AOBPreview originally published online on August 7, 2007
Annals of Botany 2007 100(4):777-789; doi:10.1093/aob/mcm163
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Modelling Wheat Growth and Yield Losses from Late Epidemics of Foliar Diseases using Loss of Green Leaf Area per Layer and Pre-anthesis Reserves
Environnement et Grandes Cultures, INRA, F-78 850 Thiverval Grignon, France
* For correspondence. E-mail mobancal{at}grignon.inra.fr
Received: 12 March 2007 Returned for revision: 30 April 2007 Accepted: 8 June 2007 Published electronically: 7 August 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Background and Aims: Crop protection strategies, based on preventing quantitative crop losses rather than pest outbreaks, are being developed as a promising way to reduce fungicide use. The Bastiaans' model was applied to winter wheat crops (Triticum aestivum) affected by leaf rust (Puccinia triticina) and Septoria tritici blotch (STB; Mycosphaerella graminicola) under a range of crop management conditions. This study examined (a) whether green leaf area per layer accurately accounts for growth loss; and (b) whether from growth loss it is possible to derive yield loss accurately and simply.
Methods: Over 5 years of field experiments, numerous green leaf area dynamics were analysed during the post-anthesis period on wheat crops using natural aerial epidemics of leaf rust and STB.
Key Results: When radiation use efficiency (RUE) was derived from bulk green leaf area index (GLAI), RUEbulk was hardly accurate and exhibited large variations among diseased wheat crops, thus extending outside the biological range. In contrast, when RUE was derived from GLAI loss per layer, RUElayer was a more accurate calculation and fell within the biological range. In one situation out of 13, no significant shift in the RUElayer of diseased crops vs. healthy crops was observed. A single linear relationship linked yield to post-anthesis accumulated growth for all treatments. Its slope, not different from 1, suggests that the allocation of post-anthesis photosynthates to grains was not affected by the late occurring diseases under study. The mobilization of pre-anthesis reserves completely accounted for the intercept value.
Conclusions: The results strongly suggest that a simple model based on green leaf area per layer and pre-anthesis reserves can predict both growth and yield of wheat suffering from late epidemics of foliar diseases over a range of crop practices. It could help in better understanding how crop structure and reserve management contribute to tolerance of wheat genotypes to leaf diseases.
Key words: Triticum aestivum, Puccinia triticina, Mycosphaerella graminicola, leaf rust, Septoria tritici blotch, growth loss, yield loss, green leaf area per layer, pre-anthesis reserves
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSLITERATURE CITED |
|---|
In northern Europe, leaf rust (Puccinia triticina) and Septoria tritici blotch (Mycosphaerella graminicola; STB) are the main leaf diseases of wheat (Triticum aestivum). In the case of susceptible cultivars, they contribute to yield losses of up to 60 % (Cornish et al., 1990; Oerke and Dehne, 1997). To limit losses, it is necessary to use both resistant cultivars and fungicide treatments. Currently, concerns over conservation of natural resources and changes in the cost/benefit ratio make a limited use of fungicides attractive. Triggered by economic thresholds, crop protection strategies, based on preventing quantitative crop losses rather than pest outbreaks, are being developed (Zadoks, 1985; Rabbinge et al., 1989; Rossing et al., 1994; de Kraker et al., 2004; Savary et al., 2006) as a means to reduce fungicide use (Paveley et al., 2000). Two protection strategies are currently under investigation. One is to relate climate scenarios at a site to a probability of epidemics based on early evaluation of the disease. The other is to link an epidemic scenario accurately to yield loss. Both require further improvement either through a better understanding of the causes of epidemic variability or through a better accuracy of crop loss models. The present study aims to improve and simplify a crop loss model in the case of leaf rust and leaf blotch in wheat.
Crop models are accurate tools to predict crop growth and yield losses (Boote et al., 1983; Teng, 1985; Madden and Nutter, 1995). Most of them are derived from Monteith's model (Monteith, 1977). To calculate biomass accumulation, it fully separates radiation interception efficiency (RIE), or absorbed photosynthetic active radiation (aPAR), from radiation use efficiency (RUE). Diseases decrease rates of aPAR (Waggonner and Berger, 1987) and/or crop RUE (Johnson, 1987). aPAR can be decreased by lesion coverage, accelerated senescence and reduced leaf area formation depending on the epidemic time course (Boote et al., 1983; Bastiaans and Roumen, 1993). However, when a disease is unevenly distributed, its vertical position on leaves, i.e. its position per layer, affects dry mass accumulation because, if the disease is on the top layer, it has a greater impact on crop growth. Thus, the vertical position of the disease also needs to be taken into account (Johnson, 1987).
Models to predict the impact of diseases on RUE are commonly derived from studies at the leaf level (Bastiaans, 1993; Garry et al., 1998; Béasse et al., 2000; Lopes and Berger, 2001). For most foliar diseases, leaf photosynthesis reduction is not proportional to the proportion of leaf area affected. To characterize this effect, Bastiaans (1991) introduced the concept of virtual lesion. The virtual lesion is defined as the percentage of diseased leaf area with a photosynthetic rate equal to zero, an area whose size fully accounts for a proportionnal drop in the rate of photosynthesis of the diseased leaf, as compared with the rate of photosynthesis of the healthy leaf. A single parameter ß, i.e. the ratio between the virtual and the visual lesion size, characterizes the effect of diseases as higher, lower or equal to the size of the visible lesion. A large variability of ß values has been found between pathosystems, ranging from 1 to 13, whether biotrophic or necrotrophic (Rabbinge et al., 1985; Garry et al., 1998; Bassanezi et al., 2001). Several studies incorporated effects of disease on photosynthesis into a crop growth model through the ß parameter to estimate growth loss at the canopy scale (Bastiaans, 1993; Béasse et al., 2000; Bassanezi et al., 2001; Robert et al., 2004). Bastiaans (1993) summed leaf layer CO2 exchanges for rice crops attacked by Pyricularia oryzae. Béasse et al. (2000) and LeMay et al. (2005) estimated each diseased (Ascochyta blight) or healthy leaf layer contribution to pea growth, according to radiation received by leaf layer. Robert et al. (2004) used the same method to account for wheat STB and leaf rust attacks in complex, although in the case of a sole year of field data (1999) and in relation to a fast epidemic development with a strong dominance of STB epidemics over leaf rust epidemics. Using the Bastiaans' equation extended to two diseases, they found that the sensitivity analysis on the ß value (between 1 and 3 for leaf rust and STB) suggested a ß value of 1 could be used for both diseases to model crop growth loss. Moreover, Robert et al. (2005, 2006) showed that a ß value of 1 can be used for both diseases to model net photosynthetic loss that includes all non-green leaf tissues in disease assessments. This result suggested that for estimating crop loss it is not necessary to distinguish between the two diseases, nor between the different symptoms developed by the disease, nor between the disease symptoms and the induced necrosis. The significant damage appears to be the loss in total green leaf area. Thus, an accurate estimation of loss in green area on each leaf layer of the canopy might be sufficient to predict the overall loss in crop growth due to the leaf rust–STB complex. Moreover, leaf damage can sometimes vary greatly with the developmental cycle of the fungus (Scholes and Rolfe, 1996; Robert et al., 2005) or with the physiological state of the host that also influences pathogen development (Snoejers et al., 2000; Erickson et al., 2003 Robert et al., 2005). Therefore, to verify in the best way the efficiency of green leaf area loss by layer to account for growth and yield losses, it is necessary to evaluate this over a wider range of crop practices and epidemic development.
In this study, losses in crop growth are first evaluated over a range of scenarios combining epidemics with crop practices. Crop growth conditions were varied according to crop management practices, i.e. nitrogen fertilization (Leitch and Jenkins, 1995; von Tiedemann, 1996), plant density (Savary et al., 1995; Lovell et al., 2002) and water supply (Shtienberg, 1991), well known to affect disease epidemics through either leaf sugar, nitrogen content and/or crop structure. aPAR was calculated either at the canopy level (aPARbulk) from the total green leaf area index (GLAI) or layer by layer (aPARlayer), by summing the calculated aPARi of each leaf layer i depending on the measured green leaf area of that given layer.
Secondly, whether or not yield losses could be simply derived from the losses in winter wheat crop growth was evaluated. An examination was made of how wheat leaf rust and blotch affected the growth–yield relationship either through a variation of harvest index (HI), as shown by Bastiaans (1993) in the case of late epidemics, and/or through the contribution to grain filling of pre-anthesis reserve mobilization, as previously documented by Gaunt and Wright (1992) and Cornish et al. (1990).
Finally, there is a discussion on how a simple model can be postulated from the results to conclude that green leaf losses per layer and the contribution of pre-anthesis reserves are sufficient to account for wheat growth and yield losses in the case of the late foliar diseases under study.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Field experiments
Five field experiments in Grignon (France, INRA-EGC Station; 48 °51'N; 1 °57'E), from 1995 to 2001, were laid out in a randomized block design with three replicates per treatment. To obtain crops (Triticum aestivum L.) with a range of epidemics, the weather, as well as full crop protection were relied on to ensure disease-free control plots during wheat development in spring. The fungicide Opus® (‘epoxiconazole’, 12·5 mg m–2) was sprayed in mid-April during stem elongation (GS 31–32) and in May when flag leaves were fully expanded (GS 39). The fungicide Alto® (‘cyproconazole’, 8 mg m–2) was sprayed in mid-June at mid-grain filling (GS 75). Crop practices (nitrogen and plant density management) were also used to obtain a range of epidemics representative of northern France.
Treatments
All crops were sown between 20 and 22 October for the different years, close to the optimal sowing date at the experimental site (Grignon). Each plot (42 m2) involved nine 30 m long rows, spaced 0·5 m apart. To limit soil-borne diseases, an early fungicide (‘cyprodinyl’, 60 mg m–2) was applied at the beginning of stem elongation (GS 30). There were no differences in plant and spike density between healthy and diseased crops (data not shown). All trials included a fungicide-treated control plot for each treatment. One treatment refers to both the non-inoculated and corresponding protected control plots (described in Robert et al., 2004). Thus, a treatment is described by a combinaison of five factors: (1) year of the experiment (5 years); (2) type of nitrogen fertilization (high or low); (3) plant density (high or low); (4) cultivar used (‘Soissons’ and ‘Récital’); and (5), water sprayed or not (yes or no). The different combinations are presented in Table 1. It should be noted that the main objective of this study was not the specific effect of treatments, rather the relationship between crop growth and yield to light absorption in crops with varying epidemics. Consequently, no attempt was made to establish a fully cross-factor experimental set-up or to analyse factor interactions.
|
The treatment ‘high density–high nitrogen’, named ‘reference treatment’, reflects the climatic potential for disease development in the different years. In high density (250 pl m–2) and high nitrogen plots, 24 g m–2 of nitrogen were broken up into three applications: 6 g m–2 in mid-March (GS 25), 10 g m–2 at the beginning of April (double ridge or beginning of stem elongation, GS 30) and 8 g m–2 close to heading (GS 55). In the low nitrogen plots, the last application was either omitted or limited to 3 g m–2. Both low and high nitrogen plots were managed to achieve a potential yield of 9 t ha–1.
Low density stands (100 pl m–2) were also sown in order to create a different crop structure. To achieve a potential yield of 4·5 t ha–1 at low plant density, corresponding plots were only supplied with 11·5 and 8·5 g N m–2, respectively, for high and low nitrogen treatments split up as follows: no application in mid-March (GS 25), to limit tillering, followed by a first application of 5 g m–2 at the beginning of April (GS 30) and a last application of either 6·5 or 3·5 g m–2 for heavy and light nitrogen treatments, respectively (GS 55). A water-sprayed treatment was also carried out to enhance the incidence of STB in 1999, as described in Robert et al. (2004). Most experimental treatments were performed with the ‘Soissons’ cultivar. The ‘Récital’ cultivar was included in the 1999 experiment to compare epidemics and their effects on growth and yield for two cultivars having slightly different precocity and yield components (Robert et al., 2004). It was also another means to create a different crop structure.
Plant sampling and assessments
By collecting two crop samples (covering 0·18 m2 each) from each plot, crop dry mass (DM) and GLAI per leaf layer were assessed weekly from heading or anthesis to physiological maturity (15 July). This time period reflected specific characteristics of STB. Even though STB develops early, between tillering and heading, its long latency period allows 3–4 new leaves to develop fully before symptoms appear. Thus, crop losses hardly occur before heading or even anthesis, and, in the present experiments, no difference in crop growth between healthy and diseased crops was found at anthesis (data not shown). On each sampling date, main stems were separated from tillers. Apart from the main stem sub-sample described below, both main stems and tillers were separately oven-dried at 80 °C for 48 h and then weighed separately for spikes and vegetative parts, to insure that disease affected main stems and tillers identically, which was the case (data not shown). Extrapolation of sub-sample data at the m2 level was then obtained from the tiller to main stem DM ratio. A sub-sample of 15 main stems selected with an individual fresh weight as close as possible to the average main stem fresh weight (±5 %) was put aside. The five upper leaf blades, stems plus leaf sheaths, as well as ears, were removed. The DM and nitrogen contents (Dumas, 1831) of the four upper leaves, stems plus sheaths, awns plus rachis, and grains were measured. Leaf protein content (g per g DM) was derived from nitrogen content data assuming a conversion coefficient of 6·25 for leaf proteins. Total water-soluble carbohydrates (WSCs expressed in g per g DM) were extracted by 70 % aqueous ethanol from oven-dried stem samples (Gaunt and Wright, 1992) and measured with a continuous flux colorimeter. Pre-anthesis reserve mobilization was calculated as the difference of leaf proteins plus stem soluble sugars (both expressed in g DM per m2) between anthesis and maturity, assuming it was reallocated to grains.
Assessment of disease severity as well as total and green leaf area
The severity of leaf rust and STB was estimated visually from heading to maturity. The standard Peterson's scale (Peterson et al., 1948) was used in the case of leaf rust. STB severity was recorded directly as the percentage of overall necrotized leaf area, including both sporulating and non-sporulating necrotic areas, as previously described in Robert et al. (2004). Leaf area index (LAI; leaf m2 per soil m2) and GLAI (leaf m2 per soil m2) were measured weekly on a sub-sample of 15 main stems per bloc and per treatment as described above. The five upper leaf blades, when still green, were colour-scanned separately to measure total, green and necrotized areas per leaf layer by image analysis using Optimas software (Media Cybernetics, Silver Spring, MD, USA). The fifth leaf from the top down was often dry at anthesis; however, the images of the scanned fourth leaf from the top down were unusable in 1997 and 1998; thus the remaining analysis concerned only the top three leaves in those years. The percentage of leaf apical necrosis and GLAI were also measured weekly on control crops, and their associated area under the disease progress curve (AUDPC) was calculated from anthesis till maturity.
Data analysis
Using data obtained by image analysis, GLAI was calculated for both control and diseased crops for each treamtent. Three different AUDPCs were calculated for the top three leaves of the diseased crops: area under the sporulating leaf rust progress curve; area under the STB progress curve (including STB and necrosis); and area under the ‘non-green’ (both diseases and necrosis) progress curve. For control crops, only the third type of AUDPC was calculated using measurements of GLAI in the control crops. To calculate, AUDPC(X) = 
j[(Xj + Xj+1) x (Tj+1–Tj)/2] was used, with X being either (a) sporulating leaf rust severity; (b) STB severity; or (c) non-green area (1–GLAI = Xrust + XSTB), and Tj being the thermal time base zero accumulated from anthesis (°Cd) till day j.
Multiple analyses of variance (ANOVAs) were conducted using Statgraphics Plus (Manugistics Inc., Rockville, MD, USA) to examine the effects of fully crossed factors on disease (AUDPC for leaf rust, STB + apical necrosis, and total), GLAI, HI and DM components (total and grain DM per mean stem, leaf and stem nitrogen and carbohydrate reserves). Effects of year, nitrogen fertilization, fungicide and replicates were analysed on the overall data set minus low plant density stands (Table 1; L–M) and cultivar ‘Récital’ (Table 1; I). As the effects of either plant density or cultivar were restricted to one experimental year each, i.e. 2001 and 1999, respectively, they were compared with the reference treatment (cultivar ‘Soissons’ high nitrogen–high density) of the same year using a multifactorial ANOVA. Statistically significant differences in means of DM, HI, WSC and nitrogen content of the plant compartments among treatments in control and diseased plots were determined with the Student–Neuman–Keuls or Bonferonni test. The overall error rate was 
= 0·05.
Plant growth modelling
A conceptual framework was used based on the Monteith approach to analyse disease impact on crop yield. It allowed the analysis of how diseases affect each component of the following relationship between yield and DM accumulation:
|
| (1) |
RIE estimation
Bulk RIE (RIEbulk) was first estimated on each day t from global GLAI(t) measured on a given day t according to Monsi and Saeki (1953):
|
| (2) |
Then, RIE was estimated, taking into account the vertical distribution of GLAI. The RIE of each leaf layer i (RIEi) was calculated according to Béasse et al. (2000), from the measurements of LAI and GLAI of varying canopy layers instead of diseased areas:
|
| (3) |
|
| (4) |
RIElayer over the grain filling period was fitted to logistic curves according to RIEmax/[1 + exp(4 x V/RIEmax) x (I–T)], where T is the cumulated degree-days (°Cd) post-anthesis with Statgraphics Plus, advanced non-linear regression tools (Manugistics Inc.). This approach was chosen due to the physiological meaning of the parameters: RIEmax, the asymptotic value of RIE; I, the x-axis of the inflexion point of the curve (°Cd), which indicates the length of light interception; and V, the rate of RIE decrease at the inflexion point I. The significance of differences between parameter values in different treatments was tested using a Student–Neuman–Keuls test for pair comparisons, with an overall error rate of = 0·05.
Estimation of RUE
To estimate RUE, it is first necessary to calculate aPAR as:
|
| (5) |
Growth during the grain filling period can be simply related to absorbed radiation as:
|
| (6) |
Growth simulations
Four simulations of crop growth from anthesis to maturity were carried out using eqn (6) for healthy and diseased crops combining different means of calculating aPAR and RUE. They were: simulation A, aPAR with RIEbulk and mean RUE for C3 species (3 g MJ–1; Jones, 1994); simulation B, aPAR with RIElayer and mean RUE for C3 species; simulation C, aPAR with RIEbulk and RUE of healthy crops calculated per treatment; and simulation D, aPAR with RIElayer and RUE of healthy crops calculated per treatment. Simulated growth was compared with observed growth, and the accuracy of the simulations was compared on the basis of coefficient of correlation, slope and root mean squared error (r.m.s.e.).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Disease development and green leaf area dynamics
Treatments produced a large range of leaf rust and STB epidemics, as shown in Table 2. AUDPC values varied from 3·4 to 58·6 (% rustsporulating °Cd) using leaf rust severity and from 346 to 574 (% STB °Cd) using STB severity plus apical necrosis. Year induced highly significant variations of both leaf rust and STB, although late nitrogen application only influenced STB and apical necrosis significantly. In one year, this was also true for density and, in a given year, it was equally true for cultivar. These epidemics resulted in areas under the GLAI progression curves ranging from 440 to 1300 m2 m–2 °Cd in the diseased crops. Treatments also produced a wide range of apical necrosis and GLAI dynamics in the control crops, with a significant effect of all factors on these dynamics. Areas under the GLAI progression curves ranged from 510 to 2008 m2 m–2 °Cd in the control crops, revealing a significant effect of fungicide application on apical necrosis and GLAI dynamics.
|
Figure 1 presents GLAI profiles for healthy and diseased crops at three dates chosen to characterize yearly disease development, i.e. anthesis, mid-epidemics and 1 week before the end of epidemics. In the control crops, the overall GLAI varied from 3·5 to 4·75 m2 m–2 at anthesis depending on the experimental treatments, leading to different crop structures. At anthesis, flag leaf, leaf 2 and leaf 3 represented from 28 to 35 %, from 35 to 43 % and from 23 to 37 %, respectively, of the GLAI of the top three leaves. GLAI differences over time increased in relation to disease incidence (Fig. 1). For most of the studied years, i.e. 1995, 1997 and 1998, epidemics developed rapidly from 250 until 550 °Cd after anthesis. They even occurred after 323 °Cd in year 2001, whereas year 1999 was characterized by an early disease development (Fig. 1F–I). Unlike all other years, in 1999, GLAI of diseased leaf layers already differed from that of control plots at anthesis. With the same sowing dates and densities, the different years thus induced both different crop structures at anthesis and different GLAI dynamics due to varying epidemics.
|
RIE estimation
RIElayer [eqns (3) and (4)] was plotted over time and fitted to a logistic curve. Table 3 details the estimated parameters. The overall coefficient of determination for each treatment averaged 0·98 (
= 0·03) in all cases > 0·89. Figure 2 presents the range of RIElayer (t) due to the different treatments.
|
|
Control plots (solid lines) show a wide range of variation by themselves, essentially between years. The main variations occurred in the case of parameters I and V, whereas RIEmax remained constant except in the case of the low density treatment, which decreased RIEmax by about a half (Table 3). I ranged from 524 to 635 °Cd between years with high nitrogen application. Based on increasing I, years could be classified from the shortest to the longest length of green area duration as follows: 1998, 1999, 2001, 1995 and 1997, but only 1998 was significantly lower than the others (P < 0·001). The omission of the late nitrogen application slightly decreased I, although not significantly (P = 0·46). V ranged from 0·0041 to 0·0078 % °Cd –1 between years with high nitrogen application, revealing a large decrease in 1999 (P < 0·001; treatment F). The omission of the late nitrogen application also slightly decreased V, but not significantly (P = 0·07) as compared with high nitrogen application rates.
Diseased plots (dashed lines) increased the variation of RIElayer(t) (Fig. 2). The main effect showed up in parameter I, which varied from 134 to 407 °Cd mostly with between-year variations; in fact, I values differed considerably across years (P < 0·001); years ranging as 1999 < 1995 < 2001 = 1998 = 1997), whereas different treatments had no significant effects on I in any given year (P > 0·16). Similarly, the parameter V varied significantly with year (P < 0·001; years ranging as follows: 1998 = 1997 = 2001 
1999 = 1995) but not with nitrogen application (P = 0·67). Except for the low density treatment, no significant differences were found in RIEmax between both year and treatments, or diseased and control plots. When compared with control plots, disease significantly reduced the duration of interception during the grain filling period from 3 % to as much as 71 % (P < 0·001), but did not significantly affect the rate of RIE decline (P = 0·06) in the high density treatments involving ‘Soissons’.
From RIElayer(t) estimations, the aPAR during the post-anthesis period was calculated for both control and diseased plots. For the control plots, aPAR ranged from 153 to 354 MJ m–2, while for diseased plots it ranged from 76 to 247 MJ m–2. When only the top three leaves, instead of the top five, were included in the RIE calculation, it led to a < 5 % underestimation of cumulative aPARlayer during the grain filling.
RUE estimation
Overall RUE and intercept were derived per treatment from the relationship between accumulated DM and aPAR (Table 4). When RIEbulk(t) [eqn(2)] was used to calculate aPAR, healthy and diseased crop RUE per treatment were not found to be significantly different in nine out of 13 cases (Table 4). RUE values were highly variable in both healthy and diseased crops, ranging respectively from 1·59 to 3·98 g MJ–1 and from 1·69 to 10·74 g MJ–1. It resulted in high coefficients of variations of 20 and 62 %, respectively, for healthy and diseased global RUE values between treatments. The fitted intercepts were also highly variable (coefficient of variation of 168 and 236 %) and, in most cases, far from zero, even though significantly different only in two cases (treatments E and F). Intercepts between healthy and diseased crops were found to be significantly different in three treatments: E (P = 0·001), L (P = 0·03) and M (P = 0·01).
|
aPAR was also calculated from RIElayer. Figure 3A–M shows the relationship between accumulated DM and aPAR accounting for leaf layers for diseased and healthy crops separately for each treatment during grain filling. When compared with the calculation in which aPAR did not account for leaf layers, coefficients of variation of RUElayer between treatments were markedly decreased, amounting to 9 and 18 % for healthy and diseased treatments, respectively. No crop RUElayer lower than 1·9 g MJ–1 (
= 0·35) or higher than 3·5 g MJ–1 ( = 0·7) was observed. Significant differences of RUElayer were observed depending on the year, but there was only one treatment with a significant difference between healthy and diseased crops (Fig. 3C; P = 0·032), suggesting that in most cases diseases did not affect RUElayer. In all cases, the relationship showed an intercept not different from zero (P > 0·11) and without any difference between healthy and diseased treatments (P > 0·12) except for treatment B (P = 0·046).
|
Growth simulations
Crop growth was simulated according to four cases, with a slope between observed and predicted values never differing from 1 (Fig. 4).When case A was compared with case B, for which a mean RUE of C3 species was used, accounting for leaf layers in aPAR calculation increased the R2 of the prediction from 0·69 (r.m.s.e. = 152; n = 26; Fig. 4A) to 0·90 (r.m.s.e. = 75; n = 26; Fig. 4B). Predictions were found without any significant bias, and intercepts were not different from zero (P = 0·66 and 0·73). When aPAR was calculated from RIEbulk using the recorded healthy RUE per treatment instead of mean RUE of C3 species, the accuracy of predictions increased from 0·69 to 0·87 (r.m.s.e. = 94) (Fig. 4C). However, the intercept was significantly different from zero (P = 0·042). When case B was compared with case D, calculating aPAR from RIElayer when healthy RUE per treatment was used, the accuracy of the prediction was 0·91 (r.m.s.e. = 71), a slight increase over case B (Fig. 4D); in case D, neither slope (P = 0·88) nor intercept (P = 0·26) was significantly different from 1 and 0, respectively.
|
Relating growth to yield
Photosynthate partitioning to grains
HI variations with treatments were analysed (Table 5). Multifactorial ANOVA showed year had a significant effect on HI and separated years into three distinct groups: (1) 1995–1997–1999; (2) 2001; and (3) 1998, with respective average values of 0·41, 0·47 and 0·50 (P < 0·001). Diseases also significantly decreased HI from an average value of 0·47 to 0·41 (
= 0·021; P < 0·001). Across all years, nitrogen did not significantly modify HI, although in 1999, HI increased in ‘Récital’ (P = 0·021) more than in ‘Soissons’. HI also increased in 2001 in low plant density stands (P = 0·039) as compared with high density ones.
|
The causes of disease effects on HI were investigated by analysing the growth–yield relationship during the grain filling period. Accumulated growth from flowering until maturity was related to yield (Fig. 5A). The relationship, established for both healthy plots and diseased plots, was highly significant (R2 = 0·90; n = 24). Data showed no significant bias on residues over the growth range. Neither slopes nor intercepts were found to be significantly different between diseased and control crops: P-values were 0·46 and 0·55, respectively. The overall slope for all data points was 0·90 (
= 0·061), not significantly different from 1 (P = 0·11), whereas an intercept of 294 g m–2 ( = 49 g m–2) was significantly different from 0 (P < 0·001), suggesting a possible contribution of pre-anthesis reserves to yield.
|
Mobilization of pre-anthesis reserves
Yield vs. total DM contributing to the grain filling is plotted in Fig. 5B. Total DM includes both photosynthesized biomass during the grain filling period and biomass remobilized from pre-anthesis reserves (see further comments in Table 6 for DM mobilization account). Neither slopes nor intercepts differed significantly between healthy and diseased data sets: P-values were 0·37 and 0·80, respectively. The overall relationship was highly significant (R2 = 0·94; n = 24), showing a slope of 0·94 (
= 0·05) not significantly different from 1 (P = 0·25; n = 24) and an intercept of 34 g DM m–2 ( = 56 g m–2) not significantly different from 0 (P = 0·54; n = 24).
|
Table 6 shows the respective contribution of DM remobilization (expressed in g m–2), i.e. stem sugars and proteins from the vegetative part to grain filling by year and treatment. Pre-anthesis reserves amounted to 207 g m–2 (
= 39 g m–2) in healthy crops, with values ranging from 108 to 334 g m–2, and to 234 g m–2 ( = 39 g m–2) in diseased crops, with values ranging from 148 to 348 g DM m–2. Year (P < 0·001), nitrogen (P = 0·02) and plant density (P = 0·01) showed a significant effect; disease did not (P = 0·057).
Stem sugars accounted for 52–80 % of pre-anthesis reserve contribution to yield. A multifactorial ANOVA indicated that two main factors affected pre-anthesis sugar contribution to grain yield. First, the year (P < 0·001) showed an important effect: two groups of years were significantly different, 1999–1995–1997 with 2001, and 1998 alone, with mean contributions of sugars of 128 g m–2 ( = 18) and 259 g m–2 ( = 15), respectively. Secondly, fungicide treatment showed a significant effect on stem WSC contribution (P = 0·03; n = 39) although lower than among years, 153 g m–2 ( = 12) and 172 g m–2 ( = 12) for control and diseased crops, respectively. Table 6 also points out that the difference decreased in the case of strong and early epidemics, i.e. in 1999.
Despite a lower contribution of pre-anthesis proteins to yield, this metabolic behaviour was more sensitive to the different factor variations. Actually, only late nitrogen application did not show any significant effect on post-anthesis protein balance (P = 0·076). Year, again, had a strong effect (P < 0·001), creating three groups: 1998–2001–1997, then 2001–1997–1999, and finally 1995, with mean protein contributions of 38, 44 and 70 g m–2 respectively. In 2001, plant density (P < 0·001) also produced a significant difference: protein contribution to grain growth was lower in low density crops. Were it possible to compare genotype, it would produce no significant difference (P = 0·06) in protein contribution to grain growth in 1999. Finally, a strong significant effect of fungicide application was observed (P = 0·001), control and diseased crops remobilizing 82 and 72 g m–2, respectively, of proteins to grains. Statistically, disease does not have an effect on the overall contribution of pre-anthesis reserves to yield. However, an effect can be detected when examining solely pre-anthesis soluble sugars or protein mobilization. In relative terms, the overall pre-anthesis reserves contributed from 15 to 36 % to yield, and this relative contribution varied significantly depending on year (P = 0·007), nitrogen application (P = 0·023) and plant density (P = 0·006), but not on fungicide application (P = 0·12) and genotype (P = 0·15). On average, in the present study, foliar diseases decreased the relative contribution of pre-anthesis reserves from 33 % ( = 2·2 %) to 29 % ( = 2·2 %).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSLITERATURE CITED |
|---|
For the range of epidemics varying widely in both timing and rate, it was confirmed that post-anthesis cumulated aPARlayer accounted well for total growth during the grain filling period: it explained from 67 to 99 % of growth variability. The relationship was better in the case of healthy crops (mean R2 = 0·96 for control vs. mean R2 = 0·84 for diseased crops), suggesting that, in that case, aPARlayer is the main factor accounting for post-anthesis crop growth, although other phenomenona may intervene in the case of diseased crops.
When leaf layers were not accounted for in aPAR calculation, the resulting fit suggested a disease effect on RUEbulk in one-third of the studied crops. However, when leaf layers were accounted for, the accuracy of the RUElayer calculation greatly increased. Moreover, the RUElayer for healthy and diseased crops was only significantly different in one out of 13 situations, suggesting that diseases do not really affect RUE. This is consistent with previous results at the leaf scale that showed that ß values (Bastiaans, 1991) are close to 1 for both leaf rust and STB when disease assessment includes all three diseased symptoms: sporulating, yellow halo and necrotized areas (Bassanezi et al., 2001; Lopes and Berger, 2001; Robert et al., 2005, 2006). If wheat leaf rust and STB do not affect the photosynthetic efficiency of green leaf parts of individual leaves, it is highly unlikely they affect RUE at the crop scale as it measures crop photosynthetic efficiency. Thus, the present results suggest that it is possible to use a ß value of 1 for both diseases, either alone or in complex across very different crop practices. This allows the full separation of RIE from RUE in the Monteith approach, and from RUE alone it is possible to estimate absolute growth losses. The fact that ß has a value of 1 is convenient for modelling purposes. Many studies have already shown that ß values are very stable with crop management and even genotype for different disease–crop combinations, such as in the case of wheat leaf rust (Robert et al., 2005) and STB (Robert et al., 2006). The present results are also in agreement with work by Bastiaans (1991) and Bassanezi et al. (2001). To our knowledge, only one study pointed out that poplar rust effects on photosynthesis strongly depended on genotype (Erickson et al., 2004). The findings allow widening the range of application from early and fast epidemics (Robert et al., 2004) to slower or later epidemics.
When leaf layers were used in aPAR calculation (RIElayer), the overall RUElayer estimated at 2·64 ( = 0·6) g MJ–1 for healthy crops was not far from the mean of C3 crops, estimated to be 3 g MJ–1 (Jones, 1994). For different wheat genotypes and climates, Bryson et al. (1995, 1997) found that RUE varied from 2·5 to 2·9 g MJ–1 of aPAR. The present range of RUElayer estimates was wider: from 1·9 ( ; = 0·35) to 3·5 ( = 1·2) g MJ–1, but variations were linked to the year more than to differences between diseased and healthy crops. In contrast, using RIEbulk, i.e. averaging the decrease of green leaf area between leaf layers, led to an important overestimation of RUEbulk, particularly in the case of early disease development (i.e. year 1999), even though the crops under the present conditions suffered mainly from STB which shows a senescence-like progression from the bottom to the top of the crop (Robert et al., 2004). This overestimation further suggests how important an aPARlayer estimation may be in the case of either biotic or abiotic stresses affecting primarily the top leaf layers, i.e. wheat leaf rust (Seck et al., 1991) or ozone (Walton et al., 1997).
Under the present conditions, when RIEbulk rather than RIElayer was used in aPAR estimation, the results were very rough. With a constant RUE of 3 g MJ–1, the accuracy of the prediction of biomass accumulation was moderate, with an R2 equal to 0·69. Still, as it accounted for 70 % of growth variations, RIEbulk may be a useful, easy-to-collect variable for policy making. However, bypassing bulk GLAI leads to an RUEbulk outside the biological range in the case of a healthy treatment. On the contrary, when RIElayer was used in aPAR estimations, the prediction of biomass accumulation was accurate, using the generic 3 g MJ–1. However, the use of specific RUE per treatment only slightly enhanced the accuracy of the biomass prediction because, in the present case, treatments only moderately affected RUE. Therefore, if treatments strongly affect healthy crop RUE, measuring or modelling RUE would enhance the prediction of disease effect on biomass accumulation. In the present case, accounting for layer GLAI dynamics was far more important than precisely estimating RUE variations. It points out the need to investigate further the relationships between crop structure resulting from climate and crop management practices, as well as senescence dynamics per leaf layer. It may be of particular importance for pathogens having few effects on asymptomatic leaf areas, i.e. those mainly damaging the crop through the senescence rate.
Confirming previous results (Bastiaans, 1993; Ayres et al., 1996), it was found that HI was modified for wheat crops subjected to post-anthesis epidemics. The results suggest that under the present experimental conditions, the earlier the diseases occurred, the more the HI decreased, most noteworthy in 1999 when both the number of grains and grain weight decreased (Robert et al., 2004). Yield was well related to total biomass accumulation during grain filling, explaining 91 % of yield variability. The slope was not significantly different from 1, pointing out that post-anthesis carbohydrate allocation to grains was not affected by leaf diseases, whether or not epidemics decreased grain number. Moreover, the positive y-axis intercept explained why HI decreased as post-anthesis growth decreased due to diseases. When directly calculated from the balance sheet of WSCs and protein, the reserve utilization completely accounted for the observed intercept between yield and post-anthesis growth, and it even slightly increased the quality of the relationship from R2 = 0·91 to R2 = 0·94. Despite large and highly significant variations in reserve mobilization between treatments, late epidemics under the present conditions did not modify the overall quantity of pre-anthesis reserves allocated to grains. This result clearly contradicts previous literature suggesting that temporary reserves are used more efficiently in adverse conditions (Schnyder, 1993; Ayres et al., 1996), although in some cases they may contribute to genotype tolerance to diseases (Gaunt and Wright, 1992). According to the present data, the contribution of pre-anthesis reserves to grain yield as a percentage of final yield was unchanged overall, showing that their relative contribution to grain yield did not increase systematically under adverse conditions. This finding also contradicts previous observations made by Gallagher et al. (1975). In the present case, depending on year, the contribution sometimes increased (treatments A, H and K), decreased (treatments M and J) or remained unchanged (other treatments). The overall stability of reserve amount distributed to grains could be explained by an opposite pattern of pre-anthesis sugar and protein mobilization. Post-anthesis WSC reallocation to grains was more efficient in the case of diseased crops, as previously proposed by Ayres et al. (1996); on the contrary, leaf diseases inhibited translocation of nitrogen from leaves to grains, leading to vegetative parts less depleted of nitrogen at the final harvest, which is true in many pathosystems (Vereet and Hoffmann, 1987; Bastiaans, 1993; Kremer and Hoffmann, 1993; Garry et al., 1996). A complex modification of reserve use by leaf diseases appears to be masked by the apparent simplicity of pre-anthesis reserve contribution to yield.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSLITERATURE CITED |
|---|
This study clearly shows that a simple model of growth and yield involving wheat attacked by leaf rust and STB can be derived from (a) an estimation of healthy crop RUE, which accounts for other stresses that were minimized in the present experiments; (b) a single relationship between growth and yield, in the case of the given genotype, providing that a constant amount of pre-anthesis reserves is remobilized to grains, whatever the epidemic; and (c) a proper estimation of aPAR per treatment, which needs to account for the green leaf area per layer over time as related to disease epidemics. Parker et al. (2004) suggested that further attention should be paid to genotype variability related to reserve pool constitution and utilization as well as green canopy size during grain filling. Having a simple way to model growth and yield opens the path to separating the effect of crop structure from that of reserve management on the tolerance of wheat genotypes to leaf diseases.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSLITERATURE CITED |
|---|
The authors gratefully acknowledge J. Troizier and his team for technical help, J. Jean-Jacques and B. Le Fouillen for work in plant sample manipulations, F. Lafouge for plant analysis, C. Lannou and P. Bancal for fruitful scientific discussions, and S. Tanis-Plant for thorough editorial advice in English. This study was funded by the Institut National de la Recherche Agronomique (INRA).
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSLITERATURE CITED |
|---|
-
Ayres PG, Colin Press M, Spencer-Phillips PTN. Effects of pathogens and parasitic plants on source–sink relationships. In: Photo assimilate distribution in plants and crops—Zamski E, Schaffer AA, eds. (1996) New York: Marcel Dekker. 479–500.
Bassanezi RB, Amorim L, Filho AB, Hau B, Berger RD. Accounting for photosynthetic efficiency of bean leaves with rust, angular leaf spot and anthracnose to assess crop damage. Plant Pathology (2001) 50:443–452.[CrossRef][Web of Science]
Bastiaans L. Ratio between virtual and visual lesion size as a measure to describe reduction in leaf photosynthesis of rice due to leaf blast. Phytopathology (1991) 81:611–615.[CrossRef][Web of Science]
Bastiaans L. Effects of leaf blast on growth and production of a rice crop. 1. Determining the mechanism of yield reduction. Netherlands Journal of Plant Pathology (1993) 99:323–334.[CrossRef][Web of Science]
Bastiaans L, Roumen EC. Effects on leaf photosynthetic rate by leaf blast for rice cultivars with different types and levels of resistance. Euphytica (1993) 66:81–87.[CrossRef][Web of Science]
Béasse C, Ney B, Tivoli B. A simple model of pea (Pisum sativum) growth affected by Mycosphaerella pinodes. Plant Pathology (2000) 49:187–200.[Medline]
Boote KJ, Jones JW, Mishoe JW, Berger RD. Coupling pests to crop simulators to predict yield reductions. Phytopathology (1983) 73:1581–1587.[Web of Science]
Bryson RJ, Sylvester-Bradley R, Scott RK, Paveley ND. Reconciling the effects of yellow rust on yield of winter wheat through measurements of green leaf area and radiation interception. Aspects of Applied Biology (1995) 42:9–18.
Bryson RJ, Paveley ND, Clark WS, Sylvester-Bradley R, Scott RK. Use of in-field measurements of green leaf area and incident radiation to estimate the effects of yellow rust epidemics on yield of winter wheat. European Journal of Agronomy (1997) 7:53–62.[CrossRef][Web of Science]
Cornish PS, Baker GR, Murray GM. Physiological responses of wheat (Triticum aestivum) to infection with Mycosphaerella graminicola causing Septoria tritici blotch. Australian Journal of Agricultural Research (1990) 41:317–327.[CrossRef][Web of Science]
Dumas JBA. Procédés de l'analyse organique. Annales de Chimie et de Physique (1831) 247:198–213.
Erickson JE, Stanosz GR, Kruger EL. Photosynthetic consequences of Marssonina leaf spot differ between two poplar hybrids. New Phytologist (2004) 161:577–583.[CrossRef][Web of Science]
Gallagher JN, Biscoe PV, Scott RK. Barley and its environment. Journal of Applied Ecology (1975) 12:319–336. V. Stability of grain weight.[CrossRef][Web of Science]
Garry G, Tivoli B, Jeuffroy MH, Citharel J. Effects of Ascochyta blight by Mycosphaerella pinodes on the translocation of carbohydrates and nitrogenous compounds from the leaf and hull to the seed of dried-pea. Plant Pathology (1996) 45:769–777.[CrossRef][Web of Science]
Garry G, Jeuffroy MH, Ney B, Tivoli B. Effects of Ascochyta blight (Mycosphaerella pinodes) on the photosynthesizing leaf area and the photosynthetic efficiency of the green leaf area of dried-pea (Pisum sativum). Plant Pathology (1998) 47:473–479.[CrossRef][Web of Science]
Gaunt RE, Wright AC. Disease–yield relationship in barley. 2. Contribution of stored stem reserves to grain filling. Plant Pathology (1992) 41:688–701.[CrossRef][Web of Science]
Hay RKM, Walker AJ. Interception of solar radiation by the crop canopy. In: An introduction to the physiology of crop yield.—Hay RKM, Walker AJ, eds. (1989) New York: Longman Scientific and Technical. 7–29.
Johnson KB. Defoliation, and growth: a reply. Phytopathology (1987) 77:1495–1497.[Web of Science]
Jones HG. Photosynthesis and respiration. Plants and microclimate—Jones HG, ed. (1994) Cambridge: Cambridge University Press. 163–214.
de Kraker J, van den Ende JE, Rossing WAH. Control strategies with reduced fungicide input for Botrytis leaf blight in lily – a simulation analysis. Crop Protection (2005) 24:157–165.[CrossRef][Web of Science]
Kremer M, Hoffman GM. Effekte von Blattinfektionen durch Drechlera tritici-repentis auf den Kohlenhydrat- und Stickstoffhaushalt von Weizenpflanzen. Journal of Plant Diseases and Protection (1993) 100:259–277.
Leitch MH, Jenkins PD. Influence of nitrogen on the development of Septoria epidemics in winter wheat. Journal of Agricultural Science (1995) 124:361–368.[Web of Science]
LeMay C, Schoeny A, Tivoli B, Ney B. Improvement and validation of pea crop growth model to simulate the growth of cultivars infected with Ascochyta blight. European Journal of Plant Pathology (2005) 112:1–12.[CrossRef][Web of Science]
Lopes DB, Berger RD. The effects of rust and anthracnose on photosynthetic competence of diseased bean leaves. Phytopathology (2001) 91:212–220.[Medline]
Lovell DJ, Parker SJ, Petghem PV, Webb DA, Welham SJ. Quantification of raindrop kinetic energy for improved prediction of splash-dispesed pathogens. Phytopathology (2002) 92:497–503.[Medline]
Madden LV, Nutter FW. Modeling crop losses at the field scale. Canadian Journal of Plant Pathology (1995) 17:124–137.
Monsi M, Saeki T. Uber der lichtfaktor den pflanzengesellschaften und seine bedeutung für die stoffproduktion. Japanese Journal of Botany (1953) 14:22–52.
Monteith JL. Climate and the efficiency of crop production in Britain. Philosophical Transactions of the Royal Society B: Biological Sciences (1977) 281:277–294.[CrossRef]
Oerke EC, Dehne HW. Global crop production and the efficacy of crop protection–current situation and future trends. European Journal of Plant Pathology (1997) 103:203–215.[CrossRef][Web of Science]
Parker SR, Welham S, Paveley ND, Foulkes J, Scott RK. Tolerance of Septoria leaf blotch in winter wheat. Plant Pathology (2004) 53:1–10.[CrossRef][Web of Science]
Paveley ND, Lockley D, Vaughan TB, Thomas J, Schmidt K. Predicting effective fungicide doses through observation of leaf emergence. Plant Pathology (2000) 49:748–766.[CrossRef][Web of Science]
Peterson RF, Campbell AB, Hannah AE. A diagrammatic scale for estimating rust intensity on leaves and stems of cereals. Canadian Journal of Research (1948) 26:496–500.
Rabbinge R, Jorritsma ITM, Schans J. Damage components of powdery mildew in winter leaf. Netherlands Journal of Plant Pathology (1985) 91:235–247.[CrossRef][Web of Science]
Rabbinge R, Ward SA, van Laar HH. Simulation and system management in crop protection (1989) Wageningen, The Netherlands: Simulation Monographs, PUDOC.
Robert C. Etude et modélisation du fonctionnement d'un couvert de blé attaqué par le complexe parasitaire Puccinia trticina–Micosphaerella graminicola (2003) Thèse de 3ème cycle de l'Institut National Agronomique Paris-Grignon.
Robert C, Bancal M-O, Nicolas P, Lannou C, Ney B. Analysis and modelling effects of leaf rust and Septoria tritici blotch on wheat growth. Journal of Experimental Botany (2004) 55:1079–1094.
Robert C, Bancal M-O, Ney B, Lannou C. Changes in wheat leaf photosynthesis due to leaf rust, with respect to lesion development and leaf nitrogen status. New Phytologist (2005) 165:227–241.[CrossRef][Web of Science][Medline]
Robert C, Bancal M-O, Lannou C, Ney B. Quantification of the effects of Septoria tritici blotch on wheat leaf gas exchange with respect to lesion age, leaf number, and leaf nitrogen status. Journal of Experimental Botany (2006) 57:225–234.
Rossing WAH, Daamen RA, Jansen MJW. Uncertainty analysis applied to supervised control of aphids and brown rust in winter wheat. Agricultural Systems (1994) 44:449–460. Part 2. Relative importance of different componets of uncertainty.[CrossRef][Web of Science]
Savary S, Teng PS, Willocquet L, Nutter FW. Quantification and modelling crop losses: a review of purposes. Annual Review of Phytopathology (2006) 44:89–112.[CrossRef][Web of Science][Medline]
Schnyder H. The role of carbohydrate storage and redistribution in the source–sink relations of wheat and barley during grain filling – a review. New Phytologist (1993) 123:233–245.[CrossRef][Web of Science]
Scholes JD, Rolfe SA. Photosynthesis in localised regions of oat leaves infected with crown rust (Puccinia coronata): quantitative imaging of chlorophyll fluorescence. Planta (1996) 199:573–582.[Web of Science]
Seck M, Roelfs AP, Teng PS. Influence of leaf position on yield loss caused by wheat leaf rust in single tillers. Crop protection (1991) 10:222–228.[CrossRef][Web of Science]
Shtienberg D. Effects of foliar diseases of wheat on the physiological processes affecting yield under semi-arid conditions. Plant Pathology (1991) 40:533–541.[CrossRef][Web of Science]
Snoeijers SS, Perez-Garcia A, Joosten MHAJ, De Wit JGM. The effect of nitrogen on disease development and gene expression in bacterial and fungal plant pathogens. European Journal of Plant Pathology (2000) 106:493–506.[CrossRef][Web of Science]
Spitters CJT, Van Roermund HJW, Van Nassau HGMG, Schepers J, Mesdac J. Genetic variation in partial resistance to leaf rust in winter wheat: disease progress, foliage senescence and yield reduction. Netherlands Journal of Plant Pathology (1990) 96:3–15.[CrossRef][Web of Science]
Teng PS. Construction of predictive models: 2. Forecasting crop losses. Advances in Plant Pathology (1985) 3:179–206.
von Tiedemann A. Single and combined effects of nitrogen fertilization and ozone on fungal leaf diseases on wheat. Journal of Plant Diseases and Protection (1996) 103:409–419.
Varlet-Grancher C, Bonhomme R, Chartier M, Artis P. Efficience de conversion de l'énergie solaire par un couvert végétal. Acta Oecologia Ecologica Plantarum (1982) 17:3–26.
Verreet JA, Hoffmann GM. Effects of infection by Septoria nodorum at different development stages of wheat on the level of production. Plant Diseases and Protection (1987) 94:283–300.
Waggoner PE, Berger RD. Defoliation, disease, and growth. Phytopathology (1987) 77:393–398.[Web of Science]
Walton S, Gallagher MW, Duyzer JH. Use of a detailed model to study the exchange of NOx and O3 above and below a deciduous canopy. Atmospheric Environment (1997) 31:2915–2931.[CrossRef]
Zadoks JC. On the conceptual basis of crop loss assessment: the threshold theory. Annual Review of Phytopatholology (1985) 23:455–473.[CrossRef]
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]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||