Quantifying different causes of leaf and tiller death in ... - CiteSeerX

New Zealand Journal of Agricultural Research, 1998, Vol. 41: 149-159 0028-8233/98/4102-0149 $7.00/0 © The Royal Society of New Zealand 1998

149

Quantifying different causes of leaf and tiller death in grazed perennial ryegrass swards S. J. R. WOODWARD AgResearch Ruakura Private Bag 3123 Hamilton, New Zealand Abstract Death of plant material in grazed perennial swards results from a number of processes, particularly leaf senescence, tiller death, and grazing. These processes must be understood in detail to manage supply of high quality pasture for, especially, dairy cattle through the reproductive season. Although measurements have been made under a variety of pasture conditions, a generalised model has not yet been established. This paper reviews experimental evidence for the rates of leaf and tiller death resulting from five independent physical and physiological mechanisms operating in a grazed ryegrass-dominant pasture. First-order (linear) models are formulated to predict the rates of leaf and stem death on vegetative and reproductive tillers in such a pasture. Leaf death rate accounts for 90% of dead matter production in the non-reproductive season, and was found to be linearly correlated to soil temperature. During reproduction, decapitation of reproductive stems, usually prior to ear emergence, is the dominant source of dead material. These models provide an essential component of the detailed modelling of grazed pasture tissue dynamics. Keywords senescence; model; pasture; Lolium perenne; reproductive development; flag-leaf; stem; ear-emergence; maturity-stage; grazing; defoliation; decapitation; temperature; flowering

A97060 Received 27August 1997; accepted 21 November 1997

INTRODUCTION In New Zealand farms, grass pasture is the principle food source for animals. During the summerautumn period, management of this food source is complicated by changes in pasture structure which result from plant reproductive development and relatively warm, dry weather. These effects may lead to accumulation of dead leaf and stem material in the sward so that, despite high pasture masses, animal production may be restricted by the low quality of herbage (Bryant 1982; Korte et al. 1984). In order to identify opportunities for improved use of summer-autumn pasture, a model of tissue cycling in grazed reproductive pastures is being developed. This model will include simulations of the effects of stocking density and water application patterns on the production and disappearance of dead material in the sward. A key component is clearly the ability to predict the rates of leaf and tiller death throughout the year. In order to make these predictions, the various physical and physiological processes driving the death of green leaf and stem material are first reviewed and then modelled. Senescence is the name given to the physiological process in plants whereby fully mature plant parts, particularly leaves, age and die. In pasture studies, senescence has generally been defined as total dead mass accumulation (e.g., Hunt 1970; Wilman& Mares-Martins 1977;Cayleyetal. 1980). This definition, however, does not recognise the separate physiological and physical processes contributing to the build-up of dead material, of which leaf senescence is only one. More recent studies have monitored the lifespan of individual leaves (Vine 1983; Chapman et al. 1984) or tillers (Chapman et al. 1984; L'Huillier 1987). This approach provides ideal data for the construction of more detailed predictive models. Previous grazing models have often assumed that a given proportion of sward leaf material senesces each day (e.g., Blackburn & Kothmann 1989; Hearne & Buchan 1990; Mohtar et al. 1994;

New Zealand Journal of Agricultural Research, 1998, Vol. 41

150 Vegetative Tillers growth

senesce

decay

trampled & lodging

Fig. 1 Schematic of tissue flows in a grazed perennial pasture model. The flows modelled in this paper are annotated with their rate response functions.

Schwinning & Parsons 1996). This proportion is called the relative senescence rate. The assumption that relative senescence rate is an environmentally determined constant seems to hold in many cases, particularly in continuously grazed swards (Cayley et al. 1980; Bircham & Hodgson 1983; Chapman et al. 1984), but may not be appropriate under rotational grazing, because relative senescence rates of ryegrass and clover leaves have been observed to increase during the regrowth phase (Hunt 1970; Robson 1973). Woodward & Wake (1994) used a simple delay model to confirm that relative senescence rates would be expected to decrease slightly following defoliation. A similar result was derived from a more detailed model by Johnson & Parsons (1985). However, further analysis showed that the overall effect of leaf age on total canopy senescence in a rotational grazing system is likely to be small, so that using a constant relative senescence rate may be sufficiently accurate for use even in intermittently defoliated swards (Woodward 1993). Deaths of whole vegetative tillers (and their leaves) result from many processes, including shading, severe grazing, pulling, dung and urine deposition, trampling, winter-kill, competition for nutrients, disease, and grazing by pests (Ong et al. 1978;Alberda&Sibma 1982;Breretonetal. 1985; Thorn et al. 1986; Matthew 1992; Bullock et al. 1994; Matthew et al. 1996). Field studies on ryegrass have suggested, however, that the instantaneous death rate of whole vegetative tillers

is independent of tiller age (Colvill & Marshall 1984; Korte 1986), although this is not true for some other grasses (Jonsdottir 1991). This is in contrast to leaf senescence rate, which is the result of interactions between leaf appearance rate and leaf life span, both of which change seasonally and in response to nitrogen (Wilman & Mares-Martins 1977; Vine 1983). In addition, reproductive tillers die when the reproductive stem matures or is grazed (Chapman et al. 1984; L'Huillier 1987). Models have not been developed to describe the rates of these processes. The object of this paper was to quantify the various biological processes contributing to the death of pasture tissue in the vegetative and reproductive fractions of the sward and to estimate the rates of these processes for New Zealand perennial ryegrass pastures as they are influenced by fluctuations in temperature.

MODEL DESCRIPTION The pasture model considers five categories of plant tissue: V, green leaf and sheath on vegetative tillers; R, green leaf and sheath on reproductive tillers (defined as those undergoing reproductive stem elongation); S, green reproductive stem; M, mature reproductive stem (defined as stems with full ear emergence); and D, dead leaf, sheath, and stem material. Although the model recognises the physiological processes occurring at the tiller and leaf levels, this is a mass-based and not a tiller-based model, and each category is quantified as kg dry matter ha"1. I followed Chapman et al. (1984) in considering only tissue on tillers which were of sufficient size to enter the grazing horizon, so that small, short-lived daughter tillers were ignored. While these may be significant in terms of tiller demographics, their contribution to herbage mass is likely to be negligible. Turnover of pasture tissue was modelled by considering the various flows between these five categories (Fig. 1). All flows were expressed as kg ha"1 d"1, and the model variables and parameters are summarised in Table 1. "Death" in this context concerns three of these flows: the conversion of vegetative green material V into dead D, the conversion of live reproductive stem S into mature stem M, and the conversion of green leaf and sheath on reproductive tillers R into dead D. However, each of these flows represents the sum of several physiological and physical processes operating in

Woodward—Leaf and tiller death

151

0.016

p

Fig. 2 Daily relative rates of perennial ryegrass tiller death in a grazed pasture near Palmerston North, New Zealand from (1) severe grazing with no visible flower head, (2) severe grazing with a visible flower head, (3) flower maturation, and (4) other causes. (Redrawn from Chapman etal. 1984).

0.004 - 0.002 - 0.000 ro

ro

9

o

ro

f-

•

s

the plant-animal system, some at leaf level, some at whole tiller level and some involving grazing. The quantification of these flows, therefore, requires an understanding of each of these processes. Definition of processes The tissue flows operating in grazed pasture have been studied experimentally by Chapman et al. (1984), who measured leaf and tiller death rates in perennial ryegrass-zlgras/w-white clover hill country pastures grazed by sheep at Ballantrae, Palmerston North, New Zealand. Their measure-

ro —i

Dec



CO CD O"

ments were leaf and tiller numbers. Chapman et al. (1984) classified grass tiller deaths from: (process 1) severe grazing without a visible flower head, (process 2) severe grazing with a visible flower head, (process 3) maturation of the flower, and (process 4) other causes in vegetative state. The results for perennial ryegrass are shown in Fig. 2. This classification of tiller death processes is ideal for the modelling of the tissue flows in Fig.

Table 1 Table of symbols. Symbol D

f Ir Is

LDR M m,

n P R S t

T V

P

|i

a ao Ola O4

Value/Units 1

kg ha" 0.21 kg animal"' d"1 kg animal"' d~' leaves leaf"1 d"' kg ha-' 0.30 animals ha ' 0.52 kg ha"' kg ha-' d °C kg ha"' 0.92 0.09 d"' kg kg-'GM d"1 kg kg-'GM d"' kg k g ' G M d 1 kg kg-'GM d"'

Description mass of dead material mass of flag leaf relative to total leaf on reproductive tillers daily intake of reproductive green leaf and sheath per animal daily intake of green stem per animal leaf death rate mass of mature stem proportion of leaf mass exported during senescence stocking density fraction of stem ingested mass of green leaf and sheath on reproductive tillers mass of green stem time in days soil temperature mass of green leaf and sheath on vegetative tillers fraction of stem killed by grazing stem maturation rate total death rate of vegetative leaf V rate of leaf senescence of V rate of V death due to severe grazing rate of V death due to other tiller death processes

New Zealand Journal of Agricultural Research, 1998, Vol. 41

152

V vegetative green

leaf senescence

\/

other tiller

deat

V

tiller death from severe ^grazing

D dead material Fig. 3 Main processes resulting in the death of vegetative leaf and sheath. 1, and much of the data presented in Chapman et al. (1984) can be used directly to parameterise the models for each of these mechanisms. By modifying the classes, this classification can be extended to include leaf senescence and to distinguish between vegetative and reproductive tiller deaths (c.f. L'Huillier 1987); I add one new process (0) and split Chapman et al.'s (1984) process 1 into two:

(process 0) leaf senescence from ageing in vegetative state, (process la) tiller death from severe grazing of vegetative tillers, and (process lb) tiller death from severe grazing of reproductive tillers without a visible flower head. This makes a total of five independent processes (0, la, lb+2, 3, and 4) responsible for the death of leaf or stem tissue in grazed grass pasture. These five processes are sufficiently detailed to characterise the rates of death modelled in this paper and are generic to pasture models at this level of detail. Death of vegetative material According to the classification of Chapman et al. (1984), the mechanisms that drive the death of vegetative material are leaf senescence (process 0), non-specific tiller death (4), and tiller death from severe grazing (la). The total rate of death of vegetative material is the sum of the rates of these three processes (Fig. 3). Leaf senescence Chapman et al. (1984) measured the number of leaves per tiller and the rate of leaf senescence per live tiller on ryegrass tillers (process 0). These data showed that the relative rate of leaf senescence in ryegrass (LDR) had a strong seasonal pattern which closely followed soil temperature (T, measured by Chapman et al. 1983) (Fig. 4):

0.035 0.030 - -

0.000 I 1 I p.....,!.,...,. 15-Mar 10-May 5-Jul 30-Aug 25-Oct 20-Dec 14-Feb

Fig. 4 Relative rates of leaf death (LDR) measured near Palmerston North, New Zealand by Chapman et al. (1984) (•) and predictions based on soil temperature (Equation 1) ( ).

Woodward—Leaf and tiller death

153

Fig. 5 Relative rates of leaf death (LDR) measured near Edinburgh, Scotland by Vine (1983) ( • ) and predictions based on soil temperature (Equation 1)

0.030 0.025 "TJ

ath

B 2

0.020 0.015 -

eaf

a> •a 0.010 0.005 0.000 1-Aug

LDR = 0.00158 (T+ 2.2) (JR.2 = 0.90) leaves leaf"1 d"1 (1) The data from the reproductive season (mid summer-early autumn) were excluded in order to isolate vegetative leaf senescence, because the typical peak in senescence rates in summer (Ong et al. 1978; Cayley et al. 1980; Chapman et al. 1984; L'Huillier 1987) results from the deaths of whole tillers. With this peak removed, Equation 1 overestimated leaf senescence in the summer period (Fig. 4). It is proposed that the significantly lower senescence rate observed in summer is a consequence of the lower average rate of leaf senescence on reproductive tillers (see below). The close correlation between annual temperature cycles and leaf senescence rate has been confirmed by other authors (Thomas & Norris 1977; Cayley et al. 1980; Lemaire & Chapman 1996; cf. Brereton et al. 1985). Also, experimental evidence suggests that leaf senescence is insensitive to water availability (Cayley et al. 1980; Barker 1983; Thorn et al. 1986; Parmenter 1991), but increases slightly after nitrogen application (Vine 1983). The experiment of Vine (1983) on ryegrassdominant pasture provides for validation of the relation in Equation 1. Leaf death rate was calculated by taking the reciprocal of the leaf death interval measured in Vine's trial and dividing by the number of leaves per tiller. These data are plotted

26-Sep 21-Nov 16-Jan 13-Mar

8-May

in Fig. 5 against predictions from Equation 1 using the reported soil temperatures (Vine 1983). Note that the seasonal pattern in Fig. 5 is for the Northern Hemisphere. The prediction of leaf death rate was good except during late spring-early autumn (MaySep), when Equation 1 again overestimated the rate of senescence of leaves. Noise in the predictions (Fig. 5) reflects the longer measurement interval for leaf lifespans compared with the relatively rapid fluctuations in soil temperature. LDR (Equation 1) cannot be used directly to predict the rate of mass senescence because older grass leaves are relatively heavier than average and because the ryegrass plant recovers 30% of the dry weight of leaves prior to their death (Robson & Deacon 1978). However, the size of the older leaves in a grazed sward is reduced so that all leaves are of roughly the same weight regardless of age (Parsons et al. 1988). The relative rate of green mass loss from leaf senescence (GQ) can therefore be calculated as (2) o = (l-m t )I£)/?kgkg = 0.00111 (T+2.2) where mt = 0.3 is the proportion of dry weight translocated back to the tiller from senescing leaves. Tiller death As well as leaf senescence, vegetative death results from the death of whole tillers (processes la and 4).

154

New Zealand Journal of Agricultural Research, 1998, Vol. 41

The main cause of vegetative tiller death is the shading of small tillers as sward mass increases in spring (Ong et al. 1978; Alberda & Sibma 1982; Colvill & Marshall 1984; L'Huillier 1987). Although this loss might represent a large number of tillers, particularly in laxly grazed swards (L'Huillier 1987), the tillers dying in this fashion are "very small" (Ong et al. 1978). For this reason Chapman et al. (1983, 1984), considering only tillers in the grazed horizon of a sheep-grazed sward, observed no such peak in vegetative tiller death in mid spring-early summer. For the purposes of this (mass-based) model, therefore, it is proposed that the death of pasture mass resulting from these deaths can be neglected. Vegetative tillers also die from other nongrazing causes (process 4), such as winter-kill (Brereton et al. 1985), disease, and grazing by pests (Ong etal. 1978). Bullock etal. (1994) and Brereton et al. (1985) found that total tiller death rate was not seasonal or temperature dependent in perennial ryegrass, and their data record the average tiller death rate as 0.0056 and 0.0024 d"1, respectively. However, Bullock et al. 's (1994) data included death from shading and grazing, and Brereton et al.'s (1985) probably included winter-kill deaths, so the lower average rate for "other" tiller deaths in Chapman et al. (1984) is preferred for this model, i.e., CT4= 0.00074 d"> (3) This would include non-grazing animal effects such as death from dung-covering (Hirata et al. 1990), urine-scorching, or trampling. Such damage might be expected to be more severe under heavy grazing pressure. L'Huillier (1987), however, observed no significant difference in autumn-winter death rate between paddocks with low or high stocking rate of dairy cows. Chapman et al. (1984) similarly commented that leaf and tiller death was largely unaffected by sheep grazing management (set stocked or rotationally grazed). The effects of grazing management and stocking density on vegetative tiller deaths are therefore neglected in the present study. Nevertheless, stocking rate effects on vegetative tiller death have been observed in at least one study (Bullock et al. 1994). The third process affecting death of vegetative material is tiller death from severe grazing (process la). This is evident in Fig. 2 from March to August, when tiller deaths from grazing could not have been from defoliation of a reproductive stem. Bullock et al. (1994) found that winter grazing with sheep

increased tiller deaths, but concluded that the mechanisms for this were difficult to explain. For the purposes of this model, therefore, stocking density effects will again be ignored and Oj a will be assumed to be constant (calculated from Chapman etal. 1984): a i a = 0.00057 d 1 (4) Total vegetative death Unlike LDR, tiller death occurs independently of age (Colvill & Marshall 1984), and relative tiller rates can therefore be converted directly into rates of mass death. The exception to this would be the early spring period where many small tillers die because of shading, as noted above. However, we follow Chapman et al. (1983) in considering only those tillers that inhabit the grazed horizon. The total death rate of vegetative green material a can therefore be estimated by adding the rates from the different processes: a = a 0 + a, a + a 4 kg kg-'GM d"1 (5) For Chapman et al.'s (1984) data this is (adding Equations 2, 3, and 4): 0=0.00111 ( T + 3.4; kg kg-'GM d~' (6) Death of vegetative material is therefore dominated by leaf senescence (Equations 1 and 2), which accounts for 90% of a at 10°C. Since Equation 1 has been checked against the data of Vine (1983) (Fig. 5), validation of Equation 6 may also be assumed. Death of reproductive stem Parsons & Robson (1981) recorded that if the transition from vegetative to reproductive tillers (in S24 ryegrass) occurs in late winter, then stem elongation typically starts six weeks later in early spring, and the ear becomes visible after a further six weeks, by late spring. In an average perennial ryegrass cultivar, the first flowers form two weeks after ear emergence, and crop maturity/harvest follows a little over a month later, in early summer (Elgersma 1990). The critical point in the reproductive development of a tiller is the onset of stem elongation (Korte et al. 1984), because the tiller can be classed irreversibly as "reproductive". The physiology of the tiller changes dramatically at this point and any subsequent stem damage results in the death of the tiller. The current model considers two categories of reproductive stem. The first (S) is stem that is

Woodward—Leaf and tiller death

155

S reproductive stem

M-S

Ms stem decapitation (processes 1b+2)^

I

\

normal stem maturation .(process 3]

M mature stem Fig. 6 Main processes resulting in the maturation of green reproductive stem.

growing, has live leaves attached, is of high digestibility, and is acceptable to grazing animals. The second (M) is stem that has ceased growing (accumulating mass), has no live leaves, is of low digestibility, and is rejected by grazing animals. In reality, of course, the distinction is not so clear: animals reject stem even before its digestibility has declined (Gallais 1972), and stem continues growing for some time after its digestibility has fallen (Johnston & Waite 1965). Nevertheless, this division is not unreasonable. Johnston & Waite (1965) observed significant changes in the structure and digestibility of reproductive ryegrass tillers around the time of ear emergence: the stem became rigid, tiller height stopped increasing, few green

leaves remained, and stem digestibility fell below 70%. Similar patterns were observed by Terry & Tilley (1964), and Gallais (1972) proposed that pasture heading provides a useful indicator of the latest stage at which pasture should be grazed. Therefore, this division may be adequate for the purpose of analysing grazing management. Stem maturation Live reproductive stem (S) matures through two main processes (Fig. 6). The first is normal stem ripening and maturity, which results in a mature flower head (process 3). In a seed crop, maturation occurs rapidly over a short time period, so that the whole crop matures within a few weeks (Elgersma 1990). In grazed pasture, however, the proportion of stems that reach maturity is very small, so stem death from maturation must be considered to be of only secondary importance (Chapman et al. 1984). For this reason, we adopt a simplified model of stem maturation which assumes that a proportion ji of live stem mass S matures each day, and that maturation occurs only over a short period of time. The rate of stem maturation is then M-Skgha-'d" 1 (7) where [l may change with time. Elgersma (1990) recorded the rate of ear emergence in 18 plots of perennial ryegrass grown for seed. Her data are given in Fig. 7. Assuming that by the end of the trial the crop was completely emerged, we can estimate that 9% of unemerged ears had emerged each day after first ear emergence on June 10. This figure may be used directly as an approximation

2500

1-Jun

Fig. 7 Average numbers of emerged ears from nine ryegrass cultivars on sandy ( • ) and clay (A) soils at Wageningen, Netherlands (Elgersma 1990), and the curve 2018(1 - exp(-0.0915 /)) ( ) where t = days after 10 June (fitted by least squares to the pooled data). It is not necessary to know the number of vernalised tillers or the total number of tillers to estimate the ear maturation rate, 0.0915 d"1.

15-Jun

29-Jun

13-Jul

27-Jul

10-Aug

New Zealand Journal of Agricultural Research, 1998, Vol. 41

156

R reproductive green PR tiller death from grazing

leaf senescence (process 0)

of stem (processes 1b+2)

\f

flag leaf death from stem maturation (process 3)

V V D dead material

Fig. 8 Main processes resulting in the death of green leaf and sheath on reproductive tillers.

of [l, so that once ear maturation has commenced, 9% of the immature ears would mature each day, i.e., - J o d -' o.O9 d"

l< 10 June t > 10 June

lengths, which gives |3= 0.92. Korte et al. (1984) suggested that hard grazed sheep might remove the whole culm (i.e.,p = 1.0, p= 0.0), but this would occur only with tillers that were still short. Predicting animals' daily intake of plant components (e.g., / s ) is beyond the scope of this paper. See Woodward (1997) for a brief outline of the method that will be used in this model. In a cutting experiment, Korte (1986) observed that further reproductive tillers continued to appear after the initial group of reproductive tillers had been decapitated in late spring (early Nov). He suggested that these may have originated from vernalised tiller buds that had previously failed to express because of apical dominance. This implies that management can influence the expression of vernalised buds, which complicates the modelling of reproductive tiller death. In the current model I assume that the effect of management on bud development may be ignored. This is a reasonable assumption in grazed swards because individual tillers develop and are grazed asynchronously, so that the pattern of reproductive tiller deaths is more evenly spread over the season (Chapman et al. 1984; L'Huillier 1987). By comparison, tillers in cut or seed crops tend to develop in phase with one another (e.g., Korte 1986).

(8)

At this rate it would take 32 days for 95% of the crop to mature. This assumes that the presence of grazing animals does not affect the rate of maturation.

Death of leaf on reproductive tillers Leaves on growing reproductive tillers (R) die either when the leaves themselves senesce (process 0) or when the reproductive stem matures (processes lb, 2, and 3) (Fig. 8).

Stem grazing Reproductive leaf senescence The main cause of tiller death in grazed pasture is In general, leaves on reproductive tillers are subject decapitation of the reproductive stem (processes lb to the same pattern of senescence as leaves on and 2) (Chapman et al. 1984; L'Huillier 1987). This vegetative tillers. Furthermore, Vine (1983) usually occurs prior to ear emergence, but Chapman observed that, prior to flag leaf emergence, leaf et al. (1984) also recorded a smaller number of tiller senescence rates were unaffected by reproductive deaths from sheep grazing tillers with emerged ears. development (cf. Chapman et al. 1984). The last When animals remove even a small portion of leaf produced by the reproductive tiller (the flag the stem, the whole tiller dies. This is easily leaf), however, survives far longer than all previous modelled. If the stocking density is n (animals ha"1), leaves on the tiller (Vine 1983). The average rate the daily intake of stem by each animal is 7S (kg of senescence of the leaves on reproductive tillers animal"1 d"1) and the average fraction eaten of any (p), therefore, is lower than that for vegetative tillers one stem is p, then, (c), and may be calculated as follows: if a rate of intake of stem = n Is kg ha"1 d"1 (9) reproductive tiller has a total of six leaves (Vine 1983), and the sixth (flag) leaf is, say, 30% heavier rate of stem death from grazing = p n 7S (io) 1 1 than the five previous leaves and does not senesce kg ha" d" prior to ear ripening, then p is: where (3 = -1+1//?. From example, Betteridge et al. 1 1 (11) (1994) measured p at 0.52 for a range of tiller p = a ( l - / ) k g k g ~ G M d -

Woodward—Leaf and tiller death where/= 1.3/6.3 = 21% is the flag leaf fraction of total reproductive leaf by weight. The lower senescence rate of flag leaves thus might account for some of the 20% over-prediction of leaf senescence in summer by Equation 1 in relation to Chapman et al.'s (1984) data (Fig. 4). Leaf death resulting from stem death As discussed above, reproductive stem matures either because of ripening or grazing (Fig. 6). For every tiller that matures by these processes, the attached reproductive green leaf (R) will also senesce (Johnston & Waite 1965). So if the rate of stem death is given in Equations 7 and 10, the associated rates of loss of R are, rate of loss of R from stem maturation = \ifR kg ha"1 d"1 rate of loss of R from stem decapitation The former process affects only flag leaves, as all other tiller leaves have senesced by the time the tiller matures. The latter process is assumed to affect all leaves with equal probability. It is assumed that stem removal (nls, Equation 9) does not directly result in death of R, as the animal will also have ingested the attached leaf (if) (i.e., / r ). SUMMARY This paper considers rates of leaf and tiller death that are based on a perennial grass pasture model with five categories of plant tissue (Fig. 1). Although it is not a tiller-based model, the physiological processes affecting leaf senescence and tiller death have been considered in calculating rates of tissue death in the vegetative and reproductive fractions of the sward. In particular, the decision to classify tillers as reproductive at the onset of stem elongation and to classify stem as "mature" at full ear formation has been justified. Furthermore, small vegetative tillers that are below grazing height have been neglected (following Chapman et al. 1984) so that the high rates of tiller death from shading in spring do not need to be considered (cf. L'Huillier 1987). The mechanisms of leaf and tiller death in grazed grass-dominant pasture have been reviewed, and simple models to predict the death rates of green leaf and stem have been formulated. The methodology of Chapman et al. (1984) of characterising leaf and tiller death according to the different physiological and physical processes provides an

157 important foundation (see Fig. 3, 6, 8). They measured the rates of death in the component subprocesses, and their experiment, therefore, not only suggested the appropriate mechanistic model of vegetative and reproductive pasture death, but also provided parameters for the model. The experiment of Chapman et al. (1984) thus provides an important theoretical and quantitative foundation for the model presented in this paper. Although their data included Agrostis and white clover, I have focussed on perennial ryegrass as the species dominating herbage dynamics in many improved grass pastures in New Zealand. A refinement to the current model would need to consider the different physiology and vulnerability to grazing of other pasture species. Leaf death rates from Chapman et al. (1984) suggested that leaf senescence is strongly correlated to soil temperature, and this relationship was validated against an independent data set (Vine 1983). Leaf senescence accounts for an estimated 90% of dead matter production in the vegetative sward, so the conclusion that senescence depends strongly on temperature but not on water availability (in temperate pastures) is important. Under high stocking rates, however, animal damage may be a more significant source of death than it was in these experiments. Leaf senescence continues to be important in the reproductive season, but is overshadowed in grazed swards by tiller death from decapitation of reproductive stems. Several processes have been identified that require a deeper understanding and further experimental examination. Background rates of vegetative tiller deaths from treading, fouling, and urine scorching need to be studied to evaluate stocking density effects on death rate (Bullock et al. 1994). Secondly, the importance of secondary reproductive tillers (Korte 1986) to herbage mass accumulation needs to be evaluated in grazed pastures. Thirdly, the quantity of reproductive stem removed by grazing animals and its relation to reproductive leaf ingestion and death needs to be measured for cattle and sheep under different stocking densities. This model is an advance on previous models because it considers changes in perennial grass pasture physiology as well as herbage mass. A future paper will integrate the tissue death models developed here with models of pasture growth, intake, and dead matter disappearance in order to identify strategies of stocking management and water application that maximise the availability of quality pasture through the late reproductive season.

158

New Zealand Journal of Agricultural Research, 1998, Vol. 41

ACKNOWLEDGMENTS Thanks to David G. McCall, David F. Chapman, and David J. Barker for their helpful suggestions during preparation of the manuscript. Thanks also to Jeff Miller for supplying data on flag leaf size.

REFERENCES Alberda, T.; Sibma, L. 1982: The influence of length of growing period, nitrogen fertilization and shading on tillering of perennial ryegrass (Lolium perenne L.). Netherlands journal of agricultural science 30(2): 127-135. Barker, D. J. 1983: The effects of summer moisture stress and its interaction with spring cutting managements on the production and persistence of a ryegrass (Loliumperenne L.) sward. Unpublished MAgSc thesis, Massey University, Palmerston North, New Zealand. 146 p. Betteridge, K.; Fletcher, R. H.; Liu, Y.; Costall, D. A.; Devantier, B. P. 1994: Rate of removal of grass from mixed pastures by cattle, sheep, and goat grazing. Proceedings of the New Zealand Grassland Association 56: 61-65. Bircham, J. S.; Hodgson, J. 1983: The influence of sward condition on rates of herbage growth and senescence in mixed swards under continuous stocking management. Grass andforage science 35:323-331. Blackburn, H. D.; Kothmann, M. M. 1989: A forage dynamics model for use in range or pasture environments. Grass and forage science 44: 283-294. Brereton, A. J.; Carton, O. T ; O'Keeffe, W. F. 1985: Tissue turnover in perennial ryegrass (Lolium perenne L.) during winter. Irish journal of agricultural research 24: 49-62. Bryant, A. M. 1982: Developments in dairy cow feeding and pasture management. Proceedings of the Ruakura Farmers Conference 34: 75-81. Bullock, J. M.; Clear Hill, B.; Silvertown, J. 1994: Tiller dynamics of two grasses—responses to grazing, density and weather. Journal of ecology 82: 331-340. Cayley, J. W. D.; Bird, P. R.; Chin, J. F. 1980: Death and decay rates of perennial pasture as affected by season. Proceedings of the Australian Society of Animal Production 13: 469. Chapman, D. F.; Clark D. A.; Land, C. A.; Dymock, N. 1983: Leaf and tiller growth of Lolium perenne and Agrostis spp. and leaf appearance rates of Trifolium repens in set stocked and rotationally grazed hill pastures. New Zealand journal of agricultural research 26: 159-168.

Chapman, D. F.; Clark D. A.; Land, C. A.; Dymock, N. 1984: Leaf and tiller or stolon death of Lolium perenne, Agrostis spp. and Trifolium repens in set stocked and rotationally grazed hill pastures. New Zealandjournal of agricultural research 27: 303-312. Colvill, K. E.; Marshall, C. 1984: Tiller dynamics and assimilate partitioning in Lolium perenne with particular reference to flowering. Annals of applied biology 104: 543-557. Elgersma, A. 1990: Seed yield related to crop development and to yield components in nine cultivars of perennial ryegrass (Lolium perenne L.). Euphytica49: 141-154. Gallais, A. 1972: Significance of the ease of use of cocksfoot. Fourrages 52: 81-88. Hearne, J. W.; Buchan, A. J. C. 1990: A model to analyse the influence of resource management policies on the economic returns from a traditional African pastoral system. Agricultural systems 33: 77-96. Hirata, M.; Sugimoto, Y.; Ueno, M. 1990: Quantification of cattle dung-related processes in bahiagrass pasture. Japan agricultural research quarterly 23: 190-195. Hunt, W. F. 1970: The influence of leaf death on the rate of accumulation of green herbage during pasture regrowth. Journal of applied ecology 7: 41-50. Johnson, I. R.; Parsons, A. J. 1985: A theoretical analysis of grass growth under grazing. Journal of theoretical biology 112: 345-367. Johnston, M. J.; Waite, R. 1965: Studies in the lignification of grasses: 1. Perennial ryegrass (S24) and cocksfoot (S37). Journal of agricultural science, Cambridge 64: 211-219. Jonsdottir, G. A. 1991: Tiller demography in seashore populations of Agrostis stolonifera, Festuca rubra and Poa irrigata. Journal of vegetation science 2: 89-94. Korte, C. J. 1986: Tillering in 'Grasslands Nui' perennial ryegrass swards: 2. Seasonal pattern of tillering and age of flowering tillers with two mowing frequencies. New Zealand journal of agricultural research 29: 629-638. Korte, C. J.; Watkin, B. R.; Harris, W. 1984: Effects of the timing and intensity of spring grazings on reproductive development, tillering, and herbage production of perennial ryegrass dominant pasture. New Zealand journal of agricultural research 27: 135-149. L'Huillier, P. J. 1987: Tiller appearance and death of Lolium perenne in mixed swards grazed by dairy cattle at two stocking rates. New Zealand journal of agricultural research 30: 15-22.

Woodward—Leaf and tiller death Lemaire, G.; Chapman, D. F. 1996: Tissue flows in grazed plant communities. In: Hodgson, J.; Illius, A. W. ed. The ecology and management of grazing systems. Wallingford, UK, CAB International. Pp. 3-36. Matthew, C. 1992: A study of seasonal root and tiller dynamics in swards of perennial ryegrass (Lolium perenne L.). Unpublished PhD thesis, Massey University, Palmerston North, New Zealand. 247 p. Matthew, C ; Hernandez-Garay, A.; Hodgson, J. 1996: Making sense of the link between tiller density and pasture production. Proceedings of the New Zealand Grassland Association 57: 83-87. Mohtar, R.; Buckmaster, D.; Fales, S. 1994: Amodel for grass growth under grazing system. Presented at the 1994 Winter Meeting of the American Society of Agricultural Engineering, Atlanta. Paper no. 947525. ASAE, 2950 Niles Road, St Joseph, Ml 49085-9659, USA. Ong, C. K.; Marshall, C ; Sagar, G. R. 1978: The physiology of tiller death in grasses: 2. Causes of tiller death in a grass sward. Journal of the British Grassland Society 33: 205-211. Parmenter, G. 1991: The effect of soil water shortage on ryegrass and white clover growing in mixtures. Unpublished PhD thesis, University of Reading, Reading, England, United Kingdom. 249 p. Parsons, A. J.; Robson, M. J. 1981: Seasonal changes in the physiology of S24 perennial ryegrass (Lolium perenne L.): 3. Partition of assimilates between root and shoot during the transition from vegetative to reproductive growth. Annals of botany 48: 111-144. Parsons, A. J.; Johnson, I. R.; Williams, J. H. H. 1988: Leafage structure and canopy photosynthesis in rotationally and continuously grazed swards. Grass andforage science 43: 1—14. Robson, M. J. 1973: The growth and development of simulated swards of perennial ryegrass: 1. Leaf growth and dry weight change as related to the ceiling yield of a seedling sward. Annals of botany 37: 487-500.

159 Robson, M. J.; Deacon, M. J. 1978: Nitrogen deficiency in small closed communities of S24 ryegrass: 2. Changes in the weight and chemical composition of single leaves during their growth and death. A nnals of botany 42: 1199-1213. Schwinning, S.; Parsons, A. J. 1996: Analysis of the coexistance mechanisms for grasses and legumes in grazing systems. Journal of ecology 84: 799-813. Terry, R. A.; Tilley, J. M. A. 1964: The digestibility of the leaves and stems of perennial ryegrass, cocksfoot, timothy, tall fescue, lucerne and sainfoin, as measured by an in vitro procedure. Journal of the British Grassland Society 19: 363-372. Thorn, E. R.; Sheath, G. W; Bryant, A. M.; Cox, N. R. 1986: Renovation of pastures containing paspalum: 3. Effect of defoliation management and irrigation on ryegrass growth and persistence. New Zealandjournal of agricultural research 29: 599-611. Thomas, H.; Norris, I. B. 1977: The growth responses of Lolium perenne to the weather during winter and spring at various altitudes in mid-Wales. Journal of applied ecology 14: 949-964. Vine, D. A. 1983: Sward structure changes within a perennial ryegrass sward: Leaf appearance and death. Grass andforage science 38: 231-242. Wilman, D.; Mares-Martins, V. M. 1977: Senescence and death of herbage during periods of regrowth in ryegrass and red and white clover, and the effect of applied nitrogen. Journal of applied ecology 14: 615-620. Woodward, S. J. R. 1993: A dynamical systems model for optimizing rotational grazing. Unpublished PhD thesis, Massey University, Palmerston North, New Zealand. 157 p. Woodward, S. J. R. 1997: Dynamical systems models and their application to optimizing grazing management. In: Peart, R. M.; Curry, R. B. ed. Agricultural systems modeling and simulation. New York, Marcel Dekker. Pp. 419^74. Woodward, S. J. R.; Wake, G. C. 1994: A differentialdelay model of pasture accumulation and loss in controlled grazing systems. Mathematical biosciences 121: 37-60.