4. 1 Introduction




Yüklə 121.04 Kb.
tarix26.04.2016
ölçüsü121.04 Kb.
4. Assemble available information on the dynamics of weed species seed populations over winter in arable fields
4.1 Introduction

Arable weeds have undergone major declines over the last century, in common with many groups of species associated with arable farming (Robinson & Sutherland 2001; other refs). The reason for these declines is clearly the increase in intensity of management associated with increases in the efficiency of herbicidal control of weeds, together with other advances in crop management and crop varieties. Although in one sense these changes in farming practice represent important advances since the efficiency of land use and net food production have increased dramatically, there are increasing concerns about the impacts such changes have had on the biodiversity of arable farmland (refs).


The impacts of intensification of agriculture on biodiversity are both direct and indirect. The direct impacts occur on pest species that are the target of control, whilst the indirect impacts effects occur on those species that may depend on these pest species. This is the situation with farmland birds, which rely on arable weeds and invertebrates living in arable fields for food. Thus, in order to develop a detailed understanding of the nature of these indirect effects on species of conservation concern, it is necessary to develop a complete understanding of the factors determining the long-term dynamics of the food species upon which they rely, as well as to predict the likely consequences of future changes in farming practice.
The seeds of arable weeds are a major source of food for many farmland birds during winter (ref), and in order to understand how changes in farming practice will affect farmland bird population it is necessary to predict the population dynamics of arable weeds. The framework for predicting weed population dynamics under contrasting management and environmental conditions is well developed (e.g. Firbank & Watkinson 1986; Doyle et al. 1990; Cousens & Mortimer 1995; Freckleton & Watkinson 1998a, b; 2001), and uses models for population dynamics, parameterised under varying conditions. The applications of this approach are varied, but importantly models of this nature can form the basis for general strategic models that predict the likely impacts of future changes to farming, such as the introduction of GM crops (Watkinson et al. 2000).
The aim of this work was to assemble information that could be used to parameterise models for the population dynamics of arable weeds, which predict the numbers of arable weed seeds available as food to farmland birds during winter. This work extends an approach developed by Freckleton & Watkinson (1998) and Lintell-Smith et al. (1999) who showed that it is possible to use literature derived information to generate robust population dynamic models for weed populations. The parameters estimated in this section were used to parameterise the models that form the basis for the final simulation model (Objective 12).
4.2 Methods
4.2.1 Approach
Fig. 4.1 shows a general outline of the life-cycle of an arable weed. The changes in numbers of weeds from one year to the next are mainly dependent on three processes: (i) seed production of the weed; (ii) survival and recruitment of new weeds from seeds in the soil and soil seedbank; (iii) the influence of management on mature weed numbers. The life-cycle shown in Fig. 4.1 is of course extremely general, and the exact nature of dynamics will depend on rotation, crop and form of management. However, despite this potential complication, a number of key variables have to be estimated for each weed population in order to parameterise models. These are: the number of seeds produced per plant, the proportion of seeds germinating and surviving in the soil, and the proportion of plants surviving control. Given these parameters it is possible to predict the number of weeds available during winter for birds.


Fig. 4.1 A schematic outline of the population dynamics of an arable weed. The key variables in determining population dynamics are the proportion of plants killed by control; the numbers of seeds produced by weed plants; and the proportions of seeds surviving in the soil and germinating per year. All of these parameters have to be estimated in order to predict the effects of changes in farming practice on arable weeds.


This model parameterisation exercise focussed on 6 species of weeds. These were Chenopodium album, Poa annua, Stellaria media, Fallopia convolvulus, Papaver rhoeas and Alopecurus myosuroides, although where possible we collated information on a variety of common species. The first four of these are very important sources of bird food during winter (Objectives 3 & 5). The latter two are known to be of economic importance, but less important as food for farming birds. It was decided to include these species in order that the impacts of changing management could be predicted separately for species of economic importance as well as for species that are significant food resources for birds.


      1. Seed Production

Seed production is defined as the number of seeds produced per weed plant. The number of seeds produced by a plant depends on its size, which in turn may depend on a number of factors, especially the number of crop and other weed plants present (Firbank & Watkinson 1986). From the point of view of predicting the population dynamics of weeds, of key importance are estimates of the seed production of plants in isolation. This number of seeds forms the baseline for predicting seed production under different conditions. In the literature, weed performance may be recorded under a variety of conditions. In reviewing the seed production of plants in the cropping phase of the rotation, we therefore distinguish estimates of maximum, minimum and mean seed production of plants, representing the production of plants in isolation, competition and the average, respectively.


As stated in our original objectives we set a priority on obtaining information for weeds growing in winter stubbles. We obtained information on seed production in stubbles from another DEFRA funded project (XXXX BTO [I referred to this as “an ongoing study conducted by the BTO (L. Robinson & P. Atkinson, unpublished data)”]). The data from this study we found to be unique in the weed ecology literature. As we emphasise below, this is an area that urgently requires further work.
Seed production is typically affected by the density of both crop and other weed plants present, with the number of seeds produced per weed plant declining as population density increases. This decline is generally found to be well described by an equation of the following form (Watkinson 1981; Firbank & Watkinson 1985):
(4.1)
The parameter sm is the mean seed production of an isolated plant, and estimated independently as outlined above. The other parameters are competition coefficients, describing the reduction in mean seed production of increasing densities of weeds (Nw) and crop plants (Nc), respectively. We collated estimates of these parameters where they were available, however these are typically only available for the most commonly studied weed species.


      1. Seed bank dynamics

Seeds of most weeds are capable of surviving in the soil for a number of years. Following a single year of seed production, the number of seeds remaining through time, as well as the numbers of seeds germinating each year, will depend on both the rate of seed mortality and the mean proportion of seeds germinating per year. Unfortunately this is not taken into account by weed ecologists. Frequently only the rate of seedbank depletion is recorded from experimental studies, which on its own is useless in modelling changes in weed numbers (e.g. see Rees & Long 1993).


We estimated rates of seed mortality and germination in the following way. Within a given time interval (typically a year) the poportion of seeds germinating and dying are defined as g and m, respectively. It is important to realise that the proportion of seeds remaining after a given time interval is not solely a function of mortality, but also of germination which may represent a considerable loss of seeds. After 1 year, the proportion of seeds that have not germinated or died is given by:
(4.2)
In equation (4.2) the proportion of seeds that survive are given by (1 – m), and the proportion that does not germinate is (1 – g), thus the proportion of seeds remaining is the product of these two quantities. More generally, after T years the proportion of seeds remaining is:
(4.3)
And the proportion of seeds to have emerged as seedlings is:
(4.4)
Given estimates of the proportion of seeds remaining, and the number to have emerged as seedlings after some time T, the rates of germination and mortality may be estimated by solving equations (4.2) and (4.3) as simultaneous equations.

The most comprehensive source of information on seed bank dynamics is the dataset collected by Roberts & colleagues over a number of years(refs). This dataset is unique in terms of the quality of data collected and large taxonomic coverage. For many species this dataset is the only source of information on seed survival. Furthermore, this study measures recruitment of seedlings, at the same time as monitoring seed bank numbers, and it is thus possible to use the above analysis to estimate m and g. Simple monitors of seedling emergence or decline in seed numbers through time cannot be used in this way as both PR and Pe have to be measured simultaneously to be able to separate mortality and germination. We re-analysed all data from these papers to estimate rates of seedbank decline for common weed species. We augmented these data with other information from literature reviewed that allowed separate estimation of germination and mortality. For most species, however, the Roberts dataset is the only source of information that allows this.




      1. Baseline densities and management

Management affects the dynamics of weed seeds in three important ways. Firstly, herbicidal control determines the proportion of weed plants that survive to yield seeds. Secondly, the timings of management events, particularly the timing of sowing and harvest, determine which weeds are capable of completing their life-cycles within the crop. And thirdly, the timing and nature of cultivation determines whether seeds remain on the soil surface or whether they become incorporated into the soil seedbank, as in winter stubbles. This section details how we estimated how cropping affects the survival of weeds in different crops, and the phenology of the weed species.


In principle it should be possible to model the effects of management through estimating the survival of weeds subject to herbicidal management in different crops. If p is the proportion of plants surviving herbicidal control, then this proportion could be estimated from field trials and used to estimate the effects of management in a model. Unfortunately this approach generally fails because the parameter of p is generally estimated with far too much error to be directly applicable (e.g. see Freckleton & Watkinson 1998b). To give an example of this, consider a stable weed population, which emerges at a density of 100 seedlings m-2. If an adult plant produces seeds that yield 100 seedlings per year, then for the weed population to be constant from one year to the next, exactly 0.99 of the emerging seedlings must be killed per annum. This balance implies that the proportion p must be estimated with an accuracy of at least 3 decimal places to be capable of accurately recreating the population behaviour in a model. If, for instance, p were estimated as 0.995 (error of +0.005), then the population would be predicted to halve each year. On the other hand, if p were estimated as 0.98 (error of –0.01), then the population would be predicted to double from one year to the next. Thus, an error range of just 0.015 in this case could lead to a range of predictions that include population doubling, constancy or halving. In practice the problem is worse than this since most weeds are capable of producing thousands of seeds per plant, or at the stand level, tens of thousands to millions of seeds m-2.
We therefore assessed the effects of management in two ways. Firstly we asked ADAS for a qualitative assessment of the efficacy of control of weeds in different crops, based on their long-term experience with herbicide trials. For each species it was recorded whether it was generally easy to control or not, together with the typical proportion of plants killed by a single herbicide application. Secondly, we used the data from the long-term IFS experiment to estimate the net and relative densities of the different weed species in different crops. We combined these data with information on seed production, to estimate typical rates of winter seed rain for the range of cropping practice encompassed within the experiment. The densities of weeds were also used to set values of p, the efficacy of control, in the modelling exercise (Objective 12).
4.2.5 Other models
Models have been developed for the population dynamics of several other weed species (Chenopodium album, Alopecurus myosuroides, Stellaria media, Poa annua). We reviewed these models in order to retrieve relevant parameter estimates for the modelling exercise.
4.3 Results
4.3.1 Seed production
Table 4.1 summarises information retrieved on the seed production of individual arable weeds, whilst Table 4.2 summarises information on the density-dependence of seed production. Seed production is rarely estimated in a systematic way in weed studies, since most weed research concentrates on the yield of crop. From those studies that have estimated weed seed production, Table 4.1 shows that most weeds are extremely fecund, generally capable of yielding thousands of seeds per plant. The figures in Table 4.1 refer to plants growing according to normal or optimal phenology. Many, but not all, species are also capable of yielding seeds in winter stubbles. Table 4.1 also summarises information on the relative production of weed seed in winter stubbles from the study of Atkinson & Robinson (unpub.). The percentages illustrate the production of weed seed in stubble as a proportion of the production under normal conditions. Table 4.1 suggests that most species are capable of yielding c. 10-15% of maximal seed production, if allowed to grow in winter stubbles.
Table 4.1 Summary of information on seed production of arable weeds at the beginning of winter and during winter. Recorded are the range of seed production per isolated plant, the mean seed production, together with the production of seed in winter stubbles, expressed as a percentage of the maximum in autumn.

Species

Range

Max

Min

Mean

Stubble Production

Reference

Fallopia convolvulus


-

10000

-

-

24.30%

Hume et al. (1983)













11900




Stevens (1932)

Polygonum aviculare

880-4010

4010

880

2898

15%







-

-

-

6380




Stevens(1932)

Poa annua













10%




Stellaria media

1000-10000

10000

1000

-

10%

Salisbury (1964)

Alopecurus myosuroides










0%




Papaver rhoeas

1000-10000

10000

1000

-

0%

Jagli (1992)




-

-

-

170000




Salisbury (1964)



















Lintell Smith (1995)

Chenopodium album

200 - 20000

20000

200

6000

6.10%

Erviö (1971)




30000 - 230000

230000

30000

100000




Harrison (1990)

Veronica persica

1000-10000

10000

1000

-

-

Harris & Lovell (1980)

Galium aparine

100-1000

1000

100

-

-

Malik & Vanden Born (1988)



















Lintell Smith 1995

Anisantha sterilis

20-1000

1000

20

100

0%

Lintell Smith et al. (1999)



















Cousens et al. (1988



















Firbank et al. (1984)

Solanum nigrum

1000-10000

10000

1000

-

-

Salisbury (1942)

Senecio vulgaris

-

10000

-

-

-

Salisbury (1976)

Capsella bursa-pastoris

10-1000

1000

10

-

-

Grime et al. (1988)




1000-10000

10000

1000

-




Salisbury (1942)













38500




Stevens (1932)

Sinapis arvensis

1000-10000

10000

1000

-

-

Mulligan & Bailey (1975)



Table 4.2. Density-dependent weed seed production in crops and winter stubbles, for species considered in the modelling exercise


Species

Parameters in crop

Parameters in winter stubble




Sm

a

Sm

a















Chenopodium album


230,000

0.1

1400

0.005

Poa annua

100

0.00233

100

0.00233

Stellaria media

1,000

0.0125

100

0.000625

Fallopia convolvulus

2,898

0.05

2,898

0.05

Papaver rhoeas

8,000

0.008

No seeds produced in winter stubbles

Alopecurus myosuroides

300

0.004

No seeds produced in winter stubbles

4.3.2 Seed bank mortality and germination


Table 4.3 summarises the estimates of seed mortality and germination from seedbank studies. Fig. 4.2 shows how these parameters translate in to rates of seedbank decline. As shown in Fig. 4.2a, species vary considerably in the rate at which seedbank numbers decline through time. Species such as Chenopodium album and Fallopia convolvulus have extremely long lived seedbanks. The likely persistence of populations, however, depend not only on the rates of decay of seedbanks shown in Fig. 4.2a, but also depend on the input of seed into the soil. Fig. 4.2b shows the rate of decline in seedbank density following an initial seeding from a single individual. It is clear in Fig. 4.2b that whereas the density of some species may decline to less than one seed in under 20 years, declines in species such as Chenopodium album take many years longer to reach such a level.
Fig. 4.2 The long-term dynamics of seedbanks of arable weed species valuable as food for birds. (a) The poportion of seeds remaining following an initial sowing. (b) The total number of weeds remaining following the seeding of a single plant.

Table 4.3 Estimates of parameters determining seed dynamics of annual weeds. The key parameters are the rate of germination of seeds (g), the rate of seed mortality (m), which have been estimated separately from uncultivated and cultivated soil. Also shown is the proportion of seed lost to germination and mortality over 6 years in cultivated soil.






Uncultivated




Cultivated







Species

g

m

g

m

Loss in 6 yrs

Polygonum aviculare

0.0286

0.12

0.134

0.242

0.92

Fallopia convovulus

0.059

0.15

0.1487

0.1996

0.9

Poa annua

0.061

0.16

0.187

0.193

0.92

Stellaria media

0.043

0.188

0.216

0.254

0.96

Senecio vulgaris

0.063

0.241

0.273

0.477

0.997

Papaver rhoeas

0.026

0.208

0.131

0.261

0.93

Spergula arvensis

0.036

0.261

0.166

0.375

0.98

Thlapsi arvense

0.038

0.081

0.235

0.141

0.92

Veronica heredifolia

0.071

0.096

0.427

0.189

0.99

Veronica persica

0.048

0.127

0.303

0.334

0.99

Chenopodium album















Fig. 4.3 compares the rate of decline in arable weed seed densities in the soil (Robinson & Sutherland 2000), with the rates of seedbank decline in Fig. 4.2. It is clear in this figure that observed rates of seed density decline are many orders of magnitude lower than rates of seedbank depletion. This indicates very clearly that the observed rate of decline in weed seed numbers is the consequence of reductions in weed seed production resulting from increases in the efficiency of control. The relatively shallow decline in mean soil seed densities, compared with the very rapid decline in the densities of seeds already in the soil seedbank, indicates that considerable seed production must occur in order to account for this difference. By comparing these rates of decline, we estimate that at least 25% of the seedbank must (this figure is generated using the rate of decline for Chenopodium album), on average, be renewed to account for this difference. This implies that average seed production per square metre in arable crops has declined from 1000 m-2 to 50 m-2 over the course of less than a century.


F
ig, 4.3
Observed rate of decline of seeds in arable soils (Robinson & Sutherland 2000), compared with the rates predicted by the fitted models (Fig. 4.2). The large difference between these two rates implies that considerable seed production occurs, despite increases in control. We estimate that at least 25% of the seedbank would have to be renewed per annum in order to account for the difference between the observed and predicted rates.

4.2.4 Baseline densities and management


Table 4.1 summarises the information on the efficacy of control supplied by ADAS for the species used in the model .
Table 4.1 Summary of information on the relative ease with which weeds are controlled in different crops, together with typical proportions of weeds killed by herbicide application.





Winter wheat

Winter rape

Sugar beet

Spring barley

Peas

Species

herbicide control

Ranking

herbicide control

Ranking

herbicide control

Ranking

herbicide control

Ranking

herbicide control

Ranking

Polygonom spp.

>95%

easy

100%

-

95%

Moderate

50-95%

difficult

50-80%

awkward

Chenopodium album

>95%

easy

100%

-

95%

Moderate

>95%

awkward

95%

easy

Senecio vulgaris

>95%

easy

95%

easy

90%

moderate

>95%

easy

95%

easy

Poa annua

100%

v. easy

90%

easy

95%

moderate

30-95

awkward

80%+

easy

Stellaria media

100%

v. easy

90%

easy

100%

easy

>95%

easy

80%+

easy

Alopecurus myosuroides

90%

moderate

95%

easy

95%

easy

40-80%

difficult

-

-


4.2 Summary
The information collated yielded parameters that were fed into models for the population dynamics of the 6 species of weeds in conventionally managed systems. These models then form the basis for the final modelling exercise (Objective 12).


Verilənlər bazası müəlliflik hüququ ilə müdafiə olunur ©azrefs.org 2016
rəhbərliyinə müraciət

    Ana səhifə