Jukuri, open repository of the Natural Resources Institute Finland (Luke) All material supplied via Jukuri is protected by copyright and other intellectual property rights. Duplication or sale, in electronic or print form, of any part of the repository collections is prohibited. Making electronic or print copies of the material is permitted only for your own personal use or for educational purposes. For other purposes, this article may be used in accordance with the publisher’s terms. There may be differences between this version and the publisher’s version. You are advised to cite the publisher’s version. This is an electronic reprint of the original article. This reprint may differ from the original in pagination and typographic detail. Author(s): Simone Bianchi, Daesung Lee, Sara Bergante, Gianni Facciotto, Jari Hynynen & Giuseppe Nervo Title: Climate-wise models of biomass productivity for hybrid poplar clones in Europe Year: 2023 Version: Published version Copyright: The Author(s) 2023 Rights: CC BY-NC 4.0 Rights url: http://creativecommons.org/licenses/by-nc/4.0/ Please cite the original version: Bianchi S, Lee D, Bergante S, Facciotto G, Hynynen J, Nervo G (2023). Climate-wise models of biomass productivity for hybrid poplar clones in Europe. iForest 16: 188-194. - doi: 10.3832/ifor4211-016 ii F o r e s tF o r e s t Biogeosciences and ForestryBiogeosciences and Forestry Climate-wise models of biomass productivity for hybrid poplar clones in Europe Simone Bianchi (1), Daesung Lee (1), Sara Bergante (2), Gianni Facciotto (2), Jari Hynynen (1), Giuseppe Nervo (2) Renewable bioenergy has the potential to contribute sustainably to the energy sector. Forestry is the main source of biomass for energy in Europe, and poplar (genus Populus) is widely used for short rotation coppice (SRC). Many studies have assessed poplar clones’ productivity but there is a lack of regional studies and links with the climate. We investigated the biomass productivity of twenty hybrid poplar clones for SRC. Clones were planted in sixteen locations across nine countries in Europe, although not all clones were replicated in all loca- tions. In each location, clones were planted in three replicated plots. All plots were harvested after four years, and the aboveground dry biomass estimated. We fitted clone-specific linear mixed models of total aboveground dry biomass production at plot level as function of climatic variables. For many clones (eight) only annual heat moisture deficit negatively affected productivity, in few cases (3) together with a quadratic term for a smoother relationship. In some other clones (five) only the mean summer precipitation positively and linearly affected productivity. On average, the variance explained by the fixed effects in those models was 56%. For the remaining clones (seven), no climate variables resulted significant. Our study explicitly investigated the quantita- tive link between water availability and poplar SRC productivity, one of the most important known factors but not often studied with a modelling ap- proach. Further, we show the most productive clones in dried conditions. We also highlight the need to larger scale regional experiments to produce models that can be used in climate change scenarios. Keywords: Hybrid Poplar, Short Rotation Coppice, Aridity Index, Water Avail- ability, Above Ground Biomass Introduction Bioenergy can play a key role in achieving the EU’s renewable energy target, pro- vided it is produced, processed and used in a sustainable and efficient way (EC/JRC 2019). Forestry is the main source of bio- mass for energy in Europe, and short rota- tion coppice (SRC) using fast growing spe- cies is an important potential component (Bentsen & Felby 2012). In Europe, the Sali- caceae family, Populus spp. and Salix spp., presents the greatest developments on an industrial level for SRC (Oliveira et al. 2020). However, the global consensus is that growing woody SRCs for energy pro- duction is not yet economically feasible (Rodrigues et al. 2020). So far, data on pro- ductivity of SRCs are rather scarce and lim- ited to small experimental-scale planta- tions (Verlinden et al. 2015); therefore bet- ter information are needed to improve the decision making process on SRCs. Poplars are among the fastest growing trees in temperate latitudes and their high productivity comes with high water de- mands, although a wide diversity in water requirements has been reported (see Nav- arro et al. 2014 and references therein). To improve and optimize the biomass produc- tion across Europe, it is important to know which clone can perform better at a given site according to its characteristics. To this purpose, a better understanding of the re- lationship between productivity and local climate is required. Various studies have analysed the produc- tivity of poplar SRCs by comparing differ- ent clones (Dillen et al. 2013, Manzone et al. 2014, Navarro et al. 2014, Verlinden et al. 2015, Fernández et al. 2016, Niemczyk et al. 2018, Oliveira et al. 2020). However, most of the mentioned studies addressed only one country, generally with a limited num- ber of clones established in few sites. Fur- ther, even when a large number of clones/ sites were used (Nielsen et al. 2014, Land- graf et al. 2020), productivity was com- pared to identify the best performing clones at the experimental site, but with- out investigating their relationships with the site characteristics. To the best of our knowledge, only few studies focused on modeling poplar SRC biomass productivity as a function of climate variables, with con- trasting results (Bergante et al. 2010, Fis- cher et al. 2014, Njakou Djomo et al. 2015). The latter was the only study we found in- volving data across different European countries. Werner et al. (2012) employed a process-based model for simulating poplar SRC productivity, but it was validated only against one clone. Navarro et al. (2014) identified more resistant clones to arid con- ditions by investigating their stomatal be- havior from only one site. The objective of this study was to over- come the gap in cross-European studies on climate-productivity relationships in poplar SRCs for bioenergy production. The final aim is to provide stakeholders with better information for decision making at a larger scale such as landscape planning (Ro- drigues et al. 2020). We used a dataset in- cluding several sites from various countries across Europe where the same poplar clones were planted in SRCs. We calibrated © SISEF https://iforest.sisef.org/ 188 iForest 16: 188-194 (1) LUKE, Latokartanonkaari 9, F-00790, Helsinki (Finland); (2) CREA, str. Frassineto 35, I-15033, Casale Monferrato, AL (Italy) @@ Simone Bianchi (simone.bianchi@luke.fi) Received: Aug 25, 2022 - Accepted: Mar 30, 2023 Citation: Bianchi S, Lee D, Bergante S, Facciotto G, Hynynen J, Nervo G (2023). Climate-wise models of biomass productivity for hybrid poplar clones in Europe. iForest 16: 188-194. – doi: 10.3832/ifor4211-016 [online 2023-06-30] Communicated by: Andrea Cutini Research ArticleResearch Article doi: doi: 10.3832/ifor4211-01610.3832/ifor4211-016 vol. 16, pp. 188-194vol. 16, pp. 188-194 http://www.sisef.it/iforest/contents/?id=ifor4211-016 mailto:simone.bianchi@luke.fi Bianchi S et al. - iForest 16: 188-194 linear mixed models to identify the rela- tionship between climate variables and clones’ productivity. Material and methods Study sites and clones We retrieved data on twenty hybrid pop- lar clones, belonging to eight different genetic groups (or “families”), planted in 16 locations across 9 European countries (Fig. 1). On average, there were 15 clones planted in each site (from 8 to 20). Tab. 1 shows more details about the location of each site, while the genetic group for each clone is reported in Tab. 2. Experimental design All plots were established in winter 2013- 2014. There was a common minimum de- sign across all locations. At least four rows with ten plants were planted in a plot (spacing 3 × 1 m) over a minimum area of 120 m2, with a buffer of one row each side to avoid edge effects. When a clone was planted in a location, there were three replicates. All plots were ploughed approx- imately to 30-40 cm, and weed were con- trolled manually or chemically until the end of July 2014. No fertilization was carried out. Plots were chosen as homogenous as possible in flat locations on arable land. Irrigation was carried out in five sites in Mediterranean conditions (Spain and Italy). Irrigation strongly and positively affected the productivity of the plots, thus introduc- ing a bias in the climate-growth relation- ship. Indeed, preliminary results showed that when a dummy variable reflecting wa- tering is included in the productivity mod- els, no significant climate-growth relation- ship is found. Furthermore, no clear details on the irrigation practices carried out were available at these sites, therefore they were excluded them from modelling. After removing those sites, each clone was em- ployed in at least other seven locations. After four years (winter 2017-2018), the 16 plants in the centre of each plot were har- vested and fresh weighted. For most loca- tions wood moisture was directly evalu- ated by comparing fresh and dry weight of samples collected from the harvested trees. Wood moisture ranged from to 47% to 65%, with an overall mean of 56%. For other sites (Bajiti, Hungary; Frassineto and San Isidoro, Italy; Guadalajara and La Canaleja, Spain), no data on wood mois- ture were available, and in these cases it was considered equal to the average level (58%), except for the Suniglia site, where the same values of nearby Casale Monfer- rato site were used. Consequently, for each plot, total aboveground dry biomass pro- ductivity (AGBp, Mg ha-1 year-1) was calcu- lated as the variable of interest. While we retrieved and kept separately the values for the three plot of each clone in most of the locations, for one (Tvrdavica, Croatia) we were able to retrieve only the average of the three plots for each clone (data from Vusić et al. 2019). Data analysis Climatic data for all the study sites were retrieved from ClimateDT (https://ibbr.cnr. it/climate-dt/), a web portal where scale- free climatic data are freely provided at global level using CRU-TS data (Harris et al. 2020) for the historical period (1901-current year – Marchi et al. 2022). Several variables of interest can be selected on the portal, based on monthly and annual values of temperature and precipitation, along with a wide array of derivative indices. Climatic values for each site were retrieved for the years 2014-2017 and averaged. We carried out a preliminary analysis of the relation- ship between the climatic variables and AGBp. According to Marchi et al. (2022), we selected the following variables as poten- tial predictors of productivity: Mean Sum- mer Precipitation (MSP, mm), calculated from May to September; a continentality index (CONT), defined as the standard de- viation of the average monthly tempera- tures; and the Annual Heat Moisture (AHM, °C mm-1), defined as (eqn. 1): (1) where MAT is the Mean Annual Tempera- ture (°C) and MAP the Mean Annual Precip- itation (mm). High MSP corresponds to high water availability during the vegeta- 189 iForest 16: 188-194 Fig. 1 - Map of the study sites, with indication of their mean productivity (m 3 ha-1 year-1) considering all clones and plots (both irrigated and non-irrigated sites). Tab. 1 – List of the locations where poplar clones were employed. Coordinates are in WGS84 decimal degrees. Sites with an asterisk were irrigated and excluded from modelling. (AHM): Annual Heat Moisture. Location Country Lat N Long E AHM (°C m-1) 1 Belgium 50.763 3.879 34.3 2 Bulgaria 42.217 24.320 56.5 3 43.511 23.367 49.3 4 Croatia 45.576 18.686 42.2 5 Czech Republic 49.992 14.578 49.4 6 Germany 47.863 12.064 24 7 47.973 10.823 23.9 8 Hungary 47.276 16.973 46.1 9 47.314 19.485 53.4 10 Italy 41.906 12.355 42.1 11* 45.142 8.516 34.6 12* 44.709 7.678 35.9 13* 44.595 7.617 30.8 14 Romania 44.192 27.308 60 15* Spain 40.666 -3.169 76.8 16* 40.513 -3.310 73.7 iF or es t – B io ge os ci en ce s an d Fo re st ry AHM= MAT +10 MAP⋅10−3 https://ibbr.cnr.it/climate-dt/ https://ibbr.cnr.it/climate-dt/ Climate-wise models for hybrid poplar clones’ biomass productivity in Europe iForest 16: 188-194 190 Tab. 2 - Details of the study clones. Codes for genetic groups: (d) Populus deltoides; (k) P. koreana; (m) P. maximowiczii; (n) P. nigra; (t) P. trichocarpa. Sites with an asterisk were irrigated and excluded from modelling. (AGBp): above ground biomass productivity (mean ± standard deviation). Clone name Genetic group Locations AGBp (Mg year-1 ha-1)1 2 3 4 5 6 7 8 9 10 11* 12* 13* 14 15* 16* AF18 (d×n) x - - - - x x x x x x x x x - - 9.536 ± 5.410 AF34 (d×n) x x x - - x x - - x x x x x x x 8.649 ± 5.468 AF8 (t×d) x x x - x x x x x x x x x x x x 7.090 ± 4.116 Antonije (d×n)×d x - - x - x x x - x x - x x x x 4.930 ± 3.298 Baldo (d×d) x x x x - x x x x x x x x x x x 7.133 ± 5.290 Brenta (d×n) x x x - x x x x x x x x x x x x 6.750 ± 4.379 Dellinois (d×d) x - - - - x x - - x x - x - x x 8.385 ± 3.300 Delrive (d×d) x - - x - x x - - x x - x - x x 7.617 ± 3.201 Grimminge d×(t×d) x x x - - x x x x x x x x x x x 5.834 ± 3.859 Hybride275 (n×m) x x x x x x x x - - x x x x x x 3.470 ± 2.805 Koreana (t×k) x x x x x x x x - - x x x x x x 5.302 ± 5.341 Matrix21 (t×m) x x x x x x x x - - x x x - x x 4.461 ± 4.034 Max 1-4 (n×m) x x x x x x x x - x x x x x x x 6.323 ± 3.411 Muur (d×n) x - - - x x x x x x x x x x - - 4.740 ± 3.359 Orion (d×n) x - - - - x x x - x x x x x - - 10.293 ± 5.348 Oudenberg (d×n) x - - - - x x x x x x x x x - - 6.783 ± 3.999 SV885 (d×n) x - - x - x x x - x x - x x - - 5.580 ± 3.024 SW 822 (d×n) x - - - - x x x x - x - x x - - 5.108 ± 1.927 Skado (t×m) x x x - x x - x x - x x x x x x 6.061 ± 4.846 Vesten (d×n) x x x - - x x x - x x x x x - - 5.709 ± 3.607 iF or es t – B io ge os ci en ce s an d Fo re st ry Fig. 2 - Comparison of observed AGBp (above ground biomass productivity, in Mg year-1 ha-1) vs. its predicted values for all clone-specifics models. Bianchi S et al. - iForest 16: 188-194 tive season, high CONT corresponds to high degree of continentality in terms of annual temperature variation, and high val- ues of AHM are due to high temperature and/or low precipitation, resulting in low water availability for the plants. Soil data were retrieved from SoilGrids (https://soilgrids.org/), a system for digital soil mapping based on a global compilation of quality-assessed and standardised soil profile data (WoSIS – Batjes et al. 2020). Af- ter preliminary analyses, the soil parame- ters retrieved were either not significant for most of the clones, or had negligible in- fluence, or were not biologically sound. Thus, all soil parameters were excluded as predictors from modelling. Unfortunately, no soil data collected directly in the field were available at the study sites. We fitted clone-specific linear mixed mod- els for AGB productivity as follows (eqn. 2): (2) where quadratic terms were used to ac- count for possible non-linearity of the re- sponse; ul are random effects for each loca- tion to account for the spatial correlation; and εlm the error for each measurement. We then selected the best model for each clone according to lower Akaike Informa- tion Criteria (AIC), analysis of the residuals, and biological interpretation of the predic- tors’ effects. All analyses were carried out using the package “lme4” (Bates et al. 2015) in the R statistical environment (R Core Team 2022). As previously mentioned, irrigated sites were excluded from the analysis. Once the models were selected, we carried out a sensitivity analysis of all predictors. We simulated AGBp as a func- tion of only one predictor, which we let to vary from the minimum to the maximum value observed in the data used for cali- brating the models, while the values of the other variables were kept equal to the ob- served mean. 191 iForest 16: 188-194 Tab. 3 - Results of clone-specific linear mixed models. (Sig.): significance of the p-values according to the standard notation. (MAE): mean absolute error; (RMSE): root mean square error. R2 is the variance explained by the model, either considering only the fixed effects (R2 marginal) or both fixed and random effects (R2 conditional). Location No. of samples Coefficient Estimate p-value Sig. MAE RMSE R2 marginal R2 conditional AF18 21 (Intercept) 7.10847 0.00034 *** 2.094 2.772 0.00 0.26 AF34 20 (Intercept) 0.10698 0.94571 - 2.194 2.778 0.41 0.41 MSP 0.01970 0.00176 ** 2.194 2.778 0.41 0.41 AF8 30 (Intercept) 0.81599 0.66112 - 1.354 1.743 0.34 0.63 MSP 0.01859 0.02315 * 1.354 1.743 0.34 0.63 Antonije 19 (Intercept) 3.91550 0.01111 * 1.006 1.243 0.00 0.75 Baldo 28 (Intercept) -3.05146 0.20270 - 1.994 2.597 0.51 0.66 MSP 0.03353 0.00546 ** 1.994 2.597 0.51 0.66 Brenta 26 (Intercept) 9.56719 0.00670 ** 1.225 1.532 0.26 0.64 AHM -0.12388 0.06367 . 1.225 1.532 0.26 0.64 Dellinois 12 (Intercept) 19.59646 0.00026 *** 1.903 2.697 0.51 0.51 AHM -0.37528 0.00716 ** 1.903 2.697 0.51 0.51 Delrive 13 (Intercept) 7.95905 0.01401 * 0.819 1.057 0.00 0.91 25 (Intercept) -3.45854 0.01018 * 1.689 2.276 0.73 0.73 25 MSP 0.03539 0.00000 *** 1.689 2.276 0.73 0.73 Grimminge 23 (Intercept) 25.82406 0.00150 ** 1.184 1.527 0.71 0.74 23 AHM -0.95448 0.00827 ** 1.184 1.527 0.71 0.74 Hybride275 23 I(AHM^2) 0.00947 0.02064 * 1.184 1.527 0.71 0.74 24 (Intercept) 42.79374 0.00294 ** 1.176 1.712 0.80 0.89 24 AHM -1.50734 0.01828 * 1.176 1.712 0.80 0.89 Koreana 24 I(AHM^2) 0.01369 0.05519 . 1.176 1.712 0.80 0.89 21 (Intercept) 36.46674 0.00358 ** 1.123 1.380 0.84 0.89 21 AHM -1.31474 0.02028 * 1.123 1.380 0.84 0.89 Matrix21 21 I(AHM^2) 0.01251 0.05660 . 1.123 1.380 0.84 0.89 25 (Intercept) 5.12385 0.00035 *** 1.100 1.497 0.00 0.63 24 (Intercept) 3.81112 0.00819 ** 0.874 0.975 0.00 0.86 Max 1-4 15 (Intercept) 18.19725 0.00005 *** 1.791 2.331 0.57 0.57 Muur 15 AHM -0.30283 0.00092 *** 1.791 2.331 0.57 0.57 Orion 21 (Intercept) 5.24868 0.00119 ** 2.116 2.617 0.00 0.26 Oudenberg 19 (Intercept) -0.40648 0.84369 - 1.053 1.654 0.49 0.68 SV885 19 MSP 0.02029 0.03270 * 1.053 1.654 0.49 0.68 18 (Intercept) 4.89019 0.00035 *** 1.238 1.632 0.00 0.21 SW 822 21 (Intercept) 19.27073 0.00566 ** 3.598 4.742 0.30 0.30 Skado 21 AHM -0.27800 0.02690 * 3.598 4.742 0.30 0.30 23 (Intercept) 13.18552 0.00210 ** 1.328 1.701 0.48 0.71 Vesten 23 AHM -0.19773 0.01444 * 1.328 1.701 0.48 0.71 21 (Intercept) 7.10847 0.00034 *** 2.094 2.772 0.00 0.26 iF or es t – B io ge os ci en ce s an d Fo re st ry AGBp=b0+b1 AHM +b2 AHM 2 +b3MSP+b4MSP 2+b5CONT +b6CONT 2+b7 AHM⋅MSP +b8 AHM⋅CONT +b9CONT⋅MSP+ul+ε lm https://soilgrids.org/ Climate-wise models for hybrid poplar clones’ biomass productivity in Europe Results Significant climate-growth relationships were found in thirteen out of twenty AGBp models (Tab. 3). For eight models, AGBp was significantly and negatively correlated with AHM using a linear term, and in three of those cases, the negative trend tended to stabilize to a minimum towards high val- ues of AHM, due to a significant positive quadratic term. For five other clones, AGBp was significantly and negatively correlated with MSP using a linear term. For the re- maining seven clones, no fixed effect was found significant. CONT was significant in only three models, but it was removed from the analysis since its inclusion pushed the coefficients for AHM and MSP to ex- treme illogical values, indicating an overfit- ting with no biological meaning. The clone-specific models showed a good fitting to the observed values of productiv- ity for most clones (Fig. 2). These was con- firmed by the conditional R2 values ob- tained, which ranged from 0.21 to 0.91 and averaged to 0.64 (Tab. 3). However, it is worth noting that only in thirteen cases the good fit of the clone-specific model was maily due to the fixed effects (AHM or MSP), while in the other cases fitting was largely explained by the random effects. For the models showing significant fixed effects, the marginal R2 values (considering only the fixed terms) still showed satisfac- tory results, ranging from 0.26 to 0.84, on average 0.56. Also, it should be stressed that the conditional values for the inter- cept only models were on average 0.68, only due to random effects. Model residu- iForest 16: 188-194 192 Fig. 3 - Residuals of the above ground biomass productivity (Mg year -1 ha-1) vs. pre- dicted values for all clone-specifics models. Fig. 4 - Lines: above ground biomass productivity (AGBp, Mg year-1 ha-1) predicted as function of Annual Heat Moisture index (AHM) for all clones. Points: observed values of AGBp across different sites. Irrigated locations (green points, not used in modeling) are also shown for comparison. Fig. 5 - Lines: above ground biomass productivity (AGBp, Mg year-1 ha-1) predicted as function of Mean Summer Precipitation (MSP) for all clones. Points: observed values of AGBp across different. Irrigated locations (green points, not used in model- ling) are also shown for comparison. iF or es t – B io ge os ci en ce s an d Fo re st ry Bianchi S et al. - iForest 16: 188-194 als did not reveal any bias against the pre- dictions, although in very few cases there was a slight tendency towards a larger vari- ance for higher predicted values (Fig. 3). The sensitivity analysis shows how the poplar clones reacted to either AHM (Fig. 4) or MSP (Fig. 5). Very low productivity values were found in the AHM range 40-50 °C mm-1, where the minimum productivity was reached for the models with a qua- dratic term. For most clones planted at irri- gated sites, the productivity was generally higher and not influenced by AHM, except for seven clones (Delrive, Matrix21, Hy- bride275, SV885, SW822, Skado and Vesten). The above evidence indicates that over a AHM threshold value of about 40-50 °C mm-1, irrigation seems to be essential to ensure a good productivity of poplar SRC for most clones. Below that threshold, the best performing clones were Dellinois, Ko- reana, Matrix21, Orion, and Skado. On the contrary, the best clones performing with irrigation were AF34, AF8, Baldo and Brenta. In the model using MSP as fixed effect, AGBp increased linearly until reach- ing around 10 Mg ha-1 year-1 for values of mean summer precipitation higher than 400 mm. Discussion Our study highlights the need of large re- gional studies for modeling poplar SRC pro- ductivity, which can be used to fit climate- wise models. Currently, there is a lack of such models in the literature on this topic. Understanding the clone-specific relation- ships between climate and poplar SRC pro- ductivity can help in improving both the profitability and sustainability of renewable biomass production, for example by reduc- ing the needs for irrigation or fertilization (Oliveira et al. 2020, Ghezehei et al. 2021, Xi et al. 2021). This is especially important given both the current concerns in water uptake at European level (Verlinden et al. 2015) and the climate change impact (Liu et al. 2022). The general level of productivity in this study (from 1 to 26 Mg ha -1 year-1) was similar to that of various European countries and clones reported in Oliveira et al. (2020), which also included some of the clones tested in this work. The selection criteria used for commercial poplar geno- types should take into account drought re- sistance together with biomass productiv- ity (Navarro et al. 2014). Our research ex- panded the findings of Bergante et al. (2010) by including sites from several coun- tries in Europe, and those of Fischer et al. (2014) and Navarro et al. (2014) by provid- ing links between clone productivity and water availability estimated by climatic datasets. Most of the clones (thirteen out twenty) were significantly affected by either AHM or MSP, similarly to the findings reported by Bergante et al. (2010). This is in line with the species’ ecology and its high require- ments of water availability (Dickmann 2001). Njakou Djomo et al. (2015) did not find a significant relationship between clone productivity and water availability, likely due to the lack of inclusion of spe- cies-specific and/or site management ef- fects. A quadratic term was used also in Marchi et al. (2022), but the inclusion of MSP in the model resulted in different (positive) relations for AHM. Also, conti- nentality did not enter any model in this study, in contrast to Marchi et al. (2022) where the same predictor did significantly affect poplar productivity. For the thirteen clones were either AHM or MSP was significant in this study, the R2 was 0.56 on average, indicating an ade- quate fit. For the rest of clones, only the random effect due to the plot location in- fluenced the model, so there was no possi- ble biological interpretation. The high val- ues of R2 even in many of those cases (on average 0.68) suggest the presence of en- vironmental factors which drastically con- trast among different sites. Indeed, these environmental factors could explain much of the variation, though were not included in the models. Some clones, namely Dellinois, Koreana, Matrix21, Orion, and Skado, showed the best performances below a AHM threshold of 40-50 °C mm-1, i.e., in areas with higher water availability. Thus, they are potential candidates for poplar SRC to be estab- lished in sites where water availability is not a constraint. On the contrary, except for the clone Orion, they had amongst the lowest productivity values at high level of AHM, i.e., in the driest conditions, and a weak positive response to irrigation. Over AHM 40-50 °C mm-1, all clones performed very poorly without irrigation. Marchi et al. (2022) in a similar study on hybrid poplar in Italy did not investigated sites with condi- tions drier than AHM 40 °C mm-1, so com- parisons are not possible. Regarding MSP, we found a linear increase in productivity in both studies, although slightly exponen- tial in Marchi et al. (2022) for MSP values until 400-500 mm, the highest in both datasets. As expected, irrigation generally improved productivity. Oliveira et al. (2020) reported that the highest levels of produc- tivity are shown in irrigated Mediterranean sites, as in this study. In our models, when irrigation was included, we did not find any significant climate-relationship. Similarly, Marchi et al. (2022) found out that irriga- tion explained most of the variance, fol- lowed by AHM. In this study, ee decided to discard the irrigated sites from the analysis to better study the climate-productivity re- lationships and find the clones less suscep- tible to drought. However, some clones (AF34, AF8, Baldo and Brenta) seemed to have the best performance when irrigation was carried out, regardless of the site arid- ity. Soil properties were excluded from mod- elling due to their low or inconclusive con- tribution to explain the model variance, while they are considered crucial for pro- ductivity (Dickmann 2001). All plots were established in agricultural land where the managers expected a good productivity for poplar. Either there was a low variation in the overall soil potential across sites (all equally favourable for good productivity), or we could not find reliable data at such level of details for all sites. Marchi et al. (2022) also highlighted the lack of high- quality, high-resolution soil data layers to be used for range-wide modelling pur- poses. Soil texture and organic matter con- tent may interact with plant growth due to their influence on soil structure and hydro- logical characteristics (Bergante et al. 2010), and may be included in further mod- elling experiments. It is worth to highlight that few replicates for each clone were analyzed in this study, especially when data from irrigated sites were discarded (from seven to eleven loca- tions). Moreover, the plot size was rela- tively small (minimum 120 m2), and it has been found that using small-scale planta- tions tend to overestimate biomass pro- duction values (Verlinden et al. 2015). Fur- ther, some of the models included a qua- dratic term for AHM that was crucial in smoothing the climate-growth relation- ship, although it can have an unplausible positive effect at AHM values higher than those included in our dataset. For the above reasons, we recommend caution in extrapolating the results of our models outside the specific range of the study. Fi- nally, our study stresses the need for larger network of trials, especially in the light of the uncertainty of climate change, as also reported by Marchi et al. (2022). Conclusions We presented climate-wise model for hy- brid poplar productivity using a dataset col- lected at European scale. Our results show water availability, expressed by climatic variables linked to precipitation and aridity, increased poplar productivity in short rota- tion coppice with a quantitative modelling approach. However, some limitations of the study, such as soil variables not in- cluded in the models, and the few repli- cates in the dataset, may not allow extrap- olating the results of the models presented here outside the specific range of the study. Nonetheless, we demonstrated that that using a larger dataset in further study, a more comprehensive model for poplar productivity in short rotation coppice can be developed as function of climatic vari- ables. Acknowledgements The data were collected within the project “Testing of poplar clones from EU member states for the use in short rotation coppice”. We wish to thank for this study: Linda Meiresonne, Marijke Steenackers (INBO, Belgium); Albena Bobeva (Execu- tive Forest Agency, Bulgaria); Davorin Ka- jba (University of Zagreb, Croatia); Jan Weger (Silva Tarouca, Czech Republic); An- drea Barban, Francesco Fabbrini (Alasia 193 iForest 16: 188-194 iF or es t – B io ge os ci en ce s an d Fo re st ry Climate-wise models for hybrid poplar clones’ biomass productivity in Europe Franco Vivai, Italy); Enrico Alasia, Fabiano Bertaina, Valerio Crescenzo (Biopoplar, Italy); Giuseppe Pignatti, Stefano Verani (CREA - Research Centre for Forestry and Wood, Italy); Randolf Schirmer (AWG Teisendorf, Germany); Boglárka Nemethne, Sándor Bordács (National Food Chain Safety Office, Hungary); Attila Benke (Sav- ar, Hungary); Catherine Bastien (INRAE, France); Hortensia Sixto (CIFOR-INIA, Spain); and Georgeta Mihai (Forest Re- search & Management Institute, Romania). This research was funded by the Acad- emy of Finland under the UNITE (Forest- Human-Machine Interplay) flagship ecosys- tem (Grant no. 337655) and by the Euro- pean Union Horizon 2020 Programme un- der the B4EST (“Adaptive breeding for pro- ductive, sustainable and resilient forests under climate change”) project (Grant no. 773383). References Bates D, Mächler M, Bolker B, Walker S (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67: 1-48. - doi: 10.18637/jss.v067.i01 Batjes NH, Ribeiro E, Van Oostrum A (2020). Standardised soil profile data to support global mapping and modelling (WoSIS snapshot 2019). Earth System Science Data 12: 299-320. - doi: 10.5194/essd-12-299-2020 Bentsen NS, Felby C (2012). Biomass for energy in the European Union - A review of bioenergy resource assessments. Biotechnology for Biofu- els 5: 1-10. - doi: 10.1186/1754-6834-5-25 Bergante S, Facciotto G, Minotta G (2010). Iden- tification of the main site factors and manage- ment intensity affecting the establishment of Short Rotation Coppices (SRC) in Northern Italy through stepwise regression analysis. Open Life Sciences 5: 522-530. - doi: 10.2478/s11535- 010-0028-y Dickmann D (2001). Poplar culture in north America. NRC Research Press, Ottawa, Canada, pp. 397. [online] URL: http://books.google. com/books?id=GRtdQYO_ejUC Dillen SY, Djomo SN, Afas NA, Vanbeveren S, Ceulemans R (2013). Biomass yield and energy balance of a short-rotation poplar coppice with multiple clones on degraded land during 16 years. Biomass and Bioenergy 56: 157-165. - doi: 10.1016/j.biombioe.2013.04.019 EC/JRC (2019). Brief on biomass for energy in the European Union. European Commission EC and Joint Research Centre - JRC, Publications Of- fice, Luxembourg, pp. 8. - doi: 10.2760/49052 Fernández MJ, Barro R, Pérez J, Losada J, Ciria P (2016). Influence of the agricultural manage- ment practices on the yield and quality of pop- lar biomass (a 9-year study). Biomass and Bio- energy 93: 87-96. - doi: 10.1016/j.biombioe.2016. 06.027 Fischer M, Trnka M, Kučera J, Fajman M, Zalud Z (2014). Biomass productivity and water use re- lation in short rotation poplar coppice (Populus nigra × P. maximowiczii) in the conditions of Czech Moravian Highlands. Acta Universitatis Agriculturae et Silviculturae Mendelianae Brun- ensis 59: 141-152. - doi: 10.11118/actaun20115906 0141 Ghezehei SB, Ewald AL, Hazel DW, Zalesny RS, Nichols EG (2021). Productivity and profitability of poplars on fertile and marginal sandy soils under different density and fertilization treat- ments. Forests 12: 869. - doi: 10.3390/f12070869 Harris I, Osborn TJ, Jones P, Lister D (2020). Ver- sion 4 of the CRU TS monthly high-resolution gridded multivariate climate dataset. Scientific Data 7: 1-18. - doi: 10.1038/s41597-020-0453-3 Landgraf D, Carl C, Neupert M (2020). Biomass yield of 37 different SRC poplar varieties grown on a typical site in North Eastern Germany. Forests 11: 1048. - doi: 10.3390/f11101048 Liu J, Li D, Fernández J-E, Coleman M, Hu W, Di N, Zou S, Liu Y, Xi B, Clothier B (2022). Varia- tions in water-balance components and carbon stocks in poplar plantations with differing wa- ter inputs over a whole rotation: implications for sustainable forest management under cli- mate change. Agricultural and Forest Meteorol- ogy 320: 108958. - doi: 10.1016/j.agrformet.2022. 108958 Manzone M, Bergante S, Facciotto G (2014). En- ergy and economic evaluation of a poplar plan- tation for woodchips production in Italy. Bio- mass and Bioenergy 60: 164-170. - doi: 10.1016/j. biombioe.2013.11.012 Marchi M, Bergante S, Ray D, Barbetti R, Fac- ciotto G, Chiarabaglio PM, Hynynen J, Nervo G (2022). Universal reaction norms for the sus- tainable cultivation of hybrid poplar clones un- der climate change in Italy. iForest - Biogeosci- ences and Forestry 15: 47-55. - doi: 10.3832/ifor 3989-015 Navarro A, Facciotto G, Campi P, Mastrorilli M (2014). Physiological adaptations of five poplar genotypes grown under SRC in the semi-arid Mediterranean environment. Trees 28: 983- 994. - doi: 10.1007/s00468-014-1012-3 Nielsen UB, Madsen P, Hansen JK, Nord-Larsen T, Nielsen AT (2014). Production potential of 36 poplar clones grown at medium length rotation in Denmark. Biomass and Bioenergy 64: 99-109. - doi: 10.1016/j.biombioe.2014.03.030 Niemczyk M, Kaliszewski A, Jewiarz M, Wróbel M, Mudryk K (2018). Productivity and biomass characteristics of selected poplar (Populus spp.) cultivars under the climatic conditions of northern Poland. Biomass and Bioenergy 111: 46-51. - doi: 10.1016/j.biombioe.2018.02.002 Njakou Djomo S, Ac A, Zenone T, De Groote T, Bergante S, Facciotto G, Sixto H, Ciria Ciria P, Weger J, Ceulemans R (2015). Energy perfor- mances of intensive and extensive short rota- tion cropping systems for woody biomass pro- duction in the EU. Renewable and Sustainable Energy Reviews 41 (3): 845-854. - doi: 10.1016/j. rser.2014.08.058 Oliveira N, Pérez-Cruzado C, Cañellas I, Rodrí- guez-Soalleiro R, Sixto H (2020). Poplar short rotation coppice plantations under Mediter- ranean conditions: the case of Spain. Forests 11: 1352. - doi: 10.3390/f11121352 R Core Team (2022). R: a language and environ- ment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Re- trieved from. [online] URL: http://www.R-proj ect.org/ Rodrigues A, Gonçalves AB, Casquilho M, Gomes AA (2020). A GIS-based evaluation of the po- tential of woody short rotation coppice (SRC) in Portugal aiming at co-firing and decentral- ized co-generation. Biomass and Bioenergy 137: 105554. - doi: 10.1016/j.biombioe.2020.105554 Verlinden MS, Broeckx LS, Ceulemans R (2015). First vs. second rotation of a poplar short rota- tion coppice: above-ground biomass productiv- ity and shoot dynamics. Biomass and Bioenergy 73: 174-185. - doi: 10.1016/j.biombioe.2014.12.012 Vusić D, Kajba D, Andrić I, Gavran I, Tomić T, Plišo Vusić I, Zečić Z (2019). Biomass yield and fuel properties of different poplar SRC clones. Croa- tian Journal of Forest Engineering 40: 231-238. - doi: 10.5552/crojfe.2019.678 Werner C, Haas E, Grote R, Gauder M, Graeff- Hönninger S, Claupein W, Butterbach-Bahl K (2012). Biomass production potential from Pop- ulus short rotation systems in Romania. GCB Bioenergy 4: 642-653. - doi: 10.1111/j.1757-1707. 2012.01180.x Xi B, Clothier B, Coleman M, Duan J, Hu W, Li D, Di N, Liu Y, Fu J, Li J, Jia L, Fernández J-E (2021). Irrigation management in poplar (Populus spp.) plantations: a review. Forest Ecology and Man- agement. 494: 119330. - doi: 10.1016/j.foreco. 2021.119330 iForest 16: 188-194 194 iF or es t – B io ge os ci en ce s an d Fo re st ry https://doi.org/10.18637/jss.v067.i01 https://doi.org/10.1016/j.foreco.2021.119330 https://doi.org/10.1016/j.foreco.2021.119330 https://doi.org/10.1111/j.1757-1707.2012.01180.x https://doi.org/10.1111/j.1757-1707.2012.01180.x https://doi.org/10.5552/crojfe.2019.678 https://doi.org/10.1016/j.biombioe.2014.12.012 https://doi.org/10.1016/j.biombioe.2020.105554 http://www.R-project.org/ http://www.R-project.org/ https://doi.org/10.3390/f11121352 https://doi.org/10.1016/j.rser.2014.08.058 https://doi.org/10.1016/j.rser.2014.08.058 https://doi.org/10.1016/j.biombioe.2018.02.002 https://doi.org/10.1016/j.biombioe.2014.03.030 https://doi.org/10.1007/s00468-014-1012-3 https://doi.org/10.3832/ifor3989-015 https://doi.org/10.3832/ifor3989-015 https://doi.org/10.1016/j.biombioe.2013.11.012 https://doi.org/10.1016/j.biombioe.2013.11.012 https://doi.org/10.1016/j.agrformet.2022.108958 https://doi.org/10.1016/j.agrformet.2022.108958 https://doi.org/10.3390/f11101048 https://doi.org/10.1038/s41597-020-0453-3 https://doi.org/10.3390/f12070869 https://doi.org/10.11118/actaun201159060141 https://doi.org/10.11118/actaun201159060141 https://doi.org/10.1016/j.biombioe.2016.06.027 https://doi.org/10.1016/j.biombioe.2016.06.027 https://doi.org/10.2760/49052 https://doi.org/10.1016/j.biombioe.2013.04.019 http://books.google.com/books?id=GRtdQYO_ejUC http://books.google.com/books?id=GRtdQYO_ejUC https://doi.org/10.2478/s11535-010-0028-y https://doi.org/10.2478/s11535-010-0028-y https://doi.org/10.1186/1754-6834-5-25 https://doi.org/10.5194/essd-12-299-2020 Bianchi et al 2023.pdf iForest_ifor4211-016_Bianchi Climate-wise models of biomass productivity for hybrid poplar clones in Europe Introduction Material and methods Study sites and clones Experimental design Data analysis Results Discussion Conclusions Acknowledgements References