Observations on T hemeda australis- Savannah 10 Papua

P. C. H EYLIGERSI

SAVANNAHS, vegetation types with a ground Fires are common throughout the dry season, cover dominated by grasses and an open tree at the end of which most of the ground cover storey, are extensive in the coastal lowlands of has been burnt. Regrowth starts after some the Central District, Territory of Papua (Fig. rain has fallen. Areas burnt early in the dry 1) . These lowlands were included in a regional season can have a fair cover at the end of it survey carried out in 1962 by a team of the and are liable to burn off again. At the time of Division of Land Research and Regional Sur­ our investigation, notwithstanding high rainfall vey, CSIRO. in June, the dry season was already well ad­ This paper describes the vegetation of savan­ vanced and fire had destroyed the ground nah near jackson's Airport, about 12 km east vegetation of a part of the area under in­ of . Emphasis is laid on correla­ vestigation. tion with edaphic conditions. The field work, Wild life in the area is very scarce and its in cooperation with the geomorphologist and influence on the vegetation is negligible. Graz­ the pedologist of the team, was done in July ing by cattle is restricted to a few fenced prop­ and August 1963. ·erties. Th e following data pertinent to this area are Two transects were selected for observations: extracted from the survey report (Mabbutt et transect 1 was located 3 km south of jackson's Airport near the Riga Road ; transect 2 about aI., 1965). 1.5 km northeast of the airport. Levels were The climate is monsoonal: strong south­ taken along each transect and at selected situa­ easterly winds prevail during the dry season tions pits were dug, varying in depth between from May till October, in which monthly rain­ 1.0 and 2.5 m. fall averages 30 mm, and light variable winds during the rest of the year with monthly rain­ falls averaging 160 mm. Table 1 gives more TOPOGRAPHIC AND SOILS DESCRI PTION OF detailed information about rainfall, together THE TRANSECTS with data on temperature and evaporation. Each transect comprised a fairly straight, The landscape, made up of strike ridges and smooth hill slope attaining 22° to 30 ° and with vales, is underlain by fairly steeply dipping minor rock outcrop, passing into a shallowly rocks of Tertiary age. Cherty shale, marl, and dissected foot slope mainly between 0° 30' and limestone form the rounded ridges; the vales 5° and up to 1200 m long, and ended at a small have been cut in less resistant tuff. Relief is in strike stream, which on transect 1 has a narrow the order of 100 m. Lithosols and regosols are bordering flood-plain. found on the ridges, brown clay soils and tex­ Much runoff is as sheet flow: slope wash on ture-contrast soils on the higher parts of the hill slopes results in small terracettes, and on flatter land, and dark clay soils and alluvial foot slopes leaves a fairly abundant lag gravel soils near and along drainage lines. The vege­ (Mabbutt and Scott, 1966) . tation is savannah, predominantly of the Each transect had a similar sequence of soils Th emeda amtralis-EttcalyptllS type, with tall (Scott, unpub lished data) . Th e hills are occu­ grass vegetation and forest along permanent pied by red regosols, which are deeply devel­ streams. oped at the lower slope of site 2. On the adjacent foot slopes texture-contrast soils are 1 D ivision of Land Research and Regional Survey, CSIRO, Canberra, Australia. Manus cript received found, whilst black clay soils occupy the September 21, 1965. remainder of the flatter country, forming a 477 478 PACIFIC SCIENCE, Vol. XX, October 1966

FIG. 1. View from the crest of a str ike ridge over the savannah country just east of, but similar to, that in which the transects were laid out. The foothill in the foreground rises fairly abruptly from the foot slopes, which end against a forested narro w flood-plain. In the farther distance is anoth er strik e ridge shadowed by clouds and backed by Bootless Inlet.

belt of 50 m width parallel with the creek at E. papltana, and some scattered A lbizia procera; transect 1, but extending over 300 m from the and a grass cover dominated by T bemeda gully at transect 2. The soil of the flood-plain australis, with Het eropogon contortus and is an organic black clay. Sebima neruosum commonly associated, the latter especially on stonier soils. A shrub layer is only locally developed; A lbizia procera and T H E VEGETATION Cycas media are the more important species. The T bem eda amtralis-Eltcalyptlts savannah This type of savannah occurs on both transects, has a tree stratum of one or more of the except for a fringe of forest and tall grass on Eucalyptlts species E. alba, E. confertiflora, and the narrow flood-plain. Tbemeda australis-Eucalyprsa Savannah In PapUa-HEYLIGERS 479

...:l Th e Trees -< '" 0 00 ::> ci r...: c-nO\ N""lN N ('(') DISTRIBUTION PATTERNS (Fi g. 2) : Ellcalyp­ z .-< z .-< tus alba occurred over the whole transect 1; its -< height ranged from 9 to 16 rn, On the black clay and adjacent textur e-contrast soils it was o 00 accompanied or predominated by E. conjerti­ .-< flora, of about the same height. Relative pro­ portions of E. alba and E. confertiflora varied between 6:4 and 2:8. On the rest of the texture­ contrast soils and the regosols E. alba was joined by E. papuana, varying in height be­ o .-< tween 10 and 20 m and mostly remaining sub­ N ~ ordinate. Relative proportions varied between .g" 9:1 and 5:5 . Somewhere In the central part .~'" of the texture-contrast soils, over a distance of N N NN ..... \O 3) . On the transition to the organic black clay -< .-< soils some specimens of E. papllana also oc­ curred. E. papttana occurred over the whole transect 2, with heights rangin g from 10 to 18 m on .-< .-< N the flat country and from 17 to 22 m on the o. \O r<\V"\ r<)V"\ ec NN N ...... slopes and along the gully. E. alba, 9-13 m '" .-< high, was concentrated on the texture-contrast 0\ 0 0\ soils, but some trees occurred along the gully o\ r---:~ OO \C":j" N N N('(')c<\t<""l and one single tree was growing on the slope. .-< On the texture -contrast soils the relative pro­ 00V\ 0 portions of E. alba and E. papllana varied be­ O"": ~NV"\O\ ('('"\ N NN r- ..... tween 8 :2 and 5:5. Not a single specimen of N .-< E. confertiflora was seen. ff") ~«: The density is in the order of 150 trees per ...... oo":!'-.::r r- O ('(') N N"'t' 0\""1'" hectare. A more exact determination of density .-< N .-< was abandoned because of signs of wartime dis­ turbance such as local cutting of trees, trees that were pushed over, and patches of rather dense eucalypt regrowth. Moreover, it is un­

TRANSECT 1

SCALE OF ~.:.:~ . TOPOGRAPHY TRANSECT 2 .' 25m I__-< .0 . 00" r-v-v-v-v- 100m o••••••••• 00• 0 .:.:.:.:.::.:.r::::::::::::::::::::::::::/:::::·:··:·:··:::·; ':':" :':':'~'::' ..

Eucalyptus Ground cover Soi 1 t ype

0.... .f. alba //// I . ..Joemeda-neteropoqon ~~ ....81ac kca I y SOl'1 group • . .Epapuana \\\ \\\ \\\ ....llJemedo-Copi//ipfdium ~.. .. :.Text ure - contrast soil group ~ . .f. confertijlora III \1, 1// . • ... Ihemeaa-sehima ::::::::: Red regosol group -<.///. Sehima - Heleropoqon series

FIG. 2. Diagram of the transects, showing the distribution and mutual proportions of the Eucalyptus species, and the distribution of the ground cover groups and of the soil types. trunk. The horizontal roots of E. papuana are flo ra, where they gradually emerge from the often oval in section with the long axis vertical, base. In the case of E. alba the trunk is usually and divide at or near the base in several hori­ a little swollen, but sometimes is considerably zontal roots above each other (Fig. 4) . so. With a trunk diameter of 24 em, the diam­ Horizontal roots run for distances up to eter of the base may vary between 30 and 45 about 25 m more or less parallel with the em. Just below the surface the base is con­ surface, often not deeper than 0.2 m, only stricted, under which constriction the main very gradually tapering. Near the base vertical roots branch off. roots branch off from horizontal roots, and are The taproot, if present, may reach a depth particularly common in some specimens of E. of 1.3 m but is usually shorter. It branches off alba. In some cases these roots run down verti­ under acute-angle, thick roots, which show the cally more than 1.5 m, but more often they same features as described above for the ver­ gradually bend away obliquely, and may even tical roots. One specimen of E. alba was found bend up and penetrate the topsoil again, where which lacked the cluster of horizontal roots they continue horizontally. but which had a very stout taproot. The way in which the main roots of E. alba Tree roots with diameters of 1 em or less, branch off from the base of the trunk is dif­ whose origin could not be traced, were found ferent from that of E. papuana and E. conjerti- scattered at depths of 1.3 m, and rootlets with Tbemeda australis-Eucalyptus Savannah in PapUa- HEYLIGERS 48 1

diameters commonly less than 1 mm were still over the whole length of both transects. It was, found at a depth of 2 m (bottom of the deepest however, more common on the flatter parts. pits). Rootlets of 1 or 2 mm diameter were Mostly varying in height between 1.5 and 3.5 seen branching off from the main roots; they m, some reached a height of 8 m and thus have a slightly thickened base. penetrated into the tree layer. The shrubs grow The conclusion is reached that the Eucalyptus in clusters and consist mostly of little-branched species studied have root systems with major shoots, which were generally shedding leaves roots strongly developed in horizontal direc­ at the time of the investigation. Examination tions, rather close to the ground level, and of the surface soil and removal of the superfi­ smaller roots penetrating to depths of 2 m and cial soil layers revealed that the shoots are over. linked by a network of creeping woody root­ stocks, often as thick as the base of the shoots Th e Shrubs (Fig. 5). From the rootstock branch off vertical Except for young specimens of the eucalypts, and oblique roots, which branching does not was the only shrub encountered seem to have any correlation with the places

FIG. 3. The overlapping distribution area on transect 1 of Eucalyptus conjertiflor« (with flaky bark) and E. papua11a (tree with light trunk and slender foliage at right) . The tree at the left (with damaged bark) is E. alba. In the foregrou nd several branches of Albizia procera shrubs ; in the background an area of eucalypt regrow th. The ground cover is formed by the Themeda-Capillipedium group . 482 PACIFIC SCIENCE, Vol. XX, October 1966

FIG. 4. Root system of Eucalyptus papuana in black clay soil of transect 2. (The camera looks obliquely down into the pit .) Under the basal whorl of main horizontal roots a small taproot can be seen, which divides into upward-bending roots. where the shoots originate. These vertical roots Tbemeda australis, Heteropogon contortus, and taper rapidly and the deepest penetrations ob­ Sebima neruosum are the commonest; Capil­ served were about 0.9 m. From them branch lipedium spicigerum and Sorgbam nitidum are off horizontal and downward rootlets of diam­ subordinate; and Cymbop ogon procel'tls, Dicbnn­ eters varying between 1 and 2.5 mm. tbium superciliatum, Tbemeda nouoguineen­ Cycas media has its main distribution on the sis, and Panictim sp. are occasionally found. The slopes; some specimens, however, were found season was not very favourable for the study of near gullies and the gallery forest margin. It other herbs. Some dried sedges were found, of was completely absent on the intermediate, which only Eleocharis monostacbyos on black flatter parts of the transects. Usually Cycas clay soils was recognizable. Among the forbs the remains low, 0.5 or 1 m high, but occasionally Papilionaceae were best represented : Lourea it reaches heights of 4 m. obcordata, Crotalaria linifolia, Alysicarpus For the sake of completeness the occurrence vaginalis, Uraria lagopodioides, Pycnospora is mentioned of a few specimens of Pandanus lutescens, T ephrosia Pmaculata, and four species sp., ghaesembilla, T imonius timon, not yet identified, of which two were twiners. and Desmodium umbellatum, growing on black Among the others were: Passifiora fo etida clay soil along the gullies and creeks. (Passifloraceae), Eooloulus alsinoides (Con­ volvulaceae), Melothria maderaspatana (Cu­ Th e Ground Layer curbitaceae), Tridax procumbens (Composi­ GENERAL COMPOSITION : The ground layer tae) ?Borreria sp. (Rubiaceae) , and several is composed mainly of grasses, of which Labiatae. Because of the general poor condition T bemeda atlStralis- Eucalyptus Savannah In PapUa-HEYLIGERS 483

FIG. 5. Root system of A lbizia procera consistmg of rootstocks connecting the trunks and givi ng an­ chorage to the shrubs or trees, and of vertical roots scattered along the roo tstocks. of these at this time of the year, no stability of the slopes (Mabbutt, unpublished attempt has been made to study their distribu­ data) . tion. COMPOSITION OF THE GRASS COVER : The A feature observed locally at transect 1 and investigation was concentrated on the grasses, common on transect 2 was a crust of greenish which generally were fruiting and therefore or reddish brown algae, which could cover as had probably reached their maximum produc­ much as 20% of the ground. These algae were tion ; Sehima had already lost most of its in­ not restricted to the flatter areas, but occurred florescences. The overall height of the grass also on the slopes. Mosses occurred sparsely and cover was about 1.1 m on the flat parts of the were commonest on the lower slopes. transects, decreasing to 0.9 m on the slopes. INFLUENCE ON TERRACETTES: Grass tussocks The culms were mostly fairly erect ; locally, may playa role in the formation of terracettes, however, they could be more or less inclined. which are especially common on steeper slopes Foliage cover averaged about 50% , but ex­ but occasionally can be seen in flatter areas. A treme local variations occurred ; almost bare number of tussocks form an obstacle to the areas could lie next to areas with nearly 100% runoff flow and soil material accumulates at the cover. Th ese changes are related to micro­ upslope side. Differences in level at the two topography-depressions have a more open sides are mostly a few centime ters, but may vegetation-and cover of the soil by grit and become more than a decimeter ; factors leading gravel, whilst severe disturbance dur ing the to enlargement of this difference includ e slope war could have been a factor in some of the angle, age of tussock, and texture of soil cases. Trees and shrubs did not seem to have material. Tu ssocks contribute in this way to a noticeable influence on the grass cover. 484 PACIFIC SCIENCE, Vo!. XX, October 1966

The grass cover was sampled near each of the length of some of the longest culms ; and the soil pits and on the steep side of the hill at the area covered by leaf litter, algae, grit and transect 2. A series of 10 plots was cut along gravel, and stones. Moreover, a series of 10 a line at right angles to the direction of the plots gave frequency figures for the species. main transect or following the contour, the The production per series, expressed as the plots spaced three paces apart. In a few cases average dry weight per sample plot of 0.2 m2, burnt areas made a deviation of the scheme appeared to be very variable and the 95% con­ necessary. A plot was delineated by a frame of fidence limits to the means showed that only 28.3 X 70.7 em, comprising an area of 0.2 m2 the series with lowest pro duction significantly (Figs. 6 and 7). Th e grasses were cut at about differed from the four with highest production. 1 em above the ground, and the species were Also, the figures for Tbemeda australis, the kept separately. Some plots seemed to contain only species with consistently high frequency also some material from earlier seasons. The and showing a tendency to increasing produc­ samples were dried in sun and wind for several tion downslope, failed to show significant dif­ days, keeping them indoors at night, after which ferences. they were weighed. At this stage they could Because of these nonsignificant differences, still have held up to 10% moisture, but this series have been combined in groups , according drying at least made the samples mutua lly to the relative impor tance of the species in the comparable. In addition, in each plot records composition of the sward (percentage of species were made of the total cover of the tussock in total dry weight of a series). Three series bases as left after cutting; the number and the remained separate because of their strongly diameter of the base of the tussocks per species ; deviating composition; the others were in-

FIG. 6. A plot on the lower hill slope of site 1, con taining three and a half tussocks of Sebima neroosum (i n centre and at left) and one of Themeda australis (at right) . T hemeda australis-Bucalyptus Savann ah In PapUa-HEYLIGERS 485

FIG. 7. The same plot after cutting shows the tusso ck bases, some leaf litter, and scattered gravel. The frame adapted from D aubenmire (1959) comprises an area of 0.2 m2 and is painted to divide by sight the area into squares of 10%, 5%, and 2¥2% to facilitate estimation of cover. corpora ted in three groups of 6, 6, and 7 series poorer quality of the tussocks, which is also respectively, called the T hemeda-H eteropogOIl reflected in the average length of the largest group, the T bemedn-Capillip edium group, and culms. the Themeda-Sehima group. The thr ee remain­ The T hem eda series consists of a pure but ing series are the T hemeda series, the T bemeda­ rather poor stand of Tbemeda australis . The Sorgbnm series, and the Sebima-Heteropogon T bemeda-Sorebnm series contains about equal series. Of the three groups the produ ction of the T bemeda-Heteropogon group and of the TABLE 2 T hemeda-Capillipedimn group are not signifi­ TOTAL G RASS P RODUCTI ON OF THE T hemeda cantly different, but the T bem eda-Sebima group australis-Eucalyptus SA VANNAH" produces significantly less than the two other groups (Table 2). AVERA GE GRASS PRO D. Other data on the groups are given in Table GROUP AN D SERIES (GM) 3. There appears to be a good correlation be­ T bemeda-Heteropogon group 126 ± 17 tween the composition of the sward by weight T bemeda-Capillipedium group 115 ± 16 and the basal cover of the tussocks. Also, total Themeda-Sehi ma group 81 ± 10 production and total basal cover show the same Tbemeda series 76 ± 32 correlation. An inverse correlation exists be­ T bemeda-Sorgbum series 119 ± 60 tween basal cover of the tussocks and the Sebima-Heteropogon series 83 ± 18 numb er of tussocks: with increasing number of • Figures give the average production of grass in grams of air-dry matter per plot of 0.2 m2 , with th e 95% confidence tussocks the basal cover decreases, due to the limits to the means. 486 PACIFIC SCIENCE, Vol. XX, October 1966 quantities of Sehima neruosttm, Sorghum nttt­ the latter with a crumbly topsoil of only 5 or 10 dzan, and H eteropogon contortus, and double em. Deeper layers have a blocky, lenticular, or that amount of T hemeda australis, with a little columnar structure, are black, dark grey, or dark Dicbantbium snperciliatnm. Th e third series, brown, and frequentl y have calcium carbonate the Sehima-Heteropogon series, with a produc­ concretions. The pH in upper layers varies be­ tion almost the same as the T bemedn-Sebima tween 5.5 and 7.0, and increases at depth to group, consists of about equal quantities of 8.5. Some 20% of the ground is covered by H eteropogon contortus and Sehima neruosum , litter, mainly of the eucalypts, and 10% is cov­ with T hemeda australis accounting for only ered by grit and gravel, both showing a great 6% of the dry matter. local variation. DISTRIBUTION PATTER NS (Fig. 2): The The­ The T hemeda-Capillipedium group occupies meda-Heteropogon group occupies the lowest the intermediate parts of the transects. At parts of both transects, with slopes varying be­ transect 1 it occurs on texture-contrast soil al­ tween 0°30' and 3°, and occurs on black clay ternating with the T hemeda-H eteropogon soil as well as on texture-contrast soil, the for­ group, at transect 2 on texture-contrast soil and mer with a 20 or 30 cm deep crumbly topsoil, adjacent regosol ; consequently it occurs over

TABLE 3

S PECI FIC D ATA OF THE G ROUND C OVER G ROU PS (T-H =T bemeda-Heteropogon group; T-C =Themeda-Capil/ipedium gro up ; T-S = T hemeda-Sehima group)

T . aus­ H . con­ C. spici­ Sorghum Sehima OT H ER tralis tortus gerton nitidu m neruosum GRASSES T OT ALS Weigh t (g/ m:!) T -H 517 87 10 13 2 629 T-C 407 76 78 3 12 1 577 T -S 218 50 3 131 4 406 Percentage composi tion of sward by weigh t T-H 82 14 2 2 100 T-C 72 13 13 2 100 T-S 53 12 1 33 100 Basal cover of tussocks ( %) T -H 4.05 0.84 0.10 0.06 0.005 5.08 T-C 3.24 0.73 0.95 0.01 0.2 0.01 5.02 T-S 1.66 0.54 0.02 1.36 0.11 3.67 Percentage areas of tussock bases T-H 81 16 2 reo T -C 63 14 19 4 100 T -S 45 15 37 3 100 Number of tussocks per m:! T-H 14.7 3.4 1.6 0.5 0.2 20.4 T·C 15.6 3.1 4.2 0.1 0.3 0.2 :3.5 T-S 15.3 4.8 0.7 6.9 0.2 27.9 Frequency ( %) T-H 96 43 10 10 2 T -C 93 33 32 3 7 3 T-S 84 43 4 73 3 Average length of longest eulms (em) T-H 143 111 135 102 95 T·C 139 111 135 115 95 T -S 117 96 105 89 Themeda australis-Eucalyptus Savannah in PapUa-HEYLIGERS 487

quite a range of topography, on slopes from distribution, which will be treated in the next 0°30' to 20° (average 8°). The soils have in section. An autecological study of the grass common a rather thin crumbly topsoil, 7- 15 em species rather than investigation of group pat­ deep, which is underlain by massive structured terns might perhaps contri bute more to an sandy loam or sandy clay loam. Deeper layers explanation. Factors also to be taken into ac­ are greyish-brown or yellow-brown clays in the count are, for instance, the frequency and in­ case of texture-contrast soil, or brown, reddish­ tensity of fires, the severity of runoff, the brown, or yellowish-red sandy Ioams to sandy lengths of periods of waterlogging, and ability clays in the case of regosol, often with man­ for germination and regeneration. Observations ganese concretions. The pH of the top layer at Katherine, N .T., Australi a, have revealed a varies between 5.5 and 8.5, of deeper layers marked response of sward composition to short­ between 4.0 and 8.0. About 30% of the ground term variations in rainfall, T bemeda australis is covered by litter, but the area covered by grit becoming dominant in drier years (Norman, and gravel is much the same as in the Tbemeda­ 1963). A similar interaction could be expected H eteropogon group. to occur in the area of investigation . Th e T bemeda-Sebim a group occupies the ROOT SYSTEMS OF GRASSES: The distribution higher parts of the transects; it occurs on rego­ of the grass roots in the sections of the pits was sol on hills with slopes varying between 11° studied by spraying the walls with a fine jet and 24° (average 15°). Only a 5-15 em-deep hose. sandy loam topsoil has a crumbly structure; The root systems of Tbem edn australis, H et­ deeper layers are sandy loams or finer textured eropog01Z contortus, Sehima neruosum, and soils, to clays, with a massive structure, dark Sorgh ml1 nitidu m appeared to be of much the brown, merging into reddish colours. Pieces of same structure. Roots spread in horizontal, ver­ weathering rock are present, sometimes already tical, and intermediate directions from the tus­ at depths of 35 em. Th e pH of the topsoil is sock base, forming a dense "corona" with a 6.5- 7.5, of deeper layers 5.0-7.0. About 38% radius of 1 or 2 dm which becomes more open of the ground is covered by litter and 25 % by at greater distances. grit and gravel, whilst stones and rock frag­ In the black clay soil the crumbly topsoil, ments cover 3.5%. especially the fine crumbly upper part, is densely The Tbemeda series occurs on the texture­ rooted; numbers of roots gradually diminish in contrast soil of transect 1. One of the main the underl ying heavy clay; usually roots are characteristics of its locality is the excessive frequent at 0.9 m and a few are still present cover of grit, which averages 53% . The T he­ at 1.8 m. Roots sometimes seem to follow the meda-Sorgbum series occurs on black clay soil, planes along which the clay cracks on drying; of which the upper horizon appeared to contain in other cases they seem to penetrate at random. red soil material, which disturbance is presum­ In the texture-contrast soil, horizontal roots ably caused by wartime road works. The Se­ spread densely through the superficial layer, bima-H eteropogon series occurs on the steep about 10 em thick, of friable sandy loam or hill slope of transect 2 (28 0), which for al­ loamy sand, with a fair amount of oblique and most half the surface is covered by rock out­ vertical roots penetrating in the underlyin g crops and fragments. massive layer, which is rather resistant to wash­ With regard to an explanation for the rather ing. Roots penetrating in the clayey subsoil at simple group distribution pattern, there is no about 0.3 m depth are usually few, about 4 per direct correlation with soil type or with slope dm-, sometimes more, which may be due to a angle. Differences in cover of grit and of stones, less massive structure of the topsoil. Deepest in the thickness of the crumbly topsoil, in the roots are observed at 1 m. Several grass roots colour of the deeper soil layers, and in the na­ penetrated deeper horizons by following the ture of concretions they contain, however, do vertical roots of the eucalypts, closely pressed not rule out the possibility of a great influence against the bark. of soil moisture regime. On the other hand, this In the red regosol, roots spread in all direc­ influence was not closely expressed by the root tions in the superficial, crumbly layer and in 488 PACIFIC SCIENCE, Vol. XX, October 1966 the upper part of the massive layer of the sandy soils, whilst E. alba showed a considerable (clay) loam, regardless of whether or not the variation within one soil type. surface shows terracettes. The density of the Observations over a period of years, also "corona" around the base of the tussocks is less during the wet season, and on a regional basis than that in texture-contrast soil. Rather many could lead to more conclusive results, but an roots penetrate deeper, into the layer with investigation into the influences of burning will weathered rock fragments. Deepest roots are probably reveal fire to be the overriding factor observed at 1.3 m. of the environment. Consequently, rooting patterns except for being somewhat more diffuse in the regosols are COMPARISON WITH AUSTRALIAN VEGETATION not very different in various kinds of soil, and only a firm, massive, sandy loam layer as en­ Savannahs comparable in physiognomy and countered in certain texture-contrast soil pro­ to a certain extent also in floristic composition files seems to be unfavourable for deeper root occur in the higher rainfall areas of tropical development. and subtropical Australia, but relevant litera­ ture is scarce and is concerned mainly with the Ecological Significance of the R esults characteristics of the ground cover for use as The investigation revealed that correlations natural pasture. between vegetation and soil conditions seem to At Katherine, N.T., Arndt and Norman have only a localised validity. For instance, the ( 1959) studied a savannah vegetation of a distribution pattern of Eucalypt1lS alba encoun­ richer floristic composition but which included tered on transect 2, viz. its restriction to texture­ most of the species encountered at Port contrast soils, is not consistent with that on Moresby. Predominant in the ground cover are transect 1, where E. alba occurs over the whole Sorghum plamosum, T bemeda australis, and transect. Also, a hypothesis based on observa­ Chrysopogon fallax. The dry matter yield for tions at transect 1, that the boundary between unburnt areas of a lightly grazed pasture was the T bemeda-Capillipedium and the T bemedn­ about 110 g/ m2 at the end of the dry season Sehima ground cover groups coincides with the (October) , and rose to about 290 g/ m2 at the boundary between texture-contrast soil and end of the wet (April-May) . For areas burnt regosol, would appear untenable when applied at the end of the dry season the dry matter to transect 2. On the other hand, black soils yield at the end of the wet season was only carry only the Tbemeda-H eteropogon group, about 140 g/m2• which however also extends over other soils. Shaw and Bisset (1955) have given figures Some conditions find ready expression in the for subtropical , which can be as vegetation (e.g., steep, rocky slopes are covered low as 60 g/m2 for dry season yield and as with a Sebima-H eteropogen ground cover, and high as 425 g/ m2 for yields at the end of the disturbed soils with a Themeda-Sorghum wet season. ground cover), but generally the determining The figures obtained for the Tbemeda ass­ factors remain unknown. The presence of an tralls-Eucalyptza savannah in Papua are much algal cover, great local variations in grass cover, higher than those mentioned above. Oven-dry­ and differences in certain soil characteristics ing of the samples instead of air-drying would mentioned in the discussion on the ground have resulted in figures about 10 to 15% lower, cover groups, suggest a great influence of soil which gives a figure of, say, 530 g/ m2, which moisture, whilst terracettes and local concen­ still is much higher than the figures for Kather­ trations of grit and gravel point to slope wash ine and Queensland. Because our observations as an important factor. On the other hand, dif­ are from only one season, it is impossible to say ferences in moisture regime were hardly, if at how near they are to the average. On the other all, expressed by rooting patterns as the root hand calculations by Fitzpatrick (1965) about systems of the grasses as well as those of E. the periods of useful pasture growth demon­ papuana appeared to be similar for different strate that the growing season at Jackson's Themeda australis-Eucalyptus Savannah in PapUa-HEYLIGERS 489

Airport could be nearly twice as long as at on two transects near Port Moresby.Observa­ Katherine, viz. 41 weeks against 22: this would tions have been made on its composition and certainly contribute to the difference in produc­ structure, and on rooting habits of trees and tion between these stations. grasses. Quantitative data for the grass cover have been determined. Patterns in the vegetation coincide only lo­ ACNOWLEDGMENTS cally with some topographic and soil factors, Grateful acknowledgment is made of the in­ and it is suggested that further investigation terest which Mr.R. A. Perry and the other may reveal burning as an agent overriding the ecologists of the Division of Land Research influences of other factors. and Regional Survey have taken in this investi­ gation, and also their criticism on the subse­ REFERENCES quent paper. I am much indebted to Mr. J. A. Mabbutt ARNDT, W ., and M. J. T.NORMAN. 1959. and Mr. R. M. Scott for the discussions on the Characteristics of native pasture on Tippera physical part of the savannah ecosystem; to Mr. clay loam at Katherine, N.T. CSIRO Aust. Div. R.Pullen for his assistance with identification Land Res. Reg. Surv. Tech. Paper No .3. of my collections; and to Dr. P.Grieg-Smith, DAUBENMIRE, R.F. 1959. A canopy-coverage Mr.G. A. McIntyre, and Mr. M. 1. Dudzinski method of vegetational analysis. No rthwest for discussions on statistics. Sci. 33:43-64. Appreciation is expressed to the head of the FITZPATRICK, E. A. 1965. Climate of the Quarantine Station of the Department of Agri­ Port Moresby-Kairuku area. In : CSIRO Aust. culture, Stock and Fisheries, Port Moresby, for Land Res. Ser. No. 14, 83- 97. putting a balance at my disposal and to the MABBUTT, J. A., et aI. 1965. Lands of the Port Director of the Department of Forests, Port Moresby-Kairuku area. CSIRO Aust. Land Moresby, for lending wartime aerial photo­ Res. Ser. No. 14. graphs. MABBUTT, J. A., and R. M. SCOTT. 1966. Last, but not least, I would like to thank Periodicity of morphogenesis and soil for­ Habau and Mou for their assistance in the field. mation in a savannah landscape near Port Moresby, Papua.Zeitschrift fiir Geomor­ SUMMARY phologie 10:69- 89. NORMAN, M. J. T. 1963. The short term ef­ T bemeda australis-Eucalyptus savannah cov­ fects of time and frequency of burning on ers extensive areas in the monsoonal coastal native pastures at Katherine, N .T. Aust. J. lowlands of Papua. The ground layer is domi­ Exp. Agr. Animal Husb. 3:26-29. nated by grasses, mainly T bemeda australis, SHAW, N . H., and W. J. BISSET. 1955. Charac­ H eteropogon contortus, and Sehima neruosum; teristics of bunch spear grass ( H eteropogon the tree layer is formed by three species of contortus (1.) Beauv.) pasture grazed by E1Icalyptus. cattle in subtropical Queensland. Aust. J. This type of savannah has been investigated Agr. Res. 6:539-552.