Long-term elevation of temperature affects organic N turnover and 1 associated N 2 O emissions in a permanent grassland soil 2 3

Abstract. Over the last century an increase in mean soil surface temperature has been observed, and it is predicted to increase further in the future. In order to evaluate the legacy effects of increased temperature on both nitrogen (N) transformation rates in the soil and nitrous oxide (N2O) emissions, an incubation experiment and modelling approaches were combined. Based on previous observations that gross N transformations in soils are affected by long-term elevated-temperature treatments we hypothesized that any associated effects on gaseous N emissions (e.g. N2O) can be confirmed by a change in the relative emission rates from various pathways. Soils were taken from a long-term in situ warming experiment on temperate permanent grassland. In this experiment the soil temperature was elevated by 0 (control), 1, 2 or 3 °C (four replicates per treatment) using IR (infrared) lamps over a period of 6 years. The soil was subsequently incubated under common conditions (20 °C and 50 % humidity) and labelled as NO315NH4 Gly, 15NO3NH4 Gly or NO3NH4 15N-Gly. Soil extractions and N2O emissions were analysed using a 15N tracing model and source-partitioning model. Both total inorganic N (NO3− + NH4+) and NO3− contents were higher in soil subjected to the +2 and +3 °C temperature elevations (pre- and post-incubation). Analyses of N transformations using a 15N tracing model showed that, following incubation, gross organic (but not inorganic) N transformation rates decreased in response to the prior soil warming treatment. This was also reflected in reduced N2O emissions associated with organic N oxidation and denitrification. Furthermore, a newly developed source-partitioning model showed the importance of oxidation of organic N as a source of N2O. In conclusion, long-term soil warming can cause a legacy effect which diminishes organic N turnover and the release of N2O from organic N and denitrification.


1.
Introduction 39 Globally, managed pastures were estimated to occupy 34.7 million square kilometres in 2000 40 and this area is projected to increase by a further 13.4% by 2050 (Tilman et al., 2001). 41 Concomitantly, the Earth's mean surface temperature has increased by 0.6°C in the past century 42 with surface temperatures expected to increase by a further 1.5-4. 5°C  sampling. During gas flux analyses the jars were sealed using a clamp and a rubber ring 164 between the jar and the lid. At other times a gap was left between the jar and the lid to allow 165 air exchange while minimising water loss. Two days after soil sampling (day -55), all jars were 166 put in a dark climate chamber at 20°C and 50% humidity and incubated for 55 days prior to 167 15 N substrate addition (day 0). 168 169 Soil gravimetric moisture data were used to determine the exact amount of dry soil in each jar, 170 and to calculate the amount of water to be added to ensure the same soil water content in each 171 jar. On day -53 the soil moisture in each jar was adjusted to a water-filled pore space (WFPS) 172 of 64%. On day -43 and -5 the jars were watered to replenish the water lost due to evaporation. 173 174 For the 15 N tracing study three different labels were used, NO3 15 NH4 Gly, 15 NO3NH4 Gly and 175 NO3NH4 15 N-Gly (at 60, 60 and 99 atm% 15 N respectively). All solutions contained 50 µg NO3-176 N, 50 µg NH4-N, and 30 µg Gly-N g -1 soil. On day 0, the substrate solution was added to each 177 jar using a needle with side-ports, to inject the solution into the soil to minimise disturbance, other extractions took place at 0.11 days (+/-0. gas to ensure a representative sample was taken. The times between t0 and t1 during each of the 201 seven different gas samplings (three before label addition and four immediately prior to 202 extraction) were 120-129, 120, 180, 233, 240, 235 and 214 minutes, respectively. Gas samples 203 were analysed within 24 h after sampling using a GC (Bruker) equipped with an electron 204 capture detector (ECD) for N2O analysis. An average of the concentrations measured in the 15 205 samples was used as the t0 concentration for all 43 jars. Fluxes were based on the ppm and time 206 difference between t0 and t1. They were calculated using the constant gas law, with ambient 207 pressure, and temperature was assumed to be 20°C (the temperature of the incubation room). 208 The fluxes were then converted to a per dry gram basis. 209 210 For the 15 N abundance of N2O, a 30 ml sample was taken at t1 and transferred to a 12 ml This was performed using a double ended needle fixed vertically in a clamp stand with the  214   ventral needle submerged 3-4 mm in a beaker of water and the gas sample held upside down  215   and pushed onto the dorsal needle. The excess pressure in the sample vial was thus released  216 causing the water to bubble until the pressure inside the vial has equilibrated with the ambient 217 atmospheric pressure. Cessation of bubbling implied equal pressure had been reached. The 15 N 218 enrichments of 15 N2O and 15 N2 were determined using an automated isotope ratio mass 219 spectrometry (Sercon Ltd 20-20), as described by Stevens et al. (1993), inter-faced to a TGII 220 cryfocusing unit (Sercon Ltd 20-20). The detection limit for atom% 15  The initial NO3and NH4 + pool sizes were determined by extrapolating the first two extraction 233 times back to time zero. The initial AA pool size was set to 30 µg N g -1 soil, corresponding to 234 the application of glycine (Gly). The initial NH4 + ads and NO3sto were based on the difference 235 between the added and initial N (Müller et al., 2004). The initial pool sizes for organic N (Nrec and for Nlab, the quick turnover time ensures that a small pool will be governed quickly by the 239 dynamics of the in-and out-flowing rates. 240 241 The N transformations are described in Table 1. The N transformations were calculated based 242 on zero or first order kinetics (Table 1). Whether Nlab and Nrec were transformed into AA or 243 The chance that the N2O in the gas sample contains zero, one or two 15 N atoms is described by 293 equations 11, 12 and 13 respectively. Where the subscripts d, n and o refer to the fractions of 294 N2O produced by denitrification, nitrification and oxidation of organic N, respectively. The 295 fraction of N2O produced by co-denitrification is 1-d-n-o as all of the N2O produced was 296 assumed to come from one of the four processes. 297 298 Chance of 0 15 N atoms: n(1-an Chance of 1 15 N atom: 2n(1-an)an + 2d(1-ad)ad + 2o(1-ao)ao + (1-n-d-o)(ad(1-a0)+a0(1-ad)) (12) 300 Chance of 2 15 N atoms: differed with temperature treatment (p<0.0001) (all pairwise comparisons were also 339 significant; p<0.0001). The total inorganic N content was in the order: T1< Tcontrol< T3<T2. 340 341 Soil NH4 + concentrations increased from 2 µg N g -1 soil to between 28 and 54 µg N g -1 soil 342 upon label addition, and subsequently decreased over the next five days to ca. 9 µg N g -1 soil 343 (Fig. 3a). Soil NH4 + concentrations did not differ as a result of the soil warming treatments on 344 either days 0 or 6. However, on day 1, treatment T1 had a lower NH4 + concentration compared 345 to all other treatments (p<0.029), while the soil NH4 + concentration in the T2 treatment was 346 higher than in the Tcontrol or T1 treatments (p<0.001). Three days after label addition the NH4 + 347 concentration in the T1 treatment remained lower compared to the T2 and T3 treatments (p 348 respectively <0.001 and 0.044). 349 350 After the initial increase in NO3due to label addition, the NO3concentrations continued to 351 slowly increase over the following six days (Fig. 3b). NO3concentrations were significantly 352 different among the treatments (p<0.001), with differences also occurring with respect to the 353 initial NO3concentrations prior to label addition (p<0.001). The highest NO3concentrations 354 occurred in the T2 treatment followed by the T3 and Tcontrol, while the lowest NO3concentration 355 was observed in the T1 treatment. 356 357

Soil N transformations 358
The modelled and observed concentrations and 15 N enrichments were in good agreement with 359 gross mineralisation of organic N to NH4 + did not differ with the previously imposed warming 364 treatments. This was because the mineralisation of labile organic N was the major contributor 365 to total mineralisation, and this rate was not significantly affected by previous warming (Table  366 2). Net mineralisation did not differ as a result of the previously imposed warming treatments. 367 Despite the fact that the release of stored NO3tended to increase with warming (p=0.096), and 368 also that cumulative ONH4 and ONrec rates tended to be different (p=0.095), no significant effect 369 on net nitrification could be observed (Table 2). 370 371

N2O fluxes 372
In response to N supply, N2O emissions immediately increased, and decreased thereafter (Fig.  373 3c). While treatments T2 and T3 had lower N2O fluxes than the control treatment (p=0.004 and 374 p=0.036, respectively) no interaction between incubation time and treatment was observed. 375 The N2O fluxes from the T2 treatment were also lower than those from the T1 treatment 376 (p=0.016). However, observed fluxes from the T1 treatment did not differ from the control 377 treatment and N2O fluxes from the T2 treatment did not differ from the T3 treatment. 378

379
The newly developed partitioning model was successful to identify cumulative N2O fluxes 380 ( Fig. 5) and N2O contribution at each extraction time (Fig. 6) associated with nitrification, 381 denitrification, co-denitrification and the oxidation of organic N between 0.11 and 5.93 days 382 after N addition. The oxidation of organic N was the main source of N2O at all sampling dates, 383 comprising between 63 and 85% of the total N2O flux (Fig. 5). The percentage contribution 384 made by organic N to N2O fluxes increased over the sampling period, rising from a minimum 385 of 40% in the control treatment, to virtually 100% across all treatments by Day 6 (Fig. 6). The 386 fluxes from organic N oxidation were the highest in the control treatment, followed by T1, and 387 lowest for T2 and T3. Significant differences were found between the control and the T2 and T3 treatment (p=0.011 and p=0.002, respectively) and between T1 and T3 (p=0.039). The amount 389 of N2O produced via denitrification was also the highest under the control treatment, followed 390 by T1 and T3. It was the lowest under T2. Compared to the control treatment, denitrification 391 contributed less to N2O under the T2 and T3 treatments (p <0.0001 and p=0.002, respectively). 392 The contribution of denitrification also differed between treatments T2 and T1 (p=0.004). Co-393 denitrification only contributed to the N2O flux during the first day after substrate addition. The 394 highest amount of N2O produced via co-denitrification was found under the control treatment, 395 followed by T1. Under T2 and T3 treatments, the contribution of co-denitrification was minor. 396 However, these differences were not significant. No significant differences were found in the 397 amount of N2O produced via nitrification. 398 peat and clay soils). Unfortunately, due to methodological restrictions were not able to detect 481 significant N2 fluxes, as they were <4 g N2-N ha -1 day -1 (Stevens and Laughlin, 1998). denitrifiers reduce O2 rather that NO3 -(Arah, 1997). Any reduction in soil moisture could 498 therefore lead to a decrease in the in situ denitrification rate. 499 500 Co-denitrification was observed to be significant in Tcontrol and T1 shortly after N addition. 501 Rates were comparable with those from true denitrification. Co-denitrification is a co-502 metabolic process which uses inorganic and organic N compounds concurrently and converts 503 it to the same end products as in denitrification. Gases produced in this process are a hybrid N-504 N species where one atom of N comes from NO2and the other one from a co-metabolised 505 compound (Spott et al., 2011). The conditions for increased co-denitrification are still not fully 506 understood, but the presence of fungi along with adequate amino acid pools appears to enhance 507 losses via this pathway (Laughlin and Stevens, 2002;Spott et al., 2011). 508 509 Laughlin and Stevens (2002) found that fungi dominated denitrification and co-denitrification 510 in grassland soils. It has been suggested that warming could increase the relative contribution 511 of fungi to the soil microbial community (Zhang et al., 2005;Pritchard, 2011). Most fungi lack 512 therefore be expected that warming would lead to an increase in N2O produced via 514 denitrification and co-denitrification. However, the opposite was found in the current 515 experiment, although the changes in co-denitrification were not significant. The reduced co-516 denitrification and total denitrification rates seem to indicate a reduction in fungal-mediated N 517 processes under elevated temperatures in these soils. Further research is required to elucidate 518 the effect of increased temperatures on N processes mediated by fungi. 519 520

5.
Conclusion 521 Sustained increases in soil temperatures over 6 years (between 2 and 3°C) led to an increase in 522 both total inorganic soil N and NO3pools. Subsequent analyses of gross N transformations, 523 during an incubation of these soils under common temperature and moisture conditions to study 524 the legacy effect of increased temperatures, revealed that mineralisation of amino acids 525 (glycine) and recalcitrant organic N decreased with previously imposed elevated temperatures. 526 This decrease in mineralisation was also correlated with a decrease in N2O emissions from 527 organic N turnover. However, elevated temperature did not cause a significant change in 528 relative N2O emissions from the different pathways as hypothesised, but it lead to an absolute 529 decrease in N2O emission rates. A new, easy to use, source partitioning method was developed 530 to determine the contribution of four different pathways to N2O emissions. Emissions of N2O 531 in the first six days after fertilisation were decreased for soils previously subjected to higher 532 temperatures as a consequence of a reduction in the rates of denitrification and the oxidation 533 of organic N. For all treatments, oxidation of organic N was the main contributor to N2O 534 emissions, and should therefore in future research not be omitted as a possible source of N2O. 535 The error bars are the standard error of the mean. * shows a significant difference in NH4-N 756 from Tcontrol (p<0.03), # shows a significant difference in NO3-N from Tcontrol (p<0.0001), and Δ 757 shows a significant difference in N2O flux from Tcontrol (p<0.05). 758 Tables 778 Table 1: Description of N transformations and average gross N fluxes per treatment (diagram shown in Fig. 2). Standard deviation between 779 brackets. K stands for Kinetics were 0 implies the use of zero-order and 1 the use of first-order kinetics in the model. The p is the p-value of the 780 one-way ANOVA, with ns (non-significant) if p>0.1 (p value in bold if < 0.05). For the holm-sidak pairwise comparisons: t tends to be different 781 from control (p<0.10). 782