Australian Weeds Vol. 2(3) Autumn 1983 107

a serious weed eventually, plant ad­ ministrators could make decisions on control measures. containment pol­ REVIEWS (continued) icies . subsidy schemes and research priorities with a clear picture of the area infested. In 1970. information on the area and distribution of E. curvula in was limited, the only accurate assessments being for a small area near Tenterfield (Auld and Scarsbrick. 1970) and for points of introduction and chance spread throughout the State (Leigh and Davidson, 1968). Area, distribution and weed potential of Eragrostis Thus in 1981 a statewide survey was curvula (Schrad.) Nees in New South Wales initiated to assess more thoroughly the area and distribution of E. curvala.

M. H. Campbell Department of Agriculture. Orange. New South Wales 2800 Methods A mail questionnaire survey was carried out in 1981-82 seeking infor­ mation from weed control bodies in Summary chromosome number and inflorescence 123 shires in New South Wales. Re­ characteristics. Clear definition of all peated requests were made until an­ A questionnaire suney of the area and genotypes will not be possible until the swers were obtained from all shires. distribution of African loyegrass (Erag­ taxonomy of the genus has been The 'shires ' included the A.C.T. and rostis carvala (Schrad.) Nees) in 123 revised (Leigh and Davidson, 1968). II 'city' shires, for example City of shires in New South Wales reyealed that Greater Lithgow. Weed officers an­ the plant occupied 45 000 ha of land Eragrostis curvula has been reported swered II questions and provided a and occurred in 56 shires and on 644 to be one of the highest producing map of the shire showing infestations. properties. Fhe shires had more th.n grasses in the summer rainfall tem­ Information for some shires was ob­ 1000 ha, and 54 properties were heavily perate and cool subtropical regions of tained from agronomists, weeds field infested. The plant was present on road­ southern Africa and of North and officers and soil conservation officers. sides, railway land and river banks in 53, South America (Leigh and Davidson, 21 and 15 shires respectiYely; from these 1968). It has also persisted over a wide The survey method was chosen situations it spread to adjacent rural range of conditions in Australia and because it has proved to be successful land used for agricultural production. shown promise as a pasture grass for other weeds in New South Wales Eragroslis carvala (consisting of a (Leigh and Davidson, 1968; Leigh and (Campbell, 1977) and was faster, number of naturalized strains) was con­ Mulham, 1964; Squires and Myers, cheaper and more practical than some sidered to be an unpalatable weed on the 1970). other methods, for example those of south coast and northern, central and Because of its drought tolerance, Smith (1975) and Cuthbertson (19'78). southern tablelands. It has the potential warm season performance and adapta­ In the survey, the only types of for further spread within favourable e.­ bility to a range of soil erosion situ­ E. curvula that could be identified with vironments and there is a strong ations, E. curvula has shown potential any degree of certainty were the cur­ argument for its control in these areas in for soil conservation. Johnston and vula-type and the short chloromelas­ New South Wales. Aveyard (1977) screened 70 acces­ type. However, no distinction was sions, selecting 16 on their ability to made between them in the survey in protect the soil during summer when case other types were also presen t. In cool season species are dormant. 1981 , identification problems were IntroductIon Factors which indicate that E. cur­ overcome by lecturing to weed officers Leigh and Davidson (1968) described vula may be a potential weed include: and showing live plant specimens, and African lovegrass (Eragrostis curvula low palatability, low crude protein by sending them a description and a (Schrad.) Nees) as a perplexing pasture content (Voigt et al., 1970), suscepti­ seedhead encased in plastic. Many species because it is regarded as a valu­ bility to frost damage and ability to weed officers also had identification able plant for animal production and invade and dominate native and intro­ assistance from agronomists and soil conservation on the one hand and duced pastures. For example, in New botanists. as an unpalatable weed on the other. South Wales, E. curvula has invaded The area of land infested with At present there are seven agro­ and dominated Phalaris aquatica pas­ E. curvula was assessed from records nomic types of E. curvula in New tures south of Braidwood, Lolium kept by inspectors, from their know, South Wales: robusta (blue, green, in­ perenne pastures nea r Bega and native ledge of their shires, from special sur­ termediate); chloromelas (tall, short); grass pastures near Cooma (Campbell, veys, or from indications of infesta­ conferta; and curvula (Leigh and unpublished data). Many landholders tions on shire maps. The degree of in­ Davidson, 1968; Johnston and in the northern and southern table­ festation was classified according to Aveyard, 1977). The types are iden­ lands and' south coast consider it to be density: class I - dense infestation; tified only with difficulty; the dif­ a weed. Thus we need to know the class 2 - scattered patches with ferences are based on leaf colour and present area and distribution of the isolated plants interspersed; and class 3 size, stalkiness, habit, plant height, plan t so that, should it be c1assifiied as - scattered plants only. 108 Auslra han Weeds Vol. 2(3) AUlumn 1983

Results and discussion compared with that occupied by other had class 2 infestations, and only seven pasture weeds. for example, Nassel/a of these had 100 ha or more. Of the 56 trichotoma L. - 680 000 ha or Hyperi­ shires infested only 19 had a total in­ Area and distribution cum perJoratum L. var. angustiJolium festation of 100 ha or more. Although this survey shows that DC. - 188000 ha (Campbell, 1977). The earliest record of E. curvula in E. ClIrvlIla is widely distributed in New Of the 56 shires with Eragrostis cur­ New South Wales was from the N.S.W. South Wales. mainly on the coast and vula, only 15 had class I infestations National Herbarium where plants tablelands (Table I. Figure I). the total and only five of these had 100 ha or coll ected at Cowra and Richmond were area occupied (45000 hal is small more (Table I). Similarly, 19 shires identified in 1900 (Leigh and David-

Table 1 Area of land and number of properties infested with Eragrostis curvula in the shires of New South Wales

Shire Area infested (ha) Number of properties infested class I class 2 class 3 Total class I class 2 class 3 Total

A.C.T. to 50 500 560 3 t5 18 Ballina 270 270 1 t Baulkham Hills 20 20 Bega Valley t 500 1000 5800 8300 14 10 20 44 Berrigan 1 2 Bombala 10 12 2 5 8 Byron 270 270 Cabonne t 1 10 50 t50 210 City of 1 1 City of Greater Lithgow to 30 90 130 4 4 200 200 10 10 City of 1 1 Conargo 1 1 Cooma-Monaro 200 5500 12 000 17700 20 30 80 130 Coonabarabran 1 Cootamundra 1 Cowra 1 Crookwe11 1 Culcaim 1 1 Dumelesq to 20 30 10 10 EUlobodalla 1 1 Evans 200 900 1 100 20 46 66 Gundagai 1 1 Guyra 2 2 Harden 1 1 ·Hawkesbury 40 60 tOO 10 10 Holbrook 1 1 Hume 20 20 2 2 Inverell 10 10 Kcmpscy 1 1 270 270 Lachlan 1 t Manilla 21 21 Mulwaree 20 20 200 240 4 4 Murrurundi 25 25 Narrabri 1 t Narrandera 1 t Nundle 25 25 Oberon 1 1 Parkes 5 20 20 45 Parry 20 20 Port Stephens 2 2 Quirindi 20 20 270 270 1 Severn 50 100 300 450 3 6 20 29 Snowy River 50 50 80 180 4 4 6 14 Tallaganda 100 200 1000 1300 7 4 30 41 Tenterfield 400 6000 6000 12400 2 50 160 212 Tweed 240 240 1 Uralla 5 5 Walcha 3 3 5 5 Wingecarribee 4 3 60 67 2 2 Wollondilly 40 25 8 73 3 4 to 17 Yarrowlumla tOO 200 80 380 20 20 TOTAL 2500 13500 29000 45000 54 13 3 457 644 Australian Weeds Vol. 2(3) Autumn 1983 109

Figure 1 Distribution of Eragrostis curvula in New Sou th Wal es on a shi re classificat ion basis

L

o

"-!.-~,~.--f--=t,~, . <-,."

Shirel with > 1000 h. 01 E. cUOful. 17 Snowy River 35 Gundagai 1 Begi;! Valley 18 Tweed 36 Guyrs t9 Yarrowl um la 2 Cooma- Monaro 31 Harden 3 Evans 38 Holbrook. •4 Ta ll aganda 39 Hume 5 Ten ter1 ield 40 Inverell [I] Shlrel with < 100 ha 01 E. cur~ula 41 Kempsey 20 Bau!kham Hill s 42 Lachlan 21 Berrigan 43 Manilla 22 Bomba!a 44 Murrurundi Shirel with 100 to 1000 h. 01 E. curtul. S 23 Cabonne 45 Narrsbri 6 A.C.T. 24 City 01 Dubbo 46 Narra ndera 1 Ballina 25 City 01 Shoa!haveo 47 Nundla B Byron 26 City 01 Wagga Wa gga 48 Oberon 9 City of Blue Mounlains 21 Conargo 49 Parkes 10 City 01 Grealer Lithgow 28 Coonabarabran 50 Parry 11 City 01 Lismore 29 Cootamundra 51 Pan Stephens t2 Hawkesbury 30 Cowra 52 Ouirindi 13 Kyog le 31 Crookwell 53 Ura lla 14 Mulwaree 32 Culcalm 54 Walcha 15 Ric hmond River 33 Oumeresq 55 Wingecafribea 16 Severn 34 Euroboda ll s 56 Woliondilly

son, 1968). Since then over 100 lines and national parks ). E cllrvllia oc­ Far North Coast County Council but is of E. curvula have been assessed in ex­ curred on rural land in 26 shires which not considered to be a weed because it periments for their value in pastures included infesta ti ons on 644 properties has bee n present for 40 years and has and soil conservation, Leigh and (Table I) . However, only 54 of these spread on ly to farm tracks, school Davidson ( 1968) recorded 13 sites and properties had class I infestations. playgrounds and other areas where W. H. Johnston (personal communica­ This compares with 3794 properties traffic has des troyed competing tion, 1982) four sites in New South infested with Nassella Iricholoma in species. The greater spread of E. cur­ Wales where various strains had been New South Wales, 283 of them having vula on the south coas t than on the sown. Leigh and Dav idson ( 1968) also class I infesta ti ons (Campbell, 1977). north coast may be due to lower rain­ recorded that E. curvula had become The spread of E. curvula in New fall on the south coast than on the naturalized, outside the controlled ex­ South Wales has proceeded along a north coast a nd thus less competition perimental areas , in nine situations in number of pathways. In 8 ega Valley from ex isting pastures, and/ or to the New South Wales by 1968. The survey Shire, the plant has spread from road­ predominance of sandy textured soil reported here shows the plant has now sides to adjacent cultivated paddocks, a (derived fr om granite) on the south become naturalized in 56 shires. pathway common in southern Africa coast which favours its growth and Almost all (53) of the infested shires (Leigh and Davidson, 1968). E. cur­ spread (Leigh and Davidson, 1968) had roadside infesta ti ons of E. curvula vula was first noticed in the 8ega more th an the heavier soil s of the north and, in 36 of these, roads ides were Valley in 1945 (E . Cochrane, Candelo, coast. classed as the main problem area for personal communicati on. 1982) but On the southern tablelands E. cur­ control of the weed (Table 2) . Other was not considered a problem until it vula (fi rst recorded at Cooma in 1948) situations in which E. curvula occurred began to spread after dry years in the infests 60 km of roadsides on the high­ we re, railway land, ri ver banks and in­ 1960s and' 1970s. Thus it appea rs to way from Cooma to Mic helago. During termittently grazed land (mainly aban­ have spread over an estimated 8300 ha the last 34 years it has spread from doned mines. forests, travelling stock in 37 years. E. curvula infests 8000 km these roadsides to approx imately routes, cemeteries, school playgrounds of roadsides in shires contro ll ed by the . 17 700 ha of the adj acent paddocks 110 Australian Weeds Vol. 2(3) Autumn 1983

Table 2 Occurrence of Eragrosris cur­ indications are that the plant can irrigated and supplied with nitrogen vula in shires in New South spread further, at least wi thin the en­ fertilizer (Squires and Myers, 1970). It Wales vironments that have already proved to has also been used for beef production be suitable. As there are large areas on in Oklahoma, U.S. A. where, over three Number of shires the south coast and northern, ce ntral years, steers gai ned an average of 0.44 and southern tablelands that are free of to 0.49 kg beast-' day-', the higher Occurrence in: 56 the weed but have a similar environ­ we ight gain on a palatable selection Declared noxious in : 18 ment to areas already infested, the po­ and the lower on an unpalatable strain Shires wi th infestat ions of: tential for spread in New South Wales (Voigt et 01., 1970). T he pastures were 1000 to 20 000 ha 5 mu st be rega rded as high. fertilized a nnually with 60 kg ha-' of 100 to 1000 ha 14 urea. Palatability was associated wi th IOto IOO ha 14 Weed potential low lignin:cellulose ratio, tallness, < 10 ha 23 In 12 of the 56 infested shires, E. cur­ late heading and wide leaf. In west Shires wit h infestations on: vula was rated by weed officers Texas, experiments have shown that by roadsides 53 amongst the ten most important weeds, correct animal managemen t an un­ inte rmi ttently grazed land I 28 but in only five shires was it rated in palatable strai n of E. curvula can be railway l

from U.S.A. and South Africa in­ dicates that with an appropriate grazing strategy the plant may not be a weed but may be used for animal pro­ duction. However, forcing animals to N graze unpalatable grasses, for example Nasse/la Iricholoma, is generally not successful over large areas or in heavy infestations, because in periods of abundance animals avoid the weed and in periods of scarcity they overgraze 1 useful species (Campbell, unpublished '" data). To overcome this problem a search for more palatable strains of Eragroslis curvula has begun in South Africa (Kruger and Grunow, 1982) and Aust­ ralia. As a result a new strain of E. cur­ vula, a conferta-type, has recently been submitted for registration as a pasture species in New South Wales (W. H. Johnston, personal communication, 1982). This strain was shown in grazing trials to be more palatable than many other strains of E. curvula (Johnston and Aveyard, 1977). How­ ever, if it is released as a pasture species, animals will have the choice of a much wider range of alternatives than other strains of E. curvula. Thus its palatability should be compared with that of widely distributed pasture species in New South Wales to ensure that it will not be ignored by grazing animals and become another weed in these situations. It must be remem­ bered that Nasse/la Iricholoma is grazed w class 1 infestation in Argentina in preference to other w closely-related grasses (Connor, 1960) t:> but has become a major weed in New Cl South Wales because animals refuse to cla ss 2 infestation graze it, selecting the more palatable pasture species available here. Irres­ • pective of the possible value of the con­ cl ass 3 infestation ferta-type, the problem of the types already established in the 56 shires in New South Wales remains.

Qmlrol At present Eragroslis curvula can be controlled on arable land by plough­ ing, sowing improved pasture, spelling from animals for a year and removing any re-infestation. On non-arable land scattered plants can be removed by chipping. Recent research has in­ dicated that the herbicide tetrapion can remove selectively E. curvula from Phalaris aqualica, native perennial grasses and Penniselum clandeslinum which will assist control on both arable and non-arable land (Campbell, un­ published data). Registration of the herbicide for these purposes is cur­ o 5 10 KMS rently in progress. Replacement of Eragroslis curvula on roadsides, rail­

Figure 2 Spread of Efagrostis curvula from the roadsides to paddocks in the 60 km between Cooma and way land, intermittently grazed land Michelago on the southern tablelands of New South Wales. and riverbanks, a major problem in its 112 Austra lian Weeds Vol. 2(3) AlAumn 1983 control. could be achieved by using tet­ strains of E. curvula would incur ex­ Cross. D. O. (1937). Yass River tussock. rapion for selective removal of E. cur­ penses to some local government weed Agricultural Gazette of New South Wales vula from native perennial grasses and control authorities and to some land­ 48:546 - 8. broadleafed plants which are common holders but. because the areas infested Cuthbertson, E. G. (1978). Advances in in these situations. Other methods weed distribution mapping. Proceedings are relatively small, immediate control of the First Conference of the Council of would also need to be used to stop its could prove to be a sound investment, Australian Weed Science Societies, spread. for example aerial sowing or keeping in mind the potential spread of . pp. 273-87. direct drilling of improved pastures the plant. In that respect, it may be Johnston, W. H. and Aveyard, J. M. after herbicide treatment and the use of wise to reflect on the fact that Cross (1977). Testing and selection of African road-making equipment free of E. cur­ (I 937) warned that Nassella trichotoma lovegrass (Eragrostis curvula) for soil vula seeds. was a potential weed in 1937. Had con­ conservation in south-western New trol been enforced then it could have South Wales. Australian Plant Intro­ Conclusion saved New South Wales annual duction Review 12:27 -40. Kruger, A. J. and Grunow, J. O. (1982). This survey shows that naturalized production losses of $11 .8 million and Voluntary intake and quality studies a control bill of $24.4 million (V ere strains of E. curvula are widely distrib­ with sheep on forty ecotypes of Erag­ uted in New South Wales but occupy a and Campbell, 1979). rostis curvula (Schrad.) Nees. M.Sc relatively small area compared to other (Agric) thesis, University of Pretoria, major pasture weeds. They are un­ Acknowledgements South Africa. palatable and have .the potential to Leigh, 1. H. and Davidson, R. L. (1968). spread in regions where the environ­ Information provided by weed officers Eragrastis curvula (Schrad.) Nees and ment favours their growth. This in­ and officers of the New South Wales some other African lovegrass . Australian dicates that E. curvula is a weed in New Department of Agriculture and assist­ Plant Introduction Review 5:21-44. South Wales that falls into Category B ance proffered by weed control bodies Leigh. J. H. and Mulham, W. E. (1964). in the classification of weeds by Amor in New South Wales is gratefully The performance of introduced dryland acknowledged. species on three soil types in the and Twentyman (1974). Category B is southern Riverine plain. C.S.I.R.O. a proclaimed species that should be Division of Plant Industries Field Station controlled by governments and by References Record 3:9-20. local government as follows: assisting Amor, R. L. andTwentyman.J. D. (1974). Smith, K. R. (1975). A new system of weed landholders through research, exten­ Objectives of, and objections to, Aust­ surveying and its use on silverleafnight­ sion. hiring of equipment and loans; ralian Noxious Weed Legislation. Jour­ shade. Journal of Agriculture of South suppressing the weed on public land; nal of the Australian Institute ofAgricul­ Australia 78:35-9. and organizing containment in regions tural Science 40:194 - 203 . Squires, V. R. and Myers, L. F. (1970). where local opinion is strongly in Auld, B. A. and Scarsbrick, B. D. (1970). Performance of warm-season perennial favour of enforced control. This could Chloromelas lovegrass in the Tentemeld grasses for irrigated pastures at Denili­ area of New South Wales. Journal of the quin, South-eastern Au stralia . Tropical be accomplished under the present Grasslands 4: 153 - 6\. weed administrative system in New Australian Institute 0/ Agricultural Science 36:296-7. Quinlivan, B. 1. (1972). An ecological South Wales. Campbell, M. H. (1977). AsseSSing the area basis for decision making. Journal of the It would require. in addition to ac­ and distribution of serrated tussock (Nas­ Australian Institute 0/ Agricultural tivities already in progress: research by sella trichotoma), St John's wort (Hyperi­ Science 38:283 -6. government bodies into the biology cum per/oratum var. angusti/olium) and Vere, D. T. and Campbell, M. H. (1979). and ecology of the weed and its control si/ton bush (Cassinia arcuata) in New Estimating the economic impact of ser­ by grazing management. extension of South Wales. New South Wales Depart­ rated tussock (Nassella 'richotoma) in information by the New South Wales ment of Agriculture, Technical Bulletin New Sou th Wales. Journal of the Aust­ Department of Agriculture on the 18. ralian Institute of Agricultural Science Connor, H. E. (1960). Nassella tussock in 45 ,35-43. danger of the weed and methods of Voigt, P. W., Kneebone, W. R.. Mcilvain. control. and special control pro­ Argentina. New Zealand Journal ofAgri­ culture 100:18 - 21. E. H., Schoop, M. C. and Webster, J. E. grammes initiated by local government Cotter, P. F., Dahl, B. E. and Stasi, C. (1970). Palatability, chemical composi­ weed authorities in shires where the (198 \). Weeping lovegrass' potential in tion and animal gains from selections of weed is already widespread or where it west Texas. Noxious Brush and Weed weeping lovegrass, Eragroslis curvula is a potential threat. Control Range and Wildlife Management (Schrad.) Nees. Agronomy Journal Statewide control of the naturalized 12:30- I. 62:673-6.