australian forestrysvc043.wic023v.server-web.com/pdf/pdf-members/afj/afj... · 2006. 12. 13. ·...

Australian Forestry

Australian Forestry is published by the Institute of Foresters of Australia (IFA) for technical, scientific andprofessional communication relating to forestry in Australia and adjacent geographic regions. The views expressedin this journal are not necessarily those of the IFA. The journal is included in the Register of Refereed Journalsmaintained by the Australian Government Department of Education, Science and Training.

The IFA gratefully acknowledges a grant provided by the Australian Government and State and Territory forestryand forest products agencies through the Forestry and Forest Products Committee to assist in the preparation andproduction of Australian Forestry.

Managing Editor: Dr Brian Turner Production Editor: Mr Alan Brown

Editorial Panel: Dr Stuart Davey Dr Ross Florence Dr Graeme SiemonMr Neil (Curly) Humphreys Dr Ian Bevege Dr John HerbohnDr Grant Wardell-Johnson Dr Humphrey Elliott

Contributions

Contributions to this journal are sought covering any aspect of forest ecology, forest management, forest policyand land use related to Australia and the South Pacific region. Contributions related to the performance ofAustralian tree genera elsewhere in the world are also welcome. Instructions to authors are given inside the backcover of each issue. Contributions should be sent to the Executive Director at the address below.

Journal subscriptions 2007

A$240 including GST within Australia A$270 per year for all other countries

All correspondence relating to subscriptions should be addressed to:

Executive Director, The Institute of Foresters of Australia, PO Box 7002, Yarralumla, ACT 2600, AustraliaPhone: 61 2 6281 3992 Fax: 61 2 6281 4693 Email: [email protected] Web: http://www.forestry.org.au

Cover

The front cover features African mahogany, Khaya senegalensis. This is a promising plantation tree for northernAustralia, tolerant of difficult sites and producing a high-value timber. The story behind the recognition of itspromise, and work to realise its potential, are described by Garth Nikles on pages 68–69 of this volume ofAustralian Forestry. He also selected the photos and prepared the captions.

Photo 1. Butt logs (mainly), from selected trees in 32-y-old, unmanaged stands of African mahogany grown in theDarwin region of the Northern Territory (NT). They were used in a study of sawn timber recovery, timber dryingschedules and wood properties, and for manufacturers’ evaluations of the timber. (Photo by courtesy of Don Reilly.)

Photo 2. Award-winning set of chess table and chairs made from timber from a sample of the logs of Photo 1. Itwon three ‘Australian Furniture of the Year Awards, 2004’ sponsored by the Furniture Industry Association ofAustralia: the Queensland and national awards in the category ‘Excellence in furniture using Australian plantationtimber’; and the Queensland ‘Best-of-the-best’ (across categories) award. (Photo by courtesy of Ray Burgess,Sight Photographics, via the manufacturer, Paragon Furniture, Brisbane.)

Photo 3. A rooted cutting of African mahogany at 10 mo from planting in a clone test established in the Darwinregion of the NT in January 2005. At 12 mo from planting, some trees were nearly 3.5 m high. (Photo by courtesyof Beau Robertson.)

Photo 4. Pods on a graft of African mahogany in a clonal seed orchard planted near Darwin in December 2001.The pods are about 35 mm in diameter some 3 mo after flowering. (Photo by courtesy of Beau Robertson.)

Photo 5. Part of an unmanaged stand of African mahogany aged 25 y derived from natural-stand seed. It wasestablished on back-filled land following surface mining for bauxite at Weipa, Queensland. The central, dominanttree is 48 cm dbhob and one of 36 superior trees selected so far and used in the conservation and geneticimprovement program at Weipa. (Photo by courtesy of Alan Bragg.)

Photo 6. A pruned, 5.5-y-old stand of African mahogany, derived from unimproved seed, planted privately at 5 m ×2.5 m (800 trees ha–1), not thinned, near Bowen, in coastal central Queensland. (Photo by courtesy of GeoffDickinson.)

ISSN 0004-9158

ACN 083 197 586

Australian Forestry

Volume 69 Number 4, 2006

ISSN 0004-9158

The Journal of the Institute of Foresters of Australia

Australian Forestry, Volume 69 Number 4, December 2006

Contents

Page

241 Guest editorial: Forestry education — where are the students, and what should we do?Peter Kanowski

243 Comparative in-ground natural durability of white and black cypress pines(Callitris glaucophylla and C. endlicheri)G.C. Johnson, J.D. Thornton, A.C. Trajstman and L.J. Cookson

248 Effective non-destructive segregation of Eucalyptus grandis logs according to radial andtangential hardnessBill Joe and Ross Dickson

257 Development and testing of seed-crop assessment models for three lowland forest eucalypts inEast Gippsland, VictoriaOwen D. Bassett, Matt D. White and Mark Dacy

270 Paropsine beetles (Coleoptera: Chrysomelidae) in south-eastern Queensland hardwoodplantations: identifying potential pest speciesHelen F. Nahrung

275 An empirical, comparative model of changes in annual water yield associated with pineplantations in southern AustraliaLeon Bren, Patrick Lane and Don McGuire

285 Vulnerability of Tasmanian giant treesWalter Herrmann

299 The effect of tree spacing on the production of flowers in Eucalyptus nitensDean R. Williams, B.M. Potts, W.A. Neilsen and K.R. Joyce

305 First thinning in sub-tropical eucalypt plantations grown for high-value solid-wood products:a reviewR. Geoff B. Smith and Paul Brennan

313 Book review: Forestry in a Global ContextCris Brack

314 Letter to the editor: Forest decline: should we manage or muddle?Vic Jurskis, John Turner and Marcia Lambert

316 Index to Volume 69

241Peter Kanowski

Australian Forestry 2006 Vol. 69 No. 4 pp. 241–242

Guest editorial

Forestry education — where are the students, and what should we do?

Professional foresters continue to be in short supply in theAustralian employment market. Around 2000 professionalforesters were in employment across the sector in Australia in2001, a 10% increase in five years1. Despite this increase,foresters continue to be in short supply in the Australianemployment market. Moreover, fewer students have ben enrollingin undergraduate forestry programs; the number of forestrygraduates declined by 40% over the past decade, to 35 graduatesnationally in 20052. As a result, professional forestry is nowidentified by forestry agencies as their most important area ofskills shortage. As many readers will know, the situation forprofessional foresters is a microcosm of that for employmentin the forest sector more generally, and also for other ruralindustries such as agriculture. It is also typical of the situationin most other countries.

The significance of this national skills shortage has beenrecognised in a number of ways. The House of RepresentativesStanding Committee on Agriculture, Fisheries and Forestry hasconducted a national inquiry into Rural Skills Training andResearch, which has completed hearings but is still to report.The Australian Department of Education, Science and Technology(DEST) commissioned a joint National Association of ForestIndustries/ A3P study of Wood and Paper Products Industry

Skills Shortage in Australia, which recently reported. TheForest and Wood Products Research and DevelopmentCorporation has commissioned a survey of the wood processingindustries to assess the need for education in advanced woodprocessing. This followed the success of such a joint industry–university advanced wood processing program in Canada and theestablishment of similar programs in Chile, New Zealand andSouth Africa.

Each of the forest industries, government, universities and theforestry profession have roles to play in addressing this skillsshortage, which is already imposing considerable direct andopportunity costs on the sector. There are some good startingpoints: a collaborative education model is already well-established and successful at the graduate research-student level,through Cooperative Research Centres such as those for Forestryand Bushfire; the forestry agencies and companies have a longhistory of supporting students through offering summer vacationwork which provides invaluable on-the-job training as well asincome for students; the Institute of Foresters of Australia hasworked with government to develop promotional material forschools, and many individual foresters have spoken to schoolstudents about the profession and its challenges and rewards;

1Australia’s State of the Forests Report, 2003http://www.affa.gov.au/stateoftheforests

2 Data from universities, collated by DPI Queensland

Fieldwork remains an essential and valued part of forestry education.Diverse activities in south-eastern Australia are illustrated in the accom-panying photographs; the scene above is in coastal forest at Kiola, NSW.

Tall native forest in Tasmania

242 Forestry education — where are the students, and what should we do?


better sharing scarce staff resources across universities.• strategic investments by the federal and state governments;

for example, the Australian Government could reduce thefee level or enhance financial assistance for forestrystudents, as it has done for other professions such as teachingor nursing; State governments could co-invest, as some havealready done, with universities in facilities or in internshipprograms. The Australian Government could also useschemes already established, such as those administered byDEST, to support forestry education initiatives.

• coordinated investment by forestry agencies and companiesin scholarships, to create a pool of scholarships nationallyto help attract good students to forestry, and raise the profileof forestry courses and professional opportunities amongstprospective students. To their credit, a number of individualcompanies such as Green Triangle Forest Products and ITChave already recognised and responded to this need, but acoordinated industry-wide strategy is likely to be moreeffective. Maintaining summer employment and internshipopportunities also remains important for students’ educationand income.

• continuing commitment by foresters to promoting theirprofession to students, with support from the IFA, forestryagencies and businesses, and universities, and in partnershipwith organizations such as the NFEAN.

The universities offering forestry degrees are now workingtogether, and with other key players (a working group the Instituteof Foresters has established, NFEAN, industry associations, theForest and Wood Products Research and DevelopmentCorporation, Australian and State government agencies, andforestry companies) to develop a more effective nationalresponse to the professional forester skill shortage.

We would welcome your assistance and ideas in this task, whichis fundamental to the future of the forest sector.

Peter KanowskiProfessor of Forestry, The Australian National UniversityCanberra

Replanting Mt Stromlo, ACT

Radiata pine logs at Tumut, NSW

and a National Forest Education and Awareness Network(NFEAN), which links forest educators in schools, has beenestablished.

The enhanced national awareness of forestry and related skillsshortages offers an opportunity to strengthen collaborationbetween those with an interest in forestry education to betteraddress this problem. Some of the key elements of a coordinatedresponse could include:

• initiatives by universities, such as Southern CrossUniversity’s extension of its undergraduate forestry programto Mt Gambier, or the proposal by the ANU, Southern CrossUniversity, and the Universities of Melbourne and Tasmaniato offer a joint graduate coursework program in forestry.The joint graduate coursework program offers a means of

243G.C. Johnson, J.D. Thornton, A.C. Trajstman and L.J. Cookson


Comparative in-ground natural durability of white and black cypress pines(Callitris glaucophylla and C. endlicheri)

G.C. Johnson1,2, J.D. Thornton3, A.C. Trajstman4 and L.J. Cookson5

1FWPRDC, PO Box 69, World Trade Centre, Victoria 8005, Australia2Email: [email protected]

3John Thornton & Associates, 35 Warana Way, Mt Eliza, Victoria 3930, Australia4CSIRO Mathematical and Information Sciences, Private Bag 33, Clayton South MDC, Victoria 3169, Australia

5Ensis – a joint venture between CSIRO FFP Ltd and Scion Australasia LtdPrivate Bag 10, Clayton South MDC, Victoria 3169, Australia

Revised manuscript received 28 April 2006

Summary

Stakes composed of outer heartwood of Callitris glaucophylla(white cypress pine), Callitris endlicheri (black cypress pine),Corymbia maculata (spotted gum) and Eucalyptus regnans(mountain ash) were exposed to in-ground accelerated testingfor 280 weeks (5.4 y). The time for each stake to reach unservice-ability (loss of 60–75% of cross-section) due to microbial attackwas recorded. The time for Ca. glaucophylla was not significantlydifferent (P > 0.05) from that of Co. maculata. Both these Class 2timbers were significantly different in durability fromCa. endlicheri which, in turn, was significantly different(P < 0.05) from E. regnans, a less durable Class 4 timber. Calibra-tion with long-term field test results for Ca. glaucophylla andCo. maculata suggested that Ca. endlicheri should remainClass 2. The sapwood of both Callitris species showed evidenceof higher natural durability than that of Pinus radiata. Othercomparisons demonstrated that no significant difference wasfound in time to unserviceability of Ca. glaucophylla from twosites (Forbes and Gilgandra, both NSW), and similarly forCa. endlicheri. Also, no significant correlation between time tounserviceability and density was found.

Keywords: durability; decay; simulation; calibration; classification;cypress pine; spotted gum; mountain ash; Callitris endlicheri; Callitrisglaucophylla; Corymbia maculata; Eucalyptus regnans

Introduction

Timber products can deteriorate during service because of decayfungi and insects. The inherent ability of timber species to resistdeterioration is called natural durability, and in Australia fourclasses are recognised: Class 1 timbers are most durable andClass 4 timbers are non-durable. Natural durability ratings arebased upon the outer heartwood, which is often more durablethan inner heartwood, while sapwood is generally considered tobe non-durable. The natural durability ratings of the outerheartwood of Callitris glaucophylla J.Thompson and L.J.Johnson(white cypress pine) and Ca. endlicheri (Parl.) F.M.Bail. (blackcypress pine) have been controversial. Based on experience andexpert opinion, Bootle (1981) considered heartwood of both

species to be of natural durability Class 1. The CSIRO tentativedurability ratings considered only Ca. glaucophylla and thatspecies was rated as Class 2 (Thornton et al. 1983). In laterpublications, Bootle (1983) and Smith et al. (1991) reallocatedCa. endlicheri to Class 2 and retained Ca. glaucophylla asClass 1. The CSIRO in-ground stake test at five sites confirmedthat Ca. glaucophylla was a Class 2 timber (Thornton et al. 1997;Cookson 2005). Callitris endlicheri was not included in that fieldtest; but both species are now listed as Class 2 in-ground timbersin the Australian Standard on natural durability, AS 5604-2005(Standards Australia 2005).

Very little research has been carried out to compare the naturaldurability of these Callitris species. Rudman (1966) recordedthe mass loss caused by a brown rot fungus attacking heartwoodblocks cut from six Ca. endlicheri and ten Ca. glaucophylla trees.From his laboratory study, he concluded that Ca. glaucophyllawas highly resistant to attack while Ca. endlicheri was onlymoderately resistant. More recent research has concentrated onthe production of a natural wood preservative using extracts ofwaste wood from Ca. glaucophylla (Powell et al. 2000).

Here we report a study that compares the in-ground durability todecay of the two Callitris spp. The durability of these species iscompared with that of several ‘yardstick’ timbers whose ratingfor natural durability is known.

Material and methods

Exposure

Unsterile sandy loam soil collected from our natural durabilitytest site at Walpeup (Victoria) was used as a test medium. Astainless steel tank, 1770 mm long, 620 mm wide and 740 mmdeep, was used to contain the test. On the bottom of the tank alayer of gravel, 100 mm thick, was placed to assist in drainageand prevent waterlogging. On top of the gravel was 350 mm ofsoil from the B horizon which, in turn, was topped with 150 mmof soil from the A horizon and humus. The soil was kept moistthroughout the study and the test stakes were installed verticallywith 75 mm of their length buried. The stakes were arranged in a

244 Natural durability of cypress pines


duplicated resolvable row × column design consisting of fiverows by eight columns. Spacing of 100 mm was maintainedbetween stakes in a row and between columns.

The tank was located in the Accelerated Field Simulator (AFS),a room kept at a temperature of about 28°C and 85% relativehumidity.

Test material

Foresters supplied billets, 300 mm long, cut from the bottom ofthe butt log of Ca. glaucophylla and Ca. endlicheri trees thatwould otherwise have been used to produce standard grade or‘run-of-mill’ sawn timbers. Leaves and fruits were also collectedto confirm species identity. The billets came from major areas ofcypress pine production in New South Wales and Queensland.Oversize stakes (40 mm × 40 mm × 150 mm) were cut from theouter heartwood of the green billets, slowly air dried and thencut to a final dimension of 20 mm × 20 mm × 100 mm.

Tables 1 and 2 show the sources and characteristics of the Callitristest material. Heartwood stakes from 16 Ca. glaucophylla and14 Ca. endlicheri trees and, in each case, six geographical sources,were included in the study, along with sapwood stakes from twoCa. glaucophylla and two Ca. endlicheri trees. Heartwood stakesfrom two trees of both Corymbia maculata (Hook.) K.D.Hilland L.A.S.Johnson (spotted gum) and Eucalyptus regnansF.Muell. (mountain ash) were included as typical representativesof Durability Classes 2 and 4 respectively. Sapwood stakes fromtwo trees of Pinus radiata D.Don (radiata pine) were alsoincluded. In all cases the stakes were of clear wood and tworeplicate stakes were included from each tree.

Inspection method

At intervals of eight weeks, each stake was removed from thetank, scraped free of soil and inspected with a scalpel to determinethe depth of decay. The condition of the stake was rated at themost severely decayed cross-section. Scores ranged from 8(sound, no loss of cross-section) to 0 (failed, total loss of cross-section) in integer steps. A stake was considered unserviceablewhen it rated 3 (loss of 60–75% of cross-section) or less. Thorntonet al. (1991) give a full description of the rating system.

Statistical analyses

Wood types (species, heartwood or sapwood) were assessed usingMinitab (Release 12) and S-PLUS (Version 4.5) according to thefollowing criteria: nonparametric Kaplan–Meier survival curves,hazard curves and median time to unserviceability. The survivaland hazard curves provide an overall visual assessment ofperformance of the wood types, and median time to unservice-ability is an index that can be used to characterise the performanceof a wood type. Survival time was the time for a stake to reachunserviceability. As some specimens had not reached a conditionof unserviceability before the end of the study, it was necessaryto take data censoring into account. It should be noted that thesize of samples for all wood types other than Ca. glaucophyllaheartwood and Ca. endlicheri heartwood are rather small (i.e.four replicates, Table 3).

The two-way ANOVA, with replicates as blocks, was used tocompare the time to unserviceability of Ca. glaucophylla heart-wood from Forbes, NSW, with Ca. glaucophylla heartwood from

Table 1. Sources and characteristics of Callitris glaucophylla billets from which outer heartwood stakes were prepared

Geographical source Heartwood Tree No.

Nearest town

State forest/ compartment

Diam.1

(mm) Sapwood thickness2

(mm) Diam.3

(mm)

Air-dry density (kg m–³)

Estim. age4

(y)

Mean rating at 4 y5

01 134 15–17 092 715 ? 02 124 10–16 092 678 ? 03

Western Creek, Qld 130 07–12 110 748 ?

3.2

04 Dalby, Qld Dunmore 148 28–32 089 721 ~70 4.5

05 Eura/34 117 10–15 080 830 >80 06 Eura/34 127 11–14 101 927 >80 07

Gilgandra, NSW Eura/430 130 12–14 106 795 >80

5.2

08 Wittenbra/274 158 16–18 116 690 ~40 09 Wittenbra/274 142 15–17 110 716 ~40 10

Baradine, NSW Wittenbra/274 140 15–21 101 724 ~40

5.2

11 Strahorn 160 21–29 116 739 ~90 12 Strahorn 145 12–15 120 904 ~90 13

Forbes, NSW Strahorn 149 18–24 114 735 ~90

3.8

14 Jimberoo 131 14–17 096 790 ? 15 Jimberoo 115 11–12 095 860 ? 16

Griffith, NSW Jimberoo 125 14–18 104 846 ?

4.5

1Diameter inside bark when green; mean of maximum and minimum diameters 2Minimum and maximum sapwood thickness when green 3Mean of minimum and maximum heartwood diameters when green 4Age of tree estimated by field foresters 5Based on a 0–8 rating scale, where 8 is sound, 3 unserviceable and 0 totally destroyed



Gilgandra, NSW, and Ca. endlicheri heartwood from Forbes withCa. endlicheri heartwood from Gilgandra. The analysis was basedon six replicates from each species and site.

Finally, the effect of density on time to unserviceability ofCa. glaucophylla and Ca. endlicheri heartwood was examinedby Spearman rank correlation analysis (Siegel 1956). At the timethe analyses were carried out, the in-ground test had run for280 weeks (5.4 y).

Results and discussion

Results of statistical analyses of the cypress pines

A major comparison of interest is between Ca. endlicheri heart-wood and Ca. glaucophylla heartwood, and the survival curvesand hazard curves for these wood types are given in Figures 1and 2. A log-rank test and a Wilcoxon test identified the survival

curves for these wood types as significantly different (P < 0.0001).To simplify the graph, the 95% confidence intervals for thesurvival curves were not plotted on Figure 1. However, whenthese were plotted (elsewhere) the intervals did not overlap anypart of the ‘Time to failure’ axis. Figure 1 indicates thatCa. endlicheri heartwood and Ca. glaucophylla heartwood haddifferent survival curves. The curve for Ca. endlicheri heartwoodwas much steeper than for Ca. glaucophylla, indicating thatCa. endlicheri heartwood declined into unserviceability muchquicker than Ca. glaucophylla heartwood. This is furthersupported by the hazard curves in Figure 2, which provideestimates of the failure rates over time. In the case ofCa. glaucophylla heartwood, the hazard curve is relatively flatcompared to the steep ascent for Ca. endlicheri heartwood.Callitris endlicheri heartwood rapidly declined in serviceabilitybetween the end of the third (156 weeks) and fourth (208 weeks)years of study.

Table 2. Sources and characteristics of Callitris endlicheri billets from which outer heartwood stakes were prepared

Geographical source Heartwood

Tree No.

Nearest town

State forest/ compartment

Diam.1

(mm) Sapwood thickness2

(mm) Diam.3

(mm)

Air-dry density (kg m–³)

Estim. age4

(y)

Mean rating at 4 y5

17 Western Creek, Qld 130 20–24 086 673 ? 0.0

18 Dalby, Qld Dunmore? 153 17–23 112 665 ~50 3.5

19 Eura/34 127 11–18 134 812

246 Natural durability of cypress pines


Natural durability ratings

Table 3 presents data on the median time (in weeks) to unservice-ability of the timber types included in this study. The durabilityof Ca. glaucophylla heartwood, despite 11 replicates still beingserviceable, was not significantly different (P > 0.05) fromCo. maculata heartwood. The results from exposure in the AFSof these species are consistent with results obtained after 35 y atfive field test sites. In the field, both species had similar servicelives although Co. maculata was slightly better in performance,and both were classified as Class 2 timbers (Cookson 2005).

On the other hand, Ca. endlicheri was less durable thanCa. glaucophylla, as the time to unserviceability of Ca. endlicheriheartwood stakes was significantly different (P < 0.05) from that

of heartwood from both Ca. glaucophylla and Co. maculata(Class 2 timbers). Most Callitris spp. stakes became unservice-able due to brown rot fungi. Callitris endlicheri was more durablethan the Class 4 timber, E. regnans heartwood (P < 0.05). Arepresentative Class 3 timber was not included in the trial, makingit difficult to know whether to reassign Ca. endlicheri to this class.However, calibration of the AFS trial with the 35-y stake test(Cookson 2005) suggests that Ca. endlicheri should remain aClass 2 timber. The final report of the long-term stake test usedthe mean of the medians from four of the test sites (excluding Innisfail,Queensland) to determine natural durability ratings. The mean valuein the field test for Ca. glaucophylla was 18.6 y (967 weeks). Thetime to median unserviceability for Ca. glaucophylla in the AFSwas 256 weeks, which is 3.8 times faster than in the field (notethat different stake sizes were used in each test). Similarly, theacceleration factor for the other naturally-durable yardstickspecies Co. maculata was more than 4.0, as in the field the meanof the median specimen lives was >16.7 y. Failure of E. regnansin the AFS was faster again at 8.3 times faster than in the field,suggesting that acceleration factors in the AFS may increaseproportionally with reducing natural durability. The averageacceleration factor for the two naturally-durable yardstick specieswas >3.9. If this factor is applied to the median service life in theAFS of 192 weeks for Ca. endlicheri, the service life in the fieldmay have been >749 weeks or >14.4 y. As Class 2 timbers havein-ground service lives of 15–25 y (Standards Australia 2005),for the 50 mm × 50 mm cross-sectioned timbers used in the fieldtest, Ca. endlicheri is likely to fall just within the service liferange for Class 2 timbers. This classification could be assigned,especially if the acceleration factor for Ca. endlicheri lies between3.9 and 8.3, due to its natural durability being lower (but withinthe range possible for Class 2) than for Ca. glaucophylla andCo. maculata.

Another interesting aspect to emerge from this research is thatthe sapwood of both Ca. glaucophylla and Ca. endlicheriappeared to have some natural durability, with median specimenlives of 130 and 156 weeks respectively (Table 3). Sapwood isnormally considered non-durable, so that it was expected thatspecimen lives would be closer to those for P. radiata sapwood(87 weeks) and E. regnans heartwood (35 weeks). However, byapplying the >3.9 times factor used above, the service life in thefield may have been >507 weeks (>9.8 y) and >608 weeks(>11.7 y) for Ca. glaucophylla and Ca. endlicheri respectively. Thiscalibration would suggest natural durability ratings of Class 3 forCallitris sapwood. In practice, however, there have been a numberof early failures (within 2–4 y) of untreated Ca. glaucophyllavineyard posts, due to their high sapwood content, especially atground line when installed small-end down (Johnson unpublisheddata). Also, Rudman (1966) found in a laboratory bioassay thatthe sapwood of Callitris spp. was susceptible to his brown rottest fungus. Therefore, Callitris spp. sapwood should continueto be considered non-durable, although the unexpected result inthe AFS is worthy of further investigation. Sapwood with higher-than-usual natural durability has been reported for Eusideroxylonzwageri Teijsm. & Binnend. (belian) (Wong and Singh 2001),and has been observed for Homalium foetidum (Roxb.) Benth.(malas) (Thornton and Johnson, unpublished data).

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

Pro

ba

bili

ty

180 230 280

Time to failure (weeks)

Figure 1. Nonparametric Kaplan–Meier survival plot for time to failureof the heartwood stakes of Ca. endlicheri (_____) and Ca. glaucophylla (_ _)

Figure 2. Nonparametric empirical hazard function plot for time tofailure, showing failure rate (rate of specimen failure at time t given that ithas not failed before time t) of the heartwood stakes of Ca. endlicheri(_____) and Ca. glaucophylla (_ _)

0.5

0.4

0.3

0.2

0.1

0.0

180 230 280

Rate

Time to failure (weeks)



Provenance and density

The time to unserviceability of Ca. glaucophylla and Ca. endlicheriheartwood, each from two geographic sources, was compared.Forested areas near Forbes and Gilgandra, located 200 km apartin north-western NSW, provided samples from trees of similarage and were therefore selected for the comparisons (Tables 1and 2). Mean ratings for decay after 4 y exposure are providedin Tables 1 and 2. The statistical analysis showed that there wasno significant difference at the 5% level between the twoprovenances within each species.

The Spearman rank correlation coefficient of time to unservice-ability and density of Ca. glaucophylla and Ca. endlicheri was0.315 and 0.248, respectively. Neither coefficient is significantlydifferent from zero. Hence, within the range included in this study,air-dry density does not have an effect on time to unserviceabilityof either Callitris species. This observation contrasts with thesituation reported for the eucalypts, in which greater densityresulted in greater in-ground durability, both between species(Chafe 1989) and within species (Johnson et al. 1996).

In summary, this research shows that Ca. endlicheri heartwoodis not as durable as Ca. glaucophylla heartwood, although thedifference is not sufficient to reduce its rating as a Class 2 timber.This outcome is fortunate, as separation of the two species in themarketplace would be difficult.

Acknowledgements

We would like to thank the staff of State Forests of New SouthWales and Queensland Forestry Research Institute who obtainedthe cypress pine billets essential for this project. Also, we aregrateful to Dr. N-K. Nguyen for assistance with statistical analysis.

ReferencesBootle, K.R. (1981) Building Timbers of New South Wales. Forestry

Commission of New South Wales, Wood Technology and ForestResearch Division, Technical Publication Number 6, 1–8.

Bootle, K.R. (1983). Wood in Australia: Types, Properties and Uses.McGraw–Hill, Sydney, 443 pp.

Chafe, S.C. (1989) Observations on the relationship between wooddurability and density. Journal of the Institute of Wood Science11, 182–185.

Cookson, L.J. (2005) The In-Ground Natural Durability of AustralianTimbers. Project No. PN04.1004. FWPRDC, Melbourne,Australia, 14 pp. http://www.fwprdc.org.au/content/pdfs/PN04.1004.pdf .

Johnson, G.C., Thornton, J.D. and Trajstman, A.C. (1996) Anaccelerated field simulator study of the natural durability ofheartwood from mature and regrowth trees. In: Proceedings, 25thForest Products Research Conference. Melbourne, 18–21November 1996, Paper 1/21, 7 pp.

Powell, M.A., Stephens, L.M., Francis, L. and Kennedy, M.J. (2000)Natural durability transfer from sawmill residues of white cypress(Callitris glaucophylla). 2: Laboratory fungal bioassays. TheInternational Research Group on Wood Preservation, DocumentNo. IRG/WP/00-20204. Stockholm, Sweden.

Rudman, P. (1966) Decay resistance in the cypress pines. AustralianForestry 30, 279–282.

Siegel, S. (1956) Nonparametric Statistics for the Behavioural Sciences.McGraw-Hill, New York.

Smith, W.J., Kynaston, W.T., Cause, M.L. and Grimmett, J.G. (1991)Building Timbers — Properties and Recommendations for theirUse in Queensland. Queensland Forest Service, TechnicalPamphlet No. 1, 117 pp.

Standards Australia (2005) Australian Standard 5604: Timber —Natural Durability Ratings. Standards Australia, Sydney,Australia.

Thornton, J.D., Walters, N.E.M. and Saunders, I.W. (1983) An in-ground natural durability field test of Australian timbers and exoticreference species. I. Progress reports after more than 10 years’exposure. Material und Organismen 18, 27–49.

Thornton, J.D., Johnson, G.C. and Nguyen, N-K. (1991) An in-groundnatural durability test of Australian timbers and exotic referencespecies. VI. Results after approximately 21 years’ exposure.Material und Organismen 26, 145–155.

Thornton, J.D., Johnson, G.C. and Nguyen, N-K. (1997) RevisedNatural Durability Classification. In-Ground Durability Ratingsfor Mature Outer Heartwood. Wall Chart, A3. CSIRO Forestryand Forest Products, Clayton, Australia.

Wong, A.H.H. and Singh, A.P. (2001) The high decay resistance in thesapwood of the naturally durable Malaysian hardwood belian(Eusideroxylon zwageri). The International Research Group onWood Preservation, Document No. IRG/WP/01-10410. Stockholm,Sweden.

248 Segregating Eucalyptus grandis logs according to hardness


Effective non-destructive segregation of Eucalyptus grandis logsaccording to radial and tangential hardness

Bill Joe1,2 and Ross Dickson3

1Forests NSW, PO Box 100, Beecroft, NSW 2119, Australia2Email: [email protected]

3Forests NSW, PO Box 46, Tumut, NSW 2720, Australia

Revised manuscript received 8 May 2006

Summary

Basic density and hardness strongly influence the quality of high-value end products, such as floor boards. Significant variationsin wood properties have been identified both within and amongstlogs of plantation-grown Eucalyptus grandis (flooded gum orrose gum), which typically feature an inner cylinder of corewoodcharacterised by low wood density, hardness and strength. Theseand other defects near the core are not easily detected until thelog is sawn open and thus committed to a particular end-use.Despite this, it is important that only logs with suitably denseand hard wood in the core are processed into high-quality flooring.

This study assessed the utility of a non-destructive tool, thePilodyn, for gauging the radial variation in density and hardnessamongst plantation-grown E. grandis logs prior to sawing.Pilodyn readings were taken at intervals across the ends of 40typical fresh flooded gum logs about 5 m long. A sub-sample of15 logs, representative of the Pilodyn penetration (P-pen)spectrum, was selected for more intensive testing. Janka hardnesstests on air-dried specimens indicated characteristically lowhardness (

249Bill Joe and Ross Dickson


in those properties (Muneri and Raymond 2001; Raymond andMuneri 2001). For processing and value recovery, it is importantto understand patterns of variation in key wood properties acrosslogs. This understanding allows the operational segregation ofwood to maximise grade yield.

Experience has shown that low corewood density and largewithin- and between-tree variability in density and brittleheartare important factors affecting successful and profitableconversion of logs from E. grandis plantations into high qualityend products (Hillis 1978). In the case of E. grandis, the stemdensity is low at the pith, then increases radially and levels offtowards the bark, depending on tree age. The pith-to-bark densitygradients tend to become less higher up the tree. Large differencesin wood properties exist between trees, including those growingin the same general location. Great radial variation in wooddensity (known to be closely related to hardness), particularly inthe lower butt logs, can result in corresponding differences indensity across the width of sawn boards. These reported variances(Bamber et al. 1969; Taylor 1973; Wilkins and Horne 1991) implythere is scope for product segregation and sorting to achieveuniform and acceptable end-products for particular markets andapplications. Sorting criteria such as colour have been found tobe less satisfactory than density as indicators of hardness.

Limited information is available on variation in wood hardnesswithin stems of eucalypts, and no non-destructive testingprocedures are available for assessing hardness in trees and logs.It is known, however, that the basic density of E. dunnii is stronglycorrelated with hardness, particularly for 9-y-old trees (Dicksonet al. 2003).

The overall density of logs can be estimated acoustically, forexample by the Director HM200 tool (e.g. Dickson et al. 2005).

The Pilodyn is another tool often used to estimate wood density(Raymond and MacDonald 1998). The depth of penetration of ablunt spring-loaded striker pin fired into fresh woody materialon the surface of the stem of a standing tree provides an indicationof density. We therefore thought that Pilodyn measurements onthe end faces of logs might be a surrogate measure of timberhardness across the section.

This study sought to explore the use of the Pilodyn to measurethe variation in wood hardness across the cross-sections of matureplantation-grown E. grandis logs. The outcomes were expectedto identify a method and apparatus suitable for sorting the supplyof plantation logs in a mill yard to minimise the effect of outlierson the uniformity of sawn boards. A further goal was to assessthe wood properties, particularly density and hardness, ofplantation-grown E. grandis logs.

Method

Our initial plan was to stratify a sample of 40 logs on the basis ofacoustic velocity (density) as assessed by the Director HM200tool, and then to examine the wood properties of a sub-sample of15 in detail.

Acoustic velocity and Pilodyn 12 J penetration (P-pen) weremeasured on 40 fresh flooded gum logs (about 5 m long) at themill of Notaras and Sons at Grafton. The logs had been harvested

from a plantation at Tuckers Nob State Forest (Compartment 71/73), near Coffs Harbour on the north coast of New South Wales.The plantation was established by APM Forests in 1965/1966on a previously-cleared agricultural site (Dean Kearney, ForestsNSW, pers. comm., 2004), thus making the trees about 38 y oldat the time of harvest.

The P-pen readings were taken on both ends of each log at threepoints within zones equidistant from the pith to bark, representingthe inner, middle and outer zones of the log (Figs 1 and 2).Readings were made where the cut transverse surface wasreasonably flat, taking care to avoid firing the pin into latewoodbands or end checks, which could distort the results.

Due to the close relationship between acoustic velocity in a logand wood stiffness (Dickson et al. 2003), it was initially expectedthat the Director readings would be the best basis for identifyinglogs with hardness values ranging from low to high. However,preliminary analysis showed that the correlation between theDirector and P-pen readings was poor.

Inner ( )

Middle ( )

Outer ( ) 5

9

3

9

1

9

Figure 1. Cross-section of a conceptual log illustrating the three zonesused in measurements. The numbers in brackets are the factors used incalculating weighted averages of the data from cross-sections.

Figure 2. The Pilodyn in use



We concluded that it would be better to select the sub-sample of15 logs from across the spectrum of wood properties using theP-pen readings alone, reasoning that the P-pen readings wouldbe more closely related to wood density, and hence hardness,than the Director readings.

An average P-pen reading was calculated (from 3 readings × 2 ends)for each log, and the logs were classed as having low, medium orhigh readings. Five logs were chosen in each class to provide samplesfor more intensive wood quality measurements in the laboratory.

Test samples, consisting of a short (about 400 mm) centre cant(bark-to-bark longitudinal diametrical section) about 150 mmthick, were cut from both ends of each log. All test samples wereplaced immediately into plastic bags and sealed to minimise anymoisture loss before further processing.

In the laboratory, the corewood diameter was recorded for eachtest sample, along with the heartwood:sapwood ratio. We definedcorewood as a zone in the core that is porous or spongy inappearance, reasonably dry, and of slightly different colour tothe surrounding wood (Fig. 3). This may also contain brittleheart.‘Corewood ratio’ was calculated by dividing the corewooddiameter by the log diameter.

A wedge was then cut from the end of each test sample to providesub-samples for determination of green and basic wood densitiescorresponding with the inner, middle and outer zones of the log.

The remainder of each test sample was sawn into boardscorresponding with each of the three zones of the log and driedto 12% moisture content under controlled conditions (25°C and65% relative humidity). The boards were then machined intoclearwood test specimens for bending strength and stiffness,compression strength (parallel-to-grain) and Janka hardness.

Janka hardness was assessed on both the radial and tangentialfaces of the specimens. All mechanical wood properties weretested in accordance with Mack (1979).

Data analysis

Individual data on Janka hardness, P-pen, wood density (green,basic and air dry), bending strength (MOR), stiffness (MOE) andcompression strength for each of the three zones were collated.An average and a weighted average value were calculated foreach log and end. The weighting factor for averaging was basedon the cross-sectional area of each zone, i.e. 1/9 (inner), 3/9(middle) and 5/9 (outer).

Analysis of variance of individual test data was used to examinevariation in wood properties within and between logs. Simpleregression and correlation were used to explore relationshipsbetween the non-destructive measurements and wood properties.

Results and discussion

Basic density

The basic density increased rapidly with increasing distance fromthe pith (Fig. 4); close to the pith (where we did not have asampling point) the density may have initially decreased (Wilkinsand Horne 1991). The extent of this initial decrease in densityfrom pith outwards has been shown, for 9.5-y-old plantation-grown E. grandis, to be more pronounced with increasing heightin the tree (Wilkins and Horne 1991).

Log measurements

Log measurements are summarised in Table 1. The size of logswas similar in each of the three classes of log within the sub-sample. The average diameter was about 32 cm at the top end(sed) and 35–36 cm at the butt end (led). The proportion ofheartwood was also similar across all three log classes (average73–77%). However, corewood ratio (corewood diameter:logdiameter) for all three log classes was similar only at the top end(average around 0.40). Significant differences were found at thebutt end. The average corewood ratio for the low and mediumP-pen log class was lowest at the butt end (about 0.33); whereasthe average value for the high P-pen log class was 0.40.

Figure 3. A sample with corewood (arrowed) as indicated by the porous-looking zone and lighter-coloured wood

Figure 4. Radial variation in basic density of plantation flooded gumlogs showing mean values (± 1 std error)

300

400

500

600

700

Inner Middle Outer

Zone

Ba

sic

de

nsity (

kg

m–

3)



The average P-pen readings were 29.1, 34.5 and 41.3 mm for thelow, medium and high P-pen log classes, respectively. Thedifferences in mean basic densities between these logs weresignificant, with densities of 564, 524 and 484 kg m–3 for thelow, medium and high P-pen log classes respectively. The basicdensity of the low and medium P-pen log classes was similar tothat quoted for E. grandis from mixed native stands (Bootle1983). However, the high P-pen log class had an average basicdensity that was about 26 kg m–3 less than the published averagevalue.

As alluded to previously, the acoustic velocity readings (Director)along the logs bore little relationship to the Pilodyn readings on

Table 2. Janka hardness on the radial and tangential faces of specimens taken from the inner, middle and outer zones of the butt and top ends of the sub-sample logs. WtAve is the weighted average of the inner, middle and outer values.

Average of butt and top end of log

Radial Janka hardness (kN) Tangential Janka hardness (kN) Class and log 00inumber

Inner Middle Outer Ave WtAve


High P-pen log class 5 1.7 2.9 6.0 3.5 4.5 2.4 3.2 5.5 3.7 4.4 6 2.2 3.6 6.0 3.9 4.8 2.1 3.7 5.3 3.7 4.4 8 1.7 3.1 4.6 3.1 3.8 1.6 3.1 4.0 2.9 3.4 9 2.0 3.3 6.1 3.8 4.7 2.0 2.6 5.9 3.5 4.4 36 1.9 3.3 5.1 3.4 4.1 1.8 2.9 4.9 3.2 3.9 Ave 1.9 3.2 5.6 3.6 4.4 2.0 3.1 5.1 3.4 4.1

Medium P-pen log class 16 2.4 4.5 6.3 4.4 5.3 2.7 4.4 6.5 4.5 5.4 18 2.6 4.9 8.4 5.3 6.6 2.4 3.8 7.4 4.5 5.6 19 2.2 3.7 6.4 4.1 5.0 2.3 3.8 5.4 3.8 4.5 23 2.3 3.8 8.4 4.8 6.2 2.3 3.4 7.8 4.5 5.7 38 2.8 3.2 4.5 3.5 3.9 2.6 3.3 4.2 3.4 3.7 Ave 2.5 4.0 6.8 4.4 5.4 2.4 3.7 6.2 4.1 5.0

Low P-pen log class 4 2.6 4.7 8.0 5.1 6.3 2.7 3.9 7.6 4.7 5.8 27 2.5 5.5 7.1 5.0 6.1 2.3 4.6 6.8 4.6 5.6 33 2.8 6.1 9.1 6.0 7.4 2.9 5.1 8.4 5.5 6.7 35 2.2 4.4 7.3 4.6 5.8 2.9 4.2 6.9 4.7 5.6 37 2.6 6.7 9.4 6.2 7.8 2.4 6.3 9.2 5.9 7.5 Ave 2.5 5.5 8.2 5.4 6.7 2.6 4.8 7.8 5.1 6.2

the ends of the logs. There was no significant difference inDirector velocity amongst the three log classes, even though thevelocity was about 7% higher for the medium P-pen log classthan for the low and high P-pen classes. This suggests that theDirector tool cannot segregate the logs for density (and hardness)as well as the Pilodyn does.

Janka hardness

Table 2 gives the radial and tangential Janka hardness values ofthe wood within the inner, middle and outer zones of each log,averaged over the butt and top sections. The average (arithmetic

Table 1. Average log and wood quality information in the 15 logs of the sub-sample of 38-y-old plantation-grown Eucalyptus grandis

P-pen log class Data set and variable Units

Low Medium High

No. of logs 5 5 5 Diameter, butt end cm 35.9 35.7 34.9 Diameter, top end cm 32.4 32.6 32.2 Corewood ratio, butt end 0.33 0.34 0.43 Corewood ratio, top end 0.40 0.39 0.40 Heartwood, butt end % 77 73 75 Heartwood, top end % 77 75 76 P-pen (log average) mm 29.1 34.5 41.3 P-pen (weighted ave, butt end) mm 24.1 30.7 35.6 P-pen (weighted ave, top end) mm 29.1 32.9 37.7 Basic density kg m–3 564 524 483 Green density kg m–3 1032 981 969 Director velocity km s–1 4.05 4.27 4.02



further refinement is desirable, this result demonstrates thepotential of the Pilodyn to identify logs in the mill yard that donot meet specifications for hardness.

Brittleheart

In this study we concentrated on understanding the effect of thecorewood on sawn board quality and the measurement ofhardness. While referring to corewood, it is also possible thatthe problematic corewood comprised brittleheart, which typicallyoccurs in the central zone of a tree (Yang 2001). Brittleheart ischaracterised by the presence of cell wall deformations and hasparticularly low strength. It is known that E. grandis has apropensity to form brittleheart and this may in part explain someof the shortfall in product grade being experienced by the mill(Hillis et al. 1973). The various tests (e.g. microscopicexamination) for identifying brittleheart were beyond the scopeof this study. Colour changes and a porous or spongy appearanceon the end of logs were the only characteristics we noted thatmay have indicated the presence of brittleheart (Fig. 3). Thepresence of compression failures on dressed sawn timber couldhave been another indication of brittleheart (Jane 1970) but nosuch failures were visible on the clearwood test specimens.

Some assessment of the presence of brittleheart was undertakensubsequent to testing of the bending specimens by examinationof fractures. Nearly all the tested specimens from the inner zone(and a few from the middle zone) were observed to have a brashor ‘carroty’ failure (Fig. 7), which strongly suggests the presenceof brittleheart in the corewood. It could be postulated that thebrash failures were associated with wood decay, but the woodwas visually sound with no evidence of fungal attack.

The likely presence of brittleheart in the corewood is furthersupported by examining the ratio of modulus of rupture (MOR)to density of the tested specimens. In Eucalyptus obliqua, materialwith brittleheart has a lower ratio of MOR to density than normalwood (Yang 2001). The mean MOR:density ratio found in thecurrent study was about 0.11 for specimens from the corewood,while for specimens in the middle and outer zones it was about

mean) and weighted average (WtAve) hardness values across thethree zones are also given in the same table for each log, as wellas for the three classes of log. The trend for hardness to increasefrom pith to bark is noticeable, indicating a systematic variationin wood quality within the stem (Figs 5 and 6). With datacombined for all the logs, the average hardness values on theradial face of the test specimens were 2.3 kN (inner), 4.2 kN(middle) and 6.9 kN (outer); values for the tangential face were2.4 kN (inner), 3.9 kN (middle) and 6.4 kN (outer). This equatesto a three-fold increase in average hardness from pith to bark,which is of practical significance for product utilisation. Thecorrelation between radial and tangential hardness was very good(r = 0.97), with hardness on the radial face being only marginallyhigher than on the tangential face. In service, tangential facehardness would be more relevant to backsawn boards, while radialface hardness is more relevant to quartersawn boards.

The standard deviation for Janka hardness overall was lowerwithin the inner zone (sd = 0.5 kN) than within either the middle(sd = 1.2–1.5 kN) or outer (sd = 1.8 kN) zones. This findingindicates that Janka hardness will nearly always be low (< 3 kN)within the inner zone of every log, regardless of size and age.This inner zone represents about 11% of wood in a log.

The overall average hardness value recorded for the outer woodin this study was up to 1.0 kN less than the 7.5 kN quoted forE. grandis from mixed native hardwood stands (Bootle 1983).However, when considering the logs in their separate classes,the low P-pen log class had an average outer wood hardness thatexceeded the published value, while the medium and high P-penclasses had average hardness values that were less.

In general, the order for both the average and weighted averagehardness values for the P-pen log classes was low>medium>high.Segregating the logs into the three classes on the basis of thePilodyn readings gave a reasonable segregation of Janka hardnessvalues. The overall weighted average Janka hardness on the radialface was 6.7 kN, 5.4 kN and 4.4 kN for low, medium and high P-pen readings, respectively. On the tangential face, the overallweighted average Janka values were 6.2 kN, 5.0 kN and 4.1 kNfor low, medium and high P-pen readings, respectively. While

0.0

2.0

4.0

6.0

8.0

Inner Middle Outer

Zone

Ja

nka h

adnes

s (k

N)

Figure 5. Janka hardness on the radial face of specimens taken fromthe three zones of the butt and top end of the logs. Mean values (± 1standard error) are shown. Note the small variation in hardness in theinner zone.

Figure 6. Janka hardness on the tangential face of specimens takenfrom the three zones of the butt and top end of the logs. Mean values(± 1 standard error) are shown. Note the small variation in hardness inthe inner zone.

0.0

2.0

4.0

6.0

8.0

Ja

nka h

adnes

s (k

N)

Inner Middle Outer

Zone



0.15–0.18. Interestingly, these ratios were consistent with andalmost the same as those found by Yang (2001), despitedifferences in the species and the age of the trees studied.Furthermore, the effect of brittleheart on the MOE:density ratiowas minimal (about 0.04), regardless of the zone from whichspecimens were obtained. This low ratio is also comparable tothat found by Yang (2001). It can be reasonably concluded thatthe 38-y-old plantation E. grandis did contain brittleheart in thecorewood. Ultimately, this finding will affect wood properties.

Bending strength and stiffness

As expected, both the bending strength (as measured by the MOR)and stiffness (as measured by the modulus of elasticity (MOE))increased in the radial direction from the pith to bark (Table 3).Factors for the average radial increase in bending propertiesranged from 2.2 to 2.8 for MOR, and from 1.7 to 2.0 for MOE,depending on log class. The MOR and MOE gradients (frompith to bark) were similar for the low and medium P-pen logs,but steeper for the high P-pen logs. However, regardless of logclass, both the inner and middle MOR and MOE measurementswere lower than those published for E. grandis from mixed nativehardwood stands. This suggests that in most logs the juvenilecore extended beyond one-third of the radius. Bootle (1983)reported average MOR and MOE values for E. grandis of122 MPa and 17 000 MPa, respectively (established from testson wood from older trees). It is clear from the results in Table 3that within the outer zone the strength and stiffness properties ofE. grandis in this study have approached those of mature wood.

In terms of wood utilisation for structural purposes, the lowerMOR and MOE values could have implications for the strengthclassification of this species when grown in plantations. Underthe Australian strength classification system (Standards Australia2000), species are classified into ‘strength groups’ based on theirmean MOR and MOE values. Currently, E. grandis is in strengthgroup SD3 for seasoned (12% moisture content) timber, a grouprequiring minimum values of 110 MPa for MOR and 16 000 MPafor MOE. This study suggests that wood from plantation-grownlogs only in the low and medium P-pen classes would retain thecurrent classification, while that from the high P-pen class mayneed to be downgraded to SD4.

Figure 7. Bending specimens from the inner zone (corewood) showinga brash or carroty type failure (arrowed) that is characteristic ofbrittleheart

Table 3. Bending strength (MOR) and stiffness (MOE) of clearwood specimens taken from the inner, middle and outer zones of the butt and top ends of the sub-sample logs. WtAve is the weighted average of the inner, middle and outer values.

Average of butt and top end of log

Bending strength MOR (MPa) Bending stiffness MOE (MPa) Class and log



High P-pen log class 5 42.4 080.4 120.1 081.0 098.2 09808 14709 19318 14612 16725 6 43.8 092.9 123.0 086.6 104.2 09638 16019 19127 14928 17037 8 52.8 086.4 115.2 084.8 098.7 11308 15133 18749 15063 16717 9 46.3 083.7 140.7 090.2 111.2 08824 13536 18903 13755 15994 36 40.7 076.5 122.8 080.0 098.2 09261 14961 21968 15397 18220 Ave 45.2 084.0 124.4 084.5 102.1 09768 14872 19613 14751 16939

Medium P-pen log class 16 63.2 093.0 156.9 104.4 125.2 13995 16689 22332 17672 19525 18 77.3 095.0 160.2 110.8 129.2 15947 16880 24762 19196 21155 19 53.7 089.6 127.5 090.3 106.7 09641 12266 16955 12954 14579 23 63.2 123.3 156.4 114.3 135.0 11543 19552 22971 18022 20562 38 61.3 093.6 136.4 097.1 113.8 11893 16104 21673 16557 18730 Ave 63.8 098.9 147.5 103.4 122.0 12604 16298 21739 16880 18910

Low P-pen log class 4 56.4 121.6 156.0 111.3 133.4 14301 17248 22845 18131 20030 27 49.3 098.9 158.2 102.1 126.3 11266 15372 22190 16276 18703 33 71.4 111.2 135.1 105.9 120.1 11834 14773 17774 14794 16114 35 82.9 097.1 125.1 101.7 111.0 11922 13743 17383 14349 15563 37 56.5 116.9 136.5 103.3 121.1 10787 16430 19096 15438 17284 Ave 63.3 109.1 142.2 104.9 122.4 12022 15513 19858 15798 17539



The Pilodyn was able to predict not only logs with low hardness(particularly in the core), but also those that comprised corewoodwith low stiffness and strength. Logs segregated into the high,medium and low P-pen classes generally had the MOR and MOEvalues increasing in that order. Essentially, all corewood in alllogs had inferior structural properties.

Although the emphasis of this study was on hardness, the MOEvalues were consistent with the Director acoustic readings forthe three log classes. It is reasonably well established that MOEis positively related to acoustic velocity along a log (Lindstromet al. 2002; Dickson et al. 2003), and so the higher MOE valuesfor the medium P-pen logs in this study (Table 3) are reflected inthe higher Director readings for that log class when comparedwith the other two log classes (Table 1).

Compression strength parallel to grain

The data on compression strength parallel to grain are summarisedin Table 4. There was a regular increase from pith to bark incompression strength, with the average values increasing radiallyfrom the inner to outer zones by factors ranging from 1.6 to 1.8depending on log class. The compression strength values weresimilar for logs in the medium and low P-pen log classes, butwere significantly higher than those in the high P-pen log class.Irrespective of log class, compression strength in the inner andmiddle zones was mostly lower than the published value of66 MPa for E. grandis (Bootle 1983). Again, this suggests thejuvenile core may extend from the pith to beyond one-third ofthe radius of the log. The compression strength in the outer zoneof the logs was comparable to if not better than that published inBootle (1983).

Compression strength is important where timber elements are incompression parallel to grain, such as in poles, posts, props orwall studs. In the past, compression strength was one of the woodproperties considered in combination with bending MOR andMOE to establish the strength group of a species (SAA 1986).Although it is no longer a requirement for the purpose of strengthgrouping species, it was still of interest in this study to be able tobenchmark this property for NSW plantation-grown E. grandis.The minimum average compression strength value formerlyrequired was 61 MPa for E. grandis to be strength grouped asSD3 (SAA 1986). Hence, based on this former requirement andthe weighted average compression strength values attained(Table 4), it would appear that only logs in the low and mediumP-pen classes would have retained the current classification, whilethose in the high P-pen class may have needed to be downgradedto SD4. The results here demonstrate that the Pilodyn wasparticularly successful in identifying logs that had lowcompression strength.

Variation in the quality of corewood amongst trees

As the corewood amongst trees is highly variable (Walker andNakada 1999), values for corewood stiffness, strength, densityand hardness may be expected to vary too. All of the trees usedin this study were expected to have a high proportion of corewood,and indeed the lowest stiffness and strength values were recordedin the inner and middle zones compared to the outer zone.

Figures 5 and 6 indicate that the hardness of the inner zones didnot vary to any great extent amongst logs. This was not the case,however, for the outer zone where there was significant variation

Table 4. Compression strength parallel to grain of clearwood specimens taken from the inner, middle and outer zones of the butt and top end of the sub-sample logs. WtAve is the weighted average of the inner, middle and outer values.

Compression strength (MPa) (average of butt and top end of log) Class and log


High P-pen log class 5 37 47 65 49.6 56.0 6 38 56 65 53.0 59.1 8 36 55 58 49.9 54.9 9 36 53 75 54.8 63.4 36 38 52 69 53.2 60.2 Ave 37.0 52.7 66.7 52.1 58.7

Medium P-pen log class 16 55 58 83 65.3 71.4 18 44 63 85 63.9 73.0 19 41 52 63 52.0 57.0 23 47 63 79 63.0 70.1 38 47 66 77 63.4 70.0 Ave 46.9 60.5 77.3 61.5 68.3

Low P-pen log class 4 49 67 82 66.1 73.4 27 43 60 81 61.2 69.5 33 43 61 69 57.7 63.5 35 47 56 65 56.3 60.3 37 45 68 81 64.5 72.7 Ave 45.5 62.2 75.8 61.2 67.9



amongst logs. Interestingly, only a small portion of logs had outerzones with hardness values greater than those quoted by Bootle(1983) for E. grandis from mixed native hardwood stands. Thismay have been due either to portions of corewood being sampledfrom the outer extremes of the zone of corewood or to anenvironmental effect.

Effects of silviculture on radial wood density and hardness

The silvicultural management of the stand which provided ourtest logs may partially explain why logs were exhibiting hardnessproblems during manufacture. The initial stocking of the standwas about 1087 stems ha–1, typically planted at 3 m × 3 m spacing.The stand was established in 1965/1966 on land that hadpreviously been cleared and used for agricultural production. Highlevels of soil fertility were probably the result of previousintensive land use as a dairy farm. The stand would probablyhave been free-growing for the first 5–7 y at a rate of about25 m3 ha–1 y–1. This rapid growth would have ensured a largetree diameter and, potentially, a large juvenile core.

It is likely that crowding of tree crowns would have progressivelysuppressed growth from between 5 and 10 y of age, so that standgrowth would have fallen to a very low value in the decade priorto thinning when 27 y old. Between thinning and harvest at age38 y, growth may have been 10–15 m3 ha–1 y–1 (Dean Kearney,Forests NSW, pers. comm., 2004). This suppression of growthin the middle half of the rotation would have resulted in the zoneof juvenile wood constituting a large proportion of the cross-sections at harvest. Further, a significant portion of the logdiameter would have exhibited variation in wood density, and alarge portion of the log cross-sectional area would have comprisedtimber of low hardness.

A method suitable for sorting logs according tohardness

The P-pen readings within logs decreased with radial distancefrom the pith. This was consistent with the expected trend thatdensity generally increased radially from the pith outwards(Bamber et al. 1969). Basic densities of samples taken from theinner, middle and outer zones of the log confirmed this trend(Fig. 4). There were noticeable differences in the P-pen readingsbetween the two ends of the log; readings at the top end weregreater than the butt end by about 3 mm. This slightly greater P-pen at the top end was the result of a minor decrease in basicdensity, as would be expected. However, the differences in densityat the top and butt ends were not statistically significant. Thecorrelation coefficient between the weighted average P-penreading and log basic density was 0.91 and 0.75 for the butt endand top end readings, respectively (Fig. 8); that is, the relationshipstill provides only a broad indication of hardness.

In this study, a weighted average P-pen reading gave the bestoverall prediction of hardness of a log. The correlation betweenthe weighted average P-pen reading and weighted average Jankahardness was 0.80 and 0.81, for hardness on the radial andtangential faces respectively (Fig. 9). This indicates that Pilodyntests on the end face of logs may assist processors to identifythose logs with particularly problematic proportions of corewood

Figure 8. Relationship between basic density of the log and theweighted average P-pen at the butt and top ends of logs

Figure 9. Relationship between the weighted average Janka hardnesson the radial and tangential faces, and the weighted average P-penreading. Data for butt and top ends of logs are combined.

with low hardness. In practice, the measurement could beundertaken in a mill yard, requiring only three P-pen readings(inner, middle and outer) and their weighting based on the cross-sectional area that the three zones occupy. A low-costprogrammable calculator could be used for this calculation.

Conclusion

The plantation logs in this study had extensive juvenile corewoodcentred on the pith. There was significant variation in thecorewood properties amongst trees. Logs with high P-penreadings had particularly soft, low density timber. Pilodynmeasurements at intervals from the pith to the periphery acrossthe end of logs were found be effective for segregating logs ofvarying hardness. We conclude that a Pilodyn could be used ina sawmill yard to segregate plantation-grown E. grandis logsinto groups with uniform hardness and structural properties.Ultimately, this may lead to greater value-recovery from forestresources.

Acknowledgements

The authors wish to acknowledge the assistance provided bySpiro Notaras and the late Brinos Notaras of J. Notaras and Sons,

300

350

400

450

500

550

600

650

20.0 25.0 30.0 35.0 40.0 45.0

Weighted average P-pen (mm)

Lo

g b

asic

de

nsity (

kg

m–

3)

Butt

Top

Linear (top)

Linear (butt)

TangentialRadial

Linear (radial)Linear (tangential)

0.0

2.0

4.0

6.0

8.0

10.0

15.0 20.0 25.0 30.0 35.0 40.0 45.0

Weighted average P-pen (mm)

Weig

hte

d a

vera

ge J

anka

hard

ness (

kN

)



Grafton, in undertaking this project and for access to and provisionof test material in their mill yard. We also wish to specially thankLindsay Nicks of Forests NSW, who undertook the complex anddemanding laboratory tests at West Pennant Hills.

This paper is dedicated to the late Brinos Notaras for hisoverwhelming passion for the timber industry and innovation inwood products.

ReferencesBamber, R.K., Floyd, A.G. and Humphreys, F.R. (1969) Wood

properties of flooded gum. Australian Forestry 33, 3–12.

Bootle, K.R. (1983) Wood in Australia — Types, Properties and Uses.McGraw-Hill, Australia, 443 pp.

Dickson, R.L., Raymond, C.A., Joe, W. and Wilkinson, C.A. (2003)Segregation of Eucalyptus dunnii logs using acoustics. ForestEcology and Management 179, 243–251.

Dickson, R., Joe, B., Johnstone, D., Austin, S. and Ribton-Turner, F.(2005) Pre-processing prediction of wood quality in peeler logsgrown in northern New South Wales. Australian Forestry 68,186–191.

Hillis, W.E. (1978) Wood quality and utilization. In: Hillis, W.E. andBrown, A.G. (eds) Eucalypts for Wood Production. CSIRO, GriffinPress Adelaide, South Australia, pp. 259–289.

Hillis, W.E., Hardie, A.D.K. and Ilic, J. (1973) The occurrence ofbrittleheart in Eucalyptus grandis in Zambia. Proc. IUFRODivision 5 Meeting, South Africa. International Union of ForestryResearch Organizations, Vienna, 2, 485–493.

Jane, F.W. (1970) The Structure of Wood. 2nd edition. Adam & CharlesBlack, London. 778 pp.

Lindstrom, H., Harris, P. and Nakada, R. (2002) Methods for measuringstiffness of young trees. Holz als Roh- und Werkstoff 60, 165–174.

Mack, J.J. (1979) Australian Methods for Mechanically Testing SmallClear Specimens of Timber. CSIRO Division of Building ResearchPaper (2nd Series) No. 31, Melbourne.

Muneri, A. and Raymond, C.A. (2001) Non-destructive sampling ofE. globulus and E. nitens for wood properties. II. Fibre lengthand coarseness. Wood Science and Technology 35, 41–56.

Raymond, C.A. and MacDonald, A.C. (1998) Where to shoot yourpilodyn: within tree variation in basic density in plantationEucalyptus globulus and E. nitens in Tasmania. New Forests 15,205–221.

Raymond, C.A. and Muneri, A. (2001) Non-destructive sampling ofEucalyptus globulus and E. nitens for wood properties. I. Basicdensity. Wood Science and Technology 35, 27–39.

SAA (1986) Timber — Classification into Strength Groups. AS 2878-1986. Standards Association of Australia, Sydney.

Standards Australia (2000) Timber — Classification into StrengthGroups. AS/NZS 2878:2000. Standards Australia, Sydney.

Taylor, F.W. (1973) Variations in the anatomical properties of SouthAfrican grown Eucalyptus grandis. Appita 27, 171–178.

Walker, J.C.F. and Nakada, R. (1999) Understanding corewood in somesoftwoods: a selective review on stiffness and acoustics.International Forestry Review 1, 251–259.

Wilkins, A.P. and Horne, R. (1991) Wood-density variation of youngplantation-grown Eucalyptus grandis in response to silviculturaltreatments. Forest Ecology and Management 40, 35–50.

Yang, J.L. (2001) Bending strength properties of regrowth eucalyptbrittleheart. Short Note. Holzforschung 55, 183–184.

257Owen D. Bassett, Matt D. White and Mark Dacy


Development and testing of seed-crop assessment models forthree lowland forest eucalypts in East Gippsland, Victoria

Owen D. Bassett1,2,4, Matt D. White3 and Mark Dacy4

1VicForests, PO Box 124, Benalla, Victoria 3672, Australia2Email: [email protected]

3Department of Sustainability and Environment, Melbourne, Victoria 3001, Australia4Formerly Department of Sustainability and Environment, Melbourne, Victoria 3001, Australia

Revised manuscript received 16 May 2006

Summary

A technique developed for assessing seed crops in Eucalyptusregnans regrowth in Central Gippsland, Victoria, south-easternAustralia, has been adapted for use with three LowlandSclerophyll Forest eucalypts from East Gippsland: E. sieberi,E. globoidea and E. baxteri. The technique was tested in the fieldby comparing the number of capsules assessed on two test treesof each species with the actual number of capsules counted onthem, and by comparing the quantity of seed assessed on seedtreesretained in two harvested coupes with the subsequent seedfallfrom those trees. The results indicate that the technique can beused to assess seed crops on a stand basis with sufficient accuracyfor field research purposes, although the assessment of individualtrees is likely to be less precise.

Keywords: seed crops; assessment; models; prediction; surveys;sampling; Eucalyptus sieberi; Eucalyptus globoidea; Eucalyptus baxteri

Introduction

Regeneration of Australia’s native eucalypt forests followingtimber harvesting frequently relies upon natural or inducedseedfall from retained trees. For example, in East Gippsland,Victoria, VicForests currently relies on the seedtree system toregenerate a large proportion of their harvested coupes, wherebyabout five to seven seedtrees per hectare are retained afterharvesting, and an intense slash-burn is used to induce seedfallfrom them and to prepare a seedbed. Regeneration using thissystem is usually successful (Bassett and White 2003). However,some inefficiencies and failures do occur (Douglas et al. 1989;Delbridge 1998; Bassett and White 2003). Some of these failuresmay be due to the use of only rudimentary visual assessments tojudge whether a particular retained seedtree actually carriesadequate seed to regenerate the harvested area around it.

A major experiment evaluating a range of silvicultural treatments,called the Silvicultural Systems Project (SSP), was establishedin lowland forest at Cabbage Tree Creek, East Gippsland, by theVictorian Department of Natural Resources and Environment(Squire 1990; Squire et al. 1991, 2004, 2006). During theestablishment and monitoring of this experiment, considerableeffort was made to understand the flowering and seed biology of

the major eucalypt species involved (Bassett 2002) and to makequantitative estimates of all inputs, including seed, to thatregeneration (Bassett and Geary in preparation). It becameevident that the rudimentary techniques in local operational usewere not adequate for this purpose. Consequently, seed cropassessment techniques used in eucalypt forests elsewhere werereviewed, and this paper describes how the technique of Harrisonet al. (1990) was adapted and tested for use with E. sieberiL.Johnson (silvertop ash), E. globoidea Blakely (whitestringybark) and E. baxteri (Benth.) Maiden & Blakely ex J.Black(brown stringybark), the three dominant eucalypt species presenton the SSP experimental site at Cabbage Tree Creek.

Various techniques for assessing eucalypt seed crops have beendeveloped in Australia. Loneragan (1979) based an assessmentof E. diversicolor F.Muell. (karri) seed crops on two parametersidentified by Cremer (1971) as being important: crown size andcapsule density. Both of these studies attempted to provideabsolute measures of a number of crown dimensions, havingrecognised that crown size is an important factor in the productionof seed within any tree (Grose 1960). However, crown size hasproved difficult to quantify simply and accurately. For example,Loneragan (1979) required a direct measure of crown widthwhich, in a field situation, may be time consuming to obtain.Alternatively, Cremer (1971) and Lockett (1991) classified crownsize using four visual categories which, although simple, wereconsidered too imprecise for our purposes.

None of these approaches made use of the relationship that isknown to exist between crown size and stem diameter, removingthe need for direct crown measurements (Harrison et al. 1990).Huxley (1931, 1932) was one of the first to recognise theexistence of relationships between the growth rate of onecomponent of an organism and that of a separate component ofthe same organism. Subsequently, Kittredge (1944), Baskerville(1965), Crow (1971), Stewart and Flinn (1981), Leech (1984),Dean and Long (1986) and Applegate et al. (1988) developedstrong regression relationships between either the stem diameteror sapwood area of various conifers, and the biomass ordimensions of their canopy components, allowing an assessor tosimply measure stem diameter or sapwood area in order to obtainan acceptable estimate of crown biomass or size. Attiwill (1962,

258 Seed crop assessment for eucalypts in Gippsland


1966) first demonstrated similar relationships in a non-uniform,broad-leaved genus (Eucalyptus). Later, Stewart et al. (1979)used such relationships to estimate the above-ground biomass ofa eucalypt forest in East Gippsland, and Pook (1984) andWhitford (1991) to estimate leaf area in E. maculata Hook.(spotted gum) and E. marginata Donn ex Smith (jarrah)respectively. More recently, Harrison et al. (1990) applied theconcept to capsule crop assessment in E. regnans F.Muell.(mountain ash) regrowth in Central Gippsland, Victoria by:

1. developing regression equations correlating stem diameter(dbhob) with the actual number of branchlets (BNa) in a treecrown, and using these regressions to obtain an estimate (BNe)of branchlet number by simply measuring dbhob

2. obtaining an estimate (CDe) of average capsule density(capsules per branchlet) by viewing a sample of branchletsthrough a spotting scope and comparing them with standardphotographs representing a series of capsule density classes

3. multiplying the above two estimates (BNe × CDe) to obtain anestimate (CNe) of the number of capsules on the tree.

The accuracy of capsule crop assessments using this techniquewas tested by first estimating CNe as above for a series of testtrees, and then felling the trees and counting the actual numberof capsules (CNa) present. These tests detected over- and under-estimates in CNe with a range of +40% to –14% (average +16%),which was considered acceptable.

To convert such estimates of the number of capsules to estimatesof the number of viable seeds, Harrison et al. (1990) then:

4. estimated the average number of viable seeds per capsule (VS)by collecting a sample of capsules, extracting the seed andtesting it for viability

5. multiplied the three estimated figures together (BNe × CDe × VS).

The accuracy of seed crop assessments using this technique wasnext tested by estimating the number of viable seeds on seedtreesto be retained in a seedtree coupe and monitoring seedfall fromthem for 5 months following a slash-burn. The results indicatedthat the technique could provide acceptably accurate estimatesof seed crops in the field. Hence it was used as the basis fordeveloping seed crop assessment models for the three majoreucalypt species in the SSP experiment at Cabbage Tree Creek.

Methods

Study site

The SSP experiment at Cabbage Tree Creek is located about5 km north-east of Cabbage Tree Creek township, East Gippsland(37°41'S, 148°45'E). Geology, soil, climate, vegetation and sitehistory are described by Stuwe and Mueck (1990). Of the nineeucalypt species present, the dominant species are E. sieberi,E. globoidea and E. baxteri.

Developing the capsule crop assessment models

Stem diameter–branchlet number regressions

Stem diameter–branchlet number regressions were developed foreach of these major species. In each case, up to 22 trees of good

form were selected, assessed for dbhob, crown health anddominance class, and felled. Wherever possible, the trees selectedwere from pre-harvest roading operations on the SSP site, orfrom harvested coupes less than 10 km distant. However, to coverthe full range of diameters likely to be encountered in E. sieberion the SSP site, it was expedient to also sample three large-diameter trees from a coupe 30 km distant. Trees were selectedacross the range of aspects where each species is typically found.Prior to felling, an area around each tree was roughly cleared sothat the crown could be felled onto it for ease of working. Oncefelled, the number of branchlets of 2 cm diameter over-bark (dob)on E. sieberi and E. globoidea, and 1.5 cm dob branchlets onE. baxteri, was tallied with the aid of a diameter gauge set atthose diameters. A branchlet was not chosen unless the gaugefitted its stem somewhere along its length. It had previously beendetermined that branchlets of these sizes carried most of thecapsule crop, and that the number of smaller branchlets containingcapsules, which were not tallied, would be insignificant. Detachedbranchlets were not tallied unless a matching branchlet stub couldbe identified on the felled tree, as branchlets were at timesdislodged from neighbouring trees during felling.

A series of linear, curvilinear, log-linear and log-curvilinearregression equations for each species was calculated from thedbhob and branchlet number data, using MINITAB (Release 7)®and the stepwise least squares procedure (Zar 1984). Data pointsidentified by MINITAB as generating large standard residualsand which could be explained in terms of crown damage weredeleted. For each species, the regression equation with the highestcoefficient of determination (R2) was then chosen for use, aftertesting for statistical validity using the four diagnostic tests below:

1. a heteroscedastic analysis of a scatter-plot of standard residualsand dbhob, to test for normality (Devore 1991)

2. a plot of the BNes estimated from the regressions against theBNas used in constructing the regressions, as a test for bias inBNe (Devore 1991)

3. a comparison of the R2 generated by a simple linear regressionof BNe against BNa, known as the ‘Fit Index’ (Schlaegel 1981;Whitford 1991), with the R2 of the original regression, tofurther test for bias in BNe

4. a comparison of a regression between BNa and (dbhob –dbhobaverage), which effectively centres each regression aboutdbhobaverage and deletes spurious correlation, with the originalregression (Devore 1991).

A 95% prediction interval for an individual observation was thencalculated using MINITAB for each chosen regression.

Crown damage in Eucalyptus sieberi

In selecting trees for use in preparing the above equations,individuals with healthy primary crowns were preferred.However, field expediency rendered it impractical to completelycomply with this aim, and consequently, a number of trees ofeach species were sampled which had suffered some degree ofprimary crown damage and subsequent epicormic recovery. WithE. globoidea and E. baxteri, such trees generally did not detractfrom the models, but damaged E. sieberi trees often appeared asoutliers in the data and were not included in the calculation of

259Owen D. Bassett, Matt D. White and Mark Dacy


the regressions for that species. Consequently, a method ofadjusting the estimated branchlet number, in accordance withthe estimated level of crown damage, was developed forE. sieberi.

Ten trees exhibiting various levels of crown damage were selectedand photographed, the dbhob of each was measured, and branchletnumbers were estimated using the appropriate regressionequation. Each tree was then felled and the actual number ofbranchlets present was tallied. The difference between the actualand the estimated number of branchlets for each tree was thenused as a measure (%) of crown damage for that tree. Sub-sequently, six photographs, representing a range of crown damagepercentages, were selected as standards against which damagedtrees could be compared as needed in later assessments.

Capsule density classes

Capsule density classes were used for each species to assess theaverage number of capsules per branchlet. The classes weredetermined following consideration of the expected natural rangeof capsule densities, the balance between detail (number ofclasses) and ease of use (class width), and the minimum resolutiondetectable by an assessor using a 60 mm spotting scope. Branch-lets were then collected from felled trees or from standing treesusing a .308 calibre rifle and photographed, and the capsulesfrom each were counted. Each branchlet was photographed intwo formats; lying prostrate in ‘horizontal format’, and heldupright in ‘sky format’; the latter enabling a clearer view of thecapsules. This process was continued until each predeterminedcapsule density class was represented by a photograph of abranchlet with a coinciding capsule density.

Testing the accuracy of capsule crop assessments

The accuracy of capsule crop assessments using the above modelswas tested by first estimating the number of capsules present ontwo test trees of each species, and then determining the actualnumber of capsules present on them. Individual test trees withgood form and crown health were selected and the dbhob of eachwas recorded and used to estimate branchlet number from therelationships already established. For each tree, four branchletsat each major point of the compass were then selected byrandomly choosing a primary branch near the tree stem andfollowing it outwards. At each subsequent junction, the branchto follow was randomly selected until a branchlet of theappropriate dob was reached. These 16 branchlets were thenvisually assessed for capsule density using the standardphotographs and a 60 mm spotting scope. The average capsuledensity for the tree was then calculated and multiplied with theestimated branchlet number to obtain an estimate of the numberof capsules on the tree.

The trees were then felled onto an area roughly cleared of under-growth, the actual branchlet numbers were counted, and thecapsules were collected and transported to the laboratory forcounting. For five of the six test trees all of these capsules werecounted manually, whereas for one E. globoidea tree (Test tree 4),from which a very large number of capsules were collected, onlya sample was manually counted. These were then weighed, along

with the remainder of capsules collected from that tree, and thusthe total number of capsules collected from it was determined.

As collection of these capsules proceeded, the branchlets andother heavy material were removed from where the crown hadfallen. A 1 m × 1 m grid was established across this area and, ateach grid point, a 0.06-m2 plot was searched for capsules whichwere considered to have been dislodged from the test tree duringfelling and capsule collection. These were counted and the totalestimated number of dislodged capsules was calculated.

Thus the actual number of capsules (CNa) on each test tree wasobtained by adding the number of collected capsules to thenumber of dislodged capsules. The accuracy of the estimatednumber (CNe) was then tested by determining whether CNa foreach tree was within the corresponding 95% prediction intervalfor CNe of each tree. This was calculated from the variance ofthe capsule density estimates and the 95% prediction intervalsfor the stem diameter–branchlet number regressions.

Testing the accuracy of field seed crop assessments

Following the testing of capsule crop assessments, the accuracyof field seed crop assessments was tested by comparing thenumber of viable seeds estimated to be in the crowns of retainedseedtrees with the subsequent seedfall from them. Seedtree coupeson the SSP site, that were intended for later slash-burning, werechosen for the field tests since most seed in the retained treescould be expected to be induced to fall by the burn (Cremer 1965;Christensen 1971; Luke and McArthur 1977; Whelan 1986).Further, two particular coupes (509/19 and 510/51) were chosensince E. sieberi, E. globoidea and E. baxteri were the dominantspecies in them, with only small numbers of two other speciespresent: E. consideniana Maiden (yertchuck) and, principally ingullies, E. botryoides Smith (southern mahogany).

Prior to harvesting, five seedtrees per hectare on each coupe wereselected on the basis of species, seed crop, spacing, stem form,and crown size and health, and were marked for retention. Twenty-two trees were marked on coupe 509/19 and 23 on coupe 510/51.Trees of all species were then assessed for capsule crops accordingto the schedule in Table 1 and according to the method describedabove for the test trees. In addition, for E. globoidea andE. baxteri, it proved possible to differentiate between older andyounger capsule crops originating from separate floweringseasons, and hence these were assessed separately. For E. sieberi,however, which frequently carries capsules from up to six separateflowering seasons on any one branchlet (Bassett 2002), it wasnot possible to differentiate between capsules from separateseasons an

australian forestrysvc043.wic023v.server-web.com/pdf/pdf-members/afj/afj... · 2006. 12. 13. ·...

Documents