OCCASIONAL PAPERS THE MUSEUM TEXAS TECH UNIVERSITY NUMBER 29 25 APRIL 1975 A NEW SUBSPECIES OF GEOMYS BURSARIUS (MAMMALIA: GEOMYIDAE) FROM TEXAS AND NEW MEXICO Robert J. Baker and Hugh H. Genoways As part of a study of the systematics and ecology of pocket gophers occurring on the high plains of Texas and eastern New Mexico, numerous populations of the plains pocket gopher, Geomys bursarius, were examined karyotypically. Four chromosomal races were de¬ scribed from this area by Baker et al. (1973). Additional studies lead us to believe that two of these races represent an undescribed sub¬ species of the plains pocket gopher. In addition to karyological evi¬ dence, specimens of this subspecies are morphologically distinct from those of all contiguous populations of Geomys bursarius major, the race to which they previously were assigned. How a widespread sub¬ species of pocket gopher could have gone undetected until now is not easily explainable. It is noteworthy, however, that Bailey (1905) did assign the first known specimen of this subspecies to Geomys arenarius, which the new subspecies does resemble superficially. Geomys bursarius knoxjonesi, new subspecies Holotype .—Adult female, skin, skull, and body skeleton, no. 19872, The Museum, Texas Tech University (TTU); from 4.1 mi. N, 5.1 mi. E Kermit, Winkler Co., Texas; obtained on 27 January 1974 by Stephen L. Williams; original no. 1303; karyotype no. TK 5074. Distribution. —Presently known from southern Cochran, Yoakum, Terry, Gaines, northwestern Martin, Andrews, Winkler, and Ward counties in western Texas, and Chavez, Eddy, and Lea counties in southeastern New Mexico (Fig. 1). This subspecies generally is re¬ stricted to deep, sandy soils of aeolian origin within this region. 2 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Fig. 1 . — Map of West Texas and eastern New Mexico showing the geographic distribution of Geomys bursarius knoxjonesi (closed circles) and adjacent samples of Geomys bursarius major (open circles) used in this study. Description .—Size small, both externally and cranially (Table 1), particularly evident in measurements of cranial length (Fig. 2); length of tail proportionally long as compared with the length of head and body. Coloration pale; upper parts buffy-brown, paler on sides and venter; some areas on venter covered with almost pure white hair; feet white. Karyotypic features .—The diploid number is 70 (Fig. 3) and the fundamental number (FN, number of arms of autosomal complement) BAKER AND GENOWAYS—NEW SUBSPECIES OF GEOMYS 3 Fig. 2.—Dorsal, ventral, and lateral views of the cranium of the adult female holotype, TTU 19872, of Geomys bursarius knoxjonesi. is 68 in Texas populations and 70 in New Mexico samples. The X chromosome is the largest element. The Y is believed to be a medium or small-sized acrocentric. New Mexico samples have a small pair of biarmed elements, whereas karyotypes from individuals from Texas are composed entirely of acrocentrics. The three smallest pairs of ele¬ ments have secondary constrictions. Texas populations consist of chromosomal race A and the New Mexico population represent chromosomal race B of Baker et al. (1973). A variant karyotype (2 N = 69, FN= 68) was described by Baker et al. (1973) for a speci¬ men assigned to G, b. knoxjonesi. Measurements .—Measurements of three samples of G. b. knox¬ jonesi are given in Table 1. External and cranial measurements (in millimeters) of the holotype (TTU 19872) are as follows: total length, 238; length of tail, 83; length of hind foot, 30; length of ear, 6; great¬ est length of skull, 40.1; condylobasal length, 38.5; zygomatic breadth, 24.7; least interorbital breadth, 5.4; mastoid breadth, 23.3; length of 4 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY 3 cfl u, 3 X (fl E o CS o CM q 00 OS O 00 o d os" ici 00 o sd 00 sd +1 +1 ics so TT CM -H q °°. -rt c~ re 00 o d os d sd o d d d +1 +1 re ICS Cl q c- 00 oo c- c- Os ics d d M 3 d d o ci d d +1 +1 Tt 00 o re o ICS d q -rt 00 re ic O re M 3 d CM o d d ci +1 CM CM CM +1 Cl Cl C- SO o o q TT OS o 'Ci Cl — M 3 o •ci SO d ICS o ici sd d +1 +1 ICS CM O ics 00 CO — OS Cl q 00 00 d o d OS d d o ci •ci re CM +1 CM CM CM +1 CM Cl 00 o so ei- m — q Os re OS CM ci o Os d d 00 o sd d re TT +1 cl re +1 re T c- 00 os o c- —■ CM o o q d o o sd ce o o 00 re ci ef +1 ■'t ■t- +1 re ef 00 , =t CM q o o q q ef o o q o 00 ics d oo" o ICS d sd re +1 d re CM +! CM re 00 os 00 er o o SO so q o oi CM d d OS o i—* ICi 00 +1 d o 00 +1 so ics c- rc o o q q CM O ics SO sd ci 00 ci d Os ICS CM +1 o CM 00 Cl re CM +1 O Cl cm X « tii jS « nj 2 ^ 2 II 1 S 2 2 Sg 2-22 10.2 Andrews, southern Cochran, Gaines, and Terry counties, Texas BAKER AND GENOWAYS—NEW SUBSPECIES OF GEOMYS 5 OS tn "T r ~~. NO o r~- d +1 PT — Os o PT oo o C Os It +1 Os r~- m SO (N Os oo o C o +1 (N Os r-i rr t SO so o t' 00 +1 q so rr r- q in O d r-’ -T (N +1 (N ri (N m O OS m q SO o •n SO ri +1 so o o n oo r- Os rl +1 SO r-i C rr o SO q OS r~~ oo rr o d in d 't +1 00 m q o os r- ^t o ri sd d ’a- +1 ■'3- T r- 00 on q o (N Os o 00 ’3" so rl +1 rl pr o Os q o o >n P'S '3- SO oo +1 r~ OS SO rl r- o o r-l m d r-’ «n n +1 rr (N oo r-i q OS o o q o so q - 3 ; cn q o — o ’ 3 " ri o so rr o - 3 - +1 +1 '~ l +1 oo Os q o q q ri so Tf o oo o SO oo q t~~i d +1 r-i 00 "3- oo o +1 SO Os co 00 o +1 r- oo , 3 : o P'S ’3- oo q fP) >n so 04 q CO q Os o r-i SO so o +1 d ■3- d ri o +1 Os' (N r-i c +1 SO Os' SO n so P'S 00 00 r- r4 SO SO OS (N o PO ri o so oo ri rn o SO r4 +1 (N r-4 (N +1 (N r4 (N +1 (N r4 oo O in q o SO so — cn m O Os 00 SO in d »n so* un t/r: SO* o SO* so o iri so ppi +1 X +1 +1 h r- OS OS >n s n P'S q tN 00 o ppl q n q ppi o ri »n (N c Os o C in d ri 00 >n (N +1 (N r4 3 O (N +1 (N (N +1 (N r4 o O ^3" '3- SO un so o oo OS L. so pr '-J q q q r4 r-i d SO OS so 0 so o un 00 ri o o r^i ri r^i +1 ppl pr 00 ■s}- +1 rr '3- +1 ’*3* c •— 1 in "o 00 ri SO q n u o ppi ON so pp| ■3- PPI r- O o ON o C ri oo’ d wn Os ri ri d oo NO rr r^i +1 P'S ■3- "3- +1 ■3- ■3- +1 m so pr rr o o ■3- 00 r- o o ^ 00 m o O rr SO d pr oi SO r^' d o oo od oi d SO C oo r4 +1 (N ri m +1 pp) rr ri +1 r-4 rr O 00 O o o o OS on q o O q ri (N O o m, pr ri C o o rr o o +1 in oo 00 +1 r~- OS r- +1 so 00 o 00 '3' '3' o o o q oo o q o m rr q o o q Os ri pr ppi vn in ppi ri ri '3- pr rr pp +1 O pr 4-1 pr rr j_i m, r4 ri (N (N Tl ri n (N Tl (N ri £ S so E S 2 C |e Is i2w.ES> ^Sw.ES> -3 2 2 S 2 u ES2SSU s £ - c El ^ c £ S £ 2 w -5 i > ^ 2 w .§ i > s £ Table 1.— Continued. 6 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY c 3 o a "a! > . N- — t^- OS oo K O « h DO (N N oo od O ad os mi +1 r- Os q — in os O O oo m’ r' cm -j-| — cm c- f' o Tf (jlJ oo Oi © O — in ‘ os cm -)-| m m N- © oo Os sd © sd so it +1 o cm cm in os r- ms d iT mi rn ” +1- in cm — r-~ —- rn oo o K oi d +1 o ms if — O it r- o so os so ■ m N; CM u oi in cm Os o rn sd it 3 in sd m O CM N" -a c CM +1 cm m N +1 rM c i O a cm (N rM +i CM CM 0 -o C cm it VO o r- m 00 00 T3 UJ m m 00 oo fM m i-J 0 a c 00 it +1 cm it r-* CM it d +1 <3\ m it NT it N U > 03 ci it N- od m d +1 oo m ai m c3 SI y cm o SO o o r- o SO it O o o O o O in O O Q u c © as rn it 00 Csl o vC IN o m it o os od O so Os m +1 cm m +1 m rM CM +1 fM CM q o o o q o o o i> o o m OS q o O m N- oo N- r-' CM od r-‘ in as ri CM Os od +1 so Os r- +1 t-- oo oo oo r- +1 C- 00 so CM o o 00 ms o o —< o o m N; O o Os oo N- r-’ rn CM in rd m in N- o ^_h‘ in +1 m oo m + 1 CM N- Tf CM CM +1 N- CM (N CM rM CM CM CM CM rM CM CM sO SO o r- in ms r- r-~ a> U. sO Crosby, Dickens, Garza, and Lubbock counties, Texas BAKER AND GENOWAYS—NEW SUBSPECIES OF GEOMYS r- Tt o O p ON o (N p (N r~ o (N o m (SI 0\ rJ o o o6 m oo d d d d d d o o d n p —; oo r-~ o d o o d o r^i •O Tf d o d d d d +1 *** +1 ym ~ t +1 -H — n •n NO ON r- NO NO fsj O m r- in fM (Nl (N 00 ,_, d o d d d d o rn •n c i d o d oo — r^i (SI +1 (SI (N (SI +1 CM (N fM +1 (SI (N fM NO 00 O (N O NO 't oo o p m p OO p O d o d d r^i d o «n d d d O in vd oo d in +1 £ m ON p +1 ^r fM NO oo ■t m as n oo 00 +i NO ON r- oo O o p o o O 00 't O O 00 in r- o o oo o d d m o d d d d d d d d d d *s{ d NO -U it ON m +1 r- NO II «n 00 m (N Tl (N (SI fM fM Tl fM Sag* cd 13 2 S 8 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY AA A* ftrt Aft *n *0 40 on aa in C c 04 no < « ii lA 4A Aa ** *4% 4 * ** A A ** ft * •* A* Fig. 3.— Karyotype of the adult female holotype, TTU 19872, of Geomys bursarius knoxjonesi. nasals, 14.0; length of rostrum, 16.5; length of maxillary toothrow, 8.0; palatofrontal depth, 14.7. Comparisons .—Populations of Geomys bursarius knoxjonesi are in contact only with populations of G. b. major and, therefore, need ex¬ tensive comparison only with this taxon. Individuals of Geomys bursarius knoxjonesi are significantly smaller in size than those of G. b. major in several cranial measurements. G. b. knoxjonesi averages smaller than major in most other characteristics (see Table 1, Figs. 4-5, and discussion below) and has a proportionally longer tail. In coloration, knoxjonesi is noticeably paler than major , being a buffy brown rather than a darker (more chocolate) brown on the upper parts. It is of interest to note that Bailey (1905:130) reported the first specimen of knoxjonesi from near Monahans, Texas, as Geomys arenarius. These two taxa do resemble each other in external coloration. The karyotype of the Texas populations of G. b. knoxjonesi is dis¬ tinguished from that of adjacent populations of G. b. major by com¬ paring fundamental numbers (70 or 72 in major , as opposed to 68 in Texas populations of knoxjonesi). New Mexican populations of knoxjonesi have a fundamental number of 70, their karyotype having BAKER AND GENOWAYS—NEW SUBSPECIES OF GEOMYS 9 a pair of small biarmed elements. No pair of small biarmed elements has been found in the karyotype of any population of G. b. major having a karyotype with a fundamental number of 70. The relationship and distinction of the four karyotypic races (A and B in knoxjonesi and C and D in major ) found in Geomys bursarius in western Texas and adjacent New Mexico is complicated by poly¬ morphisms, and these were discussed in detail by Baker et al. (1973). Their paper should be consulted for additional information. Another subspecies that approaches knoxjonesi in the northeastern part of its geographic range is G. b. jugossicularis. Morphologically, samples of knoxjonesi differ from those of jugossicularis in many of the same characteristics in which they differ from major. G. b. knox¬ jonesi is smaller in size and has a proportionally longer tail. Based on coloration, samples of knoxjonesi are not separable from our sample of jugossicularis from Kansas. The karyotype of G. b. jugossicularis was reported by Hart (1971) to have a 2N = 12 and FN = 72, identical to that recorded for some populations of G. b. major that we have examined, but the diploid and fundamental values are greater by two than any recorded for G. b. knoxjonesi. As will be seen in the discussion below, the subspecies of Geomys bursarius that are most closely related to G. b. knoxjonesi are G. b. llanensis and G. b. texensis. These two subspecies are geographically separated from knoxjonesi by intervening populations of major. The main differences among these taxa are the generally narrower skulls of texensis and llanensis , particularly evident in interorbital breadth (5.7 and 5.7, respectively, for females and 5.7 and 5.8 for males), and the proportionally shorter tails of texensis and llanensis (40.4 and 38.8 per cent of head and body length, respectively, for females and 38.7 and 36.8 per cent for males). The karyotype of knoxjonesi is indistinguishable from that of texensis and llanensis. Remarks .—Both univariate and multivariate statistical analyses were used to study the relationships among populations of Geomys bursarius in Texas and adjacent regions. Samples used in the uni¬ variate analyses include three populations of G. b. knoxjonesi, three of G. b. major from near, or adjacent to, the geographical range of knoxjonesi, and one of G. b. jugossicularis (Table 1). Males and fe¬ males were treated separately because of the high degree of secondary sexual dimorphism in this species. For the univariate analyses, single classification analysis of the variance (ANOVA) and sums of squares 10 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY simultaneous testing procedure (SS-STP) were used in a manner similar to that employed by Genoways (1973). Skull measurements were used as defined by Russell (1968) and Genoways (1973). All comparisons were made using adults (as in¬ dicated by the completed ossification of the basisphenoid and basioc- cipital bones). The univariate analyses revealed that samples of knoxjonesi were significantly different from samples of major and jugossicularis in several characteristics. This was particularly true for females. In males, the same trends as for females are present, but the picture is not as clear. This probably results from the smaller sample size and gener¬ ally higher individual variation in males. In only two characteristics (total length and length of maxillary toothrow) were the means for samples of males not significantly dif¬ ferent (ANOVA). In the remaining 10 characters, several patterns of nonsignificant subsets of means were revealed (SS-STP). For two characteristics (condylobasal length and length of rostrum), the samples were divided into two nonoverlapping subsets—one con¬ taining samples of major and jugossicularis; the other, samples of knoxjonesi . The samples of knoxjonesi were significantly smaller than those of the other two subspecies. Subsets containing samples of knoxjonesi and major overlapped only at the sample from Lubbock County and vicinity for greatest length of skull. Again, the means for knoxjonesi were significantly smaller. The other seven characteristics exhibit patterns of two or three broadly overlapping subsets. For three of these characteristics (mastoid breadth, length of nasals, and palato- frontal depth), however, samples of knoxjonesi had the smallest mean values. One characteristic in which knoxjonesi did not average smaller than major and jugossicularis was in length of tail. It appears that knoxjonesi has a proportionally longer tail in comparison with length of head and body than do major and jugossicularis (average percentage for knoxjonesi samples is 54.2, 49.2, and 55.0, as compared with 45.4, 45.9, 45.5, and 44.5 for major and jugossicularis ). Only in length of maxillary toothrow were the sample means of females not significantly different. In three characteristics (greatest length of skull, mastoid breadth, and length of rostrum), the three female samples of knoxjonesi formed a subset that did not overlap the subset formed by the samples of major and jugossicularis. Samples of knoxjonesi also are significantly smaller than all samples of major and jugossicularis, with the exception of the sample from Bailey and northern Cochran counties, Texas, and Curry and Roosevelt counties, New Mexico, which is intermediate in four characteristics (condylo- BAKER AND GENOWAYS—NEW SUBSPECIES OF GEOMYS 11 basal length, interorbital breadth, length of nasals, and palatofrontal depth). This sample of G. b. major is intermediate between typical major and knoxjonesi , these four characteristics being in subsets with each taxon. As in males, females of knoxjonesi have a proportionally longer tail (53.0, 50.1, and 54.3) than do those of major and jugos- sicularis (41.1, 43.3, and 44.4). The one sample of major that ap¬ proaches knoxjonesi in this characteristic is the one from Bailey and Cochran counties, Texas, and Curry and Roosevelt counties, New Mexico, in which the ratio of the length of tail to head and body length is 49.8. Based on the univariate analyses, it appears that G. b. knoxjonesi is a distinctly smaller subspecies than either G. b. major or G. b. jugossicularis and is more distinct from both than either is from the other. These differences are more marked in females than in males, but the same trends are present in both sexes. In females, an inter¬ mediate sample between the geographic ranges of knoxjonesi and major (Bailey and Cochran counties, Texas, and Curry and Roosevelt counties, New Mexico) is morphologically intermediate in several characteristics, although significantly different from knoxjonesi in several others. This intermediate tendency was not evident in males. Another characteristic of samples of knoxjonesi is that they possess relatively long tails in comparison with the length of head and body. In the multivariate analyses that were conducted, the OTUs were sample means. Phenetic distance coefficients were derived from stand¬ ardized characteristic values; these were clustered using UPGMA (unweighted pair-group method using arithmetic averages), and a phenogram was generated. Also, the first three principal components were extracted from a matrix of correlation among the 12 characters. A projection matrix for the first three dimensions was generated and used for plotting OTUs onto these principal components (see Geno- ways, 1973, for additional discussion of these techniques). In addition to the samples used in the univariate analyses, samples of the follow¬ ing subspecies were used in the multivariate analyses (see also speci¬ mens examined): pratincola, ammophilus , attwateri , brazensis, dutcheri, texensis, and llanensis. Additionally, several individuals from near the range of knoxjonesi , for which no chromosomal data were available, were tested to determine their morphometric relation¬ ships. These specimens originated from the following localities: 2.9 mi. S Patricia, Martin County, Texas (one female); 4.5 mi. SSW Morton, Cochran County, Texas (one male); 1 mi. SE Santa Rosa, 12 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Ml M2 M3 J MH SA K1 K2 C AT D T L K3 PR B I_I_I_I_I_I_1_I 1420 1.020 0.620 0,220 Fig. 4.—Phenograms of samples of Geomys bursarius (males left, females right) computed from distance matrices based on standardized characters and clustered by unweighted pair-group method using arithmetic averages (UPGMA). The cophenetic correlation coefficient for males is 66.2 per cent; for females, 79.5. Symbols used are as follows: Kl, G. b. knoxjonesi from Winkler and Ward counties, Texas; K2, G. b. knoxjonesi from Andrews, southern Cochran, Gaines, and Terry counties, Texas; K.3, G. 6. knoxjonesi from Chavez, Eddy, and Lea counties, New Mexico; Ml, G. b. major from Crosby, Dickens, Garza, and Lubbock counties, Texas; M2, G b. major from Collingsworth County, Texas; M3, G. b, major from Bailey and northern Coch¬ ran counties, Texas, and Curry and Roosevelt counties, New Mexico; J, G. b. jugossieularis ; AM, G. b. ammophi!us\ AT, G. b. attwateri\ B, G. b. brazensis; D, G. b. dutcheri; L, G. b. llanensis ; PR, G. b. pratincola; T, G. b. texensis; C, single male from 4.5 mi. SSW Morton, Cochran Co., Texas; MH, sample from Midland and Howard counties, Texas; PA, single female from 2.9 mi. S Patricia, in Martin County, Texas; SA, single male from 1 mi. SE Santa Rosa, Guadalupe County, New Mexico. Guadalupe County, New Mexico (one male); Midland and Howard counties, Texas (one male, seven females). The phenogram (Fig. 4) resulting from clustering of phenetic dis¬ tance coefficients for females is divided into three major groups. One sample is composed solely of G. b. ammophilus. The second group includes the three samples of G. b. major, a sample from Mid¬ land and Howard counties (which would be assigned to major based on these data), and samples of jugossieularis, llanensis, and attwateri. Within the third group, the three samples of knoxjonesi form a dis¬ tinct cluster from samples of pratincola, brazensis, dutcheri, and texensis. The specimen from near Patricia is within this group. Based on this analysis, it appears that knoxjonesi has a greater morphologi¬ cal similarity to subspecies of Geomys bursarius from central and eastern Texas than to geographically contiguous samples of G. b. major and G. b. jugossieularis. BAKER AND G ENOW AYS—NEW SUBSPECIES OF GEOMYS 13 In the phenogram (Fig. 4) for males, two major clusters are present. The upper cluster contains the three samples of major and one of jugossicularis. Also in this group are the sample from Midland and Howard counties, Texas, and the individual from Santa Rosa, New Mexico. Within the other cluster, three subclusters are evident. The upper of these contains the two Texas samples of knoxjonesi and the single specimen from south of Morton, Texas. The second subcluster contains samples of the subspecies attwateri , dutcheri, texensis , and llanensis. The last subcluster contains the New Mexican sample of knoxjonesi and samples of pratincola and brazensis. Males, as do fe¬ males, of knoxjonesi have a greater morphometric similarity to those from samples of Geomys bursarius from eastern Texas than they do to males in contiguous populations. The OTUs projected onto the first three principal components are shown in Fig. 5. For males, these two components account for 82.5 per cent of the total phenetic variation (71.2 for I and 11.3 for II) and for females 79.3 per cent (60.9 for I and 18.4 for II). Results of the factor analyses are shown in Table 2. For both sexes, size is the major influence in component I. Males and females both show high positive weighting for interorbital breadth and length of maxillary toothrow and high negative weighting for length of tail in component II. Highest weighting is for length of tail in component III in males. Females have a high negative value for length of tail and a high posi¬ tive one for length of rostrum in the third component. In the plots, samples of knoxjonesi form a cluster separated from others. The cluster is much tighter in females than in males. In both sexes, knoxjonesi is separated from major in both the first and second components. The sample of jugossicularis is separated from knox¬ jonesi in the first component. The main separation of other samples is in the second component. The sample of attwateri also may be sepa¬ rated in the first component, at least in females. The sample of llanensis appears morphologically nearest to G. b. knoxjonesi in the plot of females, whereas llanensis and texensis are nearest for males. The multivariate analyses clearly indicate that G. b. knoxjonesi is morphologically distinct from contiguous populations of G. b. major . In fact, knoxjonesi shows greater distinctness from major than do any of the other taxa included in this study; it evidently has affinities, both morphologically and karyotypically, with populations of G. bursarius from central and eastern Texas. It would appear to be more closely related to G. b. llanensis and G. b. texensis than to other races to the east. 14 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY T 0 I Fig. 5.— Two dimensional projections of the first two principal components illustrating the phenetic position of samples of Geomys hursarius (males, upper; females, lower). See Fig. 4 for key to symbols. Significance of karyotypic variation .—The karyotype serves to identify populations at the subspecies level, but the actual role of this BAKER AND GENOWAYS—NEW SUBSPECIES OF GEOMYS 15 ■<3 3 E o c r-- sO M f*n ml m m> oo O MH Os as ’'t oo Os 'O ■"fr OO a w mi sO on o o M r-* O r-~ o E c U d 1 d 1 o 1 o d o o o o i O d d c _ o a c. E E o U o _ 00 M <0 ■'t i mi o ms o o 00 o ■t wn O — ml ml mi »o mi O o O o o d o d o o o o ms oo ’^r r- Os Os ^r r- Os Os O r- o m wn 00 r^i o mi O SO Os so Os Os OO *— Os as m m Os o o o o o o o o o o o o nf mi cn »o ^j- mi o o w> tj- im o r -4 o SO 00 nt- r- m o so ■'t ’^r — mi *—• o o o o o o o o o d o o o Os SO oo sn oo „ so d» , ' cfl i2 3 = cfl i; — '£ sn 2 O C i_ CL 1 ) T3 2 £ jc (h o a n a s o o o £ £ £ ^ oo oo oo 2 ~ c c c 2 — 6