Responses of forest carbon and water coupling to thinning treatments 5 across multiple spatial scales

Water-use efficiency (WUE) represents the coupling of forest carbon and water. Little is known 25 about the responses of WUE to thinning at multiple spatial scales. The objective of this research 26 was to use field measurements to understand short-term effects of two thinning treatments (T1: 27 4,500 stems ha -1 ; and T2: 1,100 stems ha -1 ) and the control (C: 27,000 stems ha -1 ) on WUE (the 28 ratio of leaf photosynthesis to leaf transpiration, or tree growth to tree transpiration, or net stand 29 above-ground biomass (AGB) accumulation to stand transpiration) and the intrinsic WUE 30 (WUEi, the ratio of leaf photosynthesis to stomatal conductance or the net stand AGB 31 accumulation to canopy conductance) in a 16-year old natural lodgepole pine forest. Leaf-level 32 measurements were conducted in 2017, while tree- and stand-level measurements were 33 conducted in both 2016 (the normal year) and 2017 (the drought year). 39 The effects of thinning on forest carbon and water coupling differed with the spatial scales and 47 the metrics (WUE or WUEi) of water use efficiency. Lacking consistent reponses of WUE 48 metrics to thinning treatments across the spatial scales suggests that caution must be exercised 49 when transferring and modeling WUE from one spatial scale to others. Both tree-level and stand- 50 level WUE values in the more heavily thinning stands were significantly promoted under the 51 drough condition, demonstrating that thinning can improve WUE and consequently support 52 forests to cope with the drought effects. 53


Abstract 23
Background 24 Water-use efficiency (WUE) represents the coupling of forest carbon and water. Little is known 25 about the responses of WUE to thinning at multiple spatial scales. The objective of this research 26 was to use field measurements to understand short-term effects of two thinning treatments (T1: 27 4,500 stems ha -1 ; and T2: 1,100 stems ha -1 ) and the control (C: 27,000 stems ha -1 ) on WUE (the 28 ratio of leaf photosynthesis to leaf transpiration, or tree growth to tree transpiration, or net stand 29 above-ground biomass (AGB) accumulation to stand transpiration) and the intrinsic WUE

36
There was no significant effect of thinning on the tree-and stand-level WUE in 2016, while in 37 2017, only T2 exhibited significantly higher tree-level WUE (0.63 mm 2 kg -1 ) than the C (0.06 38 mm 2 kg -1 ), and the stand-level WUE values were significantly higher in the thinned stands, with 39 the means of 0.34, 0.61 and 0.7 kg m -3 for the control, T1 and T2, respectively. Stand-level 40 WUEi was, however, significantly higher in the unthinned stands than in the thinned stands. In ratio of tree growth (e.g., basal area increments (BAI)) to the whole tree transpiration 81 (Wullschleger et al., 1998). And at the ecosystem level, WUE can be quantified as the ratio of 82 gross primary production to evapotranspiration or the ratio of net primary production to 83 transpiration (e.g., Petritsch et al. (2007)). The responses of WUE at finner spatial levels (e.g., 84 leaf and individual tree) provide valuable information for understanding and predicting behaviors 85 of carbon and water processes of forest ecosystems at coarser spatial scales. 86 Holistic view on forest WUE across spatial scales is limited. Leaf-level studies are generally 87 concentrated on the leaf-level WUEi that is detected from isotopic signatures of tree tissues, 88 because when comparing with leaf-level WUE, the isotopic leaf-level WUEi only accounts for 89 the ratio of intercellular and ambient CO 2 concentration, and can cover periods of low light, low 90 temperature and dry conditions, while leaf-level WUE that is usually conducted by the gas 91 exchange method, reflects optimal conditions of trees (near light saturation and optimal ranges of 92 5 temperature) due to limitations in conducting measurements in the field (e.g., during night) 93 (Wieser et al., 2018). The leaf-level WUEi can also be measured from the gas exchange 94 technique, and the discrepancy between WUEi of two different methods has been ascribed to the 95 differences in the time scale (i.e., long term and short term) (Wieser et al., 2018) because the 96 ratio of photosynthesis to stomatal conductance is positively linearly with the ratio of the leaf 97 intercellular CO 2 concentration (Ci) to the atmospheric CO 2 concentration (Ca), and the Ci/Ca is 98 also assumed to be linearly related to the carbon isotope discrimination by approximating the 99 intercellular chloroplast concentration to Ci (Farquhar et al., 1989;Seibt et al., 2008   The mean daily temperature during the growing season of 2017 is 12.1⁰C, and the total growing Leaf-level WUE (μmol mmol -1 ) was calculated as the ratio of leaf photosynthesis rate (μmol CO 2 188 m -2 s -1 ) to leaf transpiration rate (mmol H 2 O m -2 s -1 ). Leaf-level WUEi (μmol mol -1 ) was 189 calculated as the ratio of leaf photosynthesis rate (μmol CO 2 m -2 s -1 ) to stomatal conductance  Given that the studied stand is even-aged and mono-species forest with sparse understory, NPP is 235 estimated by changes in the stand above-ground biomass (AGBstand, g) in each growing season,     factors. Stand-level WUE was log-transfomred before the two-way ANOVA analysis (Table S4) (Figure 1, Table S1). The ANCOVA test showed that thinning significantly affected tree-level WUE (p = 0.033, Table   317 S2 respectively, and only tree-level WUE between C and T2 was statistically different (p=0.009).

321
Thus, the heavier thinning significantly improved tree-level WUE in the drought year.

322
However, the ANCOVA test showed that the drought did not significantly affect tree-level WUE 323 (p=0.4, Table S2). And for each of the three groups, their tree-level WUE also did not 324 statistically differ between years (all p>0.1). These were probably due to the large variances in 325 the tree-level WUE of T1 and T2 in the non-drought year (Figure 2). In addition, there was no 326 significant interaction effect between year (drought) and thinning (p=0.84, Table S2). The ANOVA test showed that the thinning did not have significant impacts on the stand-level 332 WUE (p = 0.27, Table S4). In 2016, stand-level WUE was 1.57±0.347, 1.73±0.245, and

338
The drought significantly reduced stand-level WUE (p < 0.001, Table S4), while the interaction 339 between thinning and drought did not play a significant role (both p > 0.1) (Table S4, Figure 3).

494
Except for that, the stand-level WUE exhibited good correspondences with the tree-level WUE.

495
As expected, the unthinned stand had the least tree growth but the highest stand transpiration 507 Surprisingly, the stand-level WUEi was significantly higher in C than in the thinned stands with 508 no significant difference between T1 and T2. The net accumulation of the stand above-ground 509 biomass was significantly higher in C than T1 (p = 0.004) and T2 (p = 0.001), and there was no Thinning changed the responses of WUE to microclimate at the tree and stand levels, but to our 522 surprise, thinning did not change the sensitivity of the leaf-level WUE to PAR, and leaf-level 523 WUEi to VPD and PAR. And the relationship between the leaf-level WUE with VPD fits the 524 exponential increasing function (Figure 4) and their relationship was modulated by the transmitted solar radiation (Figures 6 and 7). Such 546 relationship between the tree-level WUE and VPD fits the exponential decay function, which is 547 in agreement with the previous study (Lindroth and Cienciala, 1996). Although there was no 548 significant correlation between tree-level WUE and leaf-level WUE and WUEi (both p >0.1) 549 probably due to limited sampling data, our study agrees with the research reporting that tree-550 29 level WUE was primarily a function of VPD (Table 1)  Thinning changed the responses of tree-level WUE to VPD, resulting in the apparent lower tree-553 level WUE in C than in T1 and T2 at each VPD level ( Figure 5). It is apparent in the Figure 5 554 that the lower responses of the tree-level WUE to VPD in C was because C generally had a lower 555 transmitted solar radiation, and thus lower responses of the tree-level WUE to VPD.

556
At the stand level, the response of the stand-level WUE to VPD was similar to the tree-level 557 response, exhibiting the exponential decay relationship in this study. This has also been 558 described by Lindroth and Cienciala (1996) and Kuglitsch et al. (2008). However, thinning 559 seemed to affect the response of the stand-level WUE to VPD, although the correlation was only 560 significant in C in our study. Besides, both the stand-level WUE and WUEi were significantly 561 correlated with transmitted solar radiation, which was probably because the stand-level WUE 562 and WUEi were primarily driven by the net above-ground biomass accumulation as discussed in 563 the previous section, and the net above-ground biomass accumulation is depdent on the light 564 availability (Jarčuška and Barna, 2011). Besides, we did not find any significant relationship 565 between the stand-level WUE and WUEi and the tree-level WUE (both p>0.2). Therefore, 566 whether the stand-level WUE can be predicted from the tree-level WUE requires further study.
567 Table 1. Correlation coeffieicnt between WUE with VPD, light intensity and soil water content. Last but not least, although thinning did not affect the sensitivity of the leaf-level WUE and 572 WUEi, the changes in microclimate resulting from thinning can still lead to the differences in 573 WUE between the control and thinned stands (e.g., leaf-level WUEi and tree-level WUE).  The data supporting this research are included within the article and its additional files.

608
Additional data are available upon request to corresponding author. 609 610