During photosynthetic induction, biochemical and stomatal limitations differ between 3 Brassica crops 4

12 Interventions to increase crop radiation use efficiency rely on understanding how biochemical 13 and stomatal limitations affect photosynthesis. When leaves transition from shade to high 14 light, slow increases in maximum Rubisco carboxylation rate and stomatal conductance limit 15 net CO 2 assimilation for several minutes. However, as stomata open, intercellular [CO 2 ] 16 increases, so electron transport rate could also become limiting. Photosynthetic limitations 17 were evaluated in three important Brassica crops: B. rapa , B. oleracea and B. napus . 18 Measurements of induction after a period of shade showed that net CO 2 assimilation by B. 19 rapa and B. napus saturated by 10 min. A new method of analyzing limitations to induction 20 by varying intercellular [CO 2 ] showed this was due to co-limitation by Rubisco and electron 21 transport. By contrast, in B. oleracea , persistent Rubisco limitation meant that CO 2 22 assimilation was still recovering 15 min after induction. Correspondingly, B. oleracea had the 23 lowest Rubisco total activity. The methodology developed, and its application here, shows a 24 means to identify the basis of variation in photosynthetic efficiency in fluctuating light, which 25 could be exploited in breeding and bioengineering to improve crop productivity.


Introduction 31
The continued growth of the global human population and its increasing urbanisation will 32 lead to increased pressure on farming systems over the next half century, and increased 33 productivity on the land we are already using will be crucial to minimize the environmental 34 impacts (Tilman, Balzer, Hill & Befort 2011). In this context, it is essential to understand 35 photosynthetic efficiency because it fundamentally affects the productivity and efficiency of 36 resource use by crops. The majority of crops use C3 photosynthesis, which requires massive 37 investment of nitrogen in leaf chloroplasts, where 21-74% of leaf soluble protein is allocated 38 to the primary CO2 fixing enzyme ribulose-1,5-bisphosphate (RuBP) carboxylase/oxygenase 39 (Rubisco; Carmo-Silva, Scales, Madgwick & Parry 2015). Furthermore, A cost of allowing 40 CO2 into the leaf for photosynthesis, is the escape of water vapour via transpiration (Farquhar 41 & Sharkey 1982;Raschke 1975). Consequently, crop biological N2 fixation and crop applied 42 N fertilisers now account for more than 44% of the total annual N entering the global 43 biosphere (Fowler et al. 2013), and crop irrigation accounts for 70% of annual global human 44 water use (Haddeland et al. 2014). 45 The major focus of studies of crop photosynthetic efficiency has been under light-46 saturating steady-state conditions. Yet these are rare for crop leaves in the field or glasshouse. 47 (Indermühle et al. 1999;Larson et al. 2014;Sage 1995) Yara, Grimsby, UK), and were transplanted to 1.5 L pots one week after emergence, in each 162 case using a soil-less compost mix (Petersfield Products, Leicester, UK) that incorporated a 163 broad range fertilizer. Checks were made daily to ensure that compost was kept moist without 164 overwatering. 165 Plants used for biochemistry were also sown, germinated and transplanted to 1.5 L 166 pots in the greenhouse, containing the same compost mix as above. then simulated by a step decrease in PPFD to 150 μmol m −2 s −1 for 30 min, followed by a step 202 increase back to 1500 μmol m −2 s −1 . Gas analysers were matched one minute before starting 203 the sun-shade-sun sequence, and measurements were logged every 10 s from one min before 204 shade until at least 28 min after shade. 205 The following key timesteps from the 10 s resolution induction curves were 206 identified. First, the end of the RuBP regeneration dominated 'fast-phase' of induction was 207 taken to be 2 min after the return to high light, following shade. Second, tci,min was the time at 208 which minimum ci was observed during induction, marking the transition between 209 predominant limitation by non-stomatal factors (which results in decreasing ci) and increasing 210 stomatal conductance (gs; which results in increasing ci). Next, tA,90 was the timepoint at μmol mol −1 and ≥ 500 μmol mol −1 ranges were interspersed randomly (e.g., 800, 200, 600, 234 100, 500, 400, 700, 300, 1000, 50), and were rotated over ten separate inductions so that 235 every [CO2] was measured at every interval between 2 and 20 min following shade 236 ( Supplementary Fig. 2). To aid with consistency of responses, measurements were made in 237 the laboratory (i.e., low light, and relatively constant temperature and humidity conditions), 238 and between inductions gas exchange was allowed to fully recover to steady state at reference 239 [CO2] of 430 μmol mol −1 . To ensure that induction measurements for a leaf could be captured 240 within a single day, two LI-6800F were used, attached adjacent to one another, either side of 241 the mid-rib. 242

Models 243
The relationship between A and incident PPFD was modelled as a non-rectangular hyperbola 244 (Long & Hallgren 1985  In the dynamic A/ci analysis, where greater measurement error and a slightly reduced 292 number of measurements made least-squares fits less reliable, genotype-level parameters 293 from the steady-state A/ci measurements were used to ensure A/ci fits provided a reasonably were dropped and the model was refit, dropping the highest ci data as necessary until a best-299 fit admissible model was found that either (a) included both AC and AJ, or (b) included AC 300 alone. When a best fit model with AC alone was reached, because identification of AJ requires 301 N ≥ 2, the uppermost ci value was dropped to prevent mis-attribution of data that could be 302 assigned to AJ and the model was refit, taking the highest ci used as a lower-bound value for 303 ci,trans. 304 Stomatal limitation (LS) was calculated from the steady-state A/ci responses following 305 Farquhar & Sharkey (1982): 306 Where, A0 is a reference net CO2 assimilation rate predicted at a ci equal to leaf external 308 [CO2], and A was the rate observed at the initial reference [CO2] of 430 μmol mol −1 . 309

Photosynthetic response to light and leaf biochemistry 353
Leaf level responses to PPFD (Fig. 1) showed mean values of Asat Rd, and θ that were highest 354 for B. rapa, slightly lower for B. napus, and lowest for B. oleracea (Fig. 1). By contrast, ϕ 355 was greater in B. oleracea and B. napus than in B. rapa. There was limited support for 356 significant differences in Rd (F2,9 = 2.22, P = 0.16) and ϕ (F2,9 = 2.56, P = 0.13) across the 357 three Brassica. However, differences in Asat were marginally significant (F2,9 = 3.03, P = 358 0.099), and there was strong evidence for a significant difference in θ (F2,9 = 9.91, P = 0.005). 359 The smaller θ for B. oleracea compared with B. napus and B. rapa, supports a more gradual 360 transition from light-to carboxylation-limited photosynthesis at higher PPFDs and was 361 significant for both individual comparisons (P ≤ 0.026). 362 The observed patterns of differences in mean Rubisco total activity and Rubisco 363 amount were consistent with marginally significant differences in mean Asat. Rubisco amount 364 and total activity were lower in B. oleracea than in B. napus and B. rapa (Table 1), though 365 these differences were not significant among the three species (F2,12 ≤ 1.6, P ≥ 0.24). activities among the three Brassica (Table 1), implying that patterns of difference in total 368 activity were strongly affected by amounts of Rubisco protein per unit leaf area. Interestingly, 369 while the lower Rubisco content of B. oleracea leaves was paired with similar total soluble 370 protein to B. rapa (P = 0.94), these two species showed marked differences in chlorophylls. 371 B. oleracea had approximately double the amount of chlorophyll a+b (P < 0.001), and lower 372 chlorophyll a:b ratios (P = 0.001) compared with B. rapa (Table 1) Table 2). The significant differences between A and gsw of B. oleracea and B. rapa were 386 associated with an increase in mean ci from 26.5 (B. oleracea) to 29.3 Pa (B. rapa), but 387 measurements were not sufficiently repeatable across the small number of replicates to 388 establish a significant difference in ci among the three species (F2,9 = 2.56, P = 0.13; Table 1). 389 The similarity in operating ci, and differences in A and gsw between the Brassica were 390 associated with differences in steady state A/ci responses ( Fig. 2; Supplementary Fig. 1).
LS were (F2,9 = 5.01, P = 0.035), specifically between B. rapa and B. oleracea (P = 0.037, 395 other comparisons P ≥ 0.089; Table 2). There was also a marginally significant difference in and J limitation ( Fig. 2; Supplementary Fig. 1). Finally, though at much higher ci than the 404 operating point, a highly significant difference was also shown for the ci at which AJ 405 transitioned to AP (F2,9 = 10.38, P = 0.006), between B napus, which had the lowest value for 406 the ci of this transition, and B. oleracea, which had the highest ( Fig. 2; P = 0.005). 407

Recovery of A during fast, mesophyll-dominated, and stomata-limited induction 409
The vast majority of recovery in A occurred while ci was decreasing, i.e., while recovery of A 410 was controlled primarily by non-stomatal factors (Fig. 3); recovery of A during this 4-5 min 411 period (tci,min, Table 3) averaged 77-84% (Rci,min, Table 3). After 30 min shade at the 412 relatively high shade-irradiance of 150 µmol m −2 s −1 , ~ 70% of recovery occurred during the 413 first 2 min (fast-phase), so slow-phase recovery prior to increases in ci accounted for ~ 10% 414 of the shade-sun difference in A (Table 3). When the fast-and slow-phase components of non-stomatal-dominated recovery were taken together, neither their combined impact on 416 recovery of A nor their combined duration were significantly different between the three By contrast with non-stomatal-dominated induction, the remaining 20% of recovery 419 in A, that was predominated by the effect of increasing gs on ci, took significantly longer in B. 420 oleracea than in B. rapa (tA,90 − tci,min, Table 3; P = 0.02), and was marginally significantly 421 longer in B. oleracea than B. napus (Tukey HSD, P = 0.055; Table 3). Mean A, gsw and ci of 422 B. oleracea had not approached their steady-state values even after 20 min of induction (Fig.  423 4a), such that tA,90 was significantly longer in B. oleracea than the other two species (Table 3; (Table 3), even though, like B. oleracea, their gsw and ci continued to increase 427 beyond 20 min, A was insensitive to this ( Fig. 3 and 5). 428

Apparent limiting biochemical factors during induction -dynamic A/ci 429
Progressive changes in Vc,max determined from dynamic A/ci responses were qualitatively 430 different between the three Brassica (Fig. 4). Increases in Vc,max during induction were: 23% 431 in B. oleracea, 33% in B.napus and 29% in B. rapa. The rate of change in Vc,max (dVc,max/dt) 432 declined smoothly (Fig. 4d), and confirmed that increases in Vc,max were predominantly over 433 the first ~ 10 min of induction in B. oleracea , ~ 12 min in B. rapa (Fig. 4a, c & d), and ~18 434 min in B. napus (Fig. 4b & d). In all three, Vc,max increased rapidly for the first 4-5 min of 435 induction, coinciding with the tci,min observed in induction measurements (Table 3). It was 436 also notable that Vc,max of B. oleracea saturated before tA,90 from the ambient induction 437 experiments, whereas increases in Vc,max of B. napus and B. rapa were continuing at their 438 tA,90, but with little subsequent effect on A (Fig. 4). 95% confidence intervals, ci was significantly less than ci,trans throughout induction for B. 447 oleracea (Fig. 5a), until ~ 10 min for B. napus (Fig. 5b), and until ~ 7 min in B. rapa (Fig.  448 5c), with ci intersecting mean ci,trans after 10-15 min induction in B. napus and B. rapa. 449 Because ci < ci,trans infers that A is limited by Vc,max, as ci < ci,trans throughout induction A of B. 450 oleracea was always Vc,max-limited, and the other two species were Vc,max limited beyond tA,90 451 (Table 3). Because ci,trans denotes a change in the slope of the A/ci response, overlap between 452 ci,trans and ci of B. napus and B. rapa during induction explains why A saturated while their gs 453 and ci continued to increase (Fig 3b & c). The initial decrease in ci always extended to ~ 4-5 min of induction, at least twice the 535 upper limit for the fast-phase is taken from the literature (e.g., Sassenrath-Cole & Pearcy 537 1992), and was used because gas exchange system mixing times meant that fast-phase 538 kinetics could not be directly parameterised. The inflection of A indicating the end of the fast 539 phase nonetheless tended to occur slightly before 2 min (e.g., Fig. 3), so the estimate of 540 photosynthetic recovery driven by Rubisco activation, at 2-3 min duration and 10%, is 541 conservative. Evidence that shade-induced Rubisco deactivation can limit midday 542 photosynthesis in field crops is consistent with previous detailed measurements of apparent 543 alternative water use strategies (Cowan & Farquhar, 1977;Cowan, 1982   Different superscripts indicate significant differences at P < 0.05 using Tukey's HSD. 841 Different superscripts indicate significant differences at P < 0.05 using Tukey's HSD. 847