• No results found

Declaration

1 General Introduction

1.2 Study objectives

In the previous section it was concluded that there is both an increasing pressure on growers to deliver agricultural products that are produced in a sustainable way and at the same time a risk that a decreasing portfolio of herbicides or continuous use of reduced herbicide rates might jeopardise weed management and economic profits in the future. There is now a broad recognition that weed management strategies have to be diversified to meet these challenges and cultural control is increasingly seen as a vital part of integrated weed management.

The general objective of this study therefore was to evaluate the relative merit of weed management strategies based on cultural control on the long term development of weed populations.

Within the DEFRA project ‘Understanding the relative establishment times of crops and weeds within the changing seedbed’ - HH3406SX, of which this Phd research was part, carrot and onion had been chosen as target crops. The former because it is the vegetable crop hardest hit by herbicide loss and the latter since it is particularly vulnerable to yield loss if weeds are insufficiently controlled. The value of home produced carrots represents by far the highest value for all field vegetables (£167 million in 2005), whereas onions are less important in terms of value but constitute a relatively high proportion of the vegetable planted area (DEFRA, 2006). Two problematic weed species, Stellaria media (L.) Vill. and Tripleurospermum inodorum

(L.) Schultz Bip. (see Figure 1-6) were initially chosen as model species.

Scentless mayweed is often referred to as Matricaria perforata but this thesis will follow Stace (1997) and use Tripleurospermum inodorum. The weed species have been chosen as both are quite prevalent in field vegetables (see Table 1.2 in Grundy et

Figure 1-6 Two weed species frequent in many crops and chosen as ‘model’

al., 2003), because for both weeds considerable amounts of biological data is available from other studies and because they possess contrasting life history characteristics (see Table 1-1).

Table 1-1 Key contrasting life history characteristics of the two model species

Stellaria media Tripleurospermum inodorum

Depth-mediated germination

practically no germination if seeds are located > 5 cm.

practically no germination if seeds are located > 2 cm.

Plant morphology prostrate upright

Flowering induction not sensitive to daylength sensitive to daylength

Due to the buffering effect of the weed seedbank, the evaluation of weed management strategies typically requires long-term field studies (Norris and Légère, 2004; Thomas

et al., 2004), especially if it involves crops grown in rotation. Such long-term trials are often not economically feasible. Population dynamic models on the other hand, are an efficient tool to compare the effect of weed management strategies on the long- term population dynamics (Kropff et al., 1999) and carry a number of advantages. In a multi-component system a model brings together the knowledge of the parts and provides a coherent view of the behaviour of the complete system. Once a model is in place, more targeted, instead of ad hoc, experimentation can be planned (France and Thornley, 1984) and using sensitivity analysis ‘Achilles heels’ can be identified in a weed’s lifecycle that can then be specifically targeted (Davis et al., 2003).

Climate change provides both threats and opportunities to agriculture (Anonymous, 2005) and when proposing alternative weed management strategies it is important to understand how changing climate conditions might affect and interact with these strategies. It has been suggested that weeds may respond stronger than crops to resource changes (light, water, nutrients or carbon dioxide) due to their larger genetic diversity (Ziska, 2004).

Specifically then, the objective of this study was to develop a modelling framework, capable of simulating the impacts of (weed management strategies based on) cultural control and climate on the long-term population dynamics of weeds in field vegetable systems. The intended aim of the model is as a management aid. The modelling framework will be further referred to in the thesis as ECOSEDYN which stands for: Effects of Cultural control and climate On SEedbank DYNamics.

12

1.3

Study approach and outline of the thesis

At the start, literature reviews were carried out to select promising cultural control components for the weed management scenarios and to make sure that the weed biology was represented as accurately as possible in ECOSEDYN (see further Chapter 2). The approach followed to meet the objectives is graphically presented in Figure 1-7. Study objective Implemented model Time System to be modelled Literature research Problem Weed management guidelines C Model output Additional experiments Knowledge Modelling as a research tool C Weed man. scenarios Conceptual model Knowledge gaps Interviews / Questionnaires C

In addition a questionnaire (see Appendix 1) was sent out to 10 carrot growers and to an agronomist at the Allium & Brassica Centre (Carl Sharp) to find out about current common practices of commercial carrot and onion growers in the UK. The questions related to cropping system characteristics and the nature and timing of farming practices (weed management and cultivation). A further visit to three large commercial carrot growers (Elveden Estate, Isleham Fresh Produce and Watton Produce), going through the same questions, yielded additional information.

It is important to emphasize that the development of a model is not a one-way process but an iterative process of revisiting previous stages when flaws are identified and new insights gained (Balci, 1994; Jackson et al., 2000). This aspect is represented graphically by the dashed arrows (feedback loops) in Figure 1-7. This feature also made it inevitable that in the thesis forward and backward references to sections of other chapters are given on several occasions. The study approach is narrated over the next paragraphs through outlining the content of the different chapters.

Chapter 2 documents the conceptual modelling phase. The literature was thoroughly

reviewed to get a comprehensive understanding of the system, to evaluate the different ways in which system components have been and can be modeled and to select the cultural control options that would form the components for the weed management scenarios. Together this resulted in the conceptual model (Robinson, 2006), a description of the simulation model addressing the biological aspects of the system. The assumptions, mathematical representation and parameterisation of those model components that operated independently of the weed management scenarios are presented in this chapter. Evaluation of available model components and parameters resulted in the identification of key areas for additional research which led to several field experiments being conducted that are discussed in the next chapters. With the project progressing, the understanding of the system increased and this inevitably led to new areas for potential research. However due to time constraints these gaps could not be addressed through experiments and therefore had to be bridged by additional assumptions in the model.

In Chapter 3 and 4, experiments are described that were aimed at evaluating existing and formulating new model components. To continue the flow of the text, model implementation is presented at the end of each chapter. If the representation depends

on the specifics of other model components or the weed management scenarios this will be referred to in the text.

Chapter 3 explores the validity and robustness of the present ways in which seed

movement by cultivation is modelled. The validity of the algebraic approach (multiplication of transition matrices) of modelling seed movement when multiple implements are involved and the assumptions in the more mechanistic models for ploughing as proposed by Colbach et al. (2000) and Roger-Estrade et al. (2001) were empirically evaluated. To compliment the experimental work that led to the transition matrices of the four implements by Mead et al. (1998), a transition matrix for the plough was generated. This chapter provides the theoretical justification, but not the exact implementation, for modelling vertical seed distribution. The model implementation of vertical seed distribution due to cultivation and harvest is given at the end of the chapter.

Chapter 4 deals with the theme ‘Plant growth and reproduction’. Experiments

identify:

• the environmental variable that best explains biomass increase of weeds during early growth

• the onset and increase of flowers relative to biomass increase

• the biomass – seed relationship for plants of different ages

• the relationship between timing and duration of flowering and plant biomass Together the results lead to the formulation of a set of model components for biomass increase, flower production and seed production over time. The mathematical representation and parameterisation of the component models for Biomass increase, Flowering and Seed production are presented in the last section of the chapter.

Chapter 5 represents the second part of the modelling process. First the specifics and

hypotheses regarding the weed management and climate change scenarios are given. Then the modelling methodology, e.g. the model structure and the analysis of model output, is explained. Lastly the results of the model simulations in ECOSEDYN are presented and discussed.

In Chapter 6 a summary of the experimental research achievements is given first. I then reflect on model development, provide explicit weed management guidelines based on model output and discuss where future work on ECOSEDYN should be focused and how this fits in with the future of weed research in general.

The key content of the chapters and the way the chapters relate to each other is graphically represented in Figure 1-8.

3 – Seed redistribution Experimental results Model implementation

6 – Synthesis and discussion Formulation of guidelines General discussion 2 – Modelling - Part I Model formulation Gaps in research 5 – Modelling - Part II Results Model / Scenario Implementation 1 – General Introduction Context Objectives

4 – Growth and Reproduction Experimental results Model implementation