558 S.-Afr. Tydskr. Plantk., 1988, 54(6): 558--564 Response of the fynbos shrubs Protea repens and Erica p/ukenetii to low levels of nitrogen and phosphorus applications
A.J. Lamb* and E. Klaussner South African Forestry Research Institute, Jonkershoek Forestry Research Centre, Private Bag X5011, Stellenbosch, 7600 Republic of South Africa
Accepted 17 June 1988
The responses of five-year-old fynbos shrubs, Protea repens (L.) L. and Erica plukenetii L., to low levels of nitrogen and phosphorus application were examined. Ammonium nitrate (2 9 N m-~, superphosphate (0,2 9 P m-2) and ammonium nitrate plus superphosphate (2 9 N m -2 + 0,2 9 Pm -2) were applied in early spring 1982. Erica plukenetii responded to the phosphorus and particularly nitrogen plus phosphorus treatment with increases in height, stem diameter, crown cross-sectional area and biomass by the end of the second growing season. These responses were lower on plots treated with nitrogen than on the control. Vegetative growth of Protea repens on the fertilized plots, particularly those treated with nitrogen and nitrogen plus phosphorus, was lower than on the control. Nitrogen addition increased shoot production in the second year while phosphorus addition increased flower production. Low levels of nutrient application are quite capable of altering morphological characters of fynbos species. Nutrient enrichment by atmospheric inputs might therefore lead to changes in species composition and could result in increased invasion of fynbos areas by alien herbaceous species.
Die reaksie van vyfjaar-oue fynbos struike, Protea repens (L.) L. en Erica plukenetii L., op die toediening van klein hoeveelhede stikstof en fosfor, is ondersoek. Ammoniumnitraat (2 9 N m-2), superfosfaat (0,2 9 P m-2) en ammo niumnitraat -plus-superfosfaat (2 9 N m -2 + 0,2 9 P m - 2) is in die vroee lente van 1982 toegedien. Erica plukenetii het op die fosfor- en veral op die stikstof- plus-fosfortoedienings gereageer, aan die einde van die tweede groei-seisoen met 'n toename in hoogte, stamdeursnee, kroonoppervlak-deursnee en biomassa. Op die persele wat met stikstof behandelde persele, veral die wat met stikstof en stikstof-plus-fosfor behandel is, was laer as op die kontrole. Die toe voeging van stikstof het loot-aanwas in die tweede jaar opgestoot terwyl fosfor blomproduksie bevorder het. Lae vlakke van voedingstof-toediening is in staat om die morfologiese karaktereienskappe van fynbos spesies te verander. Voe voedingstof-toediening is in staat om die morfologiese karaktereienskappe van fynbos spesies te verander. Voe dingstotverryking deur atmosferiese invoere mag dus lei tot veranderings in spesie-samestelling en mag daartoe lei dat fynbos-gebiede in 'n groter mate deur kruidagtige indringerspesies ingeneem word.
Keywords: Fynbos, nitrogen, nutrient additions, phosphorous, south-westem Cape *To whom correspondence should be addressed
Introduction nutrients accumulated in standing dead biomass and litter The soils of the South African Fynbos Biome are generally (Rundel & Parsons 1980). low in mineral elements and are considered to be closely Nitrogen and phosphorus have been recognized as the allied to those of Australian heathlands (Groves 1983; Mit elements most likely to control rates of growth in medi chell et al. 1984). These oligotrophic (strongly-leached) terranean ecosystems (Read & Mitchell 1983) . Phosphorus soils are characterized by a pH of less than 6,0, total nitro has been shown to increase the growth of Australian heath gen less than 0,10%, total phosphorus less than 0,02% and on oligotrophic soils (Specht 1963). Californian chaparral total exchangeable cations less than 5 meq 100 g -I soil on mesotrophic soil shows a response to additions of nitro (Specht & Moll 1983). In contrast, the other mediterra gen rather than phosphorus (Hellmers et al. 1955). Fertili nean ecosystems (California, Chile and the Mediterranean zation with a combination of phosphorus and nitrogen also Basin) have mesotrophic (moderately-leached) soils results in increased growth in Australian heath (Specht (Specht & Moll 1983). 1963; Heddle & Specht 1975) and in chaparral (McMaster The relatively high proportion of available forms of N et at. 1982). and of other nutrients in ash is generally considered to be Previous fertilizer experiments, especially those of the major factor promoting rapid re-growth following fire Specht (1963) and Siddiqi et at. (1976) involved high levels in mediterranean ecosystems (Rundel et al. 1983). Nitro of fertilizer application. Specht (1963) added 140 g m -2 su gen and phosphorus estimated as likely to be released dur perphosphate, 75 g m -2 sodium nitrate, 1,57 g m -2 copper ing a fire in 21-year-old mountain fynbos at 10nkershoek sulphate and 1,57 g m -2 zinc sulphate over a 3-year period. amounted to 15,8 and 0,64 g m-2 respectively (van Wilgen Siddiqi et al. (1976) doubled the amounts of phosphorus, & Ie Maitre 1981), but the proportion returned to the soil magnesium and potassium available immediately after a was not quantified. A fire in a coastal fynbos ecosystem in fire. creased the total N concentration by 6,6 g N m -2 at the soil This paper investigates the growth response of Protea surface (Stock & Lewis 1986a). Although no significant repens (L.) L and Erica plukenetii L. to low levels of nitro change in total phosphorus levels was found, resin-extract gen and phosphorus additions, the responses to high levels able phosphorus did increase significantly (Brown & Mit having clearly been demonstrated elsewhere (Specht 1963; chell 1986). Such increases in nutrient availability are, how Siddiqi et al. 1976). Protea repens and Erica plukenetii were ever, shortlived. Plant growth rates decline steadily as post specifically chosen for investigation as both are widely dis fire age of the vegetation increases (Rundel & Parsons tributed in the fynbos of the south-western Cape (Rourke 1979). This decline is correlated with increased amounts of 1980; Baker & Oliver 1976). As fynbos ecosystems are S. Afr. 1. Bot., 1988,54(6) 559
subject to low levels of nutrient enrichment by atmospheric The mature vegetation of the study area comprises a Tall inputs (Stock & Lewis 1986b; Brown et al. 1984; van Wyk Mid-dense Proteoid Shrubland with a Mid-dense Restioid 1984) and to a lesser extent by the application of fire retar and Ericoid Shrubland understory (Campbell et al. 1981). dants and fertilizer drift, these inputs may represent an Protea rep ens and Erica plukenetii were the dominant important input of nutrients to natural ecosystems on species on the study site which was last burnt in 1977, with nutrient-poor soils, could alter fynbos growth responses mean heights of 0,86 and 0,36 m respectively. Numerous over time and could also lead to more rapid invasion of hemicryptophytes and some geophytes and herbs were also fynbos areas by alien herbaceous species. present in the understory.
Study area Methods Sosyskloof (34° OO'S, 18° 57'E) is situated in the Jonkers Fertilizer treatments hoek State Forest, 15 km from Stellenbosch in the south Twelve 5-m x 5-m plots, separated by 5-m strips to prevent western Cape Province, South Africa. The soil is derived overlapping of fertilizer effects, were laid out in July 1982. from coarse-grained porphyritic granite of the Pre-Cape Four fertilizer treatments replicated three times, were System (Fry 1987). Total nitrogen and phosphorus levels of randomly assigned to the plots. The treatments were: no the Sosyskloof soils, as determined by the Saasveld Fores fertilizer (control), phosphorous only (P), nitrogen only try Research Centre soil laboratories, were 0,11% and (N), and nitrogen plus phosphorus (N+P). 0,012%, respectively. During the period 12 to 18 August 1982, 3,572 g m -2 of The climate of the Jonkershoek Valley is mediterranean, ammonium nitrate (1 g m -2 N) and 1,905 g m -2 of super Koppen's (1931) humid-mesothermal (type Csb) with a dry phosphate (0,2 g m -2 P) dissolved in 0,5 I m -2 of water, were summer and the average temperature of the warmest applied evenly over the soil surface. An additional},01 m-2 month below 22°e. Annual precipitation in Sosyskloof of water was applied immediately following the fertilizer averaged 1 426 mm from 1965 to 1984. Total precipitation application to wash any fertilizer solution that may have for the study periods April 1982 to March 1983 and April landed on plant foliage, onto the soil surface. An equal 1983 to March 1984 were 1 359 mm and 1 415 mm, re quantity of water (1,5 I m -2) was applied to the control spectively. plots. During the period 18 to 19 October, a further 3,572 g A relatively homogeneous site of 0,15 ha was chosen for m -2 of ammonium nitrate (1 g m -2 N) was applied to the N the fertilizer experiment. The site is situated at 500 m alti and N + P plots in the same manner. The ammonium nitrate tude and has an average slope of 17° with a SE aspect. was applied during two periods to prevent rapid leaching.
0,5 10 0,03
9
0.4 r+ 8
7 ~ 0,02 E .s T 0.3 Q; 6 Q3 E co I '6 1: en E 5 Q; Q) I U5
0,2 4
0,01 3
0,1 2
561 241 877 138 561 241 877 138 561 241 877 138 b a b a a b c d a a b c
c P N N+P c P N N+P c P N N+P
Figure 1 Mean heights, stem diameters and crown areas of Erica plukenetii shrubs at the end of the 1983 growing season after various fertilizer treatments at the beginning of the 1982 growing season. The analysis of variance significance levels were: height, P < 0,01; stem diameter, P < 0,01; crown area, P < 0,01. C, control; N, addition of nitrogen; P, addition of phosphorus; N + P, addition of both nitrogen and phosphorus. Vertical lines indicate ± S. E. The sample size is given in each column. Different letters in columns indicate significant differences between means (P < 0,05; non-parametric Student Newman-Keuls multiple comparison test). S.-Afr. Tydskr. Plantk., 1988, 54(6) 560
50
40
30
9 (f) (f) E'" 0 iii 20
10
75 69 65 61
b b b a a a a a a a a a a a N N+P c P N N+P c P N N+P c P N N+ P c P Total Branches Leaves Inflorescences
Figure 2 Mean dry weight of Erica plukenetii shrubs at the end of the 1983 growing season after various fertilizer treatments in 1982. The analysis of variance significance levels were: total mass, P < 0,07; branch mass, P < 0,05; leaf mass, P < 0,07; inflorescence mass, P < 0,3. Symbols and statistical analyses as in Figure 1.
0,15 1,0 20 I-+- 0,9 18
0,8 K-+r+ 16
0,7 • 14 0,10
E . f-- ~ 0,6 .s 12 Q; 1 Qi E '"~ 10 I 0,5 '6'" '"c:: E E >: Ol e ~r+ a:; gj () I 0,4 8
0,05 0,3 6
0,2 4
170 136 178 163 170 136 178 163 0,1 170 136 178 163 2 b b b a b b b a a a a a b b b
P N N+P C P N N+ P C P N N+P c Figure 3 Mean heights, stem diameters and crown areas of Protea repens shrubs at the end of the 1983 growing season after various fertilizer treatments at the beginning of the 1982 growing season. The analysis of variance significance levels were: height, P < 0,01; stem diameter, P < 0,05; crown area, P < 0,01. Symbols and statistical analyses as in Figure 1. S. Afr. J. Bot., 1988,54(6) 561
A nitrogen addition of 2 g m -2 and a phosphorus addition inflorescence oven-dry (7ooe to constant weight) weights of 0,2 g m -2 might be considered low but was estimated to were determined, be approximately equal to the annual nitrogen and phos The number of 1982 shoots per 1981 shoot and the number phorus turnover required to maintain the rates of biomass of 1983 shoots per 1982 shoot were counted on all Pro tea accumulation (Read & Mitchell 1983). These levels of repens shrubs after the 1983 growing season, The lengths of application are also low when compared to the amounts the 1982 and 1983 shoots were also measured on all shrubs released after fire in fynbos (Stock & Lewis 1986a; Brown after the 1983 growing season. & Mitchell 1986; van Wilgen & Le Maitre 1981). Bud and inflorescence production Rainfall occurred within one day of each application (7,3 and 2,4 mm respectively) but no overland flow was ob The number of aborted buds and the total number of in served on any of the treatment sites. Direct losses of ap florescences for both 1982 and 1983 were counted on each Protea repens shrub 4 months after the 1983 growing sea plied fertilizers from the plots were therefore assumed to be son. Total bud set was determined by adding the total num negligible. ber of inflorescences and the number of aborted buds. Only Determination of response to fertilization shrubs with buds or inflorescences were used in the analysis The height, stem diameter at ground level and maximum of the data. and minimum crown diameters of each Pro tea repens and Statistical analyses Erica plukenetii shrub were measured after the 1983 grow The data sets were plotted as histograms and visually exa ing season (April, 1984). Maximum and minimum crown diameters were used to determine the crown cross-section mined for normality. Analysis of variance using the Gene al areas, which is an estimate of canopy cover of the shrubs. ral Linear Models procedure (SAS 1985) was used to test a multisample hypothesis of equal means. The non-parame Biomass and shoot production tric Student Newman-Keuls multiple comparison test All Erica plukenetii shrubs on the fertilized plots were har (P = 0,05) was used to test between which means dif vested after the 1983 growing season. On the control plots ferences existed. where a large number of shrubs were available, 75 shrubs were randomly selected and harvested. Branch, leaf and Results Response of Erica plukenetii to nutrient addition Morphological 1983 1982 The heights and stem diameters of Erica plukenetii shrubs treated with P and N + P were significantly greater than the 1,6 r+- !f-f+ 1982 1983 100
1,4 + (5 90 0 f+ .c ...I T 30 · 0,4 20 0,2 769 661 664 760 1107 891 854 1074 1848 1318 1414 1698 10 1049 827 785 1026 a a a a a a a a a a b c d b b b b b b c P N N+P c P N N+ P c P N N+P c P N N+P Figure 4 Mean ratio of vegetative shoots to previous vegetative Figure 5 Mean shoot lengths of Prolea repens shrubs at the end of shoots in PrOlea repens shrubs at the end of 1982 and 1983 growing the 1982 and 1983 growing seasons after various fertilizer treatments seasons after various fertilizer treatments in 1982. The analysis of in 1982. The analysis of variance significance levels were: 1982 shoot variance significance levels were: 1982 shoot ratio, P < 0,01; 1983 lengths, P < 0,01; 1983 shoot lengths, P < 0,01. Symbols and statisti shoot ratio, P < 0,03. Symbols and statistical analyses as in Figure 1. cal analyses as in Figure 1. S.-Afr. Tydskr. Plantk. , 1988,54(6) 562 control after two growing seasons (Figure 1). The same pat Shoot production tern was observed for crown cross-sectional areas but only The number of 1982 shoots per 1981 shoot was ranked as the crown cross-sectional areas of shrubs treated with N + P follows: N < P < N + P < control (Figure 4). Only .the differed significantly from the control. number of 1982 shoots per 1981 shoot on shrubs treated The stem diameters and crown cross-sectional areas of with N was significantly lower than the control. shrubs treated with N were significantly lower than the con The number of new shoots per previous season's shoot in trol (Figure 1). The heights of shrubs treated with N were 1983 was higher on all treatments than in 1982 (Figure 4). also lower than the control but did not differ significantly. The number of 1983 shoots per 1982 shoot on the fertilized plots was lower than the control. Only the number of 1983 Biomass production shoots per 1982 shoot on shrubs treated with P was signifi Total biomass of Erica plukenetii shrubs was significantly cantly lower than the control. higher on the N + P treated plots than on the N treated plots The most striking difference between the 1982 and 1983 (Figure 2). No significant responses were detected among fertilizer responses was that the number of new shoots per the control, P and N + P treatments. The same effect was previous shoot on shrubs treated with N was highest in 1983 observed for the biomass of branches and leaves but not for whereas the ratio in this treatment was the lowest in 1982. inflorescence weight (Figure 2). The lengths of the 1982 shoots on shrubs on the fertilized plots were lower than the control, this effect decreasing in the order of P, Nand N + P (Figure 5). Only the lengths of Response of Protea repens to nutrient addition the 1982 shoots on shrubs treated with N + P were signifi Morphological cantly lower than the control. The heights of Protea repens shrubs on the fertilized plots Mean shoot lengths on all treatments were lower in 1983 were significantly lower than the control after two growing than in 1982 (Figure 5). The lengths of the 1983 shoots on seasons (Figure 3). Stem diameters of shrubs on the ferti shrubs on the fertilized plots showed the same pattern of lized plots were also lower than the control but only stem response as in 1982 although the lengths of the 1983 shoots diameters of shrubs treated with N differed significantly were significantly lower than the control for all fertilizer from the control. The pattern for crown cross-sectional treatments. areas was the same as for stem diameter except that crown cross-sectional areas of shrubs treated with Nand N + P Bud and inflorescence production were significantly lower than the control. Bud set in Protea repens occurs at the end of its growing 1983 3,2 1982 1983 1982 3,0 2,8 r-- f- 2,6 r--I- 2,4 ...- f- 2,2 ...- f- 2,0 r- Io-- 'f) r--I- "D f-f- :::J r D 1,8 - f-- 15 r ([; D 1,6 E T ...- f- :::J ~>-lt Z l-f- 1,4 1 l-f-- 1,2 1,0 0,8 0,6 - 0,4 82 54 70 72 29 22 17 28 50 37 35 62 0,2 . 69 40 44 56 a a a a a a a a a a a a a a a a N+P c P N N+P c P N N+P c P N N+P c P N Bud set Aborted buds Figure 6 Mean number of buds and aborted buds on Protea repens shrubs at the end of the 1982 and the 1983 growing seasons after various fertilizer treatments in 1982. The analysis of variance significance levels were: 1982 bud set, P < 0,20; 1983 bud set, P < 0,08; 1982 aborted buds, P < 0,53; 1983 aborted buds, P < 0,39. Symbols and statistical analyses as in Figure 1. S. Afr. 1. Bot. , 1988,54(6) 563 season (April). The buds develop during the next growing on shrubs treated with N + P was greater than on shrubs season (September to April) and flower between April and treated with N whereas in 1983 it had been lower. August of the following year. Bud set in 1982 therefore occurred prior to the fertilizer applications and the number Discussion of buds per shrub WaS lower on the plots to which fertilizer This study has shown that responses to nutrient addition was added, than on the control (Figure 6) . are generally similar to those found by Specht (1963) in The total number of buds set in 1983 was higher on Australian heathlands. Phosphorus addition promotes shrubs treated with P and N + P than on the control and on vegetative growth in Erica plukenetii and increases flower shrubs treated with N, respectively (Figure 6). Neither the ing in Protea repens. 1982 nor the 1983 responses differed significantly between Erica plukenetii responded positively to phosphorus treatments. addition whereas nitrogen addition resulted in a negative The number of aborted buds in 1982 was lower on the response. Read (1983) found that mycorrhizal plants had fertilized plots than on the control (Figure 6). In 1983 the access to organically complexed nitrogen sources which re number of aborted buds was higher on shrubs treated with present the greatest reserve of nitrogen in heathland soils. Nand N + P than on the control, but did not differ signifi Additional nitrogen may therefore not result in increased cantly. The most striking difference between the 1982 and nitrogen uptake in Erica piukenetii. The addition of nit 1983 responses was that the number of aborted buds was rogen plus phosphorus produced the greatest increase in highest on shrubs treated with N+P in 1983, whereas in growth. This may be attributed to the interaction between ] 982 it was the lowest in this treatment. nitrogen and phosphorus in plant nutrition (Groves 1983). The number of inflorescences per shrub in 1983, was Fertilization depressed vegetative growth of Pro tea re greatest on shrubs treated with P followed by the control, pens particularly on the Nand N + P treatments. Phos the N and the N + P treatments (Figure 7) . In 1984 the num phorus addition resulted in increased flower production. ber of inflorescences on shrubs treated with P was signifi The lack of significant growth response in Pro tea repens can cantly greater than on shrubs treated with Nand N + P, but not be ascribed to its low demand for nitrogen together with the the control (Figure 7) . The number of 1984 inflorescences potential for internal cycling of limited nutrients (Stock & Lewis 1984). The differences in phenology of the two species and the 1983 1984 3,8 timing of fertilizer addition could have contributed to the different responses observed. Erica plukenetii has a growth 3,6 1 flush in December/January but can grow throughout the r- year given the correct environmental, water and nutrient 3,4 - · requirements (E. Klaussner, unpubl. data). Fertilizer was 3,2 applied in August and October which would allow for nut rient uptake before the beginning of the main growth flush. 3,0 Vegetative growth in Protea rep ens starts at the beginning 2,8 ~ of September. Nutrient uptake may have begun before the - fertilizer was applied and additional nutrients may not have 2,6 influenced vegetative growth during the first year. The amount of fertilizer may have been too small to produce 2,4 any significant effect in the second year. Probably the most important response to nutrient addi 0,6 and ultimately led to a change in species composition and a decline of the heath vegetation (Specht et ai. 1977) . 0,4 Fynbos ecosystems are subject to nutrient enrichment by 51 24 31 39 46 27 40 20 atmospheric inputs (Stock & Lewis 1986b; Brown et al. 0,2 · b b b a a a a a a 1984; van Wyk 1984) and P and (apparently) N has been found to accumulate in upland watersheds (Verry & Tim c P N N+P c P N N+P mons 1982). This could eventually lead to changes in spe Figure 7 Mean number of Protea repens inflorescences in 1983 and cies composition in the fynbos. On the other hand, the 1984 after various fertilizer treatments in 1982. The analysis of vari ance significance levels for total inflorescences in 1983 was P < 0,16 application of fire retardants which are high in phosphates and in 1984, P < 0,01. Symbols and statistical analyses as in Figure 1. could create immediate imbalances between phosphorus and 564 S.-Afr. Tydskr. Plantk., 1988, 54(6) nitrogen resulting in negative effects on seedling regenera READ, D.l. & MITCHELL, D.T. 1983. Decomposition and mine tion. Small increases in soil fertility by atmospheric inputs ralisation processes in mediterranean-type ecosystems and in heath lands of similar structure. In: Mediterranean-type could also lead to increased invasion of fynbos areas by ecosystems: The role of nutrients, eds Kruger, F.J., Mitchell , alien herbaceous species. D.T. & Jarvis, J.U.M. Springer-Verlag, Berlin. ROURKE, J.P. 1980. The proteas of southern Africa. Tafelberg, Acknowledgements Cape Town. We thank Mr G .M. Bailey and Mrs L. van Vuuren for their RUNDEL, P.W., BATE, G.c., LOW, A.B., MILLER, P.c., MIL LER, P. & MITCHELL, D.T. 1983. Nutrient cycling processes. assistance with the field work . Dr F.J. Kruger, Mr D .C. Ie In: Mineral nutrients in mediterranean ecosystems, ed. Day, J.A. Maitre, Mr P.T. Manders, Prof. D.T. Mitchell, Mr D.M. South African National Scientific Programmes Report No. 71 , Richardson, Dr B.W. van Wilgen, Mr E.T.F. Witkowski CSIR, Pretoria. and anonymous referees provided useful comment on the RUNDEL, P.W. & PARSONS, D.J. 1979. Structural changes in chamise (Adenostoma fasciculatum) along a fire-induced age text. The work forms part of the Conservation Forestry Re gradient. 1. Range Mt 32: 462-466. search Programme of the South African Forestry Research RUNDEL, P.W. & PARSONS, D.J. 1980. Nutrient changes in two Institute, Department of Environment Affairs. chaparral shrubs along a fire-induced age gradient. Am. 1. Bot. 67: 51-58. SAS INSTITUTE INC. 1985. SAS User's Guide: Statistics, VersionS References edition, SAS Institute Inc. , Cary, North Carolina. BAKER, H.A. & OLIVER, E.G.H. 1967. Ericas in southern Africa. SIDDIQI, M.Y., MYERSCOUGH, P.J. & CAROLlN, R.C. 1976. Purnell, Johannesburg. Studies in the ecology of coastal heath in New South Wales. IV. BROWN, G. & MITCHELL, D.T. 1986. Influence of fire on the soil Seed survival, germination, seedling establishment and early phosphorus status in sand plain lowland fynbos, south-western growth in Banksia serratifolia Salisb., B. aspleniifolia Salisb. and B. Cape. S. Afr. 1. Bot. 52(1): 67-72. ericifolia L.F. in relation to fire: temperature and nutritional ef BROWN, G., MITCHELL, D.T. & STOCK, W.D. 1984. Atmos fects. Ausl. 1. Eco!. 1: 175-183. pheric deposition of phosphorus in a coastal fynbos ecosystem of SPECHT, R.L. 1963. Dark Island Heath (Ninety-Mile Plain , South the south-western Cape. 1. Ecol. 72: 547-551 . Australia). VII. The effect of fertilizers on composition and CAMPBELL, B.M. , COWLING, R.M. , BOND, W ., KRUGER, growth, 1950-60. Aust. 1. Bot. 11: 67-94. F.J. in collaboration with BANDS, D.P., BOUCHER, c., SPECHT, R.L., CONNOR, D.J. & CLIFFORD, H.T. 1977. The MOLL, E.J., TAYLOR, H.c. & VAN WILGEN, B.W. 1981. heath-savanna problem: the effect of fertilizer on sand-heath Structural characterization of vegetation in the fynbos Biome. vegetation of North Stradbroke Island, Queensland. Ausl. 1. South African National Scientific Programmes Report No. 52. Eco!.: 179- 186. CSIR, Pretoria. SPECHT, R.L. & MOLL, E.J. 1983. Mediterranean-type heathlands FRY, M. 1987. Soils of the Pella and Swartboschkloof Fynbos Re and sclerophyllous shrublands of the world: An overview. In: Medi search sites. M.Sc. thesis, Univ. of Stellenbosch. terranean-type ecosystems: The role of nutrients, eds Kruger, F.J., GROVES, R.H. 1983. Nutrient cycling in Australian heath and Mitchell , D.T. & Jarvis. J.U.M. Springer-Verlag, Berlin. South African Fynbos. In: Mediterranean-type ecosystems: The STOCK, W .D. & LEWIS, O.A.M. 1984. Uptake and assimilation of role of nutrients, eds Kruger, F.J. , Mitchell , D.T. & Jarvis, nitrate and ammonium by an evergreen fynbos shrub species. 1.U.M. Springer-Verlag, Berlin. Protea rep ens L. (Proteaceae). New Phyto!. 97: 261 - 268. HEDDLE, E.M. & SPECHT, R.L. 1975 . Dark Island Heath STOCK, W .D . & LEWIS, D.A.M. 1986a. Soil nitrogen and the role (Ninety-Mile Plain , South Australia). VTII. The effect of fer of fire as a mineralizing agent in a South African coastaIfynbos tilizers on composition and growth, 1950-1960. Aust. 1. Bot. 23: ecosystem. 1. Eco'. 74: 317-328. 151-164. STOCK, W.D. & LEWIS, O.A.M. 1986b. Atmospheric input of nit HELLMERS, H., BONNER, J.F. & KELLEHER, 1.M. 1955. Soil rogen to a coastal fynbos ecosystem of the south-western Cape fertility: a watershed management problem in the San Gabriel Province, South Africa. S. Afr. 1. Bot. 52: 273-276. Mountains of southern California. Soil Sci. 80: 189-197. VAN WILGEN, B.W. & LE MAITRE, D.C. 1981. Preliminaryes KOPPEN, W. 1931. Grundriss der klimakunde. De Gruyter, Berlin. timates of nutrient levels in fynbos vegetation and the role of fire in MCMASTER, G.S., lOW, W.M. & KUMMEROW, 1. 1982. Re nutrient cycling. S. Afr. For. 1. 119: 24-28. sponse of Adenostoma fascicu!atum and Ceanothus greggii VAN WYK, D.B. 1984. Ioon-invoere en pH van reenval in die chaparral to nutrient additions. 1. Ecol. 70: 745-756. bergopvanggebiede van die Wes-Kaap. Abstracts of the Seminar MITCHELL, D.T., BROWN, G.l . & lONGENS-ROBERTS, on atmospheric interaction related studies in the western Cape. S.M. 1984. Variation of forms of phosphorus in the sandy soils of pp. 41-43. Kirstenbosch Botanical Gardens, CSIR, Pretoria. coastal fynbos, south-western Cape. 1. Ecol. 72: 575-584. VERRY, E.S. & TIMMONS, D.R. 1982. Waterborne nutrient flow READ, D.J. 1983. The biology of mycorrhiza in the Ericales. Can. 1. through an upland-peatland watershed in Minnesota. Ecology Bot. 61: 985-1004. 63(5): 1456-1467.