TEXAS TECH UNIVERSITY Natural Science Research Laboratory Occasional Papers Museum of Texas Tech University Number 296 25 August 2010 RbpI in Geomyid Rodents: Reduced Rate of Molecular Evolution or Evidence for Selection? Robert D. Bradley, Cody W. Thompson, and Ryan R. Chambers Abstract DNA sequences from the interphotoreceptor retinoid binding protein gene ( Rbp3 ) in pocket gophers ( Geomys ) display an unusually slow rate of molecular evolution relative to other species of rodents. Rates of molecular evolution were examined in pocket gophers and other members of the rodent superfamily Geomyoidea to determine if this phenomenon was restricted to pocket gophers. DNA sequences from the Rbp3, mitochondrial 12S ribosomal RNA(12S rRNA), and mitochondrial cytochrome-6 (Cytb) genes were compared within members of Geomys, among members of the Geomyidae, and among members of the Geomyoidea to ascertain rates of molecular evolution for the three genes among the various taxa. A variety of analyses (genetic distance, Tajima’s relative rate test, Tajima’s neutrality test, coalescence theory, and Hudson, Kreitman, and Aguade test) indicated that DNA sequences affiliated with Rbp3 in species of Geomys were evolving at a rate slower than were sequences of members of the Heteromyidae. In addition, there was weak evidence suggesting that the Rbp3 gene in other pocket gopher genera ( Cratogeomys, Orthogeomys, Pappogeomys, and Thomomys ) evolved more slowly than in members of the Heteromyidae. Key words: geomyoid rodents, Geomys, interphotoreceptor retinoid binding protein, molecular evolution, pocket gophers, Rbp3 Introduction Pocket gophers of the genus Geomys are fossorial rodents distributed throughout the central plains and southeastern regions of the United States and coastal regions of northeastern Mexico (Russell 1968; Hall 1981; Baker et al. 2003; Patton 2005). Distributions of pocket gophers are affected by availability of suit¬ able soil types (Davis 1940; Baker et al. 2003), and as a result, populations generally contain few individuals and are isolated from other conspecific populations. Pocket gophers also are highly territorial leading to a solitary lifestyle with limited vagility (Williams and Baker 1976; Smolen et al. 1980) and non-overlapping home ranges. In addition, past glacial events in the central plains region are thought to have had a major impact on speciation and distributions of members of Geomys (Russell 1968; Hart 1978). Studies of genetic 2 Occasional Papers, Museum of Texas Tech University evolution indicate that pocket gophers (probably as a consequence of the above factors) have small effective population sizes and possess low levels of intrapopu- lational and intraspecific variation; however, variation among populations and species is high and overall levels of heterozygosity is low (Selander et al. 1975; Penny and Zimmerman 1976; Avise et al. 1979; Zim¬ merman and Gay den 1981; Ruedi et al. 1997). Recent studies pertaining to systematic relation¬ ships among species in Geomys have produced DNA sequence data for two mitochondrial genes (12S ribo- somal RNA- 12S rRNA, Jolley et al. 2000; cytochrome- b - Cytb, Sudman et al. 2006) and one nuclear gene (interphotoreceptor retinoid binding protein - Rbp3, Chambers et al. 2009). Although the goals of these studies were to reconstruct phylogenetic relationships among taxa, Chambers et al. (2009) noted unusually low levels of genetic divergence among species for Rbp3 relative to the other two genes. Specifically, Chambers et al. (2009) reported an average between species genetic divergence of 0.60% (0.08%-1.5%) for the Rbp3 gene, whereas similar comparisons among the same taxa yielded divergence values of 3.67% (0.6%-8.1%) for 12S rRNA and 13.8% (8.1%-21.0%) for Cytb. Although it is well known that nuclear genes evolve at slower rates than do mitochondrial genes, the low level of genetic divergence associated with Rbp3 was unexpected given the higher levels of genetic divergence reported for other rodent taxa (Stanhope et al. 1996; Weksler 2003). The goals of this study were to determine: 1) whether the low rate of molecular evolution in Rbp3, as reported by Chambers et al. (2009), is restricted to Geomys - or is it typical for other genera of pocket go¬ phers, and 2) if the rate of molecular evolution in Rbp3 is a product of the following scenarios: a) population dynamics, b) age of the geomyid lineage, c) reduction of vision as a product of a fossorial lifestyle, or d) selec¬ tive forces by examining rates of molecular evolution for genes unrelated to Rbp3 (12S rRNA and Cytb). To examine these goals, DNA sequences were obtained for Rbp3, 12S rRNA, and Cytb in other genera of pocket gophers ( Cratogeomys, Orthogeomys, Pappogeomys, and Thomomys,) and five genera of the rodent family Heteromyidae ( Chaetodipus, Dipodomys, Heteromys, Liomys , and Perognathus). The Heteromyidae (kan¬ garoo rats and pocket mice) is sister to the Geomyidae and together the two families comprise the superfamily Geomyoidea. In general, the Heteromyidae possess larger population sizes and presumably a greater de¬ pendence on vision, and therefore offer an opportunity to examine the four scenarios presented above in taxa that have different demographic and natural history traits than the Geomyidae. Methods Taxonomic sampling. —DNA sequences for Rbp3 , 12S rRNA, and Cytb were either generated in this study or obtained from GenBank for 29 individuals from the Geomyidae: Geomys (21 individuals representing 12 species), Cratogeomys (2 individuals representing 2 species), Orthogeomys (1 individual), Pappogeomys (2 individuals from 1 species), and Thomomys (3 individu¬ als representing 3 species) and 13 individuals from the Heteromyidae: Chaetodipus (3 individuals representing 3 species), Dipodomys (4 individuals representing 4 species), Heteromys (1 individual), Liomys (2 indi¬ viduals representing 2 species), and Perognathus (4 individuals representing 3 species). Three individuals representing Castor canadensis were used for outgroup comparisons. GenBank accession numbers and mu¬ seum voucher numbers are provided in Table 1. PCR and sequencing methods. —Twenty unre¬ ported Rbp3 sequences were obtained in this study. Genomic DNA was isolated from approximately 0.1 g of frozen liver or muscle tissue using the Puregene DNA isolation kit (Gentra, Minneapolis, Minnesota). Approximately 1,230 bp near the 5’ end of exon 1 of the single-copy Rbp3 gene was amplified by the poly¬ merase chain reaction (PCR, Saiki et al. 1988) using primers A, B, D, D2, E2, F, 125F, G, and I (Stanhope et al. 1992; Jansa and Voss 2000; DeBry and Sagel 2001; Weksler 2003; Chambers et al. 2009). Thermal profiles were adapted from those of Jansa and Weksler (2004): initial denaturation at 95°C for 10 min, 35 cycles of denaturation at 95°C for 25 sec, annealing at 58°C for 20 sec, and extension at 72°C for 60 sec, and a final extension at 72°C for 10 min. Bradley et al.— Rbp3 in Geomyid Rodents g co | a t -s § g 1 I I S3 Sg £ .5 • ~ a R 'S -■s 5 cn t> os cn Ph < < g, os cn CN 1 5 P ^t- VD OS o CN OS VO P H H CN os CN ^1- 00 o P-H < (N Ov VO P H H os (N 00 o Ph < Ov CN Ov so P H H ^l- Ov CN ^ 1 - 00 o P-H c vo os CN ^ 1 - 00 o P-H < 0- Ov CN ^ 1 - 00 o P-H < Ov CN ^ 1 - 00 o P-H < os CN ^ 1 - 00 o P-H < g Ov OS CN ^ 1 - 00 o Ph < g t— t— g P W P W P W P W cn CN P W P PP S' o S cn P PP C 3 -s: P PP P PP in in P PP T3 § aupopppppp B o O P PP P PP R © .o 63 § CO 1 ,-C £ »? o 1 SP "-s 'S CJ 1 O g a B Ki, p P p p P P o N" OS N" CO OS cn «5 _ / _^ r _^ , _ v CN ^_ s f _ s in o Ov Ov CN os VD o VO CN 00 cn ^H o cn o o CN vo VD Ov vo LSI vo H H H H H H hP H >—5 on H o Ph < o CN 00 00 o CN m r- VO 00 cn Ov o N- r-- cn cn OS 00 CN o O CN Ov cn O VD cn r- CN o ^-N ^ 1 - cn fN.'l CN CN CN cn o CN o (N s. VD CN Ov PS OS in Ov [Sj Ov VD Ov VD [Sj CN OS vo 5 vo r- vo VD r- VO S 00 i— 1 VD P p P P P p P P p P CO CN Q p H c n H H H 00 H H H H 00 Z p 2 P H H hP H, p. | ) 00 H 00 VD vo n- CN cn o r- in n- ,—1 r—i 00 VD C" in os Ov 00 00 00 o Ov 00 r— H 0\ 00 00 P PP 8 , bo Occasional Papers, Museum of Texas Tech University »c> £ 5 < & Ov O N" in U £ u H, o vo ov co ov CO o £ o H o U-i P P pq pq o vo cn ov r- ov vo vo ri ov GO N- Q p N- 00 vo p H GO 't p H p C/2 ^H p H pp ■J CO in N r-q w cl H. hJ, H, CO s> 00 N" vo ,—1 z 2i 'Ct CN VO OV CO OV CO vo OV CO OV CO vo OV CO OV CO VD OV CO OV CO vo OV CO OV CO IN OV CO OV CO o i> N" 00 CO OV CO o 00 00 o o OV vo o W N- 'd- 00 uo CN o OV CN in >H >H >H >H >H p q pp Pp vo < < < < < < p pq p < << P ov o 'Cl¬ in U £ U H o Ph < in tj- o ^- N ov Ov ov N" vo (N 00 o CN CN CN i-H 00 *“H ON, Ov ov N" in vo VO vo n O' P pq co CN Ov 't < i g i> OV o vo in PP pq & §> ! P H H o Ph < s r-- r- (N t" ov r i Ph < £ ^1- CO CN c p H H, o ov cn N- 00 o Ph < P H H ov 8 H | cn ^l- CN OV o o s N J Pp pq ov in co On IN CN o o o H H H S 5 5 i> ^l- CN Ov o o s ov N- cn On O O 5 VO ri Ov 5 't P H H Pp < P H H PP c H < < £ P H H o Pp < H c q q I' o i CO :§ i’2 a q a a *5 a § "CS 1 a 0 q I' 1 ^a -p 0 g a -1 u ^a P a P ,|h Q o cS q Q q p £ o < c/ o §D S O ^ -2 3f- o cn 't U £ U H £ H a Os 'O Os oi so ^i- D H H Os 8 to to D H H a Q & H H Os o O H os o ot- /o D oo U £ u I £ H to Os ' s —' Os t> 00 F" SO Os SO I/O Cl I/O o P-H SO P-H m 0 W o Os. o D H H Os D O q; a; PCR products were purified using the Exosap-II PCR purification kit (USB Corp., Cleveland, Ohio). Amplified gene products were sequenced on an ABI 3100-Avant using ABI Prism Big Dye v3.1 terminator technology (Applied Biosystems, Foster City, Califor¬ nia). Primers used to cycle sequence Rbp3 included B, D, E2, F, 125F, Geo395R, Geo609F, Geo958R, Geol405R, and 1000F, (Stanhope et al. 1992; Jansa and Voss 2000; DeBry and Sagel 2001; Weksler 2003; Chambers et al. 2009). Primers beginning with “Geo” were modified from Stanhope et al. (1992) by altering nucleotides so they matched sequences of Geomys more specifically. Cycle sequencing reactions were purified using isopropanol cleanup protocols. Sequences were assembled and proofed using Sequencher 4.9 software (Gene Codes, Ann Arbor, Michigan) and chromato¬ grams were examined to verify all base changes and to inspect sequences for heterozygous sites, which were coded following the International Union of Biochemistry (IUB) polymorphic code. MEGA 4.1 software (Kumar et al. 2007) was used to align and inspect sequences for the presence of stop codons and pseudogenes. Data Analyses .—To examine rates of molecular evolution in the three genes examined in this study (Rbp3 - 1,230 bp, 12S rRNA- 870 bp, and Cytb -1,140 bp), five methods were implemented for data analysis. First, neighbor-joining trees (Saitou andNei 1987) were generated independently using DNA sequences from each of the three genes so that taxonomic relationships and corresponding branch lengths (indicating rates of molecular evolution) could be compared among genes. The neighbor joining analyses used uncorrected-P genetic distances obtained using the MEGA 4.1 soft¬ ware (Kumar et al. 2007) for each of three respective genes. The uncorrected-P distance was selected to avoid interjecting “rules of molecular evolution” on the DNA sequences as incorporated by the various substitution models commonly used in calculating genetic distances. This choice was crucial so that rates of molecular evolution could be compared as evenly as possible among nuclear and mitochondrial genes. Average uncorrected-P distances were estimated for individuals within each genus and between genera and used to estimate levels of genetic divergence between various taxonomic groups. 6 Occasional Papers, Museum of Texas Tech University Second, Tajima’s relative rate test (Tajima 1993) using MEGA 4.1 software (Kumar et al. 2007) was used to ascertain if rates of molecular evolution differed significantly among taxa and among genes. Specifically, this test was implemented to determine if Rbp3 sequences in Geomys, and geomyids in gen¬ eral, were evolving at rates different (i.e., evidence for rate heterogeneity) than those of heteromyids relative to DNA sequences from 12S rRNA and Cytb. Pairwise comparisons of DNA sequences from each of the three genes were made between species within Geomys , between species of Geomys and other pocket gophers, and between geomyids and members of the Heteromyidae. Third, Tajima’s neutrality test (D-statistic, Tajima 1989) using MEGA 4.1 software (Kumar et al. 2007) was implemented to determine if DNA sequences were evolving under a neutral model of evolution (Kimura 1983) or under non-random models normally associated with selective forces (directional selection, balancing selection, demographic expansion or contraction, ge¬ netic hitchhiking, etc.). Specifically, the neutral model of evolution would be operative, and would remain a viable hypothesis, if rates of molecular evolution at the three loci were not significantly different among Geomys and other members of the Geomyoidea. Fourth, the Hudson, Kreitman, and Aguade test (HKAtest, Hudson et al. 1987) was used to determine if the Rbp3 was behaving in a neutral fashion relative to 12S rRNA and Cytb. The HKAtest estimates theta (0) from the following equation, 0 = 4N e p, where N e is the effective population size and p is the mutation rate. Theta is estimated for each locus based on comparing the intrapopulational genetic variability for one taxon with the interpopulational genetic variability between that taxon and a second. The DnaSP software pro¬ gram (version 5.10.01, Librado and Rozas 2009) was used to estimate theta values at each locus (Rpb3,\2S rRNA, and Cytb) for six genera of geomyoid rodents (Chaetodipus, Cratogeomys, Dipodomys, Geomys, Perognathus, and Thomomys) and one outgroup taxon (Castor). A chi-square test (P < 0.05) was used to iden¬ tify significant differences among pairwise comparisons of the three loci, with one locus representing observed values and the second locus representing expected values. Significantly different 0 values indicated a deviation from neutrality (i.e. selection), with positive selection inferred if ps for each locus were equal and the N e was unequal and purifying selection inferred if ps for each locus were unequal and the N e was equal. In other words, under a model of neutrality, (Kimura 1983; Hudson et al. 1987) all loci are expected to pos¬ sess equal ps if all taxa have the same the N e ; however if taxa have unequal N es , then positive selection acts upon individual loci producing an excess of polymorphisms between species, conversely, if taxa possess unequal ps, then purifying selection generates an excess of polymorphisms within a species. Fifth, coalescence theory was used to estimate the time of divergence from a hypothetical common ancestor based on DNA sequences from the three genes. If Rbp3 sequences coalesce at times similar to those obtained for 12S rRNA and Cytb, then the hypothesis of a slower rate of Rbp3 evolution in Geomys could be rejected. The software program BEAST vl.5.3 (Drummond and Rambaut 2007) was used to analyze the coalescence process among each gene. All taxa were grouped into all possible taxon sets (e.g., Castori- morpha, Geomyoidea, Geomyidae, Geomyini, Geomys, etc.). Two fossil calibrations of ancestral taxa (Casto- rimorpha - 54.4 MYA, McKenna 1960; Geomyoidea - 45.45 MYA, Walsh 1991) were used as priors on the tree. A normal distribution was used for all point fossil calibrations with standard deviations based on dates from the International Commission on Stratigraphy (Gradstein et al. 2004; Ogg et al. 2008). A relaxed, uncorrelated lognormal clock was used with a GTR + I + G model of substitution based on MrModeltest 2.3 (Nylander 2004) and the Akaike information criterion (Nylander 2004) for each gene. In addition, a Yule species prior was used to date nodes within each gene tree. Each dataset was analyzed twice for 10,000,000 generations (with a 10% burn-in) to obtain an appropri¬ ate effective sample size. The log files were combined using LogCombiner vl.5.2 (Drummond and Rambaut 2007) and analyzed for convergence in Tracer vl.4.1 (Rambaut and Drummond 2007). A one-way analysis of variance (ANOVA, P < 0.05) was used to compare the mean rates of substitution to determine whether genes were evolving at different rates. Bradley et al.— Rbp3 in Geomyid Rodents 7 Results Taxonomic relationships and genetic divergence .— Genetic divergence values, based on uncorrected-P distances, were estimated for individuals within each genus and between genera for the three respective genes (Table 2). Within genera values ranged from 0.66% for individuals within Geomys to 5.98% within Liomys for Rbp3, from 2.38% for individuals within Perognathus to 12.02% in Chaetodipus for 12S rRNA, and from 11.93% for individuals within Geomys to 18.67% in Cratogeomys for Cytb. In addition, these values were used to construct a neighbor-joining tree for each of the three genes (Fig. 1). Topologies recovered in the three analyses were similar, although placement of some heteromyid genera differed depending on which gene was analyzed. However, branch lengths, reflect¬ ing the number of substitutions per site, were different between genes and between taxa in each tree. For example, in all analyses, branch lengths for individual species of Geomys were substantially shorter than for other taxa. Rate heterogeneity. —Tajima’s relative rate test (Tajima 1993) depicted specific taxa that exhibited dif¬ ferential rates of molecular evolution relative to other members of the Geomyoidea based on comparisons within each of the three genes (Table 3). In most intra¬ generic comparisons, the 12sRNAgene accounted for a greater number of significantly different rates (P < 0.05) than the other two genes. However, in comparisons involving members of Geomys versus heteromyids and geomyids versus heteromyids, Rbp3 depicted a greater number of significantly different rates (Table 3). Neutral model of molecular evolution .—DNA sequences obtained from the three genes were tested independently for departure from the model of neutral¬ ity using Tajima’s neutrality test (Tajima 1989) and the HKA test (Hudson et al. 1987). Tajima’s neutrality test provided evidence of positive selection or a previ¬ ous history of having been subjected to a population bottleneck in four instances (Table 4). Two cases Table 2. Average genetic distances (uncorrected-P distances) were estimated for each of the three genes examined in this study. Values were estimated by averaging genetic distances for comparisons of selected taxa. Those with a single sequence prohibited the calculation of an average distance and are indicated by N/A. Abbreviations are as fol¬ lows: interphotoreceptor retinoid binding protein gene (RbpJf mitochondrial 12S ribosomal RNA (12S rRNA), and mitochondrial cytochrome -b (Cytb). Taxon Rbp3 12S rRNA Cytb Within Geomys 0.00661 0.03568 0.11926 Within Cratogeomys 0.1771 N/A 0.18670 Within Pappogeomys N/A N/A N/A Within Thomomys 0.02520 0.07732 0.17970 Within Chaetodipus 0.01749 0.12022 0.15510 Within Dipodomys 0.01981 0.11107 0.15341 Within Liomys 0.05976 0.05472 0.15263 Within Perognathus 0.01439 0.02375 0.16910 Within Geomyidae 0.02193 0.06869 0.15867 Within Heteromyidae 0.08326 0.16319 0.21493 Within Geomyoidea 0.07862 0.14693 0.20546 8 Occasional Papers, Museum of Texas Tech University Rbp3 Castor (1/3) 12S rRNA r"> Cratogeomys (2/2) t-3- Orthogeomys (1/1) Pappogeomys (1/2) \ Geomys (12/21) L-U homomys (3/3) Dipodomys (4/4) Chaetodipus (3/3) Perognathus (3/3) Heteromys (1/1) -> Liomys (2/2) r 1 } Castor {M3) ■3- Cratogeomys (1/1) Pappogeomys (1/2) r \ ^ Geomys (12/21) _> Thomomys (2/2) Dipodomys (4/4) " 3 - Chaetodipus (2/2) £■> Perognathus (2/2) 3- Heteromys (1/1) Cyfb ■f} Castor (1/3) £ _3" Cratogeomys (2/2) —3" Orthogeomys (1/1) ■c> Pappogeomys {M2) tE ■ J- Thomomys (3/3) £ hE 7^- Dipodomys (4/4) -J- Chaetodipus (3/3) Liomys (2/2) Perognathus (3/3) ■-3“ Heteromys (1/1) M > L/omys (2/2) 0.05 substitutions/site 0.05 substitutions/site 0.05 substitutions/site Figure 1. Neighbor joining trees obtained from uncorrected-P genetic distances estimated from DNA sequences obtained from the Rbp3 , 12S rRNA, and Cytb genes. Only genera are labeled and numbers in parentheses following each genus represent: number of species included per genus (left of slash), and number of DNA sequences included per genus (right of slash). Bradley et al.— Rbp3 in Geomyid Rodents 9 Table 3. Number of significant differences (V < 0.05) in pair-wise comparisons based on Tajima's relative rate test (Tajima 1993). Numbers to left of the slash represent the number ofsignificant comparisons and numbers to right of the slash indicate the number of comparisons attempted. Abbreviations are as follows: interphotoreceptor retinoid binding protein gene (RbpJ,), mitochondrial 12S ribosomal RNA (12SrRNA), and mitochondrial cytochrome-b (Cyt b). Taxon Rbp3 12S rRNA Cytb Within Geomys 5/66 9/66 0/66 Within Cratogeomys 0/1 0/0 0/1 Within Pappogeomys 0/0 0/0 0/0 Within Thomomys 0/3 0/1 0/3 Within Chaetodipus 0/3 0/1 0/3 Within Dipodomys 0/6 3/6 1/6 Within Liomys 1/1 0/1 0/1 Within Perognathus 0/3 0/1 1/3 Geomys to Other Geomyids 0/84 1/48 3/84 Within Geomyids 4/171 10/120 7/171 Within Heteromyids 3/78 10/55 7/78 Geomys to Heteromyids 81/156 20/132 15/156 Geomyids to Heteromyids 124/247 27/176 25/247 Table 4. Results from Tajima’s neutrality test (Tajima 1993) for each of the three genes examined. The outcome of Ta¬ jima ’s neutrality test is based on Tajima’s D statistic. Taxa with sample sizes of <3 gave inconclusive results and were not included. Abbreviations are as follows: interphotoreceptor retinoid binding protein gene fRbpi^), mitochondrial 12S ribosomal RNA (12S rRNA), mitochondrial cytochrome-b (Cytb), population bottleneck (PB), positive selection (PS), and balancing selection (BS). Taxon Rbp3 12S rRNA Cytb Within Geomys PB or PS (90% Cl) PB or PS (<90% Cl) BS Within Dipodomys BS BS BS Within Geomyids PB or PS (<90% Cl) PB or PS (<90% Cl) BS Within Heteromyids BS BS BS Within Castorimorphs BS BS BS 10 Occasional Papers, Museum of Texas Tech University involved comparisons of taxa within Geomys ( Rbp3 and 12S rRNA) and two cases involved comparisons of taxa within the Geomyidae (Rbp3 and 12S rRNA). Based on this test, heteromyid taxa (generic or family level) and Cytb sequences from all taxa appear to be evolving at neutral rates in all comparisons. In addi¬ tion, the HKAtest (Hudson et al. 1987) indicated that 0 values between Rbp3 and Cytb were significantly different (P = 0.0016). The HKAtest did not detect any other significant differences in 24 additional pairwise comparisons of genera and loci, which suggests that purifying selection was responsible for a slower rate of molecular evolution at Rbp3 in Geomys but that the remaining sequences were evolving at a neutral rate. Coalescence theory. —The mean rates of evo¬ lution (substitutions per site per million years) were 0.0023, 0.0067, and 0.0138 for Rbp3, 12S rRNA, and Cytb , respectively. The coefficient of variance for Rbp3 and 12S rRNA were high (0.4847, 0.6783) but low for Cytb (0.0872). A one-way ANOVA(F = 2.1832 x 1012, P ~ 0.0000) rejected the null hypothesis of equal rates among the three datasets, indicating independent rates of evolution for each gene. In addition, trees obtained from each of the three genes used in the BEAST analy¬ sis depicted more recent divergence times for species of Geomys based on Rbp3 than for the other two genes (Fig. 2). Discussion The observation (Chambers et al. 2009) thatRfy?3 sequences obtained from several species of Geomys were evolving at rates slower than sequences obtained from other genes for the same taxa was re-examined using genetic distances (uncorrected-P), relative rate test (Tajima 1993), neutrality tests (Tajima’s D statis¬ tic, Tajima 1989; HKAtest, Hudson et al. 1987), and coalescence theory (BEAST, Drummond and Rambaut 2007). All analyses, whether visual (comparison of genetic distances) or statistically supported (Tajima’s relative rate test, Tajima’s test of neutrality, HKAtest, or coalescence theory) indicated that species of Geomys were evolving at a rate slower compared to members of the Heteromyidae. Also, other pocket gopher genera ( Cratogeomys, Orthogeomys, Pappogeomys, and Thomomys) appeared to evolve more slowly than their heteromyid counterparts, although low sample sizes prevented meaningful statistical analyses in some cases. Although the various analyses performed in this study revealed differences in the molecular evolution of Rbp3 in geomyids and heteromyids, with geomyids consistently possessing a slower rate of evolution, it was not clear from a molecular standpoint why geomy¬ ids possessed a slower rate. To further investigate this phenomenon, we determined the number of variable sites per codon position (1st, 2nd, and 3rd) for DNA sequences obtained from the two protein-coding genes {Rbp3 and Cytb)', the 12S rRNA gene was not included as it is not a protein-coding locus. The average number of variable sites (by position) was determined at the generic and familial levels for geomyids and hetero¬ myids (Table 5). A chi-square test was used to detect differences in the observed number of variable sites (represented by the number of changes per position in Rbp3) versus the expected number of variable sites (represented by the number of changes per position in Cytb). Cytb was selected as the “expected” value to approximate a neutral rate. Significant differences (P < 0.05) were detected among taxa for Rbp3 relative to Cytb, with the genera of geomyids possessing a signifi¬ cantly lower number of substitutions, in the 1st and 3rd positions relative to the other taxa (Table 5). At least four scenarios are possible for explain¬ ing the low level of genetic variation in the Rbp3 gene in Geomys and for pocket gophers in general. First, the product of being fossorial has resulted in pocket gophers being distributed in small isolated populations, susceptible to inbreeding, and generally characterized by low levels of heterozygosity, etc. Also, it is well known that glacial periods had a major impact on the distribution and speciation of Geomys (Blair 1954; Rus¬ sell 1968; Penney and Zimmerman 1976; Heaney and Timm 1983; Mauk et al. 1999) by producing population bottlenecks during glacial maxima. These events may have acted to homogenize or constrain evolution of the Geomys genome. However, these arguments seem unlikely given that levels of genetic variation reported Bradley et al.— Rbp3 in Geomyid Rodents 11 Figure 2. Coalescence trees were generated using the BEAST analysis (Drummond and Rambaut 2007) and DNA sequences obtained from the Rbp3 (top), 12S rRNA (middle), and Cytb (bottom) genes. The GTR + I + G model of substitution and two combined runs of 10,000,000 generations (with a 10% burn-in) were used for tree construction. Two fossil calibrations of ancestral taxa (Castorimorpha - 54.4 MYA, McKenna 1960; Geomyoidea - 45.45 MYA, Walsh 1991) were used as priors on the tree. Numbers at nodes reflect approximate coalescence times. Table 5. Number of variable nucleotide sites per codon position for the protein-coding genes, interphotoreceptor retinoid binding protein gene fRbpi^) and mitochondrial cytochrome-b fCyt b), respectively. For each taxonomic group and associated genes, the following is depicted: average number of bases examined (ANBE), total number of changes per gene, number of changes at the 1st position, number of changes at the 2nd position, number of changes at the 3rd position, and statistical significance (SIGN). A chi-square test (P < 0.05) was used to determine if the number of nucleotide changes (by position) in the Rbp3 gene (treated as observed data) was significantly different relative to changes per position in the Cyt b gene (treated as expected data). Occasional Papers, Museum of Texas Tech University SIGN Yes Yes Yes No No No Yes Yes No Yes T3 98 CM 82 NO NO o r- 't cn En o (N cn 't CM 't in In cn cn (N CM CM cn cn cn Eh _o 00 O 2nd no 't in 00 cn - 32 40 cn NO Ph Eh n CM ON CM nT CM Eh O 2nd m no X (N 't 't CM ON 23 N" o 0- S—