Dune advancement 1937 — 1977 at the Mlalazi Nature Reserve , Mtunzini , Natal , South Africa , and a preliminary vegetation-succession chronology

Foredune advancement on a 2 km coastline north of the Siaya Lagoon Mouth was studied using air photos. Between 1937 and 1977 the dunes advanced about 95 m (2,4 m per year). Vegetation was dated according to its position on a profile. If a 2,4 m per year seaward advancement of the dunes is assumed, the following succession chronology is obtained: Scaevola thunbergii Foredunes from 0 to 30 years; Passerina rigida Open Dune Scrub 30 to60 years; Closed Dune Scrub 60 to 90 years and Dune Forest beginning at about 90 years. Variation in dune advancement rates on different coastal stretches and for different time intervals was observed and will be reported on later. This dune succession chronology should, therefore, only be seen as a first rough approximation.


INTRODUCTION
H untley (1977), referring to vegetation mapping in South A fric a , indicated that little e ffo rt had been expended in m onitoring the rates and kinds o f change taking place in vegetation.The a vailability o f air photos (dating from 1937 to 1979) o f M tu n z in i, N atal, afforded the author the opp ortun ity to contribute inform ation in this neglected field.Objectives o f this study were to evaluate the adequacy o f air photos in m onitoring dune advancement and vegetation changes; to measure foredune advancement and calculate advancement rates; and to in fe r a dune-succession chronology by dating the com m uni ties according to th e ir position in a pro file .
The tw o km coastline studied is part o f the M lalazi Nature Reserve and stretches from the Siaya Lagoon M outh (latitude 28° 58' South, longitude 30° 45' 45" East) north-eastwards.The coast between M lalazi Lagoon and the Tugela M outh is one o f the few coastlines in N atal where active sand deposition and foredune advancement is occurring.The dunefield there is form ed by a series o f about 16 phytogenic ridges parallel to the coast and is about 400 m broad.Edwards (1967) and M o ll (1972) described dune communities o f this region.

METHODS AND METHODOLOGICAL CONSIDERA TIONS
The dune vegetation and its advancement were studied by direct inspection o f air photos.The inform ation was transferred onto a 1:10 000 base map using a Bausch & Lom b Z T-4 Zoom Transfer Scope (= Z T S ).This instrum ent was also employed to draw the base map from the orthophoto map 2831 D D 21 'M tu n z in i' (1977).In terpretation o f aerial photo graphs was aided by a Topcon Stereoscope.Roads were m ainly used as matching lines.L im its o f the foredunes were drawn onto overlays and the ir progression was measured w ith a 1/10 mm graduated ocular.A b o u t one hundred systematic sampling lines were drawn perpendicular to the coast on the base map starting at the Siaya Lagoon M o uth and going northwards.A north-eastward longshore advance ment o f the Siaya Lagoon M o u th o f about 740 m from 1937 to 1977 was found, i.e. about 17,4 m per annum.A ir photos 54693 (Job 117, o f 1937.05.05), 6447 (Job 400, 1957.05.24), 5665 (Job 608, 1969.08.18) and the., orthophoto 2831 D D 21 (Job 498/91, June 1979) were used.G round tru th was gathered on 1974.09.15, 1975.11.03, 1978.06.29, 1980.03.27, 1980.05.15, 1980.06.11 and 1980.06.12.
The potential and lim ita tio n s o f air photos in vegetation studies have been discussed by Edwards (1972) and, in relation to dune vegetation in Z u lu la n d , by Weisser (1979).Concerning this study the follow ing points should be borne in m ind.The photographs are at different scales, and were not taken at the same time o f the year which im pairs th e ir com parability.A n o th er factor is the d iffe rin g resolution o f the air photos, the 1937 photos being the weakest.
The evaluation o f the results is com plicated because dune advancement is not a linear process.I t occurs in pulses, each pulse corresponding to the detection o f an additional dune ridge on the air photo.There is a time lag between the establishment o f Scaevola thunbergii seedlings, the accum ulation o f sand and the appearance o f the new ridge on the air photo.These dunes are at first isolated and later colaesce to a coastal parallel ridge (Figs 1 & 2).
The distance between the dune ridges is not uniform .A t Twinstreams, south o f the Siaya M outh, the average distance between the last fo u r dune crests was 25 m.Distances measured on the transect published by M o ll (1972) over the whole dune field suggest a distance o f about 22 m between crests.

RESULTS AND DISCUSSION
A ir photos o f a scale equal to o r less than 1: 30 000 were found adequate fo r m onitoring and measuring the advancement o f the foredunes provided match ing lines o r points were available.A ir photos o f 1977 and 1979 were discarded, because the small area covered provided insufficient m atching points in the optical fie ld o f the ZTS.(V an Leeuwen, 1966).

Space/time relations
The distance o f dune ridges from the seashore is directly p ro po rtio na l to th e ir age and O n a speculative level, and i f the yearly advancement rate o f 2,4 m is applied to the transect published by M o ll (1972), the positions o f the com m unities on the transect suggest the fo llo w in g succession schedule (Table 1): (1) It agrees w ith M o ll (1972) in that it takes about 10 years fo r a dune ridge to be form ed under present conditions (Fig. 2).(2) I t w ill take about 30 years to become invaded and later replaced by the Passerina rigida Open Dune Scrub (Fig. 4).

^ ' * 1
Ni^tK h fk n (3) I t w ill take about another 30 years before this scrub is replaced by Closed Dune Scrub (Fig. 4).
(4) I f protection against seawinds and saltspray is given by the seaward ridges and th e ir vegeta tio n , a Dune Forest could develop after about 90 years, beginning in the dune slacks and later spreading from there.
The re lia b ility o f the space/time relation is reduced by the variation in dune advancement rates on d ifferent coastal stretches and at d iffe re n t tim e intervals.In some places, e.g. in the northern part o f the M lalazi Reserve, there has lately been a tendency fo r the height o f the existing dunes to increase instead o f new ridge form ation and dune advancement seawards taking place.The key factor to the d iffe rin g deposition o f sand, i.e. either in the form a tio n o f new ridges or in increasing the height o f existing ones, is the establishment o f Scaevola thunbergii seedlings on the beach and the accumulation o f sand around these obstacles.I f beach conditions do not allow establishment o f Scaevola thunbergii, the landwardblown sand w ill tend to get trapped in the existing foredunes and contribute to th eir height increase.
H igh and low ridges in M o ll's p ro file suggest that they have held th eir position adjacent to the beach fo r varying periods o f tim e before being replaced and outcompeted fo r sand by new ridges.H igh dune ridges suggest longer periods near the seafront, during which sand had tim e to accumulate.The occasional destruction o f ridges by unusual storms and tides must also be considered.
A recent increase in the rate o f sand deposition is lik e ly , because o f the increased sediment load o f the Tugela R iver follow ing agricultural malpractices in the catchment areas.Part o f the sand is transported northwards by longshore d rift (O rm e in Begg, 1978).How ever, the L A N D S A T satellite image No. 1190 -07143 o f 29th January 1973, shows the sediment plume o f the Tugela extending m ainly southwards.
The space/time relationships obtained here must be regarded as a first approxim ation and should be follow ed up by studies w ith permanent plots.C. J. W ard (pers.comm.) and M acD onald and Pammenter (M S) have laid out perm anent vegetation plots and transects in the Twinstreams area and M lalazi Nature Reserve which w ill allow future refinem ent o f this interim dune vegetation-succession chronology.

DEFINITION OF TERMS
Biomass is defined as oven dry mass o f live, actively or structurally functional organic m aterial and does not include the dead wood category.The categories which were determined included: -T otal biomass, the to ta l livin g o r functional mass.
-Biomass o f the stem, m ainly wood.
-Biomass o f branches, m ainly wood.W here stem and branch biomass were not separated, this is referred to as wood biomass.
-Biomass o f current season's twigs.
-Biomass o f leaves.C urrent season's tw ig mass together w ith leaf mass, that is the to ta l mass o f current term inal grow th, is referred to as shoot biomass.
-Mass o f dead wood (branches and twigs) still attached to the plant individual.
A ll shrubs and trees w ith stem diameters equal to or greater than one centim etre at 20 cm above ground level were included in the study.Individuals w ith stems less than one centim etre in diam eter were clipped and included in an independent study o f Shrubs were defined as individuals w ith height less o r equal to 2,5 m and trees were ta lle r than 2,5 m.The use o f 2,5 m height as a basis fo r separation o f shruband tree-sized individuals is arb itrary but 2,5 m is a height above which v irtu a lly no browsing by larger herbivores present can take place.I t is also a height below which shoot grow th points are most com m only damaged after fire on the Nylsvley site.

Biomass estimation
The method used involved a destructive phase in which a num ber o f individuals o f each o f the more im portant woody plant species were measured fo r various dimensions (see A ppendix 1), then felled and the above defined parts o f the plant weighed oven dry.A ll plant m aterial was dried to constant mass at 85°C.A relational stage follow ed where dimensions were appropriately related to the various mass categories o f the plant resulting in a predictive relation fo r each species (function types in A ppendix 1).A th ird phase involved a large-scale fie ld survey in which all individuals were measured fo r the predictor variables in plots o f know n size, whereupon the predictive relations were applied to give the mass o f the various categories per u n it ground area.
Because the num ber o f d iffe re nt species involved was too large fo r equally intensive treatm ent o f each, species were divided in to convenient groups, based on plant abundance data available fro m early surveys.
The firs t group contained the fo u r species, Burkea africana, Ochna pulchra, Terminalia sericea and Strychnos pungens in each o f which a fu ll size range o f up to 49 individuals were processed in d e ta il, providing from prim ary data, the predictive equations listed in A ppendix 1. Le af area data were obtained from leaf mass data by determ ination o f Specific Leaf A rea (cm 2/g) in each species.Stem wood was calculated by subtraction o f all other biomass categories fro m to ta l biomass.
The second group contained the next three, much less abundant species, Vitex rehmannii, Combretum zeyheri and Dombeya rotundifolia.The m ethod was as in the first gi oup except that the size o f the fie ld sample was much reduced, as little as three very carefully selected representative individuals being analysed in fu ll.These restricted data were then plotted out together w ith relations fo r the species o f the firs t group and used to determ ine what constants ( if any) should be applied to the equations fo r the species o f the first group to form new predictive equations.This method resulted in quantitative approxim ations o f the biomass categories.
A th ird group comprised several distinctly rarer species.Strychnos cocculoides and Combretwn molle were subjectively matched according to a ffin ity at the generic level.Two Securidaca longipedunculata individuals were sampled in the fie ld since m atching here was less obvious.The two Combretum species (C.zeyheri and C. molle), from inspection o f several individuals, appeared sim ilar fo r each biomass category.Strychnos cocculoides appeared sim ilar to S. pungens except in respect o f the relations fo r total biomass and mass o f dead wood.
A fo u rth group included all other rarer species fo r which no fie ld mass data existed and there was no clear basis fo r matching w ith any particular other species.
Here the combined relations o f the group w ith the most reliable data (the firs t) were used.The above relations are valid fo r the peak o f grow th season (based on com pletion o f term inal grow th).A lm ost all field w ork was done in the 1975/76 season.
A fifth 'group' contained only the m ulti-stem m ed shrub Grewia flavescens.The method described above fo r individual stems was found im practical to apply owing to the p roh ibitively large numbers o f such stems in the enumeration-type survey.The alternative use o f a 'whole in divid ua l' predictor fo r dimensions such as that o f canopy diam eter were found unsuitable owing to large variation in density and spacing o f individual stems.A n o th e r test showed large scale harvest o f all G. flavescens individuals over large areas, w ithout recourse to predictive relations, was unacceptable due to several sampling problems.A fu rth e r approach, which was fin a lly accepted, made use o f G. flavescens individuals stratified upon fo ur modes o f grow th and die-back.Mean mass ratios were then applied to each o f these diffe re n t forms fo r shrub-sized and tree-sized individuals.

RESULTS
Biomass o f different plant parts fo r tree-and shrub-sized individuals o f each species is given in Table 1.Biomass proportions are summarized in Table 2.A n example o f a typical breakdown o f biomass categories in a tree species is provided by Burkea africana (Table 3).For those species in which stem wood biomass and branch wood biomass were measured separately, that is in B. africana, Ochna pulchra and Terminalia sericea which make up more than three-quarters o f the to ta l biomass o f all species, v irtu a lly twice as much stem wood as branch wood biomass was found, varying from 2,14 times more in Burkea africana to 1,37 times in Ochna pulchra.
Four species (Fig. 2) accounted fo r 81,9% o f shoot mass o r term inal production.A m ajor co ntribution o f 93,8% to dead wood mass was made by the same fo u r species.A lth ou gh Burkea africana comprised more than h a lf the tota l biomass, its productivity (term inal) was only about one-third o f the total.Conversely, the percentage con trib utio n o f Ochna pulchra to tota l term inal production is about twice that o f its percentage con trib utio n to total biomass while Grewia flavescens p ro d u ctivity rank position o f 4 drops to 10 relative to to ta l biomass.
Dead wood was also the only mass category where only three species accounted fo r more than 90% o f the tota l amount.The mass o f dead w ood as a percentage o f tota l mass was 79,3% fo r G. flavescens, 10,5% fo r Terminalia sericea and 5,6% fo r Burkea africana.The other separately considered species varied between 1,6 and 4,9% .Three o f the fo u r species w ith the greatest term inal grow th capacity also had the highest percentage o f dead wood mass.L eaf mass and leaf area differences between species follow ed the pattern o f differences in shoot mass (Fig. 2), w ith the clear exception o f Grewia flavescens.
Biomass and leaf area variatio n w ith in the study site is summarized in Table 4.There were relatively large differences in woody species leaf area index ( L A I) from one transect to another (Fig. 3) where, fo r example, L A I in transect A was 177% o f that in transect C.

DISCUSSION
The mean woody species basal (at 20 cm above ground) area (excepting Grewia flavescens) is 6,26 m ; ha'1 varying from 7,40 m : h a 1 in transect A to 4,52 n r ha'1 in transect C.This basal area is low er than the 8 n r ha'1 quoted fo r a long protected savanna woodland w ith Burkea africana dom inant in north-eastern South West A fric a /N a m ib ia (R u the rfo rd, 1978) and the 8,5 m 2 h a 1, in a B. africana dominated com m unity about 7 km from the Nylsvley study area, that had been protected from fire fo r several decades (R uth erford & K e lly , 1978).The Nylsvley study area has a low er woody species basal area than sim ilar comm unities elsewhere possibly owing to the more frequent occurrence o f fire on the Nylsvley site.This low er basal area is also reflected in a low er to ta l biomass.
The mean total biomass fo r the Nylsvley site (16 273 kg h a 1) is considerably less than the 22 300  -------------------------------------------------------------------------------- In a few cases the independently estimated total biomass does not precisely equal the sum of the constituent biomasses.This is due to one or both of the following reasons depending on species and area: 1, All computer calculations from the application of the allometric formulae onwards were carried out retaining several decimal places.This was to reduce the magnitude of round-off error that would otherwise be propagated during calculation.To obtain minimum round-off error per separate mass category, data were converted to integer form only in the final presentation but this sometimes results in imperfectly additive matrices relative to the last significant digit; 2, For the smallest shrub of some species the estimate of its stem wood mass, through subtraction, becomes marginally negative owing to the predictor variables being applied at the extreme limit of regression range.Such estimates were automatically set to zero as the most feasible estimate of stem wood mass in such individuals.Only where such shrubs occurred in exceptionally large numbers did this setting to zero slightly affect the equality between total biomass and the sum of the constituent biomasses.
Two typographical errors which appeared in the prediction equations of the original report have been corrected.The biomass results remain unchanged.

Fig. 3
Fig.3summarizes the distance measurements as interpreted from the photos.From 1937 to 1957 (20 years), the Scaevola thunbergii Foredunes advanced vice versa.The foredune advancement rate o f the last 40 years was 2,4 m per year n orth o f the Siaya M o uth.Therefore, it is possible to date the ridges and th e ir vegetation by measuring the distance from the beach.This possibility o f dating vegetation led us to extend our studies south to the farm Twinstreams.Whereas processes and dune advancement values sim ilar to those north o f the Siaya M o uth were encountered, the lack o f m atching points in the lim ite d optical fie ld o f the ZTS made precise distance measurements d iffic u lt.
advancement at the Mlalazi Nature Reserve.Total distance of the dune advance 1937-1977 was 95 m and the average rate 2,4 m per year.INTRODUCTION The South A fric a n Savanna Ecosystem P roject is being conducted on a portion o f the N ylsvley N ature Reserve (3 120 ha in extent), 10 km south o f Naboom spruit in the northern Transvaal.The basic ecological characteristics o f the study area are described in H untley & M o rris (1978), w hile the projects overall objectives and research programme are outlined in H untle y (1978).The study area lies on the edge o f the Springbok flats on a slightly raised plateau at about 1 100 m above sea level.M ost o f the W aterberg System sandstone bedrock is covered by sandy soils belonging m ainly to the H u tto n and C lovelly forms (Harm se, 1977).Mean annual ra in fa ll is about 630 mm and occurs m ainly in summer.The mean annual air tem perature is 18,6°C.The study site's past management has included light summer grazing by cattle w ith small populations o f im pala and fluctuating populations o f kudu present.Fire has occurred irregularly at approxim ately five year intervals though there is evidence o f more frequent fire in the south-western part (belt transects D , E ) o f the study area.The main vegetation type o f the study area has been classified as Eragrostis pallens -Burkea africana Tree Savanna (Coetzee etal.f1976) w ith the most extensive variation o f this being the Eragrostis pallens -Dombeya rotundifolia variation w ith dom inant trees Burkea africana and Terminalia sericea and dom inant shrubs Ochna pulchra and Grewia flavescens (Fig. 1).H untley (1977) suggests that the broad-leaved savanna o f the study area is related to the mesic and moist broad-leaved savanna biome o f A fric a .Scattered at several localities w ith in the study area are small abandoned native settlement areas which now support a flora very different to that o f the rem ainder o f the study area.* This is a publication of the South African Savanna Ecosystem Project and is an abridged form of a report (No. 36) of the South African National Scientific Programmes.** Botanical Research Institute, Department of Agriculture and Fisheries, Private Bag X101, Pretoria, 0001.The first objective o f the Savanna Ecosystem Project has been to determ ine 'the structure and dynamics o f the ecosystem as a w hole' (A n o n , 1975), and Phase I in the project includes 'the description and quantification o f structural features o f the ecosystem' (H u n tle y, 1978).The objectives o f the study described in this paper fe ll w ith in this firs t phase o f the overall ecosystem project, and were to determ ine the biomass o f the main aboveground categories o f the woody species present and the variation fro m one part o f the study area to another.This paper is largely lim ite d to presentation o f data and detailed discussion o f methods is available elsewhere (R u th e rfo rd , 1979).
Fig. 1.-View of the vegetation in a part of Transect A in which aboveground woody plant dry mass (functional and dead) was estimated at 21 602 kg ha1.
have described how five belt transects were selected to best represent the woody vegetation o f the ecosystem study site.These areas represent the three variations and the subvariations o f Eragrostis pallens -Burkea africana Savanna o f Coetzee et al. (1976).These selected areas do not include Acacia patches o f old abandoned settle ments , sandstone hills o r occasional rocky outcrops, or fire-break areas.These five areas were used by Lubke et al. (1976) fo r an intensive survey o f woody species structure including detection o f pattern o f d istrib u tion.The five areas are designated A in the north-east o f the study area through to E in the south-west o f the study area.In the present study, all or part o f each o f these belt transects was used as basis fo r estim ation o f biomass o f woody species, respective sample sizes being 0,875 ha in transect A , 1,6 ha in transect B , 0,8 ha in transect C , 0,96 ha in transect D and 0,96 ha in transect E. The to tal area sampled fo r application o f biomass relations was therefore 5,195 ha.B elt transects D and E have been set aside fo r destructive sampling w ith areas A , B and C being protected.O ur own measurements o f dimension were used fo r trees but data o f Lubke et al. (1976) (using smaller parts o f each sample strip) were used fo r shrub-sized individuals (except fo r Grewia flavescens).
Shrubs had a far greater p ro p o rtio n o f term inal growth than trees.It is particula rly in current tw ig biomass o f shrubs that exclusion o f Grewia flavescens greatly reduced the relative c o n trib u tio n o f this category.The other biomass categories are little affected by inclusion or exclusion o f G. flavescens.H ow ever, in the dead wood mass category, it is im portant to differentiate between relative con trib u tions w ith o r w ith ou t G. flavescens.It is p articularly the shrub-sized individuals where the exclusion o f G. flavescens causes a very large decrease in the relative contribu tion o f dead wood mass.That G. flavescens differs from other species in respect o f relative amounts o f tw ig biomass and dead wood mass may not be merely fo rtu ito u s.These tw o aspects are lik e ly to be linked since the large scale d ie -o ff o f older parts allows fo r new self-supported shoot grow th only from ground level.There are indications that proportions o f dead wood mass in G. flavescens can vary greatly w ith season.
136 ABOVEGROUND BIOMASS CATEGORIES OF WOODY PLANTS INA BURKEA AFRICANA -OCHNA PULCHRA SAVANNA Nylsvley site (12% ) is v irtu a lly identical to the mean p ro po rtio n (11% ) o f shrubs given fo r the Z im ba b wean C. mopane sites.The leaf production (1 100 kg h a 1) o r shoot production (1 336 kg h a 1) o f the Nylsvley site agrees w ell w ith data fo r other savanna areas (R u th e rfo rd , 1978), particularly w ith that o f the South West A fric a n site.For th e ir Colophospermum mopane sites, K e lly & W alker (1976) obtained an average shoot production o f 1 506 kg ha'1 season 1 which is 8% o f the mean tota l biomass.This p ro p o rtio n is identical to that found fo r the Nylsvley site where shoot production was also 8% o f total biomass.For the tw o dom inant woody species in the abovemen tioned Combretum apiculatum and C. zeyheri savanna com m unity, D ayton (1978) found shoot production to constitute 9% o f the tota l biomass.On N ylsvlei, the term inal shoot production by the woody species was greater than fo r example, Ochna pulchra; 3, Terminalia sericea; 4, Grewia flavescens; 5, Vitex rehmannii; 6 , Combretum zeyheri; 7, Dombeya rotundifolia; 8, Combretum molle; 9 , Strychnos pungens; 10, Strychnos cocculoides;

TABLE 4 .-Mass and leaf area data for the woody plant component of different areas of the Nylsvley study site
W alker (1976) determ ined woody plant biomass o f nine sites in Colophospermum mopane dom inated communities in a region w ith an annual ra in fa ll o f approxim ately 500 mm in south-eastern Zim babw e. W oody plant biomass ranged fro m 8 726 to 30 782 kg ha'1 and averaged 19 694 kg ha'1.This average value is very sim ilar to the biomass value fo r transect A o f the Nylsvley study site.The relative species on the Nylsvley site(s).This relatively low er degree o f dominance on Nylsvley is also apparent when compared in terms o f basal area to the long-term fire protected Burkea africana com m unity seven kilom etres from N ylsvley (R u th e rfo rd & K e lly, 1978).The relative c o n trib utio n o f shrub biomass to total woody plant biomass fo r the