Gene Duplication in an African Cichlid Adaptive Radiation Heather E. Machado, Ginger Jui, Domino A. Joyce, Christian R.L. Reilly, David H. Lunt, and Suzy C.P. Renn Heather E. Machado Department of Biology, Reed College, Portland OR 97202, USA Ginger Jui Department of Plant and Microbial Biology, University of California, Berkeley California 94720-3102, USA Domino A. Joyce Department of Biological Sciences, University of Hull, Hull HU6 7RX, UK Christian R.L. Reilly Santa Catalina School, Monterey, CA 93940, USA David H. Lunt Department of Biological Sciences, University of Hull, Hull HU6 7RX, UK Dr. Suzy C.P. Renn Department of Biology, Reed College, Portland OR 97202, USA Email contact: HM: machadoheather@gmail.com GJ: ginger.jui@gmail.com DJ: d.joyce@hull.ac.uk CR: Christian_Reilly@santacatalina.org DL: d.h.lunt@hull.ac.uk SR : renns@reed.edu

THIS IS A DRAFT PLEASE DO NOT DISTRIBUTE

1

Abstract 1

Background: Gene duplication is an important source of evolutionary innovation and can result 2 in phenotypic novelty and the divergence of lineages. We present the first large-scale study of 3 gene duplication in African cichlid fishes, identifying gene duplicates in three species belonging 4 to the Lake Malawi adaptive radiations (Metriaclima estherae, Protomelas similis, 5 Rhamphochromis “chilingali”) and one closely related species from a non-radiated lineage 6 (Astatotilapia tweddlei). 7

Results: Microarray comparative genomic hybridization reveals 134 duplicated genes among the 8 four cichlid species tested. A similar number of duplicated genes were identified for the three 9 Lake Malawi radiated lineages (M. estherae: 54, P. similis: 51, R. “chilingali”: 55), representing 10 a 38% – 49% increase relative to the non-radiated lineage (A. tweddlei: 37). Approximately 20% 11 of the duplicated genes are shared among the 3 radiated lineages. Duplicated genes include 12 several that are involved in immune response, ATP metabolism and detoxification. 13

Conclusions: These results support the hypothesis that gene duplication is a characteristic of the 14 Lake Malawi cichlid adaptive radiations and provide candidate genes for the analysis of 15 functional relevance. Comparative sequence analysis of gene duplicates can address the role of 16 positive selection and adaptive evolution by gene duplication, while further study across the 17 phylogenetic range of cichlid radiations will determine whether the patterns of gene duplication 18 seen in this study consistently accompany rapid radiation. 19

Background 20

Adaptive radiation, the evolution of genetic and ecological diversity leading to species 21 proliferation in a lineage, is thought to be the result of divergent selection for resource 22 specialization (Dobzhansky 1937; Mayr 1963; Schluter 2000). Differential selection in 23 heterogeneous environments can result in adaptive radiation when there is a genetic basis for 24 variability in organisms’ success in exploiting alternative resources (Dobzhansky 1937; Mayr 25 1963; Smith 1966; Slatkin 1980; Schluter 2000). Examples of such radiations include the 26 Cambrian explosion of metazoans (Gould 1989), the diversification of Darwin’s finches in the 27 Galapagos (Darwin 1859), variations in amphipods and cottoid fish in Lake Baikal (Fryer 1990), 28 the Caribbean anoles (Losos et al. 1998), the Hawaiian Silversword (Baldwin and Sanderson 29 1998) and the explosive speciation of the cichlid fishes in the African Great Lakes (Fryer and 30 Iles 1972). 31

The cichlid fishes are the product of an incredible series of adaptive radiations in response to the 32 local physical, biological and social environment. While cichlids can be found on several 33 continents (Farias, Orti, and Meyer 2000), the most dramatic radiations are those of the 34 haplochromine cichlids in the great lakes of East Africa. This speciose clade exhibits 35 unprecedented diversity in morphological and behavioral characteristics (Kocher 2004) and 36 accounts for ~10% of the world’s teleost fish. Interestingly, this clade also includes lineages that 37 have remained in a riverine environment and have not radiated (Seehausen 2006). The ability to 38 compare closely related cichlid lineages that have and have not undergone an evolutionary 39 radiation provides a critical tool for identifying genomic features that promote, or correlate with, 40 adaptive radiation. 41

42

Current genomic research (e.g. primates: Fortna et al. 2004; Marques-Bonet et al. 2009) supports 43 the classic work of Ohno (1970) that proposed a prominent role for gene duplication in 44 evolutionary expansion. Duplication makes extra gene copies available for dosage effects, 45 subfunctionalization, or neofunctionaliztion (Force et al. 1999), with the resultant phenotype 46 potentially contributing to an organism’s survival, adaptation and overall fitness (for review see 47 Taylor and Raes 2004). The recent and rapid speciation of the haplochromines allows the 48 possibility to identify those gene duplications that are associated with radiated lineages. 49 50 We used array-based comparative genomic hybridization (aCGH) to identify gene duplications 51 among 5689 genes for three Lake Malawi species from lineages that have undergone radiation 52 (Metriaclima estherae, Protomelas similis, Rhamphochromis “chilingali”) and one closely 53 related riverine species from a non-radiating lineage (Astatotilapia tweddlei) (Figure 1). This is 54 the first large-scale study of gene duplication among species of the African cichlid radiations. 55 56 Results 57 aCGH identification of duplicated genes 58 59 A total of 5689 of the sequenced microarray features passed quality control measures in all four 60 test species. Among these, 145 array features (representing 134 genes) were determined to have 61

an increased genomic content (i.e. copy number) for one or more heterologous species relative to 62 A. burtoni (P < 0.1 FDR corrected) (Tables 1, 2). This included duplications of 54 genes in M. 63 estherae, 51 in P. similis, and 55 in R. “chilingali”, whereas the non-radiated lineage, A. 64 tweddlei, contributed only 37 duplicated genes (Figure 2). The similar number of duplicated 65 genes identified for the three radiated lineages represents a 38% – 49% increase relative to the 66 number of duplicated genes identified in A. tweddlei. Shared duplications were prevalent among 67 the three Lake Malawi species, with 11 duplications shared among all three and 16 duplications 68 shared between two of the three (Figure 2). Five genes had greater genomic content in all four 69 species relative to A. burtoni. BLAST comparison of array feature sequence similarity to the 70 nucleotide database allows annotation and predicted function for discussion of possible adaptive 71 duplicates (see discussion below). 72 73 Gene Ontology Analysis 74 Of the 134 genes identified as duplicated in any of the test species, 38 could be annotated with 75 GO terms, inferred by sequence similarity. Two terms are over-represented. Among molecular 76 functions, nucleoside-triphosphatase activity is over-represented (P = 0.021). Of the four 77 duplicated genes in this category, one is duplicated in all test species, two are duplicated in the 78 three radiated lineages, and one is duplicated only in P. similis. Among biological processes, 79 regulation of cell proliferation is over-represented, with two genes found duplicated, one in R. 80 “chilingali” and one in M. estherae (P = 0.008). 81 82 Quantitative PCR verification 83 Four loci found to be duplicated in various combinations of radiated and non-radiated cichlid 84 lineages were validated by quantitative PCR (qPCR) (Table 2). For each locus, all species 85 showing significant duplication in the microarray analysis also showed significantly higher 86 genomic content than A. burtoni according to the qPCR analysis (Figure 3). 87 88 Discussion 89 Gene duplication is an important source of functional novelty and has a demonstrated role in 90 adaptive evolution (Taylor and Raes 2004). Such adaptations can allow for niche diversification, 91 as has been suggested for thermal adaptation (Antarctic ice fish: Chen et al. 2008; plants: Sandve 92 et al. 2008) and for metabolic novelty (C-4 photosynthesis: Monson 2003). The adaptive 93 radiations of the African cichlid fishes exhibit remarkable niche exploitation in the presence of 94 low levels of sequence divergence. However, little is known regarding the relative number of 95 duplicated genes, nor the identity of duplicated genes, within this group. If there is an increased 96 rate of gene duplication or gene duplicate retention in radiated lineages, or if particular 97 duplications are associated with radiated lineages, then their pattern and identity could provide 98 insight into the processes producing the rapid expansion of the African cichlids. The patterns 99 reported and validated here indicate shared and increased gene duplication in Lake Malawi 100 radiated lineages. Importantly, the array results are corroborated by independent assays using 101 quantitative PCR. Several candidate genes, including those that are involved in immune 102 response, ATP metabolism and detoxification, are found to be duplicated in and among lineages 103 (Table 1). Some of these gene duplications may underlie adaptive phenotypes and may be 104 integral to the process of speciation. 105 106

The evolution of immune response is a potent factor contributing to the divergence of lineages, 107 resulting from strong selection on certain loci (Sackton et al. 2007; Lazzaro and Little 2009; 108 Barreiro and Quintana-Murci 2010). Several genes associated with immune response are found 109 to be duplicated in the Lake Malawi species, including two finTRIM genes (one duplicated in P. 110 similis and the other in P. similis and R. “chilingali”). This gene family is known to play a role in 111 immunity against viral infection, and several finTRIM paralogs have been found in teleost fishes, 112 resulting from duplication and positive selection (70 in trout, 84 in zebrafish) (van der Aa et al. 113 2009). Five major histocompatibility complex (MHC) genes- two MHC class I, two MHC class 114 II, and kinesin-like protein 2- are also found duplicated in one or more of the species from the 115 radiated lineage. The MHC gene family, in addition to being involved in immunity (salmon: 116 Lukacs et al. 2007), has a history of expansion and contraction through duplication and deletion 117 (Miller, Kaukinen, and Schulze 2002). MHC gene families vary in size among teleosts, with 118 particularly large families in cichlids (Malaga-Trillo et al. 1998; Miller and Withler 1998; Sato et 119 al. 1998; Persson, Stet, and Pilstrom 1999). Additional immune related genes duplicated in the 120 Lake Malawi radiations include an immunoglobulin light chain, small inducible cytokine 121 (associated with the MHC region in stickleback: Reusch, Schaschl, and Wegner 2004), and 122 sestrin 3. In A. tweddlei, the test species from the non-radiated lineage, two immune genes, 123 kallikrein-8 and natural killer cell lecin-type receptor, are also found to be duplicated. The 124 identification of several duplicated immune function genes is consistent with previous work 125 documenting size variability and rapid expansion of immune function gene families (Drosophila: 126 Sackton et al. 2007; Bombyx mori: Tanaka et al. 2008) which may allow species to invade new 127 niches. 128 129 ATP metabolism and function is critical to many physiological processes. Two ATP synthases 130 and one ATP transporter are found duplicated among the four species. Subunits G and E of 131 vacuolar ATPases, which couple the energy of ATP hydrolysis to proton transport across 132 intracellular and plasma membranes, are duplicated in A. tweddlei and M. estherae, respectively. 133 In R. “chilingali”, the adenine nucleotide translocator (ANT) s598 is found duplicated. This 134 mitochondrial transmembrane protein is the most abundant mitochondrial protein and is integral 135 in the exchange of ADP and ATP between the mitochondria and the cytoplasm. Increased 136 expression of mitochondrial ATP synthase has been found in cold acclimated carp (Kikuchi, Itoi, 137 and Watabe 1999) and ANT genes are being studied for their potential adaptive role in thermal 138 acclimation (Itoi et al. 2005). The ATP synthase and transport genes found duplicated in this 139 study could also be associated with acclimation to ecological variation in Lake Malawi or could 140 be associated with other differential metabolic demands. 141 142 Selection on detoxification genes (those involved in the breakdown of toxic compounds) can 143 determine survival in particular environments or can contribute to expansion into new niches. 144 One example is seen in plant-herbivore interactions, where the ability of herbivores to detoxify 145 plant defense compounds can prevent exclusion of the herbivore from that food source. Gene 146 duplication has been implicated as a key process in this adaptive response (Wen et al. 2006; 147 Fischer et al. 2008). We detect duplication of detoxification genes in all three species from the 148 radiated lineage. The sulfotransferase (SULT) gene cytosolic sulfotransferase 3 is found 149 duplicated in P. similis and R. chilingali. SULT genes are detoxifying enzymes that catalyze the 150 transfer sulfonate groups to endogenous compounds and xenobiotics. Once sulfated, compounds 151 may become more easily excreted from the body. In zebrafish, ten SULT proteins have been 152

cloned, two of which show strong activity towards environmental estrogens (Liu et al. 2008). 153 Zebrafish SULTs have also been found to act on other xenobiotics (Sugahara et al. 2003). In 154 Atlantic cod, a SULT gene was found to be upregulated in response to polluted water (Lie et al. 155 2009). Two other genes involved in detoxification, arsenic methyltransferase and ferritin (heavy 156 subunit), are found duplicated in R. “chilingali”. Arsenic methyltransferase converts inorganic 157 arsenic into less harmful methylated species, and ferritin is an iron storage protein that is 158 essential for iron homeostasis, keeping iron concentrations at non-toxic levels. Another iron-159 related protein, the iron-sulfur cluster assembly enzyme, was also duplicated in R. “chilingali”. 160 It is possible that some of these gene duplicates have been retained due to a selective advantage 161 for greater degradation of environmental toxins. 162 163 The existence of gene families indicates that there is a propensity for duplication and duplicate 164 retention of certain genes. One study assigned 38% of known human genes to a gene family, 165 based on amino acid sequence similarity (Li et al. 2001). These gene families typically consisted 166 of two genes, with the largest gene family having 139 members. In the present study, several of 167 the genes found to be duplicated were members of large gene families. These include 40S and 168 60S ribosomal proteins (duplicated in R. “chilingali” and M. estherae), claudin 29a (M. 169 estherae), GTPase IMAP family member 7 (P. similis), C-type lectin domain family 4 (M. 170 estherae), high-mobility group 20B (HMG20B) from HMG-box superfamily (A. tweddlei), and 171 hox gene cluster genes (all species). Hox genes are important in the regulation of development, 172 and have been found to be associated with differential jaw development in cichlid fishes (le 173 Pabic, Stellwag, and Scemama 2009). An immunoglobulin light chain gene came from the 174 largest gene family represented, and was duplicated in P. similis. Since large gene families are 175 comprised of multiple paralogs and may possess a greater tendency for expansion, it is not 176 surprising that large gene families are well represented in our list of duplicated regions. 177 178 One important caveat to this study is that genomic content for each gene is assessed relative to 179 the array platform species A. burtoni. We identify five genes, two annotated as Hox gene cluster 180 genes, one as a Ras-related C3 botulinum toxin substrate gene and two that lack annotation, that 181 appear to be duplicated in all four test species. Alternatively, the increased genomic content in 182 the test species could be due to a deletion in A. burtoni. For example, if multiple paralogs existed 183 for a given gene, a deletion of one or more of the paralogs in A. burtoni would also result in 184 increased genomic content in the test species. 185 186 Even when gene names are not intuitively informative, analysis according to Gene Ontology 187 (GO) annotation (Ashburner et al. 2000) can generate functional hypotheses. Despite the 188 ascertainment bias inherent to GO terms, which is largely due to the nature of hypothesis-driven 189 research conducted in model organisms, this approach allows rigorous statistical analysis for 190 over- and under-representation of particular molecular functions, biological processes, and 191 cellular components. Also, applying GO terms reduces the complexity of the data, avoids 192 experimenter bias in cross-referencing between experiments and species, and facilitates 193 comparisons of experimental results obtained in different organisms and/or with different 194 platforms. While few terms were found to be overrepresented in this particular dataset 195 (nucleoside triphosphatase activity and cell proliferation) we advocate for the use of such 196 annotation schemes due to their comparative power across experiments. 197 198

In this study, the species from the radiated lineage have been selected for their peculiar and 199 compelling demographic history of rapid evolutionary radiation and adaptation as compared to 200 the riverine taxa. We find a pattern of increased gene duplication in these species, with 38-49% 201 more genes duplicated than in the non-radiated lineage. Care must be taken in interpreting this 202 increase in the context of adaptive radiation, with three primary considerations. First, only a 203 subset of genes (i.e. those present on the array with available sequence) were tested. Second, we 204 cannot rule out the possibility that gene duplicates may have become fixed in shared ancestral 205 populations due to neutral processes. Founder events, genetic bottlenecks or drift during the 206 relatively recent evolutionary past, 4-9 MY (Kornfield and Smith 2000), or 0.7-2 MY (Coyne 207 and Orr 2004), may account for the shared duplications that were observed. Sequence data from 208 multiple species will be necessary to distinguish neutral vs. adaptive evolutionary processes. 209 Third, due to the shared evolutionary history of the radiated lineages, the three species from 210 radiated lineages cannot be considered independent, as such the tantalizing results of our single 211 comparison of radiated versus non-radiated requires further support before general patterns can 212 be rigorously discussed. 213 214 Conclusions 215 Only recently have studies begun to examine the patterns of gene duplication and copy number 216 polymorphism across species in natural systems, beyond primates (e.g. Dopman and Hartl 2007; 217 Chen et al. 2008). We present the largest analysis thus far of patterns of gene duplication across 218 lineages of the African cichlid radiations. We identify several genes that are potentially 219 duplicated among the four species tested, and find a pattern of shared and increased gene 220 duplication in the Lake Malawi radiated lineage. While our inference regarding the adaptive 221 value of candidate gene duplicates must be tempered, the results of this initial study support the 222 generally accepted hypothesis that gene duplication provides one avenue to the evolution of 223 adaptive phenotypes. Assessment across a greater phylogenetic range of cichlid radiations will 224 identify consistent patterns of gene duplication associated with radiated and non-radiated 225 lineages, and comparative sequence analysis will reveal the potential contribution of natural 226 selection to gene duplicate evolution. 227 228 Methods 229 Genomic DNA, extracted from ethanol-preserved field tissue samples by standard 230 ProteinaseK/Phenol protocol, was size reduced by Hydroshear (Genome Solutions/Digilab) to 1 231 – 5 Kb. DNA (4µg) and labeled with Alexa-Fluors conjugated dCTP by Klenow polymerization 232 (Invitrogen, Bio-Prime). Each species was hybridized twice (in dye swap) against a reference 233 pool of A. burtoni genomic DNA using the A. burtoni cDNA microarray (GEO platform 234 GPL6416). After a 16 hour hybridization (67.5°C, 3.4X SSC, 0.15% SDS, 1 mM DTT, Cot-235 1DNA), arrays were washed and scanned (Axon 4100B, Genepix). 236 237 Microarray data (GEO series GSE19368) were filtered by omitting features with a lack of 238 sequence information, known ribosomal content, or that had faint array signal (<2 SD above 239 background). Only features that survived this quality control for all eight microarrays were 240 analyzed. Data were corrected for background intensity (“minimum”) and were loess normalized 241 within array using 250 conserved features (Salzburger et al. 2008). This corrects for bias 242 introduced by sequence divergence under standard normalization (van Hijum et al. 2008). 243 Duplicated genes were identified as those with increased fluorescence according to the “lmFit” 244

statistical model with “eBayes” correction and FDR adjustment for P < 0.1 significance level 245 (Smyth 2004). The reported results are underestimates of duplication levels, due to the fact that 246 diverged duplicates are less likely to be detected (Machado and Renn in review). GEL50 247 measurements (Townsend 2004) indicated that experiments were of similar statistical power (M. 248 estherae: 1.05, P. similis: 0.76, R. “chilingali”: 0.65, A. tweddlei: 0.75). Gene Ontology analysis 249 conducted with BiNGO (Maere, Heymans, and Kuiper 2005) used the annotation of the top 250 BLAST hit (e-value < 10-14) to the GO sequence file (go_200806-seqdb.fasta). 251 252 Genomic content was validated for four genes using qPCR (Table 3). gDNA concentration was 253 quantified with 1.5X SYBR Green I (Roche Applied Science) on a Nanodrop 3300 254 (Thermosavant). Triplicate qPCR reactions (Opticon MJ Research) contained 0.75x SybrGreen, 255 1x Immomix (Biolabs), 200-500 nM primers and 0.2 ng sample DNA in 10 µl reactions (95 °C- 256 10 min; 35 cycles of: 94 °C- 2 min, 60 °C- 20 sec, 72 °C- 15 sec, and 2 min extension). Copy 257 number relative to A. burtoni was calculated as CT, the cycle number at a set threshold relative 258 to the A. burtoni standard curve, standardized to an A. burtoni copy number of 1. Primer 259 efficiency was calculated with a dilution series for A. burtoni DNA and one test species (supp. 260 table S2). 261 262 263 264 Authors' contributions 265 SCPR, DHL, DJ conceived of the project. HEM, CRLR, GJ performed the experiments. HEM 266 conducted the analyses. SCPR, HEM, DHL prepared the manuscript. 267 268 Acknowledgements 269 Funded by Murdock Charitable Life Trust and NSF-OIS 0818957. Thanks to Martin J Genner 270 for Rhamphochromis “chilingali” sample. 271

272

272

273 FIG 274 Figure 1. Phylogenetic positions of experimental (stars) and reference (circle) taxa. The 275 maximum likelihood tree is based on 1785 bp mitochondrial ND2. Nodes not supported by 50% 276 maximum likelihood SH values are collapsed.277

278

279 Figure 2. Identified gene duplicates (P < 0.1 FDR) 280 281

282

282 Figure 3. qPCR validates gene copy number determined by aCGH. Abbreviations are genus and 283 species initials. 284 ** P <0.1 FDR, * P <0.2 FDR found by array analysis 285 286 287 288

289

289 Table 1: Genes duplicated relative to A. burtoni with informative BLAST hits. 290 GenBank Homology A.twe M.est P.sim R.chi BitScore CN468828* Adenine nucleotide translocator s598 ns ns ns 0.60 567 DY630000 Alcohol dehydrogenase Class VI ns ns 0.73 ns 379 DY630424 Alkylated DNA repair protein alkB homolog 7 ns 0.43 ns ns 304 DY629046 Arsenic (+3 oxidation state) methyltransferase ns ns ns 1.06 150 DY626788 ATPase, H+ transporting, lysosomal V0 subunit E ns 0.76 ns ns 87.8 DY628437 Claudin 29a (cldn29a) gene ns 0.60 ns ns 526 DY632040 Coiled-coil domain containing protein 80 ns ns 1.19 2.13 434 DY629141 Crystallin gamma M2b ns ns ns 0.43 829 DY626204 C-type lectin domain family 4 member C ns 0.38 ns ns 246 DY631088 Cystatin-B 0.45 ns ns ns 150 DY630353 Cytosolic sulfotransferase 3 ns ns 0.62 0.64 713 CN470675 Dazl gene ns ns ns 0.57 89.7 DY629967* Ferritin heavy subunit ns ns ns 0.82 1160 DY631817 Fish virus induced TRIM protein ns ns 0.59 ns 170 DY626596 Fish virus induced TRIM protein ns ns 0.41 0.44 145 DY628624 Gamma M7 crystallin ns ns ns 0.42 169 DY630388 Glutamyl-tRNA(Gln) amidotransferase 0.48 ns ns ns 347 DY626115 GTPase IMAP family member 7 ns ns 1.14 ns 370 CN471284 High-mobility group 20B 0.60 ns ns ns 163 CN469367 Hox gene cluster 1.34 1.16 0.86 1.11 183 DY627986 Hox gene cluster 1.81 1.12 0.80 1.22 95.1 DY629113 Immunoglobulin light chain ns ns 0.65 ns 482 CN468953 Iron-sulfur cluster assembly enzyme ISCU ns ns ns 0.86 610 DY628151 Kallikrein-8 precursor 1.02 ns ns ns 102 DY627800 Kinesin-like protein 2 (knsl2) ns 0.86 1.84 1.14 398 CN469578 KLR1 gene 1.04 ns ns ns 154 DY629760 LOC100150543, polyprotein 1.35 ns 0.65 0.79 141 CN468718 LOC100151545, similar to Protein KIAA0284 0.72 ns ns ns 145 DY629780 MHC class I ns 0.84 1.26 1.05 161 DY630620 MHC class IA antigen ns ns 0.42 ns 120 DY630701 MHC class II alpha subunit ns ns 0.49 ns 764 DY631898 MHC class II antigen alpha chain ns ns 0.94 ns 87.8 DY631847 Mitotic spindle assembly checkpoint protein MAD2A 0.60 ns ns ns 374 DY627079 Muscle-type creatine kinase CKM2 ns 0.41 ns ns 787 DY626009 Non-LTR retrotransposon Rex1a 0.70 ns ns ns 82.4 DY629391 Non-LTR retrotransposon Rex3_Tet 0.94 ns ns ns 122 CN469375* Peptidyl-prolyl cis-trans isomerase NIMA-interacting 1 ns 0.69 ns ns 663

DY632057 Pituitary adenylate cyclase activating polypeptide receptor 1A ns 1.73 1.98 1.79 170

DY628779 Post-GPI attachment to proteins factor 2 ns 0.87 ns ns 123 DY626114* Ras association domain-containing protein 4 ns 0.82 ns ns 1086 DY630104 Ras-related C3 botulinum toxin substrate 2 1.44 0.83 1.47 1.90 331 DY630508 Replication factor C subunit 5 1.04 ns ns ns 1234 DY628495 Ribosomal protein, large P2 (60S) ns ns ns 1.01 161 DY630832* Ribosomal protein S20 (40S) ns 0.65 ns ns 663 DY626643 Serine/threonine phosphatase gene ns 0.57 0.57 0.54 87.8 CN470072 Sestrin 3 ns 1.30 1.61 1.70 116

DY629126 Short coiled-coil protein ns ns ns 0.59 242 DY631649 SINE sequence ns 0.78 ns ns 138 DY630540 Small inducible cytokine SCYA102 ns 0.64 ns ns 1204 CN471492 Solute carrier family 9 (sodium/hydrogen exchanger) ns ns 0.63 ns 197 CN471103 Ubiquitin ns ns 1.27 ns 985 DY629776 UDP glycosyltransferase 2 family, polypeptide A1 ns 0.92 ns ns 304 CN469822 Vacuolar ATP synthase subunit G 1 0.79 ns ns ns 277

291 Legend: BitScore: the quality of the alignment for the annotated homology. A.twe: A. tweddlei; 292 M.est: M. estherae; P.sim: P. similis; Rchi: R. “chilingali”; “ns”: not significant; “*”: the 293 GenBank number is a representative for multiple array features for that gene. 294 295 Table 2: Genes duplicated relative to A. burtoni with no informative BLAST hit. 296 GenBank A.twe M.est P.sim R.chi DY631067 ns 0.78 0.80 1.02 DY626766 ns 0.81 0.98 0.68 DY629123 ns 0.87 0.83 1.41 DY630373 ns 0.89 1.00 1.23 DY632058* ns 0.90 0.72 0.71 DY627641 ns 0.76 0.85 ns DY630229 ns 0.88 0.83 ns CN471811 ns 1.35 1.22 ns DY631442 ns 1.40 1.16 ns CN470857 ns 0.67 ns 0.96 DY632097 ns 0.79 ns 0.82 CN470988 ns 1.28 ns 1.34 CN470402 ns ns 0.48 0.45 DY631821 ns ns 0.61 0.60 DY626304 ns ns 0.99 1.50 DY631315 ns ns 1.57 1.16 DY629717 ns ns 1.60 1.28 DY628642 ns ns 1.62 1.13 DY629912 1.41 2.21 1.06 1.16 DY631507 0.67 0.69 0.78 0.61 DY627911 1.04 0.87 0.49 ns DY632134* 0.94 0.86 ns 0.84 DY629482 1.39 0.71 ns 1.12 DY630867 0.97 0.64 ns ns CN470216 ns 0.39 ns ns DY631869 ns 0.39 ns ns DY632294 ns 0.41 ns ns DY627085 ns 0.44 ns ns DY630284 ns 0.54 ns ns DY628316 ns 0.64 ns ns DY630993 ns 0.67 ns ns DY631505 ns 0.72 ns ns DY626192 ns 0.75 ns ns DY631827 ns 0.86 ns ns DY626140 ns 1.05 ns ns DY632092 ns 1.09 ns ns

DY625804 ns 1.23 ns ns DY627780 ns 1.51 ns ns CN470835 ns 1.55 ns ns DY628268 ns ns 0.46 ns CN471851 ns ns 0.47 ns DY631408 ns ns 0.50 ns DY626389 ns ns 0.57 ns CN469460 ns ns 0.64 ns CN470713 ns ns 0.68 ns DY626737 ns ns 0.79 ns CN471261 ns ns 0.93 ns DY631698 ns ns 1.03 ns DY629387 ns ns 1.18 ns DY632256 ns ns 1.56 ns DY626428 ns ns ns 0.39 DY628561 ns ns ns 0.42 DY628714 ns ns ns 0.48 CN469431 ns ns ns 0.50 DY628477 ns ns ns 0.58 CN470540 ns ns ns 0.60 CN469913 ns ns ns 0.63 CN470701 ns ns ns 0.65 DY628702 ns ns ns 0.67 CN472050 ns ns ns 0.70 DY627361 ns ns ns 0.74 DY629882 ns ns ns 0.77 DY630964 ns ns ns 0.95 DY631680 ns ns ns 1.02 DY629058 ns ns ns 2.41 DY626122 1.50 ns ns ns DY628172 1.38 ns ns ns CN469125* 1.32 ns ns ns DY625919 1.18 ns ns ns DY625845 1.16 ns ns ns DY627087 1.15 ns ns ns CN470724 1.02 ns ns ns DY632007 0.99 ns ns ns DY631850 0.85 ns ns ns DY628517 0.76 ns ns ns CN470646 0.73 ns ns ns DY627338 0.72 ns ns ns CN470597 0.65 ns ns ns CN470781 0.58 ns ns ns DY628148 0.50 ns ns ns DY625884 0.49 ns ns ns

297 Legened: A.twe: A. tweddlei; M.est: M. estherae; P.sim: P. similis; Rchi: R. “chilingali”; “ns”: 298 not significant; “*”: the GenBank number is a representative for multiple array features for that 299 gene. 300 301

302 Table 3: Oligonucleotide primers used for qPCR designed against GenBank sequence available 303 for microarray features. 304 Primer Efficiency

GenBank Primer Sequence Homology Predicted Length A. burtoni Test Species

DY626766 F: TCGGTCTCCTTAACCGGATG No Hit 193 86 74

R: CTGAGTTTGGCTGCCCGTAA (P. similis)

DY627986 F: ACGAACACCCGAACGGAAAC Hox gene cluster 222 100 104

R: GGTGCACGCACATGAACTGT (M. estherae)

DY631898 F: CGTCCCAGTGAGGATGAGGA MHC class II antigen 161 82 82

R: TGATGCTGATCGGTTGATGC (R. "chilingali")

DY632057 F: ATTACTGCGAGTGCCGTCCA Pituitary adenylate cyclase activating 150 91 78

R: CTGCGCCCTGAAAGAACAGA polypeptide receptor 1A (A. tweddlei)

305 Legend: primer efficiency: percent is based on 4-fold template dilutions for A. burtoni and one 306 heterologous test species. 307

308

References 308 Ashburner, M., C. A. Ball, J. A. Blake, D. Botstein, H. Butler, J. M. Cherry, A. P. Davis, K. 309

Dolinski, S. S. Dwight, J. T. Eppig, M. A. Harris, D. P. Hill, L. Issel-Tarver, A. 310 Kasarskis, S. Lewis, J. C. Matese, J. E. Richardson, M. Ringwald, G. M. Rubin, G. 311 Sherlock, and C. Gene Ontology. 2000. Gene Ontology: tool for the unification of 312 biology. Nature Genetics 25:25-29. 313

Baldwin, B. G., and M. J. Sanderson. 1998. Age and rate of diversification of the Hawaiian 314 silversword alliance (Compositae). Proceedings of the National Academy of Sciences of 315 the United States of America 95:9402-9406. 316

Barreiro, L. B., and L. Quintana-Murci. 2010. From evolutionary genetics to human 317 immunology: how selection shapes host defence genes. Nature Reviews Genetics 11:17-318 30. 319

Chen, Z. Z., C. H. C. Cheng, J. F. Zhang, L. X. Cao, L. Chen, L. H. Zhou, Y. D. Jin, H. Ye, C. 320 Deng, Z. H. Dai, Q. H. Xu, P. Hu, S. H. Sun, Y. Shen, and L. B. Chen. 2008. 321 Transcriptomic and genomic evolution under constant cold in Antarctic notothenioid fish. 322 Proceedings of the National Academy of Sciences of the United States of America 323 105:12944-12949. 324

Coyne, J. A., and H. A. Orr. 2004. Speciation. Sinaur, Sunderland, MA. 325 Darwin, C. 1859. The Origin of Species. Bantam Books. 326 Dobzhansky, T. 1937. Genetics of the evolutionary process. Columbia University Press. 327 Dopman, E. B., and D. L. Hartl. 2007. A portrait of copy-number polymorphism in Drosophila 328

melanogaster. Proceedings of the National Academy of Sciences of the United States of 329 America 104:19920-19925. 330

Farias, I. P., G. Orti, and A. Meyer. 2000. Total evidence: Molecules, morphology, and the 331 phylogenetics of cichlid fishes. Journal of Experimental Zoology 288:76-92. 332

Fischer, H. M., C. W. Wheat, D. G. Heckel, and H. Vogel. 2008. Evolutionary origins of a novel 333 host plant detoxification gene in butterflies. Molecular Biology and Evolution 25:809-334 820. 335

Force, A., M. Lynch, F. B. Pickett, A. Amores, Y. L. Yan, and J. Postlethwait. 1999. 336 Preservation of duplicate genes by complementary, degenerative mutations. Genetics 337 151:1531-1545. 338

Fortna, A., Y. Kim, E. MacLaren, K. Marshall, G. Hahn, L. Meltesen, M. Brenton, R. Hink, S. 339 Burgers, T. Hernandez-Boussard, A. Karimpour-Fard, D. Glueck, L. McGavran, R. 340 Berry, J. Pollack, and J. M. Sikela. 2004. Lineage-specific gene duplication and loss in 341 human and great ape evolution. PLOS Biology 2:937-954. 342

Fryer, G. 1990. Comparative aspects of adaptive radiation and speciation in Lake Baikal and the 343 great rift lakes of Africa. Hydrobiologia 211:137-146. 344

Fryer, G., and T. D. Iles. 1972. The cichlid fishes of the Great Lakes of Africa: Their biology and 345 evolution. Oliver & Boyd, Croythorn House, 23 Ravelston Terrace, Edinburgh. 346

Gould, S. J. 1989. Wonderful Life: The Burgess Shale and the Nature of History. W.W. Norton. 347 Itoi, S., R. Misaki, M. Hirayama, M. Nakaniwa, C. S. Liang, H. Kondo, and S. Watabe. 2005. 348

Identification of three isoforms for mitochondrial adenine nucleotide translocator in the 349 pufferfish Takifugu rubripes. Mitochondrion 5:162-172. 350

Kikuchi, K., S. Itoi, and S. Watabe. 1999. Increased levels of mitochondrial ATP synthase beta-351 subunit in fast skeletal muscle of carp acclimated to cold temperature. Fisheries Science 352 65:629-636. 353

Kocher, T. D. 2004. Adaptive evolution and explosive speciation: The cichlid fish model. Nature 354 Reviews Genetics 5:288-298. 355

Kornfield, I., and P. F. Smith. 2000. African cichlid fishes: Model systems for evolutionary 356 biology. Annual Review of Ecology and Systematics 31:163-+. 357

Lazzaro, B. P., and T. J. Little. 2009. Immunity in a variable world. Philosophical Transactions 358 of the Royal Society B-Biological Sciences 364:15-26. 359

le Pabic, P., E. J. Stellwag, and J. L. Scemama. 2009. Embryonic Development and 360 Skeletogenesis of the Pharyngeal Jaw Apparatus in the Cichlid Nile Tilapia 361 (Oreochromis niloticus). Anatomical Record-Advances in Integrative Anatomy and 362 Evolutionary Biology 292:1780-1800. 363

Li, W. H., Z. L. Gu, H. D. Wang, and A. Nekrutenko. 2001. Evolutionary analyses of the human 364 genome. Nature 409:847-849. 365

Lie, K. K., A. Lanzen, H. Breilid, and P. A. Olsvik. 2009. Gene expression profiling in Atlantic 366 cod (Gadus morhua l.) from two contaminated sites using a custom-made cDNA 367 microarray. Environmental Toxicology and Chemistry 28:1711-1721. 368

Liu, T. A., S. Bhuiyan, R. Snow, S. Yasuda, T. Yasuda, Y. S. Yang, F. E. Williams, M. Y. Liu, 369 M. Suiko, G. Carter, and M. C. Liu. 2008. Identification and characterization of two 370 novel cytosolic sulfotransferases, SULT1 ST7 and SULT1 ST8, from zebrafish. Aquatic 371 Toxicology 89:94-102. 372

Losos, J. B., T. R. Jackman, A. Larson, K. de Queiroz, and S. Rodriguez. 1998. Contingency and 373 determinism in replicated adaptive radiations of island lizards. Science 279:2115-2118. 374

Lukacs, M. F., H. Harstad, U. Grimholt, M. Beetz-Sargent, G. A. Cooper, L. Reid, H. G. Bakke, 375 R. B. Phillips, K. M. Miller, W. S. Davidson, and B. F. Koop. 2007. Genomic 376 organization of duplicated major histocompatibility complex class I regions in Atlantic 377 salmon (Salmo salar). Bmc Genomics 8. 378

Machado, H. E., and S. C. P. Renn. in review. A critical assessment of cross-species detection of 379 gene duplicates using comparative genomic hybridization. Bmc Genomics. 380

Maere, S., K. Heymans, and M. Kuiper. 2005. BiNGO: a Cytoscape plugin to assess 381 overrepresentation of Gene Ontology categories in Biological Networks. Bioinformatics 382 21:3448-3449. 383

Malaga-Trillo, E., Z. Zaleska-Rutczynska, B. McAndrew, V. Vincek, F. Figueroa, H. Sultmann, 384 and J. Klein. 1998. Linkage relationships and haplotype polymorphism among cichlid 385 Mhc class II B loci. Genetics 149:1527-1537. 386

Marques-Bonet, T., J. M. Kidd, M. Ventura, T. A. Graves, Z. Cheng, L. W. Hillier, Z. S. Jiang, 387 C. Baker, R. Malfavon-Borja, L. A. Fulton, C. Alkan, G. Aksay, S. Girirajan, P. Siswara, 388 L. Chen, M. F. Cardone, A. Navarro, E. R. Mardis, R. K. Wilson, and E. E. Eichler. 2009. 389 A burst of segmental duplications in the genome of the African great ape ancestor. Nature 390 457:877-881. 391

Mayr, E. 1963. Animal species and evolution. Belknap Press of Harvard University Press, 392 Cambridge, Massachusetts. 393

Miller, K. M., K. H. Kaukinen, and A. D. Schulze. 2002. Expansion and contraction of major 394 histocompatibility complex genes: a teleostean example. Immunogenetics 53:941-963. 395

Miller, K. M., and R. E. Withler. 1998. The salmonid class I MHC: limited diversity in a 396 primitive teleost. Immunological Reviews 166:279-293. 397

Monson, R. K. 2003. Gene duplication, neofunctionalization, and the evolution of C-4 398 photosynthesis. International Journal of Plant Sciences 164:S43-S54. 399

Ohno, S. 1970. Evolution by Gene Duplication. Springer-Verlag. 400 Persson, A. C., R. J. M. Stet, and L. Pilstrom. 1999. Characterization of MHC class I and 401

beta(2)-microglobulin sequences in Atlantic cod reveals an unusually high number of 402 expressed class I genes. Immunogenetics 50:49-59. 403

Reusch, T. B. H., H. Schaschl, and K. M. Wegner. 2004. Recent duplication and inter-locus gene 404 conversion in major histocompatibility class II genes in a teleost, the three-spined 405 stickleback. Immunogenetics 56:427-437. 406

Sackton, T. B., B. P. Lazzaro, T. A. Schlenke, J. D. Evans, D. Hultmark, and A. G. Clark. 2007. 407 Dynamic evolution of the innate immune system in Drosophila. Nature Genetics 408 39:1461-1468. 409

Salzburger, W., S. C. P. Renn, D. Steinke, I. Braasch, H. A. Hofmann, and A. Meyer. 2008. 410 Annotation of expressed sequence tags for the east African cichlid fish Astatotilapia 411 burtoni and evolutionary analyses of cichlid ORFs. Bmc Genomics 9:1-14. 412

Sandve, S. R., H. Rudi, T. Asp, and O. A. Rognli. 2008. Tracking the evolution of a cold stress 413 associated gene family in cold tolerant grasses. Bmc Evolutionary Biology 8. 414

Sato, A., F. Figueroa, C. O'Huigin, N. Steck, and J. Klein. 1998. Cloning of major 415 histocompatibility complex (Mhc) genes from threespine stickleback, Gusterosteus 416 aculeatus. Molecular Marine Biology and Biotechnology 7:221-231. 417

Schluter, D. 2000. The Ecology of Adaptive Radiation. Oxford University Press, Oxford, UK. 418 Seehausen, O. 2006. African cichlid fish: a model system in adaptive radiation research. 419

Proceedings of the Royal Society B-Biological Sciences 273:1987-1998. 420 Slatkin, M. 1980. Ecological Character Displacement. Ecology 61:163-177. 421 Smith, J. M. 1966. Sympatric speciation. American Naturalist 100:637. 422 Smyth, G. K. 2004. Linear models and empirical Bayes methods for assessing differential 423

expression in microarray experiments. Statistical Applications in Genetics and Molecular 424 Biology 3:1-26. 425

Sugahara, T., Y. S. Yang, C. C. Liu, T. G. Pai, and M. C. Liu. 2003. Sulphonation of 426 dehydroepiandrosterone and neurosteroids: molecular cloning, expression, and functional 427 characterization of a novel zebrafish SULT2 cytosolic sulphotransferase. Biochemical 428 Journal 375:785-791. 429

Tanaka, H., J. Ishibashi, K. Fujita, Y. Nakajima, A. Sagisaka, K. Tomimoto, N. Suzuki, M. 430 Yoshiyama, Y. Kaneko, T. Iwasaki, T. Sunagawa, K. Yamaji, A. Asaoka, K. Mita, and 431 M. Yamakawa. 2008. A genome-wide analysis of genes and gene families involved in 432 innate immunity of Bombyx mori. Insect Biochemistry and Molecular Biology 38:1087-433 1110. 434

Taylor, J. S., and J. Raes. 2004. Duplication and divergence: The evolution of new genes and old 435 ideas. Annual Review of Genetics 38:615-643. 436

Townsend, J. P. 2004. Resolution of large and small differences in gene expression using models 437 for the Bayesian analysis of gene expression levels and spotted DNA microarrays. BMC 438 Bioinformatics 5. 439

van der Aa, L. M., J. P. Levraud, M. Yahmi, E. Lauret, V. Briolat, P. Herbomel, A. Benmansour, 440 and P. Boudinot. 2009. A large new subset of TRIM genes highly diversified by 441 duplication and positive selection in teleost fish. Bmc Biology 7. 442

van Hijum, S., R. J. S. Baerends, A. L. Zomer, H. A. Karsens, V. Martin-Requena, O. Trelles, J. 443 Kok, and O. P. Kuipers. 2008. Supervised Lowess normalization of comparative genome 444 hybridization data - application to lactococcal strain comparisons. Bmc Bioinformatics 9. 445

Wen, Z. M., S. Rupasinghe, G. D. Niu, M. R. Berenbaum, and M. A. Schuler. 2006. CYP6B1 446 and CYP6B3 of the black swallowtail (Papilio polyxenes): Adaptive evolution through 447 subfunctionalization. Molecular Biology and Evolution 23:2434-2443. 448

449 450