The importance of paddock trees forregional conservation in agriculturallandscapes

A discussion paper for consideration bythe Riverina Highlands RegionalVegetation Committee

Philip GibbonsMiles Boak

New South Wales National Parks and Wildlife ServiceSouthern Directorate

November 2000

2

Summary

Isolated trees and small patches of trees –paddock trees – perform a number ofecosystem services:

� they provide habitat to a range of fauna inthe Riverina Highlands, including severalspecies listed on the NSW ThreatenedSpecies Conservation Act and speciesacknowledged to be in decline throughoutthe wheat-sheep belt;

� these trees contribute to the viability ofwildlife populations in agriculturallandscapes by maintaining connectivitybetween larger patches of remnantvegetation;

� they contribute to salinity mitigation;

� in riparian areas they are important formitigating erosion;

� they help recycle nutrients leached beyondthe pasture root zone;

� they provide shade to stock and are animportant component of the visuallandscape.

Tree-cover within grazed and cultivatedpaddocks in the Riverina Highlands willgradually diminish unless there are changes toland use practices. This is because paddocktrees will senesce with time - often acceleratedby factors associated with dieback – and willnot be replaced because recruitment does nottypically occur in these environments.Estimates of annual mortality among paddocktrees suggest that this resource may be lostwithin 40 years.

Using satellite (SPOT) imagery with a 10mresolution for a 30,000ha sample of theRiverina Highlands around Holbrook, weexamined the contribution to total tree covermade by paddock trees. Woody nativevegetation covered 12% of the study area.Overall, clumps of trees up to 0.1ha(approximately 1-5 trees) represented 9% oftotal tree cover and patches up to 1harepresented 33% of total tree cover.

When assessed in isolation, paddock treesappear to have little value for conservation. Inthe regional context however, paddock treesare important for the conservation of some

vegetation communities. For woodlanddominated by E. blakelyi/E. melliodora, 8%occurred as patches <0.04ha (1-2 trees), 16%as patches <0.1ha (1-5 trees), 42% as patches<0.5ha, 55% as patches <1ha and 79% aspatches <5ha. A similar pattern occurred forcommunities dominated by Eucalyptusmacrorhyncha/E. polyanthemos, E.macrorhyncha/E. goniocalyx and E. albens.

If patches containing 1-5 paddock trees are notprotected and recruited, the communitydominated by E. blakelyi/E. melliodora will bereduced from 6.7% to 5.7% of its predictedpre-1750 distribution within the study area. Ifpatches <1ha are not protected and recruitedthen this community will be reduced to 3.1% ofits predicted pre-1750 distribution. The figuresare similar for woodland dominated by E.macrorhyncha/E. polyanthemos, E.macrorhyncha/E. goniocalyx and E. albens.

Thus, by not protecting and promotingregeneration among paddock trees, it will bedifficult to achieve a net gain in selectedvegetation communities – a stated objective inthe draft Riverina Highlands RegionalVegetation Management Plan.

To reduce the rate at which paddock trees arelost from the landscape requires measures toreduce clearing and dieback. Dieback amongpaddock trees can be reduced through anumber of measures, such as managing stockmovements and minimising herbicide drift.

Measures that can be implemented toencourage recruitment among paddock trees ingrazed and cultivated areas include: periodicreductions to stocking rates within paddocks;temporary fencing around paddock trees; andpermanent fencing around paddock trees.Regeneration events need only be applied to apaddock or stand once every 50-100 years. Thepotential increase to tree-cover that can beachieved across the landscape using thesetechniques greatly exceeds that which can beachieved by planting alone.

3

Introduction

Under present land-use practices, tree coverprovided by isolated and small clumps of trees(paddock trees) in grazed or cultivatedpaddocks within the Riverina Highlands willbe lost over time.

There are three reasons for this:

(1) These trees have a maximum life-span ofaround 500 years (Banks 1997) and willtherefore naturally senesce with time.

(2) Premature mortality among paddock trees -or dieback - often occurs as a result offactors including salinity (Kimber 1981),the interaction between increased nutrientloads and insect attack (Landsberg et al.1990), soil compaction (Yates and Hobbs1997), altered flooding regimes (Reid andLandsberg 2000) and windthrow.

(3) Such trees are cleared for a number ofpurposes such as plantation establishment,firewood collection and some forms ofcultivation.

The loss of paddock trees has been measured at2.5-11% per annum in different agriculturallandscapes in Australia (Reid and Landsberg2000). At this rate, the paddock-tree resourcewill be lost in a period of 9-40 years.

There is a general absence of eucalyptregeneration in cultivated or intensively grazedlandscapes (Wilson 1990), so paddock treesare not replaced when they are lost. Thisscenario has given rise to the description ofpaddock trees as ‘the living dead’.

To date, there has been no holistic approach tomanagement of paddock trees because theircontribution to biodiversity conservation andother ecosystem services is often undervalued.Further, vegetation mapping is undertaken at ascale such that their contribution to total tree-cover in the landscape has not been measured.

In this discussion paper we review the value ofpaddock trees – particularly in terms ofbiodiversity conservation. We also quantify thecontribution of paddock trees to total tree-cover in the Riverina Highlands. Options forappropriate management of this resource arediscussed.

Relevant conservation principles

Two overarching notions are relevant whenassessing the importance of paddock trees forbiodiversity conservation: (1) connectivity; and(2) comprehensiveness, adequacy andrepresentativeness.

Connectivity

Metapopulation theory underpins the basis forcontemporary assessment of populationviability - or extinction risk - for organisms infragmented landscapes (Hanski and Gilpin1991, Lindenmayer and Possingham 1994).This theory is based on observations thatanimal populations generally occur as a set ofsub-populations embedded within a matrix ofsub-optimal habitat. The matrix is importantfor maintenance of the metapopulation (allsub-populations together) because it enablesexchange, or dispersal, of organisms or theirgenetic material between sub-populations. Thisis important for maintaining sufficient geneticdiversity within a sub-population and enablessub-populations to be recolonised in the eventof periodic perturbations, such as drought, fireor disease.

In agricultural landscapes, there is an increasedrisk of extinction among the remaining wildlifepopulations within remnant patches ofvegetation when the matrix, or area betweenthese patches, forms a barrier to movement.For example, there have been observations ofsmall bush birds such as the Hooded Robin,Silvereye and Diamond Firetail disappearingpermanently from remnants in the wheat-sheepbelt after major droughts have caused mortality(Reid 1999). The inability of these patches tobe recolonised has been blamed for thesespecies’ permanent, localised loss (Reid 1999).

However, agricultural landscapes generallycontain some level of tree-cover between thelarger remnants. There are many species withthe capacity to feed and move throughpaddocks via the crowns of these scatteredtrees. McIntyre and Barrett (1992) suggestedthat native vegetation in agriculturallandscapes functions as a continuum for manyorganisms and therefore the distinctionbetween remnant patches and scattered trees isan artificial one. Indeed, Bennett and Ford(1997) found that total tree cover in thelandscape was a significant variableinfluencing bird species richness in northern

4

Victoria - not just the contribution made byvegetation in larger remnants.

Comprehensiveness adequacy andrepresentativeness

The guiding principles used to assess forests inthe Comprehensive Regional Assessment(CRA) process throughout Australia were:comprehensiveness, adequacy andrepresentativeness (Commonwealth ofAustralia 1997). The Native VegetationAdvisory Council considered these principlesrelevant for preparation of RegionalVegetation Management Plans for NSW(Native Vegetation Advisory Council 1999).Among other things, these principles recognisethat effective conservation requires theprotection of sufficient examples of everyvegetation community in every region. In theabsence of better information, vegetationcommunities are considered to be one of thebetter surrogates for biological variation, andtherefore the representation of sufficient, goodexamples of each vegetation community in aconservation strategy is seen to be a positivestep towards biodiversity conservation.

Contributions by paddock trees toconservation

Use of paddock trees by fauna

Most studies of fauna in agriculturallandscapes have focused on larger patches ofremnant vegetation, because they supportlarger populations of most species. However,there is a range of studies that provide us withan indication of the value of paddock trees forbiodiversity.

In Victoria, Loyn and Middleton (1991)detected 24 more bird species in agriculturalareas that contained scattered trees, comparedwith agricultural areas containing no trees. Asimilar result was found in northern Victoriawhere approximately four birds per hectarewere detected in open grassland, 11 per ha inareas with scattered trees, 17 per ha inremnants <5ha and 12 per ha in remnants>30ha (A. Bennett pers. comm. cited in Reidand Landsberg 2000). Birds detected inpaddock trees are not only the common andwidespread species. Hill et al’s (1997)observations of birds using isolated trees inSouth Australia included the Jacky Winter,Brown Treecreeper and Diamond Firetail.

Each of these species is considered to bedeclining in the wheat-sheep belt of NSW(Reid 1999).

Some of the migratory birds that visit theRiverina Highlands seasonally have beenobserved using isolated trees. These includespecies listed on the NSW Threatened SpeciesConservation (TSC) Act, such as the SuperbParrot and Regent Honeyeater. Some of thepreferred tree species for these birds (e.g. E.melliodora) occur predominantly in areas thathave been cleared for agriculture and thereforetend to be the dominant paddock-tree species.

Use of the paddock-matrix by insectivorousbats has been noted in a number of Australianstudies. Law et al. (1999) detected nine speciesin five paddock sites located in the RiverinaHighlands (Chalinolobus morio, C. gouldii,Vespadelus regulus, V. darlingtoni, V.vulturnus, Falsistrellus tasmaniensis,Miniopterus schreibersii, Mormopterusplaniceps, Nyctinomus australis). Two of thesespecies (F. tasmaniensis and M. schreibersii)are listed as Vulnerable in the TSC Act. In astudy from Victoria, an isloated tree containedan extremely large roosting colony of F.tasmaniensis (Parnaby and Cherry 1992).

Law et al. (2000) undertook a study comparingfauna in scattered trees and different-sizedpatches in areas to be converted to plantationsin North East NSW. Two of the arborealspecies recorded (the Brush-tailed Phascogaleand Squirrel Glider) were listed as Vulnerableon the TSC Act. This species occurs in theRiverina Highlands.

Paddock trees also represent a source of woodydebris important for a range of invertebrate andvertebrate species. Reid and Landsberg (2000)quoted figures from an unpublished study thatfound 210 lizards per ha in areas with scatteredtrees, but 1500 lizards per ha in areas wherethe litter and woody debris under scatteredtrees were not removed. Trees on floodplainsare an important source of woody debris forsome native fish species such as the GoldenPerch (McNally et al. 2000).

Resources provided by native paddock treesnot provided by pine plantations

The replacement of paddock trees by pineplantations is an important issue in theRiverina Highlands. Pines provide a number of

5

ecosystem services also provided by paddocktrees (see next section). In terms ofbiodiversity, pine plantations support a numberof native species, particularly where there is adense understorey of blackberries. In studiesundertaken in the Tumut region, pineplantations supported a number of mammaland bird species including the Eastern YellowRobin, Yellow-tailed Black Cockatoo, White-throated Treecreeper, White-browedScrubwren, Brown Antechinus CommonRingtail Possum, Swamp Wallaby, CommonWombat and Brushtail Possum, althoughabundance was often less in pines comparedwith remnant native vegetation (Lindenmayeret al. 1999). However, pines do not providehabitat to many of the typical woodlandspecies, including virtually all of those listed asdeclining by Reid (1999), possibly becausethey lack many of the dietary items that occurin stands of eucalypts (Suckling et al. 1976).For example, pine plantation do not providethe nectar resources required by the migratorybird species, many of which are listed on theTSC Act, and Radiata Pine does not readilydevelop hollows. Also, the value of pines as aconduit for movements by declining birdspecies that occur in the Riverina Highlandshas not been established.

Other ecosystem services provided by paddocktrees

Paddock trees perform a number of otherecosystem functions (Reid and Landsberg2000):

� Isolated and widely-spaced trees have agreater root volume per tree than trees instands and therefore have the potential tointercept and pump considerable volumesof subsurface water, thus reducing salinityrisk.

� Scattered trees reduce wind velocity atground level and therefore slow the spreadof grass fires.

� Trees among pastures recycle nutrientsleached beyond the pasture root zone.

� Trees reduce erosion potential, especiallyin gullies and on stream banks.

� Favourable conditions for some nativeunderstorey species (e.g. Microlaena

stipoides and Poa sieberiana) occur wherethere is some tree-cover.

� The vertebrate fauna supported by paddocktrees help control invertebrate populations.

Paddock trees in the Riverina Highlands

Using SPOT satellite imagery for the studyarea and a Geographical Information System(GIS), we examined the relative contributionof single paddock trees and small clumps oftrees to total native tree-cover in the westernportion of the Riverina Highlands, viz. theHolbrook 1:25,000 mapsheet – an area of30,808 ha (Figure 1). It is important to notethat the quality of native vegetation is notdistinguished between remnants and non-woody vegetation that may occur in the studyarea. Also, non-woody vegetation (e.g.grasslands) is not identified.

#

#

#

Tumut

Albury

Wagga Wagga

Figure 1. The Riverina Highlands VegetationCommittee boundary and the 1:25,000 Holbrookmapsheet used as the study area in this project.

Total native tree-cover for the Holbrookmapsheet was estimated to be 3,768ha. That is,12% of the area contained native tree-cover. Itis important to note that this estimate of tree-cover is higher than the 3,407ha for the samearea as mapped by Maguire et al. (2000) aspart of the Southern Comprehensive RegionalAssessment (CRA). The difference reflects tothe resolution of the satellite imagery used toidentify woody vegetation. The SPOT imageryhas a resolution of 10mx10m and thus is likelyto capture most isolated trees, small patchesand narrow linear strips, such as roadsides(Figure 2).

Figure 2. Comparison between woody vegetation mapped in CRA process (top) with woody vegetationderived from SPOT imagery (bottom). We estimated that tree-cover occurred over 12% of the study area.

7

There were relatively more small patches ofnative vegetation (including isolated trees)than there were larger patches of nativevegetation (Figure 3). These data indicated thatthere were only 18 patches of native vegetation>10ha in the study area, compared with 12,680patches <0.1ha (around 1-5 mature trees perpatch). There was only one patch of nativewoody vegetation with an area >50ha.

Together, single trees and clumps of trees up to0.1ha (around 1-5 mature trees) accounted for9% of the total tree-cover in the study area.Together, clumps of trees up to 1ha accountedfor 33% of the total native tree-cover in thestudy area. Clumps of trees up to 10haaccounted for 57% of total tree cover in thestudy area, with the balance (43%) occurring inpatches >10ha (Figure 4).

1

10

100

1000

10000

<0.04 0.04-0.1 0.1-0.5 0.5-1 1-5 5-10 10-50 >50Patch size (ha)

Coun

t

Figure 3. Counts of different patch-sizes ofwoody vegetation. Patches <0.04ha are likely tocontain 1-2 mature trees. Patches up to 0.1ha arelikely to contain up to 5 mature trees. There wasone patch >50ha.

0102030405060708090

100

<0.04 <0.1 <0.5 <1 <5 <10 <50 <1500

Patch size (ha)

Per c

ent

Figure 4. Cumulative total area of tree-covercontributed by different patches in the studyarea.

Changes to tree-cover in the landscape whenpatches <0.1 ha (1-5 trees) and patches <1haare removed are simulated in Figure 5. Thisillustrates not only the degree to whichvegetation cover is reduced under thesescenarios, but also the decreased level ofconnectivity that results. The mean distancebetween patches across the study areaincreased from 247m to 267m when patches<0.04ha were removed, to 294m when patches<0.1ha were removed and to 451m whenpatches <0.5ha were removed.

We examined the patch-size distribution foreach vegetation type separately by overlayingour tree-cover layer with the pre-1750 layerdeveloped for the Riverina Highlands as partof the Southern Comprehensive RegionalAssessment (Maguire et al. 2000). The patch-size distribution of vegetation in the study areavaried considerably between vegetation types.It is important to note that all areas forvegetation communities reported here arebased on tree-cover within the predicteddistribution of the vegetation types,irrespective of condition.

8

Figure 5. Total tree cover in the landscape (top),tree cover in the landscape with patches <0.1ha(1-5 trees) removed (middle) and tree cover inthe landscape with patches <1ha removed(bottom).

Vegetation dominated by E. dwyeri/Acaciadoratoxylon was predicted to occupy an area of1345ha within the study area. This typedominated the larger patches of remnantvegetation (viz. the drier ridges), with only asmall proportion occurring as scatteredpaddock trees (Figure 6). Only 0.5% of themapped extent of this community occurred inpatches <50ha, with the remaining 99.5% inone patch >50ha.

Vegetation dominated by E. macrorhyncha/E.polyanthemos or E. macrorhyncha/E.goniocalyx had a different patch-sizedistribution. Most of these vegetation typesoccurred in patches between 0.1-0.5 ha(29.5%) and 1-5ha (30.1%) (Figure 6). Overall,8% of these vegetation types occurred inpatches up to 0.04ha (1-2 trees), 17% occurredin patches up to 0.1ha (1-5 trees), 46%occurred in patches up to 0.5ha, 61% inpatches up to 1ha and 91% in patches up to 5ha.

E. dwyeri

0102030405060708090

100

<0.04 0.04-0.1

0.1-0.5

0.5-1 1-5 5-10 10-50 >50

Patch size (ha)

Per c

ent

E. macrorhyncha, E. polyanthemos, E. goniocalyx

010203040506070

<0.04 0.04-0.1 0.1-0.5 0.5-1 1-5 5-10 10-50 >50

Patch size (ha)

Per c

ent

Figure 6. The contribution of different patchsizes to the total area of the communitiesdominated by E. dwyeri/Acacia doratoxylon (top)and E. macrorhyncha/E. goniocalyx, E.macrorhyncha in the study area.

For the E. blakelyi/E. melliodora community,the majority of the mapped extent (1431ha)occurred in patches between 0.1-0.5ha (25%)and patches 1-5ha (24%) (Figure 7). For thiscommunity, 8% occurred as patches <0.04ha(1-2 trees), 16% as patches <0.1ha (1-5 trees),42% in patches <0.5ha, 55% as patches <1ha,

9

79% in patches <5ha, and 84% in patches<10ha.

E. blakeleyi, E. melliodora

0

10

20

30

40

50

60

70

<0.04 0.04-0.1 0.1-0.5 0.5-1 1-5 5-10 10-50 >50

Patch size (ha)

Per c

ent

Figure 7. The contribution of different patchsizes to the total area of the communitydominated by E. blakelyi and/or E. melliodora inthe study area.

A similar pattern occurred for the communitydominated by E. albens, which was predictedto occur over 121ha. Most of this vegetationtype occurred in patches between 0.1-0.5ha(Figure 8). Ten per cent of this vegetation typeoccurred as patches <0.04ha, 21% as patches<0.1ha, 59% as patches <0.5ha and 77% aspatches <1ha.

E. albens

0

10

20

30

40

50

60

70

<0.04 0.04-0.1 0.1-0.5 0.5-1 1-5 5-10 10-50 >50

Patch size (ha)

Per c

ent

Figure 8. The contribution of different patchsizes to the total area of the communitydominated by E. albens in the study area.

A different pattern occurred for riparianwoodland dominated by E. camaldulensis.This vegetation type covered a predicted areaof 83ha. The majority of the remnantvegetation of this type (72%) occurred in fourlinear patches >10ha (Figure 9). Isolated, andsmaller patches of trees up to 0.1ha onlyaccounted for 3% of the extant area of thisvegetation type, with patches <1ha accountingfor 15%.

E. camaldulensis

0

10

20

30

40

50

60

70

<0.04 0.04-0.1 0.1-0.5 0.5-1 1-5 5-10 10-50 >50

Patch size (ha)

Per c

ent

Figure 9. The contribution of different patchsizes to the total area of the communitydominated by E. camaldulensis in the study area.

Error associated with the data

We manually checked the accuracy of ourwoody vegetation layer by visually comparingwoody vegetation on the SPOT image with thepolygons of woody vegetation derived fromthe same image using Arcview™. This wasdone for 10, 1km2 plots located randomly inthe study area. Typically, the presence orabsence of tree canopies was obvious when theSPOT image was enlarged. Overall, our layerfor the 10km2 sample estimated there to be243ha of woody vegetation. After manuallydeleting or adding polygons to tree crownsjudged to be incorrectly classified, weestimated total woody vegetation cover overthis same area to be 225ha. This manualclassification suggested that we overestimatedwoody vegetation cover by 9%. The errorvaried in different parts of the landscape. Theassignment of woody vegetation appeared tobe most accurate on partially cleared hillyterrain. In the continuous areas of nativevegetation within the hilly parts of the studyarea, woody vegetation was underestimated.The procedure often classified croppedpaddocks in the flatter, well-watered parts ofthe landscape as woody vegetation. Therefore,some of the vegetation types on the alluvialparts of the landscape were overestimated. Afurther source of error would be attributable tonon-native woodland cover, e.g. aroundhomesteads, in windbreaks and within thetownship of Holbrook. There were no pineplantations within the study area to contributeto this.

10

Discussion

Individual trees and smaller clumps of trees upto 0.5ha represented 42-59% of remnantwoody vegetation for communities of thelower slopes and plains dominated by the treespecies E. blakelyi, E. blakelyi/E. melliodora,E. albens and E. macrorhyncha/E.polyanthemos/E. goniocalyx. Where theseisolated trees, scattered trees and clumps occurin grazed or cultivated land, this tree-cover willbe progressively lost because of the generalabsence of recruitment in these environments.

Without change to the way these patches andisolated trees are managed will undermine theability to reach vegetation targets nominatedfor these vegetation communities in the DraftRiverina Highlands Regional VegetationManagement Plan.

For example, 6.7% of the predicted pre-1750extent of the community dominated by E.blakelyi/E. melliodora occurred within theHolbrook 1:25,000 mapsheet (this estimateincludes all remnant woody vegetation in thesecommunities as mapped using SPOT imagery,irrespective of condition). Should patchescontaining approximately 1-5 trees (<0.1ha) inthe study area be lost from the landscape, thisvegetation type will be reduced to 5.7% of itspredicted pre-1750 distribution. Shouldpatches up to 1ha be lost from the landscape,this vegetation type will be reduced to 3.1% ofits predicted pre-1750 extent.

For the community dominated by E. albens,the loss of trees in patches <0.1ha will reducethe predicted pre-1750 extent of thiscommunity within the study area from 8.4% to6.6% and the loss of patches <1ha will reducethis vegetation type to 2.0% of is pre-1750distribution.

For the communities dominated by E.macrorhyncha/E. polyanthemos/E. goniocalyx,the loss of trees in patches <0.1ha wouldreduce them from 5.9% to 4.9% of theirpredicted pre-1750 distribution and losingpatches <1ha would reduce them to 2.3% oftheir predicted pre-1750 distribution within thestudy area.

We expect these patterns to generally holdacross the Riverina Highlands, except perhapsfor communities dominated by E.macrorhyncha, which may occur in generally

larger patches than suggested here. Conservingthe isolated trees and smaller patches istherefore critical for achieving the statedobjective of the Draft Riverina HighlandsRegional Vegetation Management Plan, i.e. netvegetation gain in the vegetation communitiesof the lower slopes and plains such as E.blakelyi, E. melliodora and E. albens.

According to the principles recommended bythe Native Vegetation Advisory Committee forRegional Vegetation Committees, the bestexamples of a vegetation type should be givenpriority for protection. The smaller clumps andisolated trees represent a large proportion ofmuch that remains of certain vegetation typesin the Riverina Highlands. Although thestructure and understorey floristics in thesesmaller patches will be highly modified, theyare nevertheless some of the best relativeexamples of remnant vegetation for thesetypes, given the degree to which many of thesevegetation types have been cleared. For thehighly cleared communities, it could be arguedthat all remaining vegetation of these types,regardless of condition, is important for theirconservation.

Rehabilitation efforts for these woodlandcommunities should be focused where there isexisting tree cover, unless there is a specificreason for new vegetation to be established in acertain location (e.g. riparian areas). This isbecause understorey characteristicsapproaching natural conditions can beachieved in a matter of years. Species such asgrasses, forbs and wattles can reach functionalmaturity within 1-10 years. However, trees donot reach functional maturity, that is, they donot supply large quantities of seed, nectar andhollows for decades. For example, hollowssuitable for fauna do not occur in the majorityof eucalypts until they are upward of 150-200years old (Gibbons et al. 2000).

The loss of paddock trees will also removedeep rooted perennial vegetation from a largeproportion of the Riverina Highlands. Asdemonstrated in Figure 5, the removal ofisolated trees and patches <1ha from thelandscape leaves large areas without woodyvegetation, predominantly in the lowlandareas, which are also the areas with potential tobe sites of salt discharge.

11

Further, the loss of patches smaller than 1hawould severely reduce the degree to which thelarger patches of vegetation are connectedacross the landscape (Figure 5), thus increasingfragmentation and potentially reducing theviability of populations of those species thatcan move through habitat provided byscattered trees.

As discussed previously, the principles ofcomprehensiveness, adequacy andrepresentativeness suggest that all vegetationcommunities across the landscape should besampled in a conservation network. This isimportant in the Riverina Highlands becausethe lowland vegetation types dominated byspecies such as E. blakelyi, E. melliodora, E.camaldulensis and E. albens represent habitatfor some species that are generally not found inthe drier vegetation types dominated by speciessuch as E. dwyeri. These include species suchas the Squirrel Glider and Regent Honeyeater.Failure to arrest decline among paddock treeswill undermine the ability to meet theconservation objectives ofcomprehensiveness, adequacy andrepresentativeness in the Riverina Highlands.

The paddock-tree resource, i.e. isolated treesand small clumps of trees in agricultural areas,therefore provide a number of ecosystemservices in the Riverina Highlands. The loss ofthis resource will have a number of negativeconservation implications. Without changes tolanduse that reduce the rate of dieback, limitclearing and promote recruitment amongpaddock trees, then this resource will belargely lost. Such measures are particularlynecessary among the most extensively clearedvegetation types in the region (viz. E. blakelyi/E. melliodora, E. albens and E.macrorhyncha/E. polyanthemos/E. goniocalyxcommunities).

Conclusions

1. Paddock trees (i.e. isolated trees andsmaller patches of trees) perform a numberof ecosystem services.

2. Because they are typically highly modifiedexamples of native vegetation, paddocktrees will always appear to have little valueif assessed in isolation. Their true value forconservation can only be assessed in theregional context.

3. Paddock trees were important for theconservation of those vegetationcommunities within the RiverinaHighlands that were highly cleared. Failureto protect and perpetuate paddock trees inthese vegetation communities willundermine the ability to achieve theconservation objectives ofcomprehensiveness, adequacy andrepresentativeness in this region.

4. The rate of dieback among paddock treesmust be reduced to protect this resource.Reid and Landsberg (2000) provided anumber of recommendations to reducedieback among paddock trees:

� provide buffer plantings around isolatedtrees

� avoid fertilising in the vicinity of trees

� avoid excessive nutrient build-up in thevicinity of paddock trees by managingstock movements

� avoid herbicide drift to paddock trees

� provide conditions for increasing numbersof insect-eating birds and bats.

5. Conditions suitable for eucalyptregeneration must be created for thepaddock-tree resource to be perpetuated.Eucalypt regeneration can be achieved byspelling one paddock at a time at astocking rate sufficiently low enough toachieve regeneration, preferentially timedwith years favourable for recruitment.Eucalypt regeneration has been achievedby temporarily reducing stocking rates byapproximately one third in unfertilizedpastures (data from Reid and Landsberg2000). Alternatively, trees or patches canbe temporarily fenced to prevent grazingor cultivation. Fertilizer input should alsobe reduced where regeneration is targeted.The identification of demonstration areasin which eucalypt regeneration has beenachieved may help define the specificconditions required to obtain regeneration.A regeneration event as seldom as every50-100 years would be sufficient toperpetuate eucalypts provided mortality iskept to a minimum.

12

Acknowledgements

The authors wish to thank Callan Pearson,David Lindenmayer, Mark Sheahan andAnthony Overs for their input to this paper.Comments and support from our colleagues atthe New South Wales National Parks andWildlife Service were also appreciated.Willingness by the Riverina HighlandsRegional Vegetation Management Committeeto consider our suggestions was appreciated.

References

Banks, J. C. G. (1997). Tree ages and ageing inYellow Box. Dargavel, J. Ed. The Comingof Age. Forest Age and Heritage Values.Environment Australia, Canberra, pp 17-28.

Bennett, A. F., & Ford, L. A. (1997). Land use,habitat change and the conservation of birdsin fragmented rural environments: alandscape perspective from the NorthernPlains, Victoria, Australia. PacificConservation Biology, 3: 244-261.

Commonwealth of Australia (1997).Nationally Agreed Criteria for theEstablishment of a Comprehensive,Adequate and Representative ReserveSystem for Forests in Australia. A Reportby the Joint ANZECC/MCFFA NationalForest Policy Statement ImplementationSub-Committee.Commonwealth ofAustralia, Canberra, Australia.

Gibbons, P., Lindenmayer, D. B., Barry, S. C.,& Tanton, M. T. (in press). Hollowformation in eucalypts from temperateforests. Pacific Conservation Biology.

Hanski, I., & Gilpin, M. (1991).Metapopulation dynamics: brief history andconceptual domain. Biological Journal ofthe Linnaean Society, 42: 3-16.

Hill, B. M., Laan, T., & Paton, D. C. (1997).Use of Isolated Trees by Native Fauna inthe Mt Lofty Ranges, Department ofZoology, University of Adelaide.

Kimber, P. C. (1981). Eucalypt diebacks in thesouth west of Western Australia. Old, K.M., Kile, G. A., & Ohmart, C. P. Eds.Eucalypt Dieback in Forests andWoodlands. CSIRO, Melbourne, pp 37-43.

Landsberg, J., Morse, J., & Khanna, P. (1990).Tree dieback and insect dynamics in

remnants of native woodlands on farms.Proceedings of the Ecological Society ofAustralia, 16: 149-165.

Law, B. S., Anderson, J., & Chidel, M. (1999).Bat communities in a fragmented forestlandscape on the south-west slopes of NewSouth Wales, Australia. BiologicalConservation, 88: 333-345.

Law, B. S., Chidel, M. and Turner, G. (2000).The use by wildlife of paddock trees infarmland. Pacific Conservation Biology, 6:130-143.

Lindenmayer, D. B., & Possingham, H. P.(1994). The Risk of Extinction. RankingManagement Options for Leadbeater'sPossum Using Population ViabilityAnalysis. Centre for Resource andEnvironmental Studies, The AustralianNational Unversity, Canberra.

Lindenmayer, D.B., Cunningham, R.B. andPope, M.L. 1999. A large-scale‘experiment’ to examine the effects oflandscape context and habitat fragmentationon mammals. Biological Conservation, 88:387-403.

Loyn, R. H. (1981). Middleton, W.G.D. Old,K. M., Kile, G. A., & Ohmart, C. P. Eds.Eucalypt Dieback in Forests andWoodlands. CSIRO, Melbource, pp 95-111.

Maguire, O., Thomas, V., & Hunter, S. (2000).Southern CRA/Riverina HighlandsVegetation Mapping Extension. A ReportUndertaken by the National Parks andWildlife Service Southern ComprehensiveRegional Assessment Unit for the RiverinaHighlands Regional VegetationManagement Committee.

McIntyre, S., & Barrett, G. W. (1992). Habitatvariegation: an alternative to fragmentation.Conservation Biology, 6: 146-147.

McNally, R., Parkinson, A., Horrocks, G.,Conole, L., Young, M., Tzaros, C., Koehn,J., Lieschke, J., & Nicol, S. (2000).Ecological Significance of Coarse WoodyDebris on South-Eastern AustralianFloodplains. Report Number R7007,Murray Darling Basin Commission.

Native Vegetation Advisory Council of NSW(1999). Biodiversity principles for regional

13

vegetation planning. Unpublished report.

Parnaby, H., & Cherry, K. (1992). A TrappingStudy of the Bat Fauna of Mountain AshPlantations and Old Growth Forests of theStrzelecki Ranges, Victoria.Department ofConservation and Natural Resources,Victoria.

Reid, J. W. (1999). Threatened and DecliningBirds in the New South Wales Sheep-Wheat Belt: 1. Diagnosis, Characteristicsand Management. Consultancy Report toNSW National Parks and WildlifeServiceCSIRO Wildlife and Ecology,Canberra.

Reid, N., & Landsberg, J. (1999). Tree declinein agricultural landscapes: what we stand tolose. Hobbs, R. J., & Yates, C. J. Eds.Temperate Eucalypt Woodlands inAustralia: Biology, Conservation,Management and Restoration. SurreyBeatty and Sons, Chipping Norton, pp 127-166.

Suckling, G.C., Backen, E., Heislers, A. andNeumann, F.G. 1976. The flora and faunaof Radiata Pine plantations in north-easternVictoria. Forest Commission of VictoriaBulletin No. 24.

Wilson, A. D. (1990). The effect of grazing onAustralian ecosystems. Proceedings of theEcological Society of Australia, 16 : 235-244.

Yates, C. J., & Hobbs, R. J. (1997). Temperateeucalypt woodlands: a review of theirstatus, processes threatening theirpersistence and techniques for restoration.Australian Journal of Botany, 45: 949-973.