Effects of Javan Rusa Deer (Cervus timorensis) on Native Plant Species in the Jibbon-Bundeena Area, Royal National Park, New South Wales David Keith' and Belinda Pellow^ 'NSW Department ofEnvironment and Conservation, PO Box 1967, Hurstville 2220 NSW. ^Janet Cosh Herbarium, University ofWoUongong, Wollongong 2522 NSW. Keith, D. and Pellow, B. (2005). Effects ofJavan rusa deer{Cervus timorensis) onnativeplant species inthe Jibbon-Bundeena area, Royal National Park,New SouthWales. Proceedings oftheLinnean SocietyofNewSouth Wales 126, 99-110. Areconnaissance survey and exclosure experiment were carriedoutto examine the effects ofJavanrusa deeronnativefloraandvegetationinRoyalNationalParkonthesouthernoutskirtsofSydney,Australia. Of 78 native plant species examinedduring the survey, onlynine showedno evidence ofvertebrate herbivory or physical damage and the majority ofthese plants were ferns and sedges. The other 69 species showed effectsthatincludeddefoliation(youngand/oroldleaves),removalofshoots,bark-stripping,stembreakages and destruction or consumption ofreproductive material. These effects varied in severity between species andfromplacetoplace,andwereinferredtohavebeencausedbydeerbasedonthelocalabundanceofdeer droppings, footprints and the scarcity ofother vertebrate herbivores in the area. The survey also revealed localised soil erosion associated with high densities of deer footprints and droppings. An unreplicated exclosure experiment showed that planted saplings ofSyzygiumpaniculatum, a threatened rainforest tree, suffered major defoliation, bark stripping, stem breakages and some mortality when exposed to deer for several months. Many of the surviving plants showed signs of recovery when deer were subsequently excluded, although ftill recoveryoftheirleafcanopies couldtake several seasons. The observedeffects on vegetation and individual plant species are consistent with studies on several other deer species in a range ofecosystems overseas. Amodel ofthe effects ofdeer herbivory based on plant life-history suggests that curtailment ofseedproduction and seedling recruitment are likely to be the majorimpacts ofdeeronplant populationviability. Reductions innetgrowthandsurvivalofestablishedplantsandpossiblypost-dispersal predation ofseeds are less likely to be significant influences. Manuscript received31 October2003, acceptedforpublication 18August2004. KEYWORDS: browsing, Cervus timorensis, deer, endangered ecological communities, feral animals, fire, grazing, herbivory, plant demography, Syzygiumpaniculatum, threatened species, threatening process. INTRODUCTION deer {Dama dama), which were introduced some 20 years earlier and likely to be the species reported as Javan rusa deer {Cervus timorensis) is one ofsix widespread in the Park in 1914 (Anon. 1914). High deer species that have established wild populations densities of Rusa deer are now regularly observed in Australia (Moriarty 2004a). Rusa deer were in the vicinity of Bundeena, Grays Point, Garie-Era introduced into Royal National Park in 1906 by the and various sites along the Hacking River valley. Park Trust for exhibition purposes (NSW National The deerpopulation is likely to have fluctuated since Parks and Wildlife Service 2002). The seven its introduction, although densities appear to have introducedanimalswere initiallykeptwithinafenced increased markedly since bushfires in January 1994, enclosure at 'DeerTark' on the Hacking River near which burnt more than 90% of the Royal National Warumbul. The deer soon escaped and established Park(pers. obs.).Afl;erthefires, thepopulationwithin a wild population that has persisted in the Park and Royal National Park was estimated to include less adjacent areas to the present day. Hamilton (1981) than 500 individuals. Quantitative surveys carriedout suggested that rusa deer essentially replaced fallow between 1999 and 2001 indicate that the population RUSA DEER IN ROYAL NATIONAL PARK, SYDNEY METHODS increased from 2500 to 2900 individuals during that Study area time (Moriaty unpubl. data, NSW National Parks and Jibbon-Bundeena is a 300 ha area located in the Wildlife Service 2002). far north-east comer of Royal National Park on the Rusa deer grow to 1 m tall at the shoulder and shores of Port Hacking on the southern outskirts of weigh 100-160 kg. Males develop large antlers but Sydney (lat. 34° 06'S, long. 151° 09'E). The bedrock females do not. The deer are active nocturnally, ofHawkesbury sandstone is overlain in some parts by resting in dense native vegetation by day. They are podsolised marine and aeolian sand dunes that may m dietary opportunists, apparently preferring grass but exceed 10 in depth. The dune crests and slopes are browsing opportunistically on the buds, shoots and freely drained, although the swales and flats may be leaves of woody plants and herbs (Bentley 1979). periodically waterlogged. The exposed sandstones Recentdietarydata from RoyalNational Park indicate carry shallow to skeletal yellow earths with variable that food sources vary with season and location. In drainage characteristics. native vegetation remote from settled areas, 80% of Theareaincludesamosaicofrainforests,eucalypt their average summer diet consists ofnative browse, forests,heathlandsandwetlands. Littoral rainforeston with grasses making up the remaining 20% (Moriaty the hind dunes ofJibbon and Bonnie Vale beaches is 2004b). In other seasons, the grass component dominatedby CupaniosisanacardioideswithAcmena becomes negligible inthese areas. However, the grass smithii, Glochidion ferdinandi suhsp.ferdinandi and component varies seasonally from 40% to 70% in Banksia integrifolia subsp. integrifolia. It forms part animals living close to cleared grassy areas (Moriarty of an Endangered Ecological Community (Littoral 2004b). Rainforest in NSW) under the NSW Threatened Concerns about impacts on native vegetation SpeciesConservationAct1995andincludesSyzygium have resulted in a recent preliminary determination paniculatum, which is currently listed as aVulnerable offeral deer as a Key Threatening Process under the species. The eucalypt forests are found principally on NSW Threatened Species Conservation Act 1995. A deep sands. On dune slopes they are dominated by recent survey of rangers employed by Rural Lands Corymbia gummifera and Angophora costata with Protection Boards in NSW identified deer as the an understorey ofsclerophyllous shmbs and bracken. most important emerging pest animal threat (West On sandy flats. Eucalyptus botryoides and A. costata and Saunders 2003). More than 40%) ofrespondents form a taller forest with an understorey that includes indicated that there had been a moderate to high a mixture of mesophyllous and sclerophyllous increase in the distribution and abundance of wild shrubs and herbaceous plants. Heathlands occur on deer in their area. In another recent survey more than deep sands and on the exposed sandstone plateau. 80% of land managers reported browsing of native Floristic compositionvaries between these substrates, plants and agricultural crops as an impact of feral although both communities comprise a dense to deer (Moriarty 2004a). Despite numerous detailed open cover of sclerophyll shmbs interspersed with studies on the impacts of deer on native vegetation sclerophyll sedges. Wetlands are restricted to lagoons in other countries (e.g. Okuda and Nakane 1990, in dune swales and swampy sand flats. They are Veblen et al. 1992, Kay 1993, Jane 1994, Anderson mainly herbaceous communities dominated by and Katz 1993, Mladenoff and Steams 1993, Khan Baumea juncea with B. articulata, Leptocarpus et al. 1994, Augustine and Frelich 1998, Akasi and tenax and other sedges. They are currently listed Nakashisuka 1999, McSheaand Rappole 2000, Fuller as an Endangered Ecological Community (Sydney NSW and Gill 2001, Coomes et al. 2003, Rooney et al. Freshwater Wetlands) under the Threatened 2004), there are relatively few published data from Species Consen^ationAct 1995. Australia on this subject. In Royal National Park and The township of Bundeena occupies about 100 adjacent urban areas, the high densities ofdeer, their ha within the study area, covering sand dunes, sand free movementin open areas andtheirexclusion from flats and sandstone ridges. The majority ofthe town some properties provide opportunities to examine the precinct comprises suburban dwellings on blocks effects ofdeer on native vegetation. The aims ofthis <0.1 ha, some of which are fenced to exclude deer study were to document the various impacts ofdeer from gardens. There are substantial areas planted on native vegetation, compile a list of native plant with exotic grass on road verges and in parks and species affected by deer and to quantify the effects yards. Native plants, relics ofthe original vegetation, of deer on a threatened plant species, Syzygium persist as solitary individuals orin clumps throughout paniculatitm. the town. At least some ofthe deer in the area spend 100 Proc. Linn. Soc. N.S.W., 126, 2005 D. KEITHAND B. PELLOW some of their nocturnal foraging time within the months, the gatewas closed. Deergained intermittent suburban Umits of Bundeena. They graze on grassy access to the yard after that time until the perimeter areas and are fed vegetable matter by a few local fence was raised to aheight of 1.5 m. As a qualitative residents. They also browse on a variety offorbs and control treatment, five plants were observed in an woody plants, including various vegetables, exotic adjacent yard that was maintained to exclude deer ornamental plants and local native plants. throughout the duration of the experiment. Brush- tailed possums, the only other vertebrate herbivore Survey on the site, were observed in both the treatment and Extensive reconnaissance was undertaken control yards. Effects on Syzygiumpaniculatum were throughout the native vegetation and the suburban recorded six months after deer initially gained access areato observe deerbehaviourandmovement, record as follows: >75% foliage and branchlets consumed; plant species consumed or damaged by deer and <75% foliage and branchlets consumed; main stem effects ofdeeronthe structure ofvegetation andsoils. broken within 30 cm of base; bark removed; or Observations were made opportunistically between foliage and branches unaffected. One year after deer 1999 and 2003. Observations on native plants were had been excluded, plants were recorded as either confined to bushland areas within 1 km ofBundeena dead, alive and growing new shoots or alive and not township, but at least 20 m beyond the suburban/ growing new shoots. bushland boundary. The effects were classified into the following categories: young foliage and shoots consumed; young and old foliage and shoots RESULTS consumed; bark stripped; woody stems broken; seedlings uprooted; inflorescences damaged or Vegetation structure and soils consumed; and unaffected. Qualitative observations Reconnaissance of bushland suggested that on damage to vegetation structure and soils were also structural irregularities in the vegetation, including recorded during the reconnaissance ofbushland. tracks and open areas with low densities of woody To examine the effects on soils where deer plants, were associated with deer access and activity. activity was concentrated, three sites were identified Footprints indicated that a number of tracks in the where deerwere fedby local residents. All three sites area were used by both humans and deer. However, were on deep, unconsolidated sand dune soils on the numerous tracks in the area showed no evidence of interface between eucalypt woodland and the eastern regular human usage and generally dissipated after edge ofthe suburban area. They were separated from some distance orterminatedinopenareas inthe forest one another by distances of 200 metres, and not understorey or heathland. These open areas varied includedinthereconnaissance survey, which sampled from2-50 m- andhadconspicuously lowerdensities bushland away from the suburban interface. Soils and of shrubs and groundcover than the surrounding vegetation at these sites were observed between 1999 vegetation (Fig la). Resting deer were disturbed and 2003, and qualitative descriptions of vegetation from some of these sites during reconnaissance and cover and soil level were compared to those of the presence of droppings, footprints and/or dead equivalent sites on the bushland/suburban interface remains of deer at most sites suggested that they m approximately 50 distant from each feeding site. function as deer encampments. There was little The equivalentsiteswerealso onsandandhadsimilar evidence of significant surface erosion on the sand vegetation (disturbed heathy woodland) and a similar dunes within heathlands or woodlands, except along management history to the feeding sites. the tracks (Fig. lb), which were depressed below the general soil surface. However, the wetland soils were Exclosure experiment exposed, compacted and deformed on the surface Ninety-three individuals of Syzygium by deer footprints. These areas of bare compacted paniculatum were planted in a 0.1 ha yard, grassed soil were most commonly encountered around the with Pennisetum clandestimim and enclosed within margins ofwetlands. 1 m high wire fencing. The plants were 1.0 - 1.4 m The three sites where deer were fed by local tall and were watered for three months after planting. residents on the bushland/suburban interface had a Afteronemonth, allplantswereinahealthycondition low open cover of grasses, herbs and shrubs when and growing new shoots when the yard was opened, observations began in 1999. By 2003, all three sites allowing deer access. Initially relatively few deer were denuded ofvegetation cover and had lost up to m entered the yard, but after several weeks, several 0. 6 oftopsoil (Fig. 2).Attwo ofthe sites, retaining animalsgainedregularaccessduringnights.Afterthree walls supporting built-up residential yards had been Proc. Linn. Soc. N.S.W., 126, 2005 101 RUSA DEER IN ROYAL NATIONAL PARK, SYDNEY Figure 1. Changes in the structure ofvegetation and soils related to deeractivity: a (top) clearingapproxi- mately220meastofBundeenawithinadrysclerophyllforestfromwhichallwoodyunderstoreyplantshave been eliminated and groundcoverplants havebeen thinned exposingbaresoil; b (bottom) an area nearthe beginningofthecoastwalkopenedupbydeeractivityapproximately6monthspriortophotophraph,showing deerfootprintsonbaredsoilsurfaceandlignotubersofZ,ai}i6er//ayorm<>5aexposedbysignificantsoilerosion. 102 Proc. Linn. See. N.S.W., 126, 2005 D. KEITH AND B. PELLOW Figure 2. Site on the busliland/suburban interface at the end of Scarborough Street, Bundeena, where concentrated deer activity resulted in substantial loss of soil. The relic shrub with exposed root burl is Leucopogon ericoides. Feeding of deer ceased at this site approximately two years prior to photo- graph, though the animals continue to pass through the area and maintain an exposed soil surface. substantially undermined by erosion ofthe dune. The were heavily browsed in some areas appeared to be denuded areas had high densities of deer droppings relatively unaffected in others. The removal ofshoots and footprints, and varied in area from 400 to 1075 was particularly frequent and conspicuous in many m-. The roots of trees and large shrubs (Corymbia leguminous species (family Fabaceae), with some giimmifera, Banksia integrifolia subsp. integvifolia) individuals being completely defoliated. Rainforest had been exposed and broken in the eroded sites. species, notably Acmena smithii, Cissus antarctica, Substantial volumes of sand had been transported Clerodendntm tomentosum, Rapanea howitteana downslope and deposited within the adjacent area and Syzyghimpaniculatiim, also suffered high levels of native vegetation. The equivalent sites on the of defoliation. The plant families Epacridaceae m bushland/suburban interface, approximately 50 and Proteaceae also had large numbers of affected fromeachofthe feeding sites, showedlowerdensities species. ofdeerprints and dung, comparatively little evidence Bark was removed or damaged on large woody of erosion and sedimentation, and retained an open stemsofsixspecies (Table 1). Insomecases, scattered continuous cover of grasses, herbs and scattered remains of bark and the abraided appearance of the shrubs. stem from which bark was removed suggested that damagewasprobablycausedbyantlerrutting. Inother Plant survey cases, particularly the rainforest species, bark had Seventy native plant species from 29 families been torn offin strips and may have been consumed and two introduced species showed evidence of as food. In several monocotyledonous species, damage by deer (Table 1). Young foliage and shoots including orchids, Doiyanthes and Xanthorrhoea, were preferentially browsed on most of the plant inflorescences had been consumed or destroyed. species recorded. Older and tougher leaves were less Remains of destroyed Xanthorrhoea inflorescences affected, although in many species there appeared to indicated that this occurred during the bud stage. be little distinction between consumption of young Only nine species of plant consistently showed no and old leaves (Fig.3). The effects of browsing evidence of damage by deer. Three of these were were spatially variable because some species that ferns and three were sclerophyllous sedges. Proc. Linn. Soc. N.S.W., 126, 2005 103 Family Species Young youngand Bark Woody Seedlings Inflorescences Not foliage and old foliage stripped stems uprooted damaged or affected shoots and shoots broken consumed consumed consumed Agavaceae Doryanthesexcelsa X X Anacardiaccac Euroschinusfalcata X Anthericaceae Thysanotusvirgalus X X Apiaccae Platysace linearifolia X Arecaceae Livistona auslralis X X Asparagaceae *Asparagusdensijlorus X Casuarinaceae Allocasuarinadistyla X Casuarinaceae Casuarinaglauca X X Cyperaceae Causlispentandra X Cyperaceae Lepidospermaconcava X Cyperaceae Schoenusbrevifolius X Dennstadetiaceae Ptehdium esctilentum X Dennstadetiaceae Hypolepismuelleri X Dilleniaceae Hibbertiascandens X Eiaeocarpaceae Elaeocarpusreticulatus X Epacridaceae Astrolomapinifolia X? Epacridaceae Brachyhmadaphnoides X Epacridaceae Epacrislongiflora X Epacridaceae Leucopogon ericoides X Epacridaceae Leucopogonparviflorus X Epacridaceae Monotocaelliptica X? Epacridaceae Monotocascopaha X Epacridaceae Styphelia viridis X Euphorbiaceae Phyllanthusgunnii X? X Euphorbiaceae Ricinocarpospinifolius X Fabaceae Acacia implexa X Fabaceae Acacialongifolia X Fabaceae Acaciasuaveolens X Fabaceae Aotusericoides X Fabaceae Bossiaeaensata X Fabaceae Bossiaea heterophylla X Fabaceae Dillwyniafloribunda X Fabaceae Kennediarubicunda X Fabaceae Phyllotaphyllicoides X Fabaceae *Sennapendula var. X glabrata Fabaceae Viminariajuncea X? Iridaceac Patersoniaglabrata X Lomandraceae Lomandra longifolia X l.uzuriagaceae Geilonoplesium cymosum X Moraceae Ficusrubiginosa X Myrsinaccae Rapanea howitteana X Myrtaceae Acmenasmithii X Myrtaceae Angophoracostata X X Myrtaceae Corymbiagummifera X? X X X Myrtaceae Leptospermum laevigatum X Myrtaceae Leplospermum X polygalifolium Myrtaceae Leptospermumsquarrosum X Myrtaceae Leptospermumtrinervium X Myrtaceae Melaleucanodosa X Myrtaceae Syzygiumpaniculatum X X 104 Proc. Linn. Soc. N.S.W., 126, 2005 D. KEITHAND B. PELLOW Family Species Young young and Bark Woody Seedlings Inflorescences Not foliage and old foliage stripped stems uprooted damaged or affected shoots and shoots broken consumed consumed consumed Oleaceae Notelea longifolia X Orchidaceae Caladeniacaerulea* X Orchidaceae Cyrtostytisreniformis X Orchidaceae Glossodia minorX X Caladenia caerulea hybrids* Orchidaceae Pterostylissp. * X Phormiaceae Dianellacaerulea X Poaceae Austrostipapubescens X Proteaceae Banksia ericifolia X Proteaceae Banksia integrifolia subsp. X X integrifolia Proteaceae Banksiamarginata X Proteaceae Banksia oblongifolia X Proteaceae Banksiaserrata X X Proteaceae Conospermum taxifolium X Proteaceae Hakea laevipes subsp. laevipes X Proteaceae Hakeapropinqua X Proteaceae hopogon anemonifolius X Proteaceae Lambertiaformosa X Proteaceae Persoonia levis X Proteaceae Petrophilepulchella X Proteaceae Telopeaspeciosissima X X Proteaceae Xylomelumpyriforme X Restionaceae Hypolaenafastigata X Restionaceae Leptocarpus tenax X Rutaceae Acronychiaoblongifolia X X Sapindaceae Cupaniopsisanacardioides X X Sinopteridaceae Pellaeafalcata \2S.falcata Solanaceae Solanumslelligerum X':' Ulmaceae Celtispaniculata X X Verbenaceae Clerodendrum tomenlosum X X Vitaceae Cissusantarctica Vitaceae Cissus hypoglauca X Xanthorrhoeaceae Xanthorrhoearesinifera X * Introduced species * Margaret Bradhurst, unpubl. data. Table 1. (opposite page and above) List ofplant species affected by deer. Nomenclature follows Harden (1990-2002) and recent updates (www.plantnet.rbgsyd.gov.au). Effects on Syzygiumpaniculatiim to recover when deer were excluded, shooting new All 93 individuals of Syzygium paniculatum foliage in the next growing season. The five plants in suffered some loss of foliage when deer gained the adjacent yard, where deerremained excluded, did access to the yard (Table 2). For a large majority of not suffer any appreciable loss of foliage. Scats and individuals, the level ofdefoliation was severe, with nocturnal observations indicated that deer regularly less than 25% of foliage remaining on the plant. gained access to the open yard, but not to the closed About 15% ofplants were severely damaged, having control, while common brush-tailed possums gained theirbark stripped offor their main stem broken near accesstobothyardsandweretheonlyothervertebrate ground level. However, about 90% of plants began herbivore observed at the site. Proc. Linn. Soc. N.S.W., 126, 2005 105 RUSA DEER IN ROYAL NATIONAL PARK, SYDNEY Figure 3. Defoliation caused by deer herbivory on shrubs of (a - upper left) Styphelia viridis subsp. vindis;(b-u]^^eYr'x^i)Leptospermumsquarrosum; (c - lower left) Banksia marginata; (d - lower right) Persoonialevis. Various heathland sites on sandca. 400-500metresfromtheeasternfringeofBundeena. 106 Proc. Linn. Soc. N.S.W., 126, 2005 D. KEITHAND B. PELLOW Control Treatment 6 months after Treatment oneyearafterdeer gained deergainedaccess access and were subsequently excluded Effect class Number(%) of Number(%)ofplants Alive & sprouting Alive but with Dead plants (N=5) (N=93) new foliage no new foliage >75% foliage and branchlets consumed 78 (84) 72 1 <75% foliage and branchlets consumed 3(3) 3 main stembroken within 30 cm ofbase 9(10) 7 1 bark removed 3(3) 2 1 foliage and branches unaffected 5 (100) 0(0) Table 2. Effects ofdeer on planted Syzygiumpaniculatum in the exclosure experiment. the highest content of grasses (70-90%) and faeces fromdrysclerophyll forestandheathlandhadroughly DISCUSSION equal proportions of broad-leaf and grass material. Moriarty (2004b) recorded similar dietary variation Deer consumed a wide variety ofplant material from rumen analyses, with grasses comprising the including young and old foliage, branchlets, bark majority offood inthevicinityofclearedareas, while and reproductive material of a large number of native plants other than grasses were the major food plant species from a broad taxonomic spectrum. source in other areas. The generalisation that deer limit regeneration Our list of plant species affected by deer is and reproduction in a wide variety of plant species substantially larger than previously reported lists appears to hold for different species of deer studied (Hamilton 1981), but most unlikely to be exhaustive. across a broad range of ecosystems all over the Uncommon species and herbaceous species, which world, including North America (McShea and may be consumed in their entirety, are likely to be Rappole 1999, Opperman and Merenlender 2000, under-recorded in reconnaissance surveys such as Rooney 2001, Rooney et al. 2004), Europe (Kay ours. There is also a risk that signs ofbrowsing could 1993, Fuller and Gill 2001, Rackham 2003) andAsia havebeen overlookedon someplants orthatevidence (Okuda and Nakane 1990, Kahn et al. 1994, Akasi ofbrowsing was erroneously attributed to deer. The and Nakashisuka 1999), where deer are native, and latter source of errors is unlikely to be significant Australia(Moriarty2004a),NewZealand(Jane 1994, because wallaby and possum scats were rarely seen Coomes et al. 2003) and SouthAmerica (Veblen et al. within the study area, whereas deer scats were very 1992),wheretheyhavebeenintroduced. Inbothcases, common. Alternative techniques entail different deer populations seem to have increased recently, sampling errors. Analyses of faeces and rumen either as a result of expansion into new habitats at samples, for example, face difficulties ofidentifying previouslyuninhabited locations (Forsyth etal. 2004) plantfragments,highvariabilitybetween samples and orasaresultoflandscape changeswithintheirnatural limitations that labour-intensive laboratory analysis range (Mladenoff and Steams 1993, Fuller and Gill impose on sample size. Nevertheless, it would be 2001). The large numberand diversity ofnative plant possible to compile a more comprehensive list of species affected by deer in the Bundeena-Jibbon area plants consumed by deer with increased sampling were also consistent with previous studies in Royal effort and a combined sampling approach including National Park that have shown deer to be adaptable exclosure experiments and analyses of faeces and dietary generahsts. Hamilton (1981), for example, rumen. showed that the proportion of food types consumed In addition to the direct effects ofherbivory, deer by deer varied with season and habitat. Deer faeces had substantial, though localised, impacts on soils generally contained higher proportions of shrub and and vegetation structure. Dune soils were severely herb fragments in winter and/or spring, and higher erodedatpartsoftheurbaninterfacewherefeedingby proportions of grass fragments in other seasons. humansleadtointensivedeeractivity.Thecomparison Faeces recovered from rainforest had the highest with equivalent sites nearby indicated that the severe content of shrub and herb material (ca. 80%), while impacts diminish rapidly with distance from feeding faeces from grassland/wet sclerophyll habitats had sites. However, destabilisation ofthe dune may result Proc. Linn. Soc. N.S.W., 126, 2005 107 RUSA DEER IN ROYAL NATIONAL PARK, SYDNEY in the longer tenn if the denuded sites become the will help to resolve uncertainties that arise when catalyst Ibr more widespread mobilisation of sand. interpreting our experimental results in the context of Within native vegetation, shrub cover was locally wild plant populations. reduced along deer tracks and in encampment areas In Fig. 4, we propose a model of impacts based but erosion of sand was generally minimal. Impacts on plant life histories as a means ofstructuring future on wetland soils were more marked, with increased experimental investigations on the medium- to long- exposure, compaction and surface deformation term impacts ofdeer herbivory on native vegetation. evident in all three wetlands inspected. The model proposes that deer herbivory has its The exclosure experiment showed that deer may largest impact on population viability by interrupting have very substantial impacts on the populations of two major plant life-cycle processes: seedling at least some plant species over a relatively short establishment and seed production. Compared with time frame. The extent offoliage and shoot removal established plants, seedlings have less capacity to precluded any chance of reproduction in the plants recover after defoliation and could be more palatable exposed to browsing by deer. The level of mortality due to the lowercontent offibre, tannins and phenolic caused by bark stripping and stem breakages, compounds in their leaves. Bushfires expose more while comparatively small over the duration of seedlings to browsing by deer because they release the experiment, would account for an appreciable seeds of many species from dormancy or canopy reduction in a cohort of saplings over several years. storages (Keith 1996). Populations of plant species The combined effects of foliage and shoot removal that only regenerate from seed are exposed to greater and cumulative mortality are likely to delay or risks of decline than those in which a proportion prevent the growth ofindividual plant canopies above of pre-fire established plants survive. Factors that the browse height. Syzygium paniculatum appears influence the density of post-fire deer populations, to be one of the more palatable native plants in the such as fire size and patchiness and deer dispersal study area, despite the essential oils in its foliage patterns, are potentially important in mediating the and the availability of copious grass in the vicinity. impact ofdeer on seedling recruitment. The observed effects on S. paniculatum probably A second mechanism of deer impact on the represent the more severe of those to be observed viability ofplant populations is through the reduction among wild plant populations although, in bushland, of fruit production (Fig. 4), as the resulting decline severe defoliation was observed in species from a in seed banks reduces the capacity for seedling wide range of plant families including Fabaceae recruitment. These effects are likely to be most and Epacridaceae (Fig. 2). Effects are likely to be significant in species such as terrestrial orchids particularly severe in rainforest communities, such and lilies, in which all reproductive material may as the Littoral Rainforest Endangered Ecological be consumed in a single visit, and those species Community. They also have major implications for whose reproductive effort is largely limited to the revegetation projects, which may be prone to major post-fire period (Keith 1996). While the magnitude losses ifdeer gain access to the plantings before they of reductions in fecundity remain to be quantified grow beyond browse height (e.g. Opperman and under varying levels of browsing, our qualitative Merenlender 2000). observations during reconnaissance suggest that the A number of limitations in the design of current densities of deer populations in the Jibbon/ the exclosure experiment impose constraints on Bundeenaareacouldbecausingsubstantialreductions interpretation of the results. The comparison was in seed production in a wide range ofplant species. based on unreplicated treatments with unbalanced A third life-cycle process susceptible to sample sizes. In the field, plants would be exposed interruption by deer herbivory is the survival and to deer browsing at a much younger stage and would growth of established plants. These effects are not be surrounded by grassy areas as they were in likely to be less significant in plant species that are our experiment. Despite these differences between capable of growth above the vertical reach of deer the experimental conditions and those in the field, (c. 1.5-2.0 m). However, sustained herbivory could the symptoms of deer browsing observed in the slow or block the transition from juvenile to mature experimental population were substantial (relative to growth forms and could also reduce survival in the control plants) and similar to those observed in mature individuals if bark stripping occurs at levels other rainforest and sclerophyll forest plant species observed in the exclosure experiment. These effects during reconnaissance of native vegetation. Data have resulted in failure of revegetation projects from replicated exclosures, which have recently been elsewhere in Royal National Park (e.g. Hacking River established in native vegetation (Moriarty 2004b), Valley) and in other parts of the world (Augustine 108 Proc. Linn. Soc. N.S.W., 126, 2005