The Phytogeography, Ecology and Conservation Status of Lechenaultia R.Br
Total Page：16
File Type：pdf, Size：1020Kb
Kinsia 1(1): 85 133(198?) 85 The phytogeography, ecology and conservation status of Lechenaultia R.Br. (Goodeniaceae) David A. Morrison John Ray Herbarium, Botany Building (A12), University of Sydney,N.S.W. 2006.Present address: Depart ment of Applied Biology.New South WalesInstitute of Technology,P.O. Box 123,Broadway, N.S.W. 2007. Abstract Mor son, D.A. The phytogeography, ecology and conseNation stai:us of rec,,renaulfia R.Br. (Goodeniaceae).Kingia 1(1): 85-133 (1987). Distribution maps for each of the 25 known speciesof LechenauLtiaarcpresented for 0.5' lat. x 0.5" long.g d cells,along with discussionsofthe habitat, floweing period, morphology,and conseNation status. Nineteen of the speciesare endemic (or nearly endemic)to south westernWestern Australia. An isoflor map ofthis region indicatesa high node of species-dchnesson the northem sandplains,with a band of moderatespecies-richness running south eastwardsto the southem sandplains.This species-richnessband followsthe 400mm annual rainfall isohyet,and conespondswith the kwonganheathlands and the wheatbelt. Ordination, clusterand discriminant function analysesof the distribution of these19 aechenaultiaspecies suggestsihat there are five biogeographicregions: one area on the iar nolthem sandplains,one on the nofhern sandplainsand northern part of the wheatbelt,one in the jarrah woodlandsand karri forests,one on the southem sandplains and southem part of the wheatbelt, and one in the inland mallee area. Discriminant. function analyses of t]le dist bution of these 19 Lechenaultia species among the phytogeographicregions of Beard and Barlow indicate that they are both accurate reflections df the Leche nauLtia distributions. Similarity coefficientsof speciespresence among the phltogeographicregions of Beard indicate that the Irwin, Darling andAvon BotanicalDistricts are more sinilar to eachother than they areto the Eyre and Roe BotanicalDistricts; while cladisticanalysis ofendemic species indicates that the Irwin, Roeand Eyre Districts havea more similar history. This suggeststhat there has beenfairly \idespread dispersalof somespecies in recent times. LechenaultiaLaricinais considered to be an endangeredspecies, with L. juncea,L. longiLobaand L. puLuinnris vulnerable,and, L. acutiLobaand, L. superbararc; L. chlorantha and L. ouor@are too poorly known for the conservationstatus to be known. Only three of these speciesare confidently known to be representedin conservationreserves, with two of the others thought to be present. Introduction The Goodeniaceaeare well-representedin the flora of WesternAustralia (Marchant 1973),and they are one of the commonly encounteredfamilies in the heathlandsof the south-west (Georgeet al. 1979,Lamont et al. 1984).Within this family, Lechenaulti,a R.Br. is oneofthe most interesting genera.It is restrictedto Australia and New Guinea, with all of the 25 recognisedspecies occurring in Australia (Morrison 1986).Nineteen ofthe speciesare endernic (or nearly endemic)to south-westernWestern Australia, with one specieson North-West Cape,two speciesin arid western Central Australia, one in arid easternCentral Australia, one restdcted to a small areaof the Northern Territory, and one specieswidespread across tropical northern Australia and extending to the southern coast and off-shore islands of New Guinea. 86 Kingia Vol. 1, No. I (198?) Most of the speciesare small perennial sub-shrubsor herbs,with only Lechenaultia acutiloba,L. dtDaricqtqand, L.linarioides becomingshrubs over 1 m high. The plants are widespreadbut never dominant components of the flora of south-westernWestern Australia, occurringin woodland,forest, and especiallyheath (Georgeet al. 1979,Specht 1981.Lamont et al. 1984). This paper provides previously unpublished notes on the ecology,morphology, and conservationstatus of all ofthe species,as well as presentingthe distribution data. The phytogeographyof the South-West Botanical Province of Western Australia is also analysedin relation to the distribution and phylogeny of LechenquLtia. Methods The datapresented were obtained from the herbariumcollections housed at AD, BRI, CANB, CBG, LD, MEL, NT, PERTH and SYD, plus the type collectionsof CGE and S (codesas in Holmgren et al. 1981),supplemented by field observationsof some 70% of the species(all from south-westernWestern Australia). The taxonomic schemeused wasthat of Morrison (1986). The distribution of each specieswas recordedon a map of Australia subdividedinto a 0.5" Iatitude by 0.5" longitude grid, correspondingto the 1:100,000topographic survey maps of the Division of National Mapping, Commonwealth Department of Natural Resources. Within the Lechenaultiq distribution, only the South-West Botanical Province of WesternAustralia containsenough species for the analysisofphltogeographic patterns in detail. Consequently,the phytogeographicanalyses concentrated on this area. Two species-richness(isoflor) maps of the South-West Botanical province were derived from the distribution data. The first presentedthe speciesdistribution based solely on validated herbarium records.However, as this is presumablyan incomplete record,due to sparsecollecting in someareas and land clearancein others,a secondmap presentedthe distributions as they could be conservativelyinferred from the recorded distribution. For most species,this secondmap differed frorn the first only in oneor two grid cells; and this second map was assumedto be superior to the first as a true representationofthe species'distributions. It wastherefore these inferred distributions that were used in the ph)'togeographicanalyses. The pattern of variation among the distributions of the 19 Lechenaultia species present in the South-West Botanical Province was analysedsimultaneously using an ordination technique,detrended correspondence analysis (Hill 19?9a),on the presence or absenceof the speciesin each of those 130 grid cells that occur in the South-West Botanical Province. This non-linear technique is reported to be superior to the traditional linear ordinations (Gauch et al. 1977,Hill and Gauch 1980,Gauch et al. 1981),and this appearsto be true for the data setsfor which I have madecomparisons. The pattern of vadation among the 130 grid cells was analysedby ordinating the presenceor absenceof each of the 79 LechenauLtiaspecies in each cell. The clusteringpattern ofthe 130grid cellswas investigated using a clusteringstrategy, two-way indicator speciesanalysis (Hill 1979b),on the presenceor absenceof the 19 Lechenaulti.aspecies in eachgrid cell. This polythetic divisive techniqueis reportedto be superior to traditional agglomerativetechniques (Gauch and Whittaker 1981),and this aDnearsto be also true for the data sets that I have used. D.A. Morrison, The phytogeography,ecology and conservationstal;s of Lechenaultia- 87 The predictability of the phJ.togeographicclusters formed by this analysiswas tested using discriminant function analysis (Klecka 1975). This technique derives a small number of linear functions that weight the original discriminating variables (in this case,presence or absenceof eachof the 19 Lechenaultiospecies) so as to maximisethe separation of the total scoresof a set of referencesamples summed over all of the variables.In this case,the referencesamples consisted of eachofthe 0.5" x 0.5" grid cells that make up a particular phltogeographic cluster. Each of the grid cells was then individually re-classifiedusing the discriminant functions derived from the original analysis,to seeif the they were correctly classifiedinto the particular phl'togeographic region to which they were originally assigned.This technique effectively weights the speciesby constancywithin groups (ie. minimal within-group variance and maximal between-groupvariance), and this approach to assessingclassifications has been suggestedby severalpeople (Farris 1966,Johnson 1982). The relationships between the Lechena.ultia species distdbutions and the phltogeographicregions of Beard (1980d)and Barlow (1985)that coverthe South-West Botanical Province (see Figures 6a and 6b) were also analysedusing discriminant function analysis.For the Barlow (1985)regions, assigning each of the 131grid cellsto a particular region was straightforward, as this systemis basedon a 1" x 1" grid; while for the Beard (1980d)regions, which are basedon vegetationphysiognomy, those grid cells near the boundariesof the regionswere assignedto the region that occupiedthe greatestportion of the cell. Once again, each of the grid cells was then individually re-classifiedusing the discriminant functions to seeif the they were correctly classified into the particular ph].togeographicregion to which they were originally assigned.A compadsonwas also madebetween the original Barlow (1984)regions and the modified regions of Barlow (1985), to see if there was any significant improvement in the predictivity of the modified regions comparedto the original. The relationships among the five south-westernphytogeographic regions of Beard (1980d),as related to Lechenaultia, were analysedin two ways. Firstly, the overall superficialsimilarity was assessedusing two similadty indices:-the JaccardCoefficient and the Simple Matching Coefficient (Hubalek 1982). Secondly,the historical relationshipsamong the regionswere analysedby converting tineLechenaultia morphologicalcladogram of Mordson (1987)into an areacladogram. This was done by simply superimposingthe phytogeographicregions of Beard (1980d) onto the cladogram(cf. Nelson and Platnick 1981).Unfortunately, the cladogramas it stands (Figure 7a) is