I"; 5 3‘ T“ ' {3 t; 4,1 3w \ 5'“ ' . FAQ»; , ,. “95:27:33?" . \«fly RAH?“ ‘. K: 15?”. ’1 , \ I ‘ 0 km.“ “ "' “'33-‘14?! ‘4‘.) (“kg € 'I'a’xdg ' :fi FEE: r 'V" 4 ’ s 3‘23?” ‘ rive/33;; " L . , ‘— J L 3 ‘x'ii‘sé ‘9 I‘_ ‘ 1'31"! 1 9:: ‘A ~ ‘4” 'f fil.) may. ~ .,,~'. ' w, ’rt. {' ’5 ' 1‘ ’5: , $3 >4 X'?‘:}.« I w 3%?" ' - :24; _'. "“h- 1:2:h::;‘; .0 i" 1; VJ, :5, 4m . 1‘) A! .f. 45...; . VI, ‘3‘»1‘.‘ «farm wrz‘t.» 3-. ' :i'r‘ ”’3ng {$147, :(‘txgfll * = 7“,; - 23 1'“ Inf). 33:37.3 4 , hug-VJ?" .u (: 1;?! v1- v1; , :5” ',_:::'.. -;:._,. n a m, I I: .mmfir ' 3. mam; “W ,. ‘3. N .- ‘ “J W. )1 31‘33‘ ”Wm .. Luv-'1 all: _ '-«~3.- *3. . a): 1. E9; - £37. 'M' W ‘ h fr‘igm M... ' .... "' “12% ran ‘ ‘ ‘ - It'll 3' lfi§ WWW“) l l lllllllllllllllIllI ll ‘llll l l i ‘ uamy ‘” 3 1293 00609 1155 Mldligon State University v—v— This is to certify that the thesis entitled A MORPHOMETERIC ANALYSIS OF THE FIVE SUBSPECIES 0F SUS BARBATUS, THE BEARDED PIG presented by KAREN MARI MUDAR has been accepted towards fulfillment of the requirements for Master's degree in Zoology MWSLM Major professor Date April 14, 1989 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution PLACE IN RETURN BOX to move this checkout from your record. To AVOID FINES return on or baton dd. due. DATE DUE DATE DUE DATE DUE ‘ l %==‘1‘Lr’”’1 MSU Is An Affirmative Action/Equal Opportunity Institution f- A MORPHOMETERIC ANALYSIS OF THE FIVE SUBSPECIES OF SUS BARBATUS, THE BEARDED PIG by Karen Mari Mudar A THESIS submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Zoology 1989 ABSTRACT A MORPHOMETRJC ANALYSIS OF THE FIVE SUBSPECIES OF SUS BARBATUS, THE BEARDED PIG By Karen Mari Mudar Cranial variation among the five subspecies of the bearded pig Sus barbatus is an interesting problem of morphological diversity. Multivariate statistics are employed to explore the degree of diversity and characterize patterns of morphology among the populations. Measurements taken using a truss scheme are subjected to discriminant analysis. It was shown that size differences are present between three of the five subspecies. Standard principal component analysis added little new information. However, the shear procedure (Humphries et al. 1981) demonstrated that shape discrimination between three of the five subspecies was also possible. Patterns of shape and size differences are discussed with reference to the effects of island size on body size of artiodactyls, and geological reconstruction of dispersal patterns. The smallest sub-species were found on the oceanic islands of the Philippines, the largest on the Malaysian mainland and the large islands of Sumatra and Borneo. The populations on Palawan were larger than expected, based on island size. Size of this population may be influenced by genetic relationships with the largest subspecies on the next island, Borneo. Shape differences were most pronounced in the ventral region of the skull pertaining to the tooth row. It is suggested that these subspecies may be adapting to local ecological conditions. This study indicates that Sus barbatus barbatus and Sue barbatus oi should be combined as Sus barbatus barsz and that Sus barbatus philippensis and Sue barbatus cebifrons should be combined as Sus barbatus philippensis. ACKNOWLEDGEMENTS Like a tale that grew in the telling, this thesis has expanded greatly since its first inception. Initially it was a study of the interactions between human predators and suid prey in northeastern Luzon, Philippines. By the time of submission and defence, the thesis consisted of a morphometric analysis of the five subspecies of Sus barbatus. The people who have aided in the completion of this thesis deserve most of the credit; I was primarily a catalyst. I have been honored by the assistance that these people have given me, and am grateful for the opportunity to thank them here. Dr. Karl Huttter first introduced me to the Philippines in 1979, as a member of the Bais Anthropological Project. He provided financial support in 1981, when the suid sample from northeastern Luzon was collected, and again in 1983, when the museum collections were examined. He has also been an important. source of emotional support. Dr. P. Bion Griffin provided accomodations in the field in northeastern Luzon in 1981, and initiated trading relationships between myself and the local hunters. He continued to collect skulls in 1982, and made arrangements to ship them to the United States. Both of these men have been primary support, and deserve a very special thanks. This thesis would have been logistically impossible without them. Thanks goes to the Agta hunters, especially Tatayan, Sinebu, Galpong, and the hunters at Sapinet for cooperating in trading arrangements. Employees of Acme, Southern Plywood, and Cagayan Lumber Company provided transportation, and receive my thanks, especially Pete Galimba. My traveling companions, Marcus Griffin and Melinda Allen, deserve special recognition for their tolerance and patience. After returning from the Philippines in 1981, I attended the University of Michigan for one semester. During this time, Dr. Lawrence Heaney found working space for me at the Museum of Zoology, and provided a sounding board for ideas. His comments have been instrumental in the formation of the thesis topic and are acknowledged here. After my admittance to Michigan State University and the formation of my committee, the thesis began to take shape. I would like to thank my committee, Dr. Donald Straney, Dr. Alan Holman, and Dr. Norman Sauer for their patience and assistance. I owe a special debt to Dr. Alan Holman for arranging office space for me. Dr. Straney has been particularly helpful, both as an advisor and a friend. The office staff at the Zoology Department, especially Anne Tanner and Charmane Corcoran, have handled much paperwork and found financial support for me for three years, for which I am very grateful. In 1984 I was able to visit museum collections in Southeast Asia and Europe. Support for this trip came from the University of Michigan, (through Dr. Karl Hutterer), Michigan State University, and Michigan State University Museum, (through Dr. Kurt Dewhurst, and Dr. A. Chris Carmichael). Their assistance is acknowledged here. I would also like to thank the following people for assistance while woking in their countries: National Museum of the Philippines, Manila: Dr. Pedro Gonzales, Dr. Jess Peralta, Roger Sission, and Raymundo Esquerra; National University Museum, Singapore, Mrs. Yung; J abatan Museum, Kuching, Lucas Chin and Charles Leh; Muzium Sabah, Kota Kinabalu, Anwar Sullivan, Judith John Baptist; Muzeum Zoologici Bogoriense, Bogor, Indonesia,Mr. Bogadi; British Museum, London, Dr. Juliet Jewell; Cambridge Museum, Cambridge, Dr. Joysey; U.S. Museum of Natural History, Washington, D.C., Dr. Thorringten, David Schmidt; Field Museum of Natural History, Chicago, Dr. Bruce Patterson, Dr. Robert Tirnm. I also would like to thank C.H. Diong, who showed me the only live bearded pigs I ever saw, Napoleon and Petunia, at the Singapore Zoo. My woefully small sample of Sus barbatus cebt'fmns was augmented through the efforts of Paula Heideman, who took the time from his dissertation fieldwork to collect several specimens. I greatly appreciate the effort he made, and thank him here. My fellow students are gratefully acknowledged for their tolerance and support. Frank Knight spent many hours discussing thesis writing, Dan Bennack helped proof iv numbers, and Dean and Bette Premo opened their house to me while in Indonesia. Ken Creighton and Barb Lundrigen loaned me their apartments while in Washington DC. Alan J aslow was a thoughtful host while working in Chicago. I also benefitted from class discussions with Miriam Zeldich, Don Swidersky, Sergio Dos Rey, and Jim Zablotney. Brian Mavis, Paul Welch, and Chuck Hastings were especially patient instructors into the mysteries of personal computers, MTS, Mainframe, Textedit, SPSS, and MIDAS. Their technical assistance was invaluable. There is also a group of people who deserve a special thanks. These people had no interest in the thesis topic, but provided help and encouragement nonetheless. Mr. Alfonso Lim allowed me to work in his timber concessions, and arranged for logistical support. The Bombay to Borneo Bushbashers, Mr. Nick Cerra and Mr. Pat Jackson, opened their home to me on several occasions. Dr. Alice Dewey was a wonderful guide while staying in Jogyakarta. My sister Charlotte spent numerous hours proofreading data sheets, and my brothers Joe and Marc lent equipment and money. The support of this special group has been essential to finishing this thesis, and I thank them. There is also a group of people who believed I would finish even when I didn’t, for whom any measure of thanks is inadequate: Joe, Mary, and Tim. TABLE OF CONTENTS List of Tables ................................................ List of Figures ............................................... Introduction ................................................. Materials and methods ......................................... Discussion .................................................. Conclusion .................................................. Bibliography ................................................ Appendix A. List of Specimens Examined ........................... Appendix B. A verbal description of the landmarks used. ................ Appendix C. A verbal description of the measured distances. ............. vi vii 23 60 73 76 78 82 84 LIST OF TABLES Table 1 Means, standard deviations, and sample sizes for female Sus barbatus. . Table 2 Means, standard deviations, and sample sizes for male Sus barbatus. Table 3 Criteria used for aging individuals of Sus barbatus (from Matsche 1967) Table 4 Descriptive statistics associated with measurements periodically taken from one specimen during data collection as an accuracy check (N = 5). ....... Table 5 Principal components I, II, and III for mandibles of BARB males vs. females (isometry=.353). ........................................ Table 6 Principal components I, II, and III for lateral view of BARB males vs. females (isometry = .26) ......................................... Table 7 Principal components I, II, and III for dorsal view of BARB males vs. females (isometry=.27) ......................................... Table 8 Principal components I, II, and III for ventral view of BARB males vs. females (isometry=.27). ......................................... Table 9 Principal components I, II, and III for mandibles of BARB vs. 01 males, and BARB vs. FIL females (isometry=.353) .......................... Table 10 Principal components I, II, and III for lateral view of BARB vs. 01 males and BARB vs. FIL females. .................................. Table 11 Principal components I, II, and III for dorsal view of BARB vs. OI males and BARB vs. FIL females (isometry=.27) ...................... Table 12 Principal components I, II, and III for ventral view of BARB vs. 01 males and BARB vs. FIL females. .................................. Table 13 Principal components I, II, and III for mandibles, all OTUs (isometry= .353) ............................................... Table 14 Principal components I, II, and III for lateral view, all OTUs (isomtry = .26) ................................................ Table 15 Principal components I, II, and III for dorsal view, all OTUs (isometry= .27) ................................................ Table 16 Principal components I, II, and III for ventral view, all OTUs (isometry = .27) ................................................ Table 17 Varience and correlations from principal component analysis of all OTUs. ...................................................... Table 18 Principal components I, II, and III for age series of OTU FIL. ....... Table 19 Principal components I, II, and III for age series of OTU FIL, lateral View. ....................................................... 12 14 16 24 29 3O 31 32 33 35 36 38 44 46 49 51 54 55 56 Table 20 Principal components I, II, and III for age series of OTU FIL, dorsal view (isometry=.27) ............................................ Table 21 Principal components I, II, and III for age series of OTU ........... 57 58 LIST OF FIGURES Figure 1 Distribution of wild pigs in Southeast Asia. ..................... Figure 2 Width of inferior surface as a percentage of width of posterior surface in lower canines of adult male Sus, by species and subspecies. .............. Figure 3 Geographical distribution of the five subspecies of Sus barbatus ....... Figure 4 Map of major river valleys in northeastern Luzon, where sample of S. b. philippensis was collected. ...................................... Figure 5 a-d Landmarks chosen for measurement on crania of Sus barbatus. . . . Figure 6 a-d Box truss schemes for Sus barbatus ....................... Figure 7 Canonical variates 1 and 2 plotted for all five OTUs. .............. Figure 8 Discriminant Functions 1 and 2 to show the interrelations of the different samples of Sus barbatus (Groves 1981). of the different samples of Sus barbatus (Groves 1981). ......................................... Figure 9 Plot of PC I vs. PC II for mandibles from males, all OTUs. ......... Figure 10 Plot of PC I vs. PC III for mandibles from males, all OTUs. ........ Figure 11. Plot of Size vs. Shear for dorsal view of males, all OTUs. ......... Figure 12. Patterns of contrasts from principal component analysis, mandibles. a.) BARB males vs. females (PC II) b.) BARB vs. OI males (PC II) c.) FIL vs. BARB females (PC II) (1.) all OTUs female (PC II) e.) all OTUs males (PC II) f.) FIL age series (PC II) ......................................... Figure 13. Patterns of contrasts from principal component analysis, lateral. a.) BARB males vs. females (PC II) b.) BARB vs. 01 males (PC II) c.) FIL vs. BARB females (PC II) (1.) all OTUs female (PC II) e.) all OTUs males (SPC II) f.) FIL age series (PC II) ...................................... Figure 14. Patterns of contrasts from principal component analysis, dorsal view. a.) BARB males vs. females (PC I) b.) BARB vs. 01 males (PC I) c.) FIL vs. BARB females (PC H) d.) all OTUs female (PC II) e.) all OTUs males (PC II) f.) FIL age series (PC II) ......................................... Figure 15. Patterns of contrasts from principal component analysis, ventral view. a.) BARB males vs. females (PC I) b.) BARB vs. OI males (PC D c.) FIL vs. BARB females (PC II) d.) all OTUs female (PC II) e.) all OTUs males (PC II) f.) FIL age series (PC II) ......................................... 10 18 20 25 27 41 42 43 62 63 65 66 INTRODUCTION Four species of wild pig occur in Southeast Asia: Sus barbatus Muller (bearded pig), S. scrofa Linnaeus (domestic pig), S. celebensis Muller (Celebes pig), and S. verrucosus Miller (Javanese warty pig). S. verrucosus is found on Java and the adjacent islands of Bawean and Madura (see Figure 1). S. celebensis occurs on the Celebes islands of Sulawesi, Flores, and Halmahera. S. scrofa has been introduced to all of Southeast Asia inhabited by humans, but occurs indigenously on the Malaysian peninsula and Sumatra. Populations of S. barbatus occur sympatrically with wild Sus scrofa in Malaysia and Sumatra, but are the only wild pig on Borneo, Palawan, and the Philippines. Groves (1981) recently proposed a taxonomic revision of S. barbatus as part of a revision of the genus, recognizing five subspecies. Although they exhibit intra-spcific differences in size and occupy disjunct geographical areas, these five subspecies are distinct from sympatric S. scrofa in at least one characteristic. Ratios of the width of the inferior surface to width of the posterior surface of the male lower canine for all subspecies of S. barbatus fall outside the range for the same ratio in S. scrofa (see Figure 2). For the purposes of this study, I accept the taxonomy proposed by Groves. These subspecific assignments appear to be partly a function of size differences between populations, and partly based on geographical considerations. All wild pigs on Borneo were assigned to S. b. barbatus, while all non-scrofa wild pigs on Sumatra and the Malaysian mainland were classified as S. b. oi (see Figure 3). Reasons for uniting geographically discontinuous populations in subspecies oi were not given. Groves indicates that S. b. barbatus and S. b. oi may be separated on the basis of the length of the molar row as a percentage of the total tooth row, as the molar row is relatively smaller in S. b. .22 335:3 5 sea 33 as 553329 ~ 95mg 5.! com Swoosh: .m E 223.330 .m I 97w 3 . . 7 «33.3 .o m . . - . .4 4.11%? L . rillwuni m. «.3200 lllllllllllllllllll y.......u-m.,..my, 228 .m D ll HIHMIHIHIWHI . 1-; lllllllllllll L l w IHHIHIHIHIHIHH seesaw ,_ f IIIIIIIIIII ooEom c3163.... .2 853...... s. celobonslsl s. b.phmppensls2 S. harbours3 4 S. verrucosus 5 S. scrofa 70 so so 160 110 120 130 140 150 160 170 180 N-4 (Groves;1981) . N822 (Groves;1981,Mudar,n.d.) N-21 (Groves:1981) N-19 (Grovesrl981) . N-S4 (Groves;1981) MbUMH O Figune2 Width of inferior surface as a percentage of width of posterior surface in lower canines of adult male Sus, by species and subspecies. surface in lower canines of adult male Sus, by species and subspecies. oi. This is interpreted as an example of character displacement (Groves 1981;51), as S. b. oi occurs sympatrically with S. scrofa. He also noted that oi has shorter whiskers and a shorter molar row in relation to total tooth row size than barbatus. Mohr (in Groves 1981) noted that the body of oi is more bilaterally flattened than barbatus. There appears to be no difference in size between the two subspecies. All populations found in Palawan and adjacent islands were assigned to S. b. ahoenobarbus. This subspecies is smaller than barbatus and oi. Sanborn (1952) also noted that this form exhibits a particularly short palate. All wild pigs in the Philippine archepelago with the exception of Negros and Cebu were placed in S. b. philippensis. This subspecies has a shorter maxillary diastema and a smaller body size than the three preceeding subspecies (Groves 1981). Populations of bearded pigs from the islands of Negros and Cebu were recognized by Heude (1888) as a separate species because of their small size. This distinction is maintained by Groves, who assigned the populations of these islands to a separate subspecies, S. b. cebifrons. He placed the Philippine pigs in S. barbatus, but suggested that they may be a separate species, most closely related to S. barbatus (1981;50). Species such as the bearded pig are particularly problematic to distinguish taxonomically because populations which are, potentially, closely related are allopatrically distributed. An important question concerning this species is how much variation between the populations is likely to be ecological in origin, and how much is likely to be genetic. Pervasive differences in size and shape between the populations may be a result of long isolation, or adaptation to specific ecological conditions. The objective of this study is to assess intra-specific variation in cranial morphology through an exploration of shape as well as size differences. I wish to address the following questions: What is the pattern of size differences between the different subspecies? Can the subspecies be separated on the basis of shape differences? Do these shape differences correspond in groupings to those isolated by size differences? Do shape differences consist £53.35 3m. .3 momoeomnsm 3: 2.: mo gazebo? icing—meow ”83E Mindanao . . Southeast Asia 'O C S .E (U E 2 x. o o to o 'o -m U} :0 «onto .33., cogs-0° DOQOO “h c” “vb-no an: O |..D_°>o -acz-flw 9999-90 sees‘iw wwwww' O Z.Hmwc.m mmmfinmmmd Nevfifiwmué huufifimmv.” uaufiumbw.» mm: hmmfiumumd Nemfinmmhé mmmnfihumé cvudwmfié unqfivwcd. vNE ww~.H 2:4 wwfifiwmmé bmv.u_uw~m.m mmfifimwmd musing-ac.” ”NE mefimmamS mum.Hmmc.w Nemdficoaé «Sod—”ammé m~¢.HmNm.v NN—z mmvnfimvmfi hmmafihmwfi Ncmfiugvé caudnuumd wmnfimbmné ~32 mamfivmmnfi ucmfinmvvfi bvnnfimvmd cmdfiummmé :Ndfimeé ONE :54 Hbavd maafiffind unqflcmbd meJfimeé mum.fimmc.m mu: mob.fim~m.fi vmmdufimq: EQHNcNd ommnmhbmfi hwvfimmmmfi mg: wmm.H:v.w wfimfiuwmmd ~a—.H~eb.m Naadficmhd :tmnfiemaé :2 mbwfiuvwbdfi awhfiucwmd“ wwmfinueo.» ”Can—”:36 9.33.”:me $22 mvvdfizmwm watfivmmd ”aqfifiemé mmmdefiv :m.fimvm.v 52 mvm.fi~m~.v uvnfifimmoé wmmfiuhmv.” mmfifimmmd omvfiummb.» v22 mama“ Sung. mmmfiuwwvs mmnfiuchué cmmdficmmd humawmmoé m2). ~wo.fimmc.m cmmfimmmud mwmdfibwmfi @bndfimmmé nvmfivmmmé N22 menu—“$56 maps—”mafia mwvnficmufi flawflbmad “Sud—”N216 :2 swan—H :bé cmbfimmomd mmmfihaaé mvfificwwd enqfivmvi A: 2 see; “2592 wvmnnmdhmfi cmmdwuendfi shin—“3.9m mhbdvomh.» 92 a: A an ~Nwd~ mvvé ”:59: Egg H vmefifi acqfivvmd wwwfiuwmwd w: «.34 fivmmi mmcé Hmmmé ~nb.meN.m mfiufinmmmé bmmafihmmé :2 Ham.“ “5qu mac." Hvami vnm.H¢am.m cam.fimmm.m envfiummmd 92 www.— “‘55:. weanfivmms unqficcad mvc.fimmm.m $0.“;de m2 4 five 52 mam; H9233 Nvué H vmwéa vaufiuucwéu wmmnfiawmm— =2 mum; fimmm.m~ awmfinvmmdfi mmhfifiwwmAH hmméummmd mcmdfib—wd 32 «mg; H234" uomdhwmsn: mmb.H—mb.m b09fihfibé VNDJflZbS NE Venn—“2.9m m~m.H—mm.m uamdmmwmd zmdfibomé munfiuannm :2 2:30 3.8—0 $N6§ mm36 :2 83:53)— omYHcfiYo ccm.Hch.© bvmnfimwwd cmudmmwbd hNNJ—nwabé $32 awe; Hemhé “mmfiucfimé ovvfiuumfim wacfiémbé mafifimwmé mm: cwbfiubwvé umafinmmbé NNQHmch. «Ngflcmmd memfiunmwd :22 mam.“ unmask. wccé mews amvdfiavcé amvfiucmmd vmmdficacé cm: m~v.fimom.v mmwfiummmé acmfifiobvd ucnfiuummfi wuméuofiafi was. Nah.fiwnw.b fingfibmwfi vain—5N1.» abmfiuncmé vandfimmfid v32 gméfiwvmdn hSAvanén mméfihmméfi wmvfiuhzd Nuofiubvmd an: bum; Hcfid wvvfinmvcs cwmnfiwvmd mam.fio«é.w uuodfimwvé N22 avm.Hch.m comfinwwcfi cvv.fiwmm.m awmxficmmé aaafiffiéfi ~22 ammammmmé mfivfinwvafi ficmfinacmé Ozark?” mmmnnmwmd an: 53$.”de wvwfimmuwé mumafivvw.” ave.fimv>.a mafifimuc.» was. 254 mewd mmmdfihcwfi «mmfiuwmmé vafiumafim 553.5de mm: NpQHaumé binfiwbcd VEHmeNé hoonfibvmé Navfiuomm.» 5N2 vcbfifimmmd mama—”356 Nvafimed baud—”33d oaafifiomhd 022 3:20 3.3—0 Awmémv m¢ @2353 _ 22:. 14 bwv.fiwm>.m SQHv—cd cmNA Hmcmd pan—”Nb?” wNw.Hm~v.v mNE vv>.Hmav.w omvfinmbmd Nmmdubcmé mvcfiuwai mbmfiummmé «NE wmcé vam.v ENS—£32. Quad—“mew.” cwcfinbemN mc~.HvNN.n MN: Vagficwmd bmwfiuvmwd ~NN.HN®>.¢ fizfimowmé wvm.HcNo.m NN: acvdfiabmd Smfiuwbvd mcvfiuthd N¢O.Hcmmé mwmdumboé “NS— mvwdmmmwfi NuvAuwaS mofififibbé ~mO.HNH¢.m EQHNZd AN: ”:4 HoN~d~ mchnNH—d— Nwmfinwcafi bbv.bem.m ch.H®®N.m 32 chA “5.93 use.“ HNNvdm wvwd—uvchZ bad—“36d mbmnfiwde mu: NNNA vacd m~>.HNmm.m mwmfiucmfih @wmdmcawfi wafiHmmd S 2 wmafiubbméu vmcé HmNm.N~ hcmfivaNéu NVbfiuchd «.mmdflfibmd £2 minfiwamé hmwn—uwwfié bogficwmé . bvadnomvé awv.HNNm.v £2 mhvfiuuzflv Nmmdwmwbé moo.Hmaw.m thfiummNé wean—”3N4 3 Z VNwJfivad nmwdgwmfi vmv.HNww.c magnuucwfib maNdmwuod 82 mwwfimvmwd Nam; meNd— mwmdfithd :v.Hmmv.m :v.HHe—.~. N. Z mmm.fivmcd~ Nmafiqucdu mcméfimznw “mad—”33.5 awvdvmmmfi :2 «Nmfiuvvmd w~m.Hc—m.w mmm.fi~Nw.m eccé Hmeé road—599m 6:2 mwcé H :"N.: mvmé Hawwéfi wmaé H9692 bvafinmcwdfi 3&4 vavdm a2 :54 Humwéfi mva HNc~.: 2:; H9692 ~24 Hmomda >mm.H~mmd m2 NfiwfififiwNé ~am.ch~.v w~w.HN®m.N oovfiNocN Nm—JHNVQN >2 2:4 Hebvé 2: A “mg: .m bwwdwmmhd ch.Hbmc.v mwvnfi fimmd c2 Eb.— Hmmmd wVNé HNNNS vat. HNNQV mmH.Hccm.v ”3%“;de 52 has; HmcaAN was; vaoNN mmN.— H 2:..: pawn—5:3: acfiéfimmwéfi :2 mam.wac.: mNmA Hang: wow; HVNNJA mmqfimchd Nmadwzqz NE NE.. Hmha.N~ mam: “3.322 03.4 £634 u NdethvS NNp.Homm.w N2 mNm.HwhN.v Evamuzé NmNnfimmN.” uv—J—ucaNN me.vam.N :2 3520 81m: —0 h :mde m2 653.3 36. £2: .5. no»? mEEau EB 6555.5“. c.8233 6:32 N wish. 15 .mmificafi E 535:: £83 628.8 3 53:5: .5ng do»? £955 3 Sum.— mmmmfceaa E 33:52. bmmfiuvfimé uvm.Hme.v ~mm.HNmb.m «wan—”>84" ow~.Hw~m.m a: Nov.Hmvw.: mmaé Hhmwéfi 3&4 Hcoud wch—uuovfi vcvfiubchg. w—z mac; H533; m—OA Hbcwfim 2b; H umvé— mmvd—ummmén wwmfiuazéa :2 ~ma.vaw.> w:..HNcN.w mmm.Hvow.m mwN.Hm:..m «mm.Hmba.v 02 $54 Hummfiu mum; vaofifi mum; Hmmwdu an.Hbvv.G~ acm.HmwN.= m2 wwm.Hmmb.b mZnHmwbS :m.Hab~.c awN.Hcmm.v mc¢.Hw~e.m :2 mmv.Hm:..m wmwdfihwwd avm.H~.vm.v m3.Hmcw.m awn.chm.m m2 Nem.Hva.: «2.1359: bmv.Hch.m ~ac.H¢om.w hhw.HNvo.~. N2 wm>.Hmw~.: cmcé HammA: mwm.Hmvv.b Hwod—ubvmd mvw.Hm©N.o :2 835:“: cmm.Hmvm.w amm.Hmom.m unfiHmme wwfiHan m>~.Hwflc.m an: anh.H~mm.v mawwd—ngé Nam.Hmmw.m mmcfiHcmmé wcm.HvN~.N mm: cam; chm.m ~am.cho.m vvw.Hwbv.v vnv.HNov.m acv.vam.m hm: Own; Hmwud I: A mebd vwmé H ~waé cVN.H°~@.m b©N.Hmoh.m on: wam.thv.v cm>.Hcmv.v mam.Hmva.m vcm.H°~a.N N¢~.H~mc.m 5:2 ch.NHwa~.w mmné Hmmm.» mum.Hw:.~. mum.HN:.m mean—”mafia v22 NNaA H ~ “VA: awn; Hmomdfi mwmé H 23.2 vwcé Hambd mum.Hmcm.oH mum—2 mum; Hcmmd ~N~A Hbvmd mcw.mew.o ianNwwé wimHNmmé an: awcé Hana.” cmw.Hmhm.m wwmfiummcd eoc.HNwN.N Nam.Hmvw.N :22 £54 Hmowé 2.0. H and cam..— H _ vud achwaé awN.Hnmc.v cm: chA Hmwmd «ethcmd mmb.Hwow.v aquch va.Hcmv.a mu: Nam.Hv~.m.a mum.HaaN.m mvmé Haves NuN.Hm©N.© bamfiuccmé was. cum; Hmwwfi mwmé Hand vuw.H—cw.w mam.Hmve.m NNm.H~ww.m 5N5— bacé Huvmd hmmfiua—mé wam.Homm.v vm~.Hm~.w.N enm.Hh~m.m was. 3:50 $~£C ~O :mdvv m¢0 = >0 26 differences which exist between taxa are overwhelmed in this analysis by the large size differences between taxa. I compared the results of this analysis with a discriminant analysis performed by Groves. Although not specifically stated, I infer from accompanying tables that Groves’ sample of adult males used for his analysis consisted of 39 S. b. barbatus, 15 S. b. oi, 8 S. b. ahoenobarbus, 12 S. b. philippensis, and 2 S. b. cebifrons (Groves 1981:Table 4). Sixteen measurements which depended on the presence of the right and left side of associated skull and mandible were taken. Measurement protocol in the anterior skull and tooth row was identical to the one I used. Measurements concerned with the posterior skull emphasized over-all skull dimensions which were not necessarily referenced to a homologous point. Analyses were performed on transformed variables, consisting of the original measurement divided by the basal length raised to the power of the allometry coefficient (Groves 1981:68). A comparison of the results of this analysis with a discriminant analysis performed by Groves indicates that I obtained a clearer separation using this method (see Figure 8). I attribute these results to increased sample size, and landmark-based measurements. However, discriminant analysis attempts to create a composite variable which best separates the populations, and is not sensitive to complexes of variables which may discriminate between populations. To address questions of shape change I used principal component analysis and the shear procedure. Principal Component Analysis The results of discriminant analysis indicate that size differences between the OTUs separated the five groups into three. I performed conventional principal component analysis and the shear procedure to explore possible shape differences within and between the groups. Analyses were performed on all OTUs combined, as well as groups of OTUs combined by size. 27 0 Pole wan phtllppin‘" --------- ~ I I I [.1 1», I t \ \ .Bangka Figure 8 Discriminant Functions 1 and 2 to show the interrelations of the different samples of Sus barbatus (Groves 1981). 28 In adult populations, static allometry may be reflected in loadings on the first principal component (Jolicoer, 1963). Values greater than isometry (= 1/n, where n= the square root of the number of variables used in the analysis) describe positive allometry with respect to size; values less than isometry indicate negative allometry with respect to size. BARB males vs. BARB females Sexual dimorphism, readily apparent in size and shape differences in the lower canine, and in pronounced saggital cresting in older males, may confound assessment of shape differences between OTUs. Therefore, I combined both sexes of the same OTU and performed principal component analysis to isolate cranial regions are influenced by sexual dimorphism between males and females. Unfortunately, BARB was the only OTU with an adequate sample size to permit comparison. Mandibles. All loadings on the first principal component were positive and, with the exception of the premolar tooth row (M9), above 0.650 (see Table 5). I interpret this as a size axis. Premolar row and posterior length (M8) formed a contrast with ramus height (MG) in the second principal component. This component is a contrast of a vertical dimension with a horizontal one. The general pattern of negative and positive loadings suggest that length varies with respect to height of mandible, as vertical measurements of both posterior and anterior portions (M3, 4, 6) are negative, while horizontal measurements (M5, 8, 9) are positive. Horizontal measurements in proportion to the vertical measurements are greater in males than in females. Premolar tooth row forms a part of the contrast in all three components. M- All loadings on the first principal component are large and positive, with the exception of posterior nasal length (M5) (see Table 6). The second component is a contrast of posterior nasal length (M5) with three anterior nasal measurements (M2, 25, 27). All share a common landmark in the nasal-premaxilla-maxilla suture (C). The anterior nasal measurements are in a portion of the skull which contain the dimorphic canine. The third 29 Table 5 Principal Components I, II, and III for mandibles of BARB males vs. females (isometry=.353). Variable PC I PC II PC 111 M2 .902 -.028 -.223 M3 .962 -.059 -.043 M4 .899 -.098 .112 M5 .794 .454 -.054 M6 .712 -.584 .247 M7 .900 .127 -.053 M8 .686 .533 -.308 M9 .476 .515 .675 % variance 64.86 78.94 87.45 principal component is a contrast between a measure of frontal length (MI 1) and anterior nasal height (M29). M. The first component accounted for only 25% of the variance, and contained both positive and negative loadings (see Table 7). This indicates that there is shape as well as size information in this component. It constitutes a contrast between nasal length (M4) and frontal length (M8). The second component is a contrast of nasal width (M7) with oblique nasal length (MG). All four measurements share a common landmark in the nasal-frontal suture (D). The third component did not include any strong contrasts between measurements. The third component may be a residual size axis, as no clear patterns identifying contrasting portions of the skull were isolated. m. The first component of analysis of the ventral data subset also accounted for a relatively small amount of variance, 29% (see Table 8). This component included both negative and positive numbers, and constituted a contrast between premolar row (M23) and posterior width (M3 7) with two measurements of anterior length (M32, 33) which share a common landmark, the premaxilla-maxilla suture (P). Note that all Principal Components I, II, and III for lateral view of BARB males vs. females (isometry=.26). 30 Table 6 Variable PC I PC II PC III M2 .796 -.393 .148 M5 .215 .939 .228 M11 .600 -.274 -.480 M16 .840 .164 -.129 M17 .837 .217 -.068 M18 .577 -.150 -.078 M19 .744 .139 -.033 M21 .767 -.003 -.237 M22 .690 .317 .388 M24 .633 .365 .213 M25 .696 —.571 .222 M26 .767 .171 -.211 M27 .587 -.352 .662 M28 .829 .191 -.302 M29 .647 .092 -.568 % variance 49 62 72 measurements pertaining to length at midline (M30, 33, 36, 39) exhibit negative loadings. The second component contained almost all positive numbers and may be a residual size axis. Pre-molar row (M23) and oblique palate width (M34) exhibit negative loadings. They share a landmark in the anterior PMZ. The third component is a contrast between three posterior measurements (M34, 36, and 37) bearing a common landmark, the maxilla-palatine suture (Q). Discussion. Shape differences between sexes are localized in the region of the dimorphic canine. This is exhibited most clearly by analysis of the lateral data, where major shape differences involve the relative placement of the nasal-premaxilla-maxilla suture. This was anticipated, as the region provides structural support for the large canine in males. 31 Table 7 Principal Components 1, II, and III for dorsal view of BARB males vs. females (isometry=.27). Variable PC I PC II PC III M1 .228 .241 .417 M3 .154 .206 .643 M4 .988 -.127 -.038 M6 .483 .753 -.025 M7 .020 -.805 .458 M8 -.803 .195 .538 M9 .381 .516 .635 M10 .680 .085 .528 M11 -.237 .644 .394 M12 .139 .233 .781 M13 .502 .166 .551 M14 .647 .223 .594 M15 -.199 .219 .415 % variance 25 42 69 Shape differences between sexes are also apparent in relative length of skull, which is illustrated by analysis of dorsal, mandible, and ventral data. There is a relative lengthening in males, in comparison to mandible height, or skull width. This analysis indicates that there is a significant degree of sexual dimorphism present in this subspecies. The presence of a dimorphic canine and saggital cresting in males for all subspecies suggests that this is a general characteristic. Therefore, sexes are analyzed separately, to avoid confusing sexual dimorphism with intra-specific morphological variation. BARB males vs.OI males BARB and 01 constitute one of the three groups isolated by discriminant analysis. I performed principal component analysis on males of these OTUs alone to explore possible shape differences. In general, the amount of variance explained by each component of 32 Table 8 Principal Components I,II, and III for ventral view of BARB males vs. females (isometry=.27) Variable PC I PC II PC III M19 .059 .695 .226 M23 .991 -.091 .066 M26 .270 .774 -.059 M28 .276 .862 .176 M30 -.206 .563 -.282 M31 .521 .638 .384 M32 -.562 .730 .107 M33 -.662 .482 .425 M34 .587 -.161 .657 M36 -.370 .695 -.523 M37 .869 .163 -.415 M38 .465 .529 .072 M39 -.324 .020 -.005 % variance 29 61 71 each data set was low, and comparable in magnitude to variance associated with principal component analysis performed on each OTU separately. Mandible. All loadings for the first component were greater than isometry (=.357) (see Table 9) and accounted for less than half the total variance. The second component is a contrast between ramus height (M6) and posterior length (M8), which share a landmark. Note, too, that anterior horizontal measurements M2 and M5 form a contrast to ramus height. This differs from the mandible analysis of BARB males versus females, where all these measurements shared the same sign. The third component was a contrast between anterior and posterior mandible length (M2, 5), accounting for 4% of the variance. They share a landmark in the mental prominance (U). M. All loadings on the first component were greater than isometry, but accounted for 28% of the total variance (see Table 10). The second component is a contrast between. distal nasal length (M5) and nasal height (N125). Both share the nasal- 33 Table 9 Principal Components 1, II, and III for mandibles of BARB vs. 01 males and BARB vs. FIL females (isometry=.353). Variable PC 1 PC 11 PC III BARB males vs. 01 males M2 .574 -.420 .649 M3 .907 .023 .228 M4 .881 -.002 -.269 M5 .713 -.398 -.521 M6 .602 .758 .083 M7 .881 .078 -.110 M8 .589 -.542 -.325 M9 .611 .115 -.395 % variance 53.68 68.98 72.70 FIL females vs. BARB females M2 .980 .036 -.064 M3 .956 -.043 .066 M4 .975 -.085 -.002 M5 .981 .115 -.094 M6 .965 -.247 .059 M7 .995 .008 -.038 M8 .974 .158 -.129 M9 .906 .246 .342 % variance 93.44 95.57 97.51 34 premaxilla-maxilla suture point (C) as a landmark. The third component is a contrast between two measures of anterior nasal width (M27, 29), accounting for 7.5% of the variance. Boga}. The first component accounted for only 15% of the total variance, and contained both positive and negative numbers (see Table 11). It is a contrast between nasal length (M6) and nasal width (M 7), sharing a landmark in the nasal-frontal suture (D). Loadings for the second component were all positive and generally greater than isometry (= .27). This may suggest that the second component is a measure of size differences. The third component is a contrast between two measures of nasal length (M3,M5), sharing a common landmark (C). 32%. Again, the loadings on the first component are both positive and negative numbers (see Table 12). It is a contrast between premolar row (M23), and palate length and width (M33, 34). The second component is a weak contrast of a cluster of unassociated measurements in unrelated areas of the cranium (M23, premolar row; M34, oblique palate width; M39, distal cranium length) with premolar to incisor length (MZ6), incisor row (M28), and distal length (M36). It may be tentatively considered a size factor, because of the preponderance of positive loadings. The third component is a contrast between a measure of anterior width (M3 1) and posterior length (M36). It accounts for 11.3% of the variance. Discussion. The relatively small amount of variance which is accounted for by the analyses of the data sets of BARB and 01 OTU males suggest that consistent size and shape differences between these OTUs are minimal. Similarly weak patterns of contrast and low percentages of explained variance would be found in a principal component analysis of a single population, and may be accounted for by normal intra-populational variability. Within this combined sample of BARB and OI males, patterns of variance magnitude indicates that these shape differences are more important than size differences in data Principal components I, II, and III for lateral view of BARB vs. OI males and BARB vs. FIL females. 35 Table 10 Variable J PC I 1 PC 11 PC III BARB male vs. 01 male M2 .610 -.487 -.125 M5 .304 .937 -.098 M11 .327 -.353 .562 M16 .663 .199 .321 M17 .416 .200 -.064 M18 .509 -.255 .011 M19 .604 -.006 -.047 M21 .500 .072 .322 M22 .647 .297 -.347 M24 .434 .526 -.129 M25 .602 -.581 -.240 M26 .577 .080 .502 M27 .668 -.251 -.470 M28 .466 .167 .586 M29 .396 -.086 .684 % variance 27.86 42.50 56.0 FIL females vs. BARB females M2 .923 -.277 .212 M5 .931 -.167 -.316 M11 .823 -.249 .248 M16 .925 -.275 .111 M17 .938 -.220 .062 M18 .796 , .603 .013 ' M19 .921 .374 -.011 M21 .642 .754 .076 M22 .957 -.228 .010 M24 .855 -.315 .019 M25 .863 -.275 .345 M26 .938 -.248 -.023 M27 .920 -.268 .233 M28 , .935 -.293 .009 M29 .940 -.261 .011 % variance 79.36 91.86 94.55 36 Table 11 Principal Components LI], and III for dorsal view of BARB vs.OI males and BARB vs. FIL females (isometry=.27). Variable PC I PC 11 PC III BARB males vs. 01 males M1 -.055 .539 .372 M3 .083 .243 .649 M4 -.279 .609 .269 M5 .453 .664 -.571 M6 -.534 .716 -.440 M7 .980 .153 -.090 M8 .578 .536 .441 M9 -.222 .644 .619 M10 .028 .629 .535 M11 -.399 .293 .586 M12 .205 .547 .521 M13 .145 .571 .274 M14 -.021 .777 .334 M15 .155 .306 .139 % variance 15.49 45.52 65.83 FIL females vs. BARB females M1 .944 .013 .037 M3 .968 -.041 .142 M4 .948 .199 .105 M5 .944 .154 -.263 M6 .395 .893 -.045 M7 .947 -.272 -.100 M8 .978 -.074 .093 M9 .928 .105 .305 M10 .756 .051 .442 M11 .855 .121 .372 M12 .972 .033 .096 M13 .893 .008 .016 M14 .430 -.062 .347 M15 .839 -.025 -.018 % variance 74.41 81.38 86.24 37 pertaining to horizontal measurements (dorsal and ventral subsets). Conversely, size differences were more appareant in analyses of vertical plane subsets (lateral, mandible). Analyses of the horizontal subsets delineate aspects of shape differences which are underlying factors of principal component analyses. Differences were restricted to the anterior portions of the skull, and delineate aspects of change in palate and nasal width with respect to length. Position of the premolar row (MZ3) in relation to the mid-line of the skull is also thought to be an underlying factor. BARB females vs. FIL females Principal component analysis of the smallest and largest OTUs was performed to further examine patterns of shape and size differences. Females of BARB and FIL OTUs were used, as these had the largest sample sizes. Mandible. The first component was clearly a size axis as all loadings were above 0.900 (see Table 9). The second and third component combined accounted for less than 5% of the variance, and the contrasts were weak. The second component contrasted ramus height (M6) and premolar (M9). The third component contrasted premolar row with ramus width (M8). .I_._at_e_131_. The first component was also a size axis, accounting for 80% of the variance. The second component isolated three posterior measurements (M 18, 19, 21) sharing the distal end of the M3 (K) as a landmark. All other loadings were negative and ranged from -0.315 to -0.167 (see Table 10). The third component is a contrast of two anterior measurements (M5 and 25) sharing the nasal-premaxilla suture (C). M. All loadings on the first component were well above isometry (see Table 1 1). The second component of the dorsal subset is a contrast between posterior nasal width (M7) and nasal length (M6). The third component isolated three measurements with negative loadings which formed a triangle in the posterior nasal region (M5, 6, 7). Except 38 Table 12 Principal Components LE, and III for ventral view of BARB vs. 01 males and BARB vs. 01 females Variable PC I PC 11 1 PC 111 BARB males vs. 01 males M19 .000 .440 .476 M23 .996 -.025 .003 M26 .136 .696 .002 M28 .191 .777 —.193 M30 -.298 .432 .417 M31 .517 .495 -.601 M32 -.704 .556 -.375 M33 -.749 .411 -.244 M34 .681 -.008 .114 M36 -.505 .663 .333 M37 .894 .184 -.082 M38 .451 .543 .587 M39 -.323 -.097 -.066 % variance 32.95 55.80 67.15 FIL females vs. BARB females M19 .888 .415 -.176 M23 .886 -.210 .079 M26 .958 —.186 .040 M28 .954 -.240 .018 M30 .946 -.217 .105 M31 .710 -.231 -.126 M32 .862 -.340 -.002 M33 .958 -.141 -.189 M34 .914 -.108 -.317 M36 .949 -.131 .251 M37 .681 .616 .367 M38 .980 .081 -.104 M39 .834 -.220 .055 % variance 79.36 87.10 90.27 39 for distal length (M15), the rest of the loadings were positive. The largest was 0.442, oblique cranial width (M10). 2M. Again, the first component is a size axis. Positive loadings from the second component isolated two measurements of the posterior portion of the cranium (Ml9, 37). The remainder of the loadings were negative, ranging from -0.108 to -0.340 (see Table 12). The third component contrasted a measure of posterior and anterior width (M34, 37). They share a common landmark in the maxilla-palatine suture (Q). Discussion. The results of the principal component analysis suggests that, overall, size is the best discriminator between these OTUs. The second and third components combined account for less than 16% of the variance in all the analyses. Contrasts within the second component pertain to the posterior region of the skull. Results of analyses of all the subsets concur, and many of the contrasts shared landmarks between the subsets. The position of the posterior margin of the third molar figures prominantly in these contrasts, and suggest that the position of the molar row may be an underlying factor. All OTUs1 Principal component analysis of all subsets segregated by sex were performed to examine differences among all populations. Factor loadings from principal component analyses from more than two samples are problematic to interpret, as it is difficult to isolate the OTUs which are contributing most significantly to total variance. However, plots of these components can contribute much information for assessment of general morphological divergence. In analyses of data from males and females performed separately, the first component describes a size vector in which all factor loadings have the same sign and similar values. Plots of PC I vs. PC II and PC I vs. PC 111 for all data subsets separated each OTU along PC I, indicating that size was the best descrirninator for each OTU. 1 Because of small sample size CEBI is excluded from this analysis 40 These results supported the results from discriminant analysis. BARB and OI consistantly grouped together (see Figure 9 and 10 for representative plots). AHO and FIL maintained a separate status in plots from all analyses. To isolate and remove the effects of size, the data were subjected to the shear procedure. Results indicate that shape differences occurred primarily in the horizontal planes of the skull (dorsal and ventral views). Plotting Sheared component II against size resulted in clear separation of OTUs into three groups for both sexes (see Figure 11 for representative plot). Results of standard principal component analysis and the shear procedure are discussed below. Mandible. For each sex the first component of a standard principal component analysis accounted for a high percentage of the total variance, and contained only positive numbers (see Table 13). For females, the second component is a contrast between ramus height (M6) and premolar row (M9). The third component is a contrast between posterior length (M8) and premolar row. For males the second component is a contrast between ramus height (M6) and posterior length (M8). This same contrast occurred in analysis of BARB vs. OI males (see Table 9). The third component is a contrast between anterior length (M2) and ramus height (M6). Mandible data were not subjected to the shear procedure, as the loadings from the first and second component were not correlated (females= -0.4524, males= -0.3505 (p < .05)), indicating that shape differences detected on axes II and III are independent of size differences among OTUs. However, these shape differences explain very little total variation (5%) not explain by size (92%). M. The first component of standard principal component analysis for each sex describes a size vector (see Table 14). All loadings are positive; the variance is 79% for females and 78% for males. The second principal component in the data for females isolated three measurements sharing the same landmark, distal M3 (K) in the posterior 41 PC ll FIL AHO OI BARB PCI Figure 9 Plot of PC I vs. PC II for mandibles from males, all OTUs. 42 PC In PCI Figure 10 Plot of PC I vs. PC HI for mandibles from males, all OTUs. SHEAR Hm AHO BARB SIZE Figure 11 Plot of size vs. shear for dorsal view of males, all OTUs. 44 Table 13 Principal Components 1, II, and III for mandibles, all OTUs. (isometry = .353) Variable J PC I PC 11 PC In Females M2 .422 .133 .170 M3 .283 -.266 -.222 M4 .305 -.166 .031 M5 .338 .266 .291 M6 .441 -.648 -.203 M7 .392 -.030 .153 M8 .335 .339 .371 M9 .267 .523 -.794 % variance 93.18 95.57 97.12 Males M2 .381 -.040 -.741 M3 .339 -.299 -.359 M4 .357 .019 .091 M5 .355 .424 .163 M6 .401 -.674 .401 M7 .370 .077 .167 M8 .345 .605 .031 M9 .258 .105 .309 %variance 91.25 94.10 96.86 45 region of the skull pertaining to skull height (M18, 19, 21). These measurements were a contrast with a measure of anterior height (M26). The third component contains a contrast between posterior nasal length (MS), and height (M25) sharing a landmark in the nasal- premaxilla-maxilla suture (C). As the first and second components were correlated (r= - 0.20 (p<.05)), the shear procedure was not performed. In males, these components were correlated (r= -0.67 (p<.05)); the shear procedure was performed. The pattern of factor loadings for the sheared second component for males was little different than that for PC II. Sheared component 11 for males is a contrast between posterior nasal length (M5) and nasal height M25). The second component for analysis of BARB vs. 01 male also contained the same contrast (see Table 10). Loadings on the sheared PC 111 for males were similar to those from the standard PC III. It is a contrast of a posterior height measurement (M18) with the canine region (ll/[26). Although neither measurement encompasses the cheek teeth, premolar and molar teeth provide landmarks for both. M. The first component of an analysis of data from females is a size vector, containing positive numbers ranging from 0.070 to 0.558 (see Table 15). I then performed the shear procedure, because the first two components of the principal component analysis were significantly correlated (r = - 0.67 (p < .05)). The sheared second component showed no contrasts, as all loadings were positive and with one exception, less than 0.1. The sheared second component demonstrated that there was no residual shape information present, once the information in PC II correlated with size had been removed. Interpretations concerning the second component of the standard principal component analysis would have been erroneous if I had not performed the shear procedure on this data. Shape information is contained in the third shear component. Contrasts in the sheared third component consist of distal nasal length and width (M6, 7) sharing a landmark (D) in the nasal-frontal suture, and two measures of distal width (MIO, 14). 46 ~w.mm hcda med» 3:33.» a» 35.! 3:. ram. was «mm. mg. nmw. >32 :5. :N. was. was awe. 2:. nu“. ma: mac. “mg. 2%. was to. N8. “mm. was m3. mmm.l mafi. ~22 35.! Nam... new. a“: 36. Nmm.l mam. $.32 vac. :5. c5. :2 N2. 3:. an“. $32 mm“. >2. 3:. :2 min! mm“. 33. m2 emu. v2. mam. NS mafia—oh . mm: 0mm _= 0mm m: on = 0m — u Um 0.3.3.8.? 8N. H .9358: 535.0 =u .30? .335. .8.— E can .= .— 3:22.500 Eamon—E 3 03am. 2...: u «use... .3... u .33. darn «3...... sea: a... 3...: n «33... .38.. : u .32... an... n assesses—see. sea a a... u «.32... .3... u see... 5:. u 3...... sea... a... S... n «3...: .35.: u .32.... .3... n sages. sees... a... . 47 3...... 3.... 3.8. 85......» 4. En: 3...: 3...: 3...: e2. 82 5...: New. Sn: 8... «as. a: 3...: 8...: 3...: 3...: 3... ea: .8. as... a... nae. 2... as: S... 2...: 3...: 3...: 84.. 32 3.... 5...: SN. 8... as. «as. an... 2... 3...: we... 2.. a: s8. .2... 2... 2.... 2... 8: Sn a... e2. emu. .3. E: New. 8... e2. .3... x... :2 8... «a... 2...: 8... x... 3: 2n: 3.. Sn: 5.... 5.. :2 as... 3...: 2... 2...: «a... a: 3.. Es. ........ 8... new. as. 8.32 a... PE _= 0% E 0.. = 0.. . 0.. e225, .3888. I see... 48 In males, the first component is also a size vector. A shear of the second component (r= -O.64 (p<.05)) also appears to have removed the underlying size factor, leaving very little size-independent shape information. The sheared third component. contrasts posterior nasal length (M6) and width (M7). M. The first component of a standard principal component analysis of data from female responds to a size vector (see Table 16). The second component is a contrast of two posterior measurements (M19, 37) sharing a landmark in the M3 (K), and an anterior measurement (MSZ). The third component contrasts two posterior length measurements (M19, 36). For males, the first component also reflects static allometry. The second is a contrast of premolar row (M23) and anterior width (M32). They share the proximal PM2 as a landmark. The third component is a contrast of proximal (M31,32) and distal (M38) width. Principal components I and II were not significantly correlated in either males or females (females; r= -.15,males; r=.13 (p<.05)). Therefore, the shear procedure was not performed, as it would not add new insight to the analysis. Discussion. Standard principal component analysis performed on mandible and ventral data sets for each sex, and the lateral data from females did not yield first and second components which were significantly correlated (see Table 17). Therefore, I concluded that the first components contained size information and the second contained shape information. Correlation of the first and second components produced by standard principal component analysis of lateral data sets for males, and dorsal data for both sexes prompted the use of the shear procedure. The sheared second component of dorsal data for females contained only one loading above 0. 1, indicating that the second component from the standard principal component analysis was also a size vector. Information concerning shape differences were found in the third component of the shear procedure. In all procedures, the patterns of contrast produced were unique to each sex. However, within 49 hvda Nada mwdb 8:33.» .x. me.| ave. mam.l vac. am“. .3: awn. 8:. man. vac. aha. 1:2 voc.l awe. acql mac. mag. «:2 she. vmc. :5. 55. can. «:2 won. ewe. wcm. gnu. £3. :2 N3. coo. mvw. wmfi NON. cu: a—m. nee. sum. :4. awn. a: had. mwc. v3. mugl awn. w: mnv.l Neg. 2.4.! an?! mam. :2 wcv.l 33. max! :3. mac. 9: Ncc.l flea. Nae. emu. cvm. :2 ca. ~©o. ma. mco.l mam. a: Omo.| wbc. mvc.l nvc. mvn. :2 8?th «=— 95 a 0mm :— 8 = Um — 0m 03353, 3 036.? Cu. H 5358: .580 =a is? .38.. 5.... u: E... .= ._ 35:38.5 3.55.5 2:. n «.32: £3. n 53.x: .25... n 333. "Ewe Sn :5. n «32: .51: u .32: 68.: n 3%... "815... s... N 2.5.: n N32: 58. u $33... 52.... n 333 "8:5. as 50 mg. u «32: .38: n ~33... £8. u 333 "go—use 8.? «2% $2. 3.8 8.85.. m. an... «2. £8. 2". at. £2 2:. :8. 9:. £5... *2. :2 25. Se. 26. N3. m1. 2: m8. 5:. 8c. 9:. m3. 3: Sm. 3;. Sn. 2:. c2. :2 m2. 2:. m2. m3... 2.”. S: :2. 3;. :2. 2:. «on. 2 Se. can. Sc. m3. m2. a: 89: En. Se... :3. ant p: £8. 28. So. ”8.... an. a: 28.: 3... 2:2: 3...... 2.... :2 m8. 2:. m8. at. man. 22 >2. «3. >2. 2;. Eu. 5 332 «E 0.5 E 0.5 E on = on _ on 2.3.; @2283 2 2.3. 5] ccéc cccc vmdw macaw—5r 3 ac. wwc. v:. cm: cNN.| c:.| ~cv. cc: Ev. 35.! a2. has wwm. vac. vcm. cm: ccm.| Ncc. cmw. fin: Ncm.l 5:. van. was ac. mum. new. an: ~m~.l c3. av“. was 3%. 2:. ch. cm: umc. cwu. sum. as: cmc. ch. cflc. was 5:. Sc. 3:. ca: vcv.l wcm.l won. an: 8.55% :— OL = on ~ on czar—Er fin. H hams—Sc .58 :5 for, .955; 8.. n: can .= ._ 8:23:80 3395.5 2 wish. 52 coca cmdw cvéc macaw—5r fl. ch. N2. cmc. cm cvv. cvc.| muv. mm ch.| Nu?! mmN. ha 02. :..N. 53.. cc ccfi. vc~.l cw". vm cw—.| NVN. m3. cm 13.! «Am. :N. Na cw ..l N:.| m3. an bNu. :3. EN. cm mm—.l mac. cNN. wN NmN... wmc. can. cN vcc. N:..I cmN. «N wan. wbc. "Na. cm 83: En 0m = 0m — n on m5m€¢> @2253 2 2.3. 53 each sex differences between the OTUs were more pronounced in horizontal movement of landmarks (dorsal, ventral), than in vertical displacement (lateral, mandible). For males, contrasts within dorsal and lateral subsets both involve relative movement of two landmarks in the nasal portion of the skull-the nasal-maxilla-premaxilla suture, and the nasal—maxilla suture. Another landmark, the anterior PMZ, integrates contrasts in lateral and ventral data subsets. However, mandibular data emphasizes contrasts between ramus height and length, utilizing landmarks which do not include the PM2. For females, the patterns of contrasts between data subsets are not as well-integrated as for males. Contrasts from analyses of dorsal and lateral subsets emphasize contrasts between the nasal and the frontal areas of the skull. However, several of the same landmarks are shared between the subsets- the nasal-premaxilla-maxilla suture and the frontal-parietal suture. Lateral and ventral data sets shared landmarks in the tooth row in measures of contrast, and delineate shifts in relative placement of the tooth row. Contrasts of mandibular data supports this interpretation, as the premolar row forms a contrast to ramus height. FIL age series Wild pigs in the Philippines are distinct from the other subspecies in that they are the only group found on oceanic islands, and they have the smallest body size among the subspecies. I performed principal component analysis on a static cross-section of the age classes to explore aspects of growth. This type of data is not as useful as longitudinal data, as it does not yield information on individual variation in growth rates (Cock 1966;136—137). However, static cross-sectional data can yield information on the static growth rates of the entire population, which I thought adequate for the purposes of this study. Unfortunately, the utility of this data was decreased by other aspects of the sample. Not all age classes are equally represented (Age Class 4: N= 1, A05: N=5, AC: N6=24, AC7: N=4, A08: N= 1, A010: N=2, A011: N=3). In addition, information concerning 54 Table 17 Varience and correlations from principal component analysis of all OTUs. Females Males View PC I PC 11 Corr. Shear PC I PC II Corr. Shear Mandible 93.18 2.39 .-45 No 91.25 2.85 -.35 No Lateral 79.05 13.02 -.20 No 78.32 8.02 -.67 Yes Dorsal 79.85 9.17 -.67 Yes 66.86 12.83 -.64 Yes Ventral 83.54 7.15 -.15 No 64.46 21.84 .13 No the sex of immmature individuals was not available. No shear procedures were necessary, as the first and second components in all analyses were significantly correlated. Mandible. The second component is a contrast of ramus width (M8) and height (M6) (see Table 17). The first measurement exhibits positive allometry on the first component, the second is isometric. This suggests that the ramus grows vertically compared to horizontally. The third component is a contrast of anterior length (M2) and ramus height (M6). M: The second component is a contrast of nasal length (MS) with nasal height (M25) (see Table 19). The third component is a contrast of skull height (M22) and the portion of the tooth row which includes the canine (MZG). 2055a]; The second component is a contrast between anterior nasal length (M4) and posterior width (M10) (see Table 20). The third is a contrast between anterior (M4) and posterior (M15) length. W. In this data set the second component is a contrast between anterior length and width (M26, 32) and posterior width (M34) (see Table 21). All three share a landmark in the anterior PM4. The third component is a contrast of posterior (M36) and anterior length (M30), with anterior width (M31). 55 Table 18 Principal components I, II, III for age series of OTU FIL, mandibles (isometry=.353). Variable PC I PC II PC III M2 .485 .178 -.727 M3 .327 .249 -.243 M4 .094 .143 .135 M5 .256 -.326 .193 M6 .371 .658 .501 M7 .459 -.183 .266 M8 .482 -.538 .123 M9 .012 .144 .139 %variance 94.0 96.4 97.8 56 Table 19 Principal components I, II, III for age series of OTU FIL, lateral view Variable PC I PC II PC III M2 .392 .155 .095 M5 .360 -.897 .085 M11 .179 .207 -.225 M16 .326 .087 -.017 M17 .289 .045 -.072 M22 .244 -.022 .323 M24 .073 .033 .113 M25 .308 .277 .472 M26 .347 .045 -.737 M27 .323 .165 .176 M28 .327 .095 -.106 %variance 83.4 91.0 93.9 57 Table 20 Principal components I, II, III for age series of OTU FIL, dorsal view (isometry=.27). Variable PC I L PC 11 J PC In Dorsal (isometry=.27) M1 .326 -.021 .051 M3 .384 -.079 .083 M4 .374 -.748 -.413 M6 .380 .118 .028 M7 .205 .153 .411 M8 .258 .203 -.084 M9 .269 .246 -.228 M10 .262 .364 -.121 M11 .196 .080 -.149 M12 .238 .014 .118 M13 .170 -.086 .326 M14 .274 .201 .011 M15 .103 -.326 .656 %variance 81.1 85.7 89.5 58 Table 21 Principal components I, II, III for age series of OTU FIL, ventral view Variable PC I PC 11 PC In M23 .026 .022 .108 M26 .346 -.534 -.081 M28 .326 .010 -.214 M30 .202 -.087 -.488 M31 .427 .143 .550 M32 .426 -.475 .239 M33 .405 .245 .043 M34 .345 .612 .011 M36 .270 .135 -.567 M39 .100 -.072 -.107 %variance 87.8 91.7 93.8 59 Discussion. Principal component analysis performed on lateral data sets (lateral view, mandible) produced contrasts in the second component similar to thise of all other analyses containing male individuals. Ramus height contrasts with ramus width for all analyses. It must be noted that the second component of the ager series accounted for a small percentage of the total variance (2.4 and 7.6%). Analysis of the dorsal data produced contrasts involving the canine region and palate width. This suggests that sexual dimorphism is the underlying factor influencing the results of this analysis. Results of analysis of the age series of FIL is difficult to interpret, possibly for methodological reasons. A static-cross section data set may have had an effect on results; a longitudinal data set is preferred. Sample size may have influenced the results, as several age classes were under-represented, and the age classes were of unequal time length. Results of adult BARB males and females indicate that sexual dimorphism is present in the taxa. This may have obscured growth trajectories present in the results, as juveniles were unsexed, and both male and female adults were included in the analysis. DISCUSSION Results of these analyses indicate that shape and size differences separate the five subspecies into three groups. Within each group, males and females exhibit unique patterns of contrast as a result of sexual dimorphism. An assessment of these patterns and examination of sources for variation follow below. Size Both discriminant analysis and principal component analysis isolated three size groups. From largest to smallest they are BARBoOI, AHO, and FIL-CEBI. Both males and females could be placed in the appropriate size group with a high degree of confidence. Previous studies suggest that the body size of island dwellers may influenced by the size of the island where they live (Marshall and Corrucini 1978). Foster’s study (1964) suggests that artiodactyls on islands should become smaller through time. He compared the average body size of 1 1 artiodactyl species to average body size in ancestral mainland populations. All were either the same size (12%) or smaller (88%) than mainland representatives. This hypothesis suggests that S. bar-bolus found on islands should be smaller than mainland populations. It further suggests that, if body size is only influenced by island size, ahoenobarbus should be smaller than philippensis as Palawan is smaller than Luzon. In addition, the hypothesis indicates that there should be a reduction in size through time in the fossil record for the smaller subspecies. The results of discriminant analysis indicates that the populations on small islands are smaller than samples from either the largest islands or the mainland. However, if 60 61 island size is the only influence on body size, AHO populations should be smaller than FIL, as Palawan is smaller than Luzon. The average body size of A110 is larger than FIL, suggesting that other factors, such as length of time of genetic isolation, are also pertinent. Extensive fossil material for S. barbatus from small islands for study of size reduction through time has not been recovered. Middle Pleistocene material has been recovered from the Cagayan Valley in Luzon, Philippines, but has not been identified to species (Dr. J. Peralta, personal communication). No fossil material, to date, has been recovered from Palawan. Shape Sexual dimorphism Sexual dimorphism was readily apparent in all subspecies. Male individuals had massive upper and lower canines in comparison to females, and older males exhibited saggital cresting. One goal of this analysis was to assess the effect of this dimorphism on over-all skull dimensions. All analyses of data from a vertical plane (lateral view, mandible) resulted in only two patterns of contrast. One pattern was characteristic of data composed of females only (Figure 12,13:c,d). Data sets containing only males, or males and females exhibited another pattern (Figure 12,13:a,b,e,f). For females, analyses of mandible data produced contrasts between ramus height and premolar row. In the skull, measurements in the rear centering on the posterior M3 form contrasts with measurements centered on the anterior PM2. Mandibular contrasts consisted of anterior ramus height and width for all data sets containing measurements from males. Lateral data sets from males alone and males and females combined produced a contrast of nasal length and nasal height. Patterns of contrast in the horizontal planes of the skull, dorsal and ventral data, cannot be explained by reference to sexual dimorphism, and will be discussed below. 62 Figure 12. Patterns of contrasts from principal component analysis, mandibles. a.) BARB males vs. females (PC II) b.) BARB vs. OI males (PC 11) c.) FIL vs. BARB, females (PC II) 6.) all OTUs, female (PC II) e.) all OTUs, male (PC II) f.) FIL age series (PC II) a.) c.) e.) Figure 13. 63 Patterns of contrast from principal component analyses, lateral view BARB males vs. females (PC II) FIL vs. BARB, females (PC II) all OTUs, male (SPC II) b.) d.) f ) BARB vs OI males (PC II) all OTUs, female (PC II) FIL age series (PC II) 64 Inter-populations] variability Although size was the best indicater of differences between OTUs I conducted further analyses to explore the nature of shape differences. Shape differences between OTUs were most pronounced in the data from the horizontal planes of the skull (i.e. dorsal and ventral views) (see Figures 14, 15). The shear procedure identified three clusters of OTUs which were congruent with those identified by discriminant analysis- BARB-OI, AHO, and FIL-CEBI. Similarity in size and shape between these clusters may be function of genetic relationships. Patterns of genetic relationships between the taxa are predicated on the historical geography of insular Southeast Asia. The genus Sus probably evolved on the Asian mainland; representatives first appear in the Swialik beds of the Indian Pliocene. The genus appears next appears in Middle Pleistocene contexts in Java (Badoux, 1959). During the Middle Pleistocene sea levels were an estimated 160 to 180 meters lower than currently, exposing most of the Sunda Shelf and uniting the Malaysian Peninsula, Sumatra, Java, and Borneo into one land mass. The island chain of Palawan was a peninsula of Borneo during this time. During the Late Pleistocene it was separated from Borneo by a channel. It seems unlikely, judging from faunal distributions and geological reconstructions, that the Philippine archepelago was ever part of this land mass. Heaney (1984) compared the number of species and composition of current faunal assemblages from the Philippines to those from islands on the Sunda Shelf and concluded that the assemblages from the Philippine archipelago were depauperate in the number of species that the land masses could support. This is typical for oceanic islands which, by definition, were never part of a larger land mass. . Water gaps between the eastern edge of Borneo and Mindanao were no more than 15 miles, and could have been traversed by swimming animals. Pigs can and do swim (Wallace 1881;69), but their endurance is not known. Unless introduced by man (Heaney 65 Figure 14. Patterns of contrast from principal component analyses, dorsal view. a.) BARB males vs. females (PC I) b.) BARB vs. OI males (PC I) c.) FIL vs. BARB, females (PC II) d.) all OTUs, female (SPC III) e.) all OTUs, male (SPC III) f.) FIL age series (PC II) 66 Figure 15. Patterns of contrast from principal component analyses, ventral view. a.) BARB males vs. females (PC I) b.) BARB vs. OI males (PC I) c.) FIL vs. BARB, females (PC II) d.) all OTUs, female (PC II) e.) all OTUs, male (PC II) f.) FIL age series (PC II) 67 1985), the bearded pig probably entered the Philippine archipelago by swimming water gaps between the Sulu archepelago and Mindanao. An alternate route of entry is through the Palawan chain to Mindoro or Panay, but it involves a larger water gap and therefore is considered less likely a route. Once across this boundary, pigs could have easily spread throughout the archepelago. At this time, the Philippines consisted of three islands, Negros-Cebu-Panay-Masbate, Luzon, and Mindanao. Rising of the sea level by 40 meters during the Late Pleistocene did not appreciably change the configuration of the land mass of the Sunda Shelf, but separated Palawan from Borneo. If this reconstruction is correct, we would expect to see ahoenobarbus resemble barbatus more closely than it does cebifrons and philippensis, as Palawan was connected to Borneo and not to the Philippines. We would also expect to see the subspecies in the oceanic islands resemble each other, and the species in the continental islands resemble each other. Middle Pleistocene material assigned to Sue barbatus has been recovered from Java (Badoux, 1959) although this species is not found there presently. Late Pleistocene fossil material identified as Sus barbotus has been recovered from Niah Cave (Medway 1977, 1979). Historical reconstruction of migration routes suggest that AHO has remained in genetic contact with BARB longer than with FIL-CEBI. Therefore, shape characteristics of AHO should more closely resemble BARB than FIL-CEBI. The shear precedure, while isolating AHO as a distinct group, produced ambivalent results for assignment of closer genetic relationships, although discriminant analysis indicates that, based on size, AHO most closely related to FIL-CEBI. A larger sample of ahoenobarbus would permit a more detailed comparison with both the oceanic groups and the continental groups, and perhaps assessment of degree of morphological similarity. Principal component analysis was performed on BARB and FIL females to explore the nature of shape differences between the largest and the smallest OTU. This was the 68 only analysis of dorsal data to produce contrasts in the rest portion of the skull, and not pertaining to the palate. However, one landmark on the tooth row, the M3, was included. Analysis of ventral data contrasts posterior nasal length and width, also including landmarks on the tooth row. This indicates that FIL is not a scaled-down version of BARB, but that shape parameters also differentiate the two OTUs. This conclusion is also supported by analysis of the FIL age series.The patterns of contrast demonstrated through this analysis resemble those of males in the mandible and lateral view, but are unique in the dorsal and ventral view (see Figure 11-14:f). Results of analysis of this data set indicate that the nasal region of the skull exhibits positive allometry in relation to the back of the skull. Overall, the morphometric contrasts among age classes do not parallel either contrasts among BARB males and females, or contrasts among BARB and FIL. This suggests that the growth patterns of FIL do not parallel morphometric differences between BARB and FIL. Lack of congruence in these patterns of contrast may indicate that FIL exhibits an independent developmental tragectory. That is, FIL, the smallest OTU, is not a neotenic form of BARB, the largest. In all analysis comparing dorsal views of OTUs contrasts include posterior nasal length and width. This region contains the tooth-bearing portion of the skull. Analysis of ventral data all produced unique contrasts. Analyses of data from both males and females emphasize the placement of the tooth row in relation to the longitudinal axis of the skull. Shape differences betweeen OTUs were particularly pronounced in the regions of the skull pertaining to the tooth row. Groves (1981;51) notes that the molar row forms a smaller proportion of the maxillary tooth row in S. b. oi than in S. b. barbatus. He interprets this as an example of character displacement, as oi occurs sympatrically with wild S. scnofa. Unfortunately, principal component analysis of oi vs. bar-bums excluded the molar row, because of high correlation coefficients, so this potential discriminator was not closely examined. However, analysis of the ventral data set produced contrasts between palate length and width and premolar row (Figure 14:b). 69 The diversity of loading patterns of the ventral view of the skull, a data subset which contains measurements referring to the feeding apparatus, suggests that the influences on shape configuration are not uniform. Each population may be adapting to specific ecological parameters within their respective environments, presumably relating to food resources. Testing this would involve establishing evidence of a relationship between patterns of morphological diversity and some aspects of the environment. While I am not suggesting that there will be a close relationship between morphology and feeding ecology, these results suggest that separation between populations may be accomplished through reliance on measurements of tooth row alignment. At this time, the ecological data is not available to evaluate the influence of local environment on cranial morphology. This study established by quantitative means the existance of shape and size differences between three of five subspecies of the bearded pig. There is no evidence to indicate that the current taxonomy of the bearded pig includes any inappropriate groups. S. b. philippensis should not be given specific status, as Groves suggests (1981). However, this study indicates that several of the subspecies are not valid. Lack of separation between S. b. barbatus and S. b. oi on the basis of size or shape indicates that the two subspecies are actually geographically separate populations of the same subspecies. These should be combined under the older name of S. b. barbatus. Although my sample size was inadequate for analysis of shape, discriminant analysis of size suggests that S. b. philippensis and S. b. cebifrons should also be combined under the older name of S. b. philippensis. Size differences between the subspecies are only partially accounted for by reference to size of land mass on which they are found. The largest subspecies are found on the largest islands and the mainland, while the smallest are found on the smallest islands. However, ahoenobarbus is larger than predicted by this hypothesis. Geological reconstruction of dispersal of bearded pigs suggest that ahoenobarbus remained in genetic contact with barbatus longer than with philippensis populations. The 70 larger than expected body size may be a function of a closer genetic relationship with a larger-bodied subspecies. However, it was not possible to assign a closer genetic relationship to either barbatus or philippinenesis on the basis of shape analysis with the sample available. Results of this study indicate that ventral portions of the skull pertaining to tooth row are undergoing morphological divergence. This may be in response to specific ecological conditions within their geographical ranges. The nature of the factors influencing this diversity cannot be examined until further ecological studies are conducted. Comments on methodology The results and interpretations of this analysis are not independent of the analytical techniques employed. The limitations of the truss as a way of examining individual specimens, and the assumptions of the statistical procedures will be reflected in the results. Here I wish to discuss the shortcomings of this particular analysis. The truss network represents a distinct advantage over conventional measurement protocols. The use of homologous landmarks contributed to a clearer separation of the OTUs and facilitated biologically meaningful interpretations of the results. However, the particular truss scheme which I utilized had several deficiencies which hampered the effefctiveness of this method. Mandiblesand skulls were not measured as complementary units, and thus could not be interpreted as such. Few measurements were chosen which accomodated the articulation of skull with mandible. Measurements of the ventral skull included medial-lateral distance from the midline of the palate, while mandibular measurements were taken exclusively in a lateral plane. Alteration of the truss network. to include measurements of this nature may facilitate explanation of differences between OTUs. 71 A second deficiency is that potentially useful measurements, particularly those in the tooth row, were eliminated to render the co-variance matrix mre amenable to statistical manipulation. The truss network should be tailored to accomodate the problem examined. Principal component analysis has been used as an analytical tool in separation of hybrid fishes from parental groups (Neff and Smith 1979), separation of species of fish (Humphries et al. 1981), and separation of subspecies of elk (Schonewald-Cox et al. 1985), among other studies. These studies all indicate that this technique is sensitive to differences within data sets. However, the protocols for deciding whether these results are biologically significant are not well established. Neff and Smith discuss their assumptions: An implicit assumption frequently made is that the important biological phenomena will be represented most clearly by the components in the directions of the greatest variance, permitting generalized inferences to be drawn from the first few components, especially when they explain a very large percentage of the total variation.. The validity of these (assumptions) was not examined in this study, but instead remains an assumption when generalizations are made from principal components analysis results since only the first few components were examined in detail. If there is known or hypothesized to be more than one group in the data it is often assumed that the direction of greatest variance is approximately the same for all groups (1979;192). 1 also assume that the most important biological phenomena will be represented by the components which explain the largest proportion of the total variance. In the majority of analyses in this study, and in the studies enumerated above, the largest proportion of total variance is accounted for by the first component. The underlying biological factor is assumed to be size differences between the groups examined. Interpretation of the second component of a multi-group principal component is also problematic. If the first component generally represents size, and accounts for the majority of the variance, how much variance must the second component account for to be biologically meaningful? The total amount of variance accounted for by PC I and PC 11 in the study by Humphries et al. was not recorded; Neff and Smith recorded 41% for PC I and 29% for P0 11 for the principal component analysis of Lepomis species and 96% for PC I and 3% for P0 11 for principal component analysis of Notmpis species. Schonewald-Cox 72 et al. record percentages of explained variance which range from 10—70% for components I and II. All studies suggest that these were biologically significant differences. In this study, the majority of analyses produced a first principal component which accounted for a high percentage of the total variance, while the second was gnerally low, accounting for less than 10% of the variance. The shear procedure acted on this low variance and, when plotted, separated the OTUs into non-overlapping groups. Whether these shape discriminators are significant in a biological sense is a matter of interpretation. An additional problem in interpretation of multi-group principal component analysis is assigning significance to contrasting loadings on the second component. Neff and Smith (1979;192) note that the results can be patterned in an extreme manner by the presence of a single meristic character. Humphries et al. employed principal component analysis in a comfirmatory way rather than an exploratory way, and were able to assign species to positions of positive or negative contributions to loadings on the basis of prior biological knowledge. Schonewald-Cox et al. referred to univariate statistics to interpret loadings on the second component of analysis of four OTU s of elk. In this study the problem was addressed by performing analysis of pairs of OTUs and comparing the resulting patterns, but the problem of assigning relative positions still remains. CONCLUSION The influence of environment and genetic background on skeletal morphology is a significant problem in evolutionary biology. This study has examined cranial variability among 5 closely related populations of one species of pig. This taxa is distributed both on oceanic and continental islands, and on the Southeast Asian mainland. The purpose of this study is to identify sources of morphological variation through examination of cranial skeletal morphology in these populations. To identify sources of variation I examined differences between the sexes, among all the subspecies combined, and between subsets of the total sample, separated by sex. I also examined an age series of S. b. philippensis in an effort to characterize growth patterns in one subspecies. Examination of male and female S. b. barbatus revealed that differences in cranial morphology were not isolated to areas containing the dimorphic canine. Patterns of contrast isolated general length and height dimensions, rather than local regions. Results of discriminant analysis, a precedure sensitive to size differences between populations, combined the five subspecies into three groups. These were barbatus-oi, both from land masses on the Sunda Shelf, ahoenobarbus from Palawan, and philippensis- cebifrons, both from the Philippine islands. Results of principal component analysis confirmed these groupings, and supplied additional information on the nature of the shape differences between groups. My findings indicate that head size of bearded pigs is largest on land masses on the Sunda Shelf, and smallest on the oceanic Philippines islands. They are also the smallest islands supporting populations of bearded pigs. Shape differences group the samples in an 73 74 congruent manner. Hypothese drawn from historical and island biogeography are presented but are not conclusively tested. The small sample size of ahoenobarbus do not allow for fine discrimination between hypotheses. However, the findings do suggest that, on the basis of size, ahoenobarbus is genetically closer to barbatus than philippensis, and should resemble this group more closely in shape as well. Fossil specimens are also necessary to explore the implications of these hypotheses. Analysis of cranial morphology of other endemic suid species in island Southeast Asia may reveal shed further light on this bio-geographical problem. Principal component analysis performed on the data subsets indicate that significant patterns of contrast were consistantly isolated in the ventral portion of the skull. All contrasts emphasized some aspect of the tooth row to the palate midline. The variation in results suggests that divergence is occurring in the masticatory morphology of these subspecies. The source for this variation may lie in habitat characteristics specific to each subspecies. Baseline ecological studies are necessary to test this proposition. This study established by quantitative means the existance of shape and size differences between three of the five subspecies of the bearded pig. These groups are S. b. barbatus-S. b. oi, S. b. philippensis-S. b. cebifmns and S. b. ahoenobarbus. There is no evidence to suggest that any taxa is mis-identified or should be given separate status. This study also suggests that several of the subospecific classifications are not valid. Lack of separation between S. b. arbatus and S.b. oi on the basis of size or shape indicates that the two subspecies are actually geographically separate populations of the same subspecies. These should be subsumed under the older name of S. b. barbatus. Although my sample size was inadequate for a shape analysis, analysis of size suggests that S. b. philippensis and S. b. cebifrons should also be combined under one name, S. b. philippensis. This study has demonstrated that the basic stock of bearded pigs have diverged morphologically through time. Change in shape and size as well as, possibly local extinction, has occurred concomitantly with geographic isolation. Understanding the 75 factors which have influenced this divergence and the dynamics of this process can increase our understanding of general evolutionary processes. BIBLIOGRAPHY BIBLIOGRAPHY Badoux, D. M. 1959. Fossil mammals from two fissure deposits at Punang (Java), with some remarks on migration and evolution of mammals during the Quaternary in Southeast Asia. Utrecht: Proefschr. Cock, A.G. 1966. Genetical Aspects of Metrical Growth. Quarterly Review 9_f_’ Biology Vol. 41:131-190. Cranbrook, E. 1979. A review of domestic pig remains from archaeological sites in Sarawak. Sarawak Museum Journal. Vol. 27. Dixon, W.J. et al. 1985. PM Statistical Software. University of California Press, Berkeley Foster, J.B. 1964. Evolution of mammals on islands. ISM Vol. 202:234—235. Fox, D.J. and KB. Guire 1976. Documentation for; MIDAS,Third Edition Statistical Research Laboratory, University of Michigan. Groves, C. 1981. Ancestors for the Pigs: Taxonomy and Phylogeny of the Genus §E§- Technical Bulletin No.3. Australian National University. Heaney, L.H. 1984. Mammalian species richness on islands on the Sunda Shelf, Southeast Asia. Oecologia Vol.61:11-l7 1985. Zoogeographical evidence for Middle and Late Pleistocene land bridges to the Philippine Islands. Modern Quaternary Research in Southeast Asia Vol.9:127- 144 "'- 76 77 J olicoeur,P. 1963. The multivariate generalization of the allometry equation. Biometrics, Vol. 19:497—499. Humphries, J.M., F.L. Bookstein, B. Chernoff, G.R. Smith, RE. Elder, and S.G. Poss. 1981. Multivariate Discrimination by shape in relation to size. Systematic Zoology, Vol. 30:291-308. Marshall, L.G. and RS. Corruccini 1978. Variability, evolutionary rates, and allometry in dwarfing lineages. Paleobiology, Vol.4: 10 1-1 18. Matsche, G.H. 1967. Aging EurOpean hogs by dentition. Journal 9_f Wildlife Management, Vol.31:109—113. Medway, L. 1977. Wild pig remains from the mouth of Niah Cave. Sarawak Museum Journal Vol. 25:21-33. Neff, N.A. and G. R. Smith 1979. Multivariate analysis of hybrid fishes. Systematic Zoolgy, Vol. 28:176—196. Nie, N., 0. Hull, J. Jenkins, K. Steinbrenner, D. Bent 1975. Statistical Packggg for _tlg Social Sciences. Second edition. McGraw-Hill Book 00., New York. Sanborn, 0.0. 1952. Philippine Zoological Expedition. Fieldiana: Zoology Vol.33:89-158. Schonewald-Cox, 0., J. Bayliss, and J. Schonewald 1985. Cranial morphology of Pacific coast elk (Cervus elaphus)) Journal of Mammalogy Vol. 66:63-74. Strauss, RE. and F.L. Bookstein 1982. The truss: Body form reconstructions in morphometrics. Systematic Zoology, Vol.31:113—135. Wallace, A. 1881. Island I_._if_e. Harper and Brothers, New York. APPENDIXES APPENDIX A List of specimens examined The collecting localities are grouped below according to subspecies. Specific localities are grouped by country within land mass. Acronyms are: FNMH, Field Museum of Natural History, Chicago, Illinois; NMP, National Museum of the Philippines, Manila; ZRC, Zoological Research Collections, National University, Singapore; JM, J abatan Muzium, Kuching, Sarawak; MS, Muzium Sabah, Kota Kinabalu, Sabah; MZB, Muzeum Zoologici Bogoriense, Bogor, Indonesia; BM, British Museum, London England; CZB, Cambridge Zoology Museum, Cambridge, England, USNM, U.S. Natural History Museum, Washington D.C.; UMMZ, University of Michigan, Museum of Zoology, Ann Arbor, Michigan; BG, personal collecation of Dr. Bion Griffin (these specimens will be donated to UMMZ). Sus barbatus ahoenobarbus PHILIPPINES Culion E: skull, 1 male (USNM 152244). Palawan _I_s_.: skull and mandible, 1 male (BM 94688); mandibles, 2 males (NMP, non catalogued, but inscribed with locality and date of capture). Iwahig: skull and mandible, 1 male, 3 females (NMP, none catalogued, but inscribed with locality and date of capture). Binuan: skull and mandible, 1 male (FMNH 62830). Lapulapu: mandible, 1 male (FMNH 62825). Balabec .111: skull and mandible, 3 females (BM 94687,94689,946810). Susbarbatusbarbatus MALAYSIA Sabah: Sandakan District: skull and mandible, 1 male (FMNH 68756). Gomantong Forest. Reserve: skull and mandible, 1 male (FMNH 468758). 8 miles west of Sandakan: mandible, 2 females (FMNH 33553-4). Kinabatongan District: little Kretam R, skull, 1 male 68757); mandible, 1 male (FMNH 68755). 78 79 Tawau District, Kalabakan, Sungei Tibas Camp: mandible, 1 female (FMNH 85916). INDONESIA ' Kalimantan: Pasir R: skull, 1 male (USNM 154376). Pamukang Bay: skull and mandible, 2 females (USNM 151851,154380). Mahakam R, Longiram: skull and mandible, 1 female (USNM 176197). Kalei R, Toembit: skull and mandible, 1 female (USNM 196832). Segah R, south bank: skull, 1 male (USNM 196834). Sandaren Baagoe: skull and mandible, 1 female (USNM 197667). Sungai Karangan: skull and mandible, 1 female (USNM 198302). Sungai Djambajan: skull, 1 male (USNM 198850). Labuan Pendjang: skull, 1 male (USNM 197669). Sempang R: skull and mandible, 2 females (USNM 145293,151845); skull, 1 male (USNM 145292). Sempang Kampong: skull, 1 male (USNM 145299). Klumpang Bay: skull, 1 female (USNM 151843). Landak R: skull, 2 males, 1 female (USNM 142350, 142351,142354). Pulo Pelapis: skull, 2 males (USNM 145288-89). Pulo Panebangan: skull, 2 males (USNM 145290—91). Sejok: skull, 1 male (USNM 145298). Matan R: skull, 1 male (USNM 145297). Semandang R: skull, 1 male (USNM 145295). Tjangtung: skull and mandible, 1 female (USNM 151846). Sempanahan R: skull and mandible, 1 female (USNM 151852). Pangkallahan R: skull, 1 male (USNM 151849). Balikpapan Bay: skull and mandible, 1 female (USNM 154377). Pulo Bauwal: skull and mandible, 1 female (USNM 153785). Samarinda: skull and mandible, 1 female (MZB 8366), skull, 1 male (MZB 8367). Buntok: skull and mandible, 1 male (BM 1045158). Pontianak: skull and mandible, 1 male (ZRC4.1940) E; Kalimantan: skull and mandible, 1 male (MZB 8381). BRUNEI (no other locality information) skull and mandible, 1 male (ZR04.1939). MALAYSIA Sarawak: (no other locality information) skull and mandible, 10 males, 5 females (JM 2/2,5,8—11,13,16,17,19; BM 97621, 033011,033015—16,976251); skull, 1 female (JM 2/7); mandible, 2 males, 2 females (BM 97622; JM 2/4,14,18). Mt. Dulit, 4000’el.: skull and mandible, 1 male (ZR04.1938). Borneo (no other locality data): skull and mandible, 7 males, 1 female, 80 1 sex unknown (BM 951148; H.12.631-33,635-36,638; ZRC4.1965; MS, uncatalogued); skull, 4 males, 1 female (CZB H.12.634, BM uncatalogued (Medway 83); USNM 196840; MS, uncatalogued); mandible, 4 females (MS, uncatalogued). Sus barbatus cebifrons PHILH’PINES Negros .12: Inubungan: Santa Catalina: skull and mandible, 1 female (FMNH 66322); mandible, female (FMNH 68002). Arnio: skull and mandible, 1 male (FMNH 65454). Kaigangan: skull and mandible, 1 female (FMNH 65455). Negos Oriental: Lake Balinsasayao: mandible, 3 males, 3 females (UMMZ 130420,158002,3,5,158626,158851). Susbarbatusoi MALAYSIA Malaysian Peninsula: skull and mandible, 1 male, 1 female (BM uncatalogued, (HRC 368,375»; skull, 2 males (BM uncatalogued (HCR 367,374». Perak: Ulu Bernam Estate: skull and mandible, 1 male, 1 female (ZRC4.1930-31). Paheng: skull and mandible, 1 male (ZRC4.1932). Pekan: skull and mandible, 1 female (BM uncatalogued, has ‘pekan’ on skull). R_h_ig_ Archeglagg: skull and mandible, 1 male (BM 941504). INDONESIA Sumatra: Palembang: Bajung Lencin: skull and mandible, 1 female (MZB 1713). S. Sago R: skull and mandible, 4 males, 1 female (MZB 8386—89,8392). Medan: Kota Pinang: skull and mandible, 5 males (ZRC4.1925—28,1937); skull, 1 female (ZRC4.1929). Rengat Indragiri: skull and mandible, 1 male (BM 32371). Pulo Tebing: Tinggi: skull and mandible, 2 males, 1 female (USNM 144308,144310—11). mouth of Kempar R: skull and mandible, 1 male (USNM 144304). Pulo Rangsamzskull and mandible, 1 female (USNM 144355). Sus barbatus philippensis PHILIPPINES Luzon I_§_.: Cagayan Prov.: Blabeg Cr.: skull, 1 female (UMMZ 157896). Cagayan Prov.: Bagio Stream Valley: skull and mandible, 7 males, 15 females, 13 sex unknown (BG 1,4,6,10,13, 15,18,21,22,24,26,27,36,63-65,68,72,74,75,78,81—83, 86-89,92,94,98,100,102; UMMZ 157907); skull, 4 males, 11 females, 5 sex unknown (BG 8,11,14,17,23,25,31,33, 67.69,71,76,77,91,93,97,107; UMMZ 157909,157921-22); 81 mandibles, 1 male, 5 females, 11 sex unknown (BG 3,5, 12,16,19,66,84,85,90,95,96,99,104,109,113,117). Cagayan Prov.: Ilang R. Valley: skull and mandible, 1 female (UMMZ 157956); skull, 1 male, 1 female (UMMZ 157958,157961); mandible, 1 male (UMMZ 157957). Isabella Prov.: Blos R Valley: skull and mandible, 3 females (UMMZ 157923,158000—1); skull, 1 female (UMMZ 157965), mandible, 1 male (UMMZ 157966). Isabella Prov.: Divilakan R Valley: skull and mandible, 1 male, 4 females, 1 sex unknown (UMMZ 157894—6, 157903—5); mandible, 1 male, 10 unknown sex (UMMZ 157893.157897-907). Isabella Prov.:Dimansalansan Pt.: skull and mandible, 1 male (UMMZ 157697). Catanduanis Ii: skull, 1 female (NMP, uncatalogued, (inscribed with date and location of capture)). Mindanao: Cotabato: Upi: Becrunghat: skull and mandible, 1 male (NMP, uncatalogued (FMNH 56479)). Parang Bugason: skull and mandible, 1 female (NMP, uncatalogued (FMNH 56473)). Buayan: skull and mandible, 1 female, (NMP, uncatalogued (FMNH 56471)). Pikit: mandible, 1 male (NMP uncatalogued (FMNH 56478)). Davao: Malita: Lacaron: skull and mandible, 1 female (NMP uncatalogued (FMNH 62064)). Appendix B A verbal description of the landmarks used. The points that were used in this analysis are presented below (see also Fig. X and X). Toward the nose is designated as anterior, toward the tail is designated posterior, regardless of orientation to foramen magnum. A.) Anterior tip of nasals. B.) Anterior edge of nasal-premaxilla suture. C.) N asal-premaxilla-maxilla suture point. D.) Nasal-frontal suture at midline. This point was easy to identify in all but the oldest males. E.) Anterior edge of lacrimal-maxilla suture. This suture was partly fused in a significant portion of the sample, making it one of the less reliable landmarks. F.) Midline frontal parietal suture. This point was often difficult to identify in large males, where heavy muscle attachments had deformed the parietals. G.) Frontal parietal suture, where it is dissected by the parietal crest. This was the only landmark readily discernable in this region of the skull, but I suspect it to be variable. H.) The most lateral expansion of the nuchal crest of the occipital. This is clearly an analagous point. I.) Dorsal end of the occipital at midline. The point in question could be considered analgous rather than homologous, but some reference point on this part of the skull was necessary. J.) External acoustic meatus. 82 83 K.) Distal end of third molars. I took tooth measurements at the alveolus. L.) Distal end of PM4. M.) Anterior end of PM2. I divided the tooth row into two units to make measurement compatible with juvenile tooth eruption patterns. N.) Premaxilla-maxilla suture immediately anterior to the canine. O.) Anterior premaxilla-premaxilla suture. This was not strictly a suture, as the two bones were sometimes separate. I took measurements from the midline. P.) Premaxilla-maxilla suture, at midline. Q.) Maxilla-palatine suture at midline, ventral surface. R) Posterior edge of palatine bone, at midline. S.) Anterior end of mandible at symphasis. This is a homologous landmark, but a little impractical, as the presence of the incisors sometimes hindered accuracy. Whenever possible, I removed the teeth in question before taking the measurement. T.) Anterior PM2. U.) Mental prominance of the mandible. V.) Distal end of PM4. W.) Mandibular foramen, ventral-anterior margin. This landmark was rather unsatisfactory, as it was possible to introduce inaccuracy when transferring the location of this point from the lingual to the buccal side of the mandible. However, I judged it to be the most suitable landmark in a rather uniform area of the mandible. X.) Anterior edge of the condyle. I made an effort to take measurements from the edge of the condyle pad. APPENDIX C A verbal description of the measured distances. A verbal description of the landmarks used for each of the distances measured is presented here. Cranium 1.) Anterior tip of nasals (A) to nasal-premaxilla suture (B). 2.) Nasal-premaxilla suture (B) to nasal-premaxilla-maxilla suture (0). 3.) Anterior tip of nasals (A) to nasal-premaxilla-maxilla suture (C). 4.) Anterior tip of nasals (A) to nasal-frontal suture(D). 5.) Nasal-premaxilla-maxilla suture point (0) to lacrirnal-maxilla suture (E). 6.) N asal-premaxilla-maxilla suture (0) to nasal-frontal suture (D). 7.) Nasal-frontal suture (D) to lacrimal-maxilla suture (E). 8.) Nasal-frontal suture (D) to frontal-parietal suture (F). 9.) Lacrimal-maxilla suture (E) to frontal-parietal suture (F). 10.) Frontal-parietal suture (F) to frontal-parietal suture dissected by parietal crest (G). 1 1.) Lacrimal-maxilla suture (E) to frontal-parietal suture dissected by parietal crest (G). 12.) Frontal-parietal suture dissected by parietal crest (G) to lateral nuchal crest (H). 13.) Frontal-parietal suture (F) to lateral nuchal crest (H). 14.) Lateral nuchal crest (H) to dorsal end of occipital at midline (I). 15.) Frontal parietal suture (F) to dorsal end of occipital at midline (I). 16.) Lateral nuchal crest (H) to external acoustic meatus (J). 84 85 17.) Frontal-parietal suture dissected by parietal crest (G) to external acoustic meatus (J). 18.) Frontal-parietal suture dissected by parietal crest (G) to distal M3 (K). 19.) External acoustic meatus (J) to distal M3 (K). 20.) Distal M3 (K) to distal PM4 (L). 21.) Lacrimal-maxilla suture (E) to distal M3 (K). 22.) Lacrimal-maxilla suture (E) to PM“ (L). 23.) Distal PM4 (L) to anterior PM2 (M). 24.) Nasal-premaxilla-maxilla suture (0) to distal PM4 (L). 25.) Nasal-premaxilla-maxilla suture (C) to anterior PM2 (M). 26.) Anterior PM2 (M) to premaxilla-maxilla suture at 01 (N). 27.) Nasal-premaxlla-maxilla suture (C) to premaixilla-maxilla suture at 01 (N). 28.) Premaxilla-maxilla suture at 01 (N) to anterior premaxilla- premaxilla suture (O). 29.) Nasal-premaxilla suture (B) to anterior premaxilla-premaxilla suture (O). 30.) Anterior premaxilla-premaxilla suture (O) to premaxilla-maxilla suture at midline (P). 31.) Premaxilla-maxilla suture at 01 (N) to premaxilla-maxilla suture at midline (P). 32.) Premaxilla-maxilla suture at midline (P) to PM2 (M). 33.) Premaxilla-maxilla suture at midline (P) to maxilla-palatine suture (Q). 34.) Anterior PM2 (M) to maxilla-palatine suture (Q). 35.) Posterior PM“ (L) to maxilla-palatine suture (Q). 36.) Maxilla-palatine suture (Q) to anterior edge of palatine (R). 86 37.) Distal M3 (K) to maxilla-palatine suture (Q). 38.) Distal M3 (K) to anterior edge of palatine (R). 39.) Distal edge of palatine (R) to external auditory meatus (J). Mandible 1.) Anterior end of mandible (S) to PM2 (T). 2.) Anterior end of mandible (S) to mental prominance (U). 3.) Anterior PM2 (T) to mental prominance (U). 4.) Mental prominance (U) to PM4 (V). 5.) Mental prominance (U) to mandibular foramen (W). 6.) Mandibular foramen (W) to anterior condyle (X). 7.) Distal PM 4 (V) to anterior condyle (X). 8.) Distal PM 4 (V) to mandibular foramen (W). 9.) Anterior PM2 (T) to distal PM 4 (V).