SIMONTO-Lupin: an ontogenetic simulation model for lupin species (Lupinus angustifolius, L. luteus and L. albus)
SIMONTO-Lupine: Ein Ontogenesemodell für Lupine-Arten (Lupinus angustifolius, L. luteus und L. albus)
Journal für Kulturpflanzen, 63 (10). S. 333–339, 2011, ISSN 0027-7479, DOI: 10.5073/JfK.2011.10.03, Verlag Eugen Ulmer KG, Stuttgart
The development of the infection of Colletotrichum lupini is in addition to temperature and leaf wetness duration strongly dependent on the plant development stage (BBCH). For this reason a simple ontogenetic model (SIMONTO-Lupin) for the three lupin species Lupinus angustifolius, L. luteus and L. albus was developed and integrated into the decision support system SIMCOL which deals with the disease management of the pathosystem C. lupini – L. angustifolius. For the simulation of the crop stages a temperature development rate was modelled. Besides literature data about 180 data series from national variety and fungicide trials of blue, yellow and white lupin were used as data base for modelling SIMONTO-Lupin.
For the model validation 240 data sets of blue lupin of national variety and fungicide trials for the years 2006–2009 were available. These data sets were independent of the data sets used for model development.
The model SIMONTO-Lupin represents particularly the important period of flowering. The validation of the model (difference between observed and simulated growth stage max. 7 days) for the BBCH stage 61 (beginning flowering) showed 85% and for the stage 69 (full bloom) about 75% correct forecasts. Other validation criteria like the subjective validation showed about 88% correct forecasts. The validation by means of linear regression achieved, on average, a coefficient of determination of 0.984.
Key words: Lupin, temperature development rate, growth stages, ontogenetic model
Die Befallsentwicklung von Colletotrichum lupini ist neben der Temperatur und der Blattnässedauer stark abhängig vom aktuellen BBCH-Stadium des Lupinenbestands. Aus diesem Grund wurde für das Entscheidungshilfesystem SIMCOL, das sich mit dem Pathosystem C. lupini – Lupinus angustifolius beschäftigt, ein einfaches Ontogenesemodell (SIMONTO-Lupine) für die drei Lupinenarten L. angustifolius, L. luteus und L. albus entwickelt und integriert. Zur Berechnung der Bestandesentwicklung wurde eine temperaturabhängige Entwicklungsrate modelliert. Datengrundlage für die Modellierung von SIMONTO-Lupine waren neben Literaturdaten ca. 180 Datenreihen aus Landessortenversuchen, Beiz- und Fungizidversuchen von blauer, gelber und weißer Lupine. Für die Überprüfung des Ontogenesemodells SIMONTO-Lupine standen 240 Datenreihen der blauen Lupine von Landessortenversuchen, sowie Fungizid- und Beizversuchen aus den Jahren 2006 bis 2009 zur Verfügung. Diese Daten sind unabhängig von den zur Modellerstellung verwendeten Daten.
Das Modell SIMONTO-Lupine bildet insbesondere den wichtigen Zeitraum der Blüte optimal ab. Die Trefferquote (Differenz zwischen Bonitur und Simulation max. 7 Tage) des BBCH-Stadiums 61 lag bei 85% korrekten Prognosen und des Stadiums 69 bei 75% korrekten Prognosen. Bei der subjektiven Validierung wurde eine Trefferquote von 89% erzielt. Die Validierung mittels Linearer Regression erzielte im Mittel ein Bestimmtheitsmaß von 0,984.
Stichwörter: Lupine, Temperaturabhängige Entwicklungsrate, BBCH-Stadien, Ontogenesemodell
The simulation of the development stages of cultivated plants has different purposes. The prediction of the different stages of development of one culture can be used to verify the growing performance of the plants during the season or to identify periods for the correct application of pesticides. Certainly one of the most interesting aims is the integration of an ontogenetic model in a decision support system (DSS) for crop protection. An integrated ontogenetic model can for example be used to indicate periods of major plant susceptibility, to make plans for monitoring and survey activities or for the scheduling of treatments. The Central Institution for DSS in Crop Protection (ZEPP) has increased its experience in developing ontogenesis models with the creation of the SIMONTO (SIMulation ONTOgenesis) models for cereals and oilseed rape in the last years (Falke et al., 2006, 2008; Rossberg et al., 2005).
During the realisation of the DSS for the pathosystem C. lupini – narrow leaf (blue) lupin (L. angustifolius) – a simple ontogenetic model (SIMONTO-Lupin) was developed to indicate growth stages with high susceptibility.
Plant growth stages are usually described in agriculture numerical (Large, 1954; Haun, 1973; Zadoks et al., 1974). One of the most common scales used for both mono- and dicotyledonous plants is the BBCH decimal code growth scale developed in 1992 in Germany (Hack et al., 1992). The scale describes the plant growing from sowing (decimal code 0) to the natural death of the plant (decimal code 99).
From the studies by Dracup and Kirby (1996a) a BBCH growth scale for lupin was developed and described (Römer, 2007) and it is commonly used for all purposes regarding this culture. Aim of this work was to build a model able to simulate this growth scale depending on the seasonal time for the three main lupin species (L. angustifolius, L. luteus and L. albus).
Plant growing is strongly correlated with a complex of environmental factors. Some of them like the nitrogen and water availability are correlated with the soil characteristics. Other factors like temperature, rainfall and photoperiod are directly involved in the photosynthesis activity.
Since the aim of this model is to predict the occurrence of phenological stages, a simple approach using only the temperature as a limiting factor for the plant development was used for the modelling of SIMONTO-Lupin.
The simple model basis can be written as follow:
[1] BBCH x = (f) Σ LDR
Where:
BBCH x = Plant growth stage
Σ LDR = Sum of the lupin development rate from sowing to BBCH x
Data of lupin plant development depending on temperature were derived from literature (Dracup and Kirby, 1993; Farré et al., 2004) and transformed to obtain a development rate with minimum, optimum and maximum temperature at 0, 20 and 30°C respectively. LDR was obtained by interpolating this data using the Beta-Hau function (Hau, 1988):
[2] LDR (T) = LDRopt [(T–Tmin)/ (Topt–Tmin)]n(Topt–Tmin)/(Tmax–Topt) [(Tmax–T)/(Tmax–Topt)]n
Where:
LDR (T) = Lupin development rate depending on the temperature
LDRopt = Optimal lupin development rate
T = Temperature
Tmin = Minimum temperature for plant development
Topt = Optimum temperature for plant development
Tmax = Maximum temperature for plant development
n = Equation parameter
The estimated parameters of the equation are given in Table 1, the interpolated function in Fig. 1.
Table 1. Estimated parameters for the lupin development rate (LDR)
Parameter | Value | S.E. | p | r2 |
LDRopt | 1 | 0,0445 | < 0,001 | 0,95 |
Tmin | 0,00 | 0,9863 | 1,000 | |
Topt | 18,92 | 0,6502 | < 0,001 | |
Tmax | 30,00 | 1,8601 | < 0,001 | |
N | 1,27 | 0,5343 | < 0,001 |
Fig. 1. Lupin development rate (LDR) depending on the temperature. (
Transformed data from Dracup and Kirby, 1993; Farré et al., 2004. 
Interpolated Beta-Hau function).
Compared to literature data (Dracup and Kirby, 1993; Dracup and Kirby, 1996b; Farré et al., 2004) the result of the interpolation does not modify the minimum and maximum temperature for the plant growing and estimates an optimum temperature at about 19°C.
A data set of more than 900 BBCH field surveys (from 180 data series) collected by members of the Governmental Crop protection services and the seed breeders during the years 1996 to 2006 was used as basis for model development (Table 2).
Table 2. Data set used for the SIMONTO-Lupin development
Lupin Species | Region | Year | n. Trial/Year | n. of Observation |
L. angustifolius | Brandenburg | 2000–2003 | 1 | 16 |
Saxony-Anhalt | 2001–2004 | 3 | 417 | |
Thuringia | 2002 | 3 | 151 | |
West Pomerania | 2000–2007 | 3 | 18 | |
L. luteus | Brandenburg | 1996–2002 | 1 | 49 |
Saxony-Anhalt | 2001–2006 | 3 | 30 | |
West Pomerania | 1998–2007 | 2 | 60 | |
L. albus | Brandenburg | 1996–1997 | 1 | 11 |
Saxony | 1997 | 1 | 9 | |
Saxony-Anhalt | 2001–2003 | 2 | 105 | |
Thuringia | 2002 | 2 | 41 | |
West Pomerania | 1997 | 1 | 3 |
The nearest located weather station was assigned to each trial site and for each BBCH observation a sum of LDR beginning from sowing date was calculated (Table 3, Fig. 2).
Table 3. Sum of the lupin development rate (LDR) between BBCH 0 to 9, 61, 69 and 87 differentiated for L. angustifolius, L. luteus and L. albus
Lupin Species | BBCH | n | Minimum | Maximum | Median | Mean |
L. angustifolius | 9 | 112 | 3.55 | 8.94 | 7.04 | 6.70 |
61 | 126 | 35.17 | 53.05 | 41.17 | 41.35 | |
69 | 121 | 46.15 | 70.84 | 55.32 | 56.30 | |
87 | 112 | 65.67 | 109.83 | 91.29 | 90.83 | |
L. luteus | 9 | 16 | 3.29 | 11.93 | 7.38 | 7.56 |
61 | 9 | 45.47 | 59.18 | 51.04 | 51.36 | |
69 | 7 | 53.45 | 69.65 | 68.37 | 63.80 | |
87 | 2 | 94.68 | 98.15 | 96.41 | 96.41 | |
L. albus | 9 | 25 | 4.85 | 9.94 | 7.98 | 7.88 |
61 | 30 | 34.83 | 44.36 | 40.47 | 39.96 | |
69 | 30 | 51.52 | 78.49 | 67.73 | 65.55 | |
87 | 30 | 95.92 | 112.59 | 112.16 | 109.04 |
Fig. 2. Example of the variability of the sum of the lupin development rate (LDR) for BBCH growth stages 9, 61, 69 and 87 for L. angustifolius (data sets from Table 2).
The BBCH growth stages were modelled with a logistic interpolation (Fig. 3) as follows:
Fig. 3. Example of SIMONTO-Lupine: logistic interpolation between the BBCH recorded data and the sum of LDR from sowing for L. luteus ( BBCH recorded data, Interpolating logistic function).
[3] BBCH x = 1/1 + EXP [–(a + b Σ LDR)]
Where:
BBCH x = Predicted BBCH growth stage
a and b = Equation parameters
Σ LDR = Sum of the lupin development rate from sowing to BBCH x
The parameters of the equation for the three SIMONTO-Lupin models are summarised in Table 4.
Table 4. Parameters of the three SIMONTO-Lupin models
SIMONTO-Lupin | Parameters | p | r2 | |
L. angustifolius | a | –3,109 | < 0,001 | 0,88 |
b | 0,088 | < 0,001 | ||
L. luteus | a | –2,939 | < 0,001 | 0,81 |
b | 0,070 | < 0,001 | ||
L. albus | a | –2,765 | < 0,001 | 0,87 |
b | 0,075 | < 0,001 | ||
A comprehensive model validation was carried out for L. angustifolius (for the two other species more data were necessary).
For model validation three different methods were used (Racca et al., 2010).
• Subjective validation |
• Statistical validation with regression analysis |
• Evaluation of the deviation in days (difference between the date of the observed BBCH stage in field and the simulated date) |
For this type of validation the simulated progress of ontogenesis was compared manually with the monitored BBCH in field. If there was no or only minor differences between model and data collected, the prognosis was “correct” (Fig. 4A). If the monitored results (points) were below/above the simulated curve, the model had overestimated/underestimated (Fig. 4B+C). The degree of differences between model data and field data is subjective and is determined by the person who carries out the validation (Racca et al., 2010).
Fig. 4. L. angustifolius: Result of the subjective validation – Subjective comparison between the simulated (line) and the observed BBCH in field (points).
A: weather station Burkersdorf (Thuringia), year 2006 variety Idefix, classified as correct prognosis;
B: weather station Artern (Saxony-Anhalt), year 2006, variety Idefix, classified as overestimated prognosis;
C: weather station Dornburg (Thuringia), year 2008, variety Arabella, classified as underestimated prognosis.
In total 215 data sets were analysed by this validation method. 88.4% of the records were simulated correctly, 9.8% were underestimated and 1.9% overestimated.
For the statistical validation the model output (dependant variable) was compared to the field data (independent variable) with the help of a linear regression analysis. For each linear regression the null hypothesis (a = 0, slope b = 1) was verified by a Students t-test.
In total 229 data sets were analysed by this validation method. The average of the coefficient of determination was 0.984 and the slope b was in all cases not significant, which means that there are no significant differences between field data and simulation. All results of the statistical validation are summarized in Table 5.
Table 5. L. angustifolius: Results of regression analysis of the SIMONTO-Lupin model (n = 215)
Parameter | Minimum | Maximum | Median | Mean | Sign. a and b |
r2 | 0,954 | 0,996 | 0,985 | 0,984 | |
a | –1,093 | 3,050 | 0,845 | 0,881 | |
b | 0,874 | 1,094 | 0,975 | 0,974 | |
(a) Pr [t] | 0,266 | 0,998 | 0,833 | 0,807 | 100% s |
(b) Pr [t] | 0,000 | 0,016 | 0,001 | 0,001 | 100% ns |
Unfortunately, BBCH growth stages are not strictly arithmetically dependent. Some stages can appear very early in the season and stay constant for a long period of time. The arithmetic difference between two BBCH growth stages may be small but the difference in days between the appearance of two consecutive stages may be considerable. For this validation the difference in days between the ontogenesis in field and the simulated ontogenesis was calculated. If the difference between the both dates was ±7 days the prognosis was rated “correct”. Otherwise the prognosis was pointed out as “too early” or “too late” (Fig. 5).
Fig. 5. L. angustifolius: Evaluation of the deviation in days between the date of the BBCH stage observed in field and the simulated for the start (BBCH 61) and the end (BBCH 69) of the flowering (n = 229).
In total 229 data sets were analysed by this validation method for two BBCH-stages: Start of flowering (BBCH 61) and end of flowering (BBCH 69).
BBCH 61 achieved a hit rate of 86.0% correct forecasts and 14.0% of the dates were simulated too late. BBCH 69 was simulated correctly in 75.5% of all cases, 22.7% of the simulations were too early and 1.7% too late.
The validation of the developed model SIMONTO-Lupin gives very satisfactory results.
Most of the ontogenetic models (Mirschel et al., 1993; Wernecke and Claus, 1996; Rossi et al., 1997; Falke et al., 2006, 2008; Rossberg et al., 2005) do not consider soil, water and nutrient availability as a limiting factor and simulate the plant development only depending on the temperature and the day length. A correlation between the different growth stages and the sum of the temperature or combination with temperature and photoperiod was done.
The lupin growing areas in Germany are located mainly on the north-eastern part, the growing season is short with sowing time concentrated at the end of March and harvest at the beginning of August. On the assumption that day length during the growing season in this area varies from 13–14 hours in spring to 16–17 in summer, the photoperiod was also considered as a no-limiting factor. The temperature was the only parameter included in SIMONTO-Lupin model.
The appearance of phenological stages also helps to identify precisely the beginning and end of stages most sensitive to fungal diseases. In particular, the model has its practical application in conjunction with the model SIMCOL for the integrated plant disease management of the anthracnose (Racca and Tschöpe, 2010).
This work was part of the project “Developing of a decision support system for optimising the control strategy for the blue lupin anthracnose” and funded by UFOP (Union zur Förderung von Oel- und Proteinpflanzen e.V., Claire-Waldoff-Str. 7, 10117 Berlin).
Dracup, M., E.J.M. Kirby, 1993: Patterns of growth and development of leaves and internodes of narrow-leafed lupin. Field Crops Research 34, 209-225.
Dracup, M., E.J.M. Kirby, 1996a: Pod and seed growth and development of narrow-leafed lupin in a water limited mediterranean-type environment. Field Crops Research 48, 209-222.
Dracup, M., E.J.M. Kirby, 1996b: Lupin development guide. University of Western Australia Press, Nedlands, Western Australia 6907.
Falke, K., E. Jörg, D. Rossberg, 2006: Ursachen für Abweichungen zwischen bonitierten und simulierten Entwicklungsverläufen von SIMONTO für Winterraps und Wintergetreide. Mitteilungen aus der Biologischen Bundesanstalt für Land- und Forstwirtschaft 400, 282.
Falke, K., E. Jörg, D. Rossberg, 2008: Erhebung von BBCH-Stadien bei Getreide und Winterraps. Mitteilungen aus dem Julius Kühn-Institut 417, 413-414.
Farré, I., M.J. Robertson, S. Asseng, R.J. French, M. Dracup, 2004: Simulating lupin development, growth, and yield in a Mediterranean environment. Australian Journal of Agricultural Research 55, 863-877.
Hack, H., H. Bleiholder, L. Buhr et al., 1992: Einheitliche Codierung der phänologischen Entwicklungsstadien mono- und dikotyler Pflanzen. Erweiterte BBCH-Skala. Nachrichtenblatt des Deutschen Pflanzenschutzdienstes 44, 265-270.
Hau, B., 1988: Ein erweitertes analytisches Modell für Epidemien von Pflanzenkrankheiten. Habilitationsschrift, Julius-Liebig Universität Giessen.
Haun, J.R., 1973: Visual quantification of wheat development. Agronomy Journal 65, 116-119.
Large, E.C., 1954: Growth stages in cereals: illustration of the Feekes scale. Plant Pathology 3, 128-29.
Mirschel, W., A. Schultz, K.O. Wenkel, 1993: Vergleich der Winterweizenmodelle AGROSIM-WHEAT und CERES-Wheat. Berichte der Gesellschaft für Informatik in der Land-, Forst- und Ernährungswirtschaft 5, 29-34.
Racca, P., T. Zeuner, J. Jung, B. Kleinhenz, 2010: Model Validation and Use of Geographic Information Systems in Crop Protection Warning Service. In: E.-C. Oerke et al. (Eds.), Precision Crop Protection – the Challenge and Use of Heterogeneity. Dordrecht [u.a.], Springer, 259-276.
Racca, P., B. Tschöpe, 2010: SIMCOL1+3. Erarbeitung eines Entscheidungshilfesystems zur Optimierung der Bekämpfungsstrategie für die Anthraknose der Blauen Lupine. Julius-Kühn-Archiv 428, 431.
Römer, P., 2007: Lupinen – Verwertung und Anbau. Rastatt, Gesellschaft zur Förderung der Lupine e.V., 36 S.
Rossberg, D., E. Jörg, K. Falke, 2005: SIMONTO – ein neues Ontogenesemodell für Wintergetreide und Winterraps. Nachrichtenblatt des Deutschen Pflanzenschutzdienstes 57, 74-80.
Rossi, V., P. Racca, S. Giosué, D. Pancaldi, I. Alberti, 1997: A simulation model for the development of brown rust epidemics in winter wheat. European Journal of Plant Pathology 103, 453-465.
Wernecke, P., S. Claus, 1996: Modelle der Ontogenese für die Kulturarten Winterweizen, Wintergerste und Winterraps. In: Mühle, H., S. Claus (Eds.), Reaktionsverhalten von agrarischen Ökosystemen homogener Areale, 105-120.
Zadoks, J.C., T.T. Chang, C.F. Konzak, 1974: A decimal code for the growth stages of cereals. Weed Research 14, 415-421.
Suchen