AMQ abs Vallè et al PALCOM 165-168.pub Available online http://amq.aiqua.it ISSN (online): 2279-7335 Alpine and Mediterranean Quaternary, Vol. 31 (Quaternary: Past, Present, Future - AIQUA Conference, Florence, 13-14/06/2018), 165 - 168 TESTING POLLEN-CLIMATE MODELS OVER THE LAST 200 YEARS IN N-ITALY USING INSTRUMENTAL DATA Francesca Vallè 1, Michele Brunetti 2, Roberta Pini 3, Valter Maggi 1, Cesare Ravazzi 3 1 Dept. of Environmental and Earth Sciences, University of Milano - Bicocca, Milano, Italy 2 Institute of Atmospheric Sciences and Climate, CNR, Bologna, Italy 3 IDPA - CNR, Research Group on Vegetation, Climate and Human Stratigraphy, Laboratory of Palynology and Palaeoecology, Milano, Italy Corresponding author: F. Vallè ABSTRACT: We evaluated pollen-climate models developed for past temperature reconstruction in northern Italy by comparing pollen- inferred temperature series with site-specific instrumental series in two records covering the last 200 years and provided with a pollen sample resolution of 5/9 years. Moderate to good correlation is found. We discuss the methodological approach and the possible influ- ence of human activities in the pollen-inferred reconstructions. KEYWORDS: North Italy, pollen data, instrumental temperature, pollen-climate models, quantitative summer temperature, last 200 years 1. INTRODUCTION During the last 50 years, different procedures were developed to quantita- tively estimate past environmental and climate variables from stratigraphical mi- crofossil assemblages, including pollen data (Brewer et al., 2007, Juggins & Birks, 2012). The validation of pollen-climate models and the climate reconstructions obtained from palynological data for inter- vals older than the instrumental data are only possible using statistical tests and procedures (e.g. Juggins & Birks, 2012 and reference therein). So far, only one paper presented climate reconstructions based on pollen spectra and their valida- tion against instrumental data (St. Jacques et al., 2015). No attempts have been done in Italy. In this paper we present the first re- sults concerning the development and testing of pollen-climate models to recon- struct summer temperatures over the last 200 years in northern Italy. We use two previously published high- resolution pollen records, discuss methodological issues and validate the pollen-inferred reconstructions based on the direct comparison with instrumental data ob- tained for the same sites. 2. MATERIALS AND METHODS 2.1 Study sites and pollen data The two study sites are located at different altitudes and are characterized by different mean climatological conditions (Fig. 1). Lago Grande di Avigliana (353 m a.s.l.) lies at the foothills of the western Alps: mean July temperature (reference period 1961-1990) is around 22° C. Lago di Lavarone (1115 m a.s.l.) is located in the pre- alpine region: mean July temperature is around 16°C. Sediment cores covering part of the Late Glacial and the whole Holocene were extracted from both lakes and studied for their pollen content to reconstruct the vegeta- tion history, also including land-use changes in the two pollen source areas. Pollen records covering the last 200 years from Lago Grande di Avigliana and Lago di Lavarone are presented respectively in Finsinger et al. (2006) and in Arpenti & Filippi (2007). The sequence from Lago Grande di Avigliana has a good chronological control, based on varves counting, 210Pb and 137Cs dates: mean temporal resolution of the pollen record is Fig. 1 - Map showing the location of the two study sites on T july climatology (1961-1990), adapted from Brunetti et al. (2014). LGA: Lago Grande di Avigliana (353 m a.s.l.) at the foothills of the western Alps and LAV: Lago di Lavarone (1115 m a.s.l.) in the eastern side of the pre-alpine region. https://doi.org/10.26382/AIQUA.2018.AIQUAconference ~4 years for the last 200 years. A good age control is available also for the Lago di Lavarone sequence: its pollen record has a mean sample resolution of ~9 years for the last 200 years with higher temporal resolution after ~1900 AD. 2.2 Site-specific temperature series for the last 200 years To overcome the lack of representativeness at the local scale of the commonly-used global climatological datasets, monthly instrumental temperature series have been reconstructed for Lago Grande di Avigliana (1833- 2010) and Lago di Lavarone (1807-2010). Series were obtained by means of the anomaly method (New et al., 2000; Mitchell & Jones, 2005), as described in Brunetti et al. (2012), and benefit from the availability of several long instrumental temporal series for Northern Italy (Brunetti et al., 2006). Specifically, the procedure con- sists in the independent reconstruction of the clima- tological normals over a given reference period (i.e. the climatologies) and the anomalies (i.e. the departures from the climatologies): the former, characterized by remarkable spatial gradients, require a high spatial den- sity of stations (even if with a limited temporal coverage) and a sophisticated interpolation procedure (Brunetti et al., 2014; Crespi et al., 2018) to be reconstructed. Anomalies are more coherent being linked to climate variability and, therefore, a low spatial density is enough but long temporal coverage and accurate homogeniza- tion (i.e. the procedure that removes the non-climatic signals introduced by stations and instruments reloca- tion, changes in measurement practices and so on) is mandatory. Finally, absolute values of temperature monthly series for each site can be obtained by the su- perimposition of the two fields. The summer tempera- ture series have been calculated by averaging June, July and August series. In order to highlight the tem- perature trends and to facilitate the comparison and the correlation between the instrumental records and the pollen-inferred reconstructions, the instrumental sum- mer temperature series were smoothed with loess re- gression (for LGA span=0.05, for LAV span=0.1) follow- ing the pollen resolution. PAST statistical software ver- sion 3.3 (Hammer et al., 2001) was used. 2.3 Pollen-based quantitative temperature recon- structions A three-steps approach (Juggins and Birks, 2012) was applied to obtain quantitative estimates of summer temperature. As a first step we developed an Italian modern pollen-climate calibration set that could cover the summer temperature gradients displayed during the last two centuries. 265 modern pollen samples have been selected from the European Modern Pollen Data- base (EMPD, Davis et al., 2013); each sampling loca- tion was associated to site-specific instrumental tem- perature data, reconstructed for the period 1951-2000 following the same procedure described above. Pollen taxonomy has been harmonized, both in the calibration set samples and in the fossil records. The specific train- ing set is the base for the second step, i.e. modeling the modern pollen-temperature relationship and create the 166 Fig. 2 - a) Comparison between instrumental T summer series (grey curves) and the pollen-inferred T summer reconstructions (red dotted curves) obtained for LGA and LAV with the WA method. The instrumental summer T series were smoothed with loess regression (for LGA span=0.05, for LAV span=0.1) according to the pollen resolution (black curves). Coefficient of determination between the two series are also shown. b) Percentages of two anthropogenic pollen taxa, Cannabaceae from LGA and Ostrya from LAV, which might influence the pollen-inferred temperature reconstructions. Vallè F. et al 167 transfer functions by means of two numerical methods: the weighted-averaging (WA) regression and calibration (Ter Braak & Barendregt, 1986; Juggins and Birks, 2012 and reference therein) and the weighted-averaging par- tial least square regression and calibration (WAPLS) (Ter Braak & Juggins, 1993). We finally applied the transfer functions to fossil pollen records to estimate the summer temperature values for the last 200 years. Be- fore applying the transfer functions to the fossil records the statistical performances of the developed models were checked in cross-validation. The goodness of the calibration set in terms of availability of modern analogues for each fossil pollen sample was also evaluated using the Modern Analogue Technique (MAT) and its diagnostics. The “real” valida- tion of the pollen-inferred reconstructions has been ob- tained by comparing the pollen-based temperature se- ries with the instrumental data. The development of WA and WAPLS models, the evaluation of their perform- ances, the temperature reconstructions and the MAT diagnostics were carried out using R software (R Core Team, 2017) and the rioja, R package (Juggins, 2017). 3. RESULTS Pollen-inferred summer temperature reconstruc- tions obtained for Lago Grande di Avigliana and Lago di Lavarone are shown together with the instrumental se- ries (Fig. 2). WA pollen-inferred summer temperature values obtained for LGA are almost constantly underes- timated if compared to the instrumental series. They are instead slightly overestimated for LAV but, considering the sample specific errors (not shown in Fig. 2), they include the instrumental values. Despite the shift in the absolute values at LGA, the decadal variability and the long-term trend in summer temperature instrumental series are also detectable in the pollen-inferred recon- structions, and this is true for both sites. Interestingly, the LGA pollen-inferred summer T reconstruction recog- nizes the warmer interval recorded between 1940 and 1950 AD. The direct comparison of the WA pollen-inferred summer temperatures with the instrumental series yielded a moderate correlation for LGA (R2=0.41) and a good correlation for LAV (R2=0.7). 4. DISCUSSION AND CONCLUSION During the last two centuries mean air temperature in Italy shows an increasing trend of 1°C per century with an intensification occurred in the last decades (Brunetti et al. 2006, 2013). Beside this century-long temperature increase, negative trends over shorter peri- ods have been observed at the beginning of the 19th century and between 1950-1970 AD (Brunetti et al. 2006, 2013). Pollen-inferred summer temperature re- constructions capture the general warming of the last centuries. Fig. 2 also presents the pollen percentages of two important pollen-productors which were affected by human land use changes, i.e. Cannabaceae from the pollen record of Lago Grande di Avigliana (Finsinger et al., 2006) and Ostrya from the pollen record of Lago di Lavarone. Cannabis sativa was cultivated in northern Piedmont in the 19th century until ~1950 AD (Finsinger et al. 2006 and reference therein). Whether this taxon is included or not in the pollen sum used for the tempera- ture reconstructions, it influences the reconstructed absolute values. This issue needs to be carefully con- sidered because this pollen taxon is not well repre- sented in the used calibration set. Concerning Ostrya, this species is today the main component of forest vege- tation in the pollen source area of Lago di Lavarone. Its expansion started around 1965 AD, favoured by de- creased human activities in the area (Arpenti and Filippi, 2005). We are aware that WA methods are able to solve the “non-analogues” problems but suffer from the “edge effects” (Juggins & Birks, 2012 and reference therein). Being Lago Grande di Avigliana summer temperature values at the end of the calibration set summer tem- perature gradient, the constant underestimation of the LGA pollen-inferred summer temperature might be due to this reason. This rigid shift in absolute values (which does not affect the long-term variability) is reduced by using WAPLS model (not shown in Fig. 2). Our results show that pollen records obtained from lacustrine sediment cores can be used to obtain pollen- inferred reconstructions with trends comparable to in- strumental data over the last 200 years, provided that: 1) pollen records have a robust age control and high sample resolution (ideally less than a decade), 2) proper calibration sets and adequate numerical techniques are used to develop pollen-climate models and transfer functions to be applied to fossil pollen records. The hu- man activities and their effects in the pollen relative abundances must be considered and evaluated in the process of taxa selection before applying quantitative methods. A further perspective after the calibration of pollen- based reconstructions with instrumental data will involve the comparison between the pollen-inferred summer temperature reconstructions and the dendrochronologi- cal-based summer temperature reconstructions for com- mon temporal intervals in the alpine region. If the cli- mate signals from both proxies are comparable, those proxies might be combined to build an integrate multi- proxy temperature reconstruction for northern Italy. ACKNOWLEDGEMENTS This research is funded by the CNR-DTA Project of Interest NextData, a national system for the retrieval, storage, access and diffusion of environmental and cli- mate data from mountain and marine areas (www.nextdataproject.it). Pollen data for Lago Grande di Avigliana were extracted from the European Pollen Da- tabase (EPD, http://www.europeanpollendatabase.net/). The work by data contributors and the EPD community is gratefully acknowledged. REFERENCES Arpenti E., Filippi M.L. (2007) - Evoluzione della vegeta- zione nei pressi del Lago di Lavarone (TN) negli ultimi 2200 anni. Studi Trent. Sci. Nat., Acta Geol., 82, 317-324. Pollen-climate models over the last 200 years in N-Italy Brewer S., Guiot J., Barboni D. (2007) - Pollen data as climate proxies. In: Elias SA (ed) Encyclopedia of quaternary science, (4). Elsevier, New York, 2498- 2510. Brunetti M., Maugeri M., Monti F., Nanni T. (2006) - Temperature and precipitation variability in Italy in the last two centuries from homogenized instru- mental time series. International Journal of Clima- tology, 26, 345-381. Brunetti M., Lentini G., Maugeri M., Nanni T., Simolo C., Spinoni J. (2012) - Projecting North Eastern Italy temperature and precipitation secular records onto a high resolution grid. Physics and Chemistry of the Earth, 40-41, 9-22. Brunetti M., Maugeri M., Nanni T., Simolo C. (2013) - Variazioni climatiche in Italia negli ultimi due secoli. In: Il mutamento climatico. Processi naturali e in- tervento umano. Bologna, Ed. Il Mulino, pp. 357. Brunetti M., Maugeri M., Nanni T., Simolo C., Spinoni J. (2014) - High-resolution temperature climatology for Italy: interpolation method intercomparison. International Journal of Climatology, 34, 1278- 1296. Crespi A., Brunetti M., Lentini G., Maugeri M. (2018) - 1961-1990 high-resolution monthly precipitation climatologies for Italy. International Journal of Cli- matology, 38, 878-895, Doi: 10.1002/joc.5217. Davis B.A.S., Zanon M., Collins M., Mauri A., Bakker J., Barboni D., Barthelmes A., Beaudouin C., Birks H.J.B., Bjune A.E. et al. (2013) - The European Modern Pollen Database (EMPD) Project. Vegeta- tion History and Archaeobotany, 22, 521-530. Finsinger, W., Bigler C., Krähenbühl U., Lotter AF., Am- mann B. (2006) - Human impacts and eutrophica- tion patterns during the past ~200 years at Lago Grande di Avigliana (N. Italy). Journal of Paleolim- nology, 36, 55-67. Hammer, Ø., Harper, D.A.T., Ryan, P.D. (2001) - PAST: Paleontological statistics software package for education and data analysis. Palaeontologia Elec- tronica 4(1), 9. Juggins S., Birks H.J.B. (2012) - Quantitative environ- mental reconstructions from biological data. In (Birks et al., Eds.): Tracking environmental change using lake sediments. Data handling and numerical techniques. Chapter 14, 431-494. Springer. Juggins S. (2017) - rioja: Analysis of Quaternary Sci- ence Data, R package version (0.9-15.1) (http:// cran.r-project.org/package=rioja). Mitchell T.D., Jones P.D. (2005) - An improved method of constructing a database of monthly climate ob- servations and associated high-resolution grids. International Journal of Climatology, 25(6), 693- 712. New M., Hulme M., Jones P. (2000) - Representing twentieth-century space-time climate variability. Part II: Development of 1901–96 monthly grids of terrestrial surface climate. Journal of Climate, 13, 2217-2238. R Core Team (2017) - R: A language and environment for statistical computing. R Foundation for Statisti- cal Computing, Vienna, Austria. URL https:// www.R-project.org/. St. Jacques J.M., Cumming B.F., Sauchyn D.J., Smol J.P. (2015) - The Bias and Signal Attenuation Pre- sent in Conventional Pollen-Based Climate Recon- structions as Assessed by Early Climate Data from Minnesota, USA. PLoS ONE 10(1), e0113806. Doi:10.1371/journal.pone.0113806. Ter Braak, C.J.F., Barendregt, L.G. (1986) - Weighted averaging of species indicator values: its efficiency in environmental calibration. Mathematical Biosci- ences, 78, 57-72. Ter Braak C.J.F., Juggins S. (1993) - Weighted averag- ing partial least squares regression (WA-PLS): an improved method for reconstructing environmental variables from species assemblages. Hydrobiolo- gia, 269/270, (1), 485-502. 168 Vallè F. et al Ms. received: May 7, 2018 Final text received: May 16, 2018