JXB Advance Access originally published online on April 8, 2004
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Experimental Botany, Vol. 55, No. 399, pp. 1079-1094, May 1, 2004
© 2004 Oxford University Press
Plants and the Environment |
Analysis and modelling of effects of leaf rust and Septoria tritici blotch on wheat growth
Received 5 November 2003; Accepted 20 January 2004
1 Environnement et Grandes Cultures, INRA, F-78850 Thiverval Grignon, France
2 Laboratoire Pathologie Végétale, INRA, F-78850 Thiverval Grignon, France
3 Laboratoire Mathématique, Informatique et Génome, INRA, F-78352, Jouy-en-Josas, France
4 Laboratoire Statistique et Génome, CNRS, Tour Evry2, F-91034 Evry, France
* Present address and to whom correspondence should be sent: Rothamsted Research, Harpenden, Herts. AL5 2JQ, UK. E-mail: corinne.robert{at}bbsrc.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionTheoryMaterial and methodsResultsDiscussionReferences |
|---|
A model to predict Septoria tritici blotch (STB) and leaf rust effects on wheat growth was constructed and evaluated in two steps. At the leaf scale, Bastiaans’ approach that predicts the relative photosynthesis of a wheat leaf infected with a single disease, was extended to the case of two diseases, one biotrophic and one necrotrophic by considering the leaf rust–STB complex. A glasshouse experiment with flag leaves inoculated either singly with one disease or with two diseases combined was performed to check the leaf damage model. No interaction of the two diseases on photosynthesis loss was observed when they occurred simultaneously on the same leaf. In a second step, the single-leaf model was extended to the canopy scale to model the effects of the leaf rust–STB complex on the growth of a wheat crop. The model predicts the effects of disease on the growth of an affected crop relative to the growth of a healthy crop. The canopy model accounted for different contributions to photosynthetic activity of leaf layers, derived from their position in the canopy and their natural leaf senescence. Treatments differing in nitrogen fertilization, microclimatic conditions, and wheat cultivars were implemented in a field experiment to evaluate the model. The model accurately estimated the effect of disease on crop growth for each cultivar, with differences from experimental values lower than 10%, which suggests that this model is well suited to aid an understanding of disease effects on plant growth. A reduction in green leaf area was the main effect of disease in these field experiments and STB accounted for more than 70% of the reduction in plant growth. Simulations suggested that the production of rust spores may result in a loss of biomass from diseased crops and that stem photosynthesis may need to be considered in modelling diseased crop growth.
Key words: Biomass production, crop damage, crop loss, multiple pathosystem, leaf photosynthesis, Mycosphaerella graminicola, Puccinia triticina, wheat modelling.
|
Introduction |
|---|
|
TopAbstractIntroductionTheoryMaterial and methodsResultsDiscussionReferences |
|---|
Air-borne diseases, and particularly leaf rust (Puccinia triticina) and Septoria tritici blotch (Mycosphaerella graminicola), are a major cause of yield loss in wheat (Triticum aestivum). Crop protection in Europe relies largely on preventive fungicide applications and small-grain cereals are systematically treated with two or three foliar treatments. But environmental concerns and changes in the cost/price ratio for wheat are likely to increase the demand for more accurate identification of spraying needs. Decision systems based on predicted yield loss rather than epidemic threshold must be developed to meet this demand.
Disease stress may reduce yield by altering the formation of any of the yield components, depending on the crop developmental stage at which infection occurs (Madden and Nutter, 1995) and on the duration and severity of the epidemic (Madeira and Clark, 1991). Diseases that cause foliar epidemics reduce net photosynthesis and the first step in modelling the damage is to relate the loss in photosynthetic capacity of the plants to the intensity and distribution of the disease (Bastiaans, 1991).
Foliar pathogens affect leaf photosynthetic activity directly by reducing green leaf area. When photosynthesis is unaffected in the remaining green tissue of diseased leaves, the plant can be considered to accumulate dry matter in proportion to the amount of photosynthetically active radiation (PAR) that the green (healthy) tissue intercepts. In that case, predicting the growth of the diseased crop requires only an estimate of the reduction in green leaf area (Van Oijen, 1990; Waggoner and Berger, 1987). However, the pathogen may also impair the photosynthetic activity of the apparently healthy parts of the leaves (Rabbinge et al., 1985). Bastiaans (1991) proposed the concept of ‘virtual lesion’ to relate disease severity to the loss of leaf photosynthetic activity and thus account for the possible effect of the disease on the apparently healthy tissue. The virtual lesion is defined as the proportion of leaf tissue, in addition to the visual lesion (i.e. proportion of leaf tissue with visible symptoms), in which photosynthesis is assumed to be zero. A single parameter ß accounts for the effect of the virtual lesion on host photosynthesis. Bastiaans’ model has been widely used to estimate the net photosynthesis of leaves affected by a single disease (Bassanezi et al., 2001; Bastiaans, 1991; Garry et al., 1998; Lopes and Berger, 2001) and has been extended to two diseases occurring on the same leaf, with emphasis on the interaction between the two pathogens (Lopes and Berger, 2001).
In this study, a model was established to simulate effects of the leaf rust–Septoria tritici blotch (STB) complex on winter wheat growth. The model predicts the decrease in diseased crop growth relative to a control healthy crop. The model was developed and validated in two steps. First, the consequences of leaf rust and STB on leaf photosynthetic competence were quantified. For this, a new approach was proposed to extend Bastiaans’ model (Bastiaans, 1991) to the case of two diseases, one biotrophic and one necrotrophic. Since no information was available about possible interactions between leaf rust and STB, the single-leaf model was evaluated with data from a glasshouse experiment in which adult flag leaves were inoculated with M. graminicola, with P. triticina or co-inoculated with both pathogens. In a second step, the model was extended to the canopy level by computing the impact of the disease on photosynthetic activity for each canopy layer. Finally, the model’s outputs were compared with field data of an experiment including different cultivars, nitrogen fertilization treatments, and microclimatic conditions.
This modelling approach was used to identify gaps in knowledge concerning the impact of disease on crop growth and as a basis for discussing a more general approach to the situation of a complex system including a biotrophic and a necrotrophic pathogen.
|
Theory |
|---|
|
TopAbstractIntroductionTheoryMaterial and methodsResultsDiscussionReferences |
|---|
A model was developed to estimate the effects of two diseases, one necrotrophic (STB) and one biotrophic (leaf rust), on plant growth. The vertical distribution of the diseases in the canopy was taken into account, but the diseases were considered to be horizontally homogeneous. The model predicts the decrease in diseased crop growth relative to a control healthy crop. It makes the assumption that the difference in the growth of diseased and control crops is due solely to a reduction in net photosynthesis in the diseased leaves.
Model at the leaf scale
Bastiaans (1991) formalized the relationship between net photosynthesis and the proportion of diseased leaf area for a single disease, and he introduced the concept of ‘virtual lesion’ to account for the reduction in photosynthesis in the green parts of diseased leaves. His model related the net photosynthesis rate (Px) of a diseased leaf to that of a healthy leaf (P0) by:
Px=P0(1–x)ß(1)
where x is the proportion of the leaf area covered by visible lesions and ß is defined as the ratio between the sizes of the ‘virtual’ and the ‘visual’ lesions. The value of ß indicates whether the effect of disease on photosynthesis is higher (ß >1), lower (ß <1) or equal (ß=1) to that accounted for by the observed diseased area.
Bastiaans’ demonstration is based on the hypotheses that lesions are independently distributed on the leaves and that the virtual size of an individual lesion is less than 20% of the whole leaf surface area (Bastiaans, 1991). To establish equation 1, Bastiaans uses equation 2 (Justesen and Tammes, 1960) to compute the expected proportion of visible healthy area (1–x) on a leaf infected by n lesions of individual visible size 
:
1–x=(1–)n(2)
where is the proportion of leaf area occupied by a single lesion. Equation 2 is valid for lesions that are independently distributed over the leaf, able to overlap, and of similar size.
Bastiaans’ model was then extended to the case of two diseases: STB and leaf rust. The following definitions were used: x=proportion of leaf area covered by visible lesions; y=proportion of leaf area covered by virtual lesions; =proportion of leaf area occupied by a single visible lesion; ß=ratio between virtual and visual lesion size; n=number of lesions in the leaf. STB is referred to as disease 1 and leaf rust as disease 2.
In this system, STB is necrotrophic and necroses induced by STB can grow over leaf rust lesions, whereas leaf rust develops only on photosynthetically active tissue. It is impossible to identify rust lesions clearly under STB necroses in disease assessments. Thus, in the glasshouse and field experiments, STB was recorded as the proportion of the necrotic area (sporulating and non-sporulating areas), including areas where STB necroses had covered over rust lesions, whereas rust was recorded as the area occupied by sporulating lesions on green tissue. A leaf with n1 STB lesions and n2 leaf rust lesions (including those covered by STB) was then considered.
Considering that the n1 STB lesions and the n2 leaf rust lesions are each independently distributed over the leaf, the proportion (1–y) of leaf area that is not affected by virtual lesions can be calculated, according to Bastiaans (1991):
1–y=(1–1ß1)n1x(1–2ß2)n2(3)
Parameters n1, n2, 1, and 2 were then replaced by measurable values. Based on equation 2, the proportion of leaf area that is free from visible STB lesions is:
1–x1=(1–1)n1(4)
The symmetrical formula would not be valid for leaf rust since the rust lesions that were covered by STB were not taken into account in the disease assessments. Instead, the leaf area x'2 covered by sporulating leaf rust lesions has to be used. Calculated from the disease severity assessments, the proportion of leaf area that is visually free of disease is 1–x1–x'2. Considering again that the lesions were established independently, this can be evaluated by:
1–x1–x'2=(1–1)n1x(1–2)n2(5)
Parameters x1 and x'2 are the disease severity values as measured in the experiments for STB and leaf rust, respectively. From equations 4 and 5, n1 and n2 can be calculated as:
n1=ln(1–x1)/ln(1–1)
n2=ln[1–x'2/(1–x1)]/ln(1–2)
Replacing n1 and n2 in equation 3 gives:
1–y=exp{[ln(1–1ß1)/ln(1–1)]xln(1–x1)}
xexp{[ln(1–2ß2)/ln(1–2)]xln(1–x'2/(1–x1))}
When ißi (i=1 or 2) and i are small compared to 1, ln(1–ißi) can be approximated by –ißi and ln(1–i) by –i and therefore:
1–y
(1–x1)ß1x[1–x'2/(1–x1)]ß2(6)
The proportion of leaf area free of virtual lesions (1–y) is an estimation of the net photosynthetic rate of the diseased leaf, relative to a healthy leaf and as such:
Px1x'2/P0(1–x1)ß1x[1–x'2/(1–x1)]ß2(7)
Equation 7 relates photosynthesis reduction to disease severity through two parameters, ß1 and ß2, which are assumed to be identical, whether the pathogens develop alone or are associated on the same leaf.
Integration at the canopy scale
This study’s approach was to calculate the reduction in growth in a diseased crop, relative to a healthy crop, from the disease distribution in the different leaf layers of the canopy and the contribution to photosynthesis of each of these layers. The biomass in the diseased crop was then computed relative to measured biomass in a healthy control crop. The model makes the assumption that the difference in the growth of diseased and control crops is due solely to a reduction in net photosynthesis in the diseased leaves.
The relative contributions (C) of the different layers to crop photosynthesis were first calculated for a healthy canopy, considering that they were proportional to light interception and green area. For this, the plant was divided into ten parts: stem between ear and flag leaf (S1, peduncle), flag leaf blade (L1), stem between L1 and L2 (S2=flag leaf sheath), leaf blade L2, etc. Ear photosynthesis is generally considered to be almost completely offset by ear respiration (Araus et al., 1993; Gebbing and Schnyder, 2001) and was thus neglected. The fraction of the radiation (Ri) intercepted by each canopy part was calculated using Beer’s law (Beasse et al., 2000; Varlet-Grancher et al., 1989) applied to the stem and leaf area indexes. The light extinction coefficient (k) was set at 0.66 for the leaves (Varlet-Grancher et al., 1989) and to 0.45 for the stem (Ross, 1981). Considering that the stem is not affected by the diseases, the amount of radiation intercepted by the different stem sections was pooled and the overall radiation intercepted by the plant (RT) was the sum of the radiation intercepted by the stem (RS) and by the various leaves (Ri):
RT=RS+
iRi(8)
The proportion of dead tissue due to natural senescence (S) was then removed to calculate the relative contributions of each leaf layer (Ci) and of the stem (CS) to crop photosynthesis:
Ci(t)=Rix(1–Si(t))/{j[Rjx(1–Sj(t))]+RS}(9a)
CS(t)=RS/{j[Rjx(1–Sj(t))]+RS}(9b)
where Si(t) and Sj(t) represents the proportion of dead tissue due to natural senescence in leaf layers i and j.
It was considered that the relative contributions (Cs and Ci) of the different layers to crop photosynthesis, as calculated for a healthy canopy, remained valid in the diseased canopy. This means that disease would not affect the host structure (mainly leaf size) and thus the vertical distribution of light interception. The photosynthetic activity of the diseased canopy relative to the healthy canopy, Px/P0, was then estimated from the leaf layer contributions (Ci) and their relative photosynthesis rates Px,i/P0,i as:
Px(t)/P0(t)=CS(t)+iCi(t)x[Px,i(t)/P0,i(t)](10a)
Px,i(t)/P0,i(t) was calculated using equation 7 applied to each canopy layer i as:
Px,i(t)/P0,i(t)=(1–xSTB(i,t))ßSTBx[1–xrust(i,t)/(1–xSTB(i,t))]ßrust(10b)
where xSTB(i,t) is the proportion of STB in leaf layer i, at time t, relative to the green leaf area in the healthy leaf layer and xrust(i,t) is the proportion of sporulating rust lesions in leaf layer i, at time t, relative to the green leaf area in the healthy leaf layer.
Finally, diseased crop growth (
DWx) relative to healthy crop growth (DW0) was estimated by:
DWx(t)=DW0(t)[Px(t)/P0(t)](11)
where DW is the above-ground dry weight (g per stem)
Growth of diseased crops was assumed to be proportional to ß-weighted light interception and to control crop growth. The simulations used four input variables: (1) stem and leaf area in each canopy layer; (2) the proportion of leaf area covered by STB, xSTB(i,t), and leaf rust, xrust(i,t) in each leaf layer i and at each date t between –250 dd and +900 dd from flowering; (3) plant growth in control (fungicide-treated) plots, DW0(t), at each date t between –250 dd and +900 dd from flowering; and (4) proportion of necrotic tissue resulting from natural senescence, Si(t), as measured in control plots in each leaf layer i and at each date t between –250 dd and +900 dd from flowering. Between two successive sampling dates, plant growth and natural senescence in control treated plots, as well as the proportions of leaf area occupied by STB and leaf rust in each layer of the diseased plots, were estimated daily from the measured data by linear interpolation.
|
Material and methods |
|---|
|
TopAbstractIntroductionTheoryMaterial and methodsResultsDiscussionReferences |
|---|
Overview
Two different experiments were performed to evaluate the model. The first experiment was conducted in a glasshouse to quantify the effect of leaf rust and STB on the photosynthesis of individual wheat leaves. This allowed the ß parameters to be estimated for each pathogen and to check equation 7. The second experiment was conducted in the field to evaluate the model performance in predicting the consequences of the diseases on crop growth.
Glasshouse experiment.
Plants of cv. Soissons were left for 8 weeks under alternating 16 h light periods (350 µE m–2 s–1) at 8 °C and 8 h dark periods at 0 °C for vernalization. The plants were then transferred to square pots (1.1 l) filled with commercial compost (peat substrate, Gebr. Brill Substrate, Germany) and placed in a glasshouse. Light was complemented by sodium lamps from 08.00 h to 20.00 h and the temperature ranged between 8 °C and 15 °C during the night and 15–30 °C during the day. Plants were watered daily and were treated against powdery mildew (Ethyrimol, 2 ml l–1) 3 weeks before inoculation. Plants received a standard dose of 7 g per pot of fertilizer (Osmocote 10N+11P+18K). The nitrogen content of the flag leaves was measured the day before inoculation and at the end of the experiment (Robert et al., 2004). All secondary tillers were cut before inoculation.
When plants were at the heading stage, flag leaves were inoculated either with leaf rust (treatment R), with STB (treatment S), or with both pathogens (treatment R+S). Leaf rust inoculations were performed in a settling tower (Eyal et al., 1968) with isolate B9384-1C1, previously increased on seedlings of the susceptible cultivar Michigan. A wide range of lesion densities was obtained by inoculating flag leaves with 10 mg of a mixture of leaf rust uredospores and talc, in which the spore content was 9, 5, 3, 2, 1, or 0.5 mg (11 plants inoculated per dose). Infected plants were incubated for 24 h under 100% relative humidity at 17 °C and then replaced in the glasshouse until the end of the experiment.
STB inoculations were performed with inoculum derived from a single pycnidiospore isolate of M. graminicola obtained from a naturally infected Soissons field and increased on PDA medium in a growth chamber at 18 °C. Spores were collected from 3-d-old cultures in sterile distilled water, the suspension was adjusted to 108 spores ml–1 with a haemacytometer, and one drop of surfactant (Tween 20) was added before inoculation. The spore suspension was applied to restricted areas of the leaves with a soft paintbrush. The plants were then left for 5 days (d) for incubation in a walk-in growth chamber with a 16 h light period (350 µE m–2 s–1) at 18 °C and an 8 h dark period at 17 °C, at 100% relative humidity. Plants were inoculated with STB the day after the inoculation with leaf rust.
Upon the appearance of the first symptoms, 16 plants were selected for treatment R so as to obtain a wide range of rust lesion densities. Ten plants were selected for treatment S and 15 plants for treatment R+S. For each treatment, a proportion of the plants were given the same treatment as the others but with no inoculation and were used as control healthy plants. The rust lesions started to sporulate 11 d after inoculation and the STB chlorosis appeared 9 d after inoculation.
Net photosynthesis was assessed 27 d and 32 d after leaf rust inoculation (dai) in treatments R and R+S, and 27 d after STB inoculation (28 dai) in treatment S. Gas exchange was measured in the control leaves on the same dates. On two of the healthy leaves, gas exchange was measured at four dates and at different positions along the leaf length. The net rate of photosynthesis was found to be independent of the position along the leaf (P=0.106) and there was no interaction between position and date (P=0.123). The location of the measurement along the leaf was therefore ignored when estimating the net photosynthesis rate on healthy leaves.
Assessments of net photosynthetic rate were made with a portable photosynthesis system (LI-6200; Li-Cor, Lincoln, USA) mounted with a red LED light source (6400-02, Li-Cor), at light saturation (1500 µmol photon m–2 s–1). The assessments were made on leaf sections with an area of 6 cm2 (3x2 cm). When disease symptoms appeared on the leaves, the precise location of the measurements on each leaf was chosen. One or two assessments per leaf were selected so that a large range of disease severity was represented in each treatment. The leaf sections (6 cm2) on which the measurements were performed were tagged with a permanent marker and all further assessments were done at the same places.
For leaves of treatments R and R+S, digital pictures were taken at 27 and 32 dai to estimate accurately the diseased surface area on the leaf segments analysed for gas exchanges. The proportion of leaf area occupied by sporulating lesions for leaf rust (treatments R and R+S) and by STB necrosis (treatment R+S) was measured by image analysis (Optimas, Media Cybernetics, Silver Spring, USA). At these dates, leaf rust lesions were fully sporulating and necrosis was only due to STB infections. In co-inoculated leaves, because of STB lesion growth, a part of the leaf rust lesions overlapped by STB ceased to sporulate. These lesions were not recorded in the rust assessment. In treatment S, the necrotic leaf area on the leaf segment assessed for gas exchanges was traced onto a transparency film, which allowed measurement of the local disease severity as the proportion of leaf area covered by STB
The parameters of equation 1 (see ‘Theory’ section), i.e. Bastiaans’s model (Bastiaans, 1991), were estimated for leaf rust (ßR) and STB (ßS) from the data of treatments R and S, respectively. The parameters of equation 7 were estimated from treatment R+S for leaf rust (ßR+) in the presence of STB and for STB (ßS+) in the presence of leaf rust. The values of ßS+ and ßR+ were compared with ßS and ßR as proposed by Lopes and Berger (2001).
The parameters were estimated using non-linear regression analysis. The procedure used was PROC NLIN of the software SAS (SAS 6.12, SAS institute, Cary, USA). For treatments R and R+S, the nested model method was used to compare the effect of disease on gas exchange according to the assessment date (27 dai and 32 dai). Two alternative models were compared: Y=(1–X)(ß+i) and Y=(1–X)ß, for two dates (i=[1,2]), using the lack-of-fit F-test (Weisberg, 1985).
Field experiment
A field experiment with eight treatments was conducted using cvs Soissons and Recital, two widespread wheat cultivars in France and both susceptible to leaf rust and STB. No artificial inoculation was made, but the treatments were designed to ensure variation in disease development and in yield response to disease. The treatments were cv. Soissons reference treatment, water-sprayed plots of cv. Soissons, Soissons with low nitrogen inputs, and cv. Recital, all either with or without a fungicide application to control disease. The reference treatment reflected the climatic potential for disease development in the year 1999. Water favours STB development because infection requires 12–15 h of saturating humidity (Magboul et al., 1992) and because spores are disseminated by splashing. The water-spray treatment was done to enhance the STB epidemic. The canopy was sprayed with a micro-aspersion system every evening for 30–60 min (8 mm h–1), depending on the natural rainfall. A late nitrogen application is frequently given in France to increase the grain protein content (Girard, 1997) and it may change the susceptibility of the crop to disease (Leitch and Jenkins, 1995; Tiedemann, 1996). For the low-nitrogen treatment, the last nitrogen application was omitted, which resulted in a lower leaf nitrogen content around flowering, but did not cause any change in the canopy structure. In the ‘Recital’ treatment, epidemics and their effects on growth and yield for two different cultivars, Recital and Soissons were compared.
Plots were sown on 22 October. There were no significant differences in plant density between treatments or between diseased and control plots (250 plants m–2). The experiment was laid out as a randomized block design with three replicates per treatment. Each plot was 30x1.4 m. The plots were given 240 kg N ha–1 as three applications: 60 kg ha–1 on 11 March, 100 kg ha–1 on 6 April (stem elongation), and 80 kg ha–1 on 5 May (heading), except for the low nitrogen trial, where the last application was omitted. Soil water potential was evaluated daily by tensiometers and plots were irrigated to prevent water stress. All treatments included an early fungicide spray (Unix, 0.6 l ha–1) on 7 April, to limit soil-borne diseases. The fungicide-treated control plots were also sprayed on 16 April (Opus, 1.0 l ha–1), on 5 May (Opus, 1.0 l ha–1), and on 16 June (Alto, 1.0 l ha–1).
Disease and crop growth assessments: Leaf rust and STB were assessed weekly from 15 April (530 degree-days (dd) before flowering) to 20 June (330 dd after flowering). Fifteen main stems were examined every 5 m along each plot (five observation points per plot). Diseases were recorded on each of the five upper leaf layers of the canopy. The severity of STB was recorded directly as a percentage of the leaf area. Since STB is a necrotrophic pathogen, it is difficult to distinguish disease symptoms from senescence and STB was thus recorded as the overall necrotic area on the leaves (including sporulating and no-sporulating necrotic areas). Leaf rust, as a strictly biotrophic pathogen, is unable to sporulate on necrotic tissue, and was thus recorded on photosynthetically active tissues only. The severity of leaf rust was estimated as the area occupied by the sporulating lesions relative to the whole leaf surface area. The standard severity scale of Peterson (Peterson et al., 1948) was used to assess leaf rust severity.
Crop dry weight was assessed from 6 April to physiological maturity (15 July). Two plant samples (0.18 m2 each) were collected weekly from each plot. On each sampling date, main stems were separated from tillers and then both were separately taken to dry weight for vegetative parts and spikes. The mean fresh weight of the main stems was measured and a subsample of 15 main stems was selected. The stems selected had a fresh weight within ±10% of the sample mean value. Each of the five upper leaves, stems, and ears were separated. The five upper leaves were scanned to measure the total, green, and necrotic areas by image analysis. The software used was Optimas (Media Cybernetics, Silver Spring, MD 20910, USA). The dry weights and nitrogen contents (Dumas, 1831) of the four upper leaves, ears, and stems were measured. The number of grains per spike, the mean grain weight and the grain and leaf nitrogen contents were measured each week after flowering.
The ratio between mean weight of the tiller and the main stem was calculated at each assessment date and was found constant during the grain-filling period and similar in the diseased and control plots (data not shown). This ratio was 0.79, 0.82, 0.78, and 0.87 in the Soissons reference, Soissons low nitrogen, Soissons water-sprayed, and Recital treatments, respectively. The tiller leaf size was estimated from the measured main stem leaf area with respect to this ratio. This allowed the LAI of the whole canopy (tillers and main culms) to be estimated.
Analyses of variance were conducted using the SAS system (SAS Institute, Cary, NC) and statistically significant differences in leaf area, above-ground dry weight, number of spikes m–2, yield, number of grains, and mean grain weight among treatments in control and diseased plots were determined with the Tukey test.
Simulations: The model (equation 11) was evaluated with the field data. The contributions of the different canopy layers to photosynthesis (Ci in equation 9) were calculated according to Beer’s law (Varlet-Grancher et al., 1989) from the stem and leaf area indexes in control healthy plots, and corrected by the natural senescence. Between two successive sampling dates, plant growth (above-ground dry weight) and natural senescence in control plots, as well as the proportions of leaf area occupied by STB and leaf rust in each layer of the diseased plots, were estimated daily from the measured data by linear interpolation.
Measured severity of STB and leaf rust was corrected to take into account the leaf natural senescence (apical necrosis) estimated in the control plots. Leaf area actually occupied by STB was obtained by removing the necrotic area due to natural senescence. The proportion of leaf area covered by leaf rust (xrust in equation 10b), and STB (xSTB in equation 10b) was expressed relative to the green area of the corresponding healthy layer.
Simulations started when all the leaves were fully developed (250 dd before flowering). Time was in degree-days (dd, 0 °C basis) after flowering. Simulations of the diseased crop growth were performed with the values of the parameter ß obtained in the glasshouse experiment (equation 7): ßR+=2 and ßS+=1.4. The growth of a crop affected by both STB and leaf rust was simulated, as well as the effect of STB alone. Moreover, diseased crop growth was calculated both with and without taking stem photosynthesis into account.
To evaluate the sensitivity of the model to parameters ßR and ßS, simulations were performed with ßR=2 and ßS=1 or 3 and reciprocally with ßS=1.4 and ßR=1 or 3. The case ßR=ßS=1 was also tested to evaluate the effect of the disease on the apparently healthy host tissue.
|
Results |
|---|
|
TopAbstractIntroductionTheoryMaterial and methodsResultsDiscussionReferences |
|---|
Glasshouse experiment
Before inoculation, the average nitrogen content (in mg N cm–2) of the leaves was 0.19 (
=0.01) in treatment R, 0.21 (=0.02) in treatment S, and 0.19 (=0.02) in treatment R+S. At the end of the experiment, the average N content was 0.18 (=0.03) in treatment R, 0.12 (=0.02) in treatment S, and 0.16 (=0.02) in treatment R+S. Disease data are given for the leaf segments where gas photosynthesis was measured (Fig. 1). In treatment R, rust lesion density ranged from 2.8 to 34.8 lesions cm–2 and rust severity, expressed as the proportion of diseased area, ranged between 3% and 27% at 27 dai (days after rust inoculation) and between 6% and 33% at 32 dai. In treatment S, STB covered between 17% and 100% of the leaf segments at 28 dai. In treatment R+S, leaf rust lesion density ranged between 1.2 and 29.7 lesions cm–2 and rust severity ranged between 1% and 26% at 27 dai and between 1% and 21% at 32 dai At the same dates, STB severity was from 8% to 99% and from 9% to 99%, respectively.
|
In control leaves, the net photosynthetic rate ranged from 18.7–27.5 µmol CO2 m–2 s–1. Data from treatments R and S were used to estimate ßR for leaf rust and ßS for STB with a coefficient of determination ranging between 0.63 and 0.96 (Table 1; Fig. 1). The estimated values of ßR (equation 1) were 2.26±0.35 at 27 dai and 2.08±0.39 at 32 dai. The measurement date had no significant effect on ß (F=0.18,
1=149 and 2=1) and the value of ßR was finally estimated from the pooled data (two assessment dates), at 2.17±0.26. The value of ßS (equation 1) was 1.35±0.13 at 28 dai.
|
Data from treatment R+S were used to estimate ßR+ and ßS+ (equation 7) (Table 1; Fig. 1). At 27 dai ßR+=2.05±0.65 and ßS+=1.36±0.31 and at 32 dai, ßR+=1.91±0.58 and ßS+=1.43±0.28. Since the measurement date had no significant effect on ß (F=0.08,
1=2 and 2=28), the values of ßR+ and ßS+ were estimated from the pooled data at 1.98±0.41 and 1.38±0.20, respectively. The estimated values of ßR+ (1.98±0.41) for leaf rust and ßS+ (1.38±0.20) for STB when inoculated on the same leaves were not significantly different from the values obtained for separate inoculations (ßR=2.17±0.26 and ßS=1.35±0.14). It was therefore decided to use ßrust=2 and ßSTB=1.4 for the simulations at the canopy level (equation 10b).
Field experiment
The crop development stages, expressed in Julian days and degree-days (dd, 0 °C basis) from sowing and flowering are summarized in Table 2. Soissons flowered 1699 dd after sowing (26 May) and Recital 1597 dd after sowing (19 May). Although cultivars were sown on the same day, flowering occurred around 100 dd earlier for Recital than for Soissons, relative to the date of sowing. Since the phyllochron was 100 dd for both cultivars, the difference in the development of Soissons and Recital was one phyllochron. Because diseases affect yield according to the crop stage at which infections occur, the disease developments were expressed in cumulative dd from flowering for each cultivar.
|
Epidemic development: Visual assessments of the green leaf area in the field and image analysis of detached leaves gave similar results (data not shown). Disease distribution was horizontally uniform in the diseased plots (data not shown). Field assessment of disease severity in these plots is shown in Fig. 2.
|
In the Soissons reference treatment, STB was first recorded in mid-April (500 dd before flowering) in the lower layers of the canopy (Fig. 2A), but low temperatures until 25 April and low humidity from 27 April to 5 May were unfavourable to the development of the disease, which almost disappeared. STB arose again after rainfall on 6 and 7 May (250 dd before flowering) and layer L5 reached a severity of 40% on 10 May. Rainfall subsequently became more regular and STB spread from the lower layer to upper layers; layer L5 was 50% covered by STB at 180 dd and L4 at 30 dd before flowering. L3 was 50% covered by STB at 70 dd, L2 at 160 dd, and L1 (flag leaf) at 260 dd after flowering. The disease progressed by approximately one leaf layer every 100 dd.
Leaf rust was first recorded in mid-April in the two lower layers of the canopy (L6 and L5) (Fig. 2E). It developed in all the canopy layers after 10 May (220 dd before flowering), particularly in the lower layers. However, the rust epidemic was impaired by STB. The severity of the rust infection started to decrease when STB infected a leaf layer already colonized by leaf rust. STB necroses overlapped rust lesions, which stopped sporulating and were no longer recorded in the rust severity measurements.
STB appeared at the same time in the canopy layers of the water sprayed and reference plots (Fig. 2B). However, once a leaf was infected, the STB lesions spread faster on the leaf surface of the sprayed plots. STB covered 50% of L4, L3, L2, and L1 40 dd before the reference plots. Leaf rust started to colonize the successive canopy layers (Fig. 2F) at the same rate in sprayed and reference plots. Micro-aspersion slightly enhanced the development of leaf rust as long as the STB infection was low but, since STB developed faster on sprayed plants, it caused the rust infection to decrease more quickly. For instance, rust reached a severity of 60% in layer L1 in the reference plots, but only 45% on the sprayed plots.
STB developed in a similar way in the low-nitrogen and reference plots (Fig. 2C). Leaf rust appeared at the same time in the canopy layers of both plots (Fig. 2G). However, the rust infection grew faster, and the final rust infection was more severe on the low-nitrogen plots than on all others.
In the Recital treatment, STB and leaf rust developed earlier than on Soissons, relative to the date of sowing: the diseases were about 100 dd (1 week) earlier in layers L2, L3, L4, and L5, and 40 dd earlier in L1 (Fig. 3A). Because plant development was also 1 week earlier in Recital, the differences between Soissons and Recital almost disappeared when severity of STB was related to the crop development stage (Fig. 3B). Layer L1 still showed a difference, as the STB severity increased more quickly in the reference treatment (Soissons) than in Recital. This difference disappeared when considering the necrotic area (in cm2) instead of the proportion of necrotic area (Fig. 3C).
|
Plant growth and yield: The areas of the five upper leaves in all the control and diseased plots were equivalent (Table 3). As the number of spikes m–2 was also similar in diseased and control plots (Table 3), the leaf area index (LAI) was not altered by the diseases in any treatment.
|
The individual leaf areas for cv. Soissons were similar in all plots. Leaves L3, L4, and L5 were smaller and flag leaves (L1) larger in Recital than in Soissons (Table 3). As Recital had fewer spikes m–2 than Soissons (–20%), LAI and its vertical distribution were different for Recital and Soissons.
Stem height was 76 cm (from the base to the spike), and leaf layer L1 was at 58 cm from the soil, L2 at 39 cm, L3 at 27 cm, L4 at 17 cm, and L5 at 11 cm. The stem surface area taken into account for light interception was estimated to be 30.6 cm2 for all the treatments, as one half of the stem surface area for a plant. The stem area index in control plots was 1.46 m2 m–2 of soil in the reference plot, 1.54 in the sprayed plot, 1.60 in the low-nitrogen plots, and 1.21 in the Recital plots. This represented 24.1%, 22.9%, 22.7%, and 24.7% of LAI in the reference, sprayed, low-nitrogen, and recital treatments, respectively. The stem area indexes in the control and diseased crops were assumed to be identical.
The low-nitrogen treatment did not cause any change in the final leaf size but resulted in a lower leaf nitrogen content around flowering. The nitrogen contents of leaves 1, 2, 3, and 4 in the control low-nitrogen plots were slightly lower (5–8%) (data not shown) than in the reference Soissons treatment between 90 degree-days (dd) before flowering and 300 dd after flowering.
Plant growth was similar in all fungicide-treated plots for cv. Soissons, but aerial dry weight was higher for cv. Recital from 600 dd after flowering to maturity (Fig. 4; Table 3). At flowering, above-ground dry weight and spike dry weight were not affected by the disease, suggesting a similar yield potential for diseased and control plots (Table 4). The effect of the diseases on dry weight became significant (P=0.005) at 200 dd after flowering (Fig. 4). From 200 dd after flowering, diseased and control plants continued to accumulate dry matter for 300 dd for cv. Soissons and until harvest for cv. Recital, even though the growth of diseased plants was greatly reduced. Plant growth after flowering in diseased plots was reduced by 54% in the Soissons reference, 53% in the Soissons sprayed, 55% in the Soissons low-nitrogen, and 59% in the Recital treatment.
|
|
All the diseased plots had markedly reduced final yields (Table 3). Yields were reduced by 53%, 51%, 51%, and 55% in Soissons reference, Soissons sprayed, Soissons low-nitrogen, and Recital diseased plots, compared with control plots. There was no significant difference in yield among the four treatments either in diseased or control plots. However, the number of grains per spike and the mean grain weight, which were both significantly reduced in diseased plots, differed according to the cultivar. In Soissons, the diseases reduced the number of grains per spike by 13%, 11%, and 11% in reference, sprayed, and low-nitrogen treatments (Table 3). This reduction was greater in Recital and reached 20%. Although the number of grains per spike was not significantly different among diseased plots, it was clearly higher (P=0.0001) for cv. Recital (66) than for cv. Soissons (57 in the reference treatment) in the control plots.
The mean grain weight was more affected by the disease than the number of grains per spike in all plots (Table 3). The grain weight for diseased plants of cv. Soissons was reduced by 40%, 38%, and 38% in the reference, sprayed, and low-nitrogen plots. The reduction in grain weight for cv. Recital was less (30%), suggesting compensation for the greater decrease in grain number. Grain weight was not significantly different among the four treatments in control crops, but it was higher for Recital (27.5 mg) than for Soissons (23 mg, in reference treatment) in diseased plots.
The specific nitrogen content of the leaves (mg N cm–2 of leaf) was similar in diseased and control plots until flowering. After flowering, the nitrogen content of diseased leaves became lower than in the controls (Fig. 5). The leaf nitrogen content was significantly reduced in diseased plots in layers L2 at 110 dd after flowering and in layers L1 at 240 and 340 dd after flowering. The difference for the flag leaf was greatest in the low nitrogen treatment and least in the sprayed treatment. The difference in nitrogen content between diseased and control plots reversed at 450 dd after flowering and nitrogen content then became higher in diseased leaves.
|
Simulations
Leaf layer L1 made the greatest contribution (C1) to crop photosynthesis (equation 9a), and C1 increased from 43% to 44% between 250 dd before flowering and 800 dd after flowering in the Soissons reference treatment. The relative contributions (CS) of the stems to crop photosynthesis (equation 9b) in the reference plots increased from 27% to 31% between 250 dd before flowering and 800 dd after flowering. Over the same period, layer L2 contributed 20% to crop photosynthesis. Increases in C1 and CS resulted from the natural senescence that reduced the relative contributions to crop photosynthesis of layers L3, L4, and L5 (C3=7% to 4%, C4=2% to 0%, and C5=1% to 0%). Layer L5 was fully senescent by 325 dd after flowering and L4 by 500 dd.
The distribution of STB, leaf rust, senescent tissue, and green tissue in the Soissons reference treatment are shown in Fig. 6A. Figure 6B shows the relative contribution of the canopy layers to plant growth in the control plots and its reduction due to the vertical location of diseases. Although the leaves were totally necrosed 329, 306, 313, and 365 dd after flowering in the Soissons reference, Soissons water-sprayed, Soissons low-nitrogen, and Recital treatments, respectively, the diseased crops continued to accumulate dry matter because of stem photosynthesis. Simulations indicated that not considering stem photosynthesis decreased simulated maximal plant growth by 13.9% 14.4% 13.2%, and 14.5% in the diseased crops in the Soissons reference, Soissons water-sprayed, Soissons low-nitrogen and recital treatments, respectively (Table 5).
|
|
Despite the differences in the vertical distribution of LAI and in the yield component, the model accurately estimated the effect of disease on plant growth for each cultivar (Figs 7, 8). Simulations were consistent with the experimental data. Two kinds of discrepancies were, however, observed. First, the model slightly overestimated plant growth between 100 dd and 350 dd after flowering for the Soissons reference, Soissons low-nitrogen, and Recital treatments. Simulated plant growth in the Soissons low-nitrogen trial at 120 dd after flowering was 108% of that of the experimental data, which corresponds to an overestimation of aerial dry matter of 0.2 g stem–1. There were similar differences in the Soissons reference and Recital treatment at 200 dd and 300 dd after flowering. Second, the model underestimated maximal plant growth in the sprayed plots: simulated plant growth at 450 dd after flowering was 94% of that of the experimental data, which corresponds to an underestimation of aerial dry matter of 0.2 g stem–1.
|
|
The model indicated that STB accounted for 81%, 88%, 81%, and 70% of the reduction in plant growth in Soissons reference, sprayed, low-nitrogen, and Recital treatments (Table 5).
Changes in the parameter’s values (1<ßR<3 and 1<ßS<3) affected the model’s outputs only slightly (Table 6).
|
|
Discussion |
|---|
|
TopAbstractIntroductionTheoryMaterial and methodsResultsDiscussionReferences |
|---|
Disease development in the field experiment
STB developed well in all field treatments, probably because of a high infection potential and favourable climatic conditions. The water spray enhanced disease intensification on the leaves, but did not accelerate the spread of the epidemic, probably because the drop intensity was too weak to ensure spore dispersion from leaf to leaf.
Leaf rust developed slightly faster in the low-nitrogen treatment than in the reference treatment. This was surprising, as other studies (Robert et al., 2002; Tiedemann, 1996) have shown a positive correlation between high nitrogen fertilization and biotrophic pathogen development. The nitrogen deficiency occurred after flowering, when the canopy structure was already set, and only resulted in a limited reduction in leaf nitrogen. Slight nitrogen starvation increases the soluble sugar content of the leaves (Bancal and Soltani, 2002), and may increase the carbohydrates available to the fungus. Hoffland et al. (1999) showed that a high C/N ratio increased the formation of primary lesions of Botrytis cinerea on tomatoes leaves and hypothesized that the pathogen growth was linked to high soluble carbohydrate levels.
The difference in disease development between cvs Recital and Soissons, visible when time was related to degree-days after sowing, disappeared when time was expressed in degree-days from flowering. STB infected a new canopy layer every 100 dd in both genotypes. As this represents one phyllochron, it suggests that disease development was closely linked to crop development stages. This is consistent with other studies (Beresford and Royle, 1988; Paveley et al., 2000) that have shown leaf emergence was the main limiting factor for disease progression under conditions of high infection potential and optimal climatic conditions.
STB developed at a similar rate in all plots and leaf layers and appeared to be independent of rust severity. Leaf rust colonized the canopy layers first and STB became established afterwards on all plots, causing a decrease in rust severity.
Effect of disease on yield and plant growth
As in other pathosystems (Beasse, 1998), the disease affected plant growth late (200 dd after flowering) relative to the start of the epidemics (–250 dd before flowering). This probably was a consequence of the vertical distribution of the symptoms, which had little effect on plant growth as long as STB remained on the lower leaf layers.
Air-borne diseases develop mainly after flowering in northern France. The leaf area index is generally not altered by epidemics and yield reduction results only from an alteration of assimilate production by adult leaves. The disease can reduce grain weight and sometimes the number of grains per spike (Cornish et al., 1990; Zuckerman et al., 1997). Yield components were different in Recital and Soissons control crops: Recital had fewer spikes m–2 than Soissons, more grains per spike, and similar grain weight. In the diseased plots, the numbers of spikes m–2 for Recital and Soissons were not affected, but the number of grains was reduced more on Recital than on Soissons, while the Recital grain weight was less affected than that of Soissons. However, while the diseases affected the yield components for Soissons and Recital differently, the final reductions in yield were identical. Recital plants compensated for their larger reduction in the number of grains by a smaller decrease in grain weight, because the green tissue on layer L1 persisted longer.
Modelling the disease effect on leaf photosynthesis
Bastiaans (1991) formalized the relationship between net photosynthesis and the proportion of diseased leaf area for a single disease. His model relates the net photosynthetic rate of a diseased leaf to that of a healthy leaf using a single parameter ß. This model is based on the assumptions that lesions are independently distributed over the leaves, that the visible lesion size () and the virtual lesion size (ß) are similar for all lesions on the same leaf, and that a single virtual lesion (ß) occupies less than 20% of the leaf surface area. This last assumption can be restricting when the ß value is high. Although these assumptions are probably never fully verified in field assessments, because lesion size, for instance, depends on lesion age (Robert et al., 2002) and because their distribution on the leaves depends on local effects, Bastiaans’ model has been widely used (Bassanezi et al., 2001; Garry et al., 1998; Lopes and Berger, 2001) successfully to simulate disease effects on leaf photosynthesis. The robustness of this model results from the properties of Bastiaans’ formula (equation 1) relating the decrease in net photosynthesis (Px/Po) to the proportion of leaf surface area covered by the disease (x). For small severity values, (1–x)ß is equivalent to (1–ßx) and photosynthesis in a diseased leaf decreases linearly with severity, as if there is no competition among the lesions. For intermediate values of x and ß>1, the net photosynthesis ratio (1–x)ß is higher than (1–ßx), reflecting competition among the lesions. Finally, when x tends towards 1, photosynthesis tends towards 0. These properties are independent of the lesion size and distribution or the type of competition occurring among lesions and make the model suitable for a wide range of situations.
In the present study, Bastiaans’ model (equation 1) was extended to two diseases, one necrotrophic and the other biotrophic (equation 7). Necroses induced by the necrotrophic pathogen were able to grow over the biotrophic lesions, which were then removed from biotrophic disease assessment, whereas the biotrophic pathogen infected only healthy host tissue.
In the case of two diseases that would be assessed on the whole leaf surface, including parts where the diseases overlap, equation 4 would be valid for both pathogens:
1–x1=(1–1)n1(4a)
1–x2=(1–2)n2(4b)
where x1 and x2 are the proportions of the leaf covered by diseases 1 and 2, respectively, including the surface area where lesions overlap. In this case, equation 7 becomes:
Px1x2/P0=(1–x1)ß1x(1–x2)ß2 (7b)
which is the formula proposed by Lopes and Berger (2001). With the condition that both diseases are independently distributed over the leaf, equation 7 and 7b can be related to each other by calculating x'2, i.e. the proportion of the leaf covered by disease 2, but free of disease 1, as: x'2=x2x(1–x1). The choice between the formulae mainly depends on which quantity, x2 or x'2, is easier to evaluate in the field.
If, however, the independence of the distributions of the two diseases is not verified, equation 7 seems preferable because of its greater robustness with regard to this condition. This can be illustrated, for instance, by the extreme situation where x1=0.5 and x2=x'2=0.5, meaning that the leaf is fully covered by lesions but there is no overlapping. If ß1=ß2=1 for simplicity, equation 7b predicts a reduction in net photosynthesis to 25% whereas equation 7 predicts the correct estimation of photosynthesis reduced to zero.
In the current study, the ß values obtained for leaf rust and STB were 2.2 and 1.4, respectively, which led to the conclusion that there is a little effect of these diseases on the green area of wheat leaves. The ß value obtained here for leaf rust was close to that (1.26) calculated by Bastiaans (1991) and a ß value of 1.4 for STB is in accordance with other observations (Shtienberg, 1992). No interaction was observed between rust and STB with regard to their effect on net photosynthesis. In another study, Lopes and Berger (2001) did not observe any interaction either between rust and anthracnose on bean leaves. For both leaf rust and STB, ß is close to 1, which indicates a localized effect on the leaf.
It is difficult in field assessments to distinguish STB necrosis from natural leaf necrosis. In this field experiment, the STB assessment was corrected by taking into account leaf senescence in control plots. However, STB is known to accelerate apical necrosis in diseased leaves (Magboul et al., 1992) and this estimation of ß for STB might have been slightly too high. Since ß=1.4 for STB and by definition ß=1.0 for a natural necrosis, this overestimation was considered to be negligible.
Model evaluation
Model hypotheses: The hypothesis of a random distribution of lesions on the leaves is reasonable for air-borne pathogens like STB and leaf rust. Although STB lesions tend to merge when they are close to each other, their individual visual size is relatively small compared with the leaf size, and the measured values for ß are close to 1, which suggests that virtual lesions are only slightly bigger than visual lesions. For leaf rust, maximal lesion size is around 4 mm2 at a very low lesion density (Robert et al., 2002). With ßrust close to 2, the virtual lesion size should not exceed 8 mm2, which is less than 1% of leaf size in any canopy layer. This makes the approximation used for equation 6 acceptable.
To establish this model, it was also assumed that both pathogens were independently distributed on the leaves. This may not be correct in every situation, since leaf rust is unable to grow on STB infected tissue. However, the dynamics of the diseases in the field makes this assumption acceptable: because leaf rust uredospores are dispersed by the wind and septoria pycnidiospores are dispersed by splashing from the lower leaves, leaf rust usually appears first in the higher canopy layers and begins to develop independently of STB. In this experiment, the rust severity was over 60% and 40% on canopy layers L1 and L2 of the reference plots before STB started to interfere with its progression.
To estimate diseased crop growth, it was assumed that the relative contributions of the different layers to crop photosynthesis, as calculated for a healthy canopy, remained valid in the diseased canopy. In this experiment, as generally in Northern Europe, the leaf area index values are not altered by leaf rust and STB because epidemics are too late to induce differences in leaf size and number. The approximation that photosynthesis is proportional to light interception might become less accurate as LAI decreased towards plant maturity, but the use of fungicide-sprayed crop dry weight values as inputs to estimate relative effects probably absorbed most of these uncontrolled effects.
It was also assumed that diseases were horizontally homogeneous in each leaf layer of the canopy, which was the case in these field experiments and is generally observed for these diseases.
It was supposed that stem photosynthesis was similar in the diseased and control crops. Stem nitrogen contents were similar in both diseased and control crops during crop development (data not shown) which could support this hypothesis. Moreover, it was assumed that stem photosynthesis rate was equal to leaf blade photosynthesis rate and ear photosynthesis was not taken into account. Ear photosynthesis is generally ignored in crop growth models and is considered to be offset by ear respiration (Araus et al., 1993; Gebbing and Schnyder, 2001; Thomas et al., 1989). Wang et al. (2001) have found that, in the early stages of grain growth, net photosynthetic rate of peduncle and leaf sheath was around 50% of that of the flag leaf blade and close to 100% toward the end of grain filling. Further research on stem photosynthesis is still needed for improving crop growth modelling and understanding.
Simulations: The model correctly estimated the effect of disease on plant growth for each cultivar, differing from experimental values by less than 10%. This encourages this type of approach for understanding disease effects on plant growth.
Not considering stem photosynthesis decreased simulated maximal plant growth by around 14%. This suggests stem photosynthesis may be important for biomass accumulation in diseased plants. The model allowed the effects of each pathogen on plant growth to be separated: STB accounted for between 70–88% of the reduction in plant growth, depending on the treatment.
Two kinds of differences occurred between the simulated and the experimental data. First, between 100 dd and 350 dd after flowering, during the period of strongest epidemic development, the disease effect on plant growth was slightly greater than predicted by the model in all the treatments, except the Soissons sprayed treatment. This discrepancy could be due to spore production by the fungi, especially rust. Bastiaans (1993) quantified the effect of rice leaf blast (caused by Magnaporthe grisea) on leaf net photosynthesis in order to model the influence of the disease on rice dry matter production. He found that the simulated reduction in net photosynthesis was insufficient to explain the observed reduction in shoot dry weight fully, and suggested that the matter exported in the spores, which was neglected in the model, accounted for the overestimated dry weight in the simulations. This is supported in this study by a comparison of the four experimental treatments. The dry weight was not overestimated for the sprayed plots, where the leaf rust epidemic was limited, but was maximal for the low-nitrogen plots, where the rust was most severe. A study on wheat seedlings (Robert et al., 2002) showed that the overall spore production for the whole sporulation period, under optimal conditions, was 40 mg cm–2 of sporulating area. Applied to these field data, spore production could reach up to 0.5 g per stem. It was calculated that the model overestimated the final dry weight by 0.2 g per stem for the reference plot. As these data were obtained in the field and not under optimal conditions and since the rust epidemic was partly impaired by STB, these figures are consistent. The lower nitrogen content in diseased leaves, compared with control leaves, between 100 dd and 350 dd after flowering could also be due to the uptake of assimilate by the leaf rust spores. In field epidemics, spore production could then result in significant losses of biomass and nitrogen for the diseased plant, as has been shown for wheat seedlings in controlled conditions (Robert et al., 2002), and taking into account spore production by the pathogens could be an improvement for crop loss models (Bastiaans, 1993).
Another discrepancy between these simulations and the observations was that the model underestimated maximal plant growth in the water-sprayed diseased plots. Other studies (Ayres, 1995; Shtienberg, 1992) have shown that rust increases leaf transpiration by disrupting the host cuticle during sporulation. Spraying water onto the canopy may thus have reduced the effects of the pathogens by decreasing the water stress caused by the disease.
Simulations with ßR and ßS varying between 1 and 3 did not greatly change the prediction of this model. This was probably because in the field experiment, the diseases advanced very rapidly, especially for STB, and the pathogen effects on the green parts of the leaves were negligible with regard to tissue necrosis by STB and leaf colonization by sporulating rust lesions. Surprisingly, the model was slightly more sensitive to ßR than ßS. This could be explained by the location of the rust disease on the upper leaves of the canopy. This again demonstrates the importance of taking into account the disease vertical location for damage assessment.
In a first approach, the model correctly explained the relative impact of two simultaneous epidemics on plant growth in the field and allowed a separate estimation of the effects of each pathogen. It also allowed crucial points in the system to be identified and knowledge gaps in the understanding of disease impact on plant growth to be illustrated. However, this model only predicts relative effects and, in order to extend prediction from relative plant growth to actual plant growth, it would be necessary to predict directly the control crop growth as a function of time. Moreover, to extend prediction from plant growth to yield, the next step is to determine whether crop growth is a good predictor of yield in a diseased crop relative to a healthy crop.
|
Acknowledgements |
|---|
We are grateful to M Okamoto and B Le Fouillen for great technical help in field work, E Cachet and H Autret for their help with photosynthesis measurements, and F Lafouge for carbon and nitrogen analyses. We thank M Trottet and his team for field work and installing the sprayed-water system. We thank J Rodrigues for skilful image analysis and P Belluomo for programming. We thank S St Jean for discussing stem area index estimation and B Andrieux for constructive comments on the manuscript.
|
References |
|---|
|
TopAbstractIntroductionTheoryMaterial and methodsResultsDiscussionReferences |
|---|
Araus JL, Brown HR, Febrero A, Bort J, Serret MD. 1993. Ear photosynthesis, carbon isotope discrimination and the contribution of repiratory CO2 to differences in grain mass in durum wheat. Plant, Cell and Environment 16, 383–392.[CrossRef]
Ayres PG. 1995. Effects of disease on plant water relations. In: Ayres PG, ed. Effects of disease on the physiology of the growing plant. Cambridge, UK: Cambridge University Press, 131–148.
Bancal P, Soltani F. 2002. Source-sink partitioning. Do we need Münch? Journal of Experimental Botany 53, 1929–1928.
Bassanezi RB, Amorim L, Filho AB, Hau B, Berger RD. 2001. Accounting for photosynthetic efficiency of bean leaves with rust, angular leaf spot and anthracnose to assess crop damage. Plant Pathology 50, 443–452.[CrossRef][Web of Science]
Bastiaans L. 1991. Ratio between virtual and visual lesion size as a measure to describe reduction in leaf photosynthesis of rice due to leaf blast. Phytopathology 81, 611–615.[CrossRef][Web of Science]
Bastiaans L. 1993. Effects of leaf blast on growth and production of a rice crop. 2. Analysis of the reduction in dry matter production, using two models with different complexity. Netherlands Journal of Plant Pathology 99, 19–28.
Beasse C. 1998. Effets de l‘anthracnose due à Mycosphaerella pinodes sur l’élaboration du rendement du pois (Pisum sativum L.) au champ. PhD thesis, Rennes: ENSAR – INA PG.
Beasse C, Ney B, Tivoli B. 2000. A simple model of pea (Pisum sativum) growth affected by Mycosphaerella pinodes. Plant Pathology 49, 187–200.[Web of Science]
Beresford RM, Royle DJ. 1988. Relationships between leaf emergence and latent period for leaf rust (Puccinia hordei) on spring barley, and their significance for disease monitoring. Journal of Plant Diseases and Protection 95, 361–371.
Cornish PS, Baker GR, Murray GM. 1990. Physiological responses of wheat (Triticum aestivum) to infection with Mycosphaerella graminicola causing Septoria tritici blotch. Australian Journal of Agricultural Research 41, 317–327.[CrossRef][Web of Science]
Dumas JB. 1831. Procédés de l’analyse organique. Annales de Chimie et de Physique 247, 198–213.
Eyal Z, Clifford BC, Caldwell RM. 1968. A settling tower for quantitative inoculation of leaf blades of mature small grain plants with urediospores. Phytopathology 58, 530–531.
Garry G, Jeuffroy MH, Ney B, Tivoli B. 1998. Effects of Aschochyta blight on the decrease in photosynthetizing leaf area and the reduction of photosynthetic efficiency by green leaf area of dried-pea. Plant Pathology 47, 473–479.[CrossRef][Web of Science]
Gebbing T, Schnyder H. 2001. 13C Labeling kinetics in glumes indicates significant refixation of respiratory CO2 in the wheat ear. Australian Journal of Plant Physiology 28, 1047–1053.[Web of Science]
Girard ML. 1997. Modélization de l’accumulation de biomasse et d’azote dans les grains de blé tendre d’hiver (Triticum aestivum); simulation de leur teneur en protéines à la récolte. PhD thesis, INA-PG, INRA.
Hoffland E, Beusichem ML, Jeger MJ. 1999.Nitrogen availability and susceptibility of tomato leaves to Botrytis cinerea. Plant and Soil 210, 263–272.[CrossRef][Web of Science]
Justesen JH, Tammes PML. 1960. Studies on yield losses. I. The self-limiting effect on injurious or competitive organisms on crop yield. Tijdschrift over Plantenziekten 66, 281–287.
Leitch MH, Jenkins PD. 1995. Influence of nitrogen on the development of Septoria epidemics in winter wheat. Journal of Agricultural Science 124, 361–368.[Web of Science]
Lopes DB, Berger RD. 2001. The effects of rust and anthracnose on the photosynthetic competence of diseased bean leaves. Phytopathology 91, 212–220.[Medline]
Madden LV, Nutter FW. 1995. Modeling crop losses at the field scale. Canadian Journal of Plant Pathology 17, 124–137.
Madeira AC, Clark JA. 1991. Reduction in the efficiency of light use due to disease. Anais do Instituto Superios de Agronomia 43, 153–161.
Magboul AM, Geng S, Gilchrist DG, Jackson LF. 1992. Environmental influence on the infection of wheat by Mycosphaerella graminicola. Phytopathology 82, 1407–1413.
Paveley ND, Lockley D, Vaughan TB, Thomas J, Schmidt K. 2000. Predicting effective fungicide doses through observation of leaf emergence. Plant Pathology 49, 748–766.[CrossRef][Web of Science]
Peterson RF, Campbell AB, Hannah AE. 1948. A diagrammatic scale for estimating rust intensity on leaves and stems of cereals. Canadian Journal of Research 26, 496–500.
Rabbinge R, Jorritsma, ITM, Schans J. 1985. Damage components of powdery mildew in winter wheat. Netherlands Journal of Plant Pathology 91, 235–247.[CrossRef][Web of Science]
Robert C, Bancal MO, Lannou C. 2002. Wheat leaf rust uredospore production and carbon and nitrogen export in relation to lesion size and density. Phytopathology 92, 762–768.[Medline]
Robert C, Bancal MO, Lannou C. 2004. Wheat leaf rust uredospore production on adult plants: influence of leaf nitrogen content and Septoria tritici blotch. Phytopathology (in press).
Ross J. 1981. The radiation regime and architecture of plant stands. In: Leith H, ed. Tasks for vegetation sciences 3. Dr. Junk Publishers.
Shtienberg D. 1992. Effects of foliar diseases on gas exchange processes: a comparative study. Phytopathology 82, 760–765.
Thomas MR, Cook RJ, King JE. 1989. Factors affecting development of Septoria tritici in winter wheat and its effect on yield. Plant Pathology 38, 246–257.
Tiedemann AV. 1996. Single and combined effects of nitrogen fertilization and ozone on fungal leaf diseases on wheat. Journal of Plant Diseases and Protection 103, 409–419.
Van Oijen M. 1990. Photosynthesis is not impaired in healthy tissue of blighted potato plants. Netherlands Journal of Plant Pathology 96, 55–63.
Varlet-Grancher C, Gosse G, Chartier M, Sinoquet H, Bonhomme R, Allirand JM. 1989. Mise au point: rayonnement solaire absorbé ou intercepté par un couvert végétal. Agronomie 9, 419–439.[CrossRef][Web of Science]
Wang ZM, Wei AL, Zheng DM. 2001. Photosynthetic characteristics of non-leaf organs of winter wheat cultivars differing in ear type and their relationship with grain mass per ear. Photosynthetica 39, 239–244.[CrossRef]
Waggoner P, Berger R. 1987. Defoliation, disease and growth. Phytopathology 77, 393–398.
Weisberg S. 1985. Applied linear regression, 2nd edn. New York, Chichester: J Wiley and :Sons, 96–97.
Zuckerman E, Eshel A, Eyal, Z. 1997. Physiological aspects related to tolerance of spring wheat cultivars to Septoria tritici blotch. Phytopathology 87, 60–65.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M.-O. Bancal, R. Roche, and P. Bancal Late Foliar Diseases in Wheat Crops Decrease Nitrogen Yield Through N Uptake Rather than Through Variations in N Remobilization Ann. Bot., October 1, 2008; 102(4): 579 - 590. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M.-O. Bancal, C. Robert, and B. Ney Modelling Wheat Growth and Yield Losses from Late Epidemics of Foliar Diseases using Loss of Green Leaf Area per Layer and Pre-anthesis Reserves Ann. Bot., October 1, 2007; 100(4): 777 - 789. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Robert, M.-O. Bancal, C. Lannou, and B. Ney Quantification of the effects of Septoria tritici blotch on wheat leaf gas exchange with respect to lesion age, leaf number, and leaf nitrogen status J. Exp. Bot., January 1, 2006; 57(1): 225 - 234. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||