Cereal-legume mixtures increase net CO 2 uptake in a 1 forage system of the Eastern Pyrenees

4 Mercedes Ibañez 1, 2 *; Núria Altimir 2a ; Àngela Ribas 2,3,4 ; Werner Eugster 5 ; 5 Maria-Teresa Sebastià 1,2 6 7 8 1 GAMES group, Dept. HBJ, ETSEA, University of Lleida (UdL). Av. Alcalde Rovira Roure, 9 191, 25198, Lleida, Spain. 10 11 2 Laboratory of Functional Ecology and Global Change (ECOFUN), Forest Science and 12 Technology Centre of Catalonia (CTFC). C/ de Sant Llorenç, 0, 25280 Solsona, Lleida, Spain. 13 14 3 Universitat Autònoma de Barcelona, 08193, Bellaterra, Spain. 15 16 4 Centre for Ecological Research and Forestry Applications (CREAF), 08193, Bellaterra, Spain. 17 18 5 ETH Zürich, Institute of Agricultural Sciences, Universitätstrasse 2, 8092, Zürich, Switzerland. 19 20 21 a Current address: Institute for Atmospheric and Earth System Research (INAR), University of 22 Helsinki, Physicum, Kumpula campus, Gustaf Hällströmin katu, 2 , 00560, Helsinki, Finland. 23 24

Here the activation energy, E 0 (ºC −1 ), is a linear function of SWC (E 0 = a+b·SWC); T ref is the reference 225 temperature, set as the mean temperature of a given period, here set as the mean T s of the entire 226 https://doi.org/10.5194/bg-2020-173 Preprint. Discussion started: 11 June 2020 c Author(s) 2020. CC BY 4.0 License. together (2011-2017 and for each crop season, using the nlsList function. 232 Similarly as in the diversity-interaction model (Sect. 2.3), we performed the R eco , night modelling on all 233 observed data (30-minute average), on daily-averaged data and on weekly-averaged data. Afterwards, we 234 calculated R 2 as the linear relationship between modelled and measured observations. The model 235 performed best (highest R 2 ) when using weekly-averaged data, probably due to the high day-to-day 236 variability of R eco , night and T s . 237

Net biome production (NBP) 238
Finally, in line with our third objective, we estimated the NBP during the growth period. NBP can be 239 estimated knowing the NEE; C exports, including harvest/grazing and other gas emissions such as 240 methane or volatile organic compounds; and C imports, including organic C fertilizers and sowing. In our 241 study, C exports through methane were expected not to be very significant, because methane effluxes 242 require water saturated soils, typically with standing water (Oertel et al., 2016), which was never the case; 243 and volatile organic compounds were expected to be negligible (Soussana et al., 2010). C inputs through 244 sowing and fertilizers (mostly inorganic nitrogen fertilizers, Table 1) could also be neglected as they only 245 represent a very small C amount. Thus, we estimated the NBP during the growth period as the sum of the 246 NEE budget of that period and C exported through the yield Eq.

Forage species influence on CO 2 exchange dynamics and budgets 252
Seasonal CO 2 flux dynamics evolved according to environmental conditions, forage growth and 253 management events (Fig. 2). Maximum net CO 2 uptake was achieved during spring, when temperatures 254 were mild, SWC increased, and the forage development reached its peak biomass (Fig. 2). CO 2 exchange 255 capacity of the system decreased with harvesting ( Fig. 2.a), also showed by the drastic decrease of the 256 NDVI ( Fig. 2.d). 257 The field acted as a net CO 2 sink throughout all the studied crop seasons (negative NEE, Fig. 3 gap-filled data could be used to describe CO 2 exchange dynamics and allowed us to identify this rebound 273 in the net CO 2 uptake during the fallow period of that year. 274 On the contrary, cereal monocultures generally did not show this voluntary regrowth during the fallow 275 period (Fig. 2.d), and gross and net CO 2 uptake capacity of the system decreased drastically (Fig. 2.a). The 276 exception was the wheat monoculture in 2015, when there was vegetation voluntary regrowth after the 277 harvest that resulted in net CO 2 uptake during the fallow period. 278 The diversity-interaction model (Table 2) confirmed the influence of forage species on NEE. The model 279 estimates indicated less net CO 2 uptake in cereal monocultures than in cereal-legume mixtures (Table 2,  280 negative sign in the estimate means uptake), again with a high variability within cereal monocultures. 281 Barley was the cereal monoculture with the lowest net uptake (−1.0 ± 0.3 µmol CO 2 m −2 s -1 , t = −3.39, 282 p < 0.001, Table 2) and triticale was the cereal monoculture with the highest net uptake among the 283 monocultures (−1.6 ± 0.4 µmol CO 2 m −2 s -1 , t = −4.40, p < 0.001, Table 2). Cereal-legume mixtures, 284 however, showed higher net CO 2 uptake rates (oat x vetch −2.0 ± 0.3 µmol CO 2 m −2 s −1 , t = −7.44, 285 p < 0.001, Table 2) than all cereal species in monoculture. The addition of triticale in the mixture did not 286 have a significant effect on NEE (Table 2). 287

Cereal monocultures vs. cereal-legume mixtures: NEE day light response 288
All three light response parameters exhibited pronounced seasonality, as result of phenological changes 289 and management events (Fig. 4). During the growth period, cereal-legume mixtures exhibited on average 290 slightly higher values of GPP sat than cereal monocultures, while R eco,day did not increase (Fig. 5). 291 During the fallow period, cereal-legume mixtures presented on average significantly higher GPP sat and 292 α values than cereal monocultures (Fig. 5), due to the voluntary regrowth of the vegetation (Fig. 2.d), 293 which also caused a rebound on GPP sat and α (Fig. 5). 294

Cereal monocultures vs. cereal-legume mixtures: R eco , night response to temperature and soil 295
water content 296 R eco , night models, based on the equations proposed by Reichstein et al. (2002, our Eq. 4-6), presented a 297 satisfactory fitting, with R 2 ranging from 0.19 to 0.75 across seasons (Table 3). When assessing all seasons 298 together, T s and SWC drove R eco , night (Fig. 6); with an activation energy (E 0 ) significantly dependent on 299 https://doi.org/10.5194/bg-2020-173 Preprint. Discussion started: 11 June 2020 c Author(s) 2020. CC BY 4.0 License. comparison to cereal monocultures (barley, wheat and triticale). Those differences in CO 2 fluxes between 338 cereal-legume mixtures and cereal monocultures could be explained by plant species complementarity, 339 together with mechanisms related to ecophysiological responses, including CO 2 uptake and respiration 340 (Sect. 4.1), as well as management (Sect. 4.2). the marked α and GPP sat rebound during the fallow period . Accordingly, cereal legume 345 mixtures have been reported to increase gross CO 2 uptake, not only via the increased photosynthesis of 346 legumes (Reich et al., 1997(Reich et al., , 2003, but also increasing photosynthesis of the overall community via 347 nitrogen transfer from the legume to the cereal in the mixture. Interestingly, our results showed that this 348 increase in the gross CO 2 uptake and the photosynthetic activity was not accompanied by a significant 349 increase of daytime respiration rates (R eco,day ,.