Australian Forestry 2004 Vol. 67, No. 2 pp. 101–113 101

Spatial analysis of an outbreak of lugens (: Noctuidae) in the south- west of Western Australia: does logging, vegetation type or fire influence outbreaks?

J.D. Farr1, D. Swain2 and F. Metcalf3

1Science Division, Department of Conservation and Land Management, Brain St, Manjimup, Western Australia 6258, Australia Email: [email protected] 2Forest Management Branch, Department of Conservation and Land Management, Brain St, Manjimup, WA 6258, Australia 3Fire Management Services, Department of Conservation and Land Management, 17 Dick Perry Av., Kensington, WA 6151, Australia

Revised manuscript received 29 September 2003

Summary The aerial observations revealed that areas of severe defoliation existed as distinct localities in the southern jarrah forest, north- There was an outbreak of gumleaf skeletoniser (GLS, Uraba west of Manjimup and north of Walpole (Strelein 1988a; Abbott lugens Walker) in the southern jarrah forest in 1982–1988. Aerial 1992a,b). The extent of marked defoliation in successive outbreak and observational survey data were used in a Geographic years expanded from 90 000 ha in 1983 to 300 000 ha in 1985 Information System analysis of the possible impact of forest (Abbott 1987, 1990) such that the infestation was advancing on management practices on this . In addition, this allowed approximately a north-easterly front (Strelein 1988a; Abbott 1992b). investigation of the influence of vegetation type on the distribution of GLS. A total area of 6.7 million ha was interrogated, including Jarrah is an important Australian hardwood and its exploitation 89 900 ha of land infested with GLS. Neither logging nor has played a major role in the development of WA since the prescribed burning induced increases in GLS populations. settlement of Perth by Europeans in 1829. It occurs in the south- However, a decrease in area infested with GLS was indicated for western corner of the state, extending from Gingin (80 km north prescribed burning up to 3 y prior to the outbreak. Interrogation of Perth) southwards to Albany, with the exception of some eastern of vegetation complexes indicated that the GLS outbreak was outlying populations (Abbott and Loneragan 1986). The climate initiated on marginal jarrah forest on poorly drained sites that are is mediterranean with hot dry summers and cool wet winters. The prone to inundation in winter and drought in summer. In addition, jarrah forest is divided into two main regions, northern and southern, bivoltine GLS populations were discovered on the fringes of the demarcated by the Blackwood and Preston Rivers. southern jarrah forest during the observational surveys. This In the northern jarrah forest significant exploitation began in the supports the hypothesis that bivoltine GLS were more prevalent 1850s when a convict transportation scheme provided much- in the southern jarrah forest during the outbreak as opposed to needed labour for the developing colony (Mills 1989). In contrast, univoltine GLS during non-outbreak periods. although settlement of the southern jarrah forest began in the Keywords: insect pests; population dynamics; defoliation; outbreak; 1850s, the primary attraction was grazing land and minimal landforms; vegetation types; forest management; logging; controlled disturbance of the jarrah forest occurred (Berry 1987). Significant burning; gumleaf skeletoniser; Uraba lugens; jarrah; disturbance of the southern jarrah forest began in 1911 with the marginata; marri; Corymbia calophylla; Western Australia extension of the railway south of the Blackwood River from Bridgetown to Jardee and the establishment of two sawmills Introduction (Jardee and Pemberton) to supply sleepers for the Australian transcontinental railway (Berry 1987; Mills 1989). Cutting in A severe outbreak of Uraba lugens Walker (gumleaf skeletoniser, jarrah forest was effectively uncontrolled until 1920, following GLS) occurred on jarrah ( Donn ex Smith) the passing of the Forests Act in 1918 and the formation of the in the south-west of Western Australia (WA) during 1982–1988 Forests Department (Berry 1987; Mills 1989; Heberle 1997). (Strelein 1988a; Abbott 1990, 1992a). Prior to this, GLS was Thus, by 1920 parts of the northern jarrah forest had been exploited observed to attack isolated rural trees, but otherwise was not with uncontrolled cutting for over 70 y and parts of the southern considered a problem in WA. Although GLS is briefly mentioned jarrah forest for nearly ten. In the 1920s, dedication of State forest in a 1960 Forests Department questionnaire as a potential future was initiated, cutting was restricted to over-mature trees and a problem and by Jenkins and Curry (1971), the only previous record system of tree marking to control cutting was established (Havel of an outbreak was in 1947 between Calingiri and Cowaramup 1989; Heberle 1997). In addition, a fire management strategy was (Agriculture Western Australia, Insect Collection records; Western developed involving fire exclusion in young regrowth and low- Australian Department of Agriculture 1948). Nevertheless severe intensity fire in older regrowth and virgin mature forest (McCaw defoliation by GLS in January 1983 led to a scorched appearance and Burrows 1989). Prior to European settlement, the Aboriginal of jarrah crowns, enabling the detection of high population areas practice of frequent burning of the forest maintained its open ‘park- from the air. This event was regarded as unprecedented in jarrah like’ structure (Hallam 1975; Ward et al. 2001). Graziers and karri forests and caused considerable consternation. perpetuated the burning practices used by Aboriginal people, 102 Analysis of an outbreak of Uraba in WA

particularly in the southern forests and associated coastal woodlands. But, with the expansion of European settlement and the ever-increasing accumulation of logging debris, fuel loads in the jarrah forest dramatically increased and wildfires were intense WA and devastating (McCaw and Burrows 1989; Burrows et al. 1995). From 1920 to the mid-1950s control burning was mainly restricted ANDREW to strategic strips around forest compartments. In 1954 a new fire BARLEE YANMAH LOCALITY EASTER GORDON management policy was developed involving broadscale fuel Manjimup reduction burning in spring, and this practice was expanded and IFFLEY refined over the ensuing years (McCaw and Burrows 1989) such that fire became an integral part of forest management in the jarrah GIBLETT CAREY forest.

SUTTON GLS is a well-known defoliator of eucalypts and related species Pemberton throughout Australia (Brimblecombe 1962; Campbell 1962; Harris 1972, 1974; Harris et al. 1977; Elliot and Bashford 1979;

Strelein 1988a) and more recently New Zealand (Bain 2001). WELD JOHNSTON Outbreaks in eastern Australia have been associated with drought MOSSOP WYE DEEP and flood cycles (Campbell 1962; Harris 1972; Harris et al. 1977). N LONDON Other studies suggest a decreased ability of the host plant to protect FRANKLAND SOHO PINGERUP COLLIS THAMES itself, due to environmental conditions such as drought, may induce TRENT

GLS population increase (Cobbinah et al. 1982; Farr 1985). In Severe GLS defoliation GLS 1983 defoliation 1983 eastern Australia, outbreaks are favoured where there are trees growing close together with overlapping crowns, abundant low SCALE 0 20 Kilometres undamaged foliage, dense young regeneration and abundant litter Walpole on the forest floor (Campbell 1962; Harris 1975). Therefore, forest management practices such as logging and prescribed burning Figure 1. Gumleaf skeletoniser infestation in the southern jarrah forest may influence GLS populations. In the southern jarrah forests of in 1983, determined from aerial surveillance. (From Strelein 1988a). WA, however, there are subtle differences in the biology and behaviour of GLS. Pupation is not predominantly in the forest floor; larvae feed higher in the canopy than do eastern Australian height-class and vegetation complex (Mattiske and Havel 1998) populations; and GLS is predominantly univoltine compared with — were then cross tabulated with the GLS infestation theme and the bivoltine eastern coastal bio-morph on which most of the areas where overlap occurred were summed. Because logging outbreak studies have been made (Farr 2002). history in FMIS is categorised in sequences of 10 y, interrogation of logging history was limited to this time span. The forest species This paper therefore examines the spatial distribution of GLS both height class theme describes the potential height of native tree species during and following the outbreak in south-western WA, and based on the codominant height of a stand at maturity. These data investigates the extent to which logging, fire or vegetation type were interpreted from 1950s and 1960s aerial photographs (scale influenced the population. 1:15 840) and mapped to produce a 1:25 000 scale Aerial Photography Interpretation (API) map series (Bradshaw et al. 1997). Materials and methods Vegetation complexes are based on Mattiske and Havel (1998) and cross-referenced with soils and landforms classified by Logging and vegetation interrogation Churchward et al. (1988) and Churchward (1992). Climatic zones follow Mattiske (1997): hyper-humid refers to annual rainfall Aerial survey maps of the GLS outbreak for 1983–1986 inclusive >1100 mm and total summer evaporation of <450 mm; per-humid (Abbott 1987; Strelein 1988a) were digitised and recorded on a and humid zones refer to annual rainfall >1000 mm and total CAD package, MicroStation (Bentley™). Infestation boundaries summer evaporation <525 mm and <600 mm respectively. were most clearly defined for the 1983 aerial survey (Fig. 1). In addition, infested cells within the 1983 aerial survey were ground truthed for the presence of GLS in the following forest blocks: Ground observational surveys Yanmah, Barlee/Andrew, Gordon, Easter/Iffley, Carey, Gibblet, Ground-based observational surveys were conducted from 1985 Sutton, Wye, Wye/Deep, Johnston, Weld and Soho (although to 1997. Jarrah and marri (Corymbia calophylla (Lindl.) K.D.Hill quantitative estimates of populations were not made). The 1983 and L.A.S.Johnson) canopies and understorey were visually aerial data were imported into a Forest Management Information inspected for presence or evidence of GLS from September to System (FMIS), which is a computer-based raster or grid January (the period when GLS is most apparent in the field). Geographic Information System (GIS) developed by the Assessment was set at a minimum of five minutes at each location Department of Conservation and Land Management (CALM) and within the forest. The presence/absence of GLS larvae, egg rafts, the former Forests Department. The GLS-infested areas were moulting casts and leaf skeletonising were recorded against a map represented by a single code value (‘1’) which consisted of pixels reference, and leaf samples were taken as a check. From 1985 to of 0.5 ha within the Warren, South West and Swan regions. 1989 the survey targeted the outbreak front and was conducted Overlays of three FMIS themes — logging history, forest species Australian Forestry 2004 Vol. 67, No. 2 pp. 101–113 103

Table 1. FMIS interrogation of logging history on 1983 aerial survey maps of GLS infestation (Bkg = background)

Description Non-infested GLS-infested Non-infested area GLS-infested area area (ha) area (ha) (% of (total–Bkg)) (% of (total–Bkg)) Background code* 3 620 693 391 Last logged before 1920 98 512 124 3.3 0.1 Last logged 1920–1929 59 265 834 2.0 0.9 Last logged 1930–1939 121 909 3334 4.0 3.7 Last logged 1940–1949 159 688 2075 5.3 2.3 Last logged 1950–1959 424 658 6262 14.0 7.0 Last logged 1960–1969 458 192 4204 15.1 4.7 Last logged 1970–1979 348 730 5257 11.5 5.9 Last logged 1980–1989 205 205 8768 6.8 9.8 Unlogged by 1990 1 158 878 58 696 38.2 65.5 Total 6 655 730 89 943

*Areas which are not managed by CALM, including private property, plantations, water and areas where there are no logging records irrespective of land tenure

by CALM staff from the Entomology Program. From 1990, due last logged areas for 1980–1989 can be attributed to unlogged to a decline in the GLS population, the survey was extended more forest during the period of the GLS outbreak. Consequently, over broadly to cover the Warren and South West forest regions in 65.5% of the GLS-infested area in 1983 occurred in unlogged case the outbreak areas had dispersed more widely. This survey forest, indicating it is unlikely that logging influenced the was conducted by CALM operational staff who were trained to development of GLS outbreak populations. recognise GLS and its symptoms of attack, and closely followed the strategy used by Abbott (1992b) for jarrah leafminer. For the species height-class interrogation (Table 2), GLS favoured ‘A’ class sites of 25–29 m co-dominant mature stand height for all species except karri (E. diversicolor F.Muell.). Thus GLS Fire interrogation preferred jarrah and jarrah-marri mixed forests. A slight preference Twelve years (1970/71–1981/82) of fire history information from for areas of ‘A’ and ‘B’ class karri (40 – >50 m) is also indicated. microfiche map records were captured into the GIS. Records other However, this may reflect the presence of mixed forest stands of than those within the Manjimup District boundary were not karri, marri and jarrah in these areas. available at the time of analysis. In addition, information on the A total of 295 vegetation complexes were present in the area time, intensity and type of burn were not recorded within the interrogated. Of these, 51 were present in the GLS-infested areas microfiche records and therefore were not available for this study. and are listed in Appendix I. Due to the large number of vegetation Data fields for fire frequency and year-last-burnt were created for classifications, vegetation types in GLS-infested areas with >2% each year, then combined into a single data set. Only land with of the total GLS-infested area were considered most relevant. This CALM tenure was used, ensuring that areas with no fire history threshold was calculated as such as the Manjimup townsite, private land and water bodies were not included in the unburnt area examined. The 1983 GLS total GLS-infested area – background areas Aerial Survey (captured as above) and combined fire history data number of vegetation classes in GLS-infested area set were then applied to the Manjimup District boundary and combined into a single fire frequency/year-last-burnt and GLS expressed as per cent of the total GLS-infested area. Vegetation 1983 dataset to enable interrogation. In addition, individual point complexes occupying >2% of the GLS-infested area are detailed location references from ground-based observational surveys (as in Appendixes II and III. above) for the Manjimup district were overlaid on maps of fire Bevan (BE1) was the most predominant vegetation complex and frequency and year-last-burnt GIS records. In this way individual covered 11.5% of the GLS-infested area (Appendix I). This observational records of positive and negative GLS sightings were complex comprises tall open forest of marri and jarrah on uplands χ2 counted against fire history information and analysed using in per-humid and humid zones and is the most extensive landform contingency tables. north and east of Manjimup. Yanmah (YN1) intersects this vegetation complex, covering 5.1% of the GLS-infested area. The Results Yanmah vegetation complex follows watercourses and occupies valleys in per-humid and humid zones. The principal forest species Logging and vegetation interrogation is karri, with tall open forests of marri, yarri (E. patens Benth.) A total of 6.7 million ha was interrogated, including 89 900 ha of and jarrah. land infested with large populations of GLS. The second most common vegetation complexes in GLS-infested Results of the logging interrogation are shown in Table 1. Because areas were Pingerup (Pi) and Caldyanup (CA). Pingerup occupies logged areas are divided into ten-year categories, a portion of the 7.4% of the GLS-infested areas, and occurs mainly in Mossop, 104 Analysis of an outbreak of Uraba in WA

Table 2. FMIS interrogation of forest species height class on 1983 aerial survey maps of GLS infestation. (API = Aerial Photography Interpretation, Bkg = background)

Description Non-infested GLS-infested Non-infested area GLS-infested area area (ha) area (ha) (% of (total–Bkg)) (% of (total–Bkg)) Background* 3 436 837 20 468 API data not available 410 711 12.8 API data not captured 688 138 551 21.4 0.8 All species except karri: A+ (>30 m) 15 348 2 694 0.5 3.9 All species except karri: A (25–29 m) 553 862 25 828 17.2 37.2 All species except karri: B+ (20–24 m) 662 031 7 357 20.6 10.6 All species except karri: B (15–19 m) 543 947 5 145 16.9 7.4 All species except karri: C (<15 m) 168 534 5 505 5.2 7.9 Karri: A (>50 m) 84 227 9 764 2.6 14.1 Karri: B (40–49 m) 77 767 11 179 2.4 16.1 Karri: C (<40 m) 14 275 1 454 0.4 2.1 Total 6 655 681 89 943 *Areas not managed by CALM, including private property, plantations, water and areas other than native forest

Wye, Deep and Pingerup forest blocks in the GLS-infested area Mattaband (Mty1), which comprises 5.8% of the GLS-infested north-west of Walpole (Fig. 1). This vegetation complex consists area, adjoins Pingerup and Caldyanup complexes. This complex of a mosaic of closed heaths of Myrtaceae species and sedgeland occurs higher in the landscape, and is a mixture of tall open karri of Restionaceae and Cyperaceae species on remnant marine sands. forest and an open forest of jarrah and marri. Although jarrah and marri are not ascribed as major species within this vegetation complex, they are present as scattered individuals Thus, of the predominant landform units, marginal jarrah of generally low stature. Caldyanup vegetation complex, which woodland is present in Pingerup, Caldyanup, Angove Fernley and occurs in 7.3% of GLS-infested areas (Appendix I), is a mosaic of Cattaminup which comprise 22.6% of the GLS-infested area low woodland of Allocasuarina fraseriana (Miq.) L.A.S.Johnson, (Appendix I). A common element in these five landform units is Corymbia ficifolia (F.Muell.) K.D.Hill & L.A.S.Johnson (a the occurrence of swamps (Appendix II) and humus podzols preferred host for GLS) and Banksia spp.; sedgeland and tall (Appendix III). With the exception of Angove and Fernley, where shrubland of Cyperaceae and Myrtaceae species (respectively); the landforms have developed on unconsolidated quartzite, open woodland of Melaleuca preissiana Schauer and jarrah siltstone and sandstone (respectively), the other landform units in woodland (Appendix II). Caldyanup is skirted by Fernley (2.6%), Appendices II and III have developed on igneous crystalline which is a woodland mixture of bullich (E. megacarpa F.Muell.) granites and gneisses. The most common soil is yellow duplex and yarri, and includes tall Myrtaceae shrubland, again with jarrah (Appendix III). woodland. The Fernley vegetation complex is located north-east of Walpole and includes forest blocks Frankland, Soho, Collis, Ground observational surveys Trent, Thames and London in the GLS-infested area (Fig. 1). A total of 874 observations were made of GLS occurrence from 1985 to 1997 (Table 3). A map of the outbreak front survey 1985– Table 3. GLS ground observational survey summary (Regional refers to 1990 is shown as Figure 2 and the 1990–1997 Regional Surveys Southern and Central Forest Regions) as Figure 3. Despite limited survey coverage in 1995–1997, Table 3 demonstrates the decline in GLS incidence from 94% in 1987– Survey year Survey No. forest No. GLS GLS 1988 to 12% in 1994–1995. (Sept–Jan) type blocks sightings (%) Evidence of a bivoltine GLS population was found in Chariup 1985–86 Outbreak front 82 35 42.6 (7.i.1991, MGA94 coordinates 469120, 6209400), Arklow 1987–88 Outbreak front 52 49 94.2 (24.i.1991, MGA94 coordinates 415820, 6316880) and Chalk 1988–89 Outbreak front 49 10 20.4 (5.ii.1991, MGA94 coordinates 426610, 6343770) forest blocks. At Chariup and Arklow, second-instar larvae were present at a 1989–90 Outbreak front 64 18 28.1 time when univoltine GLS in the southern jarrah forest were at 1990–91 Regional 176 24 13.6 instars 9–11. At Chalk, third- and fourth-instar larvae were present 1991–92 Regional 129 6 4.7 when univoltine GLS in the southern jarrah forest were at instar- 1992–93 Regional 124 12 9.7 10 pupation. 1993–94 Regional 115 17 14.8 Aerial survey fire interrogation 1994–95 Regional 59 7 11.8 1995–96 Regional 6 2 33.3 A total area of 486 890 ha was interrogated in respect to fire history 1996–97 Regional 18 0 0 and GLS distribution in the Manjimup district. This comprised 262 932 ha of CALM-tenured land. Figure 4 shows the distribution Australian Forestry 2004 Vol. 67, No. 2 pp. 101–113 105

of GLS in respect to year last burnt and Figure 5 shows the distribution of GLS in respect to fire SWAN frequency.

Fire history results vary for different years-last- SOUTH WEST burnt (Table 4). For example, those years where Bunbury >6% of CALM-tenured land was burnt, a two- WARREN fold or greater difference between infested and CALM REGIONS uninfested GLS areas occurred in 1973–74 and 1976–77, where the proportion of GLS-infested KEY land is greater. Conversely the proportion of GLS- GLS larvae, moulting casts, egg rafts present infested land is smaller than that of uninfested land Leaf skeletonising present in 1975–76, 1979–80 and 1980–81. Nevertheless Nil GLS the year-last-burnt fire history does show a consistent and considerable decrease in GLS- infested land burnt from 1979–82, just prior to the outbreak. However, comparison of unburnt Manjimup proportional areas for infested (20.1%) and uninfested (18.1%) GLS show little difference, Pemberton indicating no influence of fire on GLS populations (Table 4). Fire frequency (Table 5) also shows no strong differences between infested and uninfested N areas.

SCALE Interrogation of year-last-burnt fire history on the 0 50 Kilometres GLS observational survey data indicates no significant relationship (χ2 = 20.12, P > 0.05, Walpole df = 12). Similarly, for pooled burnt and unburnt sites, there was no significant relationship (χ2 = Figure 2. Ground observational surveys of GLS, 1985–1990 2.22, P > 0.05, df = 1). Significance was found for fire frequency on the observational survey data (Table 6), where a greater incidence of GLS than expected was found in areas burnt three times, but the extent of GLS infestation following two fires was lower than expected. However, this trend is not supported by an examination of the 1983 aerial survey results in Bunbury relation to fire frequency (Table 5). Discussion KEY GLS larvae, moulting casts, Logging egg rafts present Leaf skeletonising present FMIS interrogation of GLS aerial surveys during Nil GLS the initial outbreak period (1983) indicated that GLS, two generations / year Manjimup logging in the southern jarrah forest does not promote GLS populations. However, logging disturbance in forest ecosystems is known to influence outbreaks of defoliating such as the spruce budworm and jack pine budworm (Blais Pemberton 1983; Swetnam and Lynch 1989; McCullough and Kulman 1991; McCullough et al. 1998). These outbreaks are a consequence of tree species, age N and stand structure, such that outbreaks may be reduced (Anderson et al. 1987; Bergeron and Harvey 1997) or exacerbated (Blais 1983; SCALE 0 50 Kilometres Swetnam and Lynch 1989; McCullough and Kulman 1991) by various logging regimes and Walpole forest management practices.

It has been suggested that outbreaks of jarrah Figure 3. Ground observational surveys of GLS, 1990–1997 leafminer ( glyphopa Common) in the 106 Analysis of an outbreak of Uraba in WA

Figure 4. The distribution of GLS from the 1983 aerial and ground observational surveys of GLS incidence in respect to year-last-burnt in the Manjimup District

Table 4. GIS interrogation of year last burnt fire history for CALM- Table 5. GIS interrogation of fire frequency from 1970 to 1982 for tenured land in the Manjimup district on the 1983 aerial survey maps of CALM-tenured land in the Manjimup district, on the 1983 aerial survey GLS infestation maps of GLS infestation

Year last Non-infested area GLS-infested area Fire frequency Non-infested area GLS-infested area burnt ha % of total ha % of total ha % of total ha % of total

1970–71 5 418 2.2 627 2.9 0 43 810 18.1 4 278 20.1 1971–72 3 702 1.5 847 4.0 1 94 120 37.0 8 787 41.2 1972–73 5 152 2.1 108 0.5 2 89 695 37.1 7 325 34.3 1973–74 9 403 3.9 2 784 13.1 3 13 170 5.5 869 4.1 1974–75 10 573 4.4 983 4.6 4 752 0.3 81 0.4 1975–76 19 349 8.0 143 0.7 5 44 0.0 0 0.0 1976–77 17 605 7.3 2 934 13.8 Total 241 591 21 341 1977–78 24 982 10.3 3 340 15.7 1978–79 28 616 11.9 2 847 13.3 1979–80 30 834 12.8 225 1.1 1980–81 25 221 10.4 1 213 5.7 jarrah forest are associated with forest logging due to the insect’s 1981–82 16 924 7.0 1 012 4.7 perceived preference for open, low-density forest (Wallace 1970) Unburnt and the production of large quantities of young foliage following 1970–1982 43 810 18.1 4 278 20.1 logging (Mazanec 1981, 1989). However, Abbott et al. (1993) argue that logging does not influence jarrah leafminer populations Total burnt 197 781 81.9 17 062 80.0 since oviposition is not restricted to new leaves, logging Total burnt + temporarily reduces leaf area, and annual fluctuations in outbreak unburnt 241 591 21 341 distribution do not correspond to prior logging sites. Australian Forestry 2004 Vol. 67, No. 2 pp. 101–113 107

Figure 5. The distribution of GLS from the 1983 aerial and ground observational surveys of GLS incidence in respect to fire frequency in the Manjimup District

Table 6. GIS interrogation of fire frequency for CALM-tenured land in on retained trees. However, on a stand basis, the overstorey cover the Manjimup District on observational survey data (1985–1997) of is reduced for 10–20 y (Stoneman et al. 1989), significantly presence and absence of GLS at point locations. Numbers represent counts reducing leaf area index (Abbott et al. 1993). GLS prefers newly- of point locations (+ GLS = GLS present; – GLS = GLS absent) mature and mature leaves for oviposition and larval feeding Fire + GLS – GLS (Morgan and Cobbinah 1977). Therefore, it might be expected that, should logging influence GLS populations, outbreaks would frequency Obsvd Exptd % of total Obsvd Exptd % of total follow at least 10–20 y after a particular site was logged, and 0 13 9.3 17.6 26 29.7 11.0 localised outbreaks would have occurred frequently throughout 1 34 32.5 46.0 102 103.5 43.2 the southern jarrah forest since 1922–1932. However, this has 2 17 26.3 23.0 93 83.7 39.4 not occurred. By comparison, five outbreaks of the western spruce 3 10 5.7 13.5 14 18.3 5.9 budworm (Choristoneura occidentalis Freeman) occurred from 4 0 0.00 1 0.4 1944–1992 in British Columbia (Myers 1998). Moreover, in a Total 74 236 thinning trial on E. camaldulensis (Dehn.) in the Murray and Goulburn Valleys in Victoria, Harris (1975) concluded that χ2 = 10.60, P < 0.05 (df = 3) thinning minimised damage caused by a subsequent GLS outbreak. However, oviposition was recorded on stump coppice. Logging in the jarrah forest entails selective thinning, and a gap and shelterwood system (Stoneman et al. 1988, 1989) depending The principal eucalypt species within the jarrah forest are jarrah on stand structure. Regeneration following logging comes from and marri, and under management practices applied in southern ground coppice, saplings, poles and stump coppice. Seedlings, forests 1975–2001 both species were utilised and harvested. GLS lignotuberous seedlings and seedling coppice do not develop survivorship is similar on both species (Farr 2002). Thus, unlike immediately into saplings but take 15–20 y before the lignotuber the North American mixed forests where forest management has attained sufficient size (a diameter of 10 cm or a long axis of practices have favoured the retention of spruce budworm hosts 15 cm) to respond to a reduction in overstorey competition and (Blais 1983; McCullough and Kulman 1991), selective logging develop into the sapling stage (Abbott and Loneragan 1986; should not have favoured GLS. Stoneman et al. 1989). Intensive logging does increase the foliage 108 Analysis of an outbreak of Uraba in WA

Vegetation and distribution pupation occurs from February to March (late summer–early autumn) mostly on the upper tree bole and branches, although Our evidence shows that GLS predominates in jarrah–marri (height some pupation may occur on the lower bole and in leaf litter (Farr class 25–29 m) although some significant areas containing karri 2002). Forest management fuel reduction burns are mainly done (height class 40 – >50 m) were infested. GLS has been recorded in spring (September–November), with some in autumn (March– on karri (Strelein 1988a; Abbott 1992a), but its performance on May). Flame height in spring is characteristically low, with a this host is not known and there are no reports of severe defoliation preferred maximum scorch height of 6 m (jarrah canopy occurs of this species during the outbreak period. Therefore the defoliation at 15–35 m) and a low to moderate rate of fire spread (Sneeuwjagt observed for these species height classes was probably on jarrah and Peet 1985). Scorch heights during autumn tend to be almost and marri within the interpretation site. Vegetation complexes double those in spring because of drier conditions and greater where these observations might be expected include Yanmah fuel consumption (Burrows 1997). The objective of such fires is (YN1) and Mattaband (Mty1), which are composed of both karri to consume leaf litter and lower understorey, and to stimulate and jarrah–marri forests. In addition, GLS has been recorded on lignotuber formation by burns with a silvicultural objective. Thus, E. rudis Endl., E. patens and C. ficifolia (Strelein 1988a; Abbott with the exception of wildfires where crown scorch occurs, fire in 1992a), where performance on C. ficifolia is superior to that on the upper canopy and crown is not a regular occurrence; spring jarrah and marri (Farr 2002). Corymbia ficifolia is present in the prescription burns are unlikely to directly affect GLS larvae. It is Caldyanup (CA) vegetation complex, located north-east of possible that early autumn burns may cause direct mortality of Walpole (Wardell-Johnson and Coates 1996), which comprised a some pupae present in the lower forest strata; but as most pupation notable proportion of the GLS-infested area in this region. The takes place in the upper part of the tree bole (Strelein 1988a; Farr Caldyanup complex and the adjoining associated complexes of 2002), a scorch height higher than that normally prescribed would Fernley, Angove and Pingerup are all poorly-drained landform also be required. A higher frequency of high-intensity burns and/ units and together comprise a significant proportion of the area or autumn fires during the pre-outbreak period (1979–1982) may infested by GLS in 1983. Also the Yanmah (YN1) vegetation explain our results; but information on the intensity and timing of complex, which intersects Bevan, is on poorly-drained soils these burns was not available for this study. Additionally, radiant 1 containing marginal jarrah forest (Churchward 1992; R. Hearn heat from these spring and autumn fires may explain the apparent pers. comm. 2001). According to Strelein (1988b), moderate- to increase in GLS mortality in areas burnt 1979–82, particularly high-quality jarrah forest occurs on well-drained soils in the for autumn burns where fire intensity is greater due to more landform units BEy, CRy, Kb, Mty, S1, and parts of COy and Ky, abundant fuel and lower moisture contents. whereas low-quality jarrah occurs on poorly-drained soils in landform units such as A, CA, Pi, F, and parts of Ky and COy. Leaf flush following a fire may favour some forest foliage feeders These poorly-drained landform units become waterlogged in (Mazanec 1989; Mazanec and Justin 1994; McCullough et al. winter but dry out quickly in summer (Churchward et al. 1988; 1998; Abbott et al. 1999). GLS, however, prefers mature, fully Strelein 1988b), suggesting jarrah on such sites would be subject expanded leaves and given that the outbreak started in 1982 to severe stress (Davison 1997). This conforms to the hypothesis (Strelein 1988a; Abbott 1990), an increased proportion of GLS that waterlogging and drought are significant elements in inducing in areas burned 1–3 y prior (1979–1981) might be expected. GLS outbreaks in eastern Australia (Campbell 1962; Harris et al. However, the opposite occurred. The only periods where there 1977; Farr 1985). Therefore our evidence suggests that the GLS was an increase in the proportional area of burnt forest affected outbreak in Western Australia during 1982–1988 was initiated in by GLS was for 1973–74 and 1976–77. Yet these increases were marginal jarrah forest on poorly-drained sites. moderate to low, and records for these years are not substantiated by significant continuity in the data. In addition, prescription burns Furthermore, C. ficifolia near Walpole has been observed to are closely related to logging in the jarrah forest, such that prior support bivoltine GLS (Farr 2002), in contrast to the single annual to 1990 both pre- and post-logging burns were part of forest generation more commonly reported in the southern jarrah forest management practices. Therefore, should fire influence GLS (Strelein 1988a; Abbott 1992a). Therefore, the discovery of populations, this would be reflected in the logging data — but no bivoltine populations on the fringes of the southern jarrah forest, influence by logging was found in this study. However, the during the ground observational survey, supports the hypothesis partitioning of logging data into ten-year periods precluded direct that bivoltine GLS were more common in the southern jarrah forest comparison between the data sets for logging and fire. Also, the during the 1982–1988 outbreak (Farr 2002). application of pre- and post-logging burns within a relatively short time for forest management practices would mean that GLS Fire incidence may be affected by fire frequencies of two or even three burns. Although fire frequency had some significance in the For fire to cause direct mortality it must occur when the insect is observational survey, our results were not consistent. at a vulnerable stage and location within its forest habitat (McCullough et al. 1998). Consequently, we did not expect the Although fire is known to predispose trees to attack by bark beetles lower incidence of GLS infestation (suggesting GLS mortality) and wood borers (see McCullough et al. 1998 and references that we found in areas burnt 1–3 y before the outbreak. In the therein), documented examples of fire increasing populations of southern jarrah forest, GLS eggs and larvae are present in the forest foliage feeders are limited (Hadlington and Hoschke 1959; upper canopy from April to January the following year, and Volney 1988). In addition, in the southern jarrah forest, evidence that fire may induce outbreaks of jarrah leaf miner (Mazanec 1981) 1 Warren Region, Department of Conservation and Land Managment, Brain has been disputed by Abbott et al. (1993). For long-term effects St, Manjimup, Western Australia. of fire, changes in forest structure and composition are important Australian Forestry 2004 Vol. 67, No. 2 pp. 101–113 109

elements influencing insect outbreaks (McCullough et al. 1998). Abbott, I., Wills, A. and Burbidge, T. (1999) Reinfestation of Eucalyptus Indeed, for the folivores spruce budworm and western spruce marginata ground coppice by jarrah leafminer after scorch by budworm, there is strong evidence that absence of fire induces autumn or spring fires. Australian Forestry 62, 160–165. outbreaks (Blais 1983; McCune 1983; Anderson et al. 1987; Anderson, L., Carlson, C.E. and Wakimoto, R.H. (1987) Forest fire Bergeron and Harvey 1997; Bergeron and Leduc 1998). In a model frequency and western spruce budworm outbreaks in western for eucalypt decline in eastern Australia, Jurskis and Turner (2002) Montana. Forest Ecology and Management 22, 251–260. suggest that exclusion of low-intensity fire increases eucalypt Attiwill, P.M. (1994) Ecological disturbance and the conservative defoliation. Since fire is an important element within the WA jarrah management of eucalypt forests in Australia. Forest Ecology and forest (Hallam 1975; Attiwill 1994; Ward et al. 2001), it may Management 63, 301–346. follow that fire absence rather than fire presence is an important Bain, J. (2001) Uraba extends its range. Forest Health News No. 110, 1 p. factor in regulating GLS populations. Bergeron, Y. and Harvey, B. (1997) Basing silviculture on natural ecosystem dynamics: an approach applied to the southern boreal mixedwood forest of Quebec. Forest Ecology and Management Conclusion 92, 235–242. Application of GIS and FMIS to the distribution of the 1983 Bergeron, Y. and Leduc, A. (1998) Relationships between change in fire frequency and mortality due to spruce budworm outbreak in the outbreak of GLS in the southern jarrah forest enabled us to examine southern Canadian boreal forest. Journal of Vegetation Science 9, whether forest management practices such as logging and 493–500. prescribed burning influenced these outbreaks. In addition, we Berry, C. (1987) The History, Landscape and Heritage of the Warren investigated the influence of vegetation complexes. Our findings District. Geography Department, University of Western Australia, suggest logging did not promote the outbreak. Analyses of on behalf of the National Trust of Australia (WA) for the shire of vegetation complexes indicate that the outbreak was initiated in Manjimup. marginal jarrah forest prone to drought and inundation. The Blais, J.R. (1983) Trends in the frequency, extent and severity of spruce influence of prescribed burning is less clear. Although our findings budworm outbreaks in eastern Canada. Canadian Journal of Forest suggest that, overall, prescribed burning does not influence GLS Research 13, 539–547. populations, an apparent increase in GLS mortality was observed Brimblecombe, A.R. (1962) Outbreaks of the eucalypt skeletonizer. in areas burnt up to 3 y prior to the outbreak. However, the analysis Queensland Journal of Agricultural Science 19, 209–217. of the role of fire was restricted to CALM-tenured land within the Bradshaw, F.J., Collins, P.M. and McNamara, P.J. (1997) Forest Mapping Manjimup District. Investigation of fire history and frequency in in the South-West of Western Australia. Department of Conservation outbreak areas near Pemberton and Walpole would assist in further and Land Management. understanding the influence of fire. Burrows, N.D. (1997) Predicting canopy scorch height in jarrah forests. CALMScience 2, 267–274. Acknowledgements Burrows, N.D., Ward, B. and Robinson, A.D. (1995) Jarrah forest fire history from stem analysis and anthropological evidence. Australian We thank Paul Davies for mapping; John Vodopier for assistance Forestry 58, 7–16. with the GIS fire analysis; and Southern and Central Forest Region Campbell, K.G. (1962) The biology of Roeselia lugens (Walk.), the gum- personnel for observational ground surveys. Dr Ian Abbott, Peter leaf skeletonizer , with particular reference to the Eucalyptus Keppel and Dr Lachlan McCaw are thanked for comments on camaldulensis Dehn. (river red gum) forests of the Murray Valley earlier manuscripts. region. Proceedings of the Linnean Society of New South Wales 87, 316–338. References Churchward, H.M. (1992) Soils and Landforms of the Manjimup Area of Western Australia. Land Resources Series No. 10, Department Abbott, I. (1987) Review of insect problems in the jarrah forest. of Agriculture Western Australia. Unpublished report. CALM, WA. Churchward, H.M., McArthur, W.M., Sewell, P.L. and Bartle, G.A. Abbott, I. (1990) Insect outbreaks in forests of Western Australia. In: (1988) Landforms and Soils of the South Coast and Hinterland, Watt, A.D., Leather, S.R., Hunter, M.D. and Kidd, N.A.C. (eds) Western Australia. Northcliffe to Manypeaks. Division of Water Population Dynamics of Forest Insects. Intercept Ltd, UK, pp. Resources, Divisional Report 88/1, CSIRO, Australia. 95–103. Cobbinah, J.R., Morgan, F.D. and Douglas, T.J. (1982). Feeding Abbott, I. (1992a) Ecological implications of insect pests in jarrah and responses of the gum leaf skeletoniser Uraba lugens Walker to karri forests. In: Research on the Impact of Forest Management in sugars, amino acids, lipids, sterols, salts, vitamins and certain South-West Western Australia. Department of Conservation and extracts of eucalypt leaves. Journal of the Australian Entomological Land Management Occasional Paper No. 2, pp. 77–97. Society 21, 225–236. Abbott, I. (1992b) Records of Outbreaks of Defoliating Insects in Jarrah Davison, E.M. (1997) Are jarrah (Eucalyptus marginata) trees killed by Forest, South-West Western Australia, from 1960–1990. Department Phytophthora cinnamomi or waterlogging? Australian Forestry 60, of CALM, Technical Report No. 28, 29 pp. 116–124. Abbott, I. and Loneragan, O. (1986) Ecology of Jarrah (Eucalyptus Elliot, H.J. and Bashford, R. (1979). Gum Leaf Skeletonizer Moth. marginata) in the Northern Jarrah Forest of Western Australia. Forestry Commission Tasmania, Forest Pests and Diseases Leaflet Department of Conservation and Land Management, Perth, Western No. 7, 4 pp. Australia. Bulletin No. 1. Farr, J.D. (1985) The performance of Uraba lugens Walker in relation Abbott, I., VanHeurck, P. and Burbidge, T. (1993) Ecology of the pest to nitrogen and phenolics in its food. PhD thesis, University of insect jarrah leafminer (Lepidoptera) in relation to fire and timber Adelaide. harvesting in jarrah forest in Western Australia. Australian Forestry 56, 264–275. 110 Analysis of an outbreak of Uraba in WA

Farr, J.D. (2002) Biology of the gumleaf skeletonizer (Uraba lugens Mazanec, Z. (1981) Environmental conditions influencing the population Walker: Lepidoptera, Noctuidae) in the southern jarrah forest of density of the jarrah leaf-miner, Perthida glyphopa (Lepidoptera: Western Australia. Australian Journal of Entomology 41, 60–69. ). In: Old, K.M., Kile, G.A. and Ohmart, C.P. (eds) Hadlington, P. and Hoschke, F. (1959) Observations on the ecology of Eucalypt Dieback in Forests and Woodlands. CSIRO, Melbourne, the phasmid Ctenomorphodes tessulata (Gray). Proceedings of the pp. 140–146. Linnean Society of New South Wales lxxxiv, Part 2, pp. 146–159. Mazanec, Z. (1989) Jarrah leafminer, an insect pest of jarrah. In: Dell, Hallam, S.J. (1975) Fire and Heath. Australian Institute of Aboriginal B., Havel, J.J. and Malajczuk, N. (eds) The Jarrah Forest. Kluwer Studies, Canberra. Advocate Press Pty Ltd, Melbourne. Academic Publishers, Dordrecht. Chapter 9, pp. 123–131. Harris, J.A. (1972) The effect of flooding on population density of the Mazanec, Z. and Justin, M.J. (1994) Fecundity and oviposition by gum leaf skeletonizer moth Uraba lugens Walk. in Barmah State Perthida glyphopa Common (Lepidoptera: Incurvaridae). Journal Forest. Forests Commission, Victoria, Branch Report (unpublished of the Australian Entomological Society 33, 223–234. report) 25. 11 pp. Mills, J. (1989) The impact of man on the northern jarrah forest from Harris, J.A. (1974) The gumleaf skeletonizer Uraba lugens in Victoria. settlement in 1829 to the Forests Act 1918. In: Dell, B., Havel, J.J. Forests Commission, Victoria. Forestry Technical Papers and Malajczuk, N. (eds) The Jarrah Forest. Kluwer Academic (unpublished report) No. 21, pp. 12–18. Publishers, Dordrecht. Chapter 15, pp. 229–279. Harris, J.A. (1975) The influence of thinning upon defoliation by the Morgan, F.D. and Cobbinah, J.R. (1977) Oviposition and establishment gum leaf skeletonizer in river red gum forests. Forests Commission of Uraba lugens (Walker), the gum leaf skeletonizer. Australian of Victoria, Forestry Technical Paper (unpublished report) No. 22, Forestry 40, 44–55. 15–18. Myers, J.H. (1998) Synchrony in outbreaks of forest lepidoptera: a Harris, J.A., Neuman, F.G. and Ward, B. (1977) An outbreak of the gum possible example of the Moran effect. Ecology 79, 1111–1117. leaf skeletonizer, Uraba lugens Walker, in river red gum forest Sneeuwjagt, R.J. and Peet, G.B. (1985) Forest Fire Behaviour Tables near Barmah. Forests Commission, Victoria, Research Branch for Western Australia. Department of Conservation and Land Report (unpublished) No. 87, 13 pp. Management, Western Australia. Havel, J.J. (1989) Land use conflicts and the emergence of multiple Stoneman, G.L., Rose, P.W. and Borg, H. (1988) Recovery of Forest landuse. In: Dell, B., Havel, J.J. and Malajczuk, N. (eds) The Jarrah Density after Intensive Logging in the Southern Forest of Western Forest. Kluwer Academic Publishers, Dordrecht. Chapter 16, pp. Australia. Department of Conservation and Land Management 281–314. (Western Australia) Technical Report 19, 26 pp. Heberle, G. (1997) Timber harvesting of crown land in the south-west Stoneman, G.L., Bradshaw, F.J. and Christensen, P. (1989) Silviculture. of Western Australia: an historical view with maps. CALMScience In: Dell, B., Havel, J.J. and Malajczuk, N. (eds) The Jarrah Forest. 2, 203–224. Kluwer Academic Publishers, Dordrecht. Chapter 18, pp. 335– Jenkins, C.F.H. and Curry, S.J. (1971) Insect pests of forests. In: Forestry 355. in Western Australia. Forests Department of Western Australia, Strelein, G.J. (1988a) Gum leaf skeletoniser moth, Uraba lugens, in the 3rd edition, pp. 122–129. forests of Western Australia. Australian Forestry 51, 197–204. Jurskis, V. and Turner, J. (2002) Eucalypt dieback in eastern Australia: a Strelein, G.J. (1988b) Site Classification in the Jarrah Forest of Western simple model. Australian Forestry 65, 87–98. Australia. Department of Conservation and Land Management, McCaw, W.L and Burrows, N.D. (1989) Fire management. In: Dell, B., Research Bulletin 2. Havel, J.J. and Malajczuk, N. (eds) The Jarrah Forest. Kluwer Swetnam, T.W. and Lynch, A.M. (1989) A tree ring reconstruction of Academic Publishers, Dordrecht. Chapter 17, pp. 317–334. western spruce budworm history in the southern Rocky Mountains. McCullough, D.G. and Kulman, H.M. (1991) Differences in foliage Forest Science 35, 962–986. quality of young jack pine (Pinus banksiana Lamb.) on burned Volney, W.J.A. (1988) Analysis of historic jack pine budworm outbreaks and clearcut sites: effects on jack pine budworm (Choristoneura in the Prairie provinces of Canada. Canadian Journal of Forest pinus pinus Freeman). Oecologia 87, 135–145. Research 18, 1152–1158. McCullough, D.G., Werner, R.A. and Neuman, D. (1998) Fire and insects Wallace, M.M.H. (1970) Biology of the jarrah leaf miner, Perthida in northern and boreal forest ecosystems of North America. Annual glyphopa Common (Lepidoptera: Incurvariidae). Australian Review of Entomology 43, 107–127. Journal of Zoology 18, 91–104. McCune, B. (1983) Fire frequency reduced two orders of magnitude in Ward, D.J., Lamont, B.B. and Burrows C.L. (2001) Grass trees reveal the Bitterroot Canyons, Montana. Canadian Journal of Forest contrasting fire regimes in eucalypt forest after European settlement Research 13, 212–218. of southwestern Australia. Forest Ecology and Management 150, Mattiske Consulting Pty Ltd (1997) Review and interrogation of floristic 323–329. classifications in the south-west forest region of Western Australia. Wardell-Johnson, G. and Coates, D. (1996) Links to the past: local Prepared for Australian Nature Conservation Agency. RFA 001/ endemism in four species of forest eucalypts in southwestern 012197. Australian Nature Conservation Agency [Perth, WA], Australia. In: Hopper, S.D., Chappill, J.A., Harvey, M.S. and 87 pp. George, A.S. (eds) Gondwanan Heritage: Past, Present and Future Mattiske, E.M. and Havel, J.J. (1998) Regional Forest Agreement of the Western Australian Biota. Surrey Beatty & Sons, Chipping Vegetation Complexes, Pemberton, Western Australia. Cartography Norton, NSW, pp. 137-154. and Geospatial Processing, Department of Conservation and Land Western Australian Department of Agriculture (1948) Annual Report Management, Map. ending 30 June 1948. 29 pp. Australian Forestry 2004 Vol. 67, No. 2 pp. 101–113 111

Appendix I. FMIS interrogation of vegetation types (Mattiske and Havel 1998) on 1983 aerial survey maps of GLS infestation, not including those vegetation types with a zero value for GLS-infested areas (Bkg = background = area outside vegetation mapping = 2 489 216 ha)

Vegetation Name Non-infested area GLS-infested-area code hectares % of (total – Bkg) hectares % of total

A Angove 32 509 0.8 2 662 3.0 BE1 Bevan 1 52 248 1.3 10 316 11.5 BEb Bevan 4 917 0.1 364 0.4 BEy1 Bevan 1 26 299 0.6 1 356 1.5 BL Balingup 14 224 0.3 1 0.0 BU Burnett 6 534 0.2 492 0.6 CA Caldyanup 51 991 1.2 6555 7.3 CC1 Catterick 16 272 0.4 77 0.1 CL1 Corbalup 1 8 538 0.2 2 158 2.4 CM Camballup 17 061 0.4 166 0.2 CO1 Collis 1 2 690 0.1 353 0.4 COb Collis 17 775 0.4 1 456 1.6 COp1 Collis 1 6 310 0.2 68 0.1 COy1 Collis 1 15 609 0.4 3 265 3.6 CP Cattaminup 1 825 0.0 2 096 2.3 CRb Crowea 41 462 1.0 3 944 4.4 CRd Crowea 1 347 0.0 141 0.2 CRy Crowea 21 537 0.5 2 560 2.9 CT Cormint 2 647 0.1 317 0.4 D1 Dwellingup 181 440 4.4 371 0.4 DO Donnelly 1 609 0.0 620 0.7 Dc1 Dempster 1 1 074 0.0 152 0.2 Ds Dempster 1 750 0.0 241 0.3 F Fernley 12 275 0.3 2 372 2.6 GR Grimwade 10 319 0.2 157 0.2 HR Hester 23 393 0.6 81 0.1 Kb Keystone 20 025 0.5 2 334 2.6 Kg Keystone 581 0.0 223 0.3 Kp Keystone 678 0.0 925 1.0 Ky Keystone 11 055 0.3 2 322 2.6 LF Lefroy 12 784 0.3 2 840 3.2 Lp Lindesay 14 252 0.3 68 0.1 MTb Mattaband 9 400 0.2 1 659 1.9 MTp1 Mattaband 1 7 639 0.2 89 0.1 Mty1 Mattaband 1 14 017 0.3 5 217 5.8 PM1 Pemberton 13 569 0.3 2 715 3.0 Pi Pingerup 6 844 0.2 6 650 7.4 Q Quagering 12 849 0.3 1 268 1.4 QN Quindabellup 6 360 0.2 111 0.1 S1 Granite Valleys 18 652 0.4 3 100 3.5 S3 Shallow Valleys 5 277 0.1 263 0.3 TR1 Trent 3 503 0.1 1 463 1.6 V4 Granite Valleys 5 016 0.1 311 0.4 Vh2 Granite Valleys 8 053 0.2 331 0.4 Vh3 Granite Valleys 9 153 0.2 1 607 1.8 WA Warren 6 581 0.2 1 020 1.1 WH1 Wheatley 13 429 0.3 2 759 3.1 WS2 Wishart 2 599 0.1 1 0.0 YN1 Yanmah 14 745 0.4 4 581 5.1 YN2 Yanmah 4 454 0.1 794 0.9 YR Yornup 11 469 0.3 1 0.0 Non-infested GLS vegetation types 3 359 861 80.6 Total 6 655 716 89 943 112 Analysis of an outbreak of Uraba in WA

Appendix II. Comparisons of most prominent vegetation complexes in GLS-infested areas (in descending order of area covered). Developed from Mattiske and Havel (1998), Churchward et al. (1988) and Churchward (1992) (+ indicates presence; Veg = vegetation; J = E. marginata; M = C. calophylla; K = E. diversicolor; Y = E. patens; Rf = C. ficifolia; B = Banksia spp.; Ag = Agonis spp; Me = Melaleuca spp.; H = humid; Ph = per humid; Hh = hyperhumid; Va = valley; Up = upland; Slp = slope; Sw = swamp and valley floor; Sp = southern plain; ToF = tall open forest)

Veg Description Species Veg. cover Climate Land form Comments & other code J M K Y Rf B Ag Me ToF Wood- H Ph Hh Va Up Slp Sw Sp significant spp. land

BE1 Bevan 1 + + + + + + Pi Pingerup + + + + + + Closed heaths CA Caldyanup + + + + + + + + + + Allocasuarina fraseriana Mty1 Mattaband 1 + + + + + + + + + E. guilfoylei, E. jacksoni YN1 Yanmah ++++ + + + + + CRb Crowea + + + + + + + + + A. decussata COy1 Collis 1 + + + + + + + + A. fraseriana, A. decussata S1 Granite Va ++++ ++ + + + + + LF Lefroy + + + + + + + + A. decussata WH1 Wheatley + + + + + + + + + + A. decussata PM1 Pemberton + + + + + + + A. decussata A Angove + + + + + + + + + + CRy Crowea + + + + + + + + F Fernley + + + + + + + + Kb Keystone ++++ + + + + + E. guilfoylei, E. jacksoni, A. decussata Ky Keystone + + + + + + + + CL1 Corbalup 1 + + + + + + + E. rudis, E. decipiens CP Cattaminup + + + + + + + + + Australian Forestry 2004 Vol. 67, No. 2 pp. 101–113 113 . (1988) and Churchward et al rom Churchward ++ ++ ++++ + ed earth; OE = orange YE yellow BFe bog ic duricrust; LaG = lateritic gravel; FeG ferruginous FeQs Soil comparisons of the most prominent landform units in GLS-infested areas (in descending order area covered). Developed f iron pans; HPo = humus podzols; Po podzol; SoY yellow solonetzic soils) codeVeg DescriptionBE1Pi Sa Bevan Gr Pingerup SL Qs Cl + LaD + + LaG + FeG FeQs GDu + YDu + RDu RDuE + RE + OE YE BFe HPo Po SoY + (1992). (+ indicates presence; Sa = sand; Gr gravel; SL sandy loam; Qs quartoze sands and grits; Cl clay; LaD laterit = ferruginous quartoze sandstone; GDu gravelly duplex; YDu yellow RDu red RDuE duplex earth; RE r CAMty1YN1 Caldyanup MattabandCRb YanmahCOy1 CroweaS1 + CollisLF +WH1 Granite Va +PM1 Lefroy Wheatley +A Pemberton +CRy + +F + + Angove Crowea +Kb + +Ky + + Fernley + +CL1 Keystone + +CP + Keystone + + + + Corbalup + Cattaminup + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Appendix III.