Predicting yield losses caused by lodging in wheat

Predicting yield losses caused by lodging in wheat

Field Crops Research 137 (2012) 19–26 Contents lists available at SciVerse ScienceDirect Field Crops Research journal homepage: www.elsevier.com/loc...

494KB Sizes 2 Downloads 132 Views

Field Crops Research 137 (2012) 19–26

Contents lists available at SciVerse ScienceDirect

Field Crops Research journal homepage: www.elsevier.com/locate/fcr

Predicting yield losses caused by lodging in wheat P.M. Berry a,∗ , J. Spink b a b

ADAS High Mowthorpe, Duggleby, Malton, North Yorkshire YO17 8BP, UK Teagasc Crops, Environment and Land Use Programme, Oak Park Crops Research Centre, Carlow, Ireland

a r t i c l e

i n f o

Article history: Received 17 September 2011 Received in revised form 29 June 2012 Accepted 27 July 2012 Keywords: Wheat Lodging Yield Canopy photosynthesis Respiration Radiation

a b s t r a c t Lodging is a major limiting factor for wheat (Triticum aestivum L.) production, yet few studies have investigated the mechanism by which it reduces yield. This paper tests the hypothesis that lodging-induced yield losses in wheat can be predicted by calculating the reduction in canopy photosynthesis that results from lodging-induced changes to the architecture of the canopy. An existing model of canopy photosynthesis has been further developed to account for the effect of lodging-induced changes to the canopy architecture on photosynthesis and grain yield. The model predicted that lodging at 90◦ from the vertical will reduce yield by approximately 61%. The ability of the model to predict lodging-induced yield losses was tested against observations made in three separate field experiments. The model predicted 71% of the variation in the proportion of yield lost due to lodging (YLOSS ) and the best-fit line was not significantly different from the 1:1 relationship. Sensitivity analysis showed that the proportion of yield lost was relatively insensitive to the model parameters. As a result it was shown that a simplified model could

f

(L90 × 0.7 + L65 × 0.3 + L25 × 0.1)/n In this be employed without losing predictive accuracy. YLOSS = i equation i and f are the 1st and last days of grain filling, L90 is the proportion of crop area lodged at 85–90◦ from the vertical, L65 is the proportion of crop area lodged between 46◦ and 84◦ , L25 is the proportion of crop area lodged between 5◦ and 45◦ and n is the number of days of grain filling. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Structural failures that permanently displace the shoots of small grained cereals from their vertical positions are known as lodging (Pinthus, 1973). In wheat (Triticum aestivum L.), lodging can occur by buckling of the stem (Neenan and Spencer-Smith, 1975) or by failure of the root–soil complex (Crook and Ennos, 1993). Lodging reduces grain yield and causes several knock-on effects including reduced grain quality, greater drying costs and slower harvest. It is a problem that limits cereal productivity in both developed and developing countries. In the UK it has been estimated that lodging in winter wheat costs the farming industry $80M per year (Berry et al., 1998). Experiments with natural and artificially-induced lodging have measured yield losses caused by lodging which range between 0 and 80% (Fischer and Stapper, 1987; Weibel and Pendleton, 1964; Laude and Pauli, 1956; Mulder, 1954; Stapper and Fischer, 1990; Easson et al., 1993). The wide range of yield losses reported is probably due to a number of factors including the stage of development at which lodging occurs, the angle of shoot displacement, possibly whether the lodging is natural or artificially induced and genotype

∗ Corresponding author. E-mail address: [email protected] (P.M. Berry). 0378-4290/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.fcr.2012.07.019

(Acreche and Slafer, 2011). Artificial lodging at the ear emergence, milk, soft dough and hard dough stages reduced yield by 31%, 25%, 20% and 12%, respectively (Weibel and Pendleton, 1964). Stapper and Fischer (1990) showed that the quantity of yield loss could be related to the number of days lodged during the grain filling period. Fischer and Stapper (1987) showed that stems lodged at 45◦ resulted in less yield loss than stems lodged at 80◦ . Understanding the mechanism by which lodging reduces yield would help to explain and predict lodging-induced yield losses. However, little research has been carried out on wheat to understand the mechanism by which lodging reduces yield. Several mechanisms have been postulated for rice. These have been reviewed by Hitaka (1968a,b) and include reduced translocation of mineral nutrients and carbon for grain filling, increased respiration, reduced carbon assimilation within the canopy, rapid chlorosis and greater susceptibility to pests and diseases. Setter et al. (1997) showed that lodging reduces the yield of rice by self shading and reducing canopy photosynthesis. Currently there is no consensus about the most likely mechanism of yield loss in wheat. It seems plausible that lodging-induced yield losses in wheat result entirely from inefficient radiation use by the canopy. Lodging at 90◦ from the vertical causes leaves to become horizontal. This reduces the area of leaf that is sunlit and increases the flux density of radiation on the sunlit leaves. The resulting increase in photosynthesis of the sunlit leaves is proportionately smaller

20

P.M. Berry, J. Spink / Field Crops Research 137 (2012) 19–26

than the increase in flux density due to the diminishing response of photosynthesis to increasing radiation levels (Choudhury, 2000). Lodging also compresses the canopy and results in more leaves resting directly upon each other. This reduces the area of leaf exposed to diffuse radiation (i.e. radiation on a cloudy day) and reduces photosynthesis. As a result the rate of photosynthesis carried out by the whole canopy of a lodged crop will be reduced. The extent to which a reduction in canopy photosynthesis reduces yield will depend on the level of buffering from the relocation of water soluble carbohydrate from the stem to the grain and on the extent to which the crop is source or sink limited with smaller yield reductions possible in sink limited crops (Zhang et al., 2010). This paper tests the hypothesis that lodging-induced yield losses in wheat can be predicted by calculating the reduction in canopy photosynthesis that results from lodging-induced changes to the architecture of the canopy. An existing model of canopy photosynthesis which accounts for the different properties of direct and diffuse radiation (Campbell and Norman, 1998a) is developed further to account for the effect of lodging-induced changes to the canopy on grain yield. The following section presents an outline of the methods used to develop the model. This is followed by a sensitivity analysis in which the model parameters are systematically varied to demonstrate their individual effects on yield in the presence and absence of lodging. The ability of the model to predict lodging-induced yield losses is then tested against observations made in three separate field experiments. Finally, the applicability of the model and the implications of its results are discussed. 2. Modelling the effect of lodging on yield This section briefly describes an existing model of canopy photosynthesis (Campbell and Norman, 1998a) and explains how this can be developed to account for lodging-induced changes to canopy architecture. It then describes how the output of this model is used to estimate grain yield after accounting for respiratory costs and the remobilisation of carbohydrate stored in the stem.

horizontal canopy, x approaches infinity. Although x is also influenced by other attributes, for simplicity from this point forward x is abbreviated to ‘leaf angle’. The extinction coefficient for diffuse PAR (kd ) is affected by leaf angle (x) and the green area index (L) (Campbell and Norman, 1998a). For a canopy with L = 5, kd increases from 0.6 to 0.7 as x increases from 0.5 to 1.0. For a canopy with x = 0.5, kd decreases from 0.6 to 0.5 as L increases from 2 to 7. 2.3. The interception of diffuse and beam PAR Diffuse light is assumed to spread over all the leaves (shaded and sunlit) and is attenuated by kd . The flux density of diffuse PAR on the leaves (Qsh ) is calculated by summing the diffuse flux density on the leaves that arises from the sky (Qd ) and from scattering of beam radiation (Qsc ). Qd is given by (Campbell and Norman, 1998a): √ Q [1 − exp(− ˛kd L)] Qd = od (2) √ ˛kd L where Qod is the flux density of diffuse radiation above the canopy and ˛ is the absorptivity of the leaves (0.8). Qsc is the difference between the unintercepted beam radiation (Qb ) and the unintercepted beam plus down scattered beam (Qbt ). Qob is the flux density of beam radiation on a horizontal surface at the top of the canopy. √ Qbt = Qob exp(− ˛kbe L)

(3)

Qb = Qob exp(−kbe L)

(4)

The concentration of light on the sunlit leaves (Qsl ) is given by Qsl = Kbe Qob + Qsh The sunlit leaf area (L*) is given by 1 − exp(−kbe L) kbe

L∗ =

2.1. Canopy photosynthesis model Photosynthetically active radiation (PAR) from both direct and diffuse sources must be considered when calculating the amount of photosynthesis carried out by plant canopies. Direct radiation from the sun is referred to as beam radiation. Diffuse radiation is the result of beam radiation being transmitted through cloud or reflected, scattered and transmitted by foliage. It can be thought of as many beams coming from all directions. The flux density of PAR (␮mol photons m−2 s−1 ) above the canopy which is either beam (Qob ) or diffuse (Qod ) can be calculated from first principles using information about the latitude, altitude, time of year, time of day and cloud cover (e.g. Campbell and Norman, 1998b). The model described here uses hourly information about the PAR due to the potential importance of solar angle on radiation use by the plant.

2.4. Photosynthesis Choudhury (2000) proposed a model describing, at the leaf scale, the relationship between the flux density of PAR on a leaf (Qsl or Qsh ) and the rate of photosynthesis (Al ). The light response curve for this model is represented by a non-rectangular hyperbola



A1 =

Am (1 + ) − {(1 + )2 − 4}

where

The attenuation of PAR through the plant canopy is measured by the extinction coefficient. For beam PAR, the extinction coefficient (kbe ) is calculated from Eq. (1).

=

kbe =

x2

+ tan2

x + 1.774((x + 1.82)−0.733 )

(1)

where x is the mean ratio of vertical to horizontal projections of leaves and is the solar angle from the vertical. For a spherical leaf angle distribution, x equals 1 (i.e., vertical and horizontal projections are equal); for a vertical distribution, x equals 0; and for a

(6)

The shaded leaf area is then found by subtracting L* from L. Eqs. (1)–(6) therefore calculate the flux density of PAR on different areas of leaves. The next step is to calculate the photosynthetic rate of the canopy.

2.2. Extinction coefficient



(5)

2

ε˛Qsl(sh) Am

0.5

 (7)

(8)

In these equations Am is the maximum rate of photosynthesis, ε is the quantum efficiency (0.06) and  is the empirical curvature factor (0.8). ε is a physical constant which varies by a relatively small amount with temperature. The gross canopy photosynthesis (␮mol CO2 m−2 s−1 ) is calculated by multiplying the photosynthetic rates by the areas of sunlit and shaded leaf areas. Integration of these calculations over the whole day then gives the daily rate of gross photosynthesis.

P.M. Berry, J. Spink / Field Crops Research 137 (2012) 19–26

21

calculated in terms of the nitrogen content per unit ground area (mmol m−2 ground) canopy (nc ) and roots (nr ). Root respiration is estimated to be double that of the stems and leaves (Amthor, 1989). Rm = 0.00021(nc + 2nr )2(T −20)/10

Fig. 1. Schematic diagram illustrating (a) unlodged plants, (b) partially lodged plants and (c) fully lodged plants.

2.5. Lodging-induced changes to canopy architecture A crop lodged at 90◦ from the vertical has entirely horizontal leaves and the upper leaves rest directly upon the lower leaves (Fig. 1). The proportion of PAR transmitted through leaves has been estimated at only 0.05 (Monteith, 1965) and for the purposes of this model it is assumed that zero PAR is transmitted through leaves. This means that for a crop lodged at 90◦ from the vertical only a proportion of its leaves will be exposed to direct or diffuse radiation. The green area index exposed to radiation is calculated from 1 − e−L . This assumes that the flat leaves of a fully lodged crop are randomly distributed and the extinction coefficient of beam and diffuse radiation for the green area index exposed to radiation is one. This method of analysing the effect of lodging on the canopy indicates that lodging causes leaves to become more horizontal, increases the extinction coefficient of the canopy and reduces the green area index which is exposed to direct or diffuse radiation. It is assumed that for intermediate angles of lodging between 1◦ and 89◦ (Fig. 1) the green area index which is exposed to direct or diffuse radiation will change in direct proportion to the amount by which the canopy has been compressed. Canopy compression is measured in terms of the vertical space occupied by the canopy. The vertical space occupied by the canopy of a lodged crop can be calculated as a proportion of the unlodged crop from the cosine of the angle of lodging from the vertical. Using this method the green area index exposed to radiation may be calculated from Lds = Lds − ((1 − cos) × (Lds − Lds90 ))

(9)

where Lds is the unlodged green area index, Lds is the green area index exposed to diffuse or direct radiation at angle of lodging  and Lds90 is the green area index exposed to diffuse or direct radiation for a crop lodged at 90◦ from the vertical. 2.6. Calculating yield Respiration uses at least 30% of the assimilates produced by photosynthesis (Hay and Walker, 1989). Choudhury (2000) outlined a simple model to calculate respiration arising from maintenance (Rm ) and growth (Rg ) separately. Since much of maintenance respiration is to support the turnover of protein, this component is

(10)

Choudhury (2000) estimates that root N is about 14% of the canopy nitrogen. The final step of this calculation is to account for the effect of temperature (T) on the rate of respiration. Rg is calculated from the efficiency with which the gross photo-assimilate (minus the maintenance respiration) is converted into growth. This efficiency has been measured at 74% for wheat (Choudhury, 2000). The daily rate of dry matter growth during grain filling has been calculated using the mean environmental conditions for the UK, GAI and lodging during the grain filling period to allow the model to be as simple as possible. It is recognised that using a mean lodging severity and GAI may reduce the accuracy of the model due to the non-linear relationship for both lodging severity and GAI with light interception. Daily rate of dry matter growth was then multiplied by the number of days of grain filling and added to the amount of dry matter relocated to the grain from reserves of carbohydrate accumulated predominantly in the stem before flowering. UK wheat crops accumulate about 3 t ha−1 of water soluble carbohydrate by flowering (Foulkes et al., 1998) of which 40–70% is relocated to the grain under non-drought conditions (Austin et al., 1977; Bidinger et al., 1977; Gebbing et al., 1999; Yang et al., 2001). The percentage of water soluble stem carbohydrate relocated under non-drought compared with drought conditions has been measured at 50% and 83%, respectively (Yang et al., 2001) and 60% and 80%, respectively (Bidinger et al., 1977). Asseng and van Herwaarden (2003) concluded that a greater percentage of water soluble carbohydrate is relocated to the grain when there is either a high sink demand or a low photosynthetic rate during grain filling. It therefore seems likely that a low rate of photosynthesis caused by lodging would result in a greater percentage of water soluble carbohydrate relocated to the grain, although it is difficult to quantify what level of increase may be expected given that direct measurements of water soluble carbohydrate relocation under lodged and unlodged conditions have not been made. A conservative increase in water soluble carbohydrate relocation in lodged conditions has therefore been assumed for this model rising from 50% for an unlodged crop to 70% for a crop lodged at 90◦ past vertical for the entire grain filling period. For intermediate levels of lodging the percentage of water soluble carbohydrate relocated was inversely and linearly related to the reduction in daily dry matter growth from current photosynthesis caused by the lodging. 3. Materials and methods 3.1. Experiments The model of yield losses is tested with data from three experiments carried out at ADAS Rosemaund, Hereford, UK (grid reference SO5647, 52.1◦ N), on a well drained silt clay loam (Bromyard series). In the summer of 1996, plots were artificially lodged in one experiment (A96) and natural lodging was recorded in another (N96). Natural lodging was also recorded in an experiment done in 1998 (N98). The artificial lodging experiment (A96) was drilled with cv. Riband at 375 seed m−2 in 24 m × 2 m plots on 16th October 1995. Treatments were arranged as a randomised block design with artificial lodging engineered on five occasions (GS59 on the 17th June, GS65 on the 27th June, GS75 on the 17th July, GS83 the 2nd August and on the day of grain harvest on the 2nd September) and three unlodged treatments, with five replicates. Artificial lodging was induced by pushing the plants through 90◦ in the direction

22

P.M. Berry, J. Spink / Field Crops Research 137 (2012) 19–26

of drilling using a board measuring 220 cm × 15 cm. The experiment received nitrogen fertiliser, ammonium nitrate granules (34.5% Nitrogen), applied at 51 kg N ha−1 on 20th March 1996, 50 kg N ha−1 on 22nd April 1996 and 78 kg N ha−1 on 17th May 1996, and a plant growth regulator application (PGR) of Stefes CCC 700 (700 g L−1 chlormequat) at 2.5 L ha−1 on 28th April 1996. Grain yield of the whole plot was determined with a small plot combine harvester on 2nd September 1996. Natural lodging experiment (N96) was drilled with cv Mercia on 20th September 1995. It was sown into soil with a high level of residual N (116 kg N ha−1 in February 1996) at 500 and 250 seeds m−2 with or without an early PGR, New 5C Cycocel (645 g L−1 chlormequat + 32 g L−1 choline chloride) applied at GS31 (Tottman, 1987). Each treatment was replicated three times. Full details of the husbandry of this experiment are given in Berry et al. (2000). Lodging was assessed regularly as the percentage area of each plot which was either leaning between 5◦ and 45◦ , leaning between 46◦ and 85◦ or lodged flat between 85◦ and 90◦ . The area combined for grain yield determination was approximately 10 m × 2.25 m. Natural lodging experiment (N98) was sown with cvs Cadenza, Haven, Soissons and Spark on 23rd September 1997 at 20, 40, 80, 160, 320 and 640 seeds m−2 . Each treatment was replicated three times. The crop husbandry for these crops is described by Whaley et al. (2000). Lodging was assessed regularly as the percentage area of each plot which was either leaning between 5◦ and 85◦ , or lodged flat. The area combined for grain yield determination was approximately 10 m × 2.25 m. 3.2. Calculation of yield losses due to lodging Lodging induced yield losses were calculated by subtracting the observed yield from the yield in the absence of lodging. For A96 the unlodged yield for each lodged plot was taken as the yield of the plots lodged on the day of harvest to ensure that any yield differences were not due to incomplete combine recovery. For N96 the unlodged yields were taken as the unlodged treatments sown in the same block, but at the lower seed rate and treated with chlormequat. Lowering the seed rate from 500 seeds m−2 to 250 seeds m−2 and using chlormequat have been shown to have no effect on yield in the absence of lodging (Whaley et al., 2000; Berry et al., 2004). The potential yield for each lodged plot in N98 was estimated from the unlodged yields of the lower seed rate treatments (belonging to the same block and cultivar). This was done by using the relationship of grain yield (t ha−1 ) (Y) to plants m−2 (P) (Whaley et al., 2000). Y = a + b × r P − cP

(11)

This relationship has been calculated for wheat experiments at ADAS Rosemaund which included N98. Whaley et al. (2000) showed that the different cultivars altered parameter a, but

parameters b, r and c remained the same at −4.324, 0.9617 and 0.00022, respectively with an R2 value of 0.75 for the fitted curves (P < 0.001). The response curve shows yield reaches a maximum between 100 and 200 plants m−2 then either remains constant or declines with greater plant populations. The yield decline is generally caused by lodging. Parameter a is the asymptote and is taken here to represent the potential yield in the absence of lodging. The unlodged yields and plant populations of treatments at seed rates 20, 40, 80 and occasionally 160 seeds m−2 have been used to estimate the a parameter for each block/cultivar combination. The mean of at least three estimates of a have been used as the potential yield for each block/cultivar combination. 4. Results 4.1. Effect of angle of lodging Typical values for some of the key environmental and crop model parameters for a UK wheat crop are described in Table 1. Environmental parameters are a mean over the grain filling period, which is assumed to begin on 20th June and last for 40 days. The long-term mean PAR in England over this period is 8 MJ per day with six hours of sun per day. This level of daily PAR is achieved for the model by assuming six hours of sun with cloud reducing the beam PAR by 50% at all other times. The amount by which cloud reduces beam PAR depends on cloud type with cirrus causing reductions of 10–20% and nimbostratus causing reductions of 90% (Campbell and Norman, 1998b). For this study it has been assumed that cloud reduces beam PAR by 50% which would be typical of altocumulus. The crop parameters are for the mid-point of grain filling. When the standard crop and environmental parameter values described in Table 1 were used the model predicted a grain yield at 100% dry matter of 10.32 t ha−1 for an unlodged crop. Lodging at 90◦ from the vertical was predicted to reduce yield to 3.99 t ha−1 (Fig. 2). The amount of yield lost was predicted to increase exponentially per degree of lodging. Lodging from the vertical position to 10◦ was predicted to reduce yield by 0.10 t/ha whereas lodging from 80◦ to 90◦ reduced yield by a further 1.22 t ha−1 . 4.2. Sensitivity analysis The effect of varying each model parameter independently about its range (described in Table 1) on the yield of wheat when unlodged or lodged at 90◦ is described in Table 2. Standard values were used for the non varying model parameters. The ranges of the environmental parameters are typical annual variations that have been observed in the main wheat growing regions of the UK. Green area index at the mid-point of grain filling varies between about 2 and 4 (Sylvester-Bradley et al., 1997). Canopy nitrogen content may vary from 150 to 250 kg N ha−1 (1050–1750 ␮mol N m−2 ) (Sylvester-Bradley et al., 1997). The

Table 1 Model parameter values expected for UK wheat during grain filling. Model parameters

Standard value

Range

Source

Mean incident radiation (MJPAR m−2 d−1 ) a Mean daily temperature (T) (◦ C) b Green area index (L) b Maximum rate of photosynthesis (Am ) (␮mol CO2 m−2 s−1 ) b Canopy nitrogen content (mmol m−2 ground) (nc ) b Leaf angle (x) Contribution to grain yield from stem reserves (t ha−1 )

8 16 3 16 1400 0.5 1.5

6–10 13–19 2–4 12–20 1050–1750 0.2–1.5 1.2–2.1

Un Met S97 A82, G81 S97 T88, C01 A77, Y01

a

a

Mean over 40 day grain filling period taken as 20th June to 30th July. Value at mid-point of grain filling. Un – Unpublished data from UK meteorological stations; Met – www.metoffice.org; S97 – Sylvester-Bradley et al. (1997); A82 – Austin et al. (1982); G81 – Gregory et al., 1981; T88 – Thorne et al. (1988); C01 – Critchley (2001); A77 – Austin et al. (1977); Y01 – Yang et al. (2001). b

P.M. Berry, J. Spink / Field Crops Research 137 (2012) 19–26

23

Table 2 Effect of minimum and maximum parameter on yields of upright and lodged crops (t ha−1 ). Parameter

Minimum value

Mean incident radiation (MJPAR m−2 d−1 ) Mean daily temperature (T) (◦ C) a Green area index (L) Maximum rate of photosynthesis (Am ) (␮mol CO2 m−2 s−1 ) Canopy nitrogencontent (mmol m−2 ground) (nc ) a Leaf angle (x) Contribution to grain yield from stem reserves (t/ha) a

Maximum value

Zero lodging

90◦ lodging

Proportion yield loss

Zero lodging

90◦ lodging

Proportion yield loss

8.75 10.80 7.46 8.75 10.96 10.62 10.02

3.39 4.46 3.71 3.24 4.62 3.99 3.69

0.61 0.59 0.50 0.63 0.58 0.62 0.63

11.06 9.74 12.73 11.51 9.69 9.18 10.92

4.26 3.40 4.09 4.59 3.35 3.99 4.59

0.61 0.65 0.68 0.60 0.65 0.57 0.58

Includes linked effects on kd .

maximum rate of photosynthesis of wheat leaves has been measured to vary between 12 ␮mol m−2 s−1 (Gregory et al., 1981) and 20 ␮mol m−2 s−1 (Austin et al., 1982). The extinction coefficient for wheat has been measured to vary from 0.46 to over 0.6 (Thorne et al., 1988; Critchley, 2001). This indicates that x (leaf angle) may vary from less than 0.5–1.5. The yield of an unlodged crop was affected most by changes in green area index. Yield was predicted to increase from 7.46 t ha−1 for a crop with a green area index of 2 at mid-grain filling to 12.73 t ha−1 for a crop with a green area index of 4. Incident radiation and the maximum rate of photosynthesis also had a large influence on the yield of an unlodged crop. As expected yield was increased by greater incident radiation and maximum rate of photosynthesis. Changes to temperature, canopy nitrogen content and leaf angle had smaller effects on the yield of an unlodged crop. Greater temperature and canopy nitrogen content were predicted to reduce grain yield as a result of greater respiration. Increasing leaf erectness was predicted to increase yield. The proportion of yield lost when the crop was lodged at 90◦ changed only slightly when each parameter was varied about its range (Table 2). The proportion of yield lost varied from about 0.50 for crops with a small green area index to 0.65–0.68 for crops experiencing high temperatures or with a large green area index or large nitrogen content. Crops with large green area indices were predicted to lose proportionately more yield when they lodge. This is because the unlodged yield was high, but the lodged yield was similar to that of a crop with a smaller green area index. High temperatures and crops with a high nitrogen content were predicted to lose more yield when they lodge due to high maintenance respiration costs which do not change when the crop lodges.

12

Yield (t/ha)

10 8 6 4 2

0 0

15

30

45

60

75

90

Angle of lodging from vertical Fig. 2. The effect of angle of lodging on grain yield at 100% dry matter as predicted by the model of canopy photosynthesis.

At this point it is important to recognise that some of the crop parameters may not vary independently in practice. For example crops with a large green area index often have a high nitrogen content. An unlodged crop with a green area index of 4 and a nitrogen content of 1750 mmol m−2 had a predicted unlodged yield of 12.03 t ha−1 compared with 12.73 t ha−1 when the standard nitrogen content of 1400 mmol m−2 was used. The proportion of yield lost due to lodging could vary more if several parameters were altered together. For example, a crop experiencing low temperatures with a small green area index and small nitrogen content was predicted to lose 48% of its yield due to lodging. This rose to 76% yield loss in warm temperatures with a large green area index and large nitrogen content. 4.3. Testing the model 4.3.1. Observed lodging-induced yield losses The model for predicting lodging-induced yield losses has been tested against the lodging-induced yield losses observed in the A96, N96 and N98 experiments (Table 3). The artificial lodging experiment (A96) experienced no natural lodging. The crops within the treatments lodged at GS59 and GS65 partially re-erected themselves by bending at the penultimate and top node respectively to give crop heights equivalent to the lengths of the top two internodes plus the ear (for GS59 lodging) and the length of the top internode plus the ear (for GS65 lodging). These treatments remained in a partially re-erected state until harvest. The proportion of total crop height represented by the upper two internodes and top internode (including the ear) has been measured at 0.77 and 0.53, respectively (Berry et al., 2007). Based on these height proportions it was estimated that average angle of lodging for the treatments lodged at GS59 and GS65 were 40◦ and 58◦ past the vertical respectively. The treatments lodged at GS75 and GS83 did not re-erect and remained lodged at 90◦ for 26 days and 8 days, respectively. These lodging periods equate to 68% and 21% of the grain filling period respectively (Table 3). Lodging was mainly by bending of the stems and anchorage failure, although some stem buckling was observed when lodging was induced at later growth stages. Lodging at GS75 resulted in the greatest yield losses, with about 50% yield lost. Lodging at GS65 and GS83 each caused about 25% yield loss. Lodging at GS59 resulted in a yield loss of only 2% (Table 3). N96 experienced major lodging events on 12th June (GS58), 24th July (GS75), 29th July (GS78), 7th August (GS87) and 11th August (GS90). All lodging events resulted from anchorage failure and occurred after 4–12 mm rain had fallen the previous day. The higher seed rate had about three times as much lodging, averaged over the grain filling period (24th June to 4th August), compared with the lower seed rate (Table 3). The high and low seed rate treatments were observed to lose 20% and 6% yield due to lodging respectively. Two plots partially re-erected after the first lodging event at GS58 and remained in this state until the 2nd lodging event

24

P.M. Berry, J. Spink / Field Crops Research 137 (2012) 19–26

Table 3 Percentage areas lodged weighted across the grain filling period for each treatment, with measured yields after lodging and estimated potential yields in the absence of lodging. Percentage area lodged weighted across the grain filling period

Yield (t ha−1 100% dm)

Experiment

Treatment

Lodged at 5–45◦

Lodged at 45–89◦

Lodged at 5–85◦

Lodged at 85–90◦

Measured after lodging

Potential without lodging

A96 A96 A96 A96 N96 N96 N98 N98 N98 N98 N98 N98 N98 N98 N98

Lodged at GS59 Lodged at GS65 Lodged at GS75 Lodged at GS83 500 seeds m−2 250 seeds m−2 Cadenza 160 seeds m−2 Cadenza 320 seeds m−2 Cadenza 640 seeds m−2 Haven 640 seeds m−2 Soissons 160 seeds m−2 Soissons 320 seeds m−2 Soissons 640 seeds m−2 Spark 320 seeds m−2 Spark 640 seeds m−2

0 0 0 0 6 5 – – – – – – – – –

0 0 0 0 10 8 – – – – – – – – –

– – – – – – 24 19 14 2 7 19 12 5 12

a

6.59 5.07 3.67 5.15 6.90 8.01 6.53 5.52 2.78 7.28 7.01 6.49 4.47 7.31 6.76

6.74 6.74 6.74 6.74 8.56 8.56 8.33 8.63 8.66 8.92 7.62 7.31 7.31 8.05 8.05

a

a

100 100 68 21 a 43 a 10 21 41 81 22 0 30 74 1 14

Partial re-erection of lodged plants occurred in some or all of the plots within these treatments.

42 days later. The upper two internodes re-erected and the angle of lodging during this period was estimated at 40◦ past the vertical. N98 experienced most lodging on 3rd June (GS55) after 22 mm rain. Further lodging occurred on 16th June (GS65) after 20 mm rain, after which the percentage area lodged gradually increased until 17th August (harvest) and was not associated with any particularly severe rain or wind events. Some of the treatments partially re-erected after the first lodging event, but were lodged again by the second lodging event. None of the treatments were re-erected during grain filling. All lodging was due to anchorage failure. Cultivars Cadenza and Soissons had the greatest percentage area lodged during grain filling followed by Spark and Haven (Table 3). Higher seed rates resulted in more lodging for each cultivar and this was reflected in observed yield losses which ranged from 70% to 20% for 640 seeds m−2 compared with losses of 35–10% for 320 seeds m−2 (Table 3).

4.3.2. Predicted lodging-induced yield losses The mean temperature and radiation during grain filling were 16.6 ◦ C and 9.5 MJ m−2 for both A96 and N96, and 15.5 ◦ C and 7.0 MJ m−2 for N98. The model used this information with the standard crop parameter values described in Table 1 to predict the yield of an upright crop and crops lodged at 25◦ , 40◦ , 45◦ , 58◦ , 67◦ and 90◦ from the vertical for each experiment. These angles of lodging were chosen to represent the average angle of lodging that was assessed in the experiments (Table 3). This information was then used with the observed lodging severity data described in Table 3 to predict the yield for each experimental plot. The model predicted that the unlodged crops in A96 and N96 would have yielded 10.8 t ha−1 dry matter. This is greater than the observed yields of unlodged crops which varied between 6.7 and 8.6 t ha−1 (Table 3). Unlodged crops for N98 were predicted by the model to yield 9.4 t ha−1 dry matter compared with the observed yields of between 7.3 and 8.9 t ha−1 (Table 3).

0.8

Observed proportion of yield lost

7

Observed yield loss (t/ha)

6 5 4 3 2 1

0.6

0.4

0.2

0 0

0.2

0.4

0.6

0.8

0 0

1

2

3

4

5

6

7

-1

-0.2

Predicted proportion of yield lost

Predicted yield loss (t/ha) Fig. 3. Relationship between the yield losses due to lodging as predicted by the model and observed. Plot data presented. A96 (), N96 (), N98 (䊉). Plots which partially re-erected after lodging are denoted for A96 () and N96 (). Regression line for all plots; y = 0.78x + 0.14, R2 = 0.60. 1:1 relationship denoted by dashed line.

Fig. 4. Relationship between the proportion of yield lost due to lodging as predicted by the model and observed. Plot data presented. A96 (), N96 (), N98 (䊉). Plots which partially re-erected after lodging are denoted for A96 () and N96 (). Regression line for all plots; y = 1.06x + 0.017, R2 = 0.71. 1:1 relationship denoted by dashed line.

P.M. Berry, J. Spink / Field Crops Research 137 (2012) 19–26

A comparison between the predicted yield losses and the observed yield losses for each lodged plot of the three experiments is described in Fig. 3. The best-fit line for the model prediction of yield losses against observations had a slope of 0.78, a y-axis intercept of 0.14 t ha−1 and an R2 value of 0.60. The least well predicted plots included nine plots which partially re-erected in A96 and 3 plots in N98 for which the model over-predicted the yield losses by 1 t ha−1 or more. When the partially re-erected plots were removed from the analysis then the best-fit line had a slope of 0.76, a y-axis intercept of 0.39 t ha−1 with an R2 value of 0.66. The predictive performance of the model improved when it was used to predict the proportion of yield lost due to lodging. In this case the best-fit line had a slope of 1.05, a y-axis intercept of 0.017 an R2 value of 0.71 (Fig. 4). When the partially re-erected plots were removed from the analysis then the best-fit line had a slope of 1.02, a y-axis intercept of 0.048 and an R2 value of 0.77. In both cases the y-axis intercept of the best fit line was not significantly different from zero and the slope was not significantly different from (1).

5. Discussion Observations of lodging induced yield losses reported here and predictions by the yield loss model indicate that 60–75% of yield is lost when the crop is lodged flat for the entire grain filling period. This is similar to the 70–80% yield losses that have previously been observed in the UK as a result of severe lodging (Easson et al., 1993). The predicted yield losses from intermediate angles of lodging are also similar to observations made in other studies. Lodging at 45◦ has been shown to cause between one quarter and one half of the yield losses incurred from 80◦ lodging in wheat (Fischer and Stapper, 1987). This is similar to this study in which lodging at 45◦ was predicted to reduce yield by 18% compared with a 54% yield reduction following lodging at 80◦ . The yield losses observed and predicted in this study therefore appear to be representative of lodging induced yield losses that have been observed previously. The canopy photosynthesis model developed in this study predicted a significant amount of variation for the proportion of yield loss caused by lodging in a range of crops and on average the predicted values were close to the observed values. Predicting the proportion of yield lost due to lodging is a useful output because growers can calculate the absolute amount of yield lost retrospectively once the yield of the lodged crop is known. While the ability of the model developed in this study to predict the proportion of yield lost due to lodging is encouraging there were clear areas where improvement is required. A few of the test crops in this study lodged very early before flowering and were able to partially re-erect because the upper internodes were still extending. Similar effects have been observed in other artificial lodging studies (Briggs, 1989). This significantly reduced the proportion of yield lost which the model did not fully predict. This study estimated the angle of lodging of the re-erected plants from an estimate of the crop height and it is possible that the angle of lodging past the vertical was over estimated which resulted in an over-estimate of yield losses. Alternatively Eq. (9) may under-estimate the green area exposed to light of a lodged crop that has re-erected because the upper leaves are positioned in a near-normal manner unlike in a typically lodged crop. Re-erection of lodged crops is relatively rare and only occurred in these experiments because the crops were either artificially lodged or because special crop management was used to generate an unusually high lodging risk. Berry et al. (2003) calculated that the highest lodging risk of wheat occurs at the end of grain filling when the ears reach their greatest weight. Other possible reasons for the imperfect prediction of yield losses include using standardised, rather than measured, model

25

parameters for green area index, the maximum rate of photosynthesis, leaf angle and water soluble carbohydrate. While variation in these parameters was shown to only have a modest effect on the model output it is inevitable that using standard figures will have reduced the model’s predictive ability. The model over-predicted the yield in the A96 and N96 experiments. This was probably due to the absence of any mechanisms within the model to account for abiotic and biotic stresses such as water stress and disease. In general disease was less likely to have reduced yield significantly because the trials were treated with fungicides. However, low levels of disease may have been present. Water stress may have been a yield limiting factor in the summer of 1996 because the rainfall was 61% of average (74 mm) for June and July, which encompassed ear emergence to maturity, and only 70% of average rainfall in May. This level of rainfall may be expected to result in moderate levels of water stress which may partially account for the over-prediction of yield in 1996. Severe early water stress may be expected to increase lodging induced yield losses because less water soluble carbohydrate may be available to buffer against lodging-induced yield losses. A more fundamental reason for the model’s imperfect predictive accuracy may be that the mechanism of lodging-induced yield loss is different to the one tested in this study. For example the model assumes that all lodging-induced yield losses result from a reduction in assimilate supply to the growing grains and does not take account of the possible effect of pre-anthesis lodging reducing sink size. The model also assumes that the crop is primarily source limited as is generally the case in the UK (Shearman et al., 2005). Sink limited crops may be expected to lose less yield to postanthesis lodging, although Acreche and Slafer (2011) observed yield losses of 40–60% from lodging at anthesis in a more sink limited environment than tested in this study. Other mechanisms that have been suggested as causes for lodging-induced yield losses include reduced translocation of mineral nutrients and carbon for grain filling, increased respiration, rapid chlorosis and greater susceptibility to pests and diseases. In this study there was no evidence of increased pests and diseases in the lodged plots. In this study the best-fit line for the model’s prediction of the proportion of yield lost to lodging against observations had an R2 of 0.71 and was not different to the 1:1 relationship. This indicates that the primary mechanism by which lodging reduces yield may be by reducing canopy photosynthesis during grain filling. This would be in agreement with the proposed mechanism by which lodging reduces yield in rice (Setter et al., 1997). It would be possible to develop the model to account for other possible factors that may affect lodging-induced yield losses that are described above. However given the predictive accuracy of the current model a better case may be made for attempting to simplify the model without losing predictive accuracy. The sensitivity analysis demonstrated that the proportion of yield lost was only moderately sensitive to changes to the model parameters. The proportion of lodging induced yield losses estimated by the model using standard parameters (Table 1) for lodging at 90◦ , 65◦ and 25◦ past the vertical were 0.61, 0.34, 0.06, respectively. When these standard figures were used to estimate the proportion of yield lost to lodging in the experiments the best-fit line for the predicted against the observed had an R2 of 0.72. Optimisation of the yield losses associated with the different lodging severities showed that the greatest predictive ability could be achieved by using proportional yield losses of 0.7, 0.3 and 0.1 for lodging at 90◦ , 65◦ and 25◦ past the vertical. The best-fit line for the predicted against the observed had an R2 value of 0.75 and was not significantly different from the 1:1 relationship. This analysis indicates that the canopy photosynthesis model developed in this study may modestly under-estimate the yield losses from severe lodging and over-estimate the yield losses from moderate lodging. The

26

P.M. Berry, J. Spink / Field Crops Research 137 (2012) 19–26

results of the optimisation analysis can be used to formulate a single equation for predicting the proportion of yield lost (YLOSS ) for any lodged crop.

f

YLOSS =

i

(L90 × 0.7 + L65 × 0.3 + L25 × 0.1) n

(12)

In this equation i and f are the 1st and last days of grain filling, L90 is the proportion of crop area lodged at 85–90◦ from the vertical, L65 is the proportion of crop area lodged between 46◦ and 84◦ , L25 is the proportion of crop area lodged between 5◦ and 45◦ and n is the number of days of grain filling. The simplified model predicted the proportion of yield lost at least as well as the more sophisticated approach. This simplified model should now be tested on a wide range of crops and environments. To our knowledge it is not possible to test the model against data from published lodging studies due to incomplete information about the angle of lodging and duration of lodging. The method for predicting lodging induced yield losses described in this paper could be used in conjunction with models of the lodging process, such as Berry et al. (2003), to help quantify the impact on yield caused by factors which influence lodging. Rudimentary predictions schemes have also been developed to predict lodging risk from assessments of the crop made early in the growing season at GS31 or before (Berry, 1998). This would be in time for remedial measures to be taken to reduce lodging risk such as reducing and delaying nitrogen fertiliser and the use of plant growth regulators. Ultimately, these models could be combined to form a decision support guide to help growers choose the most cost effective remedial action for minimising lodging risk. Acknowledgements This work was funded by grants from the Home-Grown Cereals Authority. We thank staff from ADAS Rosemaund for their technical assistance. References Acreche, M.M., Slafer, G.A., 2011. Lodging yield penalties as affected by breeding in Mediterranean wheats. Field Crops Res. 122, 40–48. Amthor, J.S., 1989. Respiration and Crop Productivity. Springer, New York. Asseng, S., van Herwaarden, A.F., 2003. Analysis of the benefits to wheat yields from assimilate stored prior to grain filling in a range of environments. Plant Soil 256, 217–229. Austin, R.B., Edrich, J.A., Ford, M.A., Blackwell, R.D., 1977. The fate of dry matter, carbohydrates and 14C lost from the leaves and stems of wheat during grain filling. Ann. Bot. 41, 1309–1321. Austin, R.B., Morgan, C.L., Ford, M.A., Bhagwat, S.G., 1982. Flag leaf photosynthesis of Triticum aestivum and related diploid and tetraploid species. Ann. Bot. 49, 177–189. Berry, P.M., 1998. Predicting lodging in winter wheat. Ph.D. Thesis, The University of Nottingham, UK, 210 pp. Berry, P.M., Spink, J.H, Griffin, J.M., Sylvester-Bradley, R., Baker, C.J., Scott, R.K., Clare, R.W., 1998. Research to understand, predict and control factors affecting lodging in wheat. Home-Grown Cereals Authority Research Project No. 169. HGCA, London, 131 pp. Berry, P.M., Griffin, J.M., Sylvester-Bradley, R., Scott, R.K., Spink, J.H., Baker, C.J., Clare, R.W., 2000. Controlling plant form through husbandry to minimise lodging in wheat. Field Crops Res. 67, 59–81. Berry, P.M., Sterling, M., Baker, C.J., Spink, J.H., Sparkes, D.L., 2003. A calibrated model of wheat lodging compared with field measurements. Agric. Forest Meteorol. 119, 167–180. Berry, P.M., Sterling, M., Spink, J.H., Baker, C.J., Sylvester-Bradley, R., Mooney, S., Tams, A., Ennos, A.R., 2004. Understanding and reducing lodging in cereals. Adv. Agron. 84, 215–269.

Berry, P.M., Sylvester-Bradley, R., Berry, S., 2007. Ideotype design for lodging-proof wheat. Euphytica 154, 165–179. Bidinger, F., Musgrave, R.B., Fischer, R.A., 1977. Contributions of stored preanthesis assimilate to grain yield in winter wheat and barley. Nature 270, 431–433. Briggs, K.G., 1989. Studies of recovery from artificially induced lodging. Can. J. Plant Sci. 70, 173–181. Campbell, G.S., Norman, J.M., 1998a. An Introduction to Environmental Biophysics, 2nd edition. Springer, London, Chapter 15. Campbell, G.S., Norman, J.M., 1998b. An Introduction to Environmental Biophysics, 2nd edition. Springer, London, Chapter 10. Choudhury, B.J., 2000. A Sensitivity analysis of the radiation use efficiency for gross photosynthesis and net carbon accumulation by wheat. Agric. Forest Meteorol. 101, 217–234. Critchley, C.S., 2001. A physiological explanation for the nitrogen requirement of winter wheat. Ph.D. Thesis, Nottingham University, UK. Crook, M.J., Ennos, A.R., 1993. The mechanics of root lodging in winter wheat, (Triticum aestivum L.). J. Exp. Bot. 44, 1219–1224. Easson, D.L., White, E.M., Pickles, S.J., 1993. The effects of weather, seed rate and cultivar on lodging and yield in winter wheat. J. Agric. Sci. (Camb.) 121, 145–156. Fischer, R.A., Stapper, M., 1987. Lodging effects on high yielding crops of irrigated semi-dwarf wheat. Field Crops Res. 17, 245–248. Foulkes M.J., Spink J.H., Scott R.K., Clare R.W., 1998. Varietal typing trials and NIAB additional character assessments, vol. V. Home-Grown Cereals Authority Project Report No. 174, HGCA, London. Gebbing, T., Schnyder, H., Kuhbauch, W., 1999. The utilization of pre-anthesis reserves in grain filling of wheat. Assessment by steady-state (CO2)-C-13/(CO2)C-12 labelling. Plant Cell Environ. 22, 851–858. Gregory, P.J., Marshall, B., Biscoe, P.V., 1981. Nutrient relations of winter wheat 3. Nitrogen uptake, photosynthesis of flag leaves and translocation of nitrogen to grain. J. Agric. Sci. (Camb.) 93, 485–494. Hay, R.K.M., Walker, A.J., 1989. An Introduction to the Physiology of Crop Yield. Longman Scientific & Technical, UK. Hitaka, H., 1968a. Studies on the lodging of rice plants. Jpn. Agric. Res. 4, 1–6. Hitaka, H., 1968b. Experimental studies on the mechanisms of lodging and its effect on the yield of rice plants. Bull. Natl. Inst. Agric. Sci. Tokyo 15, 1–175. Laude, H.H., Pauli, A.W., 1956. Influence of yield on lodging and other characteristics in winter wheat. Agron. J. 48, 452–455. Monteith, J.L., 1965. Light distribution and photosynthesis in field crops. Ann. Bot. 29, 17–37. Mulder, E.G., 1954. Effect of mineral nutrition on lodging in cereals. Plant Soil 5, 246–306. Neenan, M., Spencer-Smith, J.L., 1975. An analysis of the problem of lodging with particular reference to wheat and barley. J. Agric. Sci. (Camb.) 85, 494–507. Pinthus, M.J., 1973. Lodging in wheat, barley and oats: the phenomenon, its causes and preventative measures. Adv. Agron. 25, 209–263. Setter, T.L., Laureles, E.V., Mazaredo, A.M., 1997. Lodging reduces the yield of rice by self-shading and reductions in canopy photosynthesis. Field Crops Res. 49, 95–106. Shearman, V.J., Sylvester-Bradley, R., Scott, R.K., Foulkes, M.J., 2005. Physiological processes associated with wheat yield progress in the UK. Crop Sci. 45, 175– 185. Stapper, M., Fischer, R.A., 1990. Genotype, sowing date and plant spacing influence on high-yielding irrigated wheat in Southern New South Wales. I. Potential yields and optimum flowering dates. Aust. J. Agric. Res. 41, 1043–1056. Sylvester-Bradley, R., Watson, N.A.R., Dewes, M.E., Clare, R.W., Scott, R.K., Dodgson, G., 1997. The Wheat Growth Guide: To Improve Husbandry Decisions. HomeGrown Cereals Authority, London, UK. Thorne, G.N., Pearman, I., Day, W., Todd, A.D., 1988. Estimation of radiation interception by winter wheat from measurements of leaf area. J. Agric. Sci. (Camb.) 110, 683–687. Tottman, D.R., 1987. Decimal code for the growth stages of cereals. Ann. Appl. Bot. 110, 683–687. Weibel, R.O., Pendleton, J.W., 1964. Effect of artificial lodging on winter wheat grain yield and quality. Agron. J. 56, 487–488. Whaley, J.M., Sparkes, D.L., Foulkes, M.J., Spink, J.H., Semere, T., Scott, R.K., 2000. The physiological response of winter wheat to reductions in plant density. Ann. Appl. Biol. 137, 167–178. Yang, J.C., Zhang, J.H., Wang, Z.Q., Zhu, Q.S., Liu, L.J., 2001. Water deficit-induced senescence and its relationship to the remobilization of pre-stored carbon in wheat during grain filling. Agron. J. 93, 196–206. Zhang, H., Turner, N.C., Poole, M.L., 2010. Source–sink balance and manipulating sink–source relations of wheat indicate that the yield potential of wheat is sinklimited in high-rainfall zones. Crop Pasture Sci. 61, 852–861.