Establishing a sampling protocol to estimate tobacco specific nitrosamines in growers bales: Results from the first year of a two year study Colin Fisher 1 & Kristen McQuerry 2 1 Plant & Soil Science, University of Kentucky 2 Applied Statistics Laboratory, University of Kentucky 2016 CORESTA
Introduction Every attempt to reduce TSNA at all stages of production and manufacture Growing and curing conditions impact TSNA in farmer packages at sale. Point of sale is first opportunity for buyer to sample cured leaf Sampling may be required anytime before redrying.
Grower marketing packages Country Max wt. cu. M kg Indonesia 0.14 35 US 0.16 41 Poland 0.19 40 Brazil 0.19 55 France 0.20 25 Philippines 0.23 23 Thailand 0.28 60 Guatemala 0.32 62 Zimbabwe 0.32 90 Indonesia 0.32 100 Thailand 0.43 70 US (big) 1.16 230/300
CORESTA Sub-Group TSNA in Air-Cured and Fire-Cured Tobacco Draft Protocol for sample handling of post-cure tobacco (a.k.a. farmer bales) included: Section 1aii: 1. Collect 2 samples per 25 to 40 kg bale 2. Open bale and collect 2 samples of at least 0.7 kg per sample from center of bale. 3. For larger bales (250 kg), core sampler can be used. Collect at least 1.4 kg of leaf per bale (at least 14 cores if core is 100 g).
Quebec 2014 Aspects of protocol queried: 1. Total amount sample from each bale (1.4 kg) 2. Number samples from each size of bale Small bales - 2 grab samples of whole leaf Big bales 14 core samples 3. Effect of inclusion of midrib, especially in cored samples
Objective: Clarify sampling protocol to optimize sample size number of samples to best represent TSNA content of whole bale Year 1: Characterize variability of TSNA within individual bales
Assumptions and Limitations: Industry most interested in lamina chemistry, rather than whole leaf or midrib Number of bales tested limited by total number of samples for chemistry analysis Only burley in this study Only NNN addressed in this presentation Lamina+Midrib data = weighted concentration
Methods Small bales - 25-100 kg 6 bales Big bales - 250 kg 6 bales
Sampling Methods: Big Bales Three 45 cm cores diagonally across top and side of each bale (2 cores from same position if one core too little sample)
Sampling Methods: Small Bales Three methods: 1. Core Tip ¾ leaf length Mid-leaf Midrib ¼ leaf length x 4
Sampling Methods: Small Bales 2. Grab Sample (protocol) 30 40 leaves from each of 4 depths of each bale
Sampling Methods: Small Bales 3. Perfect Sample + 160 individual leaves randomly selected from throughout bale. then randomly divided into 4 samples of 40 leaves each
Sample preparation: 6 big bales X top & side X 3 cores each X lamina & midrib = 72 samples 6 small bales X 20 samples/bale X lamina & midrib = 240 samples
Big Bales - Variability of NNN in Lamina + Midrib between samples from same bale NNN, µg g- 1 30 25 20 15 10 5 0 Whole Leaf (weighted Lamina + Midrib) 1 2 3 4 5 6 Bale # Max Mean Min SD
NNN, ug/g NNN, ug/g Variability of NNN in Lamina and Midrib between samples from same bale 25 20 15 10 5 0 Lamina Max Mean Min SD 1 2 3 4 5 6 Bale # 80 60 40 20 0 Midrib Max Mean Min SD 1 2 3 4 5 6 Bale # Therefore: log transformed to normalize data for statistical analysis
Big Bales: Sampling from Side vs. Top
Ln single sample Lamina+Midrib Ln single sample Lamina+Midrib Big Bales: Sampling from Side vs. Top 3.5 Ln Lamina+Midrib Single SIDE Sample vs bale mean side samples (n=3) 3.5 Ln Lamina+Midrib Single TOP Sample vs. bale mean top samples (n=3) 3.0 3.0 2.5 2.0 1.5 1.0 0.5 Bottom Side R² = 0.984 R² = 0.955 Top R² = 0.991 0.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Ln Mean Lamina+Midrib 2.5 2.0 1.5 1.0 0.5 Centre Left Right R² = 0.860 R² = 0.858 R² = 0.890 0.0 0.0 1.0 2.0 3.0 4.0 Ln Mean Lamina+Midrib
Lamina:Lam+Mid Ratio Big Bales: Ratio of Lamina NNN: Midrib NNN Bale # Lamina + Midrib Lamina Lamina/ Lam + Mid 1 21.9 18.4 0.84 2 0.8 0.6 0.78 3 8.6 6.6 0.77 4 12.9 8.7 0.67 5 5.4 4.4 0.81 6 5.4 4.4 0.82 1.0 0.8 0.6 0.4 0.2 0.0 Ratio of Lamina NNN:Midrib NNN 1 2 3 4 5 6 Bale #
NNN Lamina+Midrib Small Bales: Variability of NNN between samples in perfect samples 20 15 10 5 0 Mean Max Min SD 1 2 3 Bale # 4 5 6
Small Bale Sampling Method Burton, Dye, Bush. 1992. Distribution of tobacco constituents in tobacco leaf tissue. 1. Tobacco-specific nitrosamines, nitrate, nitrite, and alkaloids. J. Agric. Food Chem., 40 (6):1050 1055 0.8 NNN µg g -1 0.6 0.4 0.2 0.0 0 1 2 3 4 5 6 7 8 9 10 Tip Segment # Butt
Proportion of contribution to total NNN Fisher, Pearce. 2013. Assessing yield loss from hail in burley tobacco. CORESTA Agro-Phyto Meeting, Italy. NNN concentrations along length of leaf 30 25 20 15 10 5 0 0 2 4 6 8 10 Burton Segment # (1 = tip, 10 = butt) Lamina Midrib Sum L+M Straight laid small bale core samples: NNN of lamina + midrib at half way along length of leaf Lamina only NNN
Small Bales: NNN by sample method & position across all bales Sample Method & Position Indiv Lvs Perfect Sample Lamina NNN Midrib Lamina + Midrib 3.2 13.1 5.5 Hand Pull 3.5 10.9 5.3 Core Butt 5.2 13.8 8.7 Core Mid 3.0 8.0 3.9 Core Tip 2.5 7.3 2.9
Results of Single Sample NNN Position of Lamina+Midrib predicting Lamina of Individual Leaves The Rep the single sample originated Order of Significance per Single Sample Position p-value 1 1 Mid <.0001 2 Butt 0.0018 3 Tip 0.0033 4 Hand Pull 0.0505
Results of Single Sample NNN Position of Lamina+Midrib predicting Lamina of Individual Leaves The Rep the single sample originated Order of Significance per Single Sample Position p-value 1 1 Mid <.0001 2 Butt 0.0018 3 Tip 0.0033 4 Hand Pull 0.0505 2 1 Mid 0.0030 2 Tip 0.0030 3 Butt 0.0115 4 Hand Pull 0.0492
Results of Single Sample NNN Position of Lamina+Midrib predicting Lamina of Individual Leaves The Rep the single sample originated Order of Significance per Single Sample Position p-value 1 1 Mid <.0001 2 Butt 0.0018 3 Tip 0.0033 4 Hand Pull 0.0505 2 1 Mid 0.0030 2 Tip 0.0030 3 Butt 0.0115 4 Hand Pull 0.0492 3 1 Mid 0.0002 2 Tip 0.0004 3 Butt 0.0005 4 Hand Pull 0.0005
Results of Single Sample Position of Lamina+Midrib predicting Lamina of Individual Leaves The Rep the single sample originated Order of Significance per Single Sample Position p-value 1 1 Mid <.0001 2 Butt 0.0018 3 Tip 0.0033 4 Hand Pull 0.0505 2 1 Mid 0.0030 2 Tip 0.0030 3 Butt 0.0115 4 Hand Pull 0.0492 3 1 Mid 0.0002 2 Tip 0.0004 3 Butt 0.0005 4 Hand Pull 0.0005 4 1 Hand Pull 0.0007 2 Tip 0.0014 3 Mid 0.0024 4 Butt 0.0067
Predicting Whole Leaf NNN from single samples of Mid-leaf core Lamina+Midrib NNN Position Frequency Percent Hand Pull 4 16.67 Butt 4 16.67 Mid 12 50.00 Tip 4 16.67 Total Possible Iterations 24
Both bales sizes: Small data set n=6 Conclusions Large variability of NNN within single bale Gross variability diminishes as mean NNN concentration decreases Data normalized with log-transformation At approx. 3 µg g -1, range +1 2 µg g -1 Probably acceptable?
Big bales: Conclusions No possibility avoiding butts in sample NNN of lamina estimated from 0.8 x (lamina + midrib)? If not, lamina should be separated from midrib before analysis if NNN of lamina only required Single core could be sufficient, but more to reduce error
Small bales: Conclusions NNN concentration of mid-leaf core samples NNN concentration of whole leaf lamina One or two samples sufficient Sample through most of bale depth 2016 Study: 3 cores from mid-leaf and either side of mid-leaf
Acknowledgments: Sample Bales Financial Support Alliance One International British American Tobacco Philip Morris International Altria Client Services Kentucky Tobacco Research and Development Center RJ Reynolds