# Source paper archive - Source title: Ancestral polyploidy and non-linear dosage shifts define the photosynthetic transition in the Heliantheae alliance - Author(s): Bonthala, V., Kafoutchoni, K. M., Krey, S., Schluter, U., Weber, A. P. M., Tolleter, D., Stich, B. - Source landing URL: https://www.biorxiv.org/content/10.64898/2026.11.05.764543v1 - PDF URL: https://www.biorxiv.org/content/10.64898/2026.11.05.764543v1.full.pdf - PDF SHA-256: e01f16e1278a662c5e5c80a991f93d829a7a524d06b518238ddeb5dac1f29bc1 - Converter: pymupdf4llm 0.0.27 - Limits: Automated PDF text extraction; reading order, equations, tables and formatting may be incomplete or incorrect. Images are omitted, and no OCR was performed. Markup and links are neutralized for safe downloading. Consult the original PDF for the authoritative version. --- [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1 **TITLE** 2 3 **Ancestral polyploidy and non-linear dosage shifts define the photosynthetic transition in** 4 **the Heliantheae alliance** 5 6 **AUTHORS** 7 Venkata Suresh Bonthala [1,2], Konoutan M. Kafoutchoni [1], Stephanie Krey [1], Urte Schlüter [3], 8 Andreas P.M. Weber [3], Dimitri Tolleter [4], Benjamin Stich [*,1,2] 9 10 **AFFILIATIONS** 11 12 1 Julius Kühn-Institut (JKI), Institute for Breeding Research on Agricultural Crops, Rudolf 13 Schick-Platz 3a, OT Groß Lüsewitz, Sanitz, 18190, Germany 14 15 2 Institute of Quantitative Genetics and Genomics of Plants, Faculty of Mathematics and Natural 16 Sciences, Heinrich Heine University, Düsseldorf, Germany 17 18 3 Institute of Plant Biochemistry, Faculty of Mathematics and Natural Sciences, Heinrich Heine 19 University, Düsseldorf, Germany 20 21 4 Biosphere Science and Engineering Division, Carnegie Science, Stanford, CA 94305, USA 22 23 24 25 [Corresponding Author: benjamin.stich@julius-kuehn.de](mailto:benjamin.stich@julius-kuehn.de) 26 27 1 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 28 **Abstract** 29 30 Whole-genome duplications (WGDs) have played a central role in plant evolutionary 31 innovation, yet their genome-wide influence on complex physiological traits such as carbon 32 concentrating mechanisms (CCMs) remains poorly resolved. Here, we present high-quality 33 genome assemblies and annotations for 35 species within the Heliantheae alliance (Asteraceae), 34 spanning the complex photosynthetic spectrum from C 3 to C 4 . Ks-based analyses combined 35 with gene tree-species tree reconciliation identified a shared paleopolyploid event (WGD-2) 36 ~30 million years ago. Using phylogenetically informed association mapping with these 37 species, we detected 451 functional MapMan BINs with copy number variation significantly 38 associated with interspecific variation in the CO 2 compensation point. Notably, 303 of these 39 BINs contain ≥10% WGD-2-derived genes, indicating a strong enrichment of duplicated loci 40 within photosynthetically relevant functional modules. Effect-size analysis revealed that the C 3 41 to-C 4 transition is governed by threshold-dependent genomic architecture rather than gradual 42 additive change. Substantial expansions in key transcription factor modules—including HD 43 ZIP, HSF, LOB, and TCP—were associated with strong non-linear physiological shifts, 44 consistent with genomic tipping points. By integrating comparative genomics, physiological 45 trait variation, and evolutionary analysis, this study establishes a mechanistic link between 46 ancient polyploidy and contemporary photosynthetic diversification. These findings provide a 47 framework for understanding the genetic basis of carbon concentrating mechanism evolution 48 and inform future optimization of photosynthetic efficiency in Asteraceae crops such as 49 sunflower. 50 51 52 **Keywords:** Whole-genome duplications, gene copy number variations, CO 2 compensation 53 points, carbon concentrating mechanisms, Heliantheae alliance. 2 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 54 **Introduction** 55 Photosynthesis serves as the primary metabolic conduit for the synthesis of nearly all biomass 56 on earth (Evans, 2013). At the molecular level, this process is driven by highly-conserved multi 57 protein complexes that execute photon capture, electron transport, and enzymatic carbon 58 assimilation through the Calvin-Benson cycle (Eberhard et al., 2008). This complex, multi 59 stage process has been shaped by numerous independent evolutionary trajectories (Sage, 2004). 60 Over time, these trajectories have refined carbon-fixation strategies across diverse plant 61 lineages. However, the ancestral C 3 pathway remains the most prevalent mode of carbon 62 assimilation (Ehleringer et al., 1997). Despite its dominance so far, it is fundamentally 63 constrained by the oxygenase activity of ribulose-1,5-bisphosphate carboxylase/oxygenase 64 (Rubisco) under current atmospheric CO 2 concentrations (Bowes et al., 1971; Walker et al., 65 2016). Specifically, Rubisco frequently mistakes O 2 for CO 2, producing the toxic metabolite 2 66 phosphoglycolate rather than sugar-building intermediates. This constraint gives rise to the 67 energy-intensive photorespiratory cycle, which salvages 2-phosphoglycolate (Ogren, 1984; 68 Sharkey, 1988). Although this pathway is metabolically necessary, it significantly reduces 69 photosynthetic efficiency and biomass accumulation. 70 71 To circumvent these constraints, various plant lineages have independently evolved carbon 72 concentration mechanisms (CCMs). Most notably, C 4 photosynthesis has arisen in more than 73 60 distinct instances of convergent evolution (Sage, 2004; Sage et al., 2011, 2012). This 74 sophisticated adaptation relies on the spatial compartmentalization of carbon fixation. By 75 elevating CO 2 levels in spatial proximity of Rubisco, it suppresses oxygenase activity and 76 thereby enhances carboxylation efficiency. Achieving these specialized CCMs requires 77 coordinated anatomical and biochemical modifications. These include the development of 78 Kranz anatomy and the recruitment of key enzymes such as phosphoenolpyruvate carboxylase 79 (PEPC) and pyruvate, orthophosphate dikinase (PPDK) (Brown, 1975; Hatch, 1987; Lundgren 80 et al., 2014; Sage et al., 2014). 81 82 Gene duplications, whether single-gene or whole-genome events, are recognized as critical 83 genomic prerequisites for the transition of C 3 to C 4 photosynthesis (Monson, 2003). This pattern 84 has been documented primarily for core enzymes in Flaveria (Ludwig, 2011; Schulze et al., 85 2013), Alloteropsis (Bianconi et al., 2018), and Cleome (Huang et al., 2021; Hoang et al., 2023). 86 For example, the photorespiratory GLDP gene underwent duplication in the ancestral C 3 87 Flaveria species (Schulze et al., 2013). One copy subsequently evolved predominant bundle 3 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 88 sheath-specific expression. This regulatory divergence facilitated the emergence of a 89 photorespiratory CO 2 pump in C 3 –C 4 intermediate Flaveria. Similarly, orthologs of several 90 essential C 4 cycle enzymes—including PEPC, PPDK, NADP-ME, NADP-MDH, and CA— 91 were duplicated in C 3 lineages in Flaveria. These duplications created a genomic substrate in 92 which one paralog was free to undergo neofunctionalization. In parallel, dosage-driven 93 upregulation increased enzymatic capacity. Together, these changes supported the evolution of 94 the specialized NADP-ME CCM (Monson, 2003; Ludwig, 2011; Bianconi et al., 2018). 95 Collectively, the above-mentioned studies demonstrate that gene duplication promotes both 96 subfunctionalization and neofunctionalization (Blanc & Wolfe, 2004; Cusack & Wolfe, 2007; 97 Panchy et al., 2016). Together, these mechanisms enable cell-type-specific expression and 98 dynamic regulatory changes that drive the specialization of photosynthetic functions. However, 99 these few well-characterized biochemical components explain only a fraction of the C 4 100 phenotype. The broader suite of regulatory and structural genes required to orchestrate a fully 101 integrated CCM remains largely unidentified. 102 103 WGDs have been mechanistically linked to a wide range of key adaptations. These include 104 morphological shifts, such as fruit development in tomato (Tomato Genome Consortium, 2012) 105 and fiber quality in cotton (Li et al., 2015a). They have also been associated with complex 106 physiological traits, including nitrogen fixation (Young et al., 2011) and stress tolerance (Huang 107 et al., 2020). Additional examples are reviewed by Qiao et al (2022). Despite this extensive 108 catalog of phenotypic impacts, the specific functional consequences of WGDs for 109 photosynthetic performance remain underexplored. 110 111 The Asteraceae family offers a unique system for investigating the genomic basis of 112 photosynthetic evolution as related genera exist next to the crop sunflower—such as _Flaveria_, 113 _Parthenium_, and _Pectis_ —that collectively encompass the full range of photosynthetic types: C₃, 114 C₃–C₄ intermediates, C₄-like, and C₄ (Moore et al., 1987; McKown & Dengler, 2007; Hansen 115 et al., 2016; Siniscalchi et al., 2021; Adachi et al., 2023). This natural diversity across species 116 within a narrow phylogenetic context creates an opportunity to test how ancient genome 117 duplications (WGDs) may have contributed to the emergence of CCMs. In particular, it allows 118 us to examine the so far disregarded question whether interspecies variation in gene dosage 119 played a role in this process and this also for genes outside of the core photosynthetic 120 machinery. 121 4 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 122 In this study, we (1) established genomic resources for evolutionary analysis of photosynthesis 123 in Asteraceae, (2) identified 451 MapMan BIN functional categories by interspecies genome 124 wide association analysis, and (3) validated the associations in summary using prediction 125 models. Our results indicate that ancient WGD events shaped specific gene copy number 126 architectures. These architectures closely align with the emergence of photosynthetic 127 phenotypes. Together, these patterns point to a genomic mechanism that likely contributed to 128 CCM adaptation. By mapping dosage-sensitive shifts, we clarify how gene copy number 129 variation relates to photosynthetic evolution. These findings highlight potential strategies for 130 enhancing photosynthetic efficiency in sunflower, _Guizotia abyssinica_ (Padhi & Panda, 2026) 131 and its related species. 132 133 134 **Results** 135 136 **Genome assemblies and annotations** 137 In this study, we analysed a total of 38 species across 11 genera within the Asteraceae family. 138 We generated 35 new high-quality genome assemblies specifically for this work 139 (Supplementary Table S1). These represent eight genera from the Heliantheae alliance and two 140 outgroup genera: Artemisia (Anthemideae tribe) and Pallenis (Inuleae tribe). To broaden the 141 genotypic space, this dataset was supplemented with three additional high-quality genomes: 142 _Flaveria bidentis_ from our previous study (Sun et al., 2025), and the publicly available 143 assemblies of _Flaveria brownii_ (Taniguchi et al., 2021) and _Helianthus annuus_ (Badouin et al., 144 2017). 145 146 We estimated genome sizes for the 35 newly sequenced species using a k-mer-based approach. 147 Overall, these k-mer-based estimates were generally consistent with values reported in the 148 literature obtained via flow cytometry and Feulgen densitometry (Taniguchi et al., 2021; 149 [https://www.asteraceaegenomesize.com), though some deviations were observed. For example,](https://www.asteraceaegenomesize.com/) 150 we estimated the genome size for _Coreopsis lanceolata_ as 1.92 Gb, while previous estimates 151 [based on flow cytometry (https://www.asteraceaegenomesize.com) recorded a larger size of](https://www.asteraceaegenomesize.com/) 152 3.359 Gb (Supplementary Table S2). 153 5 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 154 Reference-quality assemblies were generated for nine of the 35 newly sequenced species using 155 the DeNovoMAGIC pipeline (NRGene, Nes Ziona, Israel), yielding an average N50 of over 14 156 Mb and capturing over 95% of BUSCO genes (Seppey et al., 2019). The remaining 26 species 157 were assembled at draft quality using Supernova v2.1.1 (Weisenfeld et al., 2017) and Flye v2.9 158 (Kolmogorov et al., 2019), achieving an average N50 of 528 kb and 91.8% BUSCO (Seppey et 159 al., 2019) completeness (Supplementary Figures S1–S2 and Table S3). 160 161 Repetitive elements were annotated in all 35 newly assembled genomes and the genome 162 assembly of _Flaveria bidentis_ using EDTA v2.0.0 (Ou et al., 2019), with repeat content ranging 163 from 48.58% ( _F. robusta_ ) to 80.46% ( _A. vulgaris_ ; Supplementary Figure S1). Long-terminal 164 repeat (LTR) retrotransposons (Copia, Gypsy, and other LTRs) made up on average across all 165 species the largest fraction (average 46.78%), followed by Helitrons (9.64%) and Mutators 166 (4.33%) (Supplementary Figure S3 and Table S4). Gene annotation was performed in all 35 167 newly assembled genomes and the genome assembly of _F. brownii_ using the deep learning 168 based tool Helixer v0.3 (Stiehler et al., 2021), yielding 27,179 to 80,151 predicted gene models 169 per species, with an average length of 3.2 kb (Supplementary Figure S4 and Table S3). 170 Functional annotation assigned PFAM domains, Gene Ontology (GO) terms using InterProScan 171 (Jones et al., 2014) and MapMan BIN classifications (Schwacke et al., 2019) to an average of 172 76%, 50.7% and 93% of genes, respectively, demonstrating consistent annotation depth across 173 all species. An average of 75.68% of the genes showed detectable gene expression (TPM >=0.5) 174 in at least one tissue or pooled tissues (TPM > 0) using direct RNA sequencing, thereby 175 validating the structural accuracy and coding potential of the Helixer-derived gene models 176 (Supplementary Figure S4 and Table S3). Consistent with broader patterns in plant genome 177 evolution (Badouin et al., 2017), genome size variation across Heliantheae alliance was 178 primarily driven by repeat content (R² = 0.995) rather than gene number (R² = 0.544; 179 Supplementary Figure S5). 180 181 **Evolutionary relationships among species and dating speciation** 182 We identified orthologous gene families across the genome sequences of all available 38 183 species using OrthoFinder v2.5.5 (Emms & Kelly, 2019), assigning 96.36% of all genes 184 (1,339,739) to 55,238 phylogenetic hierarchical orthogroups (HOGs). Gene assignment rates 185 varied among species, ranging from 86.98% in _H. annuus to_ 99.11% in _F. australasica_, with 186 an average of 96.73%. The number of HOGs per species varied from 19,514 in _C. maritima_ to 187 26,411 in _A. vulgaris_ (Supplementary Figure S6 and Table S5). On average, 25.68% of HOGs 6 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 188 per species were shared by all 38 species, representing highly conserved gene families. Over 189 1% of HOGs per species were specific to a single species (species-specific HOGs), while 190 73.25% were shared among a subset of species (Supplementary Figure S6 and Table S5). These 191 findings indicate both deep conservation and lineage-specific evolution of gene families. 192 Notably, _A. vulgaris_, one of two above mentioned outgroup species, exhibited the highest 193 proportion of species-specific HOGs (7.8%), followed by _C. diversifolius_ (5.5%), potentially 194 reflecting recent adaptive evolution or genome duplication events. Pairwise orthologous gene 195 sharing averaged 84.9% between species (Supplementary Figure S7). 196 197 To infer species relationships on a genome-wide level, we reconstructed phylogenetic trees 198 based on the maximum likelihood (ML) method with RAxML-ng v1.2.2 (Kozlov et al., 2019) 199 and the multi-species coalescent (MSC) method with ASTRAL-Pro3 (Zhang & Mirarab, 2022). 200 Based on partitioned multiple sequence alignments of single-copy HOGs, the ML analysis 201 yielded a species tree with consistently high bootstrap support (Figure 1). Using gene trees from 202 RAxML-ng, the MSC approach produced a fully supported species tree (posterior probability 203 = 1 at all nodes; Supplementary Figure S8). Both methods resulted in largely congruent 204 topologies, capturing major evolutionary lineages. However, both approaches revealed cross 205 genus nesting— _Pectis angustifolia_ within _Porophyllum_, and _Coreopsis lanceolata_ clustering 206 outside its genus. Such cross-genus nesting has been previously reported for Asteraceae 207 (Hansen et al., 2016; Huang et al., 2016b) and Brassicaceae (Huang et al., 2016a). Furthermore, 208 the recovered intergeneric relationships and major clade topologies within the alliance align 209 well with previous studies (Zhang et al., 2021). 210 211 Phylogenomics dating using MCMCtree (Yang, 2007) yielded divergence estimates broadly 212 consistent with a recent phylotranscriptomic analysis (Zhang et al., 2021), placing the origin of 213 the Heliantheae alliance at 32.2 million years ago (MYA). However, by leveraging 38 whole 214 genome sequences, we provide refined resolution for the most recent radiations, dating the 215 Heliantheae-Coreopsideae split to 20.6 MYA and the divergence between the specialized 216 Flaveria clades to 3.7 MYA (Figure 1). This refined timeline enables more precise associations 217 between WGD-derived gene retention and the rapid evolution of C 4 photosynthesis phenotypes. 218 219 **Identification of whole-genome duplications** 220 Given the role of whole-genome duplications (WGD) in species diversification and 221 evolutionary innovations in angiosperms (Soltis & Soltis, 2016; Van de Peer et al., 2017; One 7 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 222 Thousand Plant Transcriptomes Initiative, 2019), we systematically identified WGD events 223 across the studied 38 species to assess their prevalence and timing. We employed two 224 complementary approaches to investigate WGDs: analysis of synonymous substitution rates 225 (Ks) and gene tree-species tree reconciliation. 226 227 Using wgd v2 (Chen et al., 2024), we identified paralogous gene clusters and calculated Ks 228 values for each paralogous gene pair within each species to identify Ks peaks corresponding to 229 potential WGD events. All Heliantheae alliance species of our study exhibited a prominent Ks 230 peak between 0.43 and 0.65 (mean: 0.51), suggesting a WGD event about 30.8 MYA (Figure 1 231 and Supplementary Figure S9). This peak aligns well with the previously reported ancient 232 WGD-2 event (29 MYA) in the ancestor of _Helianthus annuus_ (Badouin et al., 2017). In 233 addition, we identified species-specific WGD events in _Pectis angustifolia_ (Ks = 0.07; 4.24 234 MYA) and _Porophyllum gracile_ (Ks = 0.04; 2.42 MYA), indicative of more recent duplications. 235 Also, the outgroup species showed distinct WGD events: _Artemisia vulgaris_ at Ks = 0.08 (4.85 236 MYA) and _Pallenis spinosa_ at Ks = 0.18 (10.91 MYA) (Figure 1 and Supplementary Table S6). 237 238 To further validate these findings, we performed gene tree-species tree reconciliation using 239 Tree2GD v2.5 (Chen et al., 2022). This analysis provided strong evidence for a shared ancient 240 WGD event in the most recent common ancestor (MRCA) of the Heliantheae alliance (Node 241 N42). A total of 82% of duplicated gene families were shared among descendant taxa, which 242 was consistent with the Ks-based inference of the WGD-2 event (Figure 1 and Supplementary 243 Table S7). The reconciliation approach also identified shared WGDs at the MRCAs of several 244 genera, including _Helianthus_ and _Parthenium_ (Node N44), _Cosmos_ and _Coreopsis_ (Node N48), 245 and _Flaveria_ and _Haploesthes_ (Node N59) (Figure 1). These results confirm that the observed 246 duplications reflect large-scale genome duplications rather than small-scale gene duplications 247 (Hakes et al., 2007). 248 249 Our results therefore highlight the dynamics of genome duplication in the evolutionary history 250 of the Heliantheae alliance and its relatives. 251 252 **WGD-2-derived gene duplicates and their distributions** 253 To assess the impact of the ancient whole-genome duplication event (WGD-2) on gene 254 retention and functional conservation across species of the Heliantheae alliance, we analyzed 255 more than 3.18 million paralog pairs (Supplementary Table S8). We focused on paralog pairs 8 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 256 with Ks values of 0.41 – 0.68 (Supplementary Figure S10), which corresponds to a duplication 257 time of ~30.84 million years ago (Figure 1). A total of 17.6% of the total paralog pairs fulfilled 258 this criterion. The proportion of WGD-2 duplicates varied widely across species (7.9% 259 42.2%), with notable differences between absolute and relative counts, reflecting lineage 260 specific evolution (χ² = 121, p = 4.03 × 10⁻¹¹, Supplementary Figure S11A). Crucially, the 261 genomic distribution and functional assignments of these retained pairs were highly non 262 random. On average, 60% of WGD-2 pairs were retained within the same Hierarchical 263 Orthologous Groups, significantly deviating from random genomic allocation expectations (χ² 264 = 183, p = 1.40 × 10⁻²¹, Supplementary Figure S11B and Supplementary Table S8). 265 Furthermore, 85% of WGD-2 pairs shared the exact same MapMan BIN classifications, 266 strongly rejecting the null hypothesis of neutral functional divergence (χ² = 79.3, p = 4.30 × 267 10⁻⁵, Supplementary Figure S11C and Supplementary Table S8). Together, these results suggest 268 that a vast majority of WGD-2 duplicates actively maintained conserved biological roles and 269 structural classifications, rather than undergoing rapid functional drift or decay following the 270 duplication event. 271 272 **Phylogenetic association mapping of gene copy number variation with CO** **2** **compensation** 273 **points** 274 Gene copy number variation (CNV) between species is a genetic polymorphism contributing 275 significantly to phenotypic variability for adaptive evolution in plants and animals 276 (Suryawanshi et al., 2016; DeBolt, 2010; Lye & Purugganan, 2019; Ishikawa et al., 2021; 277 Wilhoit et al., 2025; Pajic et al., 2019; Aoyagi Blue et al., 2021). However, its role in 278 diversifying carbon-concentrating mechanisms (CCMs) is largely unexplored. To address this 279 gap, we associated CNV across 31 species from the Heliantheae alliance including two 280 outgroups, to the CO₂ compensation point (CCP; Supplementary Table S9) as a quantitative 281 proxy for CCM efficiency (Schlüter et al., 2023) using a phylogenetic association mapping 282 (PAM) framework (Kiefer et al., 2019). This approach identified 742 high-confidence HOGs 283 (Supplementary Table S10) and 451 MapMan BINs (Supplementary Table S11) with 284 significant CNV-CCP associations (p < 0.05). 285 286 To gain insights into the functional relevance of the identified associations, we examined the 287 biological roles of the MapMan BINs and HOGs identified through the PAM framework. 288 Among the 451 MapMan BINs (containing ≥2 CNV states), twelve BINs correspond to key 289 components of photosynthesis and photorespiration (Figure 2, Supplementary Table S12 and 9 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 290 Figure S12). These include BINs related to both light-dependent and light-independent 291 reactions: _Photosystem II LHC-II complex component_ (LHCb5, BIN: 1.1.1.1.3), _Photosystem_ 292 _II assembly factor_ (LTO1, BIN: 1.1.1.3.16), _Photosystem I LHC-I complex component_ (LHCa3, 293 BIN: 1.1.4.1.3), _ferredoxin electron carriers_ (Fd1/2/3, BIN: 1.1.5.1.1), and _cyclic electron flow_ 294 _(PGRL1, BIN: 1.1.6.1.2)_ —all integral to photophosphorylation. Additionally, the identification 295 of _PTOX_ (chlororespiration, BIN: 1.1.8.2), _BFA1_ (ATP synthase assembly, BIN: 1.1.9.3.2), and 296 enzymes of the Calvin Benson cycle, such as _GAPDH_ (BIN: 1.2.3.1) and _fructose-1,6-_ 297 _bisphosphatase (BIN:_ 1.2.6), reinforces the involvement of central photosynthetic machinery. 298 Notably, two BINs involved in photorespiration ( _glycine decarboxylase L-protein_ (BIN: 299 1.3.4.3) and _hydroxypyruvate reductase HPR2/3, BIN:_ 1.3.6.2) and one associated with 300 C₄/CAM mechanisms ( _carbonic anhydrase beta-type, BIN:_ 1.4.1.2) were also identified. These 301 findings are consistent with earlier reports that link evolutionary modifications in light 302 harvesting and Calvin cycle components to C₄ photosynthesis (Mallmann et al., 2014; Reyna 303 Llorens & Hibberd, 2017). 304 305 In parallel, 31 transcription factor-related BINs were significantly associated with CCP values 306 (Figure 2, Supplementary Table S13 and Figure S13). These spanned multiple TF 307 superfamilies, including MYB, bHLH, C2H2, ERF, MADS, WRKY, AHL, BEL, bZIP, 308 DREB2, GeBP, GRAS, GRF, HD-ZIP, HSF, KNOX, LOB and TCP families. For the MYB 309 domain transcription factors this is consistent with recent findings that link MYB TFs to 310 chloroplast biogenesis and photosynthetic regulation (Frangedakis et al., 2024). Similarly, the 311 identification of bHLH, bZIP, and WRKY transcription factors aligned with evidence that these 312 TFs play key roles in regulating photorespiration (Aroca et al., 2023). 313 314 **Non-linear CNV-trait associations and effect sizes** 315 To determine whether CNV-CCP association reflected gradual dosage effects or a threshold 316 dependent architecture, we visually examined the linear relationships between CNV and CCP 317 for each significant BIN. CNV-CCP distributions revealed no consistent monotonic trends of 318 linear relationships across all associated BINs. These results indicate that the detected 319 phylogenetic associations are not explained by simple additive or linear scaling of gene copy 320 number. 321 322 Given the absence of consistent linear relationships, we tested whether these associations 323 followed a non-linear, threshold-based architecture by calculating the systematic effect size (η [2] ) 10 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 324 and genomic expansion magnitude (Cohen’s _d_ ) across all 451 BINs (see Materials and 325 Methods). These analyses indicate that the transition from C 3 to C 4 followed a tipping-point 326 framework (Figure 3A) similar to that described for speciation (Nosil et al., 2017; Dakos et al., 327 2019). 328 329 Across the significant BINs, the total explanatory power of gene dosage on CCP varied 330 substantially. The median η [2] was 0.24, and the maximum reached 0.99, even though linear 331 correlations (R [2] ) were negligible (Figure 3A; Supplementary Table S14). We then identified 332 the top five primary drivers (BINs) with the highest tipping point magnitude (defined as the 333 non-linear surplus, η [2 ] - R [2] ). These included transcription factor categories 14.5.3.1.1 (HD-ZIP), 334 14.5.3.4 (HSF), 14.5.2.5 (LOB), 14.5.7.7.1 (TCP), and BIN 50.1.3 (Oxidoreductase). Together, 335 these BINs explained between 63% and 83% of the total variance in CCP, while each showed 336 only a weak linear correlation (R [2 ] <0.22; Figure 3A, orange color circles). We also detected a 337 positive correlation (r = 0.3) between the magnitude of gene counts shift (Cohen’s _d_ ) and 338 physiological impact (Figure 3B). Although moderate in strength, this relationship suggests that 339 larger genomic expansions are more likely to generate the biological thresholds associated with 340 CCM evolution. 341 342 **Validation of the identified phylogenetic associations** 343 Because of the observed low contribution of each individual HOG or BIN to variation in the 344 CCP values, we did not validate them on an individual basis using functional genetic 345 approaches but as a group using genomic prediction models. The results suggested that the 346 BINs that were identified using the above-mentioned PAM approach resulted in a significantly 347 higher prediction performance (Friedman test, _p_ = 0.006 6 for _R_ ² and _p_ = 0.004 for RMSE; 348 Nemenyi post hoc test, _p_ < 0.05) than all BINs, whereas PAM-derived HOGs did not differ 349 significantly from all HOGs (Figure 4A). Specifically, across all examined validation schemes, 350 PAM-derived BINs and PAM-derived HOGs showed lower error (RMSE = 14.7 ppm, 15.0 351 ppm) and the higher out-of-sample _R_ ² (0.475, 0.456), outperforming models that include all 352 BINs (15.6 ppm; _R_ ² 0.408) or HOGs (15.3 ppm; _R_ ² 0.426) (Figure 4A). In measured hold-out 353 tests, these four feature sets were statistically indistinguishable ( _p_ >0.05) with best performance 354 around RMSE = 11.3 ppm and _R_ ² = 0.58, indicating minimal overfitting (Figure 4B). 355 Furthermore, we confirmed the robustness of these associations under distribution shift 356 scenarios using both independent external data (literature-based validation) and artificially 357 perturbed datasets with swapped species assignments (swap-resampled validation). Across 11 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 358 these tests, PAM-derived BINs consistently maintained a performance advantage over full BIN 359 sets (e.g., _R_ ² = 0.489 vs. _R_ ² = 0.352 for swap-resampling; Figure 4B). 360 361 To identify the most relevant gene families among the set of PAM associations, we ranked them 362 based on the absolute standardized rrBLUP effect size from the genomic prediction models 363 (Figure 4D). The top 50 PAM-derived BINs included functions relevant to CCP biology, such 364 as photosynthesis light reactions, chlorophyll and tetrapyrrole metabolism, photorespiration and 365 glyoxylate pathways, Calvin-Benson cycle, mitochondrial energy metabolism, redox and ROS 366 homeostasis, and metabolite transport (Figure 4D and Supplementary Table S15). 367 368 **Contribution of whole-genome duplication-2 to CCP-associated MapMan BINs'** 369 **functional categories** 370 After having established that copy number variation in specific MapMan BINs strongly 371 correlates with variation in CCP and the validation procedure, we next sought to determine the 372 evolutionary origin of these expansions. Specifically, we examined whether the ancient whole 373 genome duplication (WGD-2) event served as the primary genomic substrate for these adaptive 374 gene expansions. In order to do so, we quantified the proportion of WGD-2–derived genes 375 within functionally annotated MapMan BINs across all species. 376 377 Out of the 451 BINs for which CNV was significantly associated with CCP, 303 BINs (67.18%) 378 contained at least 10% of their genes derived from WGD-2 (Supplementary Table S16). To 379 determine whether this genomic retention was non-stochastic, we performed a genome-wide 380 hypergeometric enrichment test against the background of all 6,472 annotated MapMan BINs. 381 This analysis confirmed that the functional modules linked to CCP variation are non-randomly 382 enriched for WGD-2 retentions (p = 2.84 x 10 [-46] ), establishing a broad, systemic influence of 383 WGD-2 on CNV at the BIN-level. In addition, the same approach indicated that the WGD-2 384 retention in BINs of the structural and regulatory machinery of the carbon-concentrating 385 mechanism are significantly (p = 9.12 x 10 [-4] ) over-represented (Figure 5 and Supplementary 386 Table S12). Furthermore, the genomic legacy of WGD-2 extended deeply into upstream 387 regulatory networks. Twenty-one out of the 31 transcription factor (TF) BINs (67.7%) crossed 388 the 10% WGD-2 retention threshold, with proportions ranging from 12.87% to 87.30% (Figure 389 5 and Supplementary Table S13). The hypergeometric enrichment test indicated that this 390 duplication-driven expansion within TF regulatory pathways is also highly significant 391 compared to the background genome (p = 2.33 x 10 [-4] ). 12 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 392 393 394 **Discussion** 395 396 The repeated, independent emergence of carbon-concentrating mechanisms (CCMs) represents 397 one of the most significant physiological transitions in plant evolution (Sage, 2004; Sage et al., 398 2011, 2012). The Asteraceae—particularly the Heliantheae alliance—provide a unique natural 399 system to investigate this phenomenon. This is due to its exceptional photosynthetic diversity 400 and deep phylogenetic divergence allowing to bridge the gap between ancient polyploidization 401 events and present-day physiological phenotypes. 402 403 **High-quality genome assemblies and evolutionary context as foundations for trait** 404 **mapping** 405 Because none or only assemblies of limited-quality were previously available (cf. Taniguchi et 406 al., 2021), we generated new, high-quality reference resources for these lineages as part of our 407 36 new assemblies. This improvement is central to our study, as the reliability of interspecies 408 CNV-phenotype associations depends strictly on genome assembly quality. The assemblies 409 presented in our study were generated using diverse sequencing technologies, including short 410 read, linked-read and long-read platforms. Despite this heterogeneity, all assemblies exhibit a 411 high contiguity and completeness (Supplementary Figures S1-S4) and met the critical quality 412 threshold of an N50 greater than the average gene length (Supplementary Table S3), which is 413 a prerequisite for accurate gene and paralog identification (Yandell & Ence, 2012). 414 Furthermore, BUSCO completeness scores remained consistently high and displayed minimal 415 inter-species variation (Supplementary Figure S2), indicating that the gene space is sufficiently 416 captured across all assemblies to enable robust, comparable CNV estimation. Furthermore, the 417 quality of the assemblies is comparable to, or exceeds, existing Asteraceae assemblies 418 (Taniguchi et al., 2021) and is similar to assemblies from well-studied Brassicaceae species 419 (Kiefer et al., 2019; Guerreiro et al., 2023). 420 421 Annotation bias is a common challenge in multi-platform comparative studies. To address this 422 issue, we applied a uniform, centralized annotation pipeline across all species. This pipeline 423 incorporated the deep learning-based tool Helixer (Holst et al., 2026) and InterProScan (Jones 13 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 424 et al., 2014). This consistency was essential to ensure that variation in gene counts reflect true 425 biological copy number variation (CNV) and is not due to technical artifacts or inconsistent 426 gene calling. Therewith our PAM analysis was grounded on high-quality, uniformly processed 427 genomic data providing a robust comparative framework for all following analyses. 428 429 **WGD-2: A shared genomic innovation with functional legacy** 430 Whole genome duplications are widely recognized as engines of plant diversification and trait 431 innovation (Jiao et al., 2011; Soltis & Soltis, 2016; Van de Peer et al., 2017; Landis et al., 2018; 432 One Thousand Plant Transcriptomes Initiative, 2019). They also create opportunities for new 433 traits to emerge. Despite this, their effects on complex traits like photosynthesis are less clear. 434 Past studies often focused on isolated examples or a small set of pathway-specific genes 435 (Ludwig, 2011; Schulze et al., 2013; Bianconi et al., 2018; Huang et al., 2021; Hoang et al., 436 2023), but our study provides a systematic analysis of how this process shaped complex 437 photosynthetic phenotypes. 438 439 In this study, we identified a conserved WGD-2 event (K s = 0.43–0.65) that is shared across all 440 Heliantheae alliance species, placing the duplication at their most recent common ancestor 441 (Figure 1). This genomic redundancy precedes therewith the diversification of all C 3 -C 4 and C 4 442 lineages within the Heliantheae alliance (Badouin et al., 2017). Our observation that the 443 MapMan BINs associated with CO 2 compensation point (CCP) variation are significantly 444 enriched for WGD-2 derived genes illustrate that the WGD-2 event is not merely resulting in 445 ancestral “baggage” but has a functional relevance. Our results therefore directly support the 446 “pre-conditioning” hypothesis of Monson (2003): the WGD-2 event expanded the genomic 447 repertoire of the Heliantheae ancestor, providing a "modular" set of duplicated genes that later 448 acted as evolutionary hubs for the assembly of complex CCM phenotypes (Supplementary 449 Figure S11). These observations are giving an evolutionary explanation to the results of earlier 450 studies which revealed the duplication of a narrow set of core photosynthesis genes (Ludwig, 451 2011; Schulze et al., 2013; Taniguchi et al., 2021; Hoang et al., 2023) while we move beyond 452 and reveal how WGD-2 provided the genomic raw material for the systemic anatomical and 453 metabolic rewiring necessary for CCM evolution which is discussed in the following for 454 different pathways. 455 14 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 456 **WGD-2-driven expansion of photosynthetic and regulatory modules supports CCM** 457 **evolution** 458 459 **1. Metabolic specialization and energy flux** 460 While the ancient WGD-2 event established the foundational genomic scaffold for these 461 pathways, subsequent lineage-specific duplications provided the “fine-tuning” necessary for 462 tissue-specific optimization. Together, these events expanded critical metabolic nodes, such as 463 β-carbonic anhydrase (β-CA), which is essential for the initial conversion of CO 2 to HCO 3 [-] 464 (Hatch & Burnell, 1990). The retention of multiple WGD-2 derived β-CA copies, supplemented 465 by lineage-specific expansions (Figure 5; Supplementary Table S12), suggests that selection in 466 the Heliantheae favored both higher enzyme capacity and precise control of carbon flow. 467 468 Furthermore, the co-retention of duplicated Calvin-Benson cycle enzymes (GAPDH, FBPase) 469 and light-harvesting components (LHCb5, LHCa3) likely enabled subfunctionalization among 470 these gene copies (Figure 5; Supplementary Table S12). Such partitioning would help balance 471 RuBP regeneration and light capture between mesophyll and bundle sheath cells. It meets the 472 higher ATP demands of CCM-active species and reduces stress from reactive oxygen that arises 473 with increased carbon flux (Rochaix, 2011). 474 475 **2. Regulatory and anatomical innovation** 476 A clear example for the refinement of the duplication space provided by the WGD-2 event by 477 lineage-specific gene family expansions is the SCARECROW (SCR) gene family (GRAS 478 transcription factors); WGD-2 derived SCR copies likely allowed ancestral root endodermis 479 networks to be co-opted for leaf patterning, facilitating the characteristic Kranz anatomy 480 (Slewinski et al., 2012; Hughes et al., 2019; Shaar-Moshe & Brady, 2023). 481 482 This regulatory rewiring extends to bHLH and MYB transcription factor families. MYB 483 proteins likely promote CCM evolution by modulating gene networks involved in chloroplast 484 biogenesis, enhancing the anatomical and metabolic differentiation required for efficient 485 glycine shuttling in C 3 -C 4 intermediates. Such control may have facilitated the transition from 486 C 3 to C 3 -C 4 photosynthesis by optimizing organelle development (Frangedakis et al., 2024; 487 Dickinson et al., 2025). Furthermore, the synergistic binding of duplicated bHLH transcription 488 factors to C 4 promoters, such as NADP-ME (Borba et al., 2018), provides one possible 15 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 489 mechanism for functional enhancement. More gene copies and stronger promoter binding 490 probably raised local CO 2 around Rubisco, helping to reduce photorespiration. 491 492 The concurrent expansion of both core metabolic BINs and regulatory transcription factor 493 modules, driven by the synergy of WGD-2 and subsequent lineage-specific duplications, 494 appears to underlie the variation in CO 2 compensation points (CCP) across the Heliantheae 495 alliance. Furthermore, our findings demonstrate that CCM evolution is not driven by isolated 496 enzymes, but by the coordinated, modular expansion of entire metabolic and regulatory 497 landscapes. 498 499 Because of the observed low contribution of each individual HOG or BIN to variation in the 500 CCP values, we did not try to validate them on an individual basis using functional genetic 501 approaches but as a group using genomic prediction models (Dos Santos Sá et al., 2026). In 502 detail, we selected the BINs for which the CNV was significantly associated with CCP values 503 in the PAM. Those were combined and used as a group to predict CCM values of a so-called 504 validation set. We observed that genomic prediction with phylogenetically informed BINs 505 improved cross-species accuracy for CCP relative to unfiltered feature sets. This indicated that 506 the evolutionary-based grouping helps retain conserved, biologically meaningful signal and that 507 the previously described associated CNV have biological relevance. 508 509 **Non-linear Architecture Underlies CNV-CCP Associations** 510 The linear effect sizes (R [2] ) for individual BIN copy numbers are often modest, but the total 511 explanatory power (η [2] ) is substantially higher (Figure 3; Supplementary Table S14). This 512 difference suggests that gene dosage effects on CCP are not additive. Rather, these effects 513 appear shaped by non-linear genetic architectures, such as evolutionary thresholds or tipping 514 points (Nosil et al., 2017; Dakos et al., 2019). This observation can be explained by the fact that 515 duplicated genes do not act alone. They work within complex regulatory networks that include 516 epistatic interactions and feedback loops (Kaste et al., 2024). Epigenetic regulation and context 517 dependent gene expression may further disconnect raw copy numbers from the resulting traits 518 (Hennacy & Jonikas, 2020; Cummins, 2021). In summary, whole-genome duplication gives the 519 raw material for evolutionary innovation. However, translating these expansions into functional 520 physiological adaptations requires complex, multi-layered mechanisms, and non-linear 521 mechanisms. 522 16 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 523 524 **Conclusion** 525 By integrating high-quality genome assemblies with experimental measurements across 35 526 species spanning C 3, C 3 -C 4 intermediate, and C 4 photosynthetic types, we established a 527 genome-wide interspecies framework linking carbon-concentrating mechanism (CCM) 528 evolution to genomic features within the Heliantheae alliance. 529 Crucially, we identified a shared whole-genome duplication event dated to approximately 30 530 million years ago (Figure 1), which established a foundational genomic scaffold that subsequent 531 lineage-specific duplications refined to drive contemporary trait variation. Phylogenetic 532 association mapping revealed systematic gene copy number variations across critical functional 533 categories (Figure 2), with photosynthetic associations enriched in functional modules 534 containing at least 10% WGD-2-derived genes. 535 Rather than involving isolated loci, these expansions are distributed across coordinated 536 biochemical and regulatory BINs—encompassing both metabolic enzymes and upstream 537 regulatory transcription factors. This pattern indicates that CCM evolution relied on the 538 retention and diversification of duplicated functional modules. 539 Therewith, this work provides a refined genomic blueprint for enhancing photosynthetic 540 efficiency in major Asteraceae crops, including sunflower. Instead of focusing on single 541 enzymes, breeders could aim to increase the copy number of these regulatory genes, through 542 genome editing or by inducing polyploidy and selecting the best-performing lines, as a way to 543 improve photosynthesis in crops such as sunflower. 544 545 546 **Materials and Methods** 547 548 **Species selection and physiological characterization** 549 We selected 35 Asteraceae species from the Heliantheae alliance and closely related tribes 550 (Supplementary Table S1) spanning the full photosynthetic continuum, from C 3 to C 4 types. All 551 species were grown under controlled conditions (24°C/21°C day/night; 16h photoperiod). 552 These standardized settings ensured that physiological measurements were directly comparable 17 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 553 across species. To quantify the carbon concentrating mechanism (CCM) efficiency, we 554 measured CO 2 compensation points (CCP) using infrared gas exchange measurements (LI 555 COR 6800). CCP values were derived from A/c i curves under high light (1500 µmol m [-2] s [-1] ) 556 and a controlled vapor pressure deficit of 1.5 kPa. The detailed growth conditions and gas 557 exchange protocols are described in the Supplementary Information. 558 559 **Multi-omics sequencing and genome assembly** 560 To establish a high-resolution genomic framework, we generated _de novo_ genome assemblies 561 for 35 species. High-molecular-weight DNA was extracted using an optimized CTAB-based 562 protocol specifically adapted for long-read library preparation (Arseneau et al., 2017; Xin & 563 Chen, 2012). Genome sequencing was realized using a combination of short-read, linked-read, 564 and long-read sequencing technologies (Supplementary Table S1 and S2). To support gene 565 model prediction and quantify gene activity, we performed multi-organ transcriptomic 566 sequencing (RNA-seq) for 27 species and direct-RNA sequencing for 8 more species. Genome 567 assemblies were generated using assemblers, such as Flye (Kolmogorov et al., 2019), 568 Supernova (Weisenfeld et al., 2017), and DeNovoMAGIC (NRGene, Nes Ziona, Israel). 569 Assembly quality was evaluated based on contiguity and completeness via BUSCO analysis 570 (Seppey et al., 2019). Detailed descriptions of DNA and RNA extraction, library preparation, 571 and assembly parameters are provided in the Supplementary Information. 572 573 **Phylogenomics and divergence dating** 574 Gene families, defined as hierarchical orthogroups (HOGs), were inferred across 38 species 575 using OrthoFinder v2.5.5 (Emms & Kelly, 2019). Species phylogeny was constructed using 576 two complementary approaches. First, we performed maximum-likelihood analysis of 577 concatenated single-copy orthologs using RAxML-ng (Kozlov et al., 2019). Second, we applied 578 a multi-species coalescent framework using ASTRAL-Pro3 (Zhang & Mirarab, 2022). 579 Divergence times were estimated using a Bayesian relaxed-clock model implemented in 580 MCMCtree (Yang, 2007). This analysis was calibrated with three fossil-based reference points. 581 Partition models, bootstrap parameters, and calibration priors are detailed in the Supplementary 582 Information. 583 584 **Identification of whole-genome duplication (WGD)** 585 Paralogous gene clusters were identified across species. For each paralog pair, synonymous 586 substitution rates (Ks) were calculated to detect historical polyploidy events. WGD-associated 18 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 587 Ks peaks were detected using kernel density estimation. Confidence intervals (90%) were 588 obtained through resampling. To distinguish shared from lineage-specific duplication events, 589 gene duplication nodes were mapped onto the species tree using Tree2GD (Chen et al., 2022). 590 A detailed description of the Ks filtering pipeline and reconciliation criteria is provided in the 591 Supplementary Information. 592 593 **Phylogenetic association mapping (PAM)** 594 To link genomic architecture with physiological traits, we applied the phylogenetic association 595 mapping framework of (Kiefer et al., 2019). This approach accounts for evolutionary 596 relatedness via a variance-covariance matrix derived from the species phylogeny. Normalized 597 gene copy numbers (HOGs and MapMan BINs) were regressed against CCP values across 31 598 species. Predictive performance was evaluated using ridge-regression best linear unbiased 599 prediction (rrBLUP; Endelman, 2011). Model robustness was assessed through repeated 600 stratified cross-validation. Although we performed phylogenetic association mapping using 31 601 species for which CCP measurements were available, we excluded two species (A. vulgaris and 602 P. spinosa), for which the WGD-2 peak was absent, from this analysis. The mathematical 603 derivation of the normalization formulas, the structure of the predictive models, and the cross 604 validation schemes are elaborated in the Supplementary Information. 605 606 **Quantifying effect sizes and evolutionary tipping points** 607 To characterize the non-linear relationship between CNV and CCP, we calculated two 608 complementary statistics across all 451 BINs: the systematic effect size (η [2] ) and the genomic 609 expansion magnitude (Cohen’s _d_ ). η [2] quantifies the proportion of total variation in CCP 610 explained by differences in copy number states and provides an overall measure of the 611 explanatory power of a given MapMan BIN with respect to CCP. To quantify the magnitude of 612 genomic shift between photosynthetic states, we additionally calculated Cohen’s _d_ . Together, 613 these metrics allowed the detection of potential evolutionary tipping points (Nosil et al., 2017; 614 Dakos et al., 2019). 615 616 617 **Data availability** 618 Sequencing data for all 35 species have been deposited in the NCBI BioProject database 619 [(https://www.ncbi.nlm.nih.gov/bioproject) under accession number PRJNA1260212. Sequence](https://www.ncbi.nlm.nih.gov/bioproject) 19 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 620 data for _Flaveria bidentis_ was separately obtained from BioProject PRJNA1259992. Genome 621 [assemblies and annotations have been archived on FigShare and are available at](https://figshare.com/account/projects/247898/articles/28946543) 622 [https://doi.org/10.6084/m9.figshare.c.8797566.](https://doi.org/10.6084/m9.figshare.c.8797566) 623 624 625 **Code availability** 626 All scripts used in this study are available on GitHub at 627 [https://github.com/vsbonthala/Gain4Crops.](https://github.com/vsbonthala/Gain4Crops) 628 629 630 **Funding** 631 We appreciate funding by the European Union H2020 Program (project GAIN4CROPS, GA 632 no. 862087, to B.S. and A.P.M.W.) and the Deutsche Forschungsgemeinschaft (Cluster of 633 Excellence for Plant Sciences [CEPLAS] under Germany’s Excellence Strategy EXC-2048 634 under project ID 390686111 to A.P.M.W. and CRC TRR 341 “Plant Ecological Genetics” to 635 B.S. and A.P.M.W). Open Access funding enabled and organized by Projekt DEAL. 636 637 638 **Author contributions** 639 BS conceptualized the study and acquired funding. BVS performed the genome assembly, 640 annotation, and all phylogenomic analyses—including species tree construction, WGD 641 identification, association mapping, and quantification of WGD-derived CNVs—and drafted 642 the manuscript with input from all authors. SK managed plant growth and tissue collection, US 643 conducted the CCP measurements, and KMK validated the association mapping through 644 genomic predictions. 20 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 645 646 647 **Acknowledgments** 648 We gratefully acknowledge the following institutions and genebanks for generously providing 649 seed and germplasm materials essential for this study: the USDA-ARS National Arid Land 650 Plant Genetic Resources Unit (Parlier, CA, USA), the USDA-ARS Western Regional Plant 651 Introduction Station at Washington State University (Pullman, WA, USA), the Leibniz Institute 652 of Plant Genetics and Crop Plant Research (IPK, Gatersleben, Germany), the Royal Botanic 653 Gardens, Kew (UK), and the Botanical Garden of Heinrich Heine University Düsseldorf 654 (Germany). The authors acknowledge the computational infrastructure and support provided by 655 the Center for Information and Media Technology at Heinrich Heine University Düsseldorf and 656 the German Network for Bioinformatics Infrastructure (de.NBI, [https://www.denbi.de/) that](https://www.denbi.de/) 657 contributed to the research results reported within this study. The authors gratefully 658 acknowledge the computing time provided to them on the high-performance computers 659 [Noctua2] at the NHR Center PC2. These are funded by the Federal Ministry of Education and 660 Research and the state governments participating on the basis of the resolutions of the GWK 661 for the national high-performance computing at universities (www.nhr-verein.de/unsere 662 partner). 663 We thank Prof. Tomas Morosinotto (University of Padova, Italy) and Dr. Giovanni Finazzi 664 (French Alternative Energies and Atomic Energy Commission, CEA, France) for their help 665 with the photosynthetic phenotyping. 666 667 668 **Declaration of interests** 669 The authors declare no competing interests. 670 671 672 **References** 21 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 673 Adachi, S., Stata, M., Martin, D. G., Cheng, S., Liu, H., Zhu, X. G., & Sage, R. F. (2023). The 674 Evolution of C4 Photosynthesis in Flaveria (Asteraceae): Insights from the Flaveria linearis 675 Complex. Plant physiology, 191(1), 233–251. https://doi.org/10.1093/plphys/kiac467 676 Aoyagi Blue, Y., Kusumi, J., & Satake, A. (2021). Copy number analyses of DNA repair genes 677 reveal the role of poly (ADP-ribose) polymerase (PARP) in tree longevity. iScience, 24(7), 678 102779. https://doi.org/10.1016/j.isci.2021.102779 679 Aroca, A., García-Díaz, I., García-Calderón, M., Gotor, C., Márquez, A. J., & Betti, M. (2023). 680 Photorespiration: regulation and new insights on the potential role of persulfidation. Journal of 681 [experimental botany, 74(19), 6023–6039.](https://doi.org/10.1093/jxb/ery029) https://doi.org/10.1093/jxb/erad291 682 Arseneau, J. R., Steeves, R., & Laflamme, M. (2017). Modified low-salt CTAB extraction of 683 high-quality DNA from contaminant-rich tissues. _Molecular ecology resources_, _17_ (4), 686– 684 [693. https://doi.org/10.1111/1755-0998.12616](https://doi.org/10.1105/tpc.021410) 685 Badouin, H., Gouzy, J., Grassa, C. J., Murat, F., Staton, S. E., Cottret, L., Lelandais-Brière, C., 686 Owens, G. L., Carrère, S., Mayjonade, B., Legrand, L., Gill, N., Kane, N. C., Bowers, J. E., 687 Hubner, S., Bellec, A., Bérard, A., Bergès, H., Blanchet, N., Boniface, M. C., … Langlade, N. 688 B. (2017). The sunflower genome provides insights into oil metabolism, flowering and Asterid 689 evolution. Nature, 546(7656), 148–152. https://doi.org/10.1038/nature22380 690 Bianconi, M. E., Dunning, L. T., Moreno-Villena, J. J., Osborne, C. P., & Christin, P. A. (2018). 691 Gene duplication and dosage effects during the early emergence of C4 photosynthesis in the 692 grass genus Alloteropsis. Journal of experimental botany, 69(8), 1967–1980. 693 https://doi.org/10.1093/jxb/ery029 694 Birchler, J. A., Riddle, N. C., Auger, D. L., & Veitia, R. A. (2005). Dosage balance in gene 695 regulation: biological implications. Trends in genetics: TIG, 21(4), 219–226. 696 https://doi.org/10.1016/j.tig.2005.02.010 697 Blanc, G., & Wolfe, K. H. (2004). Functional divergence of duplicated genes formed by 698 polyploidy during Arabidopsis evolution. The Plant cell, 16(7), 1679–1691. 699 https://doi.org/10.1105/tpc.021410 700 Borba, A. R., Serra, T. S., Górska, A., Gouveia, P., Cordeiro, A. M., Reyna-Llorens, I., 701 Knerová, J., Barros, P. M., Abreu, I. A., Oliveira, M. M., Hibberd, J. M., & Saibo, N. J. M. 702 (2018). Synergistic Binding of bHLH Transcription Factors to the Promoter of the Maize 22 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 703 NADP-ME Gene Used in C4 Photosynthesis Is Based on an Ancient Code Found in the 704 Ancestral C3 State. Molecular biology and evolution, 35(7), 1690–1705. 705 https://doi.org/10.1093/molbev/msy060 706 Bowes, G., Ogren, W. L., & Hageman, R. H. (1971). Phosphoglycolate production catalyzed 707 by ribulose diphosphate carboxylase. _Biochemical_ _and_ _biophysical_ _research_ 708 _communications_, _45_ (3), 716–722. https://doi.org/10.1016/0006-291x(71)90475-x 709 Brown WV. (1975). Variations in anatomy, associations, and origins of Kranz tissue. American 710 Journal of Botany, 62, 395–402 711 Campbell, J. Y., & Thompson, S. B. (2008). Predicting excess stock returns out of sample: Can 712 anything beat the historical average?. The Review of Financial Studies, 21(4), 1509-1531. 713 https://doi.org/10.1093/rfs/hhm055 714 Chen, D., Zhang, T., Chen, Y., Ma, H., & Qi, J. (2022). Tree2GD: a phylogenomic method to 715 detect large-scale gene duplication events. Bioinformatics (Oxford, England), 38(23), 5317– 716 5321. https://doi.org/10.1093/bioinformatics/btac669 717 Chen, H., Zwaenepoel, A., & Van de Peer, Y. (2024). wgd v2: a suite of tools to uncover and 718 date ancient polyploidy and whole-genome duplication. Bioinformatics (Oxford, England), 719 40(5), btae272. https://doi.org/10.1093/bioinformatics/btae272 720 Cummins P. L. (2021). The Coevolution of RuBisCO, Photorespiration, and Carbon 721 Concentrating Mechanisms in Higher Plants. Frontiers in plant science, 12, 662425. 722 https://doi.org/10.3389/fpls.2021.662425 723 Cusack, B. P., and Wolfe, K. H. (2007). When gene marriages don't work out: divorce by 724 subfunctionalization. Trends Genet. 23, 270–272. https://doi.org/10.1016/j.tig.2007.03.010 725 Dakos, V., Matthews, B., Hendry, A. P., Levine, J., Loeuille, N., Norberg, J., Nosil, P., Scheffer, 726 M., & De Meester, L. (2019). Ecosystem tipping points in an evolving world. _Nature ecology_ 727 _& evolution_, _3_ (3), 355–362. https://doi.org/10.1038/s41559-019-0797-2 728 DeBolt S. (2010). Copy number variation shapes genome diversity in Arabidopsis over 729 immediate family generational scales. Genome biology and evolution, 2, 441–453. 730 https://doi.org/10.1093/gbe/evq033 23 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 731 Dickinson, P. J., Triesch, S., Schlüter, U., Weber, A. P. M., & Hibberd, J. M. (2025). A 732 transcription factor module mediating C2 photosynthesis in the Brassicaceae. EMBO reports, 733 26(12), 3024–3031. https://doi.org/10.1038/s44319-025-00461-1 734 Eberhard, S., Finazzi, G., & Wollman, F. A. (2008). The dynamics of photosynthesis. Annual 735 review of genetics, 42, 463–515. https://doi.org/10.1146/annurev.genet.42.110807.091452 736 Ehleringer, J. R., Cerling, T. E., & Helliker, B. R. (1997). C4 photosynthesis, atmospheric CO2, 737 and climate. Oecologia, 112(3), 285–299. https://doi.org/10.1007/s004420050311 738 Emms, D. M., & Kelly, S. (2019). OrthoFinder: phylogenetic orthology inference for 739 comparative genomics. Genome biology, 20(1), 238. https://doi.org/10.1186/s13059-019 740 1832-y 741 Endelman, J. B. (2011). Ridge regression and other kernels for genomic selection with R 742 package rrBLUP. The plant genome, 4(3). https://doi.org/10.3835/plantgenome2011.08.0024 743 Evans J. R. (2013). Improving photosynthesis. _Plant physiology_, _162_ (4), 1780–1793. 744 https://doi.org/10.1104/pp.113.219006 745 Frangedakis, E., Yelina, N. E., Billakurthi, K., Hua, L., Schreier, T., Dickinson, P. J., Tomaselli, 746 M., Haseloff, J., & Hibberd, J. M. (2024). MYB-related transcription factors control chloroplast 747 biogenesis. Cell, 187(18), 4859–4876.e22. https://doi.org/10.1016/j.cell.2024.06.039 748 Freeling, M., and Thomas, B. C. (2006). Gene-balanced duplications, like tetraploidy, provide 749 predictable drive to increase morphological complexity. Genome Res. 16, 805–814. 750 https://doi.org/10.1101/gr.3681406 751 Freeling, M., Scanlon, M. J., & Fowler, J. E. (2015). Fractionation and subfunctionalization 752 following genome duplications: mechanisms that drive gene content and their consequences. 753 Current opinion in genetics & development, 35, 110–118. 754 https://doi.org/10.1016/j.gde.2015.11.002 755 Guerreiro, R., Bonthala, V. S., Schlüter, U., Hoang, N. V., Triesch, S., Schranz, M. E., Weber, 756 A. P. M., & Stich, B. (2023). A genomic panel for studying C3-C4 intermediate photosynthesis 757 in the Brassiceae tribe. Plant, cell & environment, 46(11), 3611–3627. 758 https://doi.org/10.1111/pce.14662 24 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 759 Hakes, L., Pinney, J. W., Lovell, S. C., Oliver, S. G., & Robertson, D. L. (2007). All duplicates 760 are not equal: the difference between small-scale and genome duplication. Genome biology, 761 8(10), R209. https://doi.org/10.1186/gb-2007-8-10-r209 762 Hansen, D. R., Jansen, R. K., Sage, R. F., Villaseñor, J. L., and Simpson, B. B. (2016). 763 Molecular phylogeny of Pectis (Tageteae, Asteraceae), a C4 genus of the Neotropics, and its 764 sister genus Porophyllum. Lundellia 19, 6–38. https://doi.org/10.25224/1097-993X-19.1.6 765 Hatch, M.D. (1987) C4 photosynthesis: a unique blend of modified biochemistry, anatomy and 766 ultrastructure. Biochimica et Biophysica Acta (BBA) - Reviews on Bioenergetics, 895(2), 81– 767 106. Available at https://doi.org/10.1016/S0304-4173(87)80009-5 768 Hatch, M. D., & Burnell, J. N. (1990). Carbonic anhydrase activity in leaves and its role in the 769 first step of c(4) photosynthesis. Plant physiology, 93(2), 825–828. 770 https://doi.org/10.1104/pp.93.2.825 771 Hennacy, J. H., & Jonikas, M. C. (2020). Prospects for Engineering Biophysical CO2 772 Concentrating Mechanisms into Land Plants to Enhance Yields. Annual review of plant 773 biology, 71, 461–485. https://doi.org/10.1146/annurev-arplant-081519-040100 774 Hoang, N. V., Sogbohossou, E. O. D., Xiong, W., Simpson, C. J. C., Singh, P., Walden, N., van 775 den Bergh, E., Becker, F. F. M., Li, Z., Zhu, X. G., Brautigam, A., Weber, A. P. M., van Haarst, 776 J. C., Schijlen, E. G. W. M., Hendre, P. S., Van Deynze, A., Achigan-Dako, E. G., Hibberd, J. 777 M., & Schranz, M. E. (2023). The Gynandropsis gynandra genome provides insights into 778 whole-genome duplications and the evolution of C4 photosynthesis in Cleomaceae. The Plant 779 cell, 35(5), 1334–1359. https://doi.org/10.1093/plcell/koad018 780 Holst, F., Bolger, A. M., Kindel, F., Günther, C., Maß, J., Triesch, S., Kiel, N., Saadat, N., 781 Ebenhöh, O., Usadel, B., Schwacke, R., Weber, A. P. M., Bolger, M. E., & Denton, A. K. 782 (2026). Helixer: ab initio prediction of primary eukaryotic gene models combining deep 783 learning and a hidden Markov model. Nature methods, 23(4), 732–739. 784 https://doi.org/10.1038/s41592-025-02939-1 785 Huang, C. H., Sun, R., Hu, Y., Zeng, L., Zhang, N., Cai, L., Zhang, Q., Koch, M. A., Al 786 Shehbaz, I., Edger, P. P., Pires, J. C., Tan, D. Y., Zhong, Y., & Ma, H. (2016a). Resolution of 787 Brassicaceae Phylogeny Using Nuclear Genes Uncovers Nested Radiations and Supports 25 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 788 Convergent Morphological Evolution. Molecular biology and evolution, 33(2), 394–412. 789 https://doi.org/10.1093/molbev/msv226 790 Huang, C. H., Zhang, C., Liu, M., Hu, Y., Gao, T., Qi, J., & Ma, H. (2016b). Multiple 791 Polyploidization Events across Asteraceae with Two Nested Events in the Early History 792 [Revealed by Nuclear Phylogenomics. Molecular biology and evolution, 33(11), 2820–2835.](https://doi.org/10.1038/s41438-020-00422-w) 793 https://doi.org/10.1093/molbev/msw157 794 Huang, C. F., Liu, W. Y., Lu, M. J., Chen, Y. H., Ku, M. S. B., & Li, W. H. (2021). Whole 795 [Genome Duplication Facilitated the Evolution of C4 Photosynthesis in Gynandropsis gynandra.](https://doi.org/10.1111/nph.19267) 796 Molecular biology and evolution, 38(11), 4715–4731. https://doi.org/10.1093/molbev/msab200 797 Huang, L., Ma, Y., Jiang, J., Li, T., Yang, W., Zhang, L., Wu, L., Feng, L., Xi, Z., Xu, X., Liu, 798 J., & Hu, Q. (2020). A chromosome-scale reference genome of Lobularia maritima, an 799 ornamental plant with high stress tolerance. Horticulture research, 7(1), 197. 800 https://doi.org/10.1038/s41438-020-00422-w 801 Hughes, T. E., Sedelnikova, O. V., Wu, H., Becraft, P. W., & Langdale, J. A. (2019). Redundant 802 SCARECROW genes pattern distinct cell layers in roots and leaves of maize. Development 803 (Cambridge, England), 146(14), dev177543. https://doi.org/10.1242/dev.177543 804 Iohannes, S. D., & Jackson, D. (2023). Tackling redundancy: genetic mechanisms underlying 805 paralog compensation in plants. The New phytologist, 240(4), 1381–1389. 806 https://doi.org/10.1111/nph.19267 807 Ishikawa, A., Stuart, Y. E., Bolnick, D. I., & Kitano, J. (2021). Copy number variation of a fatty 808 acid desaturase gene Fads2 associated with ecological divergence in freshwater stickleback 809 populations. Biology letters, 17(8), 20210204. https://doi.org/10.1098/rsbl.2021.0204 810 Jiao, Y., Wickett, N. J., Ayyampalayam, S., Chanderbali, A. S., Landherr, L., Ralph, P. E., 811 Tomsho, L. P., Hu, Y., Liang, H., Soltis, P. S., Soltis, D. E., Clifton, S. W., Schlarbaum, S. E., 812 Schuster, S. C., Ma, H., Leebens-Mack, J., & dePamphilis, C. W. (2011). Ancestral polyploidy 813 in seed plants and angiosperms. Nature, 473(7345), 97–100. 814 [https://doi.org/10.1038/nature09916](https://doi.org/10.1038/nature09916) 815 Jones, P., Binns, D., Chang, H. Y., Fraser, M., Li, W., McAnulla, C., McWilliam, H., Maslen, 816 J., Mitchell, A., Nuka, G., Pesseat, S., Quinn, A. F., Sangrador-Vegas, A., Scheremetjew, M., 817 Yong, S. Y., Lopez, R., & Hunter, S. (2014). InterProScan 5: genome-scale protein function 26 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 818 classification. Bioinformatics (Oxford, England), 30(9), 1236–1240. 819 https://doi.org/10.1093/bioinformatics/btu031 820 Kaste, J. A. M., Walker, B. J., & Shachar-Hill, Y. (2024). Biophysical carbon concentrating 821 mechanisms in land plants: insights from reaction-diffusion modeling. bioRxiv: the preprint 822 server for biology, 2024.01.04.574220. https://doi.org/10.1101/2024.01.04.574220 823 Katoh, K., & Standley, D. M. (2013). MAFFT multiple sequence alignment software version 824 7: improvements in performance and usability. Molecular biology and evolution, 30(4), 772– 825 780. https://doi.org/10.1093/molbev/mst010 826 Kiefer, C., Willing, E. M., Jiao, W. B., Sun, H., Piednoël, M., Hümann, U., Hartwig, B., Koch, 827 M. A., & Schneeberger, K. (2019). Interspecies association mapping links reduced CG to TG 828 substitution rates to the loss of gene-body methylation. Nature plants, 5(8), 846–855. 829 https://doi.org/10.1038/s41477-019-0486-9 830 Kozlov, A. M., Darriba, D., Flouri, T., Morel, B., & Stamatakis, A. (2019). RAxML-NG: a fast, 831 scalable and user-friendly tool for maximum likelihood phylogenetic inference. Bioinformatics 832 (Oxford, England), 35(21), 4453–4455. https://doi.org/10.1093/bioinformatics/btz305 833 Kolmogorov, M., Yuan, J., Lin, Y., & Pevzner, P. A. (2019). Assembly of long, error-prone 834 reads using repeat graphs. Nature biotechnology, 37(5), 540–546. 835 https://doi.org/10.1038/s41587-019-0072-8 836 Landis, J. B., Soltis, D. E., Li, Z., Marx, H. E., Barker, M. S., Tank, D. C., & Soltis, P. S. (2018). 837 Impact of whole-genome duplication events on diversification rates in angiosperms. American 838 journal of botany, 105(3), 348–363. https://doi.org/10.1002/ajb2.1060 839 Lanfear, R., Frandsen, P. B., Wright, A. M., Senfeld, T., & Calcott, B. (2017). PartitionFinder 840 2: New Methods for Selecting Partitioned Models of Evolution for Molecular and 841 Morphological Phylogenetic Analyses. Molecular biology and evolution, 34(3), 772–773. 842 https://doi.org/10.1093/molbev/msw260 843 Li, F., Fan, G., Lu, C., Xiao, G., Zou, C., Kohel, R. J., Ma, Z., Shang, H., Ma, X., Wu, J., Liang, 844 X., Huang, G., Percy, R. G., Liu, K., Yang, W., Chen, W., Du, X., Shi, C., Yuan, Y., Ye, W., 845 … Yu, S. (2015a). Genome sequence of cultivated Upland cotton (Gossypium hirsutum TM-1) 846 provides insights into genome evolution. Nature biotechnology, 33(5), 524–530. 847 https://doi.org/10.1038/nbt.3208 27 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 848 [Ludwig M. (2011). The molecular evolution of β-carbonic anhydrase in Flaveria. Journal of](https://doi.org/10.1093/jxb/eru186) 849 experimental botany, 62(9), 3071–3081. https://doi.org/10.1093/jxb/err071 850 Lundgren, M. R., Osborne, C. P., & Christin, P. A. (2014). Deconstructing Kranz anatomy to 851 understand C4 evolution. Journal of experimental botany, 65(13), 3357–3369. 852 https://doi.org/10.1093/jxb/eru186 853 Lye, Z. N., & Purugganan, M. D. (2019). Copy Number Variation in Domestication. Trends in 854 plant science, 24(4), 352–365. https://doi.org/10.1016/j.tplants.2019.01.003 855 Lynch, M., and Force, A. (2000). The probability of duplicate gene preservation by 856 subfunctionalization. Genetics 154, 459–473 857 Mallmann, J., Heckmann, D., Bräutigam, A., Lercher, M. J., Weber, A. P., Westhoff, P., & 858 Gowik, U. (2014). The role of photorespiration during the evolution of C4 photosynthesis in 859 the genus Flaveria. eLife, 3, e02478. https://doi.org/10.7554/eLife.02478 860 [McKown, A. D., & Dengler, N. G. (2007). Key innovations in the evolution of Kranz anatomy](https://doi.org/10.1038/s41586-019-1693-2) 861 and C4 vein pattern in Flaveria (Asteraceae). American journal of botany, 94(3), 382–399. 862 https://doi.org/10.3732/ajb.94.3.382 863 Monson RK. (2003). Gene duplication, neofunctionalization, and the evolution of C4 864 photosynthesis. Int J Plant Sci. 164(S3): S43–S54 865 Moore, B. D., Franceschi, V. R., Cheng, S. H., Wu, J., & Ku, M. S. (1987). Photosynthetic 866 Characteristics of the C(3)-C(4) Intermediate Parthenium hysterophorus. Plant physiology, 867 85(4), 978–983. https://doi.org/10.1104/pp.85.4.978 868 Nosil, P., Feder, J. L., Flaxman, S. M., & Gompert, Z. (2017). Tipping points in the dynamics 869 of speciation. _Nature ecology & evolution_, _1_ (2), 1. https://doi.org/10.1038/s41559-016-0001 870 Ogren W.L. (1984). Photorespiration - Pathways, regulation, and modification. Annu. Rev. 871 Plant Physiol. Plant Mol. Biol. 35: 415–442 872 Ohno, S. (1970). Evolution by Gene Duplication. New York: Springer-Verlag. 873 One Thousand Plant Transcriptomes Initiative (2019). One thousand plant transcriptomes and 874 the phylogenomics of green plants. Nature, 574(7780), 679–685. 875 https://doi.org/10.1038/s41586-019-1693-2 28 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 876 Ou, S., Su, W., Liao, Y., Chougule, K., Agda, J. R. A., Hellinga, A. J., Lugo, C. S. B., Elliott, 877 T. A., Ware, D., Peterson, T., Jiang, N., Hirsch, C. N., & Hufford, M. B. (2019). Benchmarking 878 transposable element annotation methods for creation of a streamlined, comprehensive pipeline. 879 Genome biology, 20(1), 275. https://doi.org/10.1186/s13059-019-1905-y 880 Padhi, S. K., & Panda, D. (2026). Prospects and challenges of neglected oilseed crop Niger 881 [Guizotia abyssinica (L.f.) Cass] for nutrition, health and industry. Oil Crop Science, 11(2), 882 117–128. https://doi.org/10.1016/j.ocsci.2026.05.001 883 Pajic, P., Pavlidis, P., Dean, K., Neznanova, L., Romano, R. A., Garneau, D., Daugherity, E., 884 Globig, A., Ruhl, S., & Gokcumen, O. (2019). Independent amylase gene copy number bursts 885 correlate with dietary preferences in mammals. eLife, 8, e44628. 886 https://doi.org/10.7554/eLife.44628 887 Panchy, N., Lehti-Shiu, M., & Shiu, S. H. (2016). Evolution of Gene Duplication in Plants. 888 Plant physiology, 171(4), 2294–2316. https://doi.org/10.1104/pp.16.00523 889 Qiao, X., Zhang, S., & Paterson, A. H. (2022). Pervasive genome duplications across the plant 890 tree of life and their links to major evolutionary innovations and transitions. Computational and 891 structural biotechnology journal, 20, 3248–3256. https://doi.org/10.1016/j.csbj.2022.06.026 892 Reyna-Llorens, I., & Hibberd, J. M. (2017). Recruitment of pre-existing networks during the 893 evolution of C4 photosynthesis. Philosophical transactions of the Royal Society of London. 894 Series B, Biological sciences, 372(1730), 20160386. https://doi.org/10.1098/rstb.2016.0386 895 Rochaix J. D. (2011). Regulation of photosynthetic electron transport. Biochimica et biophysica 896 acta, 1807(3), 375–383. https://doi.org/10.1016/j.bbabio.2010.11.010 897 Sage R. F. (2004). The evolution of C4 photosynthesis. The New phytologist, 161(2), 341–370. 898 https://doi.org/10.1111/j.1469-8137.2004.00974.x 899 Sage, R. F., Christin, P. A., & Edwards, E. J. (2011). The C(4) plant lineages of planet Earth. 900 Journal of experimental botany, 62(9), 3155–3169. https://doi.org/10.1093/jxb/err048 901 Sage, R. F., Sage, T. L., & Kocacinar, F. (2012). Photorespiration and the evolution of C4 902 photosynthesis. Annual review of plant biology, 63, 19–47. https://doi.org/10.1146/annurev 903 arplant-042811-105511 29 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 904 Sage, R. F., Khoshravesh, R., & Sage, T. L. (2014). From proto-Kranz to C4 Kranz: building 905 [the bridge to C4 photosynthesis. Journal of experimental botany, 65(13), 3341–3356.](https://doi.org/10.1093/jxb/erad250) 906 https://doi.org/10.1093/jxb/eru180 907 [Schlüter, U., Bouvier, J. W., Guerreiro, R., Malisic, M., Kontny, C., Westhoff, P., Stich, B., &](https://doi.org/10.1105/tpc.113.114520) 908 Weber, A. P. M. (2023). Brassicaceae display variation in efficiency of photorespiratory 909 carbon-recapturing mechanisms. Journal of experimental botany, 74(21), 6631–6649. 910 https://doi.org/10.1093/jxb/erad250 911 Schulze, S., Mallmann, J., Burscheidt, J., Koczor, M., Streubel, M., Bauwe, H., Gowik, U., & 912 Westhoff, P. (2013). Evolution of C4 photosynthesis in the genus flaveria: establishment of a 913 photorespiratory CO2 pump. The Plant cell, 25(7), 2522–2535. 914 https://doi.org/10.1105/tpc.113.114520 915 Schwacke, R., Ponce-Soto, G. Y., Krause, K., Bolger, A. M., Arsova, B., Hallab, A., Gruden, 916 K., Stitt, M., Bolger, M. E., & Usadel, B. (2019). MapMan4: A Refined Protein Classification 917 and Annotation Framework Applicable to Multi-Omics Data Analysis. Molecular plant, 12(6), 918 879–892. https://doi.org/10.1016/j.molp.2019.01.003 919 Seppey, M., Manni, M., & Zdobnov, E. M. (2019). BUSCO: Assessing Genome Assembly and 920 Annotation Completeness. Methods in molecular biology (Clifton, N.J.), 1962, 227–245. 921 https://doi.org/10.1007/978-1-4939-9173-0_14 922 Shaar-Moshe, L., & Brady, S. M. (2023). SHORT-ROOT and SCARECROW homologs 923 regulate patterning of diverse cell types within and between species. The New phytologist, 924 237(5), 1542–1549. https://doi.org/10.1111/nph.18654 925 Sharkey TD. (1988). Estimating the rate of photorespiration in leaves. Physiologia Plantarum 926 73, 147–152 927 Siniscalchi, C. M., Edwards, R. D., Gomez, J. L., Moore, E.R., Mandel, J.R. (2021). 928 Photosynthesis metabolism in the Compositae: current knowledge and future directions. Taxon 929 70: 339–350. https://doi.org/10.1002/tax.12426 930 Slewinski, T. L., Anderson, A. A., Zhang, C., & Turgeon, R. (2012). Scarecrow plays a role in 931 establishing Kranz anatomy in maize leaves. Plant & cell physiology, 53(12), 2030–2037. 932 https://doi.org/10.1093/pcp/pcs147 30 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 933 Soltis, P. S., & Soltis, D. E. (2016). Ancient WGD events as drivers of key innovations in 934 angiosperms. Current opinion in plant biology, 30, 159–165. 935 https://doi.org/10.1016/j.pbi.2016.03.015 936 Sun, T., Bonthala, V. S., Stich, B., Luginbuehl, L. H., & Hibberd, J. M. (2025). Conserved 937 spatial patterning of gene expression in independent lineages of C 4 plants. _The New_ 938 _phytologist_, _249_ (1), 24–38. https://doi.org/10.1111/nph.70475 939 Suryawanshi, V., Talke, I. N., Weber, M., Eils, R., Brors, B., Clemens, S., & Krämer, U. (2016). 940 Between-species differences in gene copy number are enriched among functions critical for 941 adaptive evolution in Arabidopsis halleri. BMC genomics, 17(Suppl 13), 1034. 942 https://doi.org/10.1186/s12864-016-3319-5 943 Suyama, M., Torrents, D., & Bork, P. (2006). PAL2NAL: robust conversion of protein 944 sequence alignments into the corresponding codon alignments. Nucleic acids research, 34(Web 945 Server issue), W609–W612. https://doi.org/10.1093/nar/gkl315 946 Taniguchi, Y. Y., Gowik, U., Kinoshita, Y., Kishizaki, R., Ono, N., Yokota, A., Westhoff, P., 947 & Munekage, Y. N. (2021). Dynamic changes of genome sizes and gradual gain of cell-specific 948 distribution of C4 enzymes during C4 evolution in genus Flaveria. The plant genome, 14(2), 949 e20095. https://doi.org/10.1002/tpg2.20095 950 Tomato Genome Consortium (2012). The tomato genome sequence provides insights into 951 fleshy fruit evolution. Nature, 485(7400), 635–641. https://doi.org/10.1038/nature11119 952 Van de Peer, Y., Mizrachi, E., & Marchal, K. (2017). The evolutionary significance of 953 polyploidy. Nature reviews. Genetics, 18(7), 411–424. https://doi.org/10.1038/nrg.2017.26 954 Walker, B. J., VanLoocke, A., Bernacchi, C. J., & Ort, D. R. (2016). The Costs of 955 Photorespiration to Food Production Now and in the Future. _Annual review of plant biology_, _67_, 956 107–129. https://doi.org/10.1146/annurev-arplant-043015-111709 957 Weisenfeld, N. I., Kumar, V., Shah, P., Church, D. M., & Jaffe, D. B. (2017). Direct 958 determination of diploid genome sequences. Genome research, 27(5), 757–767. 959 https://doi.org/10.1101/gr.214874.116 960 Wilhoit, K., Yamanouchi, S., Chen, B. J., Yamasaki, Y. Y., Ishikawa, A., Inoue, J., Iwasaki, 961 W., & Kitano, J. (2025). Convergent Evolution and Predictability of Gene Copy Numbers 31 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 962 Associated with Diets in Mammals. Genome biology and evolution, 17(2), evaf008. 963 https://doi.org/10.1093/gbe/evaf008 964 Xin, Z., & Chen, J. (2012). A high throughput DNA extraction method with high yield and 965 quality. _Plant methods_, _8_ (1), 26. https://doi.org/10.1186/1746-4811-8-26 966 Yandell, M., & Ence, D. (2012). A beginner's guide to eukaryotic genome annotation. Nature 967 reviews. Genetics, 13(5), 329–342. https://doi.org/10.1038/nrg3174 968 Yang Z. (2007). PAML 4: phylogenetic analysis by maximum likelihood. Molecular biology 969 and evolution, 24(8), 1586–1591. https://doi.org/10.1093/molbev/msm088 970 Young, N. D., Debellé, F., Oldroyd, G. E., Geurts, R., Cannon, S. B., Udvardi, M. K., Benedito, 971 V. A., Mayer, K. F., Gouzy, J., Schoof, H., Van de Peer, Y., Proost, S., Cook, D. R., Meyers, 972 B. C., Spannagl, M., Cheung, F., De Mita, S., Krishnakumar, V., Gundlach, H., Zhou, S., … 973 Roe, B. A. (2011). The Medicago genome provides insight into the evolution of rhizobial 974 symbioses. Nature, 480(7378), 520–524. https://doi.org/10.1038/nature10625 975 Yu King Hing, N., Aryal, U. K., & Morgan, J. A. (2021). Probing Light-Dependent Regulation 976 of the Calvin Cycle Using a Multi-Omics Approach. Frontiers in plant science, 12, 733122. 977 https://doi.org/10.3389/fpls.2021.733122 978 Zhang, C., Huang, C. H., Liu, M., Hu, Y., Panero, J. L., Luebert, F., Gao, T., & Ma, H. (2021). 979 Phylotranscriptomic insights into Asteraceae diversity, polyploidy, and morphological 980 innovation. Journal of integrative plant biology, 63(7), 1273–1293. 981 https://doi.org/10.1111/jipb.13078 982 Zhang, C., & Mirarab, S. (2022). ASTRAL-Pro 2: ultrafast species tree reconstruction from 983 multi-copy gene family trees. Bioinformatics (Oxford, England), 38(21), 4949–4950. 984 https://doi.org/10.1093/bioinformatics/btac620 985 32 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 986 **Figures** 987 988 989 990 **Figure 1: Species tree of 38 species belonging to** _**Heliantheae alliance**_ **,** _**Anthemideae**_ **and** 991 _**Inuleae**_ **tribes of** _**the Asteraceae**_ **family.** The focal species names are coloured according to 992 their genus, while the tip-points are coloured according to their photosynthetic phenotype. All 993 nodes showed a bootstrap of 100, otherwise mentioned in grey below the nodes. The internal 994 nodes with brown triangles (F1, F2 and F3) indicate fossil calibrations obtained from the 995 [TimeTree database (http://www.timetree.org) used in speciation dating by MCMCtree in the](http://www.timetree.org/) 996 PAML v4.10.7 package (Yang, 2007). The numbers in black above the nodes indicate the 997 estimated divergence time in millions of years ago (MYA). The star in black indicates a shared 998 whole-genome duplication (WGD) event called WGD-2, which occurred 30.84 million years 999 ago (MYA) in the ancestors of the _Heliantheae alliance_ species. The stars in green and orange 1000 colours indicate the additional and recent WGD events that occurred in respective species. The 1001 stars in purple and red colours indicate distinct WGD events that occurred in the respective 1002 species. The green rectangles above the nodes indicate WGD events shared by the descendant 1003 taxa. The first position indicates the node ID, and the second indicates the proportion of 1004 duplicate gene families shared by the descendant taxa in the rectangle. 1005 33 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1006 1007 1008 **Figure 2: Phylogenomic distribution of CO** **2** **compensation point (CCP)-associated** 1009 **MapMan BINs for photosynthesis and transcription factor families.** This integrated 1010 analysis depicts the distribution of gene copy number variations (CNVs) across the species tree 1011 (derived from Figure 1) for 43 CCP-associated functional categories. The heatmap is partitioned 1012 into two vertical blocks: 12 MapMan BINs related to **photosynthesis** (left) and 31 BINs related 1013 to **transcription factor families** (right). Heatmap cell values represent normalized gene copy 1014 numbers. Species-specific physiological traits are displayed alongside the tree, including 1015 photosynthetic types (colored points) and CCP values (proportional circle sizes). Grey cells in 1016 the heatmap indicate taxa excluded from the PAM analysis due to the unavailability of CCP 1017 values. 1018 1019 34 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1020 1021 1022 **Figure 3: Genomic tipping points characterize the C** **3** **-to-C** **4** **physiological transition.** **(A)** 1023 **Decoupling of linearity and explanatory power.** Total effect sizes (η [2] ) plotted against linear 1024 R [2] for 451 MapMan BINs significantly associated with CCP values. The five most extreme 1025 'tipping points' (highest η [2] - R [2] gap) are highlighted: transcription factor BINs: 14.5.3.1.1 (HD1026 ZIP), 14.5.3.4 (HSF), 14.5.2.5 (LOB), 14.5.7.7.1 (TCP), and 50.1.3 (Oxidoreductase). Their 1027 position in the upper-left quadrant indicates that they are massive drivers of CCP variation that 1028 cannot be explained by linear models. **(B) Magnitude-Impact Scaling.** Correlation between 1029 the scale of genomic change ( _d_ ) and driver power (η [2] ). The significant positive trend confirms 1030 that larger shifts in gene copy number are the primary engine behind high-magnitude 1031 physiological shifts. 1032 1033 35 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1034 1035 **Figure 4:** A) Distribution of genomic prediction performance metrics (out-of-sample R² and 1036 root mean square error RMSE) across feature set, and B) validation schemes. C) Normalized 1037 squared rrBLUP effects for PAM-significant and background gene-family predictors. D) Top 1038 50 CCP predictors across feature sets (HOGs and BINs): ranked effects and overlap. 1039 36 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1040 1041 1042 **Figure 5: Genomic contribution of the WGD-2 event to CCP-associated MapMan BINs.** 1043 This integrated figure illustrates the proportion of CCP-associated CNVs originating from the 1044 WGD-2 event, mapped across the Asteraceae phylogeny (based on the tree topology in Figure 1045 1). The heatmap displays the retention patterns for functional categories associated to the CO 2 1046 compensation point (CCP), partitioned into **photosynthesis-related BINs** (left) and 1047 **transcription factor families** (right). The heatmap gradient represents the percentage of the 1048 total CNVs within each BIN that is traceable to the WGD-2 event. Gray cells represent taxa 1049 excluded from the PAM and WGD analyses due to missing physiological CCP data or the 1050 absence of discernible WGD-2 genomic peak. 1051 1052 37 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1053 **Supplementar Figures** 1054 1055 **Figure S1: Assembly quality and genomic annotations.** The bar plot displays the assembly 1056 quality and genomic annotations for the newly assembled 35 species. Assembly size in 1057 gigabases (A), Assembly contiguity (log10 N50 (Kb)) (B), Percentage of assembly 1058 completeness (BUSCO scores) (C), Repetitive sequences (D), Genic regions (E), and Assembly 1059 gaps (F) of the assembly of each species. Green bars represent species with assemblies 1060 (reference-quality) generated using short-read and mate-pair libraries, followed by scaffolding 1061 with 10X linked reads using DeNovoMAGIC (NRGene, Nes Ziona, Israel). Purple bars 1062 represent species assembled with a single stLFR library using Supernova v2.1.1 (Weisenfeld et 1063 al., 2017), and orange bars represent species assembled with an Oxford Nanopore long-read 1064 library using Flye v2.9 (Kolmogorovet al., 2019) followed by scaffolded with a single stLFR 1065 linked-read library. The species are sorted based on the descending order of N50 values. 1066 1067 **Figure S2: BUSCO completeness across 35 species assemblies.** The horizontally stacked bar 1068 plot displays the BUSCO (Benchmarking Universal Single-Copy Orthologs) completeness 1069 scores for the newly assembled 35 species, assessing the quality of their genome assemblies. 1070 Each bar represents a species and is colour-coded to reflect the proportion of complete (light 1071 blue), duplicated (blue), fragmented (yellow), and missing (red) BUSCO genes. 1072 1073 **Figure S3: Distribution of transposable elements across species.** The horizontally stacked 1074 bar plot shows the composition of the top ten transposable elements (TE) groups across 35 1075 species. Each bar represents a species, with the width of each coloured segment corresponding 1076 to the percentage of the assembly occupied by a specific TE group. The top ten TE groups are 1077 colour-coded and include elements such as Copia, Gypsy, the long-terminal repeats (LTRs), 1078 Helitrons and Mutators, among others. 1079 1080 **Figure S4: Comparative genomic features across species assemblies.** The horizontal bar plot 1081 consists of seven subplots, each representing a different genomic feature for the newly 1082 assembled 35 species and _Flaveria brownii_ (Cyan colour bar). The features include (A) the total 1083 number of genes, (B) the mean gene length in base pairs, the percentage of genes with UniProt 1084 descriptions (C), PFAM protein domains (D), Gene Ontology terms (E), Expressed (F), and 1085 MapMan Ontology BINs (G). We considered genes with transcripts per million (TPM) of 0.5 1086 or greater as expressed or those with TPM values greater than 0 in the case of gene expression 1087 measured using direct RNA sequencing data (as indicated by "*"). The bar with "**" indicates 1088 that the publicly available RNA-seq data was used to calculate the expression. The bars are 1089 colour-coded, and species are sorted as in Supplementary Figure S1. 1090 1091 **Figure S5: Correlation between genome size, transposable elements (TEs), and gene** 1092 **content across 35 species.** The two plots display the relationships between genome size and 38 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1093 two genomic features: (A) a solid positive correlation (R²: 0.995) is observed between the 1094 increase in transposable elements (TEs) and the increase in genome size, indicating that TE 1095 expansion is a significant contributor to genome size variation. (B) A moderate correlation (R²: 1096 0.544) is observed between gene content and genome size, suggesting that gene content plays 1097 a lesser role in genome size changes than TEs. In both plots, species are grouped and colour1098 coded by genus, highlighting the trends across taxonomic groups. 1099 1100 **Figure S6: Phylogenetic Hierarchical Orthogroups (HOGs) across species.** This figure 1101 displays an overview of HOGs and their distribution across multiple species. It provides insights 1102 into the percentage of genes assigned, HOG counts, and the types of HOGs identified by 1103 OrthoFinder v2.5.5 **(Emms & Kelly, 2019)** . A) Percentage of genes in each species assigned 1104 to HOGs. B) Total number of HOGs identified for each species. C) Distribution of HOG types 1105 across species: This stacked bar plot represents the proportion of different HOG types in each 1106 species. We categorised HOGs into three groups - HOGs containing all species, HOGs 1107 containing more than one species but not all, and species-specific HOGs, i.e., HOGs with just 1108 one species. D) Distribution of genes in HOG types: Similar to subplot 3, this plot focuses on 1109 the distribution of genes in HOGs. It shows the percentage of genes associated with HOGs that 1110 are either present in all species, shared across some species, or species-specific. 1111 1112 **Figure S7: Pairwise heatmap of orthologous gene sharing between 38 species.** This heatmap 1113 represents the percentage of genes shared as orthologs between each pair of species, as 1114 determined by OrthoFinder v2.5.5 (Emms & Kelly, 2019). Each cell in the heatmap shows the 1115 percentage of orthologous genes shared between two species, ranging from low (red) to high 1116 (blue). Higher percentages suggest closer evolutionary relationships or similar gene content, 1117 while lower percentages indicate more divergent genomes or gene content. 1118 1119 **Figure S8: Estimation of Species tree using ASTRAL-Pro3.** We reconstructed a species tree 1120 using ASTRAL-Pro3 (Zhang & Mirarab, 2022), incorporating gene trees constructed by 1121 RAxML-ng (Kozlov et al., 2019) of 5,618 Hierarchical Orthogroups (HOGs) that contain all 1122 38 species. Quartet scores at each internal node are represented by pie charts, illustrating the 1123 relative proportions of the three possible quartet relationships: Q1 (Concordance), Q2 1124 (Alternative), and Q3 (Conflict). Each pie chart visually depicts the consensus among the gene 1125 trees. All internal nodes have a posterior probability (PPP) of 1, indicating strong support for 1126 the branching structure. 1127 1128 **Figure S9: Ks distribution density with predicted peaks for 38 Species.** This figure presents 1129 the density distributions of synonymous substitution rates (Ks) computed using wgd v2 (Chen 1130 et al., 2024) across 38 species, with each subplot representing a single species. We identified 1131 the predicted peaks of whole-genome duplication events using Gaussian Mixture Modelling 1132 (GMM). To assess the robustness of these predictions, we performed resampling 1,000 times, 39 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.05.764543; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.05.764543) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1133 each time selecting half of the Ks values, to calculate 90% confidence intervals for each peak. 1134 Subplots are colour-coded according to the genus of each species. The density plots highlight 1135 the distribution of Ks values, indicating possible whole-genome duplication (WGD) or other 1136 duplication events across species. 1137 1138 **Figure S10: Representation of whole-genome duplication-2 in Ks distribution.** The WGD 1139 peak range corresponds to the minimum and maximum Ks of peaks identified across species. 1140 The red and purple lines indicate the combined lower and upper bounds of 90% confidence 1141 intervals (CI) of the WGD peak across all focal species. 1142 1143 **Figure S11: Duplicated gene pairs derived from whole-genome duplication-2 (WGD-2).** 1144 The bar plot displays (A) the percentage of duplicated genes derived from whole-genome 1145 duplication-2 (Ks = 0.41-0.68). The Chi-square test revealed a statistically significant 1146 difference in the proportion of duplicated gene pairs derived from WGD-2 across species (χ² = 1147 121, p-value: 4.03 × 10⁻¹¹). (B) The proportion of duplicated gene pairs derived from WGD-2 1148 sharing the same or different hierarchical orthologous groups (HOG; χ² = 183, p = 1.40 × 10⁻²¹). 1149 (C) The proportion of duplicated gene pairs derived from WGD-2 sharing the same or different 1150 MapMan BIN (χ² = 79.3, p = 4.30 × 10⁻⁵). Here, _Artemisia vulgaris_ and _Pallenis spinosa_ were 1151 excluded from the analysis because both lack a WGD-2 peak within the Ks range of 0.41-0.68. 1152 1153 **Figure S12: Association between Gene Copy Numbers of BINs related to photosynthesis** 1154 **and Carbon Compensation Point (CCP).** Boxplot showing the distribution of carbon 1155 compensation point (CCP) values (Y-axis) across species grouped by gene copy number 1156 variation (X-axis). Each boxplot represents one MapMan BIN associated with CCP. 1157 1158 **Figure S13: Association between Gene Copy Numbers of BINs related to transcription** 1159 **factors and Carbon Compensation Point (CCP).** Boxplot showing the distribution of carbon 1160 compensation point (CCP) values (Y-axis) across species grouped by gene copy number 1161 variation (X-axis). Each boxplot represents one MapMan BIN associated with CCP. 40