Journal of Applied Botany and Food Quality 89, 212 - 216 (2016), DOI:10.5073/JABFQ.2016.089.027 1Council for Agricultural Research and Economics (CREA), Research Centre for Fodder Crops and Dairy Productions (FLC), Lodi, Italy 2 CREA, Research Unit for Mediterranean Agro-Pastoral Systems (AAM), Sanluri, Italy 3 Indena S.p.A., QC/R&D Laboratories, Settala (Milan), Italy Detection and exploitation of white lupin (Lupinus albus L.) genetic variation for seed γ-conglutin content P. Annicchiarico1*, M. Romani1, S. Barzaghi1, B. Ferrari1, A.M. Carroni2, P. Ruda2, E. De Combarieu3, L. Pagni3, V. Tedesco3 (Received February 10, 2016) * Corresponding author Summary The seed γ-conglutin protein fraction of white lupin has particular pharmacological interest, but its industrial production is hindered by low content in the seed. This study provides an unprecedented assessment of genotypic and environmental variation for seed con- tent and production of γ-conglutin, exploring also the ability of Near-Infrared Spectroscopy (NIRS) to predict seed γ-conglutin content. Significant (P < 0.01) genetic variation for seed γ-conglutin content emerged among ten genotypes (cultivars or breeding lines) across three environments (range: 1.59-2.02 %) and five genotypes in other two environments (range: 1.47-1.80 %). Genotype variation was found also for seed protein content and γ-conglutin proportion on total protein, the latter trait having higher impact than the former on genotype variation for seed γ-conglutin content. The production of γ-conglutin per unit area was affected also by genotype yielding ability beside genotype seed γ-conglutin content. No genotype × environment interaction was detected for any γ-conglutin trait. NIRS- based prediction based on cross-validations was only moderately accurate for seed γ-conglutin content (R2 = 0.66), while being accurate for seed protein content (R2 = 0.95). In conclusion, breeding for higher seed γ-conglutin content is feasible using data from very few test sites and, to some extent, NIRS-based predictions. Introduction Greater cultivation of rain-fed cool-season grain legumes could improve the sustainability of European agriculture in many re- spects (Nemecek et al., 2008). White lupin (Lupinus albus L.) is a traditional food legume with good potential as a high-protein feed crop in several European regions (Lucas et al., 2015), owing to its moderately high grain yield, outstanding seed protein content, and good content of essential amino acids (PaPiNeau and HuygHe, 2004; sujak et al., 2006; aNNiccHiarico, 2008). In addition the white lupin seed contains 8-12 % of oil with excellent nutritional characteristics (BoscHiN et al., 2008), and possesses various pro- perties that make it particularly useful as an ingredient of functio- nal, nutraceutical or healthy foods (arNoLdi, 2005; duraNti et al., 2008). The γ-conglutin protein fraction has particular interest in this respect, owing to its proved ability to control glycaemia via interaction and binding with insulin and its insulin-mimetic properties (BertogLio et al., 2011; Lovati et al., 2012). A patent, which awaits commercial exploitation, has been granted for the pharmacological use of γ-conglutin for the treatment of type II diabetes (morazzoNi and duraNti, 2004). Large-scale production of white lupin seed γ-conglutin is hindered by its modest content in the seed, which represents only about 4-5 % of the total protein (duraNti et al., 2008). Optimizing its industrial production requires information on the extent of genetic, environmental, and genotype × environment interaction effects that may influence its content. Earlier work on other seed quality traits of white lupin, such as oil, tocopherol or alkaloid contents, revealed that the relative size of these effects may vary largely depending on the specific trait (aNNiccHiarico et al., 2014), with implications for crop improvement strategies and optimal production environments. Despite its practical importance, we found no published information on the extent of genetic and environmental variation for seed γ-conglutin of white lupin. If exploitable genetic variation existed, it would be important to know whether it is associated mainly with overall protein content or with the proportion of γ-conglutin on total protein. Only the latter case would prompt selection programs to specific analyses of genotype seed γ-conglutin content (which are more expensive than protein content analyses). In that case, Near- Infrared Spectroscopy (NIRS) procedures could be explored as a means to reduce the costs of seed γ-conglutin analyses for plant breeding material or seed lots from different production environments. NIRS has been already used to predict seed coat proportion in intact seeds of lupin (aLomar et al., 2010) and contents of amino acids and protein in the seed (BistoN et al., 1983; kaffka, 1988). The main objective of this study was generating information on the extent of genetic and environmental variation for γ-conglutin in white lupin seed. In addition we report preliminary information on the relationship of seed γ-conglutin with total protein content and the ability of NIRS procedures to predict seed γ-conglutin content. Materials and methods Plant material and cultivation Seven elite breeding lines (7-50, 685, 722, B-17, B-20, MB-29 and MB-38) issued by a white lupin breeding program for climatically- contrasting Italian regions that is described in aNNiccHiarico et al. (2011), and three commercial varieties (Lucky, Rex, Rumbo), were grown in three autumn-sown field experiments with the aim to gather preliminary information on the genetic variation for seed γ-conglutin. The experiments were performed in the subcontinental- climate site of Lodi (Northern Italy) in the seasons 2011-12 and 2012-13, and the Mediterranean-climate site of Sanluri (Sardinia) in 2012-13. Climate and soil characteristics of these test environments are reported in Tab. 1. White lupin underwent low-temperature stress during winter in Lodi and terminal drought stress mainly in Sanluri (featuring modest spring rainfall), in agreement with the general climatic features of the sites. Soil characteristics, including active lime < 0.5 % and pH < 7.5, were always suitable for lupin cultivation (PaPiNeau and HuygHe, 2004). These experiments adopted a ran- domized complete block design with three replications, plots of 3 m2 size, and 20 and 30 germinating seeds/m2 in 2011-12 and 2012- 13, respectively. In these and following experiments, the seed was inoculated with NPPL HiStick (Becker Underwood) before sowing. Seed γ-conglutin analyses were carried out for each genotype- environment combination on a seed sample of 50 g that was obtained by bulking same amounts of seed from each experiment replicate. White lupin variation for seed γ-conglutin content 213 Two breeding lines that displayed high and moderate seed γ-conglutin content in these experiments (lines 7-50 and MB-38, respectively), and three varieties (Lucky; Multitalia; Rumbo), were grown in autumn-sown field experiments in Lodi and Sanluri in the season 2013-14. To further enhance the environmental differences between testing sites, the evaluation in Sanluri was performed in a field whose soil active lime and pH (Tab. 1) were definitely higher than those required for optimal lupin growth (PaPiNeau and HuygHe, 2004), unlike Lodi. Grain yield in Lodi was limited by biotic stresses (particularly Fusarium spp.). Rainfall and temperature patterns were relatively favourable in both locations (Tab. 1). These experiments adopted a randomized complete block design with four replications, plots of 6 m2 size, and 30 germinating seeds/m2. Genotype grain yields were assessed in all field replicates, whereas seed γ-conglutin and protein contents were assessed on two replicates per experiment using 50 g seed samples. Chemical and NIRS analyses Lupin seed samples were ground to 1 mm particle-size in three grinding steps using a “Pulverisette 14” Variable Speed Rotor Mill (Fritsch GmbH, Idar-Oberstein, Germany) with different sieve rings. The γ-conglutin content of each ground sample was assessed by Indena’s proprietary HPLC-UV method, whose details cannot be disclosed completely. A single, exhaustive extraction of γ-conglutin from the lupin flour was obtained in a Tris-Glycine buffer (extraction solvent/flour ratio: 200). After centrifugation, the supernatant was analyzed twice on a reverse phase HPLC, using an apolar stationary phase butyl silicagel and an acetonitrile/water/TFA gradient. De- tection was carried out by ultraviolet at 210 nm. Quantification was carried out against a γ-conglutin reference standard. This method was assessed for linearity (0.01- 0.4 mg/ml), precision (relative standard deviation < 4 %), and stability of the sample solution (up to 48 hours). Seed protein content was determined as described by kjeLdaHL (1883). NIRS analyses were performed on 46 seed samples, which included: (i) 30 genotype-environment combinations of experiments in the seasons 2011-12 and 2012-13; (ii) ten genotype-environment com- binations of experiments in 2013-14 (using seed of the first field replicate); (iii) six additional seed samples relative to other breeding lines that were evaluated only in Lodi in 2011-12 or 2013-14. Two ground seed replications of each sample were analyzed in Petri dishes (re-blending and re-loading the dish between measurements), averaging results of the two measurements. We used a FTNIR spectrometer (NIRFlex N500, Büchi, Italy) in the whole NIRS range (10,000-4,000 cm-1). NIRS spectra were collected in reflectance mode, adding 64 scans with a 4 cm-1 resolution. Spectra were con- verted in absorbance mode before data analysis. Statistical analysis Genotype data of seed γ-conglutin from Lodi and Sanluri in the seasons 2011-12 and 2012-13 underwent an analysis of variance (ANOVA) aimed to compare the genotype means across the three environments and the environment means, using genotype × en- vironment interaction as the error term. Plot data of seed γ-conglutin and protein contents, proportion of γ-conglutin on total protein, and grain and γ-conglutin dry matter (DM) yield from Lodi and Sanluri collected in the season 2013-14 underwent an ANOVA aimed at assessing genotype, environment and genotype × environment interaction effects and comparing genotype means across environments. Genotype γ-conglutin DM yield on a plot basis was computed as the product of plot yield by genotype seed γ-conglutin content on the site. Data from these environments were also used for simple and multiple regression analyses of seed γ-conglutin content as a function of protein con- tent and/or γ-conglutin proportion of total protein. All ANOVA and regression analyses were performed using Statistical Analysis Software (SAS Institute Inc., Cary, NC, USA). NIRS calibrations to quantify the seed contents of γ-conglutin and protein were calculated using the software PLS Toolbox (Eigenvector Research Inc, Manson, WA, USA). We used different spectra pre- treatments to improve NIRS calibration performances. Owing to the limited number of available samples, each calibration curve was assessed through cross-validations that used 11 % of the spectra for predictions, repeating the procedure nine times using contiguous data blocks. Results Significant (P < 0.01) genetic variation among ten genotypes emer- ged for mean seed γ-conglutin content across three test environ- ments in the cropping seasons 2011-12 and 2012-13. Genotype val- ues ranged from 1.59 % to 2.02 %, both exhibited by breeding lines. Control varieties displayed moderate to low values (1.60 -1.83 %). Tab. 1: Climate and soil characteristics, and mean white lupin grain yield and seed γ-conglutin content, for two test locations (Lodi, Northern Italy; Sanluri, Sardinia) in different growing seasons. Lodi Sanluri Variable 2011-12 2012-13 2013-14 2012-13 2013-14 Rainfall Nov-Feb (mm) 124 365 427 266 432 Rainfall Mar-Jun (mm) 222 598 270 148 163 Lowest winter temp (°C) -17.0 -11.3 -5.7 -2.3 0.2 Mean temperature Nov-Feb (°C) 2.4 5.0 4.9 10.8 16.1 Mean temperature Mar-Jun (°C) 15.7 14.2 16.6 11.2 16.6 Soil active lime (%) a ~0 ~0 ~0 0.3 2.6 Soil pH 6.2 6.3 6.2 7.4 8.1 Grain yield (t/ha) b 6.41 5.10 3.30 6.24 1.15 Seed γ-conglutin content (% w/w) b 1.96 1.64 1.73 1.88 1.52 a According to Drouineau (1942). b Averaged across 10 genotypes in 2011-12 and 2012-13, and five genotypes in 2013-14. 214 P. Annicchiarico, M. Romani, S. Barzaghi, B. Ferrari, A.M. Carroni, P. Ruda, E. De Combarieu, L. Pagni, V. Tedesco On average, the environment of Lodi 2012-13 displayed lower seed γ-conglutin content (1.64±0.05 %) than Lodi 2011-12 (1.96±0.05 %) and Sanluri 2012-13 (1.88±0.05 %) (P < 0.05). One breeding line (7-50) with high seed γ-conglutin content (1.95 %; Tab. 2), and a sec- ond line (MB-38) with moderate γ-conglutin content (1.72 %), were selected for further testing in Lodi and Sanluri in 2013-14 along with three commercial varieties. The ANOVA performed on data from the season 2013-14 detected no genotype × environment interaction for every γ-conglutin trait, namely, seed γ-conglutin content, proportion of γ-conglutin on total protein, and γ-conglutin yield per unit area. In contrast, genotype × environment interaction (P < 0.05) was detected for seed protein content and grain yield. Large variation for seed γ-conglutin content of genotypes across environments was found also in this data set (P < 0.01), with values ranging from 1.47 % in Multitalia to 1.80 % in the line 7-50 (Tab. 2). The ranking of four genotypes for seed γ-conglutin content across two environments in 2013-14 was iden- tical to that across three environments in 2011-12 and 2012-13 (Tab. 2), confirming the high consistency of genotype responses across environments for this trait. Significant (P < 0.01) genotype variation across environments of the season 2013-14 emerged for all other traits. Line 7-50 achieved top- values of seed γ-conglutin content via high values of both seed pro- tein content and proportion of γ-conglutin on total protein (Tab. 2). The variety Rumbo exhibited low seed γ-conglutin content because of low proportion of γ-conglutin on total protein, since its protein content was the highest one (Tab. 2). Results of simple and multiple regression analyses, which are summarized in Tab. 3, confirmed that: (i) seed γ-conglutin content of the genotypes was largely determined by the proportion of γ-conglutin in the protein fraction (with 81 % of genotype variation explained by this variable); (ii) additional infor- mation on seed protein content had some value for predicting seed γ-conglutin content (as showed by its significance in the multiple regression model). The production of γ-conglutin per unit area of the genotypes was also influenced by genotype grain yielding ability besides seed γ-conglutin content. The variety Lucky and the line 7-50, which com- bined high values of both variables, were top-ranking for γ-conglutin yield (Tab. 2). On average, the unfavorable cropping environment of Sanluri in 2013-14 exhibited, in comparison to Lodi in the same season, not only almost three-fold lower grain yield but also lower seed γ-conglutin content (Tab. 4). The latter response was the result of distinctly low- er seed protein content, while the proportion of γ-conglutin on total protein did not differ between environments (Tab. 4). A NIRS-based calibration curve for seed γ-conglutin content was established, which held five latent variables and used a standard normal variate and a gap segment 2nd derivative as pretreatments to process the spectra. It displayed moderate predicting ability based on cross-validations (R2 = 0.66, with a root mean square error of 0.14). Predictions for individual observations tended to shrink the range of γ-conglutin content relative to observed values (Fig. 1). A NIRS- based calibration curve for seed proteins content, which used four latent variables and a standard normal variate plus Savitzky-Golay 1st derivative algorithm as pretreatments, achieved much higher pre- diction (R2 = 0.95, with a root mean square error of 0.81). Discussion Our study provides an unprecedented assessment of the extent of genotypic and environmental variation for seed content and produc- tion of seed γ-conglutin in white lupin. Its indications, while being valuable in the absence of other information, cannot be considered as conclusive, owing to the fairly limited sample of genotypes and test environments they are based upon. Our findings indicate that white lupin selection for high seed con- tent of γ-conglutin is feasible, on the basis of (i) the large genetic variation detected for this trait, and (ii) the consistency of genotype differences across environments. The latter result is supported by the lack of genotype × environment interaction in the ANOVA for every γ-conglutin trait, and the consistent ranking of four genotypes for seed γ-conglutin content across two sets of environments. The consistency of genotype responses for seed γ-conglutin content fa- cilitates genotype selection, which may rely on results from very few test environments, and ensures that selection gains are maintained in Tab. 2: Mean values over two climatically-contrasting Italian locations of seed γ-conglutin and protein contents (in % w/w) measured by HPLC and Kjeldahl methods, respectively, and grain and γ-conglutin dry matter (DM) yield, for five white lupin genotypes. Genotype γ-conglutin Protein γ-conglutin Grain DM γ-conglutin DM γ-conglutin % a % a % on protein a (t/ha) a (kg/ha) a % b 7-50 1.80 a 34.15 b 0.053 a 2.35 b 42.4 ab 1.95 a Lucky 1.73 ab 32.50 c 0.053 a 2.77 a 47.8 a 1.83 ab MB-38 1.61 bc 33.77 b 0.048 ab 1.84 c 29.6 c 1.72 ab Rumbo 1.51 c 35.83 a 0.042 b 1.59 c 24.0 d 1.60 b Multitalia 1.47 c 32.15 c 0.046 b 2.59 ab 38.1 b - SE (DF) c 0.04 (8) 0.34 (8) 0.002 (8) 0.10 (24) 1.8 (24) 0.09 (18) Column means with different letter differ at P < 0.05 according to Duncan’s test. a Over two environments (Lodi and Sanluri, 2013-14). b Over three environments (Lodi 2011-12 and 2012-13; Sanluri, 2012-13). c SE, standard error of genotype mean; DF, associated degrees of freedom. Tab. 3: Ability of simple and multiple linear regressions models to explain seed γ-conglutin content variation of five white lupin genotypes across two climatically-contrasting Italian environments as a func- tion of seed protein content and/or γ-conglutin proportion on total protein. Model variables R2 P a γ-conglutin proportion on total protein 0.81 0.05 Seed protein content 0.01 0.05 γ-conglutin proportion on total protein 0.99 0.05 / 0.05 + Seed protein content a P level significance of explanatory variables in the model. White lupin variation for seed γ-conglutin content 215 different cropping environments. Other seed quality traits of white lupin displayed less favourable responses, for example tocopherol content, whose genetic improvement proved hindered by high geno- type × environment interaction (aNNiccHiarico et al., 2014). The scope for selecting for higher seed γ-conglutin content is reinforced by the possibility to achieve values of γ-conglutin proportion on total protein that exceed the 4-5 % range reported in literature (duraNti et al., 2008), as in the case of the genotypes 7-50 and Lucky (5.3 %; Tab. 2). Our NIRS calibration results, although preliminary, suggest that NIRS-based predictions for seed γ-conglutin content are not suf- ficiently accurate to substitute completely for chemical analyses. However, NIRS data could be exploited in preliminary selection stages for higher seed γ-conglutin content, performing the costly chemical analysis for this trait only on breeding lines promoted to final selection stages. We verified by earlier work that NIRS cali- bration for seed γ-conglutin content based on spectra recorded from whole grain samples, which are exploitable also for selection of in- dividual seeds, are characterized by too modest predicting ability to be of practical use (R2 = 0.27; ferrari et al., 2015). NIRS prediction based on ground seed samples can be used in breeding programs to select genotypes on the basis of seed γ-conglutin content of their bulked progeny seed. The greater dependency of seed γ-conglutin content of lupin cul- tivars from γ-conglutin proportion on total protein than from seed protein content, which emerged from regression analyses, has prac- tical implications for breeding programs. It suggests that selection for higher protein content, which could reliably be performed by NIRS measurements, may have limited value for increasing seed γ-conglutin content. Our results show that high γ-conglutin production per unit area, which is the ultimate target for industrial production of γ-conglutin, is largely affected also by high grain yielding ability of the geno- types. Therefore, breeding strategies aimed to improve crop yields, e.g., by exploiting innovative plant types, stress tolerance, global genetic resources and adaptation to specific climatic conditions (HuygHe, 1997; aNNiccHiarico et al., 2010), remain of utmost im- portance even when targeting higher yield of this protein fraction. The current information on environmental variation for γ-conglutin traits is limited. Its results for the season 2013-14 suggest that a favour- able environment may increase the crop production of γ-conglutin not only via higher grain yield but also via higher seed γ-conglutin content that derived, in this case, from higher seed protein content. Results for three relatively favourable cropping environments in the seasons 2011-12 and 2012-13 failed to reveal distinct and/or consis- tent differences between the subcontinental-climate site of Lodi and the Mediterranean-climate site of Sanluri. In conclusion, this study indicates that white lupin seed content of γ-conglutin can be enhanced through plant breeding. Selection costs can be reduced by adopting very few test sites because of limited genotype × environment interaction, and by relying on NIRS-based predictions in early stages of selection. Acknowledgements The results of this study were generated by the projects “White lupin for functional foods” funded by INDENA and “Plant Genetic Resources – FAO Treaty” funded by the Italian Ministry of Agricul- tural and Forestry Policies. References aLomar, d., mera, m., erraNdoNea, j., miraNda, H., 2010: Prediction of seed coat proportion in narrow-leafed and yellow lupins by near-infrared reflectance spectroscopy (NIRS). Crop Pasture Sci. 61, 304-309. aNNiccHiarico, P., 2008: Adaptation of cool-season grain legume species across climatically-contrasting environments of southern Europe. Agron. J. 100, 1647-1654. aNNiccHiarico, P., BoscHiN, g., maNuNza, P., arNoLdi, a., 2014: Quality of Lupinus albus L. (white lupin) seed: extent of genotypic and environ- mental effects. J. Agric. Food Chem. 62, 6539-6545. aNNiccHiarico, P., Harzic, N., carroNi, A.M., 2010: Adaptation, diver- sity, and exploitation of global white lupin (Lupinus albus L.) landrace genetic resources. Field Crops Res. 119, 114-124. aNNiccHiarico, P., maNuNza, P., Proietti, s., 2011: White lupin tolerance to winter cold, terminal drought and soil lime: patterns of genetic vari- ation and their exploitation in breeding for southern Europe. In: Naga- Tab. 4: Mean values over five white lupin genotypes of seed γ-conglutin and protein contents (in % w/w) measured by HPLC and Kjeldahl methods, respec- tively, and grain and γ-conglutin dry matter (DM) yield, for two contrasting cropping environments (Lodi, northern Italy; Sanluri, Sardinia) in the season 2013-14. Environment γ-conglutin Protein γ-conglutin Grain DM γ-conglutin DM % % % on protein (t/ha) (kg/ha) Lodi 1.73 a 37.69 a 0.046 a 3.30 a 52.2 a Sanluri 1.52 b 29.68 b 0.051 a 1.15 b 18.9 b SE (DF) a 0.05 (2) 0.36 (2) 0.002 (2) 0.12 (6) 2.4 (6) Column means with different letter differ at P < 0.05. a SE, standard error of genotype mean; DF, associated degrees of freedom. Fig. 1: Scatter plot of measured vs. predicted seed γ-conglutin content (in % w/w) for 46 genotype-environment combinations. Prediction based on a Near-Infrared Reflectance spectroscopy calibration curve, using cross-validations (continuous line = regression line; broken line = 1:1 line) 216 P. Annicchiarico, M. Romani, S. Barzaghi, B. Ferrari, A.M. Carroni, P. Ruda, E. De Combarieu, L. Pagni, V. Tedesco nowska, B., Kachlicki, P., Wolko, B. (eds.), Lupin crops – An oppor- tunity for today, a promise for the future, 99-103. International Lupin Association, Canterbury. arNoLdi, a., 2005: Optimized processes for preparing healthy and added value food ingredients from lupin kernels, the European protein-rich grain legume. Aracne, Rome. BertogLio, j.c., caLvo, m.a., HaNcke, j.L., Burgos, r.a., riva, a., morazzoNi, P., PoNzoNe, c., magNi, c., duraNti, M., 2011: Hypogly- cemic effect of lupin seed gamma-conglutin in experimental animals and healthy human subjects. Fitoterapia 82, 933-938. BistoN, r., dardeNNe, g., siNdic, m., dardeNNe, P., WatHeLet, B., sousa, B. de, 1983: Near infra-red spectroscopy as a rapid and accurate technique for estimating Lupinus seed protein and Vicia faba seed amino acids and protein. In: Thompson, R., Casey, R. (eds.), Perspectives for peas and lupins as protein crops, 247-255. Martinus Nijhoff, The Hague, Netherlands. BoscHiN, g., d’agostiNa, a., aNNiccHiarico, P., arNoLdi, a., 2008: Ef- fect of genotype and environment on fatty acid composition of Lupinus albus L. seed. Food Chem. 108, 600-606. DrouiNeau, G., 1942: Dosage rapide du calcaire actif du sol. Ann. Agron. 12, 411-450. duraNti, m., coNsoNNi, a., magNi, c., sessa, f., scarafoNi, a., 2008: The major proteins of lupin seed: characterisation and molecular proper- ties for use as functional and nutraceutical ingredients. Trends Food Sci. Tech. 19, 624-633. ferrari, B., BarzagHi, s., aNNiccHiarico, P., romaNi, m., PagNi, L., de comBarieu e., 2015: Development of a NIR calibration to select white lupin genotypes with high gamma-conglutin fractions. In: Capraro, J., Duranti, M., Magni, C., Scarafoni, a. (eds.), Developing lupin crop into a major and sustainable food and feed source, 100. University of Milan, Milan. HuygHe, C., 1997: White lupin (Lupinus albus L.). Field Crops Res. 53, 147-160. kaffka, k.j., 1988: Determination of amino acids in lupine by near infrared reflectance spectroscopy. Acta Alimentaria 17, 3-11. KjeLdaHL, J., 1883: Neue Methode zur Bestimmung des Stickstoffs in or- ganischen Körpern. Z. Anal. Chem. 22, 366-383. Lovati, m.r., maNzoNi, c., castigLioNi, s., ParoLari, a., magNi, c., duraNti, m., 2012: Lupin seed γ-conglutin lowers blood glucose in hyperglycaemic rats and increases glucose consumption of HepG2 cells. Br. J. Nutr. 107, 67-73. Lucas, m.m., stoddard, f., aNNiccHiarico, P., frías, j., martíNez- viLLaLueNga, c., sussmaNN, d., duraNti, m., seger, a., zaNder, P., Pueyo, j.j., 2015: The future of lupin as a protein crop in Europe. Front. Plant Sci. 6, 705. morazzoNi, P., duraNti, m., 2004: The use of lupin conglutin for the treat- ment of type II diabetes. Patent WO2004071521. Nemecek, t., voN ricHtHofeN, j.-s., duBois, g., casta, P., cHarLes, r., PaHL, H., 2008: Environmental impact of introducing grain legumes into European crop rotations. Eur. J. Agron. 28, 380-393. PaPiNeau, j., HuygHe, c., 2004: Le lupin doux protéagineux. Editions France Agricole, Paris. sujak, a., kotLarz, a., stroBeL, W., 2006: Compositional and nutritional evaluation of several lupin seeds. Food Chem. 98, 711-719. Address of the authors: P. Annicchiarico, M. Romani, B. Ferrari, S. Barzaghi, Council for Agricul- tural Research and Economics (CREA), Research Centre for Fodder Crops and Dairy Productions (FLC), viale Piacenza 29, 26900 Lodi, Italy E-Mail: paolo.annicchiarico@crea.gov.it A.M. Carroni, P. Ruda, CREA, Research Unit for Mediterranean Agro-Pasto- ral Systems (AAM), Podere Ortigara, 09025 Sanluri, Italy L. Pagni, E. De Combarieu, V. Tedesco, Indena S.p.A., QC/R&D Laborato- ries R&D, via Don Minzoni 6, 20090 Settala (Milan), Italy © The Author(s) 2016. This is an Open Access article distributed under the terms of the Creative Commons Attribution Share-Alike License (http://creative- commons.org/licenses/by-sa/4.0/).