(4) 433

the year, possibly because arils and elaiosomes deterior­ of the CSIR, through its Nature Conservation ate more quickly in the field . Alternatively, seeds Programme and the University of Cape Town. presented at the surface may be more eagerly sought when the only alternative source is buried beneath the References litter. BOND, W.J. & BREYTENBACH, G.J. 1985. Ants, rodents In the low density stand there was some evidence that and seed predation in Proteaceae. S. Afr. 1. Zool. 20: ants and rodents compete for seeds: in presentations 150-154. open exclusively to them , an average of 72 % and 76% of BOND, W. & SLINGSBY, P. 1983. Seed dispersal by ants in seeds, respectively, were removed after 24 hours. Ants shrublands of the Cape Province and its evolutionary are generally faster at locating and removing seeds than implications. S. Afr. 1. Sci. 79: 231- 233. rodents (Bond & Breytenbach 1985) and may take a BOND, W. & SLTNGSBY, P. 1984. Collapse of an ant- plant larger proportion of seeds falling into the litter layer if mutualism: the argentine ant (Jridomyrex humilis) and seeds are in short supply. Predation levels in dense A . myrmecochorous Proteaceae. Ecology 65 : 1031-1037. saligna and A. cyclops stands are similar, and it is likely CRA WLEY, M.J. 1987. What makes a community invasible? that reduced canopy cover in A. saligna would lead to a In: Colonization, succession and stability, eds Gray A .J., relative increase in predation rates as found in A. cyclops Crawley, M.J. & Edwards, P.J., pp. 429-453, Blackwell , stands. London . DAVID, J.H.M. 1980. Demography and population dynamics Implications of dispersal and predation for Acacia seed of the striped fieldmouse, Rhabdomys pumilio, in alien bank dynamics A cacia vegetation on the Cape Flats, Cape Peninsula, South Africa. Ph.D. thesis, Univ. of Cape Town. In fynbos habitats being invaded by acacias, resident DAVID, J.H.M. & JARVIS, J.U.M. 1985. Population rodent populations could potentially consume the entire fluctuations , rcproduction and survival in the striped Acacia seed crop, thus reducing the invasiveness of the fieldmouse Rhabdomys pumilio on the Cape Flats, South species, were it not for the presence of ants which rapidly Africa. 1. Zool. Lond. (A) 207: 251- 276. move seeds below ground to their nests. Just as ants are DEAN, S.J. , HOLMES, P.M. & WEISS, P.W. 1986. Seed important in maintaining seed banks of indigenous biology of invasive alien plants in South Africa and Proteaceae (Bond & Slingsby 1984; Bond & Breyten­ Namibia. In: The ecology and management of biological bach 1985), they might have played a critical role in invasions in southern Africa, eds Macdonald, I.A.W. , maintaining and accumulating alien Acacia seed banks. Kruger, F.J. & Ferrar, A.A. , pp. 157- 170, Oxford Whilst the acacias have been spread into uninfested University Press, Cape Town. habitats by generalist frugivores (Glyph is et al. 1981; DIXON, W.J. 1985. BMDP statistical software. University of Knight 1988) , they appear to be preadapted to the local California Press, Berkeley. short-distance dispersal mutualism. The significance of FUGGLE, R .F. 1981. Macro-climatic patterns within the fynbos biome. National Programme for Environmental this is firstly , that seeds escape rodent predation, and Sciences, Fynbos Biome Project, CSIR, Pretoria. secondly, in a fire-prone vegetation type such as fynbos , GILL, A.M. 1985. Acacia cyclops G. Don (Leguminosae ­ populations are protected from fire: especially important Mimosaceae) in Australia: distribution and dispersal. Jl R. for non-resprouting A. cyclops. If Acacia seeds are Soc. W. Aust. 67: 59- 65. buried alongside indigenous seeds, their rapid GILL, A.M. & NESER, S. 1984. Acacia cyclops and Hakea germination after fire (Milton 1980; Taylor 1983; Jeffery sericea at home and abroad. In: Proc. IV Int. Conf. on et al. 1988) may give them a competitive edge over the Medecos, Univ. , ed. Dell , B. , pp. indigenous species. Together with a high production of 57-58, Nedlands. relatively non-parasitized, undamaged seeds (Milton GLYPHIS, J.P. , MILTON, S.J. & SIEGFRIED, W.R. 1981. 1980; Holmes & Rebelo 1988) , this may enable the Dispersal of Acacia cyclops by birds. Gecologia (Berlin) 48 : acacias to accumulate larger soil-stored seed banks than 138- 141. HERRE RA, C.M. 1985. Determinants of plant- animal the indigenes, resulting in the aliens rapidly forming coevolution: the case of mutualistic dispersal by dense stands with successive fires. The possible vertebrates. Gikos 44: 132-141. deleterious effects of the presence of alien acacias on the HOLMES, P.M. 1989. Decay rates in buried alien Acacia seed indigenous ant-plant mutualism remains to be populations of different density. S. Afr. 1. Bot. 55: investigated. 299- 303. HOLMES, P.M., DENNILL, G.B. & MOLL, E.J. 1987. Acknowledgements Effects of feeding by native alydid insects on the seed I am grateful to W.J. Bond and A.G. Rebelo for viability of an alien invasive weed, Acacia cyclops. S. Afr. 1. comments on earlier drafts of the manuscript. M. W. Sci. 83 : 580-581. HOLMES, P.M. & REBELO, A.G. 1988. The occurrence of Fraser measured the PCC in three of the A. cyclops plots seed-feeding Zulubius acaciaphagus (Hemiptera, Alydidae) and R .S. Knight and M.W. Fraser permitted use of and its effects on Acacia cyclops seed germination and seed unpublished seed-fall data for those plots. I thank the banks in South Africa. S. Afr. 1. Bot. 54: 319- 324. Cape Town City Council , the Cape Divisional Council HOWE, H.F. & SMALLWOOD, J . 1982. Ecology of seed and Department of Nature and Environmental Conser­ dispersal. Ann. Rev. Ecol. Syst. 13: 201- 228. vation for access to study areas. The project was HURLBERT, S.H. 1984. Pseudoreplication and the design of supported by the Foundation for Research Development field experiments. Ecol. Monogr. 54: 187- 211. 434 S.-Afr.Tydskr. Plantk. , 1990, 56(4)

JEFFERY, D.J., HOLMES, P.M. & REBELO, A.G. 1988. RICHARDSON, D.M. , MACDONALD, LA.W. & Effects of dry heat on seed germination in selected FORSYTH, G.G. 1989. Reductions in plant species indigenous and alien legume species in South Africa. S. Afr. richness under stands of alien trees and shrubs in the fynbos 1. Bot. 54: 28- 34. biome. S. Afr. For. 1. 149: 1-8. KNIGHT, R .S. 1986. Fruit displays of indigenous and invasive SEELEY, T. & HEINRICH, B. 1981. Regulation of alien plants in the south-western Cape. S. Afr. J. Bot. 52: temperature in the nests of social insects. In: Insect 249-255. thermoregulation, ed. Heinrich, B., pp. 159- 233, Wiley, KNIGHT, R.S. 1988. Aspects of plant dispersal in the SW New York. Cape with particular reference to the roles of birds as SHAUGHNESSY, G.L. 1986. A case study of some woody dispersal agents. Ph.D. thesis, Univ. of Cape Town. plant introductions to the Cape Town area. In: The ecology KRUGER, F.l., RICHARDSON, D.M. & VAN WILGEN, and management of biological invasions in southern Africa, B.W. 1986. Processes of invasion by alien plants. In: The eds Macdonald, LA.W., Kruger, F.J. & Ferrar, A.A. , pp. ecology and management of biological invasions in 37-46, Oxford University Press, Cape Town. southern Africa, eds Macdonald, LA.W. , Kruger, F.J. & Ferrar, A.A., pp 146- 156 , Oxford University Press, Cape SIEGEL, S. 1956. Nonparametric statistics for the behavioral Town. sciences. McGraw-Hill, Auckland. MACDONALD, T.A.W. & RICHARDSON, D.M. 1986. SIEGFRIED, W.R. 1983 . Trophic structure of some Alien species in terrestrial ecosystems of the fynbos biomc. communities of fynbos birds. Jl S. Afr. Bot. 49: 1-43. In: The ecology and management of biological invasions in SLINGSBY, P. & BOND, W.J. 1985. The influence of ants on southern Africa. eds Macdonald, LA.W. , Kruger, F.J. & the dispersal distance and seedling recruitment of Ferrar, A.A., pp. 77-92, Oxford University Press, Cape Leucospermum conocarpodendron (L.) Buck (Proteaceae). Town. S. Afr. 1. Bot. 51: 30- 34. MIDDLEMISS, E. 1963. The distribution of A. cyclops in the SMITHERS, R.H.N. 1983. The mammals of the southern Cape Peninsula area by birds and other animals. S. Afr. 1. African subregion. Univ. of Pretoria. Sci. 59: 419- 420. SORENSEN, A.E. 1985. Seed dispersal and the spread of MILEWSKI, A.V. & BOND, W.l. 1982. Convergence of weeds. In: Proc. VT Tnt. Symp. Bio\' Control Weeds, cd. myrmecochory in mediterranean Australia and South Delfosse, E.S., pp. 121-126, Vancouver, Canada. Agric. Africa. In: Ant- plant interactions in Australia, ed. Can. Buckley R.C. , pp. 89- 98 , Dr W. Junk Publishers, The TA YLOR, H.C. 1983. The vegetation of the Cape of Good Hague. Hope Nature Reserve. Bothalia 14: 779-784. MILTON, S.]. 1980. Australian acacias in the south-western TAYLOR, H.C., MACDONALD, S. & MACDONALD, Cape: preadaptation, predation and success. Proc. 3rd Nat. LA.W. 1985. Invasive alien woody plants in the Cape of Weeds Conf. S. Africa, pp. 69- 78, Balkema, Cape Town. Good Hope Nature Reserve. II. Results of a second survey MILTON, S.J. & HALL, A.V. 1981. Reproductive biology of from 1976 to 1980. S. Afr. J. Bot. 51: 21-29. Australian acacias in the south-western Cape Province, VAN DEN BERG, M.A. 1980a. Natural enemies of Acacia South Africa. Trans. R. Soc. S. Afr. 44: 465-487. cyclops A. Cunn. ex G. Don and Acacia saligna (Labill.) MOLL, E.J. , CAMPBELL, B.M. , COWLING, R.M. , Wend\. in Western Australia. 1. Lepidoptera. BOSSI, L., JARMAN, M.L. & BOUCHER, C. 1984. A description of major vegetation categories in and adjacent Phytophylactica 12: 165-167. to the fynbos biome. S. Afr. Nat. Sci. Prog. Report No. 83, VAN DEN BERG, M.A. 1980b. Natural enemies of Acacia CSIR, Pretoria. cyclops A. Cunn. ex G. Don and Acacia saligna (Labill.) MOONEY, H.A. & DRAKE, J.A. 1987. The ccology of Wend\. in Western Australia. IT. Coleoptera. biological invasions. Environment 29: 10-37. Phytophylactica 12: 169- 171. O'DOWD, D.J. & GILL, A.M. 1986. Secd dispersal VAN DEN BERG, M.A. 1980c. Natural cnemies of Acacia syndromes in Australian Acacia. In: Seed dispersal, ed. cyclops A. Cunn. ex G. Don and Acacia saligna (Labill.) Murray, D.R. , pp. 87-121 , Academic Press, Australia. Wend\. in Western Australia. III. Hemiptera. PIETERSE, P.l. & CAIRNS, A.L.P. (in press). Phytophylactica 12: 223- 226. Investigations on the removal by animals of Acacia WINTERBOTTOM, J.M. 1970. The birds of the alien Acacia longifolia () seed from the soil surface at Banhoek thickets of the south-western Cape. Zoologica Africana 5: in the SW Cape. S. Afr. 1. Plant Soil. 49-57.