Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Resilience of two Banksia species to global change: Comparing results of bioclimatic modelling, demographic and translocation studies

Resilience of two Banksia species to global change: Comparing results of bioclimatic modelling,... International Journal of Biodiversity Science and Management 2 (2006) 59–72 Resilience of two Banksia species to global change: Comparing results of bioclimatic modelling, demographic and translocation studies 1 2 E. T. F. Witkowski and Byron B. Lamont Restoration and Conservation Biology Research Group, School of Animal, Plant and Environmental Sciences, University of the Witwatersrand, Johannesburg, South Africa Department of Environmental Biology, Curtin University of Technology, Perth, Australia Key words: Allee effect, Australia, BIOCLIM, climatic profiles, distribution range, habitat degradation, habitat loss, landscape fragmentation, long-distance dispersal, persistence, population modelling SUMMARY Banksia goodii (rare and endangered) and B. gardneri var. gardneri (widespread) are closely-related rhizomatous evergreen sub-shrubs of southwestern Australian scrub-heath and woodland. They have 17 and 177 known populations, respectively, mostly small remnants due to landscape fragmentation from agricultural activities. Bioclimatic pro- files developed using BIOCLIM indicate that B. gardneri tolerates a wider range of climatic conditions than B. goodii, which has a very narrow predicted range. These results do not match what one would predict from their comparative biology. Specifically, their post-fire survival and resprouting vigour, rates of seedling growth and soil penetration, and sus- ceptibility to seedling predators are similar. A field trial established along a steep climatic gradient of growing season length, showed that both species could extend their ranges beyond the distributions predicted by BIOCLIM, especially B. goodii. Seedlings of both species survived for at least 8 years at sites with two (but not three) months shorter and one month longer growing season than experienced by natural B. goodii populations. The rarity of B. goodii is a result of its recent origin, dispersal limitation, possibly habitat special- ization (dense woodland), and the impacts of habitat degradation and fragmentation within its current range. Under these circumstances, it is highly improbable that any sort of bioclimatic modelling could predict its potential climatic envelope. A stage-based model of B. goodii shows that under natural conditions the species is stable because of extremely low natural adult mortality (indeed, undetected), but plant losses due to human interference cannot be compensated due to its low levels of sexual reproduction. Population growth of B. goodii, and its potential for recovery, depends on population size and survival of seedlings and juveniles. Reproductive output per adult increases with pop- ulation size, with populations of < 8 plants being sterile. Conservation management should target these factors as well as prevent the destruction of existing adults through additional land clearing or other threatening processes. Thus, habitat degradation, fragmentation and loss are likely to have a greater influence than any predicted global Correspondence: E. T. F. Witkowski, Restoration and Conservation Biology Research Group, School of Animal, Plant and Environmental Sciences, University of the Witwatersrand, Private Bag 3, PO Wits, 2050, Johannesburg, Gauteng, South Africa. Email: ed@gecko.wits.ac.za 59 Resilience to global change in banksias Witkowski and Lamont climate change on species survival. Translocation experiments appear to have far more scope in predicting the outcomes of climate change than bioclimatic modelling. As a con- sequence of poor dispersal ability and low reproductive rates, as well as habitat fragmenta- tion, these banksias cannot migrate as the climate changes, but they already have a wide climatic tolerance and seem likely to withstand climate change in situ. INTRODUCTION Global environmental change is occurring at an and understanding the present-day climatic limits unprecedented rate (Vitousek 1994). With current of species, allows for the exploration of how distri- levels of increase, CO is expected to double pre- butional patterns might change in response to cli- industrial concentrations by 2200 and result in a matic change (Busby 1988; Huntley 1995). This will predicted increase in mean global surface air aid in the development of strategies for long-term temperature of 2°C (range 0.9–3.5°C) by 2100 management and conservation of species. (Vitousek 1994). There is also an anticipated in- The South-West Botanical Province of Western crease in the intensity of the hydrological cycle, Australia (SWBP; Beard 1980) is a mediterranean with an increasing frequency of severe floods and climate region renowned for its floristic richness droughts. This unprecedented rate of climate (Hopper 1979; Lamont et al. 1984), and is a recog- change begs the questions of whether plant species nized hotspot for floristic diversity (Hopper and will be ‘resilient’ to this change, what functional Gioia 2004). SWBP also has the highest concentra- characteristics are associated with species that tion of declared rare flora in Australia, and it is might persist, and can these be predicted. estimated that Western Australia has the highest Species ranges have expanded/contracted rates of plant extinction and endangerment of during interglacial/glacial periods and moved all Mediterranean climate regions (Greuter 1994). polar-wards/equatorially (Clark et al. 1998). How- Banksia, a pan-Australian genus, has been des- ever, can species track current rapid climatic cribed as ‘the most characteristic genus’ of the changes and remain within areas which match their SWBP (Speck 1958), where it may dominate vegeta- bioclimatic niches? There are few direct tests of tion on the poorest soils. Climate, most notably species range extension within natural habitats. rainfall, has changed in this region in recent times. The success of introduced species throughout the Pittock (1988) noted that there has been a mean globe indicates that many, probably most, species 3–5% drop in rainfall per decade over the preced- are not present in all suitable habitats. While physi- ing 70 years in SW Australia. Clearing for agricul- cal barriers to dispersal prevent them from occur- ture is a likely contributory cause to this. If this ring in distant but otherwise suitable areas, it is trend continues over the next 50 years (a drop in unknown whether plant species can extend their rainfall of 20%), there will be a dramatic decline in climatic ranges per se. Clearly they do extend their the population size of some species (Burgman and ranges over evolutionary time, but at a rate too slow Lamont 1992). Modelling the greenhouse effect to match the speed of present climate change. has indicated drier and warmer winters can be In general, alien plants on new continents occur expected in SW Australia by 2030, although within their native climatic ranges (Drake et al. some models predict no change (Pittock 1988; 1989, and references therein). Henderson-Sellers and Blarg 1989). Extensive At the regional (biome) and global scales, cli- areas of this region have recently received less than mate defines the broad limits to the distribution of 80% of the long-term average since 1975 in response plant taxa and the dominance of plant life forms to shifts in the synoptic weather systems influencing (e.g. Rutherford and Westfall 1986; Woodward and SW Australia (Hope et al. 2006). Furthermore, Williams 1987). Bioclimatic modelling has been projected changes in the synoptic systems in used to determine the influence of particular clima- response to increasing concentrations of green- tic variables on the distribution of species, vegeta- house gases will lead to increasingly dry conditions tion types and biomes (Busby 1986; Hill et al. 1988; in the twenty-first century (Hope 2006). Eeley et al. 1999). Recognizing the dynamic nature The 63 Banksia species found in Western of species distributions from the past to the present, Australia are thought to be confined to the SW 60 International Journal of Biodiversity Science and Management Resilience to global change in banksias Witkowski and Lamont corner by climate. The genus cannot tolerate namely brevidentata, gardneri and hiemalis, each of drought conditions corresponding to a rainfall of which is more widespread than B. goodii, but only < 250 mm/annum (Lamont and Connell 1996; var. gardneri occasionally co-occurs with B. goodii Cowling and Lamont 1998). Within this area (George 1981; Taylor and Hopper 1988). The two species composition and density vary with substrate, study species are very similar in growth and leaf landscape features, climate, vegetation type and forms, floral morphology and reproductive biol- fire history. We pose the question: what is the rela- ogy, and have the same pollinators (George 1981; tive importance of expected future changes in Taylor and Hopper 1988; Witkowski and Lamont climate as a result of increased atmospheric con- 1997). Reproductive potential, either sexual or centrations of CO and other ‘greenhouse’ gases, vegetative, is low in both species, and they have compared with other forms of human-mediated slow rates of growth as seedlings, or resprouts after environmental change on the distribution of fire. Seed production by both species is lower than banksias in the SWBP? Many studies have used for many other banksias (Witkowski et al. 1991; bioclimatic modelling to examine potential changes in the distributions of species or eco- systems/biomes in response to predicted climate change scenarios (Kadmon et al. 2003). However, few actual field manipulations have been under- taken to determine whether a species can survive beyond its current range. This study aims to determine whether bioclimatic and population modelling can predict the actual and potential distributions of banksias under global climate change scenarios. An experimental translocation study along an extensive climatic gradient (length of growing season) was established for two Banksia species in order to determine the potential for climatic range extension. Population modelling was also undertaken to assess the effects of various factors on the long-term persistence of populations of various sizes (Dreschler et al. 1999). STUDY SPECIES AND AREA Our study species were the prostrate Banksia goodii (nomenclature follows Green 1985), a rare and endangered species restricted to 17 populations distributed between Albany and the Porongurup Range, within a small area of 38 km (N–S) by 14 km (E–W) in the woodlands of southwestern Australia (George 1981; Taylor and Hopper 1988; Witkowski and Lamont 1997), and its more widespread and Figure 1 Distribution of Banksia goodii and B. gardneri. abundant close relative, B. gardneri var. gardneri (a) Distributions predicted by BIOCLIM, plus trans- (henceforth B. gardneri) (Figure 1). The biology of location sites. 1 = Walpole National Park, 2 = Albany, these species has been studied in detail (Lamont 3 = Mt Barker, 4 = Stirling Range, 5 = Katanning. Con- et al. 1993; Witkowski and Lamont 1997). Banksia tours for length of the growing season in months goodii is considered cladistically terminal within the (Prescott formula) are also shown. (b) Actual known section Prostratae (Thiele and Ladiges 1996), populations of B. gardneri compared with its BIOCLIM which consists of 6 species of Banksia with a pros- predicted distribution. (c) Actual known populations trate growth form (George 1981). Its closest relative of B. goodii compared with its BIOCLIM predicted is B. gardneri, which occurs as one of three varieties, distribution International Journal of Biodiversity Science and Management 61 Resilience to global change in banksias Witkowski and Lamont Witkowski and Lamont 1997), but similar to some climate, vegetation and biogeography of the whole other resprouters (Cowling et al. 1990). Pre- region with respect to banksias are given in Lamont dispersal granivory by weevils is extremely variable and Markey (1995), Richardson et al. (1995), between populations of both species and some- Lamont and Connell (1996) and Witkowski and times most seeds are devoured. B. gardneri does Lamont (1997). not reproduce vegetatively, whereas B. goodii does, albeit at a slow rate (Witkowski and Lamont 1997). B. goodii is classified as rare and endangered MATERIALS AND METHODS (Western Australian Government 1999; IUCN Bioclimatic modelling 2001). Both species resprout after fire, with no evidence BIOCLIM is a bioclimatic analysis and prediction of mortality after several fires. Two field trials system used to predict the distribution of species and one glasshouse trial consistently showed that (or other ‘entities’; Nix 1982, 1986; Busby 1988, B. gardneri has a higher germination percentage 1991). BIOCLIM generates a set of variables (clima- than B. goodii, which partially trades off with vegeta- tic indices) considered to have biological signifi- tive reproduction in B. goodii, while seed store per cance and that describe the range, extremes and adult does not differ between species (Witkowski seasonality of climatic conditions. These variables and Lamont 1997). Both species are probably are interpolated across the geographical surface (at equally susceptible to the stochastic effects of (a) various scales on the basis of longitude, latitude and generalist fungal pathogens (especially Phytoph- altitude). By matching the known distribution of a thora cinnamomi and stem cankers (Witkowski et al. taxon to the variable surfaces, a statistical summary 1991; Lamont et al. 1995)); (b) changes in fire of the values of the climatic indices for that taxon season/intensity/frequency on seed store and is produced. This summary, referred to as a bio- release, and adult survival; (c) post-fire environ- climatic profile or envelope, provides a quantitative mental conditions on seedling establishment and description of the climatic environment occupied adult survival (Lamont and Witkowski 1995); (d) by the taxon (Nix 1986). Points on the variable grid granivory and seedling herbivory (Cowling et al. surfaces that match the bioclimatic profiles of the 1990; Witkowski et al. 1991); (e) land degradation species can be identified to delimit its potential due to poor pastoralism practices (habitat modi- distribution (i.e. the area of suitable climatic condi- fication); (f) clearing for roads and agriculture tions; Busby 1991; Lindenmayer et al. 1991). The (habitat fragmentation and loss); and (g) alien underlying assumption of most BIOCLIM analyses plant invasions. is that species can only colonize and survive in Extensive clearing for agriculture (mainly cere- areas with climates fitting their current bioclimatic als, but also sheep pasture in the study area, but profiles. The model also does not consider the cereals in other parts of SW Australia) and roads potential interactions between climatic variables over a period of 100 years (Allison and Hobbs 2004) (Nix 1986). has fragmented the landscape of southwestern Seventeen distribution records of B. goodii and Australia. Most B. goodii populations are now only 177 records of B. gardneri (longitude, latitude and represented by remnants on roadsides, while altitude) were analysed using BIOCLIM, against others occur on farms, in state forest and a nature 24 climatic variables, including various tempera- reserve. The rarity of B. goodii is a function of its ture and rainfall indices. The outputs comprised small geographic range rather than scarcity within descriptions of the climatic profiles of each species its range per se (Witkowski and Lamont 1997), and and predicted distribution maps. In addition, the its small geographic range appears related to its locations of the five climatic gradient translocation more recent origin relative to B. gardneri (Thiele sites (see below) were also analysed using BIOCLIM and Ladiges 1996). to obtain their climatic profiles. The study area was centered 400 km south of Perth (35°00′S, 117°55′E) near Albany. The soils Seedling establishment along a steep climatic gradient of the native banksia populations are infertile sands usually overlying laterite (Griffin et al. 1985; Sixty intact seeds of B. goodii and B. gardneri Witkowski and Lamont 1997). More details on the were planted at each of five sites, which represented 62 International Journal of Biodiversity Science and Management Resilience to global change in banksias Witkowski and Lamont a climatic gradient of growing season length. 11–100, and (c) > 100 (6, 6 and 5 for B. goodii, and Seeds were collected from 40 serotinous cones 49, 83 and 45 for B. gardneri, respectively (Taylor harvested from 20 randomly selected plants from and Hopper 1988)), no differences in population the largest and most fecund population, and hence size is evident between species (X = 0.87, d.f. = 2, likely to have the highest genetic diversity (Lamont p = 0.65). Population sizes of B. gardneri on farms, et al. 1993). Seeds were extracted from the cones roadsides and reserves clearly match the variation by burning them until the follicles opened, mimick- in population sizes of B. goodii, both within the ing the natural seed release process (Witkowski wooded region occupied by B. goodii and through- et al. 1991). Using the Prescott formula, the out its range. number of growing months can be determined from mean monthly rainfall and evaporation Distribution and bioclimatic modelling (Bartlett 1975). Growing seasons of six (Katanning), seven (N. Stirling Range), eight Banksia goodii has a very restricted distribution, (Mt Barker), nine (Albany) and ten (Walpole- within which many of its 17 populations are fairly Nornalup National Park) months were chosen disjunct, while B. gardneri is far more widespread (Figure 1). This is a steep climatic gradient of and occupies a greater altitudinal range (Figure 1; from a six- to ten-month growing season over a Table 1). Both species occur in areas with similar distance of only about 200 km. Within these temperatures, but B. goodii receives higher rainfall areas, sites were selected with soil and other and slightly lower temperature extremes than environmental conditions which, as far as possible, B. gardneri (Table 1). matched typical B. goodii sites. Three composite Similarly, the distribution of B. gardneri pre- soil samples (0–100 mm depth) were collected per dicted on the basis of its bioclimatic profile, greatly site and soil depth was determined. exceeded and included that of B. goodii (Figure 1a). In late June 1990, a 30-m plot was cleared of The predicted distribution of B. gardneri generally vegetation (above- and belowground) at each site, included its distribution records, but omitted some and seeds sown into six 1-m quadrats, spaced 1 m marginal populations in the N, and especially apart. Seedling emergence and survival were moni- a block in the NE corner, but otherwise filled in tored in September, October and December 1990, the gaps between the remaining populations, and April, June and November 1991, May 1992, January extended its current distributional limits to the S, W 1993, January 1996 and June 1999. and E (Figure 1b). Overall extensions to the drier N, and wetter W and S were negligible. The pre- dicted distribution of B. goodii greatly exceeded the total area currently occupied but still omitted some Soil analyses marginal populations, most notably the two in the Soil samples were passed through a 2-mm sieve and SW (Figure 1c). There was no predicted extension −1 analyzed for NO ,NH (0.1 mol L KCl extraction 3 4 to the N or S, a slight contraction in the W, and and Varian autoanalyzer), available P (citric acid slight extension to the E. extract and molybdenum blue spectrophoto- metry), total Fe and K, and exchangeable K, Ca, Mg −1 and Na (0.1 mol L ammonium acetate extraction Seedling establishment along a climatic gradient and atomic absorption spectrometry), organic The climatic profiles of the five translocation sites carbon (Walkley Black) and pH (1 : 5 H O w/v) are given in Table 2. Seedlings of both species by CSBP and Farmers, Bayswater, Perth. Soil emerged during late winter and spring at all five texture was analysed using the hydrometer method. sites (Figure 2). Total emergence was 14.3% for B. goodii and 32.7% for B. gardneri. All seedlings at the driest site, Katanning, died after the first summer drought in 1991, with B. goodii succumbing RESULTS about 2 months before B. gardneri. Mortality Comparison of population sizes was also complete at the Albany site with few seed- Comparing the numbers of populations of each lings remaining by early summer 1991, and com- species which fall into the ranges (a) 1–10, (b) plete absence by the following autumn. This was International Journal of Biodiversity Science and Management 63 Resilience to global change in banksias Witkowski and Lamont Table 1 Position, altitude and climate profiles of the rare species, Banksia goodii, compared with its widespread closest relative, B. gardneri. Climate profiles were produced using BIOCLIM (Busby 1991). Data are means ± SD. P values are based on t-tests (unequal variances); CV = coefficient of variation Environmental variables B. goodii B. gardneri P Number of localities 16 177 Altitude (masl) 83.5 ± 26.7 169.3 ± 89.9 < 0.001 Latitude (°E) 34.87 ± 0.04 34.58 ± 0.21 < 0.001 Longitude (°S) 117.76 ± 0.13 118.14 ± 0.38 < 0.001 Mean annual temperature (°C) 15.3 ± 0.1 15.1 ± 0.2 < 0.001 Mean annual temperature range (°C) 19.9 ± 0.4 21.5 ± 1.4 < 0.001 Mean minimum temperature for coldest month (°C) 6.4 ± 0.3 5.6 ± 0.6 < 0.001 Mean maximum temperature for warmest month (°C) 26.3 ± 0.2 27.1 ± 0.9 < 0.001 Mean temperature for coldest 3 months (°C) 11.5 ± 0.2 10.9 ± 0.6 < 0.001 Mean temperature for warmest 3 months (°C) 19.4 ± 0.1 19.7 ± 0.4 < 0.001 Mean temperature for wettest 3 months (°C) 11.7 ± 0.3 11.5 ± 0.7 0.229 Mean temperature for driest 3 months (°C) 19.4 ± 0.1 19.7 ± 0.4 < 0.001 Mean annual rainfall (mm) 878 ± 62 623 ± 148 < 0.001 Highest mean monthly rainfall (mm) 129 ± 13 85 ± 26 < 0.001 Lowest mean monthly rainfall (mm) 24 ± 1 18 ± 3 < 0.001 CV of monthly rainfall (%) 51.5 ± 2.8 46.1 ± 4.9 < 0.001 Mean rainfall for wettest 3 months (mm) 352 ± 34 238 ± 67 < 0.001 Mean rainfall for driest 3 months (mm) 84 ± 2 64 ± 11 < 0.001 Mean rainfall for coldest 3 months (mm) 332 ± 30 231 ± 63 < 0.001 Mean rainfall for warmest 3 months (mm) 85 ± 4 65 ± 12 < 0.001 Table 2 Position, altitude, length of growing season (Prescott formula) and climate profiles for the five sites used in the climate gradient tranlocation study. CV = coefficient of variation Walpole Stirling National 1 1 1,2 Site Katanning Ranges Mt Barker Albany Park Length of growing season (months) 6 7 8 9 10 Altitude (masl) 320 250 300 80 20 Latitude (°E) 33.70 34.30 34.63 34.88 34.97 Longitude (°S) 117.55 117.77 117.67 117.78 116.99 Mean annual temperature (°C) 15.3 15.1 14.7 15.3 15.6 Mean annual temperature range (°C) 25.0 23.4 22.9 19.8 18.6 Mean temperature for coldest 3 months (°C) 10.1 10.4 10.0 11.5 12.1 Mean temperature for warmest 3 months (°C) 21.2 20.3 19.8 19.4 19.6 Mean temperature for wettest 3 months (°C) 11.1 11.4 10.0 11.5 12.2 Mean temperature for driest 3 months (°C) 21.2 20.3 19.8 19.4 19.6 Mean minimum temperature for coldest month (°C) 4.6 5.0 4.8 6.4 7.5 Mean maximum temperature for warmest month (°C) 29.6 28.4 27.8 26.2 26.1 Mean annual rainfall (mm) 465 483 749 910 1197 Highest mean monthly rainfall (mm) 78 70 111 135 198 Lowest mean monthly rainfall (mm) 12 14 22 24 27 CV of monthly rainfall (%) 61.5 51.0 50.4 52.5 62.0 Mean rainfall for wettest 3 months (mm) 212 197 297 366 537 Mean rainfall for driest 3 months (mm) 42 50 73 85 92 Mean rainfall for coldest 3 months (mm) 211 197 297 346 483 Mean rainfall for warmest 3 months (mm) 42 50 73 85 102 Naturally occurring populations of Banksia gardneri present Naturally occurring populations of B. goodii present 64 International Journal of Biodiversity Science and Management Resilience to global change in banksias Witkowski and Lamont −1 Table 3 Soil physical and chemical (µgg dry mass) properties (mean ± SD of three samples) at the five sites where seeds were sown along a climatic gradient (length of the growing season) in southwestern Australia. Values with the same letter are not significantly different at p < 0.05 (Student-Neuman-Keuls multiple range test) 1 1 1,2 Site Katanning Stirling Range Mount Barker Albany Walpole Length of growing season 6 7 8 9 10 (months) a a a a a Available P 1.0 ± 0.0 1.0 ± 0.0 1.0 ± 0.0 1.7 ± 1.2 1.0 ± 0.0 a a a a a NO 2.0 ± 0.0 2.0 ± 0.0 2.0 ± 0.0 2.0 ± 0.0 2.0 ± 0.0 a a a a a NH 3.7 ± 2.1 4.0 ± 0.0 3.3 ± 0.6 5.0 ± 1.0 5.3 ± 0.6 b b ab a b Total K 31 ± 6 30 ± 3 45 ± 3 57 ± 19 24 ± 2 c c b a c Organic C (%) 0.2 ± 0.0 0.4 ± 0.2 1.6 ± 0.9 2.6 ± 0.6 0.6 ± 0.1 b bc a c d Fe 136 ± 23 93 ± 17 464 ± 214 72 ± 12 31 ± 4 a a a a a pH 5.9 ± 0.1 5.9 ± 0.1 6.1 ± 0.7 5.6 ± 0.2 5.6 ± 0.2 b b a a b Exchangeable Ca 37 ± 1 78 ± 40 425 ± 197 430 ± 80 100 ± 10 b b a a b Exchangeable Mg 15 ± 1 19 ± 12 81 ± 45 100 ± 45 38 ± 5 b b b a b Exchangeable Na 5 ± 1 5 ± 1 15 ± 7 26 ± 7 9 ± 2 b b a a b Exchangeable K 10 ± 5 14 ± 2 35 ± 5 38 ± 16 5 ± 2 b b a b b Gravel (%) 0.1 ± 0.0 0.4 ± 0.2 4.0 ± 1.1 1.0 ± 0.5 0.1 ± 0.0 a b b c b Coarse sand (%) 32.4 ± 4.5 20.6 ± 8.8 11.7 ± 1.4 2.8 ± 0.8 21.0 ± 2.0 b b b c a Medium sand (%) 45.8 ± 3.2 44.1 ± 3.2 40.3 ± 4.1 12.0 ± 1.5 54.9 ± 0.2 d c b a d Fine sand (%) 19.9 ± 1.4 32.1 ± 5.7 42.9 ± 5.2 76.8 ± 3.9 21.0 ± 2.0 c c b a c Silt (%) 1.6 ± 0.3 2.8 ± 0.5 4.7 ± 0.9 7.9 ± 1.5 3.0 ± 0.2 a a a a a Clay (%) 0.3 ± 0.1 0.5 ± 0.3 0.4 ± 0.0 0.5 ± 0.2 0.2 ± 0.0 Soil depth (mm) > 1000 300 150 500 > 1000 Naturally occurring populations of Banksia gardneri present Naturally occurring populations of B. goodii present due to herbivory, most probably by kangaroos DISCUSSION (Macropodidae), bandicoots (Dasyuridae) and BIOCLIM analysis showed that B. gardneri usually wingless grasshoppers (Acrydidae). The Albany occurs in areas with similar temperatures but lower translocation experiment was, however, repeated at rainfall than B. goodii. This is supported by the two additional sites, separated by 3 km, where both observation that B. gardneri took slightly longer species naturally occurred. Herbivory was found to than B. goodii to be eliminated from the driest site, be site-specific, and protection from mammals at Katanning, and had greater survival in the next one site resulted in high levels of seedling recruit- driest, Stirling Range. Furthermore, for the past ment (Witkowski and Lamont 1997). At the remain- 20 million years, the area within which B. goodii ing sites along the climatic gradient (Stirling presently occurs, has been wetter than the overall Ranges, Mt Barker and Walpole-Nornalup National area within which B. gardneri occurs (Lamont and Park), seedling survival was high and similar Markey 1995). Greater tolerance to drought may between species, and remained unchanged allow B. gardneri to survive adverse conditions better throughout 1992 and 1993. By 1993, plants at than B. goodii. The creeping/prostrate growth form Mt Barker had shown the greatest growth. There of both species would be disadvantageous in the was essentially no change to this scenario between wettest areas of the southwest (represented by 1993 and mid-1999, except that the Stirling Range Walpole-Nornalup National Park), where suppres- site was accidentally cleared. sion by excessive litter and/or competition for light The soils of the five sites selected were similar from tall woodland/forest trees would be intense. (Table 3), consisting of over 90% sand, with no However, both species showed comparable levels differences in available P, NO and NH ,or inpH 3 4 of survival under both higher and lower moisture and clay content. Cation concentrations tended to conditions than found in their current ranges. be higher at Albany and Mt Barker. International Journal of Biodiversity Science and Management 65 Resilience to global change in banksias Witkowski and Lamont intervals result in more heavily shaded plants covered with litter, which inhibits both seed pro- duction and seedling establishment. Morphologi- cally, the broader and only slightly lobed leaf lamina of B. goodii , plus a 32% greater leaf area (Witkowski and Lamont 1997), makes it more suited to a shady habitat than B. gardneri. Species climatic profile models are designed to predict potential distribution ranges, but their accuracy depends totally on using data from actual distribution ranges. Hence, it is an explicit assumption in bioclimatic modelling in predicting changes in a species range in response to expected climate change, that the present range provides an adequate description of its climatic profile or envelope (Kadmon et al. 2003). Bioclimatic model- ling predicted that B. goodii would occupy an area, similar to, or slightly less than, its actual known present distribution. Outlying populations were excluded, while a continuous distribution was pro- duced around the ‘core’ of its current distribution. It is clear that when a species has a very restricted distribution, bioclimatic modelling via BIOCLIM may not totally predict its current geographic range. However, to be fair, the model fitted the dis- tribution data fairly well, and at a fine scale factors other than climate, such as topography, soils, and moisture availability, as well as biotic factors, come Figure 2 Emergence and number of seedlings of into play. For B. gardneri, which occupies a greater Banksia goodii and B. gardneri surviving over time at known range than B. goodii, BIOCLIM did little Katanning, Stirling Range, Mt Barker, Albany and more than define its current distributional bound- Walpole National Park aries, ‘filling in’ large areas to the W and SW but missing the NE corner (although extending poten- The growth of B. goodii seedlings/juveniles, for tial habitats 25 km to the E). example, was vigorous at Mt Barker, 25 km north BIOCLIM develops a ‘climate surface’ based on of its current range, with reduced vigour in the data from weather stations, and then interpolates Stirling Range (60 km N) and at Walpole between these point sources on the basis of latitude, (80 km SW). Seedlings of both species establish longitude and altitude. Because climate stations are equally fast, including rates of root penetration sparsely spread in SW Australia, the interpolated (an index of drought resistance among seedlings, ‘climate surface’ may be too coarse to predict in Witkowski 1991), which were identical. However, detail the present distribution of species with both species may be excluded once they are over- narrow distributions, such as B. goodii. Nevertheless, topped by the taller trees found at the wettest site, climate modelling may still be useful for species although B. gardneri is likely to be excluded first with broader distributions, as well as under condi- because B. goodii always occurs in wooded areas. tions of large changes in climate variables for On the other hand, B. gardneri genotypes of popula- species with narrow distributions (Kadmon et al. tions from woody sites (e.g. those co-occurring with 2003). In addition, B. goodii is a neo-endemic and is B. goodii) may be equally vigorous. probably in the process of extending its distribu- In the context of the present locations of tional range. The very poor dispersal ability of B. goodii, fragmentation of the landscape means B. goodii suggests that range extension in this greater variation in fire frequency. Longer fire species is extremely slow (see below). Thus, if there 66 International Journal of Biodiversity Science and Management Resilience to global change in banksias Witkowski and Lamont were a higher density of weather stations in the these Banksia spp. survived well beyond their native region, BIOCLIM would still not be able to predict distributions along the gradient, a longer-term the potential range of B. goodii because its distribu- perspective is needed, as persistent range extension tion is unlikely to be currently limited by climatic can only be confirmed if they are able to success- factors, but rather by dispersal barriers and insuffi- fully reproduce. Due to their long juvenile period cient time for expansion. This argument could be (Witkowski and Lamont 1997; Dreschler et al. extended to B. gardneri, which may also be in the 1999), the plants in the gradient study have not yet process of extending its range. reached reproductive age or size. An added compli- A very thorough assessment of the performance cation is that small B. goodii populations are sterile of climate profile models has been undertaken within their native distribution (see below), despite using the distribution of 192 woody species in profuse flowering and being of a similar size and Israel, a geographic region with similar steep vigour as highly fecund plants from the large popu- climatic gradients to the SWBP (Kadmon et al. lations (Lamont et al. 1993). Hence the gradient 2003). Predictably, accuracy tends to increase (and study plants may, in time, be found to be sterile, variance decrease) with increasing number of although not necessarily in response to an unsuit- distribution records (or observations), although able climate/environment, but due to small popu- this is not linear. Species with < 25 observations are lation size per se. Overall the gradient study has most adversely affected, those with 50 observations shown that the environmental tolerances of these are accurately modelled, and there is no detectable banksias are far greater than would be expected, improvement in accuracy above 100 observations. based on present distribution alone, and these Based on this factor alone, the B. gardneri model broad tolerances are not merely an ability to accli- (177 observations) should be more accurate than mate. Similar findings for other species are re- the B. goodii model (17 observations) (Figure 1). viewed in Gaston (2003) and Kadmon et al. (2003). Interestingly, Kadmon et al. (2003) found that the Given these results, it is unlikely that any improve- position of the species niche along a rainfall gradi- ments to BIOCLIM could adequately predict the ent, from the Mediterranean region to desert, had results obtained from the gradient study. a negative effect on predictive accuracy. Hence, A general pattern of range limitation in species Mediterranean species are less influenced by clima- is the relative importance of abiotic versus biotic tic conditions than desert species. Any factor pre- limiting factors. The distribution of the majority of venting a species from utilizing its potential climatic species appears to be delimited in one direction distribution, whether anthropogenic, biotic or by physical stresses and in another by biological dispersal, will lead to poor predictive accuracy stresses (Brown et al. 1996). In Western Australia (Kadmon et al. 2003), as concluded here for the major physical gradient is aridity (Lamont and B. goodii, and B. gardneri to a lesser extent. Connell 1996). B. goodii and B. gardneri are limited The translocation study showed that both by the aridity barrier to the N and E, and by the species could extend their ranges beyond those biotic barrier of dense shade (taller denser eucalypt predicted by BIOCLIM, and could occupy both forests) to the wetter W and SW of their respective much drier and much moister sites. In particular, ranges. The two species differ in that B. gardneri is BIOCLIM was a poor predictor of the potential more tolerant of aridity than B. goodii, while the distribution of B. goodii. It is only at the wettest site, latter may be more tolerant of deep shade. How- where the plants are presently situated in the open ever, considering anticipated climate change over (not under the shade of the typical eucalypt wood- the next 100–200 years in SW Australia (1–2 months land) that they are likely to be excluded by biotic reduction in length of the growing season at factors, when they eventually become heavily present rates; extrapolating from Pittock 1988), the shaded by dense forest canopy. These contrasting difference between the two species is minor. The results provide a warning to global change scientists present small range of the more recently evolved reliant on this type of climate modelling. Although B. goodii (relative to B. gardneri; Thiele and Ladiges clearly of immense value, modelling of species with 1996) can be explained primarily by evolutionary narrow ranges only provides hypotheses that factors, interacting with biological factors, such as should be tested where possible with field data. its exceptionally low rate of dispersal (see below). However, although the seedlings and juveniles of Thus, future changes in distribution are likely to be International Journal of Biodiversity Science and Management 67 Resilience to global change in banksias Witkowski and Lamont the result of the interwoven effects of both evolu- enough or quickly enough to keep within their tionary and environmental changes. present bioclimatic envelopes as defined by their The rapid rates of plant dispersal during glacial/ known distributional ranges. interglacial periods implied by past records (since The potential fate of B. gardneri and B. goodii the beginning of the Holocene; Clark et al. 1998) exemplifies that of many plant species with in- are far too high to have been produced by tradi- herent short-distance dispersal. In the event of tional notions of life history and dispersal mecha- predicted climate change, and in the absence nisms (Clark et al. 1998). Seed dispersal distances of of human-mediated conservation efforts, B. goodii these two prostrate Banksia spp. appear to be very could become extinct, while B. gardneri would prob- low. Seeds have wings for wind dispersal, but owing ably suffer a considerable reduction in its range. to the low heights of cones (< 30 cm above the soil However, other features of the biology of B. goodii surface), observed dispersal distances have all been will probably work against extinction. The highly < 1 m (unpublished observations), and even under fragmented landscapes of today, with extensive optimal windy conditions are unlikely to exceed unfavourable habitat barriers to banksia migration, 10 m. Typically, long-lived species, which resprout such as sheep pastures and roads, provide an addi- after recurrent fire (or other severe disturbances), tional and potentially insurmountable challenge. such as these banksias, often produce few seeds and Potential safe sites may be so few and far between seedlings (Witkowski and Lamont 1997). These that effective natural colonisation and migration banksias are serotinous (canopy seed storage), with will be virtually impossible, because there are too seed store peaking 15 years after fire, and declining few ‘stepping stones’. Little is known about the life thereafter (Dreschler et al. 1999). Short dispersal spans of resprouting banksias. For the shrubby distances are exacerbated by (a) low seed produc- Banksia attenuata, longevity has been estimated at tion and (b) infrequent dispersal events – seed 300 years and mortality after fire at about 5% release from the cones is triggered by fire, usually at (Enright and Lamont 1992). This species relies a frequency of 8–30 years. Such figures would only on occasional seedling recruitment after fire to allow these species to move 0.2 km over the next replace loss of some pre-fire plants. However, no 200 years, assuming fires every decade, and a very adult mortality of B. goodii or B. gardneri was evident optimistic distance of 10 m dispersal per fire. A following several fires. Zero recruitment (from 126 one-month reduction in growing season over that adults) and two recruits (from 144 adults) by four time is equivalent to a distance of 30–100 km winters after two fires for B. goodii indicates pro- (Figure 1). The probability of episodic long- longed population stability in the absence of distance dispersal to suitable distant sites (He et al. human interference. Zero recruitment of B. 2004) is also extremely low within this highly frag- gardneri populations was also found (Albany and mented landscape. Neither models of leptokurtic Mount Barker; unpublished data). In B. goodii,we dispersal profiles (fat-tailed curves) or non- estimate it takes about 20 years for recruits to standard mechanisms of dispersal (Clarke 1998; become fire-tolerant and start flowering, while fire Higgins et al. 2003a), except direct human manipu- retards flowering of adults for at least 3 years lation, are likely to make any meaningful difference (Lamont et al. 1993; unpublished) and we expect to these estimates due to the exceptionally low B. gardneri to be similar. For both species, we recom- levels of seed production and infrequency of seed mend a mean fire rotation of 10–15 years, and at release. Indeed, migration may even be halted regular intervals this should be extended to 15–20 altogether under these conditions (Higgins et al. years (Dreschler et al. 1999) on the grounds that this 2003b). Hence, the ranges of these banksias are should be sufficient time for seedlings to become highly likely to be dispersal-limited. Furthermore, fire-tolerant, keep litter in check, and ensure occa- the probable sizes and distribution of patches suit- sional rejuvenation of vegetative growth, while pro- able for Banksia seedling establishment (within this viding sufficient time for seedling establishment now highly disturbed and fragmented landscape) and substantial seed set. For B. goodii, longevity is indicates that even this very low rate of spread is estimated as 400–500 years, based on clump sizes probably very optimistic (Bergelson et al. 1993). In and growth rates (Witkowski and Lamont 1997, the event of predicted global climate change, these Dreschler et al. 1999), and slightly less for B. gardneri banksias are unlikely to be able to disperse far (300–400 years) due to smaller clump sizes and 68 International Journal of Biodiversity Science and Management Resilience to global change in banksias Witkowski and Lamont because, unlike B. goodii, it does not reproduce further ‘nibbling away’ of the populations (clear- clonally. ing of land for agriculture) is prevented. If in the Population numbers of B. goodii have decreased future, atmospheric CO can be brought under over the last 50–60 years due to clearing for agricul- control or reduced, and climate change is reversed, ture and for roads, resulting in smaller, fragmented these banksias may be able to persist over the populations. Furthermore, habitat fragmentation unfavourable period and begin to ‘thrive’ once has itself indirectly affected the reproductive poten- again. In summary, as shown by their biology and tial of B. goodii (and by comparison almost certainly the gradient study, B. goodii and B. gardneri are B. gardneri). Habitat fragmentation has resulted in long-term persisters, and resilience to marked reduced sexual reproduction in B. goodii. Mean climatic change is already built into these species, seed production per plant decreases as populations decreasing the imperative to migrate. become smaller, the so-called Allee effect (Allee 1949). Small populations of < 8 plants are sterile, apparently due to inbreeding depression rather CONCLUSIONS than lack of pollinators, despite being as vigorous (vegetative and flower production) as plants from For B. goodii and B. gardneri, the major obstacle to larger populations (Lamont et al. 1993). Indeed, persistence is clearing of land for agriculture and the Allee effect reduces seed production/plant to roads, which leads to habitat fragmentation, and some extent in all populations, especially those with reduction in routes for plants to migrate along in < 80 plants. Similar effects are likely in B. gardneri. the event of climate change. Short-distance seed Hence, in the unlikely event of successful migration dispersal in these prostrate species will result in to a new site, small founder populations may be rates of migration far too slow to keep up with the sterile or have very low rates of population growth. predicted movement of the climatic envelope. This Our data show that there was less than a 1 in 500 may be a general rule for taxa with short-distance chance of intact seeds of B. goodii stored on the dispersal. Despite the fact that B. gardneri has 177 plant becoming a one year-old seedling after fire, populations, > 10 times more than B. goodii, they compared with a 1 in 3 chance of manually removed have the same distribution of population sizes. This seeds sown at the same location reaching the same suggests that habitat fragmentation has had similar stage. Larger B. goodii populations have extremely effects on both species. Furthermore, without stable population dynamics (Witkowski and meaningful conservation outside nature reserves, Lamont 1997; Dreschler et al. 1999), and are likely the intervening areas between nature reserves will to persist for long periods. However, small B. goodii not sustain these species. As shown for these bank- populations (and probably B. gardneri as well) sia species, land-use change is now, and for some resulting from habitat fragmentation, represent decades will probably remain, the single most examples of ‘extinction debt’ (Lande 1987; Tilman important of the many interacting components et al. 1994; McCarthy et al. 1996). Even in the of global change affecting ecological systems. absence of further land clearing and active planting Resilience to marked climatic change is already of seeds/seedlings, these populations will doubtless built into these species, decreasing the imperative become extinct eventually. In particular, species to migrate. Historically, ecology has concentrated with poor dispersal abilities, such as these banksias, on small-scale studies, but in order to determine may be more susceptible to local extinctions in the responses of species to potential future climate response to habitat fragmentation. Owing to the change, large-scale studies, as shown by the trans- almost immortal nature of these banksias, particu- locations experiment here, are needed to ade- larly B. goodii, and their apparently much greater quately and reliably make predictions (Gaston climatic tolerances than their present ranges would 2003). The gradient study has provided a very use- suggest, it is likely that they will hang on in situ for ful contrast to the typical, often depressing, results centuries at their present locations under sub- of bioclimatic simulations in response to predicted optimal climatic conditions. This might involve future climate change. This contrast begs the a slow decrease in population sizes from one dis- question of how many other species show similar turbance to the next (sensu the ‘storage effect’ of resilience to climate change as these Banksia spp. lottery models; Warner and Chesson 1985), even if In addition, can one develop an a priori predictive International Journal of Biodiversity Science and Management 69 Resilience to global change in banksias Witkowski and Lamont model that could determine whether a species is Malcolm Briggs and Mike O’Donoghue of the ‘resilient’ to climate change? Western Australian Department of Conservation and Land Management for logistic support. This paper was prepared while Ed Witkowski was an ACKNOWLEDGEMENTS Adjunct Senior Fellow at Curtin University. This We thank Neil Gibson, Andrew Kennedy and project was funded by Environment Australia Dave Richardson for assistance with the BIOCLIM (Canberra), Roadside Conservation Committee analyses, Craig Walton and Darryl Abbott for field (Perth), Richard Ward Endowment Trust and Wits and laboratory assistance, and Laurie Anderson, University (Johannesburg). REFERENCES Allee WC. Group survival value for Philodina roseola, Cowling RM, Lamont BB and Enright NJ. Fire and a rotifer. Ecology 1949;30:395–7 management of south-western Australian bank- Allison HE and Hobbs RJ. Resilience, adaptive capa- sias. Proceedings of the Ecological Society of Australia city, and the “lock-in trap” of the Western 1990;16:177–83 Australian agricultural region. http/www.ecology Drechsler M, Lamont BB, Burgman MA, Akcakaya andsociety.org/vol9/iss1/art3. Ecology and Society HR, Witkowski ETF and Supriyadi. Modelling 2004;9(1):3 the persistence of an apparently immortal Banksia Bartlett WM. Western Australian Year Book. Perth: species after fire and land clearing. Biological Australian Bureau of Statistics, Western Australia Conservation 1999;88:249–59 Office; 1975 Drake J, di Castri F, Groves R, Kruger F, Mooney HA, Beard JS. A new phytogeographical map of Western Rejmanek M and Williamson M (eds). Biological Australia. Western Australian Herbarium Research invasions: a global perspective. New York: Wiley; 1989 Notes 1980;3:37–58 Eeley HAC, Lawes MJ and Piper SE. The influence of Bergelson J, Newman JA and Floresroux EM. Rates of climate change on the distribution of indigenous weed spread in spatially heterogeneous environ- forests in Kwazulu-Natal, South Africa. Journal of ments. Ecology 1993;74:999–1011 Biogeography 1999;26:595–617 Brown JH, Stevens GC and Kaufman DM. The Enright NJ and Lamont BB. Recruitment variability in geographic range: size, shape, boundaries, and the resprouting shrub Banksia attenuata and non- internal structure. Annual Review of Ecology and sprouting congeners in the northern sandplains Systematics 1996;27:597–623 heaths of south-western Australia. Acta Oecologica Burgman MA and Lamont BB. A stochastic model 1992;13:727–42 for the viability of Banksia cuneata populations: Gaston KJ. The structure and dynamics of geographic environmental, demographic and genetic effects. ranges. Oxford Series in Ecology and Evolution. Journal of Applied Ecology 1992;29:719–27 Oxford: Oxford University Press; 2003 Busby JR. Potential impacts of climate change on George AS. The genus Banksia L.f. (Proteaceae). Australia’s flora and fauna. In Pearman GI (ed.), Nuytsia 1981;3:239–473 Greenhouse: planning for climate change. Melbourne: Greuter W. Extinctions in the Mediterranean areas. CSIRO; 1988:387–98 Philosophical Transactions of the Royal Society of Busby JR. BIOCLIM – A bioclimatic analysis and pre- London B 1994;344:41–6 diction system. In Margules CR and Austin MP Green JW. Census of Western Australian plants. South (eds), Nature conservation: cost effective biological Perth: Western Australian Herbarium; 1985 surveys and data analysis. Canberra: CSIRO; Griffin EA and Associates. Vegetation survey of Bakers 1991:64–8 Junction and Millbrook Nature Reserve. Unpublished Clark JS, Fastie C, Hurtt G, Jackson ST, Johnson C, et al. Report No. 23. Perth: Department of Conservation Reid’s paradox of rapid plant migration. Bioscience and Land Management; 1985 1998;48:13–24 He TH, Krauss SL, Lamont BB, Miller BP and Cowling RM and Lamont BB. On the nature of Enright NJ. Long distance seed dispersal in a Gondwanan species flocks: diversity of Proteaceae metapopulation of Banksia hookeriana inferred in Mediterranean south-western Australia and from a population allocation analysis of Amplified South Africa. Australian Journal of Botany 1998; Fragment Length Polymorphism data. Molecular 46:335–55 Ecology 2004;13:1099–109 70 International Journal of Biodiversity Science and Management Resilience to global change in banksias Witkowski and Lamont Henderson-Sellers A and Blarg R. The Greenhouse Effect: Australia. Australian Journal of Botany 1995;43: Living in a warmer Australia. Kensington: NSW 283–303 University Press; 1989 Lamont BB, Wills RT and Witkowski ETF. Threats to Higgins SI, Nathan R and Cain ML. Are long-distance the conservation of southwestern Australian dispersal events in plants usually caused by non- Proteaceae. Acta Horticulturae 1995;387:9–18 standard means of dispersal? Ecology 2003a;84: Lamont BB and Witkowski ETF. A test for lottery 1945–56 recruitment among four banksia species based Higgins SI, Lavorel S and Revilla E. Estimating plant on their demography and biological attributes. migration rates under habitat loss and fragmenta- Oecologia 1995;101:299–308 tion. Oikos 2003b;101:354–66 Lande R. Extinction thresholds in demographic Hope P. Projected future changes in synoptic systems models of territorial species. American Naturalist influencing southwest Western Australia. Climate 1987;130:624–35 Dynamics 2006; in press Lindenmayer DB, Nix HA, McMahon JP, Hutchinson Hope P, Drosdowsky W and Nicholls N. Shifts in the MF and Tanton MT. The conservation of synoptic systems influencing southwest Western Leadbeater’s possum, Gymnobelideus leadbeateri Australia. Climate Dynamics 2006; in press (McCoy): a case study of the use of bioclimatic Hopper S. Biogeographical aspects of speciation in modelling. Journal of Biogeography 1991;18:371–83 the south-west Australian flora. Annual Review of McCarthy MA, Lindenmayer DB and Drechsler M. Ecology and Systematics 1979;10:399–422 Extinction debts and risks faced by abundant Hopper S and Gioia P. The Southwest Australian species. Conservation Biology 1997;11:221–6 Floristic region: evolution and conservation of a Nix H. Environmental determinants of biogeography global hot spot of biodiversity. Annual Review of and evolution of Terra Australis. In Barker WR and Ecology and Systematics 2004;35:623–50 Greenslade PJM (eds), Evolution of the flora and Hopper S, van Leeuwen SJ, Brown A and Patrick S. fauna of arid Australia. Frewville: Peacock; Western Australia’s Endangered Flora. Perth: Depart- 1982:47–66 ment of Conservation and Land Management; Nix H. A biogeographical analysis of Australian elapid 1990 snakes. In Longmore R (ed.), Atlas of elapid snakes of Huntley BJ. How vegetation responds to climate Australia. Canberra: Bureau of Flora and Fuana; change: evidence from palaeovegetation studies. 1986:4–15 In Pernetta JC, Leemans R, Endler D and Pittock AB. Actual and anticipated changes in Humphrey S (eds), Impacts of climate change on eco- Australia’s climate. In Pearman GI (ed.), Green- systems and species: environmental context. Gland, house: planning for climate change. Australia: Division Switzerland: IUCN; 1995:43–63 of Atmospheric Research, CSIRO; 1988:35–51 IUCN. IUCN Red List Categories v. 3.1. IUCN Species Richardson DM, Cowling RM, Lamont BB and van Survival Commission, Gland, Switzerland; 2001 Hensbergen HJ. Coexistence of Banksia species Kadmon R, Farber O and Danin A. A systematic in southwestern Australia: the role of regional analysis of factors affecting the performance of and local processes. Journal of Vegetation Science climatic envelope models. Ecological Applications 1995;6:329–42 2003;13:853–67 Rutherford MC and Westfall RH. Biomes of Southern Lamont BB and Connell SW. Biogeography of Banksia Africa: an objective categorization. Memoirs of the in southwestern Australia. Journal of Biogeography Botanical Survey of South Africa 1986;63:1–94 1996;23:295–309 Speck NH. The vegetation of the Darling-Irwin botanical Lamont BB, Hopkins AJM and Hnatiuk RJ. The flora – districts and an investigation of the distribution pattern composition, diversity and origins. In Pate JS and of the family Proteaceae in south western Australia. PhD Beard JS (eds), Kwongan: Plant life of the Sandplain. thesis. Perth: University of Western Australia; 1958 Nedlands: University of Western Australia Press; Taylor A and Hopper S. The Banksia Atlas. Canberra: 1984:27–50 Australian Government Publishing Service; 1988 Lamont BB, Klinkhamer PGL and Witkowski ETF. Thiele KR and Ladiges PY. A cladistic analysis of Bank- Population fragmentation may reduce fertility to sia (Proteaceae). Australian Systematic Botany zero in Banksia goodii: a demonstration of the Allee 1996;9:661–733 effect. Oecologia 1993;94:446–50 Tilman D, May RM, Lehman CL and Nowak MA. Lamont BB and Markey A. Biogeography of fire-killed Habitat destruction and the extinction debt. and resprouting Banksia species in south-western Nature 1994;371:65–6 International Journal of Biodiversity Science and Management 71 Resilience to global change in banksias Witkowski and Lamont Vitousek PM. Beyond global warming: ecology and Witkowski ETF and Lamont BB. Does the rare Banksia global change. Ecology 1994;75:1861–1976 goodii have inferior vegetative, reproductive or Warner RR and Chesson PL. Coexistence mediated ecological attributes compared with its widespread by recruitment fluctuations: a field guide to the co-occurring relative B. gardneri? Journal of storage effect. American Naturalist 1985;125:769–87 Biogeography 1997;24:469–82 Western Australian Government. Wildlife conserva- Witkowski ETF, Lamont BB and Connell SJ. Seed bank tion (rare flora) notice 1999. Government Gazette, dynamics of three co-occurring banksias in south WA; 17 December 1999:6194–9 coastal Western Australia: the role of plant age, Witkowski ETF. Growth and competition between cockatoos, senescence and interfire establish- seedlings of Protea repens (L.) L. and the alien ment. Australian Journal of Botany 1991;39:385–97 invasive, Acacia saligna (Labill.) Wendl. in relation Woodward FI and Williams BG. Climate and plant to nutrient availability. Functional Ecology 1991; distribution at global and local scales. Vegetatio 5:101–10 1987;69:189–97 72 International Journal of Biodiversity Science and Management http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png International Journal of Biodiversity Science & Management Taylor & Francis

Resilience of two Banksia species to global change: Comparing results of bioclimatic modelling, demographic and translocation studies

Download PDF
14 pages

Loading next page...
 
/lp/taylor-francis/resilience-of-two-banksia-species-to-global-change-comparing-results-tn1X0sR1qE

References (58)

Publisher
Taylor & Francis
Copyright
Copyright Taylor & Francis Group, LLC
ISSN
1745-1604
eISSN
1745-1590
DOI
10.1080/17451590609618099
Publisher site
See Article on Publisher Site

Abstract

International Journal of Biodiversity Science and Management 2 (2006) 59–72 Resilience of two Banksia species to global change: Comparing results of bioclimatic modelling, demographic and translocation studies 1 2 E. T. F. Witkowski and Byron B. Lamont Restoration and Conservation Biology Research Group, School of Animal, Plant and Environmental Sciences, University of the Witwatersrand, Johannesburg, South Africa Department of Environmental Biology, Curtin University of Technology, Perth, Australia Key words: Allee effect, Australia, BIOCLIM, climatic profiles, distribution range, habitat degradation, habitat loss, landscape fragmentation, long-distance dispersal, persistence, population modelling SUMMARY Banksia goodii (rare and endangered) and B. gardneri var. gardneri (widespread) are closely-related rhizomatous evergreen sub-shrubs of southwestern Australian scrub-heath and woodland. They have 17 and 177 known populations, respectively, mostly small remnants due to landscape fragmentation from agricultural activities. Bioclimatic pro- files developed using BIOCLIM indicate that B. gardneri tolerates a wider range of climatic conditions than B. goodii, which has a very narrow predicted range. These results do not match what one would predict from their comparative biology. Specifically, their post-fire survival and resprouting vigour, rates of seedling growth and soil penetration, and sus- ceptibility to seedling predators are similar. A field trial established along a steep climatic gradient of growing season length, showed that both species could extend their ranges beyond the distributions predicted by BIOCLIM, especially B. goodii. Seedlings of both species survived for at least 8 years at sites with two (but not three) months shorter and one month longer growing season than experienced by natural B. goodii populations. The rarity of B. goodii is a result of its recent origin, dispersal limitation, possibly habitat special- ization (dense woodland), and the impacts of habitat degradation and fragmentation within its current range. Under these circumstances, it is highly improbable that any sort of bioclimatic modelling could predict its potential climatic envelope. A stage-based model of B. goodii shows that under natural conditions the species is stable because of extremely low natural adult mortality (indeed, undetected), but plant losses due to human interference cannot be compensated due to its low levels of sexual reproduction. Population growth of B. goodii, and its potential for recovery, depends on population size and survival of seedlings and juveniles. Reproductive output per adult increases with pop- ulation size, with populations of < 8 plants being sterile. Conservation management should target these factors as well as prevent the destruction of existing adults through additional land clearing or other threatening processes. Thus, habitat degradation, fragmentation and loss are likely to have a greater influence than any predicted global Correspondence: E. T. F. Witkowski, Restoration and Conservation Biology Research Group, School of Animal, Plant and Environmental Sciences, University of the Witwatersrand, Private Bag 3, PO Wits, 2050, Johannesburg, Gauteng, South Africa. Email: ed@gecko.wits.ac.za 59 Resilience to global change in banksias Witkowski and Lamont climate change on species survival. Translocation experiments appear to have far more scope in predicting the outcomes of climate change than bioclimatic modelling. As a con- sequence of poor dispersal ability and low reproductive rates, as well as habitat fragmenta- tion, these banksias cannot migrate as the climate changes, but they already have a wide climatic tolerance and seem likely to withstand climate change in situ. INTRODUCTION Global environmental change is occurring at an and understanding the present-day climatic limits unprecedented rate (Vitousek 1994). With current of species, allows for the exploration of how distri- levels of increase, CO is expected to double pre- butional patterns might change in response to cli- industrial concentrations by 2200 and result in a matic change (Busby 1988; Huntley 1995). This will predicted increase in mean global surface air aid in the development of strategies for long-term temperature of 2°C (range 0.9–3.5°C) by 2100 management and conservation of species. (Vitousek 1994). There is also an anticipated in- The South-West Botanical Province of Western crease in the intensity of the hydrological cycle, Australia (SWBP; Beard 1980) is a mediterranean with an increasing frequency of severe floods and climate region renowned for its floristic richness droughts. This unprecedented rate of climate (Hopper 1979; Lamont et al. 1984), and is a recog- change begs the questions of whether plant species nized hotspot for floristic diversity (Hopper and will be ‘resilient’ to this change, what functional Gioia 2004). SWBP also has the highest concentra- characteristics are associated with species that tion of declared rare flora in Australia, and it is might persist, and can these be predicted. estimated that Western Australia has the highest Species ranges have expanded/contracted rates of plant extinction and endangerment of during interglacial/glacial periods and moved all Mediterranean climate regions (Greuter 1994). polar-wards/equatorially (Clark et al. 1998). How- Banksia, a pan-Australian genus, has been des- ever, can species track current rapid climatic cribed as ‘the most characteristic genus’ of the changes and remain within areas which match their SWBP (Speck 1958), where it may dominate vegeta- bioclimatic niches? There are few direct tests of tion on the poorest soils. Climate, most notably species range extension within natural habitats. rainfall, has changed in this region in recent times. The success of introduced species throughout the Pittock (1988) noted that there has been a mean globe indicates that many, probably most, species 3–5% drop in rainfall per decade over the preced- are not present in all suitable habitats. While physi- ing 70 years in SW Australia. Clearing for agricul- cal barriers to dispersal prevent them from occur- ture is a likely contributory cause to this. If this ring in distant but otherwise suitable areas, it is trend continues over the next 50 years (a drop in unknown whether plant species can extend their rainfall of 20%), there will be a dramatic decline in climatic ranges per se. Clearly they do extend their the population size of some species (Burgman and ranges over evolutionary time, but at a rate too slow Lamont 1992). Modelling the greenhouse effect to match the speed of present climate change. has indicated drier and warmer winters can be In general, alien plants on new continents occur expected in SW Australia by 2030, although within their native climatic ranges (Drake et al. some models predict no change (Pittock 1988; 1989, and references therein). Henderson-Sellers and Blarg 1989). Extensive At the regional (biome) and global scales, cli- areas of this region have recently received less than mate defines the broad limits to the distribution of 80% of the long-term average since 1975 in response plant taxa and the dominance of plant life forms to shifts in the synoptic weather systems influencing (e.g. Rutherford and Westfall 1986; Woodward and SW Australia (Hope et al. 2006). Furthermore, Williams 1987). Bioclimatic modelling has been projected changes in the synoptic systems in used to determine the influence of particular clima- response to increasing concentrations of green- tic variables on the distribution of species, vegeta- house gases will lead to increasingly dry conditions tion types and biomes (Busby 1986; Hill et al. 1988; in the twenty-first century (Hope 2006). Eeley et al. 1999). Recognizing the dynamic nature The 63 Banksia species found in Western of species distributions from the past to the present, Australia are thought to be confined to the SW 60 International Journal of Biodiversity Science and Management Resilience to global change in banksias Witkowski and Lamont corner by climate. The genus cannot tolerate namely brevidentata, gardneri and hiemalis, each of drought conditions corresponding to a rainfall of which is more widespread than B. goodii, but only < 250 mm/annum (Lamont and Connell 1996; var. gardneri occasionally co-occurs with B. goodii Cowling and Lamont 1998). Within this area (George 1981; Taylor and Hopper 1988). The two species composition and density vary with substrate, study species are very similar in growth and leaf landscape features, climate, vegetation type and forms, floral morphology and reproductive biol- fire history. We pose the question: what is the rela- ogy, and have the same pollinators (George 1981; tive importance of expected future changes in Taylor and Hopper 1988; Witkowski and Lamont climate as a result of increased atmospheric con- 1997). Reproductive potential, either sexual or centrations of CO and other ‘greenhouse’ gases, vegetative, is low in both species, and they have compared with other forms of human-mediated slow rates of growth as seedlings, or resprouts after environmental change on the distribution of fire. Seed production by both species is lower than banksias in the SWBP? Many studies have used for many other banksias (Witkowski et al. 1991; bioclimatic modelling to examine potential changes in the distributions of species or eco- systems/biomes in response to predicted climate change scenarios (Kadmon et al. 2003). However, few actual field manipulations have been under- taken to determine whether a species can survive beyond its current range. This study aims to determine whether bioclimatic and population modelling can predict the actual and potential distributions of banksias under global climate change scenarios. An experimental translocation study along an extensive climatic gradient (length of growing season) was established for two Banksia species in order to determine the potential for climatic range extension. Population modelling was also undertaken to assess the effects of various factors on the long-term persistence of populations of various sizes (Dreschler et al. 1999). STUDY SPECIES AND AREA Our study species were the prostrate Banksia goodii (nomenclature follows Green 1985), a rare and endangered species restricted to 17 populations distributed between Albany and the Porongurup Range, within a small area of 38 km (N–S) by 14 km (E–W) in the woodlands of southwestern Australia (George 1981; Taylor and Hopper 1988; Witkowski and Lamont 1997), and its more widespread and Figure 1 Distribution of Banksia goodii and B. gardneri. abundant close relative, B. gardneri var. gardneri (a) Distributions predicted by BIOCLIM, plus trans- (henceforth B. gardneri) (Figure 1). The biology of location sites. 1 = Walpole National Park, 2 = Albany, these species has been studied in detail (Lamont 3 = Mt Barker, 4 = Stirling Range, 5 = Katanning. Con- et al. 1993; Witkowski and Lamont 1997). Banksia tours for length of the growing season in months goodii is considered cladistically terminal within the (Prescott formula) are also shown. (b) Actual known section Prostratae (Thiele and Ladiges 1996), populations of B. gardneri compared with its BIOCLIM which consists of 6 species of Banksia with a pros- predicted distribution. (c) Actual known populations trate growth form (George 1981). Its closest relative of B. goodii compared with its BIOCLIM predicted is B. gardneri, which occurs as one of three varieties, distribution International Journal of Biodiversity Science and Management 61 Resilience to global change in banksias Witkowski and Lamont Witkowski and Lamont 1997), but similar to some climate, vegetation and biogeography of the whole other resprouters (Cowling et al. 1990). Pre- region with respect to banksias are given in Lamont dispersal granivory by weevils is extremely variable and Markey (1995), Richardson et al. (1995), between populations of both species and some- Lamont and Connell (1996) and Witkowski and times most seeds are devoured. B. gardneri does Lamont (1997). not reproduce vegetatively, whereas B. goodii does, albeit at a slow rate (Witkowski and Lamont 1997). B. goodii is classified as rare and endangered MATERIALS AND METHODS (Western Australian Government 1999; IUCN Bioclimatic modelling 2001). Both species resprout after fire, with no evidence BIOCLIM is a bioclimatic analysis and prediction of mortality after several fires. Two field trials system used to predict the distribution of species and one glasshouse trial consistently showed that (or other ‘entities’; Nix 1982, 1986; Busby 1988, B. gardneri has a higher germination percentage 1991). BIOCLIM generates a set of variables (clima- than B. goodii, which partially trades off with vegeta- tic indices) considered to have biological signifi- tive reproduction in B. goodii, while seed store per cance and that describe the range, extremes and adult does not differ between species (Witkowski seasonality of climatic conditions. These variables and Lamont 1997). Both species are probably are interpolated across the geographical surface (at equally susceptible to the stochastic effects of (a) various scales on the basis of longitude, latitude and generalist fungal pathogens (especially Phytoph- altitude). By matching the known distribution of a thora cinnamomi and stem cankers (Witkowski et al. taxon to the variable surfaces, a statistical summary 1991; Lamont et al. 1995)); (b) changes in fire of the values of the climatic indices for that taxon season/intensity/frequency on seed store and is produced. This summary, referred to as a bio- release, and adult survival; (c) post-fire environ- climatic profile or envelope, provides a quantitative mental conditions on seedling establishment and description of the climatic environment occupied adult survival (Lamont and Witkowski 1995); (d) by the taxon (Nix 1986). Points on the variable grid granivory and seedling herbivory (Cowling et al. surfaces that match the bioclimatic profiles of the 1990; Witkowski et al. 1991); (e) land degradation species can be identified to delimit its potential due to poor pastoralism practices (habitat modi- distribution (i.e. the area of suitable climatic condi- fication); (f) clearing for roads and agriculture tions; Busby 1991; Lindenmayer et al. 1991). The (habitat fragmentation and loss); and (g) alien underlying assumption of most BIOCLIM analyses plant invasions. is that species can only colonize and survive in Extensive clearing for agriculture (mainly cere- areas with climates fitting their current bioclimatic als, but also sheep pasture in the study area, but profiles. The model also does not consider the cereals in other parts of SW Australia) and roads potential interactions between climatic variables over a period of 100 years (Allison and Hobbs 2004) (Nix 1986). has fragmented the landscape of southwestern Seventeen distribution records of B. goodii and Australia. Most B. goodii populations are now only 177 records of B. gardneri (longitude, latitude and represented by remnants on roadsides, while altitude) were analysed using BIOCLIM, against others occur on farms, in state forest and a nature 24 climatic variables, including various tempera- reserve. The rarity of B. goodii is a function of its ture and rainfall indices. The outputs comprised small geographic range rather than scarcity within descriptions of the climatic profiles of each species its range per se (Witkowski and Lamont 1997), and and predicted distribution maps. In addition, the its small geographic range appears related to its locations of the five climatic gradient translocation more recent origin relative to B. gardneri (Thiele sites (see below) were also analysed using BIOCLIM and Ladiges 1996). to obtain their climatic profiles. The study area was centered 400 km south of Perth (35°00′S, 117°55′E) near Albany. The soils Seedling establishment along a steep climatic gradient of the native banksia populations are infertile sands usually overlying laterite (Griffin et al. 1985; Sixty intact seeds of B. goodii and B. gardneri Witkowski and Lamont 1997). More details on the were planted at each of five sites, which represented 62 International Journal of Biodiversity Science and Management Resilience to global change in banksias Witkowski and Lamont a climatic gradient of growing season length. 11–100, and (c) > 100 (6, 6 and 5 for B. goodii, and Seeds were collected from 40 serotinous cones 49, 83 and 45 for B. gardneri, respectively (Taylor harvested from 20 randomly selected plants from and Hopper 1988)), no differences in population the largest and most fecund population, and hence size is evident between species (X = 0.87, d.f. = 2, likely to have the highest genetic diversity (Lamont p = 0.65). Population sizes of B. gardneri on farms, et al. 1993). Seeds were extracted from the cones roadsides and reserves clearly match the variation by burning them until the follicles opened, mimick- in population sizes of B. goodii, both within the ing the natural seed release process (Witkowski wooded region occupied by B. goodii and through- et al. 1991). Using the Prescott formula, the out its range. number of growing months can be determined from mean monthly rainfall and evaporation Distribution and bioclimatic modelling (Bartlett 1975). Growing seasons of six (Katanning), seven (N. Stirling Range), eight Banksia goodii has a very restricted distribution, (Mt Barker), nine (Albany) and ten (Walpole- within which many of its 17 populations are fairly Nornalup National Park) months were chosen disjunct, while B. gardneri is far more widespread (Figure 1). This is a steep climatic gradient of and occupies a greater altitudinal range (Figure 1; from a six- to ten-month growing season over a Table 1). Both species occur in areas with similar distance of only about 200 km. Within these temperatures, but B. goodii receives higher rainfall areas, sites were selected with soil and other and slightly lower temperature extremes than environmental conditions which, as far as possible, B. gardneri (Table 1). matched typical B. goodii sites. Three composite Similarly, the distribution of B. gardneri pre- soil samples (0–100 mm depth) were collected per dicted on the basis of its bioclimatic profile, greatly site and soil depth was determined. exceeded and included that of B. goodii (Figure 1a). In late June 1990, a 30-m plot was cleared of The predicted distribution of B. gardneri generally vegetation (above- and belowground) at each site, included its distribution records, but omitted some and seeds sown into six 1-m quadrats, spaced 1 m marginal populations in the N, and especially apart. Seedling emergence and survival were moni- a block in the NE corner, but otherwise filled in tored in September, October and December 1990, the gaps between the remaining populations, and April, June and November 1991, May 1992, January extended its current distributional limits to the S, W 1993, January 1996 and June 1999. and E (Figure 1b). Overall extensions to the drier N, and wetter W and S were negligible. The pre- dicted distribution of B. goodii greatly exceeded the total area currently occupied but still omitted some Soil analyses marginal populations, most notably the two in the Soil samples were passed through a 2-mm sieve and SW (Figure 1c). There was no predicted extension −1 analyzed for NO ,NH (0.1 mol L KCl extraction 3 4 to the N or S, a slight contraction in the W, and and Varian autoanalyzer), available P (citric acid slight extension to the E. extract and molybdenum blue spectrophoto- metry), total Fe and K, and exchangeable K, Ca, Mg −1 and Na (0.1 mol L ammonium acetate extraction Seedling establishment along a climatic gradient and atomic absorption spectrometry), organic The climatic profiles of the five translocation sites carbon (Walkley Black) and pH (1 : 5 H O w/v) are given in Table 2. Seedlings of both species by CSBP and Farmers, Bayswater, Perth. Soil emerged during late winter and spring at all five texture was analysed using the hydrometer method. sites (Figure 2). Total emergence was 14.3% for B. goodii and 32.7% for B. gardneri. All seedlings at the driest site, Katanning, died after the first summer drought in 1991, with B. goodii succumbing RESULTS about 2 months before B. gardneri. Mortality Comparison of population sizes was also complete at the Albany site with few seed- Comparing the numbers of populations of each lings remaining by early summer 1991, and com- species which fall into the ranges (a) 1–10, (b) plete absence by the following autumn. This was International Journal of Biodiversity Science and Management 63 Resilience to global change in banksias Witkowski and Lamont Table 1 Position, altitude and climate profiles of the rare species, Banksia goodii, compared with its widespread closest relative, B. gardneri. Climate profiles were produced using BIOCLIM (Busby 1991). Data are means ± SD. P values are based on t-tests (unequal variances); CV = coefficient of variation Environmental variables B. goodii B. gardneri P Number of localities 16 177 Altitude (masl) 83.5 ± 26.7 169.3 ± 89.9 < 0.001 Latitude (°E) 34.87 ± 0.04 34.58 ± 0.21 < 0.001 Longitude (°S) 117.76 ± 0.13 118.14 ± 0.38 < 0.001 Mean annual temperature (°C) 15.3 ± 0.1 15.1 ± 0.2 < 0.001 Mean annual temperature range (°C) 19.9 ± 0.4 21.5 ± 1.4 < 0.001 Mean minimum temperature for coldest month (°C) 6.4 ± 0.3 5.6 ± 0.6 < 0.001 Mean maximum temperature for warmest month (°C) 26.3 ± 0.2 27.1 ± 0.9 < 0.001 Mean temperature for coldest 3 months (°C) 11.5 ± 0.2 10.9 ± 0.6 < 0.001 Mean temperature for warmest 3 months (°C) 19.4 ± 0.1 19.7 ± 0.4 < 0.001 Mean temperature for wettest 3 months (°C) 11.7 ± 0.3 11.5 ± 0.7 0.229 Mean temperature for driest 3 months (°C) 19.4 ± 0.1 19.7 ± 0.4 < 0.001 Mean annual rainfall (mm) 878 ± 62 623 ± 148 < 0.001 Highest mean monthly rainfall (mm) 129 ± 13 85 ± 26 < 0.001 Lowest mean monthly rainfall (mm) 24 ± 1 18 ± 3 < 0.001 CV of monthly rainfall (%) 51.5 ± 2.8 46.1 ± 4.9 < 0.001 Mean rainfall for wettest 3 months (mm) 352 ± 34 238 ± 67 < 0.001 Mean rainfall for driest 3 months (mm) 84 ± 2 64 ± 11 < 0.001 Mean rainfall for coldest 3 months (mm) 332 ± 30 231 ± 63 < 0.001 Mean rainfall for warmest 3 months (mm) 85 ± 4 65 ± 12 < 0.001 Table 2 Position, altitude, length of growing season (Prescott formula) and climate profiles for the five sites used in the climate gradient tranlocation study. CV = coefficient of variation Walpole Stirling National 1 1 1,2 Site Katanning Ranges Mt Barker Albany Park Length of growing season (months) 6 7 8 9 10 Altitude (masl) 320 250 300 80 20 Latitude (°E) 33.70 34.30 34.63 34.88 34.97 Longitude (°S) 117.55 117.77 117.67 117.78 116.99 Mean annual temperature (°C) 15.3 15.1 14.7 15.3 15.6 Mean annual temperature range (°C) 25.0 23.4 22.9 19.8 18.6 Mean temperature for coldest 3 months (°C) 10.1 10.4 10.0 11.5 12.1 Mean temperature for warmest 3 months (°C) 21.2 20.3 19.8 19.4 19.6 Mean temperature for wettest 3 months (°C) 11.1 11.4 10.0 11.5 12.2 Mean temperature for driest 3 months (°C) 21.2 20.3 19.8 19.4 19.6 Mean minimum temperature for coldest month (°C) 4.6 5.0 4.8 6.4 7.5 Mean maximum temperature for warmest month (°C) 29.6 28.4 27.8 26.2 26.1 Mean annual rainfall (mm) 465 483 749 910 1197 Highest mean monthly rainfall (mm) 78 70 111 135 198 Lowest mean monthly rainfall (mm) 12 14 22 24 27 CV of monthly rainfall (%) 61.5 51.0 50.4 52.5 62.0 Mean rainfall for wettest 3 months (mm) 212 197 297 366 537 Mean rainfall for driest 3 months (mm) 42 50 73 85 92 Mean rainfall for coldest 3 months (mm) 211 197 297 346 483 Mean rainfall for warmest 3 months (mm) 42 50 73 85 102 Naturally occurring populations of Banksia gardneri present Naturally occurring populations of B. goodii present 64 International Journal of Biodiversity Science and Management Resilience to global change in banksias Witkowski and Lamont −1 Table 3 Soil physical and chemical (µgg dry mass) properties (mean ± SD of three samples) at the five sites where seeds were sown along a climatic gradient (length of the growing season) in southwestern Australia. Values with the same letter are not significantly different at p < 0.05 (Student-Neuman-Keuls multiple range test) 1 1 1,2 Site Katanning Stirling Range Mount Barker Albany Walpole Length of growing season 6 7 8 9 10 (months) a a a a a Available P 1.0 ± 0.0 1.0 ± 0.0 1.0 ± 0.0 1.7 ± 1.2 1.0 ± 0.0 a a a a a NO 2.0 ± 0.0 2.0 ± 0.0 2.0 ± 0.0 2.0 ± 0.0 2.0 ± 0.0 a a a a a NH 3.7 ± 2.1 4.0 ± 0.0 3.3 ± 0.6 5.0 ± 1.0 5.3 ± 0.6 b b ab a b Total K 31 ± 6 30 ± 3 45 ± 3 57 ± 19 24 ± 2 c c b a c Organic C (%) 0.2 ± 0.0 0.4 ± 0.2 1.6 ± 0.9 2.6 ± 0.6 0.6 ± 0.1 b bc a c d Fe 136 ± 23 93 ± 17 464 ± 214 72 ± 12 31 ± 4 a a a a a pH 5.9 ± 0.1 5.9 ± 0.1 6.1 ± 0.7 5.6 ± 0.2 5.6 ± 0.2 b b a a b Exchangeable Ca 37 ± 1 78 ± 40 425 ± 197 430 ± 80 100 ± 10 b b a a b Exchangeable Mg 15 ± 1 19 ± 12 81 ± 45 100 ± 45 38 ± 5 b b b a b Exchangeable Na 5 ± 1 5 ± 1 15 ± 7 26 ± 7 9 ± 2 b b a a b Exchangeable K 10 ± 5 14 ± 2 35 ± 5 38 ± 16 5 ± 2 b b a b b Gravel (%) 0.1 ± 0.0 0.4 ± 0.2 4.0 ± 1.1 1.0 ± 0.5 0.1 ± 0.0 a b b c b Coarse sand (%) 32.4 ± 4.5 20.6 ± 8.8 11.7 ± 1.4 2.8 ± 0.8 21.0 ± 2.0 b b b c a Medium sand (%) 45.8 ± 3.2 44.1 ± 3.2 40.3 ± 4.1 12.0 ± 1.5 54.9 ± 0.2 d c b a d Fine sand (%) 19.9 ± 1.4 32.1 ± 5.7 42.9 ± 5.2 76.8 ± 3.9 21.0 ± 2.0 c c b a c Silt (%) 1.6 ± 0.3 2.8 ± 0.5 4.7 ± 0.9 7.9 ± 1.5 3.0 ± 0.2 a a a a a Clay (%) 0.3 ± 0.1 0.5 ± 0.3 0.4 ± 0.0 0.5 ± 0.2 0.2 ± 0.0 Soil depth (mm) > 1000 300 150 500 > 1000 Naturally occurring populations of Banksia gardneri present Naturally occurring populations of B. goodii present due to herbivory, most probably by kangaroos DISCUSSION (Macropodidae), bandicoots (Dasyuridae) and BIOCLIM analysis showed that B. gardneri usually wingless grasshoppers (Acrydidae). The Albany occurs in areas with similar temperatures but lower translocation experiment was, however, repeated at rainfall than B. goodii. This is supported by the two additional sites, separated by 3 km, where both observation that B. gardneri took slightly longer species naturally occurred. Herbivory was found to than B. goodii to be eliminated from the driest site, be site-specific, and protection from mammals at Katanning, and had greater survival in the next one site resulted in high levels of seedling recruit- driest, Stirling Range. Furthermore, for the past ment (Witkowski and Lamont 1997). At the remain- 20 million years, the area within which B. goodii ing sites along the climatic gradient (Stirling presently occurs, has been wetter than the overall Ranges, Mt Barker and Walpole-Nornalup National area within which B. gardneri occurs (Lamont and Park), seedling survival was high and similar Markey 1995). Greater tolerance to drought may between species, and remained unchanged allow B. gardneri to survive adverse conditions better throughout 1992 and 1993. By 1993, plants at than B. goodii. The creeping/prostrate growth form Mt Barker had shown the greatest growth. There of both species would be disadvantageous in the was essentially no change to this scenario between wettest areas of the southwest (represented by 1993 and mid-1999, except that the Stirling Range Walpole-Nornalup National Park), where suppres- site was accidentally cleared. sion by excessive litter and/or competition for light The soils of the five sites selected were similar from tall woodland/forest trees would be intense. (Table 3), consisting of over 90% sand, with no However, both species showed comparable levels differences in available P, NO and NH ,or inpH 3 4 of survival under both higher and lower moisture and clay content. Cation concentrations tended to conditions than found in their current ranges. be higher at Albany and Mt Barker. International Journal of Biodiversity Science and Management 65 Resilience to global change in banksias Witkowski and Lamont intervals result in more heavily shaded plants covered with litter, which inhibits both seed pro- duction and seedling establishment. Morphologi- cally, the broader and only slightly lobed leaf lamina of B. goodii , plus a 32% greater leaf area (Witkowski and Lamont 1997), makes it more suited to a shady habitat than B. gardneri. Species climatic profile models are designed to predict potential distribution ranges, but their accuracy depends totally on using data from actual distribution ranges. Hence, it is an explicit assumption in bioclimatic modelling in predicting changes in a species range in response to expected climate change, that the present range provides an adequate description of its climatic profile or envelope (Kadmon et al. 2003). Bioclimatic model- ling predicted that B. goodii would occupy an area, similar to, or slightly less than, its actual known present distribution. Outlying populations were excluded, while a continuous distribution was pro- duced around the ‘core’ of its current distribution. It is clear that when a species has a very restricted distribution, bioclimatic modelling via BIOCLIM may not totally predict its current geographic range. However, to be fair, the model fitted the dis- tribution data fairly well, and at a fine scale factors other than climate, such as topography, soils, and moisture availability, as well as biotic factors, come Figure 2 Emergence and number of seedlings of into play. For B. gardneri, which occupies a greater Banksia goodii and B. gardneri surviving over time at known range than B. goodii, BIOCLIM did little Katanning, Stirling Range, Mt Barker, Albany and more than define its current distributional bound- Walpole National Park aries, ‘filling in’ large areas to the W and SW but missing the NE corner (although extending poten- The growth of B. goodii seedlings/juveniles, for tial habitats 25 km to the E). example, was vigorous at Mt Barker, 25 km north BIOCLIM develops a ‘climate surface’ based on of its current range, with reduced vigour in the data from weather stations, and then interpolates Stirling Range (60 km N) and at Walpole between these point sources on the basis of latitude, (80 km SW). Seedlings of both species establish longitude and altitude. Because climate stations are equally fast, including rates of root penetration sparsely spread in SW Australia, the interpolated (an index of drought resistance among seedlings, ‘climate surface’ may be too coarse to predict in Witkowski 1991), which were identical. However, detail the present distribution of species with both species may be excluded once they are over- narrow distributions, such as B. goodii. Nevertheless, topped by the taller trees found at the wettest site, climate modelling may still be useful for species although B. gardneri is likely to be excluded first with broader distributions, as well as under condi- because B. goodii always occurs in wooded areas. tions of large changes in climate variables for On the other hand, B. gardneri genotypes of popula- species with narrow distributions (Kadmon et al. tions from woody sites (e.g. those co-occurring with 2003). In addition, B. goodii is a neo-endemic and is B. goodii) may be equally vigorous. probably in the process of extending its distribu- In the context of the present locations of tional range. The very poor dispersal ability of B. goodii, fragmentation of the landscape means B. goodii suggests that range extension in this greater variation in fire frequency. Longer fire species is extremely slow (see below). Thus, if there 66 International Journal of Biodiversity Science and Management Resilience to global change in banksias Witkowski and Lamont were a higher density of weather stations in the these Banksia spp. survived well beyond their native region, BIOCLIM would still not be able to predict distributions along the gradient, a longer-term the potential range of B. goodii because its distribu- perspective is needed, as persistent range extension tion is unlikely to be currently limited by climatic can only be confirmed if they are able to success- factors, but rather by dispersal barriers and insuffi- fully reproduce. Due to their long juvenile period cient time for expansion. This argument could be (Witkowski and Lamont 1997; Dreschler et al. extended to B. gardneri, which may also be in the 1999), the plants in the gradient study have not yet process of extending its range. reached reproductive age or size. An added compli- A very thorough assessment of the performance cation is that small B. goodii populations are sterile of climate profile models has been undertaken within their native distribution (see below), despite using the distribution of 192 woody species in profuse flowering and being of a similar size and Israel, a geographic region with similar steep vigour as highly fecund plants from the large popu- climatic gradients to the SWBP (Kadmon et al. lations (Lamont et al. 1993). Hence the gradient 2003). Predictably, accuracy tends to increase (and study plants may, in time, be found to be sterile, variance decrease) with increasing number of although not necessarily in response to an unsuit- distribution records (or observations), although able climate/environment, but due to small popu- this is not linear. Species with < 25 observations are lation size per se. Overall the gradient study has most adversely affected, those with 50 observations shown that the environmental tolerances of these are accurately modelled, and there is no detectable banksias are far greater than would be expected, improvement in accuracy above 100 observations. based on present distribution alone, and these Based on this factor alone, the B. gardneri model broad tolerances are not merely an ability to accli- (177 observations) should be more accurate than mate. Similar findings for other species are re- the B. goodii model (17 observations) (Figure 1). viewed in Gaston (2003) and Kadmon et al. (2003). Interestingly, Kadmon et al. (2003) found that the Given these results, it is unlikely that any improve- position of the species niche along a rainfall gradi- ments to BIOCLIM could adequately predict the ent, from the Mediterranean region to desert, had results obtained from the gradient study. a negative effect on predictive accuracy. Hence, A general pattern of range limitation in species Mediterranean species are less influenced by clima- is the relative importance of abiotic versus biotic tic conditions than desert species. Any factor pre- limiting factors. The distribution of the majority of venting a species from utilizing its potential climatic species appears to be delimited in one direction distribution, whether anthropogenic, biotic or by physical stresses and in another by biological dispersal, will lead to poor predictive accuracy stresses (Brown et al. 1996). In Western Australia (Kadmon et al. 2003), as concluded here for the major physical gradient is aridity (Lamont and B. goodii, and B. gardneri to a lesser extent. Connell 1996). B. goodii and B. gardneri are limited The translocation study showed that both by the aridity barrier to the N and E, and by the species could extend their ranges beyond those biotic barrier of dense shade (taller denser eucalypt predicted by BIOCLIM, and could occupy both forests) to the wetter W and SW of their respective much drier and much moister sites. In particular, ranges. The two species differ in that B. gardneri is BIOCLIM was a poor predictor of the potential more tolerant of aridity than B. goodii, while the distribution of B. goodii. It is only at the wettest site, latter may be more tolerant of deep shade. How- where the plants are presently situated in the open ever, considering anticipated climate change over (not under the shade of the typical eucalypt wood- the next 100–200 years in SW Australia (1–2 months land) that they are likely to be excluded by biotic reduction in length of the growing season at factors, when they eventually become heavily present rates; extrapolating from Pittock 1988), the shaded by dense forest canopy. These contrasting difference between the two species is minor. The results provide a warning to global change scientists present small range of the more recently evolved reliant on this type of climate modelling. Although B. goodii (relative to B. gardneri; Thiele and Ladiges clearly of immense value, modelling of species with 1996) can be explained primarily by evolutionary narrow ranges only provides hypotheses that factors, interacting with biological factors, such as should be tested where possible with field data. its exceptionally low rate of dispersal (see below). However, although the seedlings and juveniles of Thus, future changes in distribution are likely to be International Journal of Biodiversity Science and Management 67 Resilience to global change in banksias Witkowski and Lamont the result of the interwoven effects of both evolu- enough or quickly enough to keep within their tionary and environmental changes. present bioclimatic envelopes as defined by their The rapid rates of plant dispersal during glacial/ known distributional ranges. interglacial periods implied by past records (since The potential fate of B. gardneri and B. goodii the beginning of the Holocene; Clark et al. 1998) exemplifies that of many plant species with in- are far too high to have been produced by tradi- herent short-distance dispersal. In the event of tional notions of life history and dispersal mecha- predicted climate change, and in the absence nisms (Clark et al. 1998). Seed dispersal distances of of human-mediated conservation efforts, B. goodii these two prostrate Banksia spp. appear to be very could become extinct, while B. gardneri would prob- low. Seeds have wings for wind dispersal, but owing ably suffer a considerable reduction in its range. to the low heights of cones (< 30 cm above the soil However, other features of the biology of B. goodii surface), observed dispersal distances have all been will probably work against extinction. The highly < 1 m (unpublished observations), and even under fragmented landscapes of today, with extensive optimal windy conditions are unlikely to exceed unfavourable habitat barriers to banksia migration, 10 m. Typically, long-lived species, which resprout such as sheep pastures and roads, provide an addi- after recurrent fire (or other severe disturbances), tional and potentially insurmountable challenge. such as these banksias, often produce few seeds and Potential safe sites may be so few and far between seedlings (Witkowski and Lamont 1997). These that effective natural colonisation and migration banksias are serotinous (canopy seed storage), with will be virtually impossible, because there are too seed store peaking 15 years after fire, and declining few ‘stepping stones’. Little is known about the life thereafter (Dreschler et al. 1999). Short dispersal spans of resprouting banksias. For the shrubby distances are exacerbated by (a) low seed produc- Banksia attenuata, longevity has been estimated at tion and (b) infrequent dispersal events – seed 300 years and mortality after fire at about 5% release from the cones is triggered by fire, usually at (Enright and Lamont 1992). This species relies a frequency of 8–30 years. Such figures would only on occasional seedling recruitment after fire to allow these species to move 0.2 km over the next replace loss of some pre-fire plants. However, no 200 years, assuming fires every decade, and a very adult mortality of B. goodii or B. gardneri was evident optimistic distance of 10 m dispersal per fire. A following several fires. Zero recruitment (from 126 one-month reduction in growing season over that adults) and two recruits (from 144 adults) by four time is equivalent to a distance of 30–100 km winters after two fires for B. goodii indicates pro- (Figure 1). The probability of episodic long- longed population stability in the absence of distance dispersal to suitable distant sites (He et al. human interference. Zero recruitment of B. 2004) is also extremely low within this highly frag- gardneri populations was also found (Albany and mented landscape. Neither models of leptokurtic Mount Barker; unpublished data). In B. goodii,we dispersal profiles (fat-tailed curves) or non- estimate it takes about 20 years for recruits to standard mechanisms of dispersal (Clarke 1998; become fire-tolerant and start flowering, while fire Higgins et al. 2003a), except direct human manipu- retards flowering of adults for at least 3 years lation, are likely to make any meaningful difference (Lamont et al. 1993; unpublished) and we expect to these estimates due to the exceptionally low B. gardneri to be similar. For both species, we recom- levels of seed production and infrequency of seed mend a mean fire rotation of 10–15 years, and at release. Indeed, migration may even be halted regular intervals this should be extended to 15–20 altogether under these conditions (Higgins et al. years (Dreschler et al. 1999) on the grounds that this 2003b). Hence, the ranges of these banksias are should be sufficient time for seedlings to become highly likely to be dispersal-limited. Furthermore, fire-tolerant, keep litter in check, and ensure occa- the probable sizes and distribution of patches suit- sional rejuvenation of vegetative growth, while pro- able for Banksia seedling establishment (within this viding sufficient time for seedling establishment now highly disturbed and fragmented landscape) and substantial seed set. For B. goodii, longevity is indicates that even this very low rate of spread is estimated as 400–500 years, based on clump sizes probably very optimistic (Bergelson et al. 1993). In and growth rates (Witkowski and Lamont 1997, the event of predicted global climate change, these Dreschler et al. 1999), and slightly less for B. gardneri banksias are unlikely to be able to disperse far (300–400 years) due to smaller clump sizes and 68 International Journal of Biodiversity Science and Management Resilience to global change in banksias Witkowski and Lamont because, unlike B. goodii, it does not reproduce further ‘nibbling away’ of the populations (clear- clonally. ing of land for agriculture) is prevented. If in the Population numbers of B. goodii have decreased future, atmospheric CO can be brought under over the last 50–60 years due to clearing for agricul- control or reduced, and climate change is reversed, ture and for roads, resulting in smaller, fragmented these banksias may be able to persist over the populations. Furthermore, habitat fragmentation unfavourable period and begin to ‘thrive’ once has itself indirectly affected the reproductive poten- again. In summary, as shown by their biology and tial of B. goodii (and by comparison almost certainly the gradient study, B. goodii and B. gardneri are B. gardneri). Habitat fragmentation has resulted in long-term persisters, and resilience to marked reduced sexual reproduction in B. goodii. Mean climatic change is already built into these species, seed production per plant decreases as populations decreasing the imperative to migrate. become smaller, the so-called Allee effect (Allee 1949). Small populations of < 8 plants are sterile, apparently due to inbreeding depression rather CONCLUSIONS than lack of pollinators, despite being as vigorous (vegetative and flower production) as plants from For B. goodii and B. gardneri, the major obstacle to larger populations (Lamont et al. 1993). Indeed, persistence is clearing of land for agriculture and the Allee effect reduces seed production/plant to roads, which leads to habitat fragmentation, and some extent in all populations, especially those with reduction in routes for plants to migrate along in < 80 plants. Similar effects are likely in B. gardneri. the event of climate change. Short-distance seed Hence, in the unlikely event of successful migration dispersal in these prostrate species will result in to a new site, small founder populations may be rates of migration far too slow to keep up with the sterile or have very low rates of population growth. predicted movement of the climatic envelope. This Our data show that there was less than a 1 in 500 may be a general rule for taxa with short-distance chance of intact seeds of B. goodii stored on the dispersal. Despite the fact that B. gardneri has 177 plant becoming a one year-old seedling after fire, populations, > 10 times more than B. goodii, they compared with a 1 in 3 chance of manually removed have the same distribution of population sizes. This seeds sown at the same location reaching the same suggests that habitat fragmentation has had similar stage. Larger B. goodii populations have extremely effects on both species. Furthermore, without stable population dynamics (Witkowski and meaningful conservation outside nature reserves, Lamont 1997; Dreschler et al. 1999), and are likely the intervening areas between nature reserves will to persist for long periods. However, small B. goodii not sustain these species. As shown for these bank- populations (and probably B. gardneri as well) sia species, land-use change is now, and for some resulting from habitat fragmentation, represent decades will probably remain, the single most examples of ‘extinction debt’ (Lande 1987; Tilman important of the many interacting components et al. 1994; McCarthy et al. 1996). Even in the of global change affecting ecological systems. absence of further land clearing and active planting Resilience to marked climatic change is already of seeds/seedlings, these populations will doubtless built into these species, decreasing the imperative become extinct eventually. In particular, species to migrate. Historically, ecology has concentrated with poor dispersal abilities, such as these banksias, on small-scale studies, but in order to determine may be more susceptible to local extinctions in the responses of species to potential future climate response to habitat fragmentation. Owing to the change, large-scale studies, as shown by the trans- almost immortal nature of these banksias, particu- locations experiment here, are needed to ade- larly B. goodii, and their apparently much greater quately and reliably make predictions (Gaston climatic tolerances than their present ranges would 2003). The gradient study has provided a very use- suggest, it is likely that they will hang on in situ for ful contrast to the typical, often depressing, results centuries at their present locations under sub- of bioclimatic simulations in response to predicted optimal climatic conditions. This might involve future climate change. This contrast begs the a slow decrease in population sizes from one dis- question of how many other species show similar turbance to the next (sensu the ‘storage effect’ of resilience to climate change as these Banksia spp. lottery models; Warner and Chesson 1985), even if In addition, can one develop an a priori predictive International Journal of Biodiversity Science and Management 69 Resilience to global change in banksias Witkowski and Lamont model that could determine whether a species is Malcolm Briggs and Mike O’Donoghue of the ‘resilient’ to climate change? Western Australian Department of Conservation and Land Management for logistic support. This paper was prepared while Ed Witkowski was an ACKNOWLEDGEMENTS Adjunct Senior Fellow at Curtin University. This We thank Neil Gibson, Andrew Kennedy and project was funded by Environment Australia Dave Richardson for assistance with the BIOCLIM (Canberra), Roadside Conservation Committee analyses, Craig Walton and Darryl Abbott for field (Perth), Richard Ward Endowment Trust and Wits and laboratory assistance, and Laurie Anderson, University (Johannesburg). REFERENCES Allee WC. Group survival value for Philodina roseola, Cowling RM, Lamont BB and Enright NJ. Fire and a rotifer. Ecology 1949;30:395–7 management of south-western Australian bank- Allison HE and Hobbs RJ. Resilience, adaptive capa- sias. Proceedings of the Ecological Society of Australia city, and the “lock-in trap” of the Western 1990;16:177–83 Australian agricultural region. http/www.ecology Drechsler M, Lamont BB, Burgman MA, Akcakaya andsociety.org/vol9/iss1/art3. Ecology and Society HR, Witkowski ETF and Supriyadi. Modelling 2004;9(1):3 the persistence of an apparently immortal Banksia Bartlett WM. Western Australian Year Book. Perth: species after fire and land clearing. Biological Australian Bureau of Statistics, Western Australia Conservation 1999;88:249–59 Office; 1975 Drake J, di Castri F, Groves R, Kruger F, Mooney HA, Beard JS. A new phytogeographical map of Western Rejmanek M and Williamson M (eds). Biological Australia. Western Australian Herbarium Research invasions: a global perspective. New York: Wiley; 1989 Notes 1980;3:37–58 Eeley HAC, Lawes MJ and Piper SE. The influence of Bergelson J, Newman JA and Floresroux EM. Rates of climate change on the distribution of indigenous weed spread in spatially heterogeneous environ- forests in Kwazulu-Natal, South Africa. Journal of ments. Ecology 1993;74:999–1011 Biogeography 1999;26:595–617 Brown JH, Stevens GC and Kaufman DM. The Enright NJ and Lamont BB. Recruitment variability in geographic range: size, shape, boundaries, and the resprouting shrub Banksia attenuata and non- internal structure. Annual Review of Ecology and sprouting congeners in the northern sandplains Systematics 1996;27:597–623 heaths of south-western Australia. Acta Oecologica Burgman MA and Lamont BB. A stochastic model 1992;13:727–42 for the viability of Banksia cuneata populations: Gaston KJ. The structure and dynamics of geographic environmental, demographic and genetic effects. ranges. Oxford Series in Ecology and Evolution. Journal of Applied Ecology 1992;29:719–27 Oxford: Oxford University Press; 2003 Busby JR. Potential impacts of climate change on George AS. The genus Banksia L.f. (Proteaceae). Australia’s flora and fauna. In Pearman GI (ed.), Nuytsia 1981;3:239–473 Greenhouse: planning for climate change. Melbourne: Greuter W. Extinctions in the Mediterranean areas. CSIRO; 1988:387–98 Philosophical Transactions of the Royal Society of Busby JR. BIOCLIM – A bioclimatic analysis and pre- London B 1994;344:41–6 diction system. In Margules CR and Austin MP Green JW. Census of Western Australian plants. South (eds), Nature conservation: cost effective biological Perth: Western Australian Herbarium; 1985 surveys and data analysis. Canberra: CSIRO; Griffin EA and Associates. Vegetation survey of Bakers 1991:64–8 Junction and Millbrook Nature Reserve. Unpublished Clark JS, Fastie C, Hurtt G, Jackson ST, Johnson C, et al. Report No. 23. Perth: Department of Conservation Reid’s paradox of rapid plant migration. Bioscience and Land Management; 1985 1998;48:13–24 He TH, Krauss SL, Lamont BB, Miller BP and Cowling RM and Lamont BB. On the nature of Enright NJ. Long distance seed dispersal in a Gondwanan species flocks: diversity of Proteaceae metapopulation of Banksia hookeriana inferred in Mediterranean south-western Australia and from a population allocation analysis of Amplified South Africa. Australian Journal of Botany 1998; Fragment Length Polymorphism data. Molecular 46:335–55 Ecology 2004;13:1099–109 70 International Journal of Biodiversity Science and Management Resilience to global change in banksias Witkowski and Lamont Henderson-Sellers A and Blarg R. The Greenhouse Effect: Australia. Australian Journal of Botany 1995;43: Living in a warmer Australia. Kensington: NSW 283–303 University Press; 1989 Lamont BB, Wills RT and Witkowski ETF. Threats to Higgins SI, Nathan R and Cain ML. Are long-distance the conservation of southwestern Australian dispersal events in plants usually caused by non- Proteaceae. Acta Horticulturae 1995;387:9–18 standard means of dispersal? Ecology 2003a;84: Lamont BB and Witkowski ETF. A test for lottery 1945–56 recruitment among four banksia species based Higgins SI, Lavorel S and Revilla E. Estimating plant on their demography and biological attributes. migration rates under habitat loss and fragmenta- Oecologia 1995;101:299–308 tion. Oikos 2003b;101:354–66 Lande R. Extinction thresholds in demographic Hope P. Projected future changes in synoptic systems models of territorial species. American Naturalist influencing southwest Western Australia. Climate 1987;130:624–35 Dynamics 2006; in press Lindenmayer DB, Nix HA, McMahon JP, Hutchinson Hope P, Drosdowsky W and Nicholls N. Shifts in the MF and Tanton MT. The conservation of synoptic systems influencing southwest Western Leadbeater’s possum, Gymnobelideus leadbeateri Australia. Climate Dynamics 2006; in press (McCoy): a case study of the use of bioclimatic Hopper S. Biogeographical aspects of speciation in modelling. Journal of Biogeography 1991;18:371–83 the south-west Australian flora. Annual Review of McCarthy MA, Lindenmayer DB and Drechsler M. Ecology and Systematics 1979;10:399–422 Extinction debts and risks faced by abundant Hopper S and Gioia P. The Southwest Australian species. Conservation Biology 1997;11:221–6 Floristic region: evolution and conservation of a Nix H. Environmental determinants of biogeography global hot spot of biodiversity. Annual Review of and evolution of Terra Australis. In Barker WR and Ecology and Systematics 2004;35:623–50 Greenslade PJM (eds), Evolution of the flora and Hopper S, van Leeuwen SJ, Brown A and Patrick S. fauna of arid Australia. Frewville: Peacock; Western Australia’s Endangered Flora. Perth: Depart- 1982:47–66 ment of Conservation and Land Management; Nix H. A biogeographical analysis of Australian elapid 1990 snakes. In Longmore R (ed.), Atlas of elapid snakes of Huntley BJ. How vegetation responds to climate Australia. Canberra: Bureau of Flora and Fuana; change: evidence from palaeovegetation studies. 1986:4–15 In Pernetta JC, Leemans R, Endler D and Pittock AB. Actual and anticipated changes in Humphrey S (eds), Impacts of climate change on eco- Australia’s climate. In Pearman GI (ed.), Green- systems and species: environmental context. Gland, house: planning for climate change. Australia: Division Switzerland: IUCN; 1995:43–63 of Atmospheric Research, CSIRO; 1988:35–51 IUCN. IUCN Red List Categories v. 3.1. IUCN Species Richardson DM, Cowling RM, Lamont BB and van Survival Commission, Gland, Switzerland; 2001 Hensbergen HJ. Coexistence of Banksia species Kadmon R, Farber O and Danin A. A systematic in southwestern Australia: the role of regional analysis of factors affecting the performance of and local processes. Journal of Vegetation Science climatic envelope models. Ecological Applications 1995;6:329–42 2003;13:853–67 Rutherford MC and Westfall RH. Biomes of Southern Lamont BB and Connell SW. Biogeography of Banksia Africa: an objective categorization. Memoirs of the in southwestern Australia. Journal of Biogeography Botanical Survey of South Africa 1986;63:1–94 1996;23:295–309 Speck NH. The vegetation of the Darling-Irwin botanical Lamont BB, Hopkins AJM and Hnatiuk RJ. The flora – districts and an investigation of the distribution pattern composition, diversity and origins. In Pate JS and of the family Proteaceae in south western Australia. PhD Beard JS (eds), Kwongan: Plant life of the Sandplain. thesis. Perth: University of Western Australia; 1958 Nedlands: University of Western Australia Press; Taylor A and Hopper S. The Banksia Atlas. Canberra: 1984:27–50 Australian Government Publishing Service; 1988 Lamont BB, Klinkhamer PGL and Witkowski ETF. Thiele KR and Ladiges PY. A cladistic analysis of Bank- Population fragmentation may reduce fertility to sia (Proteaceae). Australian Systematic Botany zero in Banksia goodii: a demonstration of the Allee 1996;9:661–733 effect. Oecologia 1993;94:446–50 Tilman D, May RM, Lehman CL and Nowak MA. Lamont BB and Markey A. Biogeography of fire-killed Habitat destruction and the extinction debt. and resprouting Banksia species in south-western Nature 1994;371:65–6 International Journal of Biodiversity Science and Management 71 Resilience to global change in banksias Witkowski and Lamont Vitousek PM. Beyond global warming: ecology and Witkowski ETF and Lamont BB. Does the rare Banksia global change. Ecology 1994;75:1861–1976 goodii have inferior vegetative, reproductive or Warner RR and Chesson PL. Coexistence mediated ecological attributes compared with its widespread by recruitment fluctuations: a field guide to the co-occurring relative B. gardneri? Journal of storage effect. American Naturalist 1985;125:769–87 Biogeography 1997;24:469–82 Western Australian Government. Wildlife conserva- Witkowski ETF, Lamont BB and Connell SJ. Seed bank tion (rare flora) notice 1999. Government Gazette, dynamics of three co-occurring banksias in south WA; 17 December 1999:6194–9 coastal Western Australia: the role of plant age, Witkowski ETF. Growth and competition between cockatoos, senescence and interfire establish- seedlings of Protea repens (L.) L. and the alien ment. Australian Journal of Botany 1991;39:385–97 invasive, Acacia saligna (Labill.) Wendl. in relation Woodward FI and Williams BG. Climate and plant to nutrient availability. Functional Ecology 1991; distribution at global and local scales. Vegetatio 5:101–10 1987;69:189–97 72 International Journal of Biodiversity Science and Management

Journal

International Journal of Biodiversity Science & Management – Taylor & Francis

Published: Jun 1, 2006

Keywords: ALLEE EFFECT; AUSTRALIA; BIOCLIM; CLIMATIC PROFILES; DISTRIBUTION RANGE; HABITAT DEGRADATION; HABITAT LOSS; LANDSCAPE FRAGMENTATION; LONG-DISTANCE DISPERSAL; PERSISTENCE; POPULATION MODELLING

There are no references for this article.