Figure 1.
Spearman’s rank correlation coefficient for deoxynivalenol (DON) contamination in Swedish spring oats at harvest and different weather factors estimated for 14-day moving windows during the growing season. Red indicates a positive correlation and blue a negative correlation (both p ≤ 0.01) between DON contamination and a particular weather variable, with a darker colour indicating a higher value of the correlation coefficient. Tmin-daily minimum temperature, Tmean-daily mean temperature, Tmax-daily maximum temperature, RH-mean relative humidity, PREC-precipitation, VPD-vapour pressure deficit.
Figure 1.
Spearman’s rank correlation coefficient for deoxynivalenol (DON) contamination in Swedish spring oats at harvest and different weather factors estimated for 14-day moving windows during the growing season. Red indicates a positive correlation and blue a negative correlation (both p ≤ 0.01) between DON contamination and a particular weather variable, with a darker colour indicating a higher value of the correlation coefficient. Tmin-daily minimum temperature, Tmean-daily mean temperature, Tmax-daily maximum temperature, RH-mean relative humidity, PREC-precipitation, VPD-vapour pressure deficit.
Figure 2.
Spearman´s rank correlation coefficient for deoxynivalenol (DON) contamination in Swedish spring barley at harvest and different weather factors estimated for 14-day moving windows during the growing season. Red indicates a positive correlation and blue a negative correlation (both p ≤ 0.01) between DON contamination and a particular weather variable, with a darker colour indicating a higher value of the correlation coefficient. Tmin-daily minimum temperature, Tmean-daily mean temperature, Tmax-daily maximum temperature, RH-mean relative humidity, PREC-precipitation, VPD-vapour pressure deficit.
Figure 2.
Spearman´s rank correlation coefficient for deoxynivalenol (DON) contamination in Swedish spring barley at harvest and different weather factors estimated for 14-day moving windows during the growing season. Red indicates a positive correlation and blue a negative correlation (both p ≤ 0.01) between DON contamination and a particular weather variable, with a darker colour indicating a higher value of the correlation coefficient. Tmin-daily minimum temperature, Tmean-daily mean temperature, Tmax-daily maximum temperature, RH-mean relative humidity, PREC-precipitation, VPD-vapour pressure deficit.
Figure 3.
Spearman´s rank correlation coefficient for deoxynivalenol (DON) contamination in Swedish spring wheat at harvest and different weather factors estimated for 14-day moving windows during the growing season. Red indicates a positive correlation and blue a negative correlation (both p ≤ 0.01) between DON contamination and a particular weather variable, with a darker colour indicating a higher value of the correlation coefficient. Tmin-daily minimum temperature, Tmean-daily mean temperature, Tmax-daily maximum temperature, RH-mean relative humidity, PREC-precipitation, VPD-vapour pressure deficit.
Figure 3.
Spearman´s rank correlation coefficient for deoxynivalenol (DON) contamination in Swedish spring wheat at harvest and different weather factors estimated for 14-day moving windows during the growing season. Red indicates a positive correlation and blue a negative correlation (both p ≤ 0.01) between DON contamination and a particular weather variable, with a darker colour indicating a higher value of the correlation coefficient. Tmin-daily minimum temperature, Tmean-daily mean temperature, Tmax-daily maximum temperature, RH-mean relative humidity, PREC-precipitation, VPD-vapour pressure deficit.
Figure 4.
Spearman´s rank correlation coefficient for deoxynivalenol (DON) contamination in Lithuania grown spring wheat at harvest and different weather factors estimated for 14-day moving windows during the growing season. Red indicates a positive correlation and blue a negative correlation (both p ≤ 0.01) between DON contamination and a particular weather variable, with a darker colour indicating a higher value of the correlation coefficient. Tmean-daily mean temperature, RH-mean relative humidity, PREC-precipitation, VPD-vapour pressure deficit.
Figure 4.
Spearman´s rank correlation coefficient for deoxynivalenol (DON) contamination in Lithuania grown spring wheat at harvest and different weather factors estimated for 14-day moving windows during the growing season. Red indicates a positive correlation and blue a negative correlation (both p ≤ 0.01) between DON contamination and a particular weather variable, with a darker colour indicating a higher value of the correlation coefficient. Tmean-daily mean temperature, RH-mean relative humidity, PREC-precipitation, VPD-vapour pressure deficit.
Figure 5.
Spearman´s rank correlation coefficient for deoxynivalenol (DON) contamination in Polish winter wheat at harvest and different weather factors estimated for 14-day moving windows during the growing season. Red indicates a positive correlation and blue a negative correlation (both p ≤ 0.01) between DON contamination and a particular weather variable, with a darker colour indicating a higher value of the correlation coefficient. Tmean-daily mean temperature, PREC-precipitation.
Figure 5.
Spearman´s rank correlation coefficient for deoxynivalenol (DON) contamination in Polish winter wheat at harvest and different weather factors estimated for 14-day moving windows during the growing season. Red indicates a positive correlation and blue a negative correlation (both p ≤ 0.01) between DON contamination and a particular weather variable, with a darker colour indicating a higher value of the correlation coefficient. Tmean-daily mean temperature, PREC-precipitation.
Figure 6.
Variable importance in the Random Forest-based model for Sweden grown spring barley. PREC-precipitation, RH-mean relative humidity, Tmax-daily maximum temperature, Tmean-daily mean temperature, WS-mean wind speed, WD-wind direction. PREC_106-PREC 15.07–28.07, RH _092-RH 01.07–14.07, Tmax_099-Tmax 08.07–21.07, Tmax_106-Tmax 15.07–28.07, Tmean_099-Tmean 08.07–21.07, Tmean_106-Tmean 15.07–28.07, WD_001-WD 01.04–14.04, WD_057-WD 27.05–09.06, WS_008-WS 08.04–21.04, WS_106-WS 15.07–28.07.
Figure 6.
Variable importance in the Random Forest-based model for Sweden grown spring barley. PREC-precipitation, RH-mean relative humidity, Tmax-daily maximum temperature, Tmean-daily mean temperature, WS-mean wind speed, WD-wind direction. PREC_106-PREC 15.07–28.07, RH _092-RH 01.07–14.07, Tmax_099-Tmax 08.07–21.07, Tmax_106-Tmax 15.07–28.07, Tmean_099-Tmean 08.07–21.07, Tmean_106-Tmean 15.07–28.07, WD_001-WD 01.04–14.04, WD_057-WD 27.05–09.06, WS_008-WS 08.04–21.04, WS_106-WS 15.07–28.07.
Figure 7.
Distribution of the minimal depth of the variable and its mean in the Random Forest-based model for Sweden grown spring barley. PREC-precipitation, RH-mean relative humidity, Tmax-daily maximum temperature, Tmean-daily mean temperature, WS-mean wind speed, WD-wind direction. WD_057-WD 27.05–09.06, Tmax_099-Tmax 08.07–21.07, Tmean_099-Tmean 08.07–21.07, Tmean_106-Tmean 15.07–28.07, RH _092-RH 01.07–14.07, WD_036-WD 06.05–19.05, PREC_106-PREC 15.07–28.07, Tmax_106-Tmax 15.07–28.07, WS_106-WS 15.07–28.07, WS_008-WS 08.04–21.04.
Figure 7.
Distribution of the minimal depth of the variable and its mean in the Random Forest-based model for Sweden grown spring barley. PREC-precipitation, RH-mean relative humidity, Tmax-daily maximum temperature, Tmean-daily mean temperature, WS-mean wind speed, WD-wind direction. WD_057-WD 27.05–09.06, Tmax_099-Tmax 08.07–21.07, Tmean_099-Tmean 08.07–21.07, Tmean_106-Tmean 15.07–28.07, RH _092-RH 01.07–14.07, WD_036-WD 06.05–19.05, PREC_106-PREC 15.07–28.07, Tmax_106-Tmax 15.07–28.07, WS_106-WS 15.07–28.07, WS_008-WS 08.04–21.04.
Figure 8.
Variable importance in the Random Forest-based model for Sweden grown spring wheat. PREC-precipitation, RH-relative humidity, Tmax-daily maximum temperature, WS-wind speed, WD-wind direction, VPD-vapour pressure deficit. PREC_022-PREC 22.04–05.05, PREC_085-PREC 24.06–07.07, RH _029-RH 29.04–12.05, RH_036-RH 06.05–19.05, Tmax_099-Tmax 08.07–21.07, VPD_036-VPD 06.05–19.05, WS_008-WS 08.04–21.04, WS_050-WS 20.05–02.06, WS_057-WS 27.05–09.06, WS_092-WS 01.07–14.07.
Figure 8.
Variable importance in the Random Forest-based model for Sweden grown spring wheat. PREC-precipitation, RH-relative humidity, Tmax-daily maximum temperature, WS-wind speed, WD-wind direction, VPD-vapour pressure deficit. PREC_022-PREC 22.04–05.05, PREC_085-PREC 24.06–07.07, RH _029-RH 29.04–12.05, RH_036-RH 06.05–19.05, Tmax_099-Tmax 08.07–21.07, VPD_036-VPD 06.05–19.05, WS_008-WS 08.04–21.04, WS_050-WS 20.05–02.06, WS_057-WS 27.05–09.06, WS_092-WS 01.07–14.07.
Figure 9.
Distribution of the minimal depth of the variable and its mean in the Random Forest-based model for Sweden grown spring wheat. PREC-precipitation, RH-relative humidity, Tmax-daily maximum temperature, WS-wind speed, WD-wind direction, VPD-vapour pressure deficit. RH_036-RH 06.05–19.05, PREC_106-PREC 15.07–28.07, WS_050-WS 20.05–02.06, WD_099-WD 08.07–21.07, WS_057-WS 27.05–09.06, WS_092-WS 01.07–14.07, VPD_036-VPD 06.05–19.05, PREC_085-PREC 24.06–07.07, Tmax_099-Tmax 08.07–21.07, RH _001-RH 01.04–14.04.
Figure 9.
Distribution of the minimal depth of the variable and its mean in the Random Forest-based model for Sweden grown spring wheat. PREC-precipitation, RH-relative humidity, Tmax-daily maximum temperature, WS-wind speed, WD-wind direction, VPD-vapour pressure deficit. RH_036-RH 06.05–19.05, PREC_106-PREC 15.07–28.07, WS_050-WS 20.05–02.06, WD_099-WD 08.07–21.07, WS_057-WS 27.05–09.06, WS_092-WS 01.07–14.07, VPD_036-VPD 06.05–19.05, PREC_085-PREC 24.06–07.07, Tmax_099-Tmax 08.07–21.07, RH _001-RH 01.04–14.04.
Figure 10.
Variable importance in the Random Forest-based model for Lithuania grown spring wheat. PREC-precipitation, Tmean-daily mean temperature. PREC_022-PREC 22.04–05.05, Tmean_008-Tmean 08.04–21.04, Tmean_015-Tmean 15.04–28.04, Tmean_022-Tmean 22.04–05.05, Tmean_029-Tmean 29.04–12.05, Tmean_36-Tmean 06.05–19.05, Tmean_085-Tmean 24.06–07.07, Tmean_092-Tmean 01.07–14.07, Tmean_099-Tmean 08.07–21.07, Tmean_106-Tmean 15.07–28.07.
Figure 10.
Variable importance in the Random Forest-based model for Lithuania grown spring wheat. PREC-precipitation, Tmean-daily mean temperature. PREC_022-PREC 22.04–05.05, Tmean_008-Tmean 08.04–21.04, Tmean_015-Tmean 15.04–28.04, Tmean_022-Tmean 22.04–05.05, Tmean_029-Tmean 29.04–12.05, Tmean_36-Tmean 06.05–19.05, Tmean_085-Tmean 24.06–07.07, Tmean_092-Tmean 01.07–14.07, Tmean_099-Tmean 08.07–21.07, Tmean_106-Tmean 15.07–28.07.
Figure 11.
Distribution of the minimal depth of the variable and its mean in the Random Forest-based model for Lithuania grown spring wheat. Tmean-daily mean temperature, PREC-precipitation. Tmean_008-Tmean 08.04–21.04, Tmean_099-Tmean 08.07–21.07, Tmean_106-Tmean 15.07–28.07, Tmean_015-Tmean 15.04–28.04, Tmean_001-Tmean 01.04–14.04, PREC_022-PREC 22.04–05.05, Tmean_036-Tmean 06.05–19.05, Tmean_085-Tmean 24.06–07.07, PREC_071-PREC 10.06–23.06, Tmean_022-Tmean 22.04–05.05.
Figure 11.
Distribution of the minimal depth of the variable and its mean in the Random Forest-based model for Lithuania grown spring wheat. Tmean-daily mean temperature, PREC-precipitation. Tmean_008-Tmean 08.04–21.04, Tmean_099-Tmean 08.07–21.07, Tmean_106-Tmean 15.07–28.07, Tmean_015-Tmean 15.04–28.04, Tmean_001-Tmean 01.04–14.04, PREC_022-PREC 22.04–05.05, Tmean_036-Tmean 06.05–19.05, Tmean_085-Tmean 24.06–07.07, PREC_071-PREC 10.06–23.06, Tmean_022-Tmean 22.04–05.05.
Figure 12.
Variable importance in Random Forest-based model for Poland grown winter wheat. PREC-precipitation, Tmean-daily mean temperature. PREC_029-PREC 29.05–11.06, PREC_036-PREC 05.06–18.06, PREC_050-PREC 19.06–02.07, PREC_057-PREC 26.06–09.07, PREC_064-PREC 03.07–16.07, PREC_092-PREC 31.07–13.08, Tmean_015-Tmean 15.05–28.05, Tmean_057-Tmean 26.06–09.07, Tmean092-Tmean 31.07–13.08, Tmean_099-Tmean 08.08–21.08.
Figure 12.
Variable importance in Random Forest-based model for Poland grown winter wheat. PREC-precipitation, Tmean-daily mean temperature. PREC_029-PREC 29.05–11.06, PREC_036-PREC 05.06–18.06, PREC_050-PREC 19.06–02.07, PREC_057-PREC 26.06–09.07, PREC_064-PREC 03.07–16.07, PREC_092-PREC 31.07–13.08, Tmean_015-Tmean 15.05–28.05, Tmean_057-Tmean 26.06–09.07, Tmean092-Tmean 31.07–13.08, Tmean_099-Tmean 08.08–21.08.
Figure 13.
Distribution of the minimal depth of the variable and its mean in the Random Forest-based model for Poland grown winter wheat. PREC-precipitation, Tmean-daily mean temperature. PREC_057-PREC 26.06–09.07, Tmean_099-Tmean 08.08–21.08, PREC_092-PREC 31.07–13.08, PREC_064-PREC 03.07–16.07, Tmean_057-Tmean 26.06–09.07, PREC_050-PREC 19.06–02.07, PREC_036-PREC 05.06–18.06, Tmean_015-Tmean 15.05–28.05, PREC_029-PREC 29.05–11.06, Tmean092-Tmean 31.07–13.08.
Figure 13.
Distribution of the minimal depth of the variable and its mean in the Random Forest-based model for Poland grown winter wheat. PREC-precipitation, Tmean-daily mean temperature. PREC_057-PREC 26.06–09.07, Tmean_099-Tmean 08.08–21.08, PREC_092-PREC 31.07–13.08, PREC_064-PREC 03.07–16.07, Tmean_057-Tmean 26.06–09.07, PREC_050-PREC 19.06–02.07, PREC_036-PREC 05.06–18.06, Tmean_015-Tmean 15.05–28.05, PREC_029-PREC 29.05–11.06, Tmean092-Tmean 31.07–13.08.
Figure 14.
Location of field trials conducted in Sweden (SE), Poland (PL) and Lithuania (LT).
Figure 14.
Location of field trials conducted in Sweden (SE), Poland (PL) and Lithuania (LT).
Table 1.
Performance (accuracy, sensitivity and specificity) of the four models used to predict the risk of a deoxynivalenol (DON) contamination level >200 µg kg−1 in Swedish oats, based on the test dataset.
Table 1.
Performance (accuracy, sensitivity and specificity) of the four models used to predict the risk of a deoxynivalenol (DON) contamination level >200 µg kg−1 in Swedish oats, based on the test dataset.
Model | Accuracy (%) | Sensitivity 1 (%) | Specificity 2 (%) |
---|
Decision Tree | 68 | 71 | 67 |
Random Forest | 66 | 41 | 80 |
Support Vector Machine Linear | 70 | 75 | 67 |
Support Vector Machine Radial | 65 | 50 | 73 |
Table 2.
Performance (accuracy, sensitivity and specificity) of the four models used to predict the risk of a deoxynivalenol (DON) contamination level >200 µg kg−1 in Swedish spring barley, based on the test dataset.
Table 2.
Performance (accuracy, sensitivity and specificity) of the four models used to predict the risk of a deoxynivalenol (DON) contamination level >200 µg kg−1 in Swedish spring barley, based on the test dataset.
Model | Accuracy (%) | Sensitivity 1 (%) | Specificity 2 (%) |
---|
Decision Tree | 60 | 80 | 50 |
Random Forest | 77 | 63 | 85 |
Support Vector Machine Linear | 40 | 60 | 30 |
Support Vector Machine Radial | 73 | 80 | 70 |
Table 3.
Performance (accuracy, sensitivity and specificity) of the four models used to predict the risk of a deoxynivalenol (DON) contamination level >200 µg kg−1 in Swedish spring wheat, based on the test dataset.
Table 3.
Performance (accuracy, sensitivity and specificity) of the four models used to predict the risk of a deoxynivalenol (DON) contamination level >200 µg kg−1 in Swedish spring wheat, based on the test dataset.
Model | Accuracy (%) | Sensitivity 1 (%) | Specificity 2 (%) |
---|
Decision Tree | 65 | 33 | 100 |
Random Forest | 60 | 58 | 62 |
Support Vector Machine Linear | 60 | 60 | 60 |
Support Vector Machine Radial | 80 | 90 | 70 |
Table 4.
Performance (accuracy, sensitivity and specificity) of the four models used to predict the risk of a deoxynivalenol (DON) contamination level >1250 µg kg−1 in Lithuanian spring wheat, based on the test data set.
Table 4.
Performance (accuracy, sensitivity and specificity) of the four models used to predict the risk of a deoxynivalenol (DON) contamination level >1250 µg kg−1 in Lithuanian spring wheat, based on the test data set.
Model | Accuracy (%) | Sensitivity 1 (%) | Specificity 2 (%) |
---|
Decision Tree | 95 | 100 | 93 |
Random Forest | 84 | 74 | 88 |
Support Vector Machine Linear | 90 | 83 | 93 |
Support Vector Machine Radial | 90 | 83 | 93 |
Table 5.
Performance (accuracy, sensitivity and specificity) of the four models used to predict the risk of a deoxynivalenol (DON) contamination level >200 µg kg−1 in Polish winter wheat, based on the test data set.
Table 5.
Performance (accuracy, sensitivity and specificity) of the four models used to predict the risk of a deoxynivalenol (DON) contamination level >200 µg kg−1 in Polish winter wheat, based on the test data set.
Model | Accuracy (%) | Sensitivity 1 (%) | Specificity 2 (%) |
---|
Decision Tree | 75 | 59 | 83 |
Random Forest | 71 | 62 | 77 |
Support Vector Machine Linear | 69 | 81 | 63 |
Support Vector Machine Radial | 70 | 81 | 65 |
Table 6.
Summary of data on deoxynivalenol (DON) concentration in cereal grain from field trials conducted in Sweden, Poland and Lithuania.
Table 6.
Summary of data on deoxynivalenol (DON) concentration in cereal grain from field trials conducted in Sweden, Poland and Lithuania.
Species | Sweden | Poland | Lithuania |
---|
All | DON > 200 µg kg−1 Grains | All | DON > 200 µg kg−1 Grains | All | DON > 1250 µg kg−1 Grains |
---|
Oats | 80 | 29 | | | | |
Spring barley | 53 | 19 | | | | |
Spring wheat | 70 | 36 | | | 90 | 27 |
Winter wheat | | | 317 | 108 | | |
Table 7.
Growth stages with estimated dates and their respective 14-day windows for oats, spring wheat and spring barley grown in Sweden, spring wheat grown in Lithuania, and winter wheat grown in Poland.
Table 7.
Growth stages with estimated dates and their respective 14-day windows for oats, spring wheat and spring barley grown in Sweden, spring wheat grown in Lithuania, and winter wheat grown in Poland.
Country | Species | Zadoks Growth Scale | Date (dd.mm–dd.mm) | Data Frame |
---|
Sweden | oats | Germination GS0 | 27.04–30.05 | DF_022-DF_050 |
Seedling growth GS1 | 05.05–25.05 | DF_029-DF_050 |
Tillering GS2 | 11.05–12.06 | DF_036-DF_064 |
Stem elongation GS3 | 27.05–29.06 | DF_057-DF_078 |
Booting GS4 | 10.06–05.07 | DF_071-DF_085 |
Heading (Inflorescence emergence) GS5 | 20.06–13.07 | DF_078-DF_092 |
Flowering/Polination (Anthesis) GS6 | 27.06–15.07 | DF_085-DF_099 |
Milk development GS7 | 04.07–22.07 | DF_092-DF_099 |
Dough development GS8 | 08.07–23.07 | DF_092-DF_106 |
Ripening GS9 | 12.07–27.07 | DF_092-DF_106 |
spring wheat, spring barley | Germination GS0 | 20.04–17.05 | DF_015-DF_036 |
Seedling growth GS1 | 27.04–10.06 | DF_022-DF_057 |
Tillering GS2 | 10.05–18.06 | DF_036-DF_071 |
Stem elongation GS3 | 21.05–01.07 | DF_043-DF_085 |
Booting GS4 | 04.06–10.07 | DF_057-DF_092 |
Heading (Inflorescence emergence) GS5 | 11.06–17.07 | DF_064-DF_099 |
Flowering/Polination (Anthesis) GS6 | 11.06–23.07 | DF_078-DF_106 |
Milk development GS7 | 18.06–22.07 | DF_078-DF_106 |
Dough development GS8 | 29.06–27.07 | DF_085-DF_106 |
Ripening GS9 | 02.07–27.07 | DF_092-DF_106 |
Lithuania | spring wheat | Flowering, anthesis: Full flowering, 50% of anthers mature GS65 | 10.06–14.07 | DF_071-DF_092 |
Milk development GS7 | 01.07–28.07 | DF_092-DF_106 |
Dough development GS8 | 08.07–23.07 | DF_099-DF_106 |
Poland | winter wheat | Tillering GS2/Stem elongation GS3 | 01.05–14.05 | DF_001 |
Heading GS5/Flowering GS6 (beginning) | 15.05–04.06 | DF_015-DF_022 |
Flowering GS6/Milk development GS7/Dough development GS8 | 05.06–25.06 | DF_036-DF_043 |
Dough development GS8/Ripening GS9 | 19.06–16.07 | DF_050-DF_064 |
Harvest | 31.07–21.08 | DF_092-DF_099 |