ABSTRACT Gillichthys mirabilis, the longjaw mudsucker, is a Gobiid fish that lives under a wide variety of temperature ranges in bays and estuaries along the Pacific coast of North America. A population genetics study could suggest whether genetic differences associated with differential physiological adaptations have occurred. Samples of Gillichthys mirabilis from three locations: Hayward, CA, San Diego, CA, and San Felipe, Mexico, were analyzed using isozyme electrophoresis and Amplified Fragment Length Polymorphism (AFLP), a recently developed DNA fingerprinting technique. Three questions were posed: 1) Are the populations of Gillichthys mirabilis separate or interbreeding? 2) If the populations are interbreeding, to what extent does interbreeding occur?, and 3) Is AFLP a sensitive and reliable technique to use for a population genetics study of this species? In the isozyme work, staining for malate dehydrogenase revealed an allele that was present only in the San Diego fish, suggesting little gene flow among the three populations. Preliminary AFLP results show some differences between the Hayward and San Diego samples, as well as the potential for AFLP to be a powerful technique. More in-depth studies, however, will be needed to draw definitive conclusions about the population structure of Gillichthys mirabilis INTRODUCTION Gillichthys mirabilis, the longjaw mudsucker, is a fish of the Gobiidae family. It is commonly found in bays and estuaries along the Pacific coast of North America, from Tomales Bay (about 50 miles north of San Francisco) to Baja California, as well as in the Sea of Cortez (Fields, 1995) (Fig. 1). Samples from three locations: Hayward, California, San Diego, California, and San Felipe, Mexico (Fig. 2), were analyzed to determine whether the populations at these locations were separate or interbreeding. Tused two different methods to analyze the samples: isozyme electrophoresis, which enabled me to study the samples at the protein level, and Amplified Fragment Length Polymorphism (AFLP), a recently developed DNA fingerprinting technique. A third purpose of the project was to determine whether AFLP is a sensitive and reliable technique for studying population genetics of this species. Gillichthys can reside under a wide range of temperature and water conditions. It can burrow in the mud to regulate its body temperature, and has been known to survive at temperatures as high as 35° C, though it prefers temperatures from 9 to 23° C (Love, 1991). The temperature of its natural habitat can vary greatly. Annual temperatures on the coast of Northern California might range from 10 to 18° C, while in the Sea of Cortez the range might be from 9 to 30°C (Fields, 1995). Gillichthys can also survive in a wide range of water conditions, ranging from water that is nearly fresh to water that is two and a half times saltier than sea water (Love, 1991). A population genetics study of Gillichthys might suggest differential physiological adaptations that have occurred among Gillichthys populations, especially since it can live in such varied environments. Current literature on the population genetics of Gillichthys is very limited. Fields (1995) found no differences between the enzyme kinetics, amino acid sequences, and CDNA sequences of Az lactate dehydrogenase in Gillichthys from the San Diego area and the northern part of the Sea of Cortez. Based upon his studies, Fields suggests that Gillichthys uses behavioral thermoregulation rather than biochemical adaptation to survive in the warm northern Sea of Cortez and that the isolation of the California and Sea of Cortez fish from each other may have occurred relatively recently. The study of one protein, however, is not enough to draw conclusions about population genetics; more in- depth studies are needed. Cathleen Davies (personal communication) is currently studying the population structure of Gillichthys using protein electrophoresis, but she has not yet published her results. No one has used the more sensitive DNA techniques to study Gillichthys's population structure. We decided to use both isozyme electrophoresis and AFLP to analyze the samples of Gillichthys. Studying the fish at both the protein and DNA levels provided several advantages. Isozyme electrophoresis is an established method from which it would be likely for me to get results in the short eight week time frame of this study. Using a DNA technique on Gillichthys, however, provides the potential to detect differences that protein work probably could not, since with DNA fingerprinting, many loci are easily studied at once and post-translational effects are not a factor. THEORY Isozyme Electrophoresis Isozyme electrophoresis involves three main steps: extraction of protein, resolution of the protein on a non-denaturing polyacrylamide gel, and staining for the activity of the protein of interest (Fig. 3). When proteins are subjected to gel electrophoresis, they migrate different distances based mainly upon their net charge and size. A more negatively charged, smaller protein will migrate faster. Because proteins with the same function that have been coded for by different loci or by different alleles within a locus usually migrate at different rates and produce different banding patterns, one can deduce the genotype of an individual for a particular protein using isozyme electrophoresis. For example, the top section of Figure 4 shows a monomeric protein. If an individual is homozygous for the faster allele, this is represented by the single band which has migrated farther. If the individual is homozygous for the slower allele, this is represented by the single band that has not migrated as far. Finally, if the individual is heterozygous, it will have two bands, one for each allele. The situation is similar for a dimeric protein, except that the heterozygote will have three bands: one for each of the homodimers and one (the middle band) for the heterodimer, which consists of one component of each of the homodimers. These patterns are due to allozymes, which are encoded by allelic variations at a single locus. The situation increases in complexity (five bands for the heterozygote of a tetrameric protein, etc.) as the number of subunits in of the protein increases (Richardson et al., 1986). With proteins that have more than one locus coding for them, each isozyme, as the product of individual loci are called, will show a banding pattern as described above. Proteins are present in different amounts in different tissue, so the number of and intensity of isozymes will vary among tissue types (Richardson et al., 1986). In my study, however, the same type of tissue was analyzed for each fish, so the type and intensity of isozymes present should be similar. The proteins are stained through a chemical reaction, where the enzyme of interest catalyzes an oxidation-reduction reaction which is coupled with a dye (Richardson et al., 1986) AFLP AFLP was developed by Vos et al. (1995). The method involves multiple steps: isolation of genomic DNA, double restriction digest, ligation of adapters, preamplification, selective amplification and labeling, and autoradiography (Fig. 5). Genomic DNA is cut by a four base cutter (Msel) and a six base cutter (EcoRl). Synthesized oligo nucleotide adapters are then ligated to the ends of the restricted fragments (Fig. 6). In the preamplification step, primers whose sequence are identical to the adapter sequence and sticky end sequence with an additional random base on the 3 end are used. During the PCR reaction, the double stranded fragments with adapters on the end denature, and the primers bind selectively to the fragments with the complementary adapter/sticky end sequence as well as the complement to the random base (Fig. 7). The PCR reaction amplifies these fragments. This provides selectivity by amplifying only fragments with adapters followed by the complement to the random base. In the selective amplification and labeling step, the primers are similar, except they have an additional two random bases to increase selectivity. In this reaction, only the fragments with both the complementary adapter sequence/sticky end sequence and three bases complementary to the three on the primer are amplified (Fig. 8). This increases selectivity again by amplifying only a subset of the fragments that were amplified in the preamplification. Labeling of the fragments with radioactivity also occurs in this step, as the EcoRl primer in this step is labeled with3 The product of the selective amplification and labeling step is resolved on a polyacrylamide sequencing gel. Since each fragment is a unique size, it should migrate at a unique rate and appear as a unique band when visualized. Figures 9 and 10 show what AFLP banding patterns might look like. The gel is visualized by autoradiography. In summary, AFLP produces a banding pattern that is different for each individual, but that is more similar between more closely related individuals. The function and origin of each amplified fragment is not known. Different banding patterns can be produced for the same individual by using primers with different random bases. Differences between individuals can be evaluated by calculating a difference index = + different bands/+ total bands. Pair-wise comparisons could then be made, and populations compared, by using one way ANOVA. MATERIALS AND METHODS Eight Gillichthys specimens were caught using baited minnow traps in the estuaries around San Felipe, Mexico. Over thirty specimens each were caught in the same manner from the estuaries around San Diego, California and in the Cargill salt ponds in Hayward, California. The fish were kept in an aquarium which was filled with sea water at ambient temperature (18° C). When the fish were needed for analysis, they were removed from the aquarium and immediately stored in a -70° C freezer. Isozyme Electrophoresis As mentioned above, for the isozyme work the major steps involved extracting the protein, resolving the proteins on a non-denaturing polyacrylamide gel, and staining the gel for activity of the protein of interest (Lin, 1993) (Fig. 3). Two proteins were stained for: lactate dehydrogenase (LDH), a glycolytic enzyme, and malate dehydrogenase (MDH), an oxidative enzyme. We dissected approximately 0.25 grams of white muscle tissue from each individual. The tissue was ground in 0.75 mL of homogenization buffer, which consisted of 20% glycerol, 20 mM imidazole (pH 7.0), and 3 mM B¬ mercaptoethanol. The homogenized samples were spun in a 4° C centrifuge at 14000 g for 30 minutes to 1 hour. The supernatant was immediately removed and stored on ice or in a 4° C refrigerator until used. If the supernatant was stored for more than a few hours before use, it was centrifuged again for 30 minutes at 14000 g immediately before use. The polyacrylamide gel was composed of two layers: the running gel and the stacking gel. The stacking gel solution for one gel was composed of 3.56 mL double distilled (dd) H2O, 1.50 mL 4X stacking gel buffer (0.5 M Trizma base, pH 6.9), 0.60 mL 30% acrylamide/1% bis-acrylamide, 6 uL TEMED, and 40 uL 10% ammonium persulfate. The running gel solution for one gel was made up of 4.725 mL dd HO, 4.125 mL 4X running gel buffer (1.5 M Trizma base, pH 8.0), 4.125 mL 30% acrylamide/1% bis¬ acrylamide, 8.25 uL TEMED, and 0.15 mL 10% ammonium persulfate (Lin, 1993). The running gel was poured into the gel plates and allowed to polymerize. The stacking gel solution was then poured onto the polymerized running gel. To resolve the samples, we placed the gel in a gel rack and filled the chambers with electrophoretic buffer. The type of electrophoretic buffer varied depending on the protein we planned to stain. For MDH the buffer consisted of 191 uM glycine, titrated with Trizma base to pH 8.95, while for LDH the buffer was titrated to a pH of 8.3. I prepared the sample solutions by adding 15 uL of loading buffer to 40 uL of the supernatant. Blank solutions with water instead of supernatant were also prepared. For MDH gels, 55 uL of sample solution were loaded, while for LDH gels 25 uL were loaded. The gels were run in a 4° C cold room, at 12 mA until the dye reached the running gel layer at which time the voltage was increased to 25 mA, or at 12 mA overnight and 25 mA for a few minutes before the gel was disconnected. When the dye front had run off the bottom of the gel, the electrophoresis was stopped, and the gel was removed from the apparatus for staining. I made staining cocktails with the following reagents: for MDH, 44 mL pH 9.0 staining buffer (0.2 M Tris-HCl), 10 mg NAD', 100 uL (1.25 mg) phenazine methosulfate (PMS), 2 mL nitro blue tetrazolium (NBT), and 1 mL 2 ML-malate pH 7.0. For LDH, 45 mL pH 8.0 staining buffer (0.2 M Tris-HCl), 10 mg NAD', 100 uL (1.25 mg) PMS, 2 mL NBT, and 110 uL lactic acid were mixed. The gel was placed in the staining solution at room temperature until all the bands appeared; this usually required about 10 minutes. At this point, the staining solution was replaced with dd HO and the gel was washed twice for 10 minutes. AFLP The main steps of AFLP were isolation of genomic DNA, double restriction digest, ligation of adapters, preamplification, selective amplification and labeling, resolution on a polyacrylamide gel, and visualization through autoradiography (Fig. 5). The protocol was based upon the methods described by Vos, et al. (1995), and many of the modifications were made upon suggestion by Tony DeTomasso. Our main DNA extraction method was Macherey-Nagel Nucleospin C& T DNA extraction from cells and tissue (see Macherey-Nagel for protocol). We also used a phenol and 1:24 chloroform/isoamyl extraction method and a salt extraction method (courtesy of Dr. Grant Pogson). All of these methods produced comparable results. The double restriction digest was set up by incubating 500 uM DNA, 5 units EcoRl, and 5 units Msel with 3 uL New England Biological (NEB) Buffer 2 and 3 uL IOX NEB BSA. An appropriate amount of dd HO was added to make a final reaction volume of 30 uL. Adapters (Vos et al., 1995) were ligated by incubating the double digest product with the following reagents at room temperature overnight: 2 uL 20 mM ATP, 4 uL 10X Boehringer-Mannheim ligation buffer, 1 uL 50 uM Msel adapter, 1 uL 5 MM EcoRl adapter, 1 uL Boehringer-Mannheim ligase, and 1 uL dd H2O. The single strands of the adapters should be annealed together before use. This is done by heating the appropriate concentration of adapter to 95° C for five minutes and letting the mixture cool to room temperature before freezing it. I made the preamplification reaction by first diluting the ligation product 2:3. 60 uL of O.IXTE (10 mM Tris-HCl and 0.1 mM EDTA, pH 7.5) were added to each ligation product to make a total solution of 100 uL. Each preamplification reaction was then made by mixing 2 uL 1OX PCR buffer, 2 uL 2 mM dNTP's, 0.5 uL 30 uM Msel primer, 0.5 uL 30 uM EcoRI primer, 4 uL 2:3 diluted ligation reaction solution, 0.09 uL Taq polymerase, and 10.91 uL dd H2O. The random bases for the Msel and EcoRl primers were thymine and adenine, respectively. Before the PCR cycles were started, the samples were heated at 72° C for two minutes. The PCR cycles were as follows: 94° C for 45 seconds, 56° C for 45 seconds, and 72° C for 1 minute, all for twenty cycles. Al mL solution of the lOX PCR buffer consists of 670 uL 1 M Trizma Base, 67 uL 1 M MgCh, 83 uL 2 M(NH2)SO4, 7 uL 14 M B-mercaptoethanol, and 173 uL dd HO NaOH was then added to adjust the pH to 8.6. When the preamplification cycles were complete, I ran 8 uL of the preamplification solution on a 1% agarose gel to check the efficacy of the preamplification step. The remaining 12 uL of the preamplification solution were diluted 1:20 with O.IXTE. 3 uL of this diluted preamplification solution were mixed with 1 uL IOX PCR buffer, 1 uL 2 mM dNTP's, 0.25 uL 10 uM labeled EcoRl primer (random bases AGG or ACT), 0.25 UL 30 uM Msel primer (random bases TGA or TAC), 0.05 uL Taq, and 4.45 uL dd H2O to give a total reaction volume of 10 uL. We prepared the reactions on ice, spun them for a few seconds in a microfuge, transferred the samples back to ice, and then transferred the samples directly into a PCR machine preheated to 95° C. The 36 PCR cycles were as indicated in Table 1. The labeled EcoRl primers were prepared beforehand. For a 25 uL sample, 11.25 uL dd HO, 2.5 uL 100 uM primer stock, 2.5 uL polynucleotide kinase buffer, 7.5 uLy 5P ATP, and 1.25 uL T4 polynucleotide kinase (9.30 U/uL) were added together. The solution was incubated at 37° C for 30 minutes and then at 65° C for 15 minutes. We then stored the labeled primers in a lead pig in a freezer designated for radioactivity. Aster the selective amplification and labeling step was completed, the samples were prepared for loading. We added 10 uL loading dye to each sample and denatured the samples at 95° C for five minutes. We then transferred the samples to ice and loaded 5 ul of each sample into the gel. The gel was run at 55 W until the first dye front ran completely off the gel, which took two to three hours. The dried gel was visualized by autoradiography. RESULTS Isozyme Electrophoresis On the gels that were stained for LDH, all of the individuals from Hayward, San Diego and San Felipe showed the same simple two-band pattern. The lack of variation between the San Diego and San Felipe populations is consistent with Fields's (1995) results. Figure 11, with one individual from San Felipe and eight from San Diego shows the banding pattern that was characteristic for all the individuals that were stained for LDH. All of the individuals from Hayward and San Felipe also showed the same banding when stained for MDH. The slowest band is the darkest, and there are two lighter bands below the darkest one (Fig. 2 and 3). The individuals from San Diego, however, show variation. As can be seen in Figure 14, the San Diego individuals show variation in the fastest band. The individuals show three variations, which are evident in the individuals labeled SDI, SD2, and SD3. SDI shows the same banding patterns as the Hayward and San Felipe fish. SD2, however, shows three bands instead of one at the fastest position. SD3 shows one band at the fastest position, but it has migrated farther than it did than in the Hayward and San Felipe individuals. SDI is therefore homozygous for the slower allele, SD2 is heterozygous, and SD3 is homozygous for the faster allele. Table 2 shows the number of individuals with each genotype for the populations. 10 individuals from Hayward, 30 individuals from San Diego, and 8 individuals from San Felipe were analyzed. The San Diego population was tested for Hardy-Weinberg Equilibrium (Appendix 1). It was the only population tested since it was the only one that exhibited variation. The population was almost exactly in Hardy-Weinberg Equilibrium, with a insignificant X¬ test statistic of 0.1391 (critical value was 3.841) and expected values that matched the observed values almost exactly. (18 homozygous slow, 10 heterozygous, and 2 homozygous fast were observed, while 17.8, 10.6, and 1.6 were the expected values.) AFLP Figure 9 shows a gel obtained when we were trying to optimize the AFLP conditions. All the lanes are from the same sample, but each lane reflects different reaction conditions. Each lane shows the same bands, although some bands are lighter in some lanes than others. Figure 10 shows a gel on which there are several different individuals. H2 and H3 are individuals from Hayward while SD14, SD17, and SD19 are from San Diego. The two Hayward fish have bands in common with each other, as do the three San Diego individuals. The Hayward and San Diego fish also have bands in common with each other, yet there are also several distinct differences between the banding patterns. One difference is the bands indicated on the San Diego individuals which are lacking in the Hayward fish. DISCUSSION Isozyme electrophoresis All of the LDH gels showed a simple two-band pattern. The lighter band on top is most likely the heart type homotetramer and the dark band is most likely due to the muscle isoform. MDH is a dimeric protein and is coded for by three loci: one mitochondrial locus and two cytosolic loci. Hayward individuals, San Felipe individuals, and some San Diego individuals exhibit a simple three-band pattern, one band for each isoform. The darkest band on top is due to the mitochondrial isoform of MDH, and the two lighter bands below are due to the two cytosolic isoforms of MDH. The second, faster cytosolic isoform shows allelic variation in the San Diego population, with some individuals homozygous for the slow allele, some heterozygous, and some homozygous for the fast allele. The San Diego population did not significantly deviate from Hardy-Weinberg Equilibrium expectations. This means that the assumptions underlying the Hardy¬ Weinberg Equilibrium, namely,: random mating between phenotypes, large population, no selection between genotypes, no differential migration, and no mutation are generally being met. This suggests that MDH is not under selection in San Diego. If it were being selected upon, it could not be used as a marker for population structure and dynamics because it would not be neutral. The Hayward and San Felipe populations are fixed for the slow allele, suggesting that there is very little or no gene flow among the three populations. The rationale behind this conclusion is as follows: since the fast allele is not present in either the Hayward or San Felipe populations, neither of these populations is interbreeding with the San Diego population. Since San Diego is located south of Hayward and north of San Felipe, it is highly unlikely those two populations would interbreed with each other if they were not interbreeding with the San Diego population, due to the geographic distance separating them. It should be noted that my sample sizes of 10, 30, and 8 for the Hayward, San Diego, and San Felipe populations, respectively, are relatively small. It is possible that I did not detect the fast allele in the Hayward and San Felipe populations simply because of chance. However, Dr. Jen-jen Lin has also analyzed both San Diego and San Felipe Gillichthys for MDH. Of the fisty individuals from San Felipe she analyzed, all had the same simple pattern. The MDH results are somewhat unexpected if one were looking for a temperature- related cline; the samples in the coldest and warmest environments show no variation from each other while the samples from the intermediate environment show variation from both. There are several possible theories that could explain the presence of the fast allele in San Diego only. One possibility is that the slow allele is actually the ancestral form and the mutation for the fast allele is one that has occurred in the San Diego area only. Another possibility is that the frequency of the fast allele is actually increasing among Gillichthys farther south down the Pacific coast, but that I cannot see this effect because I did not have any samples from the Pacific side of Baja California. If this were case, the absence of the fast allele in San Felipe can be explained either by founder effect or by random genetic drist. Founder effect means that the few fish that journeyed (geographic movement would most likely be from oceanic dispersal of larvae rather than the swimming of adult fish) around the tip of Baja California into the Sea of Cortez to found the population happened to all be homozygous for the slow allele. The individuals from San Felipe would mostly be descended from the same founding fish and therefore are all homozygous for the slow allele. For random genetic drift to cause the homozygosity of the slow allele in San Felipe, the few fish that journeyed around the tip of Baja California into the Sea of Cortez had both the slow and fast allele. Because the small population size, these fish would not be in Hardy-Weinberg Equilibrium, and the genotype frequencies would not necessarily remain stable over time. By chance, the frequency of the fast allele could have been reduced to nothing or almost nothing. From that point on, all the fish in the San Felipe area would be homozygous for the fast allele (Hartl, et al., 1989). Although the analysis of the MDH results suggest that there is low gene flow among the three populations, the study is far from complete. Larger sample sizes, and samples from more locations, particularly the Pacific coast of Baja California would be needed before a firmer conclusion about the pattern of MDH genotypes among Gillichthys can be drawn. In addition, analysis of many more protein loci would also be needed to draw definitive conclusions about the population structure and dynamics of Gillichthys. AFLP The data from the AFLP technique were not sufficient to permit a firm conclusion to be drawn about population structure. As shown in Figure 10, however, there were some differences between the Hayward and San Diego individuals. Although these results are preliminary, they do show promise that AFLP can potentially find differences between samples of Gillichthys. Figure 9 also shows promise, not in terms of population differences but in terms of the reliability of AFLP. Figure 9 shows several lanes run with the same individual, with each lane reflecting slightly different reaction conditions. That the lanes show the same banding patterns, with only differences in the intensities of the bands, suggests that AFLE was working as planned; that the fragments being amplified are the ones we want to amplify, and are not just random fragments. If random fragments were being amplified, the fragments and therefore the bands would not be consistent over several trials. It should be noted that although we did produce promising AFLP results, numerous modifications and optimization efforts were required before consistent results were obtained. The modifications we made are included in the materials and methods, but several which we found to be particularly important involved: the amount of DNA used, the temperature of the ligation reaction, the concentration of primer, and the preparation of the selective amplification and labeling step PCR reaction on ice. Although technically the amount of restriction enzyme should determine how many fragments are cut, we found that adjusting the DNA concentrations affected the results. 500 uM gave the best results for Gillichthys, but for my colleague Maya Hayden, who was using the AFLP technique on the squid Loligo opalescens, 300 uM DNA gave the best results. It is important that the concentration of DNA not be too high, because an excess of DNA can inhibit AFLP by remaining uncut and thereby interfering with the ligation reaction. Decreasing the ligation temperature from 37° C for three hours to room temperature overnight may also have improved our results. This modification produced a less harsh environment which may have increased the likelihood of the proper DNA fragments, adapters, and enzymes coming into contact. It probably would also decrease the likelihood of the adapters denaturing. We added more Msel primer than EcoRl primer in the selective amplification and labeling step for the same reasons that more Msel adapter is added than EcoRl adapter Because Msel cuts much more frequently, there are many more Msel-Msel cut fragments. Supplying more Msel primer ensured that there were adequate Msel primers for both the many Msel-Msel cut fragments and the less abundant Msel-EcoRl cut fragments. The PCR reaction for the selective amplification and labeling step was prepared on ice. This was important to protect the Taq from denaturing and to prevent unplanned reactions from occurring. In summary, AFLP results may are consistent with a difference between the Hayward and San Diego populations. AFLP was able to provide consistency within an individual and to show variation between individuals, which suggests that it can be a powerful technique. A downside of AFLP, however, is that it may require a lot of modification before consistent and reliable results are obtained. CONCLUSIONS MDH results, which show the presence of a fast allele of the second cytosolic isoform exclusively in San Diego individuals, provide some evidence that there is very little or no gene flow between the Hayward, San Diego, and San Felipe populations of Gillichthys mirabilis. The presence of the fast allele in San Diego may be explained by several possible theories. Preliminary AFLP results, which suggest variation between the Hayward and San Diego populations, do not disprove the no gene flow hypothesis. However, more in-depth studies that include larger sample sizes, samples from more locations, and analyses of more protein loci are required before definitive conclusions about the population structure and dynamics of Gillichthys mirabilis can be drawn. Finally, the ability of AFLP to provide consistency within individuals and show variation between individuals suggests that this method has the potential to be a powerful DNA fingerprinting technique. The original protocol, however, may require some modification to make it suitable for the particular species being studied. ACKNOWLEDGMENTS I received help from numerous people throughout this project and would like to thank some of them here. Thanks to my advisors Dennis Powers and George Somero for initiating the project and for their support throughout the quarter. A huge thanks to my mentor Gary Villa for his patience, help, and support. Thanks to Jen-jen Lin, who taught me the isozyme electrophoresis techniques and to Tzung Yang, who took me Gillichthys- trapping. Trish Schulte and the other members of the Powers lab provided a lot of input and support; thank you all. Thanks also to Tony DeTomaso for his help with AFLP and to Grant Pogson for his DNA Salt Extraction protocol. Finally, I would like to thank Maya Hayden, who was my partner in developing the AFLP technique, for her company and support. LITERATURE CITED Barlow, G.W. (1961). Intra- and interspecific differences in rate of oxygen consumption in gobiid fishes of the genus Gillichthys. Biol. Bull. 121: 209-229. Fields, P.A. (1995). Adaptation to environmental temperature in two genera of coastal fishes, Paralabrax and Gillichthys. PhD thesis, University of California, San Diego. Hartl, D.L., and Clark, A.G. (1989). Random Genetic Drist. In Principles of Population Genetics, Second Edition. Sunderland, MA: Sinauer Associates. Lin, J.-J. (1993). Thermal adaptation of cytoplasmic malate dehydrogenases of teleost fishes. PhD thesis, University of California, San Diego. Love, R.M. (1991). Longjaw Mudsucker (Gillichthys mirabilis). In Probably more than you want to know about the fishes of the Pacific Coast, pp. 167-168. Santa Barbara, CA: Really Big Press. Murphy, R.W., Sites, Jr., J.W., Buth, D.G., and Haufler, C.H. (1996). Proteins: Isozyme Electrophoresis. In Molecular Systematics, 2nd Edition (ed D.M. Hillis, C. Moritz, and B.K. Mable), pp. 51-120. Sunderland, MA: Sinauer Associates. Richardson, B.J., Baverstock, P.R., and Adams, M. (1986). Electrophoresis. In Allozyme Electrophoresis, pp. 15-30. San Diego, CA: Academic Press. Vos, P., Hogers, R., Bleeker, M., Reijans, M., van de Lee, T., Hornes, M., Frijters, A., Pot, J., Peleman, J., Kuiper, M., and Zabeau, M. (1995). AFLP: a new technique for DNA fingerprinting. Nucleic Acids Research. 23, 4407-4414. APPENDIX 1 The Hardy-Weinberg Equilibrium Distribution model is a method of interpreting the population genetics of a sample. If a sample is in Hardy-Weinberg Equilibrium, it is believed to meet the following assumptions: 1) There is random mating between phenotypes 2) The population is large 3) There is no selection between genotypes 4) There is no differential immigration or emigration of genotypes 5) There is no mutation. In Hardy-Weinberg equilibrium, the genotypes will maintain a stable equilibrium with genotype frequencies of AA = p’, Aa = 2pq, and aa = q', where A and a are alleles, p is the gene frequency of A calculated from observed values, and q is the gene frequency of a also calculated from observed values (Murphy, et al., 1996). The test for Hardy-Weinberg equilibrium involves comparing the expected number of genotypes, which are calculated from the values of p", 2pq, and qu', with the observed numbers using the X2- test (Richardson, et al., 1986). The calculations for testing the San Diego population for Hardy-Weinberg Equilibrium were as follows: ÖBSERVED VALUES: 18 Homozygous Slow, 10 Heterozygous, 2 Homozygous Fast p (slow) = (2 x18 + 10)/2n = 0.77 q (fast) = 1 - 0.77 = 0.23 p==0.593, 2pq =0.354, q’=0.053 EXPECTED VALUES = (expected genotype frequency) x (+ of individuals screened) 17.8 Homozygous Slow, 10.6 Heterozygous, 1.6 Homozygous Fast X* Value = L (Observed-Expected)'/Expected = 0.1391 Critical value (o5, 1) = 3.841 0.1391 is less than 3.841, so the X* Value is not significant. Therefore, the population is in Hardy-Weinberg Equilibrium If the population was not in Hardy-Weinberg equilibrium, this would indicate that the variation has a non-genetic basis or that one or more of the assumptions are not being met Table 1-PCR cycles for the selective amplification and labeling step Duration Temp Duration 4of Duration Temp Temp cycles 1 min 45 sec 72°C 45 sec 650C 940 C 45 sec 72°C 1 min 64°C 940C 45 sec 45 sec 1 min 45 sec 630C 72°C 94° C 1 min 45 sec 72° C 45 sec 62°C 94°C 72°C 1 min 45 sec 45 sec 61°C 94° C 72°C 45 sec 1 min 45 sec 60°C 94° C 45 sec 720C 1 min 45 sec 590 C 94° C 45 sec 45 sec 72°C 1 min 580C 1 94°C 72°C 45 sec 1 min 45 sec 570C 94°C 1 min 45 sec 72°C 27 45 sec 560C 94° C Table 2 - The number of individuals with each genotype. Homozygote Heterozygote Slow 10 Hayward 10 18 San Diego 8 San Felipe Homozygote Fast 2 gaa k L 2 4 — — — — M n Ldn Figure 2 Hayward L San Diego San Felipe Figure 3 Extraction of Protein Non-Denaturing Polyacrylamide Gel Electrophoresis Staining for Activity of the Protein Figure 4 A AjA Monomer A2 E A Dimer A AjA2 AA A2 E A2A2 Figure 5 Isolation of Genomic DNA Double Restriction Digest Ligation of Adapters Preamplification Selective Amplification and Labeling Polyacrylamide Gel Electrophoresis Autoradiography 2 2 1 15 i 5 1 + 1 2 0 L 1 — E 1 Figure 9 . SD8 Figure 10 .. . W . . . . . H2 H3 SD SD SD 14 17 19 Bands that are present in the San Diego individuals but not the Hayward individuals Figure 13 FIGURE LEGENDS Figure 1 - Distribution of Gillichthys mirabilis: the coast of California and Baja California, and the Sea of Cortez. Figure 2-Samples were obtained from Northern California, Southern California, and Sea of Cortez locations. Figure 3 - The major steps of isozyme electrophoresis. Figure 4 -Possible gel patterns of monomeric and dimeric proteins. Figure 5 - The major steps of AFLP. Figure 6-After the double restriction digest, adapters are ligated to the ends of the fragments. Figure 7 -Preamplification involves the denaturing of double-stranded DNA, selective annealing of primers, and amplification. Figure 8 -Primers used in the selective amplification labeling step have three random bases to increase selectivity. EcoRI primers are labeled with radioactivity. Figure 9-The lanes indicated are the same sample under different reaction conditions. The lanes show identical banding patterns, with differences in the intensity of bands. Figure 10 - There are differences between the banding patterns of Hayward individuals (H2 and H3) and San Diego individuals (SD 14, SD17, and SD19). Figure 11 - LDH gel with the simple, two-band pattern. These individuals are from San Felipe and San Diego; however, all the individuals sampled from the three location exhibited this same pattern. Figure 12 -MDH gel with Hayward samples. Each individual has the same, three-band pattern. Figure 13 - MDH gel with San Felipe samples. As with the Hayward samples, each individual has the same, three-band pattern. Figure 14 - MDH gel with San Felipe and San Diego individuals. The San Diego individuals show variation at the second cytosolic isoform. SDI is homozygous for the slow allele, SD2 is heterozygous, and SD3 is homozygous for the fast allele.