Populations of arable weed species show intra-specific variability in germination base temperature but not in early growth rate

Key plant traits affecting growth performance can differ among and within species, influencing competitive plant community dynamics. We determined the intra-specific variability of germination base temperature among 13 arable weed species and the seedlings’ early post-emergence relative growth rate among 21 species in climate chamber and green house experiments. Intra-specific variability was quantified with two seed populations (originating from contrasting climate in Germany & France) for the germination base temperature of 6 species and for the early growth rate of 16 species. Inter-specific variability for both traits was always higher than intra-specific variability. Within a given species, we found that germination base temperatures were higher in seeds stemming from colder climate populations. Seedling relative growth rates did not differ between seed populations. Models simulating weed growth should reflect these differences in germination traits among populations, especially when they are used for weed community assembly studies in a local to regional extent.


Unfunded studies
Enter: The author(s) received no specific funding for this work.    Introduction   29 Over the last decades, there has been an increase in trait-based analyses and modelling, both in 30 ecology in general [1] and in weed ecology in particular [2,3] followed by a debate on the importance 31 and implications of intraspecific trait variability. Initially, the assumption of many trait-based 32 approaches was that intraspecific variability is negligibly small compared to the interspecific 33 (between-species) variability, and that species can be characterised by mean trait values [4]. This has 34 since been challenged by a number of studies [5,6]. Trait-based research projects therefore need to 35 critically consider whether to include intra-specific trait variability (ITV) in their set-up, based on 36 species, scale, and scope of the study [7]. 37 The mechanisms that cause intra-specific variability as a species' response to its environmental 38 conditions include adaptation (genetic variability) and phenotypic plasticity [8]. They are expressed in 39 a range of functional plant and seed traits [4,9]. Functional traits have been defined as measurable 40 features which interact with ecological factors through specific functions in order to explain plant 41 fitness components like growth, reproduction and survival [9,10]. Sometimes, they are classified as 42 hard and soft traits. A hard, usually physiological trait would be a feature which is difficult to 43 quantify, but describes a plant function more accurately in contrast to more easily measured, often 44 morpho-anatomical soft traits that can sometimes provide proxies for a hard trait [11,12]. The study of traits in weed research aims to understand which processes drive weed community 54 composition and population dynamics. One of the main goals is to contribute to a more 55 environmentally-friendly weed management [2,13]. This includes attempts to make the modelling of 56 weed populations generic rather than species-specific, based on a mechanistic representation of 57 processes and involved species traits [3,14,15]. 58 Arable fields, compared to other habitats of wild plant species, are characterised by a high frequency 59 of disturbances through and strong competition from crops. The right timing of germination and 60 establishment together with the ruderal growth and regeneration strategy are therefore of special 61 importance for plant establishment and fitness in weed plants [16,17]. 62 Selection pressure favours species' responses to local environmental cues that synchronise 63 germination with periods that are optimal for seedling survival and establishment as well as 64 reproduction [18][19][20]. Likewise, plants which compete better for light, nutrients and water have a 65 higher fitness (i.e. the number of seeds produced per germinated seed), are more tolerant to 66 resource deficits, or have higher plasticity in response to stresses [21]. 67 The high level of disturbance and the ruderal strategy, suggest a high ability of adaptation in weed 68 species, and consequently a high intra-specific trait variability. This must be represented in trait-69 based approaches in weed ecology to make them widely usable and valid in a more general context. 70 Existing trait databases are extensive sources for trait information, but they are usually of limited use 71 when it comes to intraspecific trait variation under variable environmental settings [1,3]. 72 Only a small number of studies investigated intra-specific trait variability in arable weeds, mainly for 73 soft traits like canopy height, specific leaf area, or biomass. These studies concentrated on trait 74 variability as a response to different cropping systems, or explore the effect of variability on 75 community assembly patterns and on ecosystem functioning [22][23][24]. 76 Even less studies analysed intra-specific variability in traits associated more to processes in 77 population dynamics, like germination and early growth. In this paper, we aim to close this gap and 78 analyse the intra-specific variability of two key traits in arable weed species: germination base 79 temperature and early relative growth rates. Base temperature has been identified as one key 80 parameter necessary for modelling weed seed germination [25] determining the days where 81 germination is possible as well as controlling the speed of germination, emergence and post-82 emergence growth. The latter also depends on the early growth rate of seedlings which is an 83 important ecophysiological trait in crop/weed competition models [26]. 84 Results of earlier studies were not consistent on the extent of intra-specific variability in these two 85 traits, or other traits associated to germination [27][28][29][30][31]. Differences between populations were 86 found for some species, not for others not. It is possible that the population effect is only evident 87 when populations come from locations with sufficiently strong differences in environmental 88 conditions. No differences were found between populations from near-by regions like Northern vs. 89 Eastern France [32] and Central vs. Northern Italy [33,34]. 90 In our study, we therefore examine experimental results obtained with seeds from two regions with 91 contrasting environmental conditions. We hypothesised that between-population differences in 92 germination base temperature and seedling relative growth rate exist. We expected that base 93 temperatures as well as early relative growth rates would be lower in the location with colder 94 climate and lower solar radiation, due to lower resource availability. Experiment on germination base temperature 100 Seed germination of 13 species was tested at four to six constant temperatures. Seeds were laid out 101 in petri dishes (Ø 9 cm) lined with a double layer of filter paper and moistened with 5 ml deionised 102 water, in some cases with 5 ml solution of KNO3 (10 mmol) to break seed dormancy. Four replicates 103 of 50 or 100 seeds per treatment were placed in temperature chambers with a 12-hour photoperiod. 104 Petri dishes were checked for germination at least once a day and moistened when the filter paper 105 started to fall dry. Seeds were considered germinated once a radicle was clearly visible. Experiments 106 lasted for approx. 4 weeks each, or until no more germination occurred during 7 days. Temperature 107 in the chambers was monitored with data loggers (HOBO UX100-001/ Voltcraft PL-125-T2) every 10 108 minutes. The average of all measurements during the experiment was used for calculations.
where LA is leaf area (cm²) of the plant on sampling date d, LA0 is (initial) leaf area at emergence 155 (cm²) , TTd is thermal time from emergence to sampling day d (°C·d), and RGR the relative growth 156 rate ( cm²·cm -2 ·°C -1 ·d -1 ). Leaf area was logn-transformed and parameters fitted by linear regression 157 over TTd. Data points beyond the phase of exponential growth were left out of analysis. The end of 158 this period was determined as the date when the local slope of logn(LA) vs TTd dropped below 1/10 of 159 the slope at emergence (when TTd=0). Local slopes were calculated as the derivate of a polynomial 160 fitted to logn(LA) vs TTd (S1 Fig.). 161 Thermal time TTd was calculated as 162 where Ti is the mean temperature of day i (°C) and Tb is the species-specific base temperature (°C). 164 For twelve species, we used the germination base temperature obtained in our own experiment to German arable fields (Table 1). Seeds were mostly sourced from the experimental gardens and 173 experimental fields of Rostock University, but in a few cases obtained from commercial suppliers 174 aiming for seed lots from Northern Germany (S1 Table). The seeds from French provenance used in 175 the experiment on relative growth rate were provided by the INRA weed seed collection (UMR 176 Agroécologie, INRA Dijon). Seeds were stored in paper bags at room temperature and subjected to 177 cold stratification prior to the experiments if necessary to break dormancy.  Early summer Analysis of inter-specific and intra-specific variability 203 Data 204 To analyse inter-specific and intra-specific variation of the two plant traits, we pooled the results of 205 the described experiments in Rostock with results of similar experiments obtained in Dijon (Table 2). 206 The Dijon experiments were carried out on weed species frequently found in arable crops in 207 temperate France, choosing species that differed in terms of clade, emergence season, seed mass 208 [14,21]. Here, we only use data from Dijon for species that are also important in Northern Germany. 209 For clarity, "Rostock" and "Dijon" will refer to the locations of experiments while "French" and 210 "German" will refer to seed populations originating from the two different regions. 211 We included four sources of variation into our analysis: 1. species (=inter-specific variability), 2. seed 215 provenance (= between-population variability), 3. differences in the experimental conditions 216 between experimental locations (contributing to between-individual variability), and for the growth 217 rate experiment 4. differences in the experimental conditions between repetitions within a location 218 (also contributing to between-individual variability). 219 Inter-specific and intra-specific variability in germination base temperature 220 The effects of species and population on germination base temperature were tested with a two-way 221 Anova. Pairwise tests were run on germination base temperatures between populations of the same 222 species, comparing the estimates by calculating the z-score 223 where TbG and TbF are the germination base temperature estimates and se_TbG or se_TbF the 225 corresponding standard errors for German and French populations, respectively. The significance of 226 the difference was calculated as 227

. 228
Inter-specific and intra-specific variability in relative growth rate 229 Mixed effect models were used to analyse early relative growth rates with species, seed population 230 and experimental location as fixed factors. We included repetition as a random factor to account for 231 the nested structure of the data and the unbalanced number of plants (3 to 10) within each 232 combination of species, population, location, and repetition [40]. A Type III Anova with 233 Satterthwaite's method for denominator degrees of freedom and F-statistic was used to test the 234 significance of effects of species, population and experimental location. 235 We also compared intra-specific and inter-specific variability by means of variation coefficients. First, 236 we averaged measured relative growth rates per species and population. Then we calculated the 237 coefficient of variation between population means per species (intra-specific variation) or between 238 species means per population (inter-specific variation). 239 We estimated marginal means to specify the relative growth rate per species and population and 240 conducted a pair-wise comparison of means to test for differences between populations. If the 241 difference was not significant, we estimated the species relative growth rate from both populations 242 combined. 243 Within the paper, we present analyses of relative growth rate variability without the data collected 244 from French seeds in the Rostock experiment. The results differed only slightly when incorporating 245 these "transplanted seeds" and are included in the Supporting information. We decided against using 246 this data in the main analysis because it introduced additional variability and further un-balance to 247 our analysis (by growing seeds in an environment different to their origin and only transplanting one 248 population, not both). 249 Relationship between base temperature and relative growth rate 250 We tested the relation between base temperature, provenance and early relative growth rate (RGR) 251 by fitting a linear model including all species with RGR for both provenances, followed by an ANOVA. 252  266 267 For the six species for which we had measurements for both seed populations, the base 268 temperatures for the German populations were all clearly higher than for the French populations 269 (Figure 2), with a significant difference for two species (Table 3

277
Relative growth rates 278 Relative growth rates were strongly dependent on species which accounted for about a third of the 279 variation (η² = 0.303, p<0.001). Provenance of seeds had no effect in itself (p= 0.93), but we found a 280 significant species:provenance interaction ( η² = 0.151, p <0.001) accounting for half as much 281 variation as species. Differences in the experimental conditions (environment, equipment and 282 materials, handling) between the several runs explained nearly half the variance in the data (Intra-283 correlation Coefficient ICC = 0.47). This variation could be attributed mainly to the repetitions. 284 Experimental location had no significant effect ( Table 2 in Supp. Inf. S1 File). 285 Variation in relative growth rate between species was about three times the mean variation within 286 species (Figure 3). Species-specific relative growth rates varied between 0.0101 (V. hederifolia) and 287 0.0446 cm²cm -²°C -1 d -1 (S. nigrum) ( Table 5). The pairwise comparison of means between populations 288 for each species showed a significant difference of relative growth rate only for one species, S. 289 nigrum ( Table 4 in Supp Inf. S1 File).  Relationship between base temperature and growth rate 303 We found a positive relationship between germination base temperature and early relative growth 304 rate in the studied species. A linear model for the French populations had a fit of R² = 0.74 and for 305 German populations of R² = 0.85. The slopes were not significantly different when testing for the 306 influence of seed origin on the relation between the two variables (t = -2.022, p > 0.05). Species, 307 which germinate at higher temperatures, grow more per degree-day than species with low 308 germination base temperatures. 309 Our results on base temperature of weed species from Northern German provenance are in good 321 correspondence with the ranges reported for European arable weed species in the literature before 322 (Table 3). Generally, we found that for a given species, base temperatures were higher for seeds 323 originating from the colder climate. At first sight, this result is slightly surprising: we expected base 324 temperatures to decrease as an adaptation to lower soil temperatures in Rostock compared to Dijon. Base temperature may control germination only for spring emergence periods 346 Germination base temperature has been generally assumed as giving a cue of favourable growing 347 conditions for seeds to start germination and emergence, but also halt germination in autumn before 348 the seedlings' ability to survive winter could be decreased by late germination [57,60,61]. Our 349 experiments suggest that this relationship may only hold true for the spring emergence periods and 350 therefore for species that are facultative or strictly spring-germinating. In a previous analysis, an 351 influence of base temperature could only be found on the onset of the spring emergence period but 352 not the onset of autumn emergence or end of either of these [14]. 353

363
Of the two functions of base temperature, the spring part may be more important for population 364 fitness than the autumn part. Most weed species display pronounced emergence flushes. In autumn, 365 the flush will start as soon as moisture is available. When temperatures decrease down to the base 366 temperature, probably most non-dormant seeds will have already germinated and grown to a stage 367 tolerant to frost. In spring on the other hand, the flush may start as soon as temperatures rise, with 368 many seeds germinating at a time when temperatures may still quickly fall below zero, potentially 369 leading to high losses due to frost. In spring, a potential loss of very young seedlings would therefore 370 be much more dramatic in numbers than in autumn. 371 We can find more evidence for this mechanism when looking at species that are facultative autumn 372 and spring germinators or strictly limited to one of the seasons, especially when this habit differs 373 between sites. A shift in germination habit has been reported for a number of species [61]. G. 374 dissectum is an example species that only germinates in autumn in Dijon, but in Rostock also in 375 spring. The shift in germination period may explain the large difference of base temperature for G. 376 dissectum (4°C in Rostock vs. 0°C in Dijon) [63]. We suspect G. dissectum to be (mostly) dormant in 377 spring in Dijon. The base temperature therefore has no function to prevent seeds from germinating 378 too early in spring, and subsequently selection pressure may never have favoured adaptation 379 towards a higher base temperature. Similar differences in germination timing between provenances 380 have been reported for C. bursa-pastoris [64] and C. canadensis [60]. 381 Interaction of climate and management controls germination in early and late summer 382 We propose that the main factors driving germination and emergence of arable weeds in spring and 383 autumn months are weather and management and their interaction. Management is similarly 384 dependent on favourable periods as various weed life cycle stages. 385 The summer germinating weed species are the only ones that emerge at the same time in Rostock 386 and Dijon. At this time (end of April/beginning of May), soil temperatures become similar in Dijon 387 and Rostock after a long spell of colder soils in Rostock (Figure 1). Emergence timing coincides with 388 the last tillage and sowing operations for maize cultivation. In contrast to most other crops, maize 389 sowing is also synchronous between the two locations ( Figure 5). Base temperatures in summer 390 germinators were always higher in Rostock, but it remains unclear which adaptation and selection 391 processes happened, because there seems to be no function for base temperature in these species. 392 Disparities between location-specific emergence patterns can arise due to different soil and crop 393 management practices [65] and the pattern eventually is memorized by genetic adaptation. The later 394 start of tillage and cultivation for winter-sown crops in Dijon (optimizing crop management) leads to 395 selection in the weed species to emerge later (after last tillage). A similar effect of management 396 timing on germination was found for Datura stramonium seeds from three Southern European 397 populations [33]. 398 Intra-specific variability in early growth rates is much smaller than inter-specific variability 399 Early relative growth rates for three of the four species measured in Rostock agree with literature 400 values [21]. Only the growth rate for V. hederifolia was slower than in any other previously measured 401 weed species. To our knowledge, this study is the first to investigate inter-population differences in 402 early growth rates of (non-woody) plant species, apart from [31]. Other studies on early growth rate 403 often measured above-ground biomass rather than leaf area [21,26,66-70], which makes it hard to 404 compare ranges. 405 Our results on relative growth rate were most strongly determined by species identity. This high 406 inter-specific variability is of course related to the original species choice in Dijon that aimed to 407 include a range of contrasting species into the experiments. We found a significant population effect 408 on relative growth rate only for one species. This result is in agreement with earlier studies which 409 found no maternal effects on life cycle stages later than germination [56]. 410 In general, the inter-population or intra-specific variability decreased when a species was tested 411 more than once, probably by a combination of lowering the experimental error as well as exploring a 412 wider range of growing conditions with each population leading to some overlap and convergence. 413 For some species, we found a very high variability in growth rates between successive trial with the 414 same species and population indicating a plastic response to varying environmental conditions. 415 Higher inter-specific variation compared to intra-specific variation in growth responses under these 416 non-stressful conditions is in contrast to earlier evidence that within-species variation in growth 417 responses can be as high as among-species variation, albeit under stressful conditions [71]. The 418 differences may be attributed to only two locations having been tested here (vs. several locations in 419 Europe in [71]) and more numerous species chosen to represent among-species variation (15 vs. 4 420 species). It is therefore important to consider the species distributions when choosing populations 421 for testing intra-specific variation. Nonetheless, growing conditions (with or without stress), species 422 identity and population selection all influence the balance of inter-and intra-specific variation and 423 should be considered in future studies. 424 Early relative growth rate measured in relation to thermal time may ignore sensitivity to light 425 quality 426 For A. myosuroides, the high intra-specific variability in our data (Figure 3 We have shown that local adaptation can be strongly trait-specific, posing the question whether 441 certain traits are under higher selection pressure or are simply more plastic. With respect to 442 implications for future weed models, our results provide additional criteria for explicitly 443 incorporating intra-specific trait variability within the part simulating germination. In the context of 444 typical weed modelling research where the scale of study is often local to regional and centred on 445 sites with a weed community rather than one single species, intra-specific trait variability should be 446 included if the focus is on response traits like the community assembly. It might be negligible if the 447 focus is laid more on effect traits such as ecosystem functioning or net primary productivity. 448 We suggest to increase accuracy of future modelling exercises within a study region by using 449 germination base temperatures from local seed populations. There seems to be no need to 450 specifically measure early growth rates from local populations, although further experiments with 451 new populations increase the precision of species trait averages and ranges. 452