Genetic and developmental constraints drive parallelism in flower evolution

Regulatory architecture and standing variation drive parallelism in floral evolution | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Regulatory architecture and standing variation drive parallelism in floral evolution View ORCID Profile Kevin Sartori , View ORCID Profile Natalia Wozniak , Anahid Powell , View ORCID Profile Ushio Fujikura , View ORCID Profile Christian Kappel , View ORCID Profile Tian-Feng Lü , View ORCID Profile Yang Dong , View ORCID Profile Stefanie Rosa , View ORCID Profile Michael Lenhard , View ORCID Profile Adrien Sicard doi: https://doi.org/10.1101/2025.10.28.684842 Kevin Sartori 1 Uppsala Biocenter, Department of Plant Biology, Swedish University of Agricultural Sciences , BOX 7080, 750 07, Uppsala, Sweden Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Kevin Sartori Natalia Wozniak 2 Institut für Biochemie und Biologie, Universität Potsdam , Karl-Liebknecht-Str. 24-25, 14476 Potsdam-Golm, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Natalia Wozniak Anahid Powell 2 Institut für Biochemie und Biologie, Universität Potsdam , Karl-Liebknecht-Str. 24-25, 14476 Potsdam-Golm, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site Ushio Fujikura 2 Institut für Biochemie und Biologie, Universität Potsdam , Karl-Liebknecht-Str. 24-25, 14476 Potsdam-Golm, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Ushio Fujikura Christian Kappel 2 Institut für Biochemie und Biologie, Universität Potsdam , Karl-Liebknecht-Str. 24-25, 14476 Potsdam-Golm, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Christian Kappel Tian-Feng Lü 3 State Key Laboratory of Plant Diversity and Specialty Crops, Institute of Botany, Chinese Academy of Sciences , 20 Nanxincun, Xiangshan, Beijing 100093, China 4 China National Botanical Garden , Beijing 100093, China Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Tian-Feng Lü Yang Dong 3 State Key Laboratory of Plant Diversity and Specialty Crops, Institute of Botany, Chinese Academy of Sciences , 20 Nanxincun, Xiangshan, Beijing 100093, China 4 China National Botanical Garden , Beijing 100093, China Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Yang Dong Stefanie Rosa 1 Uppsala Biocenter, Department of Plant Biology, Swedish University of Agricultural Sciences , BOX 7080, 750 07, Uppsala, Sweden Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Stefanie Rosa Michael Lenhard 2 Institut für Biochemie und Biologie, Universität Potsdam , Karl-Liebknecht-Str. 24-25, 14476 Potsdam-Golm, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Michael Lenhard Adrien Sicard 1 Uppsala Biocenter, Department of Plant Biology, Swedish University of Agricultural Sciences , BOX 7080, 750 07, Uppsala, Sweden Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Adrien Sicard For correspondence: adrien.sicard{at}slu.se Abstract Full Text Info/History Metrics Preview PDF Abstract Evolution repeatedly gives rise to similar phenotypes, suggesting that constraints and biases make some evolutionary trajectories more likely than others, yet the mechanisms underlying this predictability remain poorly understood. Using repeated transitions to self-fertilization in a model Brassicaceae genus, we show that independent reductions in petal size arise from dosage changes at the same pleiotropic growth regulator, JAGGED ( JAG ). Petal development exhibits particularly high sensitivity to JAG dosage, resulting in organ-specific morphological shifts with minimal effects on other organs and limited pleiotropy. In both selfing species, these changes were drawn from functionally divergent variants whose polymorphism patterns are consistent with long-term maintenance under weak selection. Our results demonstrate how developmental network architecture and the maintenance of functional variation at pleiotropic loci combine to channel evolution toward predictable outcomes. Main Evolution often produces similar traits across unrelated lineages, suggesting that outcomes are shaped by predictable constraints rather than chance. Such convergent phenotypes often arise in species exposed to similar ecological pressures, implying that adaptive solutions are limited 1 . Beyond ecology, developmental constraints within Gene Regulatory Networks (GRNs) also restrict the range of viable phenotypes 2 , 3 . Because GRNs carry historical legacies, closely related species often share the same constraints and thus respond similarly to shifts in selection pressures 4 . Although specific cases of genetic convergence have been identified, the general mechanisms that drive repeated phenotypic evolution remain poorly understood 5 . Cross-pollination is common in plants and often enforced by self-incompatibility systems, which make reproduction dependent on animal pollinators 6 – 10 . To attract pollinators, plants invest in large floral displays 11 – 13 . Yet self-incompatibility has repeatedly broken down throughout evolution, producing self-fertilizing (selfing) lineages, likely as reproductive assurance when mates are scarce 14 . Remarkably, independent transitions to selfing often converge on similar floral changes, collectively termed the selfing syndrome, with a hallmark being a pronounced reduction in flower and petal size 6 , 14 . Genetic studies indicate that floral traits often evolve independently from one another or from vegetative traits, suggesting adaptive changes targeting specific organ dimensions 15 – 23 . Yet because most growth regulators in plants are pleiotropic, influencing multiple organs 24 , such specific changes likely depend on rare organ-specific regulators or on regulatory elements within pleiotropic genes. It remains, nevertheless, unclear which general growth regulators can be modulated at an organ-specific level and how such targeted changes in organ size occur. If such a regulation is limited, it could help explain phenotypic convergence following transitions to selfing. Beyond pleiotropy, the effect size and availability of adaptive variation in ancestral populations may further constrain evolution 25 . In this study, we exploit the Capsella genus as a model to study the convergent evolution of flowers after independent transitions to selfing 26 – 33 . In this genus, the selfing species C. rubella ( Cr ) and C. orientalis ( Co ) evolved independently from an outcrossing C. grandiflora ( Cg )-like ancestor following the breakdown of self-incompatibility, diverging ∼100 kya and within the last 2 million years, respectively ( Fig. 1a ) 28 , 34 – 36 . Both exhibit a fully established selfing syndrome, including a pronounced reduction in petal size ( Fig. 1 ) 37 . Download figure Open in new tab Fig. 1: Convergent evolution of petal size following the transition to selfing in Capsella . a , Schematic representation of the evolutionary history within the Capsella genus. b, Heatmap illustrating the distribution of cell size within the petals of the three Capsella species. Scale bar: 400 μm. c, Non-metric multidimensional scaling (NMDS) plot based on Bray-Curtis dissimilarities in cellular features among C. grandiflora , C. rubella , and C. orientalis petals (n = 5 per species). d, Average cell count per cell area bin (n = 5 per species). e, Cell counts within the proximal and distal regions of the petals. f, Representative images showing the distribution of cell division events across the epidermis of stage 1 petals. Fluorescently labelled nucleotide analogue 5-ethynyl-2’-deoxyuridine (EdU; green) was used to visualise newly synthesised DNA, and primary cell walls were stained with propidium iodide (PI; red). Scale bar: 50 μm. g, Quantification of EdU-labelled nuclei in stage 1 petals. Letters indicate significant differences as determined by one-way ANOVA with Tukey’s HSD test (α = 0.05). Convergent petal size reduction via similar reduction of distal cell proliferation A major phenotypic change in Cr and Co is a pronounced reduction in petal size, accompanied by fewer cells, implicating altered cell proliferation as a central mechanism 38 . To examine the developmental basis of these changes, we analyzed cellular patterns along the proximal-distal axis of petals, incorporating various descriptors of cell shape and directional growth ( Extended Data Table 1 ). Petals exhibit two distinct cell clusters. A proximal cluster of large, elongated cells and a distal cluster of small, rounded cells, separated by a transition zone where cell size and anisotropy gradually decrease ( Fig. 1B and Extended Data Fig. 1 ). While this pattern is largely conserved across species, cellular features along this axis can distinguish the petals of selfers from those of the outcrosser (one-way ANOVA: F (1,13) = 28.54, p = 0.00013, η² = 0.687) ( Fig. 1b, 1d , and Extended Data Fig. 1, and 2 ). Selfers differ primarily in the distal region, where cell number is reduced roughly five-fold compared to the outcrosser, resulting in a two-fold reduction in length and a three-fold decrease in width ( Fig. 1e and Extended Data Fig. 1c; Extended Data Table 2 ). Early petal growth occurs via cell proliferation, which initially takes place throughout the primordia. Once the petal primordia reach approximately 90 µm in length, proliferation rapidly becomes restricted to the distal region, thereby determining the size of the distal section, while cell elongation is largely confined to the proximal domain ( Fig. 1 and Extended Data Fig.3 ). In selfers, cell proliferation in the distal margin, as judged from the number of S-phase cells, is reduced three- to four-fold relative to the outcrosser, consistently across both lineages ( Fig. 1f and 1g ). Thus, despite independent origins, petal size reduction in these selfing species arises through the same developmental mechanism: restricted growth at the distal margin due to decreased cell division. Independent selfing lineages co-opted JAG to reduce petal size The convergence in petal development suggests that similar genetic mechanisms underlie size reduction in the two selfing species. Previous studies indicated that half of the loci associated with petal size are shared between Cr and Co , with PAQTL2 on chromosome 2 contributing most to phenotypic evolution ( Extended Data Fig. 4a ) 38 , 39 . To refine the location of PAQTL2 in each selfing species, we introgressed the outcrossing Cg alleles into each selfer, creating near-isogenic lines ( NILrg1 and NILog2 ) and screened for recombinants within this region ( Fig. 2 ). By analyzing petal size segregation in the progeny, we narrowed down the causal mutation (s) to 14.6 kb in Cr and 130 kb in Co . The most informative recombinants from NILrg1 indicated that the causal mutation(s) are situated between 9,061,779 and 9,076,392 bp on chromosome 2. We confirmed this region by generating a quasi-isogenic line ( qILrg ) for the 14.6-kb interval, where Cr homozygotes exhibited ∼20% smaller petals than Cg homozygotes, recapitulating the PAQTL2 effect ( Extended Data Fig. 4c and 7a ). In NILog2 , the most informative recombinant localized the causal mutation to 9,019,066–9,149,619 bp on chromosome 2, accounting for a 25% petal reduction in Co ( Fig. 2b and Extended Data Fig. 7a ). This region overlaps with the locus identified in Cr , confirming that the same genomic locus underpins PAQTL2 in both selfing species. Download figure Open in new tab Fig. 2. Allelic variation within JAG contribute to petal size reduction in indepdent selfing lineages. a and b Fine mapping of PA_QTL2 in C. rubella ( Cr ) and C.orientalis (Co) , respectively. The color coded bars on the left repsent the genotype of the most informative recombinants. The position on scaffold 2 are indicated on the top in base pair (bp). The bar chart on the right shows the average petal size for each alternative homozygous plants for each recombinant lines. Red filled bars indicate significant differences between alternative homozygous genotypes, P <0.05 (Student’s t-test). c and d Quantitative complementation assay illustrated with an interaction plot of the crosses between NILrg c or NILog d with jagr-1 +/-. Letters indicate significant differences as determined by one-way ANOVA with Tukey’s HSD test (α = 0.05). The 14.6-kb interval contained only two genes, including the Arabidopsis thaliana JAGGED ( JAG ) ortholog, a zinc-finger transcription factor, known to control lateral organ growth and distal cell proliferation ( Fig. 2a ) 40 . In an EMS-mutagenized C. rubella screen, we identified a mutant phenocopying the A. thaliana jag mutation ( Extended Data Fig. 6a ) 40 . Bulked-segregant analysis revealed a splice-site mutation in the third exon of JAG , preventing proper splicing and introducing a premature stop codon ( Extended Data Fig. 5 ). The defect was rescued by reintroducing a wild-type Cr allele, confirming JAG inactivation as causal ( Extended Data Fig. 6b ). We thus used this mutant in a quantitative complementation test 41 to test whether JAG underlies PAQTL2 in both selfers. In this assay, natural alleles were crossed to a jag+/– background, where functional differences between the natural alleles are expected to be amplified only when the loss-of-function mutation affects the same gene. Consistently, in both cases, the differences between the selfer and outcrosser alleles were stronger when combined with the mutant alleles ( Fig. 2c and d ). Linear model analysis revealed a significant interaction between the tested ( Cg1 and Cr or Cg2 and Co ) and control allele (WT and crjag ) ( β = −0.38976 ± 0.08710, t = −4.475, p = 8.56 × 10⁻ 6 for NILog ; and β = 0.311 ± 0.043, t = 7.245, p = 2.27 × 10⁻ 12 for qILrg ), indicating that polymorphism within JAG contributed to a reduction in petal size in both cases. To further confirm JAG’s role, we transformed genomic constructs corresponding to the Cg ( JAGg1 ) and Cr ( JAGr ) alleles segregating in NILrg , as well as the Cg ( JAGg2 ) and Co ( JAGo ) alleles from NILog , into the A. thaliana jag-2 mutant ( Extended Data Fig. 4d and e ). All alleles rescued the mutant phenotype, but Cg1 and Cg2 alleles produced larger petals than Cr and Co alleles, respectively. These results further support that allelic variation in JAG underlies convergent petal size reduction in both selfing species. JAG is essential to promote distal petal cell proliferation Inactivation of JAG in Cr causes pleiotropic effects, leading to a ≥10% reduction in the size of most organs, with particularly strong effects on leaves (30%), petals (60%) ( Extended Data Fig. 6 ), and fruit 42 . To uncover the developmental basis, we compared wild-type and crjag petals. Silhouette overlays showed that the proximal petal region was largely unaffected, whereas the distal region was almost entirely missing ( Fig. 3a and 3b ). Cellular analysis revealed a ∼5-fold reduction in total cell number ( Fig. 3c and 3d ). This reduction in cell number was particularly pronounced in the transition and distal zones ( Fig. 3d ). Cell elongation in the proximal region and the position of the transition zone were largely unaffected, indicating that overall petal growth patterning is not altered by JAG ( Fig. 3b ). However, growth in the marginal part of the petal was almost completely abolished, with only a few small rounded cells formed within the distal area. This was associated with more than a twofold reduction in the number of cell proliferation events at the onset of cell proliferation within the distal part of the petal primordia ( Fig. 3e and f ). Together, these results indicate that JAG is essential for promoting and maintaining a high rate of cell proliferation at the petal margin. Lack of function almost completely abolishes growth in the distal region. These observations suggest that JAG does not affect the overall growth pattern of the petal but instead promotes cell proliferation within the distal margin, providing strong support for its proposed function as a distal growth promoter during petal development 43 and explaining the cellular basis of convergent petal size reductions in Capsella ( Fig. 1 ). Download figure Open in new tab Fig. 3. Loss of JAG function strongly inhibits cell proliferation during petal growth. a , Average shape of Cr and crjag petals. b, Heatmap illustrating the distribution of cell size within the petals of Cr and crjag . Scale bar: 200 μm. c, Total number of cell within Cr and crjag petals (n = 3 per genotype). d, Average cell count per 200 μm section along the longitudinal petal axis (n = 3 per genotype). Numbers on the y-axis indicate the starting point of each section. e, Representative images showing the distribution of cell division events across the epidermis of stage 1 petals. Fluorescently labelled nucleotide analogue 5-ethynyl-2’-deoxyuridine (EdU; green) was used to visualize newly synthesized DNA, and primary cell walls were stained with propidium iodide (PI; red). Scale bar: 20 μm. f, Quantification of EdU-labelled nuclei in stage 1 petals. Letters indicate significant differences as determined by one-way ANOVA with Tukey’s HSD test (α = 0.05). Cis-regulatory changes in JAG drive petal size reduction in selfing Capsella species To assess the impact of allelic variation at JAG , we compared petal morphology in the NILog and qILrg lines ( Fig. 4 a-d ). Consistent with the loss of JAG function, both showed a selective reduction in distal petal growth, with total petal area decreased by 22% and 15%, respectively ( Extended Data Fig. 7 ). This reduction reflected a lower total cell number, most pronounced at the distal tip. Strikingly, the effect was specific to petals; other floral organs and leaves were unaffected, in contrast to the broad reductions seen in crjag . These results suggest that JAG activity has been selectively modified during petal development. Download figure Open in new tab Fig. 4. Developmental constraint on petal size evolution via heightened JAG sensitivity. a , Average petal shape of Cg , Cr and Co petals (left panel), of quasi Isogenic lines ( QIL ) homozygous for the Cg1JAG (QILgg) or CrJAG (QILrr) allele (middle panel) and of Near Isogenic lines ( QIL ) homozygous for the Cg2JAG ( NILLgg ) or CoJAG ( NILLoo ) allele (right panel). b, Total number of cell within QILgg and QILrr and within NILgg and NILoo petals (n = 3 per genotype). c-d , Average cell count for 10 section along the longitudinal axis within QILgg and QILrr ( c ) and within NILgg and NILoo ( d ) petals (n = 3 per genotype). Numbers on the y-axis indicate the starting point of each section. e, Output of the Capsella petal development model with a divergent polarity field. The effect of JAG on basic growth parameters was progressively reduced, as indicated at the top of each graph from left to right. The rightmost graph shows the impact of decreased JAG activity on petal size. Letters indicate significant differences as determined by one-way ANOVA with Tukey’s HSD test (α = 0.05). Among the polymorphisms distinguishing the tested JAG alleles, four resulted in amino acid substitutions, but none affected conserved regions or the C2H2 zinc-finger domain. Furthermore, promoter-swap experiments revealed a cis-regulatory basis. A. thaliana jag-2 mutants transformed with the Cr coding sequence under the Cg1 promoter ( pJAGg1::JAGr ) developed significantly larger petals than those carrying the reciprocal construct ( pJAGr::JAGg1 ) (Fig. S4 D). Moreover, jag-2 ; pJAGr::JAGg1 petals were similar in size to jag-2 ; JAGr petals and significantly smaller than jag-2 ; JAGg1 petals ( Extended Data Fig. 4 ). Thus, in Cr , differences in JAG activity are explained by cis-regulatory rather than protein-coding changes. Because both selfers share the same non-synonymous substitutions, these are unlikely to contribute in Co, also suggesting cis-regulatory evolution in this species. Subtle changes in JAG dosage drive petal size evolution with minimal pleiotropy Single-cell analysis revealed reduced JAG expression in both developing leaves and petals of the selfers compared to the outcrosser ( Extended Data Fig. 8 ). Because JAG promotes cell division 44 , its reduced activity likely explains the lower proliferation observed at the petal margin. Yet, only petals showed strong size reduction, suggesting heightened sensitivity to JAG dosage. This is consistent with the severe petal defects of jag mutants and the established role of JAG in driving distal petal growth. To test whether JAG dosage could explain petal size differences, we used an existing petal growth model based on the Growing Polarized Tissue (GPT) framework 45 ( Extended Data Fig. 9 ). In this model, tissue polarity is generated by diffusion of a polariser (POL) produced by the basal organiser (PROXORG), which drives growth along the proximal-distal axis (Kpar). The distal factors (DGRAD, DISTORG) promote lateral (Kper) and divergent growth from the petal midline, respectively. The model reproduces A. thaliana petals and is consistent with JAG functioning as DGRAD, with the phytohormone auxin signalling contributing to the polarity field establishment 45 . However, it failed to capture Capsella petal shapes ( Extended Data Fig. 9 ). Adjusting the model by broadening DGRAD and DISTORG domains and allowing DGRAD to also influence Kpar yielded realistic Cg petals ( Fig. 4 , Extended Data Fig. 9 and Methods ). With this new model, the complete loss of DGRAD, combined with fragmented auxin distribution (as in 45 ), produced short and narrow ‘jagged’ petals phenocopying the crjag mutants. Quantitative tuning showed that even modest reductions in JAG activity had strong effects on petal size. A 10% decrease in its effects on Kper and Kpar led to a 23% reduction in petal size, while a 25% reduction in JAG resulted in more than a twofold decrease in petal size ( Fig. 4e ). To test whether organ-specific sensitivity could explain the minimal pleiotropy of change in JAG expression, we next used an existing leaf growth model 46 ( Extended Data Fig. 10 ). In this framework, a PGRAD gradient along the proximal-distal axis regulates growth parallel to the axis (Kpar), while the broadly expressed factor LAM promotes perpendicular growth (Kper), which is further inhibited by MID along the central axis. At later stages, LATE reduces Kpar and enhances Kper activation by LAM. Because crjag leaves show a 14% reduction in length and 18% in width, we hypothesised that JAG contributes to both PGRAD and LAM activity, consistent with its expression at proximal side of leaf and sepal margins 40 , 47 , 48 . Calibrating the model to match the crjag phenotype and running both the petal and leaf models for a similar number of iterations under different JAG dosage revealed that petals were far more sensitive to JAG reduction. At 75% activity, petal area had already decreased by ∼30%, whereas leaf area was reduced by less than 15%, and this contrast becomes even stronger at lower dosage levels ( Extended Data Fig. 10 ). This analysis shows that modest decreases in JAG activity strongly affect petal growth but have only minor effects on other organs. Such heightened sensitivity may represent a developmental constraint and help explain why JAG has been repeatedly co-opted to reduce petal size in different lineages. Because JAG plays a central role in promoting petal growth while contributing less to other organs, even subtle changes in its activity can drive major evolutionary shifts in petal morphology with minimal pleiotropic consequences. Standing variation facilitates parallel petal size evolution A notable observation was that polymorphisms affecting JAG function in both selfing lineages showed similar overdominant effects ( Fig. 5 ). In both NILog and qILrg heterozygotes had significantly larger petals than homozygotes (effect sizes of 0.342 and 0.186 mm²; SE = 0.132 and 0.0871; df = 19 and 20; t = 2.582 and 2.136; Bonferroni-adjusted p = 0.0183 and 0.0452, respectively). Similarly, in complementation crosses, Cg/crjag heterozygotes produced larger petals than Cg/Cr heterozygotes ( Fig. 2 ). JAG , therefore, appears to be prone to overdominant mutations, likely due to dosage-sensitive responses in petal development. Download figure Open in new tab Fig. 5. Standing genetic variation in Cg contribute to convergence. a , Relationship between petal size and JAG genotypes in the NILog and qILrg lines. Data points are gray dots; boxplots show petal area distribution per genotype. Letters denote significant differences from Tukey tests after a mixed-effects model with plant as a random effect. Bonferroni-adjusted p-values compare heterozygotes to larger homozygotes. b, Venn diagram depicts lineage-specific and shared polymorphic sites between the two Capsella selfer lineages versus their outcrosser ancestor. Convergent polymorphisms, where the same allele was fixed in both of the selfers, are highlighted in yellow. c, Manhattan-like plot displays genotype-phenotype (petal length) associations and overdominant loci. Each point tests a genomic locus’s association with petal length; the x-axis shows position; the y-axis shows -log10(p), significance from linear model testing genotype effect on the phenotype. The dashed line marks p=0.05. Overdominant loci, where heterozygotes show the largest mean, are in red. Such overdominance, if maintained by selection, could promote allele reuse by keeping functionally divergent variants in the outcrosser population. To test this, we examined allele frequencies across natural populations at polymorphic sites segregating between the haplotypes used in transgenic assays ( Cg1-Cr and Cg2 - Co ). Of 642 polymorphic sites, 84 converged between selfers. Yet, none of these sites were fixed, as selfer alleles still occurred at ≥4% in Cg . This lack of strong differentiation excludes a scenario of parallel evolution via identical independent changes at the same locus post-divergence. Both selfers also carried species-specific alleles (16 in Cr , 26 in Co ), indicating that independent de novo changes may also have contributed ( Extended Data Fig. 11 ). Because the high number of convergent polymorphisms suggested evolution via standing genetic variation, we tested whether selfer/outcrosser polymorphisms segregating in Cg were associated with petal size ( Fig. 5 and Extended Data Fig. 12 ). Among 233 polymorphisms with sufficient genotype counts, 15 were significantly correlated with petal length (FDR-adjusted P < 0.05). Of these, nine showed overdominance, with heterozygotes having larger petals. Six of these nine were fixed in both selfers, and five were associated with smaller petals ( Extended Data Fig. 12 ). This suggests that collateral evolution through the fixation of multiple small-effect variants could also contributed to petal size reduction. Because Cg/Cr and Co lineages diverged approximately 1 Mya, this scenario implies that these polymorphisms have persisted for thousands of generations ( Fig. 1 , 49 , 50 ). Such persistence could plausibly be explained by heterozygote advantage, given the importance of petal size for pollinator attraction. However, none of the convergent overdominant polymorphisms showed significant deviations from Hardy-Weinberg equilibrium (HWE) or evidence of heterozygote excess in Cg ( Extended Data Fig. 13 ). Nonetheless, in a population with a large effective population size, such as Cg 51 , weak overdominant selection could maintain allele frequencies near a stable equilibrium without causing apparent departure from HWE 52 . To test this hypothesis, we estimated selection coefficients acting on petal morphology by correlating petal size with reproductive efficiency in a hybrid outcrossing population ( Extended Data Fig. 13b ). Using these estimates, together with the observed heterozygote effects, we modelled allele trajectories for each convergent overdominant polymorphism ( Extended Data Fig. 13c ). Despite low selection coefficients (average s against homozygotes ≈ 0.0056), overdominant selection was sufficient to maintain these polymorphisms over millions of generations (probability of heterozygote maintenance: 1 with selection and drift vs. 0.18 under drift alone) ( Extended Data Fig. 13c ). Consistent with our above analysis, only rare simulations (<0.05%) led to HWE departures. Together, these results indicate that weak overdominant selection can maintain functionally divergent alleles at polymorphic frequencies in large outcrossing populations, thereby enabling repeated recruitment of standing variation and facilitating phenotypic convergence through collateral evolution. Implications for convergent evolution of selfing lineages Petal reduction in both Capsella selfers arose through the identical developmental mechanism and the co-option of the same genetic solution ( Extended Data Fig. 14 ). In both lineages, cis-regulatory changes reduce the activity of JAG , a general growth promoter, leading to decreased cell proliferation in the distal petal margin. Although reduced JAG expression is detectable in several organs, only petals are strongly affected, likely because petal growth is more sensitive to JAG dosage. This heightened sensitivity appears to impose a developmental constraint on evolution by enabling large morphological changes in petals while limiting pleiotropic effects, explaining at least in part gene reuse and the convergent evolution observed between independent transitions to self-fertilisation. In addition, the reuse of standing variation also appears to have facilitated phenotypic parallelism. Variants segregating at intermediate frequency in Cg , where they influence petal size, have been similarly fixed in both selfing lineages. Weak overdominant selection associated with improved pollinator attraction explains the long-term persistence of these variants in Cg, and thus their repeated use. Together, our findings illustrate how the interplay between standing genetic variation and developmental constraints interact to bias evolutionary trajectory toward repeatable outcomes. Methods Plant material This study used Capsella rubella , C. orientalis , and C. grandiflora accessions previously described by 37 . Near-isogenic lines (NILs) segregating for PAQTL_2 were generated by introgressing the Cg926 allele into Cr1504 ( NILrg ) and Co1983 ( NILgo ) through five and three rounds of backcrossing, respectively, followed by two additional generations of selfing while maintaining heterozygosity at the PAQTL_2 locus. Progeny from these two NILs were used for fine mapping. The crossing scheme used to generate the quasi-isogenic line (qILrg) is shown in Extended Data Fig. 4c . The jag-2 EMS mutant in Landsberg-erecta (L -er ) background showing recessive floral phenotype was previously identified to have a single nucleotide change (C/T) in the second exon resulting in a premature stop codon 40 . Here, we have identified the Crjag mutant in C. rubella Cr22.5 background by screening a C. rubella ( Cr22.5 ) ethyl-methanesulfonate (EMS) mutagenized population for plants with reduced petal size 53 . The genetic basis and phenotypic characterization of this mutant are presented in this manuscript. Growth conditions Seeds were surface-sterilized, plated on half-strength Murashige-Skoog medium supplemented with gibberellic acid (Duchefa Biochemie; 3.3 mg/L GA), and vernalized for two days at 4°C. Germination occurred at 22°C for ten days. To initiate the transition to flowering, seedlings were placed at 4°C for ten days. Subsequently, plants were grown in a growth chamber under long-day conditions (16 hours light/8 hours dark) with a temperature cycle of 22°C during the day and 16°C at night, 70% humidity, and a light intensity of 150 μmol m⁻²s⁻¹. Morphological measurements To characterize the effect of PAQTL_2, diverse genetic constructs, and jag mutations, sizes of the various organs were measured. Sepals, petals and anthers were measured from the 15th and 16th fully opened flowers on the main stem. Carpel length was measured from unfertilized flowers. Six petals, six sepals, six anthers and two carpels were measured per plant. Dissected flower organs were flattened and scanned at a resolution of 3200 dpi (HP ScanJet 4370). Measurements were conducted on digitized images by using ImageJ 54 . Leaf size was measured on the fully expanded 6th leaf. The leaves were flattened and scanned at a resolution of 300 dpi using an HP ScanJet 4370. Measurements were then taken from the digitized images using ImageJ. 54 . Morphometric analysis of petal outlines Morphometric analysis of petal shape described as closed outlines was performed using Elliptic Fourier Descriptors (EFD) 55 . First, the digital images were converted into binary images using ImageJ 54 . Coordinates of boundaries were extracted using the Python module scikit-image 56 . Outline coordinates were then analysed and visualised using the Momocs package implemented in R 57 . Developmental analysis of petal growth Petal cell size and cell number were determined from dried-gel agarose prints 58 of whole petals from fully open flowers. Cell outlines were captured using and Olympus BX 51 light microscope and AxioCam HRc camera (Carl Zeiss). The cell-outline images were merged using the photomerge function of Adobe Photoshop and further processed with the Python module scikit-image. First, cell-outlines were segmented using adaptive thresholding. Then, binary images of cell borders were dilated, skeletonized and curated by overlapping them with the original images in GIMP ( https://www.gimp.org/ ). Cell areas and centroid coordinates were extracted and used to estimate the average cell number and area per segment along the petal longitudinal axis. The analysis of differences in cell number and size between genotypes was performed using R 59 . Curated segmented images were then analysed with ImageJ by defining Region Of Interest (ROI) and measuring diverse morphological features for each particle. Heatmaps were generated in ImageJ using the ROI colour coder function. Confocal imaging To visualize petal primordia, young flower buds were fixed in a solution containing 50% methanol and 10% acetic acid. mPS-PI staining was carried out following the protocol by Ichihashi et al. (2011) 60 , with minor modifications. Imaging was performed using a confocal laser scanning microscope (LSM780, Carl Zeiss), with excitation at 488 nm and emission collected between 520 nm and 720 nm. Z-stack images were acquired at a resolution of 512 × 512 with a step size of 0.5–1 µm along the z-axis. Image stacks were processed and analyzed using MorphoGraphX 61 , 62 . Surfaces were detected using the “edge detect” tool with a threshold of 10,000. Initial meshes were generated with 5 µm cubes and subsequently subdivided three times before projecting the mPS-PI signal located 4-6 µm from the mesh surface. EDU staining Cell division events were visualized using a combination of 5-ethynyl-2’-deoxyuridine (EdU) incorporation and modified pseudo-Schiff propidium iodide (PI; Sigma-Aldrich) staining, following the protocol described in Riglet et al. (2024) with minor modifications 63 . Inflorescences were harvested and placed on EdU staining medium consisting of 0.22% (w/v) Murashige and Skoog basal salts with vitamins (Duchefa Biochemie, M0222.0050), 3.5% (w/v) sucrose, 0.8% (w/v) agarose, and 10 μM EdU (Invitrogen A10044, Thermo Fisher Scientific), adjusted to pH 5.7. Samples were submerged in the same medium lacking agarose and incubated for 15 hours in a growth chamber under long-day conditions (16-hour light/8-hour dark photoperiod) at 22°C, with an average light intensity of 150 μmol m⁻²s⁻¹. Following incubation, samples were dehydrated through a graded ethanol series (25%, 50%, 75%, 90%, and 100%; 10 minutes each), then rehydrated through the same series in reverse. Tissues were then treated overnight at 37°C with 20 mM α-amylase (Sigma-Aldrich A4551) in phosphate buffer (pH 7.0) supplemented with 2 mM NaCl and 0.25 mM CaCl₂. After washing twice with water and once with 100 mM Tris buffer (pH 8), the samples were incubated for 1 hour in a solution containing 10 μM Alexa Fluor 488-azide (Invitrogen A10266) in 100 mM Tris buffer (pH 8). Subsequently, tissues were incubated for 30 minutes in a click-reaction solution containing 10 μM Alexa Fluor 488-azide, 100 mM Tris (pH 8), 1 mM CuSO₄, and 100 mM ascorbic acid. Samples were washed three times with water. Young flower buds were dissected in water to expose petals and incubated in 1% periodic acid for 30 minutes, followed by two water washes. Samples were then stained with 0.01 μg μl⁻¹ PI for 3 hours, cleared for 2 hours in chloral hydrate solution, and mounted in Hoyer’s solution (30 g gum arabic, 200 g chloral hydrate, 20 g glycerol, and 50 ml water). Imaging was performed using a confocal laser scanning microscope (LSM780, Carl Zeiss) with excitation at 488 nm. Emission was collected between 493–547 nm for EdU and 622–718 nm for PI. Z-stack images were acquired at 512 × 512 resolution with a step size of 0.5–1 µm along the z-axis. EdU signals were detected and quantified using ImageJ. Genetic mapping To refine the position of PAQTL_2, we conducted a segregation analysis in the NILrg background. Initial mapping using approximately 400 progeny plants identified a 138.095 kb region on chromosome 2, located between markers Bs5 (position 8,964,289) and SH3_2 (position 9,102,384) ( Extended Data Table 3 ). To further narrow down this interval, we screened about 1,000 additional progeny plants from NILrg to identify individuals with recombinant chromosomes between these two markers. The positions of the recombination breakpoints were determined by genotyping these recombinants with additional markers within the interval. Selfed progenies from the 15 most informative recombinant lines were then genotyped to identify individuals homozygous for either parental allele. Petal size was quantified in these homozygous plants, and a t-test was used to compare the average petal size between genotype classes for each recombinant to localize the causal mutation in relation to the position of each recombination breakpoint. For fine-mapping PAQTL_2 in the NILgo background, we first identified a NILgo line segregating for a ∼1 Mb region surrounding PAQTL_2, located between markers JAGI (position 8,127,464) and m400 (position 9,149,587) on chromosome 2 ( Extended Data Table 4 ). Approximately 1,400 NILgo progeny plants were screened to identify recombinants between these markers. Each recombinant was genotyped with additional markers within the focal region to pinpoint the recombination breakpoints. Selected recombinant plants were selfed, and progeny were genotyped to identify individuals homozygous for either the C. grandiflora or C. orientalis allele in the segregating region. As described above, petal sizes of both genotype classes were measured and compared using a t-test, enabling us to further refine the position of the causal mutation. Molecular cloning and plant transformation C. rubella ( JAGr ), C. orientalis ( JAGo ), two C. grandiflora ( JAGg1 and JAGg2 ) and two promoter swap genomic constructs: pJAGg1::JAGr and pJAGr::JAGg1 were used to test the existence of allelic variation influencing petal size within the JAG locus. The 8.3 kb JAGr , 9.4 kb JAGg , 9.4 kb pJAGg::JAGr and 8.3kb pJAGr::JAGg1 genomic constructs were amplified in two PCR reactions from genomic DNA of plants homozygous for the C. rubella allele ( NILrr) and plants homozygous for the C. grandiflora allele ( NILgg1) as described in Extended Data Table 5 . The 8.3 kb JAGo and a 10.1 kb JAGg2 genomic constructs were amplified in three PCR reactions from genomic DNA of plants homozygous for the C. orientalis allele ( NILoo) and plants homozygous for the C. grandiflora allele ( NILgg2) as described in Extended Data Table 5 . Primers used are presented in Extended Data Table 6 . The genomic fragments were first assembled and subcloned into a modified version of pBluescript II KS (StrataGene, pBlueMLAPUCAP) using the InFusion® HD Cloning Plus kit (Clontech) and sequenced. The reconstituted cassettes were transferred into the AscI site of the plant transformation vector pBarMAP, a derivative of pGPTVBAR 64 . These genomic constructs were then used to transform Crjag and jag-2 plants by floral dip 65 . Note that since NILrg and NILog were generated from crosses with different Cg926 individuals and since C. grandiflora is a highly heterozygous outcrosser, the two C. grandiflora alleles were not identical in sequence and were therefore named JAGg1 and JAGg2 . Identification of the Crjag mutation To identify the Crjag mutation, we followed a mapping-by-sequencing approach 66 . 500 F2 progeny of a cross between a mutant that we initially termed heart-stopping (hts) because of the defect in fruit development and Cr1504 were screened to isolate plants with a WT or mutant phenotype. About 100 plants for each genotype were pooled and sequenced using the Illumina NextSeq 500 instruments (2 x 100 cycles). DNA-seq data were processed using Trimmomatic 67 to remove adapter sequences. Quality control was done using FastQC ( http://www.bioinformatics.bbsrc.ac.uk/projects/fastqc ). Reads were then mapped at per sample level against the C.rubella reference genome (Cru_183 from phytozome.org ) using BWA-MEM 68 . Variant calling was performed using SAMtools 69 . Those calls were then further processed using R 59 to extract Cr22.5 (background of crjag mutation)/ Cr1504 polymorphisms. The ratio between the number of homozygous and heterozygous calls within 200 kb windows was used to localize the mutation. RNA splicing analysis To detect splicing differences, RNA was first extracted from Crjag and Cr22.5 inflorescences (containing the inflorescence meristem and the flower buds from stage 1 to 9) using Trizol (Life Technologies). RNAs were treated with TurboDNAse (Ambion) and reverse transcribed using oligo(dT) and the Superscript III Reverse Transcriptase (Invitrogen). Primers onw545 and onw546 were used to amplify full-length JAG transcripts. Modelling of petal growth The Capsella petal growth model is based on a 2D Arabidopsis thaliana petal growth model developed by Susanna Sauret-Güeto et al. , 2013 45 . Models were specified and manipulated using the MATLAB-based application GFTbox 70 . Growth patterns are guided by spatially distributed factors across a tissue framework, referred to as the canvas. In this model, the initial canvas consists of approximately 2,300 elements and measures 30 µm in both width and length. Each factor is assigned a value at either the segments or vertices of the canvas, where it promotes growth rates through a linear function ( Extended Data Fig. 8 ). The model integrates two interconnected regulatory networks: – The Polarity Regulatory Network (PRN), which defines tissue polarity and orients growth along the canvas. – The Growth Rate Regulatory Network (GRRN), which modulates the local growth rates. Polarity is established by the propagation of the factor POL, from a proximal organizer (PROXORG) toward a distal organizer called DISTORG. In the most realistic divergent model, DISTORG is expressed at the distal petal margin where it consumes POL, creating a divergence in the polarity field. This causes the growth direction to shift away from the petal midline and toward the margins in the distal region. Another key factor, DISGRAD, is predominantly expressed in the distal region, where it promotes growth perpendicular to the mid-petal axis (Kpar, or Kper). This model successfully replicates Arabidopsis thaliana petal development and is supported by molecular data, which, in accordance with our observations, suggest that JAG functions as a distal factor DGRAD and that auxin biosynthesis and signaling are involved in the establishment of the polarity field. Although this model accurately simulates A. thaliana petal growth, it fails to reproduce the petal shape of Capsella species, despite very similar overall growth rates ( Extended Data Fig. 9 ). In C. grandiflora , petal growth dynamics are similar to those of A. thaliana , but the length-to-width ratio is significantly lower. Simply increasing Kper with the basic growth parameter did not generate realistic Cg petal shapes, as it only widened the distal region, maintaining a lobed shape, whereas Cg petals are generally more rounded. More realistic outlines were obtained by expanding both the DISTORG and DGRAD domains. However, reducing DGRAD activity in this modified model did not significantly decrease petal length, as observed in crjag mutants ( Fig. 3 ). To address this discrepancy, we modified the model so that DGRAD also affected Kpar, which resulted in petals highly similar in shape and size to those of Cg . Furthermore, abolishing DGRAD activity while accounting for the fragmented distal auxin distribution in jag petals (as in 45 ) led to a significant decrease in petal length, resulting in ‘jagged’ petals approximately 1.5 mm long, closely resembling crjag mutants. The models are provided as supplemental dataset and the parameters of the models used are presented in Extended Data Table 7 . Leaf growth was modeled using the organizer-based fixed polarity model described by Kuchen et al. (2012) 46 . In this model, tissue polarity is established by diffusion of a polarity factor (POL) from a proximal organiser (PROXORG) at the base of the canvas. After an initial equilibration phase, the POL gradient is fixed and deforms passively with the tissue during growth. Growth parallel to the polarity field is promoted by PGRAD, while LAM promotes growth perpendicular to the polarity. MID inhibits perpendicular growth along the midline, and LATE modulates both parallel and perpendicular growth. All growth regulators remain fixed to their initial spatial distributions throughout development. Because crjag mutant leaves exhibit a 14% reduction in length and an 18% reduction in width, we assumed that JAG contributes similarly to both PGRAD and LAM activity. The model was therefore calibrated to reproduce the phenotypic differences observed in crjag leaves. We then ran both the petal and leaf models for the same number of iterations while varying the level of JAG activity according to its overall contribution to organ growth. Population genetics We focused our analysis on polymorphisms segregating between the genotypes used for transgenic analysis, since they carry the alleles responsible for variation in petal size between mating strategies. Multiple sequence alignment of the four accessions’ JAG sequences ( C. rubella , C. grandiflora Cg1 and Cg2 , and C. orientalis ) was performed using MAFFT. Polymorphisms were identified from this alignment, which was then used to generate a consensus JAG sequence including all indels and the most frequent allele at each polymorphic site, using EMBOSS cons (parameters: -setcase 0-plurality 1). This consensus sequence replaced the native JAG sequence from the C. rubella reference genome (version 183). We used this custom reference for read mapping of publicly available resequencing datasets (GATK pipeline), including 180 C. grandiflora individuals from a single population, 51 C. rubella , and 16 C. orientalis 71 – 73 . The resulting VCF file contained 3,300 polymorphisms in the JAG coding and cis-regulatory regions. Indels were normalized using bcftools (norm followed by filter --IndelGap 5), applying parsimony and left-alignment, which reduced the set to 2,371 polymorphisms. We further curated the VCF applying the following filters: (i) remove polymorphisms located within large indels based on haplotypes used in transgenic experiments (accounted for ∼75% of the polymorphisms) (ii) remove polymorphisms absent from the genotypes used for introgression and complementation experiments; (iii) simplify indel annotations not fully resolved during normalization; and (iv) inspect ambiguous sites using IGV to resolve ambiguous or missing information. After filtering, the final dataset contained 340 polymorphisms for the Cg-Cr comparison and 536 for the Cg–Co comparison, among which 164 were shared. Among the 164 shared polymorphisms, i.e. the major allele in Cg differs from the major allele of Cr and Co , 84 were called convergent, i.e. the fixed allele in Cg (frequency >75%) differs from the fixed allele in Cr and Co (frequency >90%). To facilitate downstream association analyses, multiallelic sites in the VCF were decomposed into biallelic pseudo-loci. For each multiallelic site, the list of alternative (ALT) alleles was extracted, and each ALT allele was treated as a separate locus. Genotypes were recoded as allele counts (0, 1, or 2 copies of the ALT allele), with missing data set to NA. This procedure was applied to all variants, and the resulting expanded genotype matrix was used for downstream analyses. Phenotypic association analysis and detection of overdominance Phenotypic measurements for Cg individuals with available genomic information were retrieved from 26 and filtered to remove individuals with missing values. Genotypic data were converted into an individual-by-locus matrix, coded numerically, and merged with the phenotypic dataset. For each trait–locus combination, we tested for overall genotype effects using linear models with the genotype as a categorical predictor. Loci with fewer than 10 individuals in any genotype class were excluded from analysis. The overall significance of genotype effects was assessed by ANOVA, and per-term p-values were adjusted for multiple testing using the Benjamini–Hochberg false discovery rate (FDR) procedure. To formally test for overdominance, we applied an additive–dominance model in which heterozygotes were explicitly coded as a binary variable. For each locus–trait pair with adequate sample size in all three genotype classes, we fitted the model Phenotype ∼ additive + dominance and identified significant dominance effects (p < 0.05) where heterozygotes exceeded both homozygotes in mean phenotype. Overdominant candidates were defined as loci that (i) showed a significant dominance term, and (ii) had heterozygous mean phenotypes greater than both homozygotes in the filtered dataset. Overlap between these statistical and phenotypic criteria was used to generate the final list of overdominant loci. Candidate loci were annotated with SNP metadata and visualized using boxplots and Manhattan-like genome position plots, with overdominant loci highlighted. The list of phenotype based overdominant polymorphisms was compared to the list of convergent polymorphisms. Identification of overdominant loci through inbreeding coefficient Observed genotype counts per locus ( n AA, n AB, n BB) were calculated from the expanded genotype matrix. From these, observed heterozygosity ( H ₒ), expected heterozygosity ( H ₑ), and the inbreeding coefficient ( F IS = 1 − H ₒ/ H ₑ) were computed. Hardy–Weinberg equilibrium (HWE) exact tests were performed using the HardyWeinberg package in R. Loci were classified as overdominant if they satisfied all of the following criteria: (i) significant deviation from HWE ( p H ₑ. Chromosome-wise plots of F IS against genomic position were generated, marking overdominant loci and convergent overdominant (based on phenotype) candidates. Simulation of the influence of overdominant selection on allele trajectories We conducted a common garden experiment at the University of Potsdam, Germany, using an F₂ population derived from a cross between Cg and Cr . Plants homozygous for the Cg (functional) allele at the self-incompatibility locus were selected, and their petal size was quantified. Reproductive efficiency was estimated as the number of seeds produced per fruit (which reflects animal pollination), normalised by the average number of ovules developed by each individual and used as a proxy for pollination-mediated female fitness. The relationship between petal size and reproductive efficiency was assessed using Pearson’s correlation. A linear regression model was then fitted with reproductive efficiency as the response variable and petal area as the predictor. Regression coefficients and R² values were extracted to describe the strength and direction of the relationship. For visualization, scatterplots with fitted regression lines and annotated regression equations were generated. Allele frequency dynamics were simulated using a discrete-generation Wright–Fisher model. The model incorporated: Effective population size N e =500,000, 10 6 generations and an initial allele frequency p 0 =0.1 and selection gradient derived from genotype-phenotype relationships. For each replicate (500 replicates), genotype frequencies were adjusted according to selection coefficients, and the next generation was sampled multinomially to account for genetic drift. Simulations were also conducted under neutrality (no selection) for comparison. Fitness of genotypes was modelled linearly as a function of deviation from the mean phenotypic value: where β is the selection coefficient relating phenotype to fitness, Gi is the genotype’s phenotypic value, and G - is the population mean. Overdominance was assessed by comparing heterozygote fitness to the two homozygotes. A stable internal equilibrium allele frequency p ^ was estimated when overdominance was detected. For each simulation replicate, we recorded allele frequencies across generations and at the final generation: homozygote and heterozygote counts, observed heterozygosity ( Ho ), expected heterozygosity ( He ), and Fis . Hardy–Weinberg equilibrium was tested using exact tests, and the proportion of replicates deviating from equilibrium at α=0.05 was reported. Allele frequency trajectories were summarized as mean ± 95% confidence intervals across replicates. Final-generation allele frequencies were visualized as histograms, and probabilities of maintaining polymorphism were estimated as the proportion of replicates with 0 < heterozygote frequency < 1. Statistical analyses Statistical analyses were performed using R (version 4.2.1). Two-tailed Student’s t-test assuming unequal variances was used for two-sample comparisons. For multiple comparisons, we conducted Tukey’s HSD tests, performed using the R/agricolae package. Data visualization was done using the R/ggplot2 package or R/base. In all boxplots, the middle lines represent the median, the lower and upper hinges correspond to the first and third quartile, respectively. The upper and lower whiskers add or subtract 1.5 interquartile ranges to/from the 75 and 25 percentile, respectively. Overdomiance was tested using Linear mixed-effects models,fitted using the lmer function in the lme4 package. Estimated marginal means were obtained using the emmeans package. Planned contrasts between genotypes were tested with Bonferroni-adjusted p-values, and degrees of freedom were approximated using the Kenward-Roger method. Data availability The genome sequence data used in this manuscript have been previous deposited to NCBI’s Sequence Read Archive (SRA) or the European Nucleotide Archive (ENA) under the accession number SRA: PRJNA1233196, PRJNA1233196, PRJNA275635 and PRJEB6689 and are publicly available. All materials used in this study are available upon request. Author contributions K.S., N.W. and A.P. performed and analysed the experiments. U.F. generated transgenic lines, and T.F.L. contributed to phenotypic analyses of mutant lines. C.K. carried out the bioinformatic analyses. A.S. acquired funding, administered the project and, together with M.L., S.R. and Y.D., supervised the research. A.S., N.W., K.S. and A.P. drafted the manuscript, and all authors contributed to revising and editing the final version. Ethics declarations Competing interests The authors declare no competing interests. Acknowledgments We thank Fredric Hedlund and Kathrin Hesse for plant care and members of the Lenhard and Rosa groups for discussion and comments on the article. The computation and data handling were provided by the Swedish National Infrastructure for Computing (SNIC) at Uppmax, partially funded by the Swedish Research Council through grant agreement no 2018-05973. This work was supported by: Deutsche Forschungsgemeinschaft grant SI1967/2 (AS) Swedish Research Council grant 2018-04214 (AS, KS) and theNovo Nordisk Fonden NNF22OC0079830 (AS, KS) Footnotes This new version includes reformating and polishing of the manuscript. References 1. ↵ Stern , D. L . The genetic causes of convergent evolution . Nat. Rev. Genet . 14 , 751 – 64 ( 2013 ). OpenUrl CrossRef PubMed 2. ↵ Blount , Z. D. , Lenski , R. E. & Losos , J. B . Contingency and determinism in evolution: Replaying life’s tape . Science (80-.) . 362 , ( 2018 ). 3. ↵ Hether , T. D. & Hohenlohe , P. A . GENETIC REGULATORY NETWORK MOTIFS CONSTRAIN ADAPTATION THROUGH CURVATURE IN THE LANDSCAPE OF MUTATIONAL (CO)VARIANCE . Evolution (N. Y ). 68 , 950 – 964 ( 2014 ). OpenUrl 4. ↵ Stephen Jay Gould . Wonderful Life: The Burgess Shale and the Nature of History . ( W.W. Norton & Company , 1989 ). 5. ↵ Hoitinga , P. K. & Birkeland , S . Pathway-level convergence: an underexplored aspect of convergent evolution . Trends Genet . ( 2025 ) doi: 10.1016/j.tig.2025.06.009 . OpenUrl CrossRef 6. ↵ Tsuchimatsu , T. & Fujii , S . The selfing syndrome and beyond: diverse evolutionary consequences of mating system transitions in plants . Philos. Trans. R. Soc. B Biol. Sci . 377 , ( 2022 ). 7. Shimizu , K. K. & Tsuchimatsu , T . Evolution of Selfing: Recurrent Patterns in Molecular Adaptation . Annu. Rev. Ecol. Evol. Syst . 46 , 112414 – 054249 ( 2015 ). OpenUrl 8. Wright , S. I. , Ness , R. W. , Foxe , J. P. & Barrett , S. C. H . Genomic Consequences of Outcrossing and Selfing in Plants . Int. J. Plant Sci . 169 , 105 – 118 ( 2008 ). OpenUrl CrossRef Web of Science 9. Cruden , R . Pollen-Ovule Ratios: A Conservative Indicator of Breeding Systems in Flowering Plants Evolution . Biol. J. Linn. Soc . 32 – 46 ( 1977 ). 10. ↵ Barrett , S. C. H . The evolution of plant sexual diversity . Nat. Rev. Genet . 3 , 274 – 84 ( 2002 ). OpenUrl CrossRef PubMed Web of Science 11. ↵ Blarer , A. , Keasar , T. & Shmida , A . Possible mechanisms for the formation of flower size preferences by foraging bumblebees . Ethology ( 2002 ) doi: 10.1046/j.1439-0310.2002.00778.x . OpenUrl CrossRef Web of Science 12. ↵ Fenster , C. B. , Cheely , G. , Dudash , M. R. & Reynolds , R. J . Nectar reward and advertisement in hummingbird-pollinated Silene virginica (Caryophyllaceae) . Am. J. Bot . ( 2006 ) doi: 10.3732/ajb.93.12.1800 . OpenUrl Abstract / FREE Full Text 13. ↵ Goodwillie , C. et al. Correlated evolution of mating system and floral display traits in flowering plants and its implications for the distribution of mating system variation . New Phytol . 185 , 311 – 321 ( 2010 ). OpenUrl CrossRef PubMed Web of Science 14. ↵ Sicard , A. & Lenhard , M . The selfing syndrome: a model for studying the genetic and evolutionary basis of morphological adaptation in plants . Ann. Bot . 107 , 1433 – 1443 ( 2011 ). OpenUrl CrossRef PubMed 15. ↵ Frary , A. , Fritz , L. A. & Tanksley , S. D . A comparative study of the genetic bases of natural variation in tomato leaf, sepal, and petal morphology . Theor. Appl. Genet . ( 2004 ) doi: 10.1007/s00122-004-1669-x . OpenUrl CrossRef 16. Slotte , T. , Hazzouri , K. M. , Stern , D. , Andolfatto , P. & Wright , S. I . Genetic architecture and adaptive significance of the selfing syndrome in Capsella . Evolution 66 , 1360 – 74 ( 2012 ). OpenUrl CrossRef PubMed Web of Science 17. Sicard , A. et al. Genetics, evolution, and adaptive significance of the selfing syndrome in the genus Capsella . Plant Cell 23 , 3156 – 71 ( 2011 ). OpenUrl Abstract / FREE Full Text 18. Fishman , L. , Kelly , A. J. & Willis , J. H . Minor quantitative trait loci underlie floral traits associated with mating system divergence in Mimulus . Evolution (N. Y ). 56 , 2138 – 2155 ( 2002 ). OpenUrl 19. Fishman , L. , Beardsley , P. M. , Stathos , A. , Williams , C. F. & Hill , J. P . The genetic architecture of traits associated with the evolution of self-pollination in Mimulus . New Phytol . ( 2015 ) doi: 10.1111/nph.13091 . OpenUrl CrossRef PubMed 20. Lin , J. Z. & Ritland , K . Quantitative trait loci differentiating the outbreeding Mimulus guttatus from the inbreeding M. platycalyx . Genetics 146 , 1115 – 1121 ( 1997 ). OpenUrl Abstract / FREE Full Text 21. Goodwillie , C. , Ritland , C. & Ritland , K . The genetic basis of floral traits associated with mating system evolution in Leptosiphon (Polemoniaceae): an analysis of quantitative trait loci . Evolution 60 , 491 – 504 ( 2006 ). OpenUrl CrossRef PubMed Web of Science 22. Georgiady , M. S. , Whitkus , R. W. & Lord , E. M . Genetic analysis of traits distinguishing outcrossing and self-pollinating forms of currant tomato, Lycopersicon pimpinellifolium (Jusl.) Mill . Genetics 161 , 333 – 344 ( 2002 ). OpenUrl Abstract / FREE Full Text 23. ↵ Grillo , M. A. et al. Genetic architecture for the adaptive origin of annual wild rice , Oryza nivara. Evolution (N. Y ). ( 2009 ) doi: 10.1111/j.1558-5646.2008.00602.x . OpenUrl CrossRef PubMed 24. ↵ Stern , D. L. & Orgogozo , V . The loci of evolution: how predictable is genetic evolution? Evolution 62 , 2155 – 77 ( 2008 ). OpenUrl CrossRef PubMed Web of Science 25. ↵ Dittmar , E. L. , Oakley , C. G. , Conner , J. K. , Gould , B. A. & Schemske , D. W . Factors influencing the effect size distribution of adaptive substitutions . Proc. R. Soc. B Biol. Sci . 283 , 20153065 ( 2016 ). 26. ↵ Sicard , A. et al. Standing genetic variation in a tissue-specific enhancer underlies selfing-syndrome evolution in Capsella . Proc. Natl. Acad. Sci . 113 , 13911 – 13916 ( 2016 ). OpenUrl Abstract / FREE Full Text 27. Slotte , T. , Hazzouri , K. M. , Stern , D. , Andolfatto , P. & Wright , S. I . Genetic architecture and adaptive significance of the selfing syndrome in Capsella . Evolution 66 , 1360 – 74 ( 2012 ). OpenUrl CrossRef PubMed Web of Science 28. ↵ Slotte , T. et al. The Capsella rubella genome and the genomic consequences of rapid mating system evolution . Nat. Genet . 45 , 831 – 5 ( 2013 ). OpenUrl CrossRef PubMed 29. Jantzen , F. et al. Retracing the molecular basis and evolutionary history of the loss of benzaldehyde emission in the genus Capsella . New Phytol . 224 , 1349 – 1360 ( 2019 ). OpenUrl PubMed 30. Sas , C. et al. Repeated Inactivation of the First Committed Enzyme Underlies the Loss of Benzaldehyde Emission after the Selfing Transition in Capsella Report Repeated Inactivation of the First Committed Enzyme Underlies the Loss of Benzaldehyde Emission after the Selfi . 3313 – 3319 ( 2016 ) doi: 10.1016/j.cub.2016.10.026 . OpenUrl CrossRef PubMed 31. Wozniak , N. J. et al. A Similar Genetic Architecture Underlies the Convergent Evolution of the Selfing Syndrome in Capsella . Plant Cell 32 , tpc.00551.2019 ( 2020 ). 32. Woźniak , N. J. et al. Convergence and molecular evolution of floral fragrance after independent transitions to self-fertilization . Curr. Biol . 34 , 2702 – 2711 .e6 ( 2024 ). OpenUrl PubMed 33. ↵ Fujikura , U. et al. Variation in Splicing Efficiency Underlies Morphological Evolution in Capsella . Dev. Cell 44 , 192 – 203 .e5 ( 2017 ). OpenUrl PubMed 34. ↵ Brandvain , Y. , Slotte , T. , Hazzouri , K. M. , Wright , S. I. & Coop , G . Genomic Identification of Founding Haplotypes Reveals the History of the Selfing Species Capsella rubella . PLoS Genet . 9 , e1003754 ( 2013 ). OpenUrl CrossRef PubMed 35. Bachmann , J. A. , et al. Genetic basis and timing of a major mating system shift in Capsella . bioRxiv ( 2018 ) doi: 10.1101/425389 . OpenUrl Abstract / FREE Full Text 36. ↵ Hurka , H. , Friesen , N. , German , D. A. , Franzke , A. & Neuffer , B . ‘Missing link’ species Capsella orientalis and Capsella thracica elucidate evolution of model plant genus Capsella (Brassicaceae) . Mol. Ecol . ( 2012 ) doi: 10.1111/j.1365-294X.2012.05460.x . OpenUrl CrossRef PubMed Web of Science 37. ↵ Wozniak , N. , Kappel , C. , Marona , C. , Altschmied , L. , Neuffer B. , Sicard , A. Chapter 2: A Common Molecular Basis to the Convergent Evolution of the Selfing Syndrome in Capsella . 38. ↵ Woźniak , N. J. et al. A Similar Genetic Architecture Underlies the Convergent Evolution of the Selfing Syndrome in Capsella . Plant Cell 32 , 935 – 949 ( 2020 ). OpenUrl Abstract / FREE Full Text 39. ↵ Sicard , A. et al. Genetics, evolution, and adaptive significance of the selfing syndrome in the genus Capsella . Plant Cell 23 , 3156 – 3171 ( 2011 ). OpenUrl Abstract / FREE Full Text 40. ↵ Ohno , C. K . The Arabidopsis JAGGED gene encodes a zinc finger protein that promotes leaf tissue development . Development ( 2004 ) doi: 10.1242/dev.00991 . OpenUrl Abstract / FREE Full Text 41. ↵ Turner , T. L . Fine-mapping natural alleles: quantitative complementation to the rescue . Mol. Ecol . 23 , 2377 – 2382 ( 2014 ). OpenUrl CrossRef 42. ↵ Dong , Y. , et al. Epigenetic regulation of fruit shape determination by JAGGED gene in Capsella rubella . at doi: 10.21203/rs.3.rs-8265099/v1 ( 2026 ). OpenUrl CrossRef 43. ↵ Sauret-Güeto , S. , Schiessl , K. , Bangham , A. , Sablowski , R. & Coen , E . JAGGED Controls Arabidopsis Petal Growth and Shape by Interacting with a Divergent Polarity Field . PLoS Biol . 11 , ( 2013 ). 44. ↵ Schiessl , K. , Muiño , J. M. & Sablowski , R . Arabidopsis JAGGED links floral organ patterning to tissue growth by repressing Kip-related cell cycle inhibitors . Proc. Natl. Acad. Sci . 111 , 2830 – 2835 ( 2014 ). OpenUrl Abstract / FREE Full Text 45. ↵ Sauret-Güeto , S. , Schiessl , K. , Bangham , A. , Sablowski , R. & Coen , E . JAGGED Controls Arabidopsis Petal Growth and Shape by Interacting with a Divergent Polarity Field . PLoS Biol . 11 , e1001550 ( 2013 ). OpenUrl CrossRef PubMed 46. ↵ Kuchen , E. E. et al. Generation of Leaf Shape Through Early Patterns of Growth and Tissue Polarity . Science (80-.) . 335 , 1092 – 1096 ( 2012 ). OpenUrl Abstract / FREE Full Text 47. ↵ Dinneny , J. R . The role of JAGGED in shaping lateral organs . Development 131 , 1101 – 1110 ( 2004 ). OpenUrl Abstract / FREE Full Text 48. ↵ He , X. et al. JAG modulates sepal flatness by regulating cell growth direction, interacts with AS2, and is antagonized by TCP24 . Cell Rep . 44 , 115950 ( 2025 ). 49. ↵ Hurka , H. , Friesen , N. , German , D. a , Franzke , A. & Neuffer , B. ‘Missing link’ species Capsella orientalis and Capsella thracica elucidate evolution of model plant genus Capsella (Brassicaceae) . Mol. Ecol . 21 , 1223 – 38 ( 2012 ). OpenUrl CrossRef PubMed Web of Science 50. ↵ Bachmann , J. A. et al. Genetic basis and timing of a major mating system shift in Capsella . New Phytol . 224 , 505 – 517 ( 2019 ). OpenUrl CrossRef PubMed 51. ↵ Foxe , J. P. et al. Recent speciation associated with the evolution of selfing in Capsella . Proc. Natl. Acad. Sci. U. S. A . 106 , 5241 – 5 ( 2009 ). OpenUrl Abstract / FREE Full Text 52. ↵ Lachance , J . Detecting selection-induced departures from Hardy-Weinberg proportions . Genet. Sel. Evol . 41 , 15 ( 2009 ). 53. ↵ Streubel , S. , Fritz , M. A. , Teltow , M. , Kappel , C. & Sicard , A . Successive duplication-divergence mechanisms at the RCO locus contributed to leaf shape diversity in the Brassicaceae . Development ( 2018 ). 54. ↵ Abramoff , M. D. , Magalhaes , P. J. & Ram , S. J. ImageJ. Biophotonics Int . ( 2004 ) doi : DOI: 10.1016/S0022-3115(02)01073-5 . OpenUrl CrossRef 55. ↵ Kuhl , F. P. & Giardina , C. R . Elliptic Fourier features of a closed contour . Comput. Graph. image Process . 18 , 236 – 258 ( 1982 ). OpenUrl CrossRef 56. ↵ van der Walt , S. et al. scikit-image: image processing in Python . PeerJ 2 , e453 ( 2014 ). OpenUrl CrossRef PubMed 57. ↵ Claude , J . Morphometrics with RNo Title . ( Springer-Verlag New York , 2008 ). doi: 10.1007/978-0-387-77789-4 . OpenUrl CrossRef 58. ↵ Horiguchi , G. , Fujikura , U. , Ferjani , A. , Ishikawa , N. & Tsukaya , H . Large-scale histological analysis of leaf mutants using two simple leaf observation methods: Identification of novel genetic pathways governing the size and shape of leaves . Plant J . 48 , 638 – 644 ( 2006 ). OpenUrl CrossRef PubMed Web of Science 59. ↵ R Core Team . R: A language and environment for statistical computing. R Found. Stat. Comput. Vienna , Austria . ( 2018 ). 60. ↵ Ichihashi , Y. et al. Key Proliferative Activity in the Junction between the Leaf Blade and Leaf Petiole of Arabidopsis . Plant Physiol . 157 , 1151 – 1162 ( 2011 ). OpenUrl Abstract / FREE Full Text 61. ↵ Barbier de Reuille , P. , et al. MorphoGraphX: A platform for quantifying morphogenesis in 4D . Elife 4 , 1 – 20 ( 2015 ). OpenUrl CrossRef PubMed 62. ↵ Strauss , S. et al. Using positional information to provide context for biological image analysis with MorphoGraphX 2.0 . Elife 11 , ( 2022 ). 63. ↵ Riglet , L. et al. Hibiscus bullseyes reveal mechanisms controlling petal pattern proportions that influence plant-pollinator interactions . Sci. Adv . 10 , ( 2024 ). 64. ↵ Becker , D. , Kemper , E. , Schell , J. & Masterson , R . New plant binary vectors with selectable markers located proximal to the left T-DNA border . Plant Mol. Biol . 20 , 1195 – 1197 ( 1992 ). OpenUrl CrossRef PubMed Web of Science 65. ↵ Clough , S. J. & Bent , A. F . Floral dip: A simplified method for Agrobacterium-mediated transformation of Arabidopsis thaliana . Plant J . ( 1998 ) doi: 10.1046/j.1365-313X.1998.00343.x . OpenUrl CrossRef 66. ↵ James , G. V. et al. User guide for mapping-by-sequencing in Arabidopsis . Genome Biol . ( 2013 ) doi: 10.1186/gb-2013-14-6-r61 . OpenUrl CrossRef PubMed 67. ↵ Bolger , A. M. , Lohse , M. & Usadel , B . Trimmomatic: A flexible trimmer for Illumina sequence data . Bioinformatics 30 , 2114 – 2120 ( 2014 ). OpenUrl CrossRef PubMed Web of Science 68. ↵ Li , H . Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM . arXiv Prepr. arXiv 00 , 3 ( 2013 ). 69. ↵ Li , H. et al. The Sequence Alignment/Map format and SAMtools . Bioinformatics 25 , 2078 – 9 ( 2009 ). OpenUrl CrossRef PubMed Web of Science 70. ↵ Kennaway , R. , Coen , E. , Green , A. & Bangham , A . Generation of Diverse Biological Forms through Combinatorial Interactions between Tissue Polarity and Growth . PLoS Comput. Biol . 7 , e1002071 ( 2011 ). OpenUrl CrossRef PubMed 71. ↵ Agren , J. A. et al. Mating system shifts and transposable element evolution in the plant genus Capsella . BMC Genomics 15 , 602 ( 2014 ). 72. Koenig , D. et al. Long-term balancing selection drives evolution of immunity genes in capsella . Elife 8 , 1 – 27 ( 2019 ). OpenUrl CrossRef PubMed 73. ↵ Josephs , E. B. , Lee , Y. W. , Stinchcombe , J. R. & Wright , S. I . Association mapping reveals the role of purifying selection in the maintenance of genomic variation in gene expression . Proc. Natl. Acad. Sci. U. S. A . 112 , 15390 – 15395 ( 2015 ). OpenUrl Abstract / FREE Full Text View the discussion thread. Back to top Previous Next Posted March 19, 2026. Download PDF Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Regulatory architecture and standing variation drive parallelism in floral evolution Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Regulatory architecture and standing variation drive parallelism in floral evolution Kevin Sartori , Natalia Wozniak , Anahid Powell , Ushio Fujikura , Christian Kappel , Tian-Feng Lü , Yang Dong , Stefanie Rosa , Michael Lenhard , Adrien Sicard bioRxiv 2025.10.28.684842; doi: https://doi.org/10.1101/2025.10.28.684842 Share This Article: Copy Citation Tools Regulatory architecture and standing variation drive parallelism in floral evolution Kevin Sartori , Natalia Wozniak , Anahid Powell , Ushio Fujikura , Christian Kappel , Tian-Feng Lü , Yang Dong , Stefanie Rosa , Michael Lenhard , Adrien Sicard bioRxiv 2025.10.28.684842; doi: https://doi.org/10.1101/2025.10.28.684842 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Evolutionary Biology Subject Areas All Articles Animal Behavior and Cognition (7635) Biochemistry (17697) Bioengineering (13894) Bioinformatics (41951) Biophysics (21455) Cancer Biology (18592) Cell Biology (25507) Clinical Trials (138) Developmental Biology (13380) Ecology (19903) Epidemiology (2067) Evolutionary Biology (24321) Genetics (15610) Genomics (22509) Immunology (17737) Microbiology (40398) Molecular Biology (17182) Neuroscience (88618) Paleontology (667) Pathology (2833) Pharmacology and Toxicology (4825) Physiology (7641) Plant Biology (15158) Scientific Communication and Education (2046) Synthetic Biology (4296) Systems Biology (9825) Zoology (2271)