How seeds and growth dynamics influence plant size and yield: Integrating trait relationships into ontogeny

Body size is relevant to multiple dimensions of life. The an organism influences its ecological interactions and impact on ecosystem processes, most life-history traits correlate with body (Peters, 1983; Woodward et al., 2005). In plants, large individuals compete better for available resources, are less stress tolerant have higher resilience disturbance (Falster & Westoby, 2003; Kunstler 2016; Niklas 2003), contributing differences in strategies (Grime, 2001; 1998). Plant also critical vital rates, as it determines seedling survival, flowering maturation times, reproductive output (Moles Leishman, 2008; Westerband Horvitz, 2015). Furthermore, varies by orders magnitude within among plant species, extensive research has attempted explain this variation (e.g. Koch 2004; Niklas, 2007; Vasseur 2012). For example, climate, soil fertility, biogeography, regimes, growth form phylogeny determine (Goldberg 2017; McCarthy Moles 2009). However, while much progress been made describing the role evolutionary drivers size, known about proximal mechanisms that operate during ontogeny drive mature size. Plants differ widely their ability acquire allocate biomass from juvenile stages (Dayrell 2018; Henn Damschen, 2021; Poorter This partly because roles different morphological, physiological phenological (M-P-P) change development. Although one important sources variation, we still do not fully understand how interaction between development drives During ontogeny, at least three types M-P-P can size: seed rate duration vegetative (Violle 2007). Seed mass other organs via cascading effects (Roach Wulff, 1987). heavier seeds often germinate earlier season grow into larger seedlings 2004). More leaves roots stage confers early advantage hoarding regardless rates resource acquisition per unit or time (Kidson 2000). initial potentially leads leaves, stouter stems longer, roots, thus plants overall (Niklas, Indeed, previous studies found positive relationships both intra- interspecific levels (Fenner, Jakobsson Eriksson, 2000; Lush Wien, 1980), well a scaling organ sizes whole (Price 2007, 2014; West 1999). Also, global analyses functional traits, co-vary same axis trait (Díaz 2004, Pierce 2014). Therefore, heavy yield might amplify effect plants. addition contribute High gain produce increasingly Growth usually measured relative (RGR, increase pre-existing time; Blackman, 1919). RGR be decomposed underlying components reflecting photosynthetic efficiency (NAR, net assimilation rate), allocation patterns (LMR, leaf ratio), costs area (SLA, specific area) (Poorter, 1990). Thus, achieve high RGRs enhancing and/or investing more photosynthetically active tissues (Poorter Remkes, interplay (Sun Frelich, 2011). Finally, depends devoted growth. By increasing extent growth, species even smaller slow herbs genus Petasites largest British flora relatively long period compared herbs, yet very small (Grime 1988; Hodgson TRY database, Kattge 2020, request no. 8910, accessed 13 March 2020). theoretical models fit distributions assume delayed reproduction associated (Cohen, 1976; Kozłowski, 1992). Moreover, case report annual flower later tend (Bolmgren Cowan, Sun contributes To shaping need consider correlations. comprises developmental such seedling, (Gatsuk 1980). explained without considering ontogenetic cascades, is, variations shape due shifts trait–size (McNamara, approach only appropriate if linear throughout but case. stronger influence than (Stanton, 1984). relationship develop (Larocque Marshall, 1993). addition, numerous examined trade-offs Gleeson Tilman, 1994; Maranon Grubb, 1993; Rose, 2005; Shipley Peters, 1990; Swanborough 1996), few considered when assessing these (Cornelissen, 1999; Niinemets, 2006). Because correlations observed may those maturity (Laughlin Mason 2013), understanding causes requires multivariate approach. Comparisons crops wild progenitors show evolution under cultivation generally increased (Milla Milla Matesanz, 2017). changes occurred crop enlargement, shortening lengthening life cycles and, ultimately, increases (Gómez-Fernández 2022; Harlan 1973; Meyer Purugganan, 2013). So far, however, linked directly indirectly domestication further improvement differentially affected there selection pressures, human behaviours two (Abbo Selection desirable characteristics over undesirable neutral ones disrupted pattern trait–trait root domesticated counterparts (Roucou differential yields, relationships, poorly understood. Here, aimed disentangle herbaceous crops. Instead examining each individually, asked interact through direct indirect (Figure 1). We chose 18 phylogenetically diverse crops, including wild, landrace improved accessions crop, study system provides huge itself. Additionally, allowed us investigate after improvement, impacted yields. were grown common environmental conditions assessed stages: mature. expected seed, dynamics all yield, exerted interactions. particular, (1) what importance mass, account size? And which mainly arise?; (2) interactions?; (3) yields depend drivers? grew conditions. components, total 378 individual estimated (seedling, mature) harvested fruiting stage. Using path analyses, contribution variation. results independently analysing (wild vs. landraces) (landraces cultivars). grain fruit investigated selected taxonomically our experiment (Table obtained lots accessions, 126 (see Table S1 accession identifiers donors, (2020) literature progenitor assignment). (W) existing taxa closely represent ancestor (L) (I) genotypes subjected traditional agricultural practises intensive modern breeding, respectively. Our belong four groups: C3 cereals (13%), C4 legumes (26%) forbs (i.e. neither graminoids nor legumes; 48%), various families: Poaceae (22%), Amaranthaceae (5.5%), Asteraceae Boraginaceae Brassicaceae Linaceae Malvaceae Pedaliaceae Solanaceae (11%) Fabaceae (28%). them annuals cultivated (56%), (22%) fruits (22%). Wild May August 2019 CULTIVE lab glasshouse Universidad Rey Juan Carlos, Móstoles, Spain. sown peat-filled flats germinated 15 days sowing. When radicle emerged testa, transplanted 3.6 L square pots (15 × 20 cm). filled sand supplemented slow-release fertiliser (5 g L−1 Basacote Plus 6M, Compo). experimental were: mean temperature ± SD = 24 5°C, humidity 57 16%, radiation light hours 892 204 μmol m−2 s−1. followed destructively non-destructively (Pérez-Harguindeguy first method consists harvesting category regular intervals. Albeit used, precludes investigation level. second repeatedly measure proxies individual. It accurate information level, no data used mixture methods follows. experiment, divided focal calibration Several below) intervals obtain dry area. Data generate prediction equations area, out non-destructive traits. masses areas then monitoring date using equations. Further details procedures described subsection analyses. Seeds weighed individually Mettler Toledo MX5 microbalance (1 μg precision; Toledo). Approximately 2 weeks sowing, weights (light, medium heavy) (n 3 replicates plants). Each was monitored every 8 (8 times total), frequently At date, height, canopy diameter, number tillers, length longest diameter basal stem measured. Previous shown Tracey 2016). following recorded: germination (cotyledon[s] visible), (first true late (several tillers), buds open flowers/first set). Eight nine destructive harvests status (either domesticate) entire period, covering 284 After measuring harvested. Harvested washed separated stem, leaf, root, litter, (bud, fruit) fractions. fraction included petioles rachises. scanned laminae grayscale resolution 400 dpi Epson Expression 10000 XL scanner (Seiko Corporation) calculated Photoshop CS6 (Adobe Systems, Inc.). oven-dried 60°C weighed. Total (g) adding fractions harvest date. Due anomalous excluded prior analysis. All performed r v.4.2.0. (R Core Team, 2021). estimate Duration expressed sowing appearance flowers. plant, minimum maximum recorded (or size) final size), referred study, refers reaches sexual maturity. Juvenile intermediate reached closest midpoint period. Overall, ranged 0.001 0.49 stage, 0.02 4.07 0.13 66.8 conventional standardised (Pommerening Muszta, approach, (calculated log ratio interval) observationally independent response variable size). suffers another problem – decreases dependence, comparisons criticised (Turnbull 2008). avoid problems, size-standardised (sRGR) fitting curve extracting reference fruiting, infructescences enclosed organza bags (a transparent, permeable synthetic fabric) prevent dispersal. collected summer (July–August). weight proxy yield. Harvest index sum Since determined contained seeds. (borage, cabbage, lettuce white clover), indicator agronomic separate datasets compiled. First, full dataset sRGR initial, 377). From dataset, derived: (domestication dataset; n 269) (improvement 215). had formed dataset. 201 belonging 14 comprising accessions. evaluate sizes), index, ran LMMs lme function nlme package (Pinheiro Models (categorical variable: progenitor, landrace, cultivar) group cereal, forb, legume) fixed factors, identity nested random factor intercept. based variables. variables ln-transformed improve normality. presence heteroscedasticity (evaluated Levene's test), variance structure modelled ‘varIdent’ specification function. significance factors anova.lme sequential (type I) sums squares amount calculating marginal conditional pseudo-R2 r.squaredGLMM MuMIn (Barton, Multiple comparison tests statuses glht multcomp false discovery correction (Hothorn Figure S3 4 S4 S5 5 S6 model fitted sNAR, sLMR sSLA), replacing specifying bivariate 1; 2). Prior analysis, non-linear them, scaled (mean 0, diversity groups exogenous categorical control design. hypothesised reported Gómez-Fernández 2013; assumed affiliation, since close history (C3 cereals, clades) respond similarly (Reich 2003). Second, examine subsequent modulation Path B separately tested expectation altered sRGR, Individual specified A. coefficient coded ordinal (0 1 cultivar). Alternatively, test differed progenitors, landraces cultivars, conducted multigroup analysis whether coefficients varied statuses. Third, affect end, C, extended version B, additional paths assumptions. Yield especially annuals, re-allocate (Weiner shorten intensively evolution, being higher-yielding (Sadras, (i) (ii) (iii) cultivation. several phylogenetic signals (Martin, 2021), affiliation. S1) B. evaluated directed separation (d-sep; Shipley, 2009), combines independence claims single Fisher's C statistic. A consistent statistic significantly χ2 distribution (p > 0.05). AIC score know support (Shipley, allow scales. assess predictor direct, follows Direct linking Indirect product along variable. effects, taking quantified pseudo-R2. d-sep tests, AIC, coefficients, pseudo-R2, piecewiseSEM (Lefcheck, There considerable across studied largest-seeded smallest-seeded (faba bean: 548 mg amaranth: 0.57 mg). ca. quarter range worldwide (Westoby lesser extent, 0.10 peanut 0.27 g−1 day−1 amaranth, 25 tomato 43 clover, Response greatly Mature 1.25 clover 33.4 millet, 1.46 lentil 28 millet. variability, substantial variability 55-day period; S1). 90% 70% respectively, received statistical support, indicated goodness metrics (Figures 3-5). Domesticates seeds, seedlings, 2a,d,e,f,g; S2). trends proportion part S2; On hand, did statuses, 2b,c,h; Domestication sizes, breeding Heavier ultimately 3a; S3). strongly interacted ontogeny. promoted stages, so driven rapid longer growing periods 3b). provided slower S3), clear causal S3a). sNAR component accounted S3b). run defining direction 4; Tables S5). slightly decreased negative 4a; models, main became 4b). cultivars S4). size-cascading changed Evolution (direct path: 0.18; 5a; S6). Of mediated 18.5% Final determining finally 5b). Large shorter