7830 - Laserfiche WebLink

Home Browse Search

jorr/oni umillwla upperLCR] Bill WillUms R. SUIR. ~ Gila R. Green R. cypIwJ a Ilea- L J Sevier R. -.n. ~k,... bicolo< }70 O.s 0.4 0.3 0.2 Nci's D 0.1 0.0 0.. ) ~. b umillwla =:~J' SaIl R. roINJIa Gila R. GRell R. QPIta tkps Bear LI---,- Sevier g-wu bia1Ior IGn. >Ies iHl lby on nd lIy ric 1C- .tic cal :he nd sla pi- Jm lite G. ;ed G. FIG. 1 Analysis of evolutionary relationships among Gila a. lJ'GMA top%gy3!l based on Nei's distances31 calculated from 28 presumptive a1lozyme loci: mAH-A. sAH-A. AK -A. ADH-A. mAA T-A. sAA T-A. CBP-1. CBP-2. CK-A. FH-A. GP-1. G3PDH-A. GPI-B. mlDH'-A. s/DfF-A. LDH-A. LDH-B. LDH-C, mMDH-A. sMDH-A. sMDH-B. mMEP-A. PEPA. PEPB. PEPD. PEPS. PG.1-A. and sSOD-A Conditions for resolution described in refs 13 and 32. b. Most parsimonious tree (59.3 steps) resolved from allozyme allele frequencies using FREQPARS33. Alternative phylogenetic topologies were evaluated by comparing total tree lengths. c. lPGMA topology34 based on pairwise esti- mates of sequence divergence35 from mtONA restriction site's' as described in refs 13 and 36. Each mtONA sample was characterized with the following restriction endonucleases: BamHl. Bell. BgIII. BstElI. EcoRI. Hindlll. M1ul. Neal. Ndel, NIleI. PvulI. Sacl. Xbal and Xhol. d. Strict consensus of five most parsimonious trees (62 steps. consistency index" O. 74, retention index,. 0.81) resolved from a branch and bound analysis of presence/absence of mtDNA restriction sites using PAlp37. Numbers on the nodes represent proportion of 1.000 bootstrap replicates in which the particular node was ans per by ~15 - , hy- ria, use ]ila Nas Iro- 'ere b), :ion fere ntly lent 'ms. e in the .ittle INA INA asin lore :vier Relationships provided by allozymes were consistent with predictions from morphologyl8.19 and geography20. Therefore, relationships based on mtDNA probably do not reftect organ- ismal phylogeny, with the similarity of G. alraria and cypha mtDNA being especially problematic. Discordance could result from hybridization'6, stochastic inheritance of ancestral poly- morphisms21.22 or homoplasy2J, with the last two typified by poor phylogenetic resolution21.2J. Bootstrap analysis24, however, indicated the mtDNA phylogeny was well resolved; the node supporting monophyly of G. atraria and cypha was in 85% of 1,000 replicates (Fig. ld). Additionally, forcing G. atraria out- side a clade composed of cypha, elegans and robusta (as indi- cated by allozymic and morphological data) produced a mtDNA-based tree of 71 steps, 9 (15%) longer than the most parsimonious trees (Fig. ld). For allozyme data, forcing G. cypha and arraria into a sister-group relationship requires 82.2 steps, 22.9 (39%) longer than the most parsimonious allozyme- based tree. Therefore, discordance of mtDNA and allozyme topologies was most probably due to introgression of mtDNA among species. Similarity of mtDNAs from isolates (such as G. robusta jordani, seminuda, upper Little Colorado River robusla) with NATURE VOL 362 1 "PRIL 1993 ;. OO"l LETTERS TO NATURE c el~ps umillwla Sill R. J Bill Williams R. GiIaR. GRlcn R. robruIiI jorr/oni UlJlltf La ' t:yp/ttI Sevier R. Ilea- L. J-.n. JbicoJo, PynmidL. KIamaIh R. I S.o 4:0 3~0 2~0 I :0 Sequence divergencc (%) I 0.0 d 100 ek,... __wIa Salt R. 98 Bill Willilms R. Gila R. robruIiI Green R. jorr/oni upper La QPIta 5cvierR. J-.n. Bc.L. PynmidL JbicoJo, KIamadI R. 100 monophyletic in the 50% majority rule consensus tree. METHODS. Samples for phylogenetic analysis were obtained from the follow- ing locations (sample sizes for a110zymes and mtONA respectively~ G. auaria. Sevier R., Piute Co.. UT (18. 2) and Spring Cr (Bear lk. drainage) Bear lake Co.. 10 (15. 3k G. bicoIor. Pyramid lk.. Wash.:le Co~ NV (10. 3) and Sprague R (Klamath R drainage). Klamath Co.. OR (0. lk G. cypha, lower Uttle Colorado R. Coconino Co.. AI (20. 5k G. elegans. lk. Mohave. Mohave Co.. AI (20. 4-from captive stock at Dexter National Fish Hatchery and Technology Center. Dexter. l\M13k G. robusta jordani. pluvial White R. drainage. Uncoln Co.. NV (20. 4-from captive stock at Dexter National Fish Hatchery and Technology Center. Dexter. I\Mk G. rabusea. Burro Cr. (Bill Williams R. drainage). Yavapai Co.. AI (30. 4). Aravaipa Cr. (Gila R. drainage). Graham Co. (17.2). and Cherry Cr. (Salt R. drainage). Gila Co. (20. 6). Green R.. Moffat Co~ UT (20.7). and Chevelon Cr. (upper UtUe Colorado R. drainage). Navajo Co.. AI (21. 3k G. seminuda, Virgin R~ Mohave Co.. AI (17. 4). Samples of rabusea used only for analysis of introgression (Table 1) were from: Colorado R. near Rifle. Garfield Co.. Oebeque Canyon. Mesa Co.. CO and Gunnison R.. Delta Co.. CO. mtDNAs of G. cypha and elegans indicated introgression occur- red relatively recently. Ever-increasing aridity during Pleis- tocene2S and Holocene26 almost certainly altered aquatic habi- tats dramatically and may have promoted hybridization to pro- duce existing mosaics. The greater divergence between G. cypha and alraria mtDNAs indicates a far earlier hybridization event. Recent introgression between alraria and cypha would also be reftected in nuclear genes, yet diagnostic allozymes failed to provide such evidence (Table I). Therefore, phylogenetic simi- larity must have resulted from an old episode of introgression, retlecting ancestral drainage connection between the now- independent Bonneville and Colorado basins. Gila is the most morphologically diverse genus of minnows in western North America. Our results support the hypothesis that introgressive hybridization has played a significant role in generating this diversitylJ by providing additional genetic vari- ation for selection and drift to mould into locally distinctive phenotypes2.H. Thus, Colorado River Gila represent a complex of self-maintaining, genetically distinctive species which are capable of exchanging genetic material (comparable to syngameons in plants2). Botanists re~ize the importance of introgressive hybridization in evolution .21. Our results for Gila