Transcript
U N C O R R E C T E DP R O O F 12 A watershed study on genetic diversity: Phylogenetic analysis of the Platypleura 3 plumosa (Hemiptera: Cicadidae) complex reveals catchment specific lineages 4 B.W. Price a, * , N.P. Barker b , M.H. Villet a 5 a Department of Zoology and Entomology, Rhodes University, P.O. Box 94, Grahamstown 6140, South Africa 6 b Department of Botany, Rhodes University, P.O. Box 94, Grahamstown 6140, South Africa 7 9 a r t i c l e i n f o 10 Article history: 11Received 29 May 200912Revised 2 October 200913Accepted 7 October 200914Available online xxxx15 Keywords: 16Catchment diversification17Vicariance model18Watersheds19Cicadidae20 2 1 a b s t r a c t 22 Historical biogeography studies have at their disposal a small suite of vicariance models to explain 23 genetic differentiation within and between species. One of these processes involves the role of river 24 catchments and their associated watersheds, in driving diversification and is applicable to both aquatic 25 and terrestrial organisms. Although the idea of catchments structuring the genetic history of aquatic 26 organisms is reasonably well understood, their effect on terrestrial organisms has largely been over- 27 looked, withrelevant studies beinglimitedinscope. SouthAfrica presents aperfect test-bedfor elucidat- 28 ing this mechanism of diversification due to its rich biodiversity, range of climatic environments and 29 many large river catchments. Here we use the cicadas of the Platypleura plumosa complex to highlight 30 the importance of catchments and their associated watersheds in driving diversification of terrestrial 31 invertebrates that lack an aquatic life-stage. Population structure was found to correspond to primary 32 and in some cases secondary catchments; highlighting the need to include information on catchment 33 structure when formulating hypotheses of population diversification. Recognizing that climate change 34 in the near future is likely to alter the environment, and particularly precipitation patterns, insight into 35 recent patterns of population change related to catchments may be useful in a conservation context. 36 Ó 2009 Published by Elsevier Inc. 3738 39 1. Introduction 40 Planetary motion has been suggested to drive cycles of climate 41 change that cause the fragmentation or extinction of lineages of 42 organisms (‘‘Orbitally-Forced Range Dynamics” sensu Dynesius 43 and Jansson, 2000; Jansson and Dynesius, 2002; Jansson, 2003), 44 leading to changes in populations and communities that drive 45 biodiversification (Green and Sadedin, 2005). The severity of such 46 climate changes varies predictably with latitude ( Jansson, 2003). 47 Local anomalies in this pattern are referred to the historical bio- 48 geographical processes of dispersal and vicariance. Dispersal pro- 49 cesses are generally contingent on the characteristics of the 50 organism, while vicariance is generally attributable to changing 51 environmental features. Historical biogeographers have at their 52 disposal a suite of vicariance models to explain genetic differenti- 53 ation of species that includes continental drift, sea level change, 54 and landscape features as barriers or refugia. 55 The generation of new landscape features such as seas, rivers, 56 valleys, and mountains are essentially events in the histories of 57 catchments that, while they direct the movement of water, also 58 act as basins of biological diversity. Although the role of catch- 59 ments and their corresponding watersheds in structuring the ge- 60 netic history of organisms such as plants through hydrochory 61 (e.g.Levin et al., 2003) and aquatic organisms (e.g. redfin barbs: 62 Swartz et al., 2007, 2009; atyid shrimp:Hughes et al., 1995, 63 1996) is documented, the impact of different catchments oninver- 64 tebrates with aquatic stages is varied, with evidence both for (e.g. 65 Jackson and Resh, 1992; Wishart and Hughes, 2001) and against 66 (e.g.Hughes et al., 1998, 1999) Q2 a catchment effect, dependent pri- 67 marilyonthedispersalabilityoftheorganism(Wishart,2000).The 68 effectofcatchmentsonterrestrialorganismshasonlybeenconsid- 69 ered recently (e.g.Measey et al., 2007), with some studies being 70 limited in scope, focussing on either the organism (e.g.Garrick 71 et al., 2007; Price et al., 2007) or the habitat (Pearson and Raxwor- 72 thy, 2009), rather than the landscape. Recognizing that climate 73 change in the near future is likely to alter the environment, and 74 particularly precipitation patterns, insight into recent patterns of 75 population change related to landscape, and particularly catch- 76 ments, may be useful in a conservation context. 77 Ifcatchmentsandwatershedsareimportantinstructuringpop- 78 ulations during periods of climatic oscillations, several hypotheses 79 should be fulfilled. 80 (1) Different(primaryorsecondary)catchmentsshouldcontain 81 genetically distinct lineages. 1055-7903/$ - see front matter Ó 2009 Published by Elsevier Inc.doi:10.1016/j.ympev.2009.10.011 * Corresponding author. Fax: +27 (0) 46 622 8959. E-mail addresses:
[email protected](B.W. Price),
[email protected](N.P. Barker),
[email protected](M.H. Villet).Molecular Phylogenetics and Evolution xxx (2009) xxx–xxx Contents lists available atScienceDirect Molecular Phylogenetics and Evolution journal homepage:www.elsevier.com/locate/ympev YMPEV 3400 No. of Pages 10, Model 5G23 October 2009 ARTICLE IN PRESS Please cite this article in press as: Price, B.W., et al. A watershed study on genetic diversity: Phylogenetic analysis of the Platypleura plumosa (Hemiptera:Cicadidae) complex reveals catchment specific lineages. Mol. Phylogenet. Evol. (2009), doi:10.1016/j.ympev.2009.10.011 U N C O R R E C T E DP R O O F 82 (2) Sister clades should occur in adjacent (primary or second- 83 ary) catchments. 84 (3) The proportion of genetic structure that can be explained 85 using isolation by distance (IBD) should be small in comparison 86 to that explained by catchment association. 87 (4) The time to most recent common ancestor (TMRCA) should 88 correspond to periods of increased aridity if vicariance is involved 89 in cladogenesis as catchments are fixed landscape features, not- 90 withstanding river capture, and orogeny. 9192 South Africa presents a perfect test-bed for this mechanism of 93 diversificationduetoits(a)highplantandanimaldiversityofthere- 94 gion,amongthehighestintheworld;(b)numerouscatchmentsthat 95 span various environmental gradients and topographic features; 96 and(c)reasonablywell-understoodclimaticandgeologicalhistory. 97 Almost a third of South Africa’s vegetation (28.2% of land area) 98 can be classified as semi-arid, including the Succulent-Karoo (SK), 99 Nama-Karoo (NK), and Albany Thicket (AT) biomes (Mucina and 100 Rutherford, 2006). The extent of these three biomes varies from 101 widespread (NK, 19.5%) to restricted (SK, 6.5%; AT, 2.2%) (Mucina 102 and Rutherford, 2006). Albany Thicket is considered the oldest of 103 these three biomes, being of putatively Eocene srcin (Mucina 104 and Rutherford, 2006). The Nama- and Succulent-Karoo biomes 105 are thought to have developed as a result of increased aridity in 106 the late Miocene, coincident with the formation of the Benguela 107 current (Siesser, 1980), continental uplift (Partridge and Maud, 108 1987; Artyushkov and Hofmann, 1998) and the rainshadoweffects 109 associated with these two phenomena (Levyns, 1964; Scott et al., 110 1997; Mucina and Rutherford, 2006; Verboom et al., 2008). 111 Within the Plio-Pleistocene (5MYA – present), Q3 aridity and sea- 112 sonality have increased with the intensification of the Benguela 113 current (Siesser, 1980) and global climate fluctuations in response 114 to Milankovitch oscillations. Evidence for the dramatic changes in 115 South Africa’s climate in response to global glacial cycles through- 116 out the Pleistocene can best be illustrated by comparing the Last 117 Glacial Maximum (LGM; 21± 2kya) (Mix et al., 2001; Gasse 118 et al., 2008) and the Holocene Altithermal (7± 1kya) (Partridge 119 et al., 1999). In the LGM, temperatures have been estimated at 120 5 ° C cooler than present (Heaton et al., 1983; Talma and Vogel, 121 1992) with the region experiencing only50–70% of its current an- 122 nual rainfall (Partridge et al., 1999). 123 Sedentary invertebrates with short generation times are good 124 tools to capture the geographic partitioning of genetic structure 125 that results from processes such as climate cycling (Garrick et al., 126 2007). In addition they are likely to survive in small, isolated pop- 127 ulations, thus reducing the chances of local extinctions and pre- 128 serving the continuity of their phylogeographical signal. As a 129 result,theseorganismsmaybebettersuitedtostudiesofhistorical 130 biogeography than most vertebrates, which have received the 131 majority of attention in the region. 132 Although cicadas (Hemiptera: Cicadidae) are capable of flight, 133 theymaybeconsideredsedentaryorganisms. Theirunusual lifecy- 134 cle consists of 11months–17years of sedentary development that 135 occurs underground on the roots of a single plant, followed by 136 about 3weeks of mobile adult life above ground. Furthermore, 137 estimates of dispersal abilityin cicadas have ranged fromless than 138 150m (Simoes and Quartau, 2007) to 300m from their place of 139 emergence (Karban, 1981; Williams and Simon, 1995). Thus it 140 seems that long-range dispersal may take many generations in 141 cicadas (Maier, 1982; Arensburger et al., 2004). 142 The Platypleura plumosa (Germar) group provides a model clade 143 forexploringtheeffectofcatchmentsandwatershedsonthehistor- 144 icalbiogeographyofthesemi-aridregionofSouthAfricabecauseof 145 their(1)distributionacrossseveralcatchmentsoccupyingaclimatic 146 gradient (arid west–moist east); (2) short life cycles (like many 147 insects);(3)relativelylowvagility(partlybecauseadultsareonlyac- 148 tiveforafewweekseachyear);and(4)identicalsmallrangeofhost 149 plant associations, primarily on Acacia karroo , which removes the 150 confoundingeffectsofspeciationduetohostplantshifts. 151 Although the group is currently composed of two nominal spe- 152 cies, P. plumosa (Germar) and P. hirtipennis (Germar) (Villet, 1997), 153 morphological andacousticvariationsuggeststhat fiveother cryp- 154 tic species may be present in this group, referred to here as Platy- 155 pleura ‘ karooensis ’, P. ‘ gariepflumensis ’, P. ‘ olifantsflumensis ’, P. 156 ‘ breedeflumensis ’, P. ‘ gamtoosflumensis ’; speciesnames awaitformal 157 description. 158 Theaimofthisstudyistotesttheeffectofcatchmentsandwater- 159 sheds in shaping the genetic structure of the Platypleura plumosa 160 complexbothatthepopulationlevelandbetweenthespecies. 161 2. Materials and methods 162 2.1. Sampling and laboratory protocol 163 Cicadas belonging to the P. plumosa group were collected into 164 95% ethanol from a total of 119 sites, covering their known geo- 165 graphical range (Table S1,Fig. 1). In addition, two closely related 166 southern African species of Platypleura (Price et al.m unpublished 167 data) were included as outgroups for the analysis (Table S1). 168 Total genomic DNA was extracted from the muscles following 169 the Chelex Ò 100 protocol (Walshet al., 1991). Small pieces of wing 170 ortymbalmuscletissue(c.2mm 3 )werehomogenizedin300 l l5% 171 Chelex Ò 100solutionandincubatedinaheatingblockat105 ° Cfor 172 15min, vortexing the samples every five minutes. Following the 173 incubation period, the supernatant was removed for subsequent 174 use in PCR amplifications. 175 Portionsof mitochondrial CytochromeOxidaseI (COI) and large 176 subunit ribosomal 16S RNA (16S) were amplified from each sam- 177 ple. In addition portions of two nuclear genes: elongation factor 178 1 alpha (EF-1 a ) and calmodulin (CAL) were amplified for two rep- 179 resentatives of each of the major mitochondrial clades (Table S2). 180 Primers used for amplification and sequencing reactions were 181 COI: either COI-F1 (5 0 -TAATATAAACTATTAACCTTCAAAGT-3 0 ) or 182 COI-F2 (5 0 -TCTACTAATCACAAAGATATYGGAAC-3 0 ) (designed for 183 this study) with C1-N-2568 (Brady et al., 2000); Q4 16S: 16Sar with 184 16Sbr (Palumbi et al., 1991); EF-1 a : EF1-F650-mod (Arensburger 185 et al., 2004) with EF1-N-1419 (Sueur et al., 2007); CAL: Cal-60- 186 For (Buckley et al., 2006) with Cal-2-Rev (UBC insect primer kit). 187 PCR amplifications were performed in 25–50 l l reactions using 188 the following protocol: 35 cycles of denaturation at 95 ° C for 45s, 189 annealing at 48 ° C (16S and COI), or 58 ° C (EF-1 a and CAL) for 45s 190 and extension at 72 ° C for 2min. PCR products were confirmed by 191 electrophoresisof5 l l PCRproductstainedwithethidiumbromide 192 and5 l l trackingdyeina1%agarosegel, andvisualizedusingaUV 193 trans-illuminator. 194 PCR products were purified using the Invitek Invisorb MSB Ò 195 Spin PCRapace purification kit (Invitek, Germany) and sequenced 196 inbothdirections.Thesequencingreactionswerecarriedoutusing 197 the ABI Big Dye Sequencing kit v.3.1, according to the manufac- 198 turer’s instructions. Sequence trace files were generated using an 199 ABI 3100 genetic analyzer sited at Rhodes University. Trace files 200 were checked and edited using Genestudio Pro v.1.0 (GeneStudio, 201 Inc.). The sequence data were imported into MEGA v.4 (Tamura 202 et al., 2007), aligned using the Clustal W algorithm (Thompson 203 et al., 1994) and checked manually. 204 2.2. Phylogenetic analysis 205 Two sets of analyses were undertaken, the first on the mito- 206 chondrial gene dataset (MT) for each sample and the second on 2 B.W. Price et al./Molecular Phylogenetics and Evolution xxx (2009) xxx–xxx YMPEV 3400 No. of Pages 10, Model 5G23 October 2009 ARTICLE IN PRESS Please cite this article in press as: Price, B.W., et al. A watershed study on genetic diversity: Phylogenetic analysis of the Platypleura plumosa (Hemiptera:Cicadidae) complex reveals catchment specific lineages. Mol. Phylogenet. Evol. (2009), doi:10.1016/j.ympev.2009.10.011 U N C O R R E C T E DP R O O F 207 tworepresentative taxa for eachof the major mitochondrial clades 208 with the combined nuclear and mitochondrial datasets (ALL). 209 To ascertain whether the data from each gene region could be 210 combined into a single data set, the Incongruence Length Differ- 211 ence (partition homogeneity) test (Farris et al., 1994) with 1000 212 replicates was conducted in PAUP Ã v.4.0b10 (Swofford, 2003) with 213 invariant characters removed (Cunningham, 1997) as the number 214 of variable characters differed between each gene region. 215 Parsimony analysis was conducted using PAUP Ã as follows: a 216 simple heuristic search with TBR branch swapping enforced was 217 conducted to approximate the length of the shortest tree. Follow- 218 ing this a heuristic search with 1000 random addition replicates 219 was conducted, starting from random trees and keeping a single 220 tree less than or equal to the shortest tree, for each replicate. If 221 thesecondsearchfoundshortertreestheprocesswasrepeatedun- 222 til no shorter trees were found. A strict consensus tree was then 223 generated from the resulting parsimony trees. Confidence in each 224 node was assessed with 100 full heuristic bootstrap replicates in 225 PAUP Ã . 226 The most appropriate model of sequence evolution for the 227 Bayesian analysis was selected for each of the partitions in the 228 two datasets (MT and ALL) using the AIC test (Akaike, 1974) as 229 implemented in MrModeltest v.2.2 (Nylander, 2004), models se- 230 lected are summarized inTable S3. 231 Partitioned Bayesian analyses were conducted using MrBayes 232 v.3.1.2 (Huelsenbeck and Ronquist, 2001) carried out on the Uni- 233 versity of Oslo Bioportal (www.bioportal.uio.no). Although there 234 isnodefinitivemethodforpartitioningadataset,thedatasetswere 235 partitioned first by gene, then by coding properties (introns or 236 exons). In addition all protein coding regions were partitioned by 237 codon position to incorporate rate variation (Marshall et al., 238 2006). This resulted in three partitioning strategies (no partition; 239 each gene; each gene and each codon) for both the MT and ALL 240 datasets The effect of partitioning strategy was estimated using 241 pairwise comparison of Bayes Factors ( sensu Nylander et al., 242 2004; Brandley et al., 2005) with 2lnBF A–B P 10 indicative of 243 strong support for partitioning strategy A vs B, followingKass 244 andRaftery(1995).Eachanalysiscomprisedfourindependentruns 245 of 10 million generations each, using random starting trees with 246 four chains (one cold, threehot), samplingevery1000 generations. 247 Model parameters for each partition as selected by MrModeltest 248 (Table S3) were achieved using the ‘‘Lset nst=rates=; Prset state- Fig. 1. Sample localities with primary catchments delimited and cladogram representing the most-parsimonious tree and majority rule Bayesian cladogram of all datacombined. Bootstrap (left) and posterior probability (right) support are indicated at each node. B.W. Price et al./Molecular Phylogenetics and Evolution xxx (2009) xxx–xxx 3 YMPEV 3400 No. of Pages 10, Model 5G23 October 2009 ARTICLE IN PRESS Please cite this article in press as: Price, B.W., et al. A watershed study on genetic diversity: Phylogenetic analysis of the Platypleura plumosa (Hemiptera:Cicadidae) complex reveals catchment specific lineages. Mol. Phylogenet. Evol. (2009), doi:10.1016/j.ympev.2009.10.011 U N C O R R E C T E DP R O O F 249 freqpr=” commands. A branchlength prior was set using 250 ‘‘brlenspr=Unconstrained:Exponential(500.0)”. All other parame- 251 ters (excluding branchlengths and topology) were unlinked across 252 partitions using the ‘‘unlink” command. Stationarity in each analy- 253 sis was assessed using the potential scale reduction factor (PSRF) 254 data and plots of likelihood scores, tree length and average stan- 255 dard deviation of split frequencies against generation. These plots 256 showed that the analyses reached stationarity well within the first 257 1 million generations. Thus the first 1000 trees generated in each 258 analysis were discarded as ‘‘burn-in”, ensuringthat onlytrees gen- 259 erated at stationarity were used to calculate the posterior 260 probabilities. 261 2.3. Landscape genetic analysis 262 To assess the spatial component of the genetic structure three 263 alternatives were tested: (a) Genetic structure present due to iso- 264 lationbydistance;(b)Geneticstructurepresentduetocatchments 265 and associated watersheds; (c) Genetic structure present due to 266 other undefined barriers. The COI dataset was used for all analyses 267 as it is the most informative. Primary and secondary catchments 268 were identified based on South Africa’s Water Research Commis- 269 sion’s river region classification (Midgley et al., 1994). The contri- 270 bution of isolation by distance (IBD) was assessed using the 271 Mantel test (Mantel, 1967) as implementedinGeneAlEx6.1(Peak- 272 all and Smouse, 2006) with 999 permutations. While the identifi- 273 cation of barriers was carried out using Barrier v.2.2 (Manni et al. 274 2004), whichhighlights areas of pronouncedgeneticdiscontinuity. 275 For this analysis, samples were connected using Delauney triangu- 276 lation; barriers were then identified using Monmonier’s maximum 277 distance algorithm. The number of barriers to be identified was 278 stipulated a priori in Barrier, based on the number of secondary 279 catchments occupied by each of the species concerned (# catch- 280 ments – 1). To confirm the relative contribution of catchments to 281 population structure, samples were then grouped according to 282 either primary or secondary catchments and AMOVA was con- 283 ducted on these predefined groups using GeneAlEx with 999 per- 284 mutations followed by pairwise comparisons of U pt . 285 2.4. Molecular dating 286 The lack of fossil data representative of this group and the ab- 287 sence of an adequate geological vicariance event to calibrate the 288 molecular clock resulted in the use of previously estimated rates 289 of sequence evolution for the COI region. Although the application 290 of a global molecular clock is problematic (Heads, 2005), the COI 291 region is suggested to be the most reliable gene when enforcing 292 a molecular clock for insects because it possesses the most consis- 293 tent rates of sequence evolution between lineages (Gaunt and 294 Miles, 2002). The COI data were tested for the applicability of a 295 molecular clock using the likelihood ratio test (LRT). Two separate 296 maximum likelihood analyses were conducted in PAUP Ã using the 297 model settingsas selectedinModeltest v.3.7(Posada and Crandall, 298 1998) with or without the clock enforced. A v 2 test of the differ- 299 ence between the two likelihood values ( a =0.05) was then used 300 to assess whether or not the molecular clock significantly influ- 301 enced the ML analysis (Felsenstein, 1981). 302 The mean rate of sequence evolution was set at 2.3% per MY 303 using the estimate of Brower (1994). The age of each node and 304 itscorresponding95%confidenceintervalwerethenestimatedun- 305 der the GTR+I+G model as suggested by MrModeltest using an 306 uncorrelated lognormal (UCLN) relaxed clock in BEAST v.1.4.8 307 (Drummond and Rambaut, 2007). The data was partitioned into 308 three codon positions with the substitution model, rate heteroge- 309 neitymodel andbase frequencies unlinked; inaddition a Yule spe- 310 ciation prior and a UPGMA starting tree were used. This analysis 311 was run using the resources of the Computational Biology Service 312 Unit, Cornell University (http://cbsuapps.tc.cornell.edu/beast.- 313 aspx), with two independent runs of 20 million generations, fol- 314 lowing a discarded burn-in of 200,000 generations. Stationarity 315 andtheestimationoftheeffectivesamplesize(ESS)intheanalysis 316 were confirmed by inspection of the MCMC samples using Tracer 317 v.1.4 (Rambaut and Drummond, 2007). The two runs were com- 318 bined using LogCombiner and then summarized in TreeAnnotater, 319 part of the BEAST package. 320 3. Results 321 3.1. Data characteristics 322 The final molecular data set comprised 2842bp, of which 1524 323 comprised mitochondrial COI and 16S data. The EF-1 a data was 324 made up of portions of both exons (3, 4, and 5) and introns, and 325 the CAL data comprised one intron. All coding portions of the gene 326 datasets aligned readily while gaps corresponding to insertion or 327 deletion events were included in the 16S, EF-1 a , and CAL datasets. 328 As indels were parsimony informative, they were coded as binary 329 charactersandusedinallanalyses;multiple-baseindelsweretrea- 330 ted as one character when recoded. Sequence characteristics and 331 models selected for each partition are summarized inTable S3. 332 Ofthe2842basepairs,428characterswerevariable(15%);ofthese 333 313 (11%) were parsimony informative including outgroups. The 334 ILD tests indicated that the data sets were not significantly differ- 335 ent so the individual gene datasets were combined into two data- 336 sets (MT, p =0.72; ALL, p =0.93). All analyses of the MT and ALL 337 data sets resolved seven well-supported clades within the group, 338 but the topological arrangement of some of the clades lacked sup- 339 port in the analyses (Figs. 1 and 2). 340 3.2. Phylogenetic analysis – mitochondrial dataset 341 Parsimony analysis of the mitochondrial data yielded 19 most- 342 parsimonious trees with a tree length of 496 steps (CI=0.591; 343 RI=0.949). Comparison of the three partitioning strategies sug- 344 gested that partitioning data by gene provided a slightly better 345 log likelihood than not partitioning the data (2lnBF 1–2 =7.8). 346 However, partitioning the data by gene and codon provided a sig- 347 nificantlybetterloglikelihoodthaneithernotpartitioningthedata 348 (2lnBF 1–4 =11.4) or partitioning the data by gene alone (2lnBF 2– 349 4 =11.0), thus the Bayesian analysis that utilized the data parti- 350 tioned by both gene and codon is presented. The majority rule 351 Bayesian phylogeny (Fig. 2) recovered the monophyly of each pro- 352 posed species, yet failed to support many of the ingroup relation- 353 ships, resulting in a polytomous phylogram for the group (Fig. 2). 354 3.3. Phylogenetic analysis – combined dataset 355 Parsimony analysis of the combined mitochondrial and nuclear 356 data recovered the monophyly of the proposed seven species, 357 yielding one most-parsimonious tree with a tree length of 406 358 steps (CI=0.685; RI=0.763). The Bayesian analysis recovered the 359 monophyly of each species with the same topology as the parsi- 360 mony analysis; however the level of support of some nodes dif- 361 fered from the parsimony analysis (Fig. 1). 362 3.4. Landscape genetic analysis 363 Catchment association was strong at the species level, with five 364 ofthesevenspecieseachbeingrestrictedtoasingleprimarycatch- 365 ment (Hypothesis 1,Fig. 1) with only one example of primary 366 catchment sympatry involving P. plumosa , P. ‘ karooensis ’, and P. 4 B.W. Price et al./Molecular Phylogenetics and Evolution xxx (2009) xxx–xxx YMPEV 3400 No. of Pages 10, Model 5G23 October 2009 ARTICLE IN PRESS Please cite this article in press as: Price, B.W., et al. A watershed study on genetic diversity: Phylogenetic analysis of the Platypleura plumosa (Hemiptera:Cicadidae) complex reveals catchment specific lineages. Mol. Phylogenet. Evol. (2009), doi:10.1016/j.ympev.2009.10.011 U N C O R R E C T E DP R O O F 367 ‘ breedeflumensis ’. Platypleura plumosa and P. hirtipennis were dis- 368 tributed over multiple primary catchments (Fig. 1). 369 At the population level, isolation by distance was not a signifi- 370 cant factor in the population structure of any species (Table 1), ex- 371 cept for P. ‘ olifantsflumensis ’ ( R 2 =0.8272; p =0.008) and to a much 372 lesser extent P. plumosa ( R 2 =0.0115; p =0.037). The Barrier analy- 373 ses (data not shown) highlighted barriers to gene flow which cor- 374 responded to catchment watersheds in P. ‘ olifantsflumensis ’ (one 375 barrier,correspondswithsecondarycatchmentwatershed), P. ‘ kar- 376 ooensis ’ (one barrier, corresponds with secondary catchment wa- 377 tershed) and P. plumosa (ten barriers, one corresponds to a 378 primary catchment watershed and four correspond to secondary 379 catchment watersheds). The barriers highlighted by Barrier 380 showed little correspondence with primary or secondary catch- 381 mentsfor P. hirtipennis (fourbarriers,onecorrespondstoaprimary 382 catchment watershed) and did not correspond to any secondary 383 catchments for P. ‘ breedeflumensis ’ (two barriers, no correspon- 384 dence to the secondary catchment watersheds). 385 Groupingsamplesbycatchmentresultedintheidentificationof 386 groupsthat were significantly different usingAMOVAfor P. plumo- 387 sa (primary and secondary catchments), P. ‘ karooensis ’ (secondary 388 catchments), P. ‘ olifantsflumensis ’ (secondary catchments), and P. ‘ 389 breedeflumensis’ (secondary catchments), although this effect 390 was weak in P. ‘ breedeflumensis ’(Table 1); no significant variation 391 was found for P. hirtipennis , calculations were not performed for 392 P. ‘ gariepflumensis ’ and P. ‘ gamtoosflumensis ’ due the restriction of 393 these samples to one secondary catchment each (Fig. 1andTable 394 1). Population parameters and U pt values are summarized inTable 395 1; pairwise U pt values for species inhabiting more than two catch- 396 ments can be found in thesupplementary data( P. plumosa :Tables 397 S4 and S5; P. hirtipennis :Table S6). 398 3.5. Molecular dating 399 Maximum likelihood analysis of the COI data under the 400 GTR+I+G model with and without a clock resulted in two likeli- 401 hoodvalues(clock,ln L = À 4314.9564;noclock,lnL= À 4230.8537). 402 A v 2 test of the difference in likelihood values was significant 403 (2ln L =168,df=119, p =0.02),sothedatinganalysisusedarelaxed 404 clockinBEAST.Thedatingestimatessuggestthatcladogenesisinthe 405 groupbeganwiththesplitbetween P. ‘ karooensis ’andtheremaining 406 species approximately 2.3MYA, followed by a radiation with rapid Fig. 2. Majority rule Bayesian phylogram based on the mitochondrial (MT) dataset, images depict the gross morphology of the species within the group. Symbols representwell-supported clades and correspond toFig. 3. B.W. Price et al./Molecular Phylogenetics and Evolution xxx (2009) xxx–xxx 5 YMPEV 3400 No. of Pages 10, Model 5G23 October 2009 ARTICLE IN PRESS Please cite this article in press as: Price, B.W., et al. A watershed study on genetic diversity: Phylogenetic analysis of the Platypleura plumosa (Hemiptera:Cicadidae) complex reveals catchment specific lineages. Mol. Phylogenet. Evol. (2009), doi:10.1016/j.ympev.2009.10.011
