The Concept of Interaction

18.2.1 General background

Some of the topics addressed by Krupa et al. (1998a) relate to the application of existing scientific knowledge of environment/crop interactions to the global spatial grid for revealing patterns of their spatial variability at the present time. Other topics are more fundamental for understanding the dynamic relationships that occur at any spatial level, from global to the local scale. The contents of this chapter are in the latter context, without considering the spatial term. The focus is on the combination of factors that impact production of a given crop, but with strong emphasis on climate-change factors.

The joint effects of elevated UV-B, [CO2], surface [O3], temperature, moisture, diseases and insect pests on crops are ideally dealt with in an integrated manner, if we are to understand or predict their potential limitations to crop production (Krupa et al., 1998a). Such an effort requires a dynamic systems perspective. Due to the complexity of the task, it can only be implemented using advanced computer technologies and complex models yet to be developed. Experimental and observational findings of the joint effects, (even taken pairwise) of the aforementioned variables and their interactions on plant responses are fragmented and are only beginning to be studied in a quantitative manner.

Ever since cultivation of crops began, and indeed even before, plants have been exposed to increases in [CO2], extremes of temperature, moisture, light and fluctuations in nutrients. All of these factors represent needed resources that can become growth limiting. In addition, crops have been and are being exposed to increases in tropospheric [O3] and perhaps, in the future, increases in ground-level UV-B. Added to these factors are the effects of plant diseases, insect pests and competition from weeds. Therefore, agricultural production is in part a result of the ability of a crop to grow under the impacts of these multiple stress factors and within the management practices.

There is considerable interest in how plants respond to multiple environmental stresses in a changing environment. In this context, studies of plant stress have followed the normal course of examining each of the single main effects first, followed by studies of interactions of some of the effects (Forseth, 1997).

18.2.2 What is meant by an interaction?

The detection of interactions among effects (individual environmental stimuli such as some level of [CO2] or UV-B) on measurements of plant growth is not a simple task, and we will begin by defining 'interaction' as it is used here, as well as whether these interactions detected are meaningful. Some confusion exists between strict statistical and biological definitions and methods of detecting interactions. Statistical interactions are precisely defined as quantitatively determined interaction sum of squares components in linear models that partition variance. Because statistical sophistication is not always present in otherwise reasonable papers that attempt to measure interactions, the following terminology is used here, although we are aware that other definitions of these terms can be found.

Two effects are additive only when they affect the measurement in a linear or arithmetic manner, and the combined effect is the sum of the individual effects. This additivity can involve the sign of the function, if the two individual effects influence the measurement in opposite directions.

Anything other than additive is considered as an interaction. Two kinds of interactions that can be identified are known by multiple terms:

• If the combined effect is less than the sum of the individual effects, the relative interaction is ameliorative, or competitive, or the effects are said to cancel or nullify one another.

• If the combined effect is greater than the sum of the individual effects, the interaction is synergistic, or the effects are said to reinforce or amplify one another.

Simple statistical tests can be used to show whether the combined effects are significantly greater or less than the additive expectation. The same ideas apply to two factors that act in an opposing manner, but terminology becomes confusing. Such cases show most clearly that statistical definitions of interactions are not always synonymous with biological definitions; many would consider cases of cancelling factors to be important biological interactions irrespective of whether the cancelling effects interact statistically. A consideration of such cancelling effects will determine the importance of studies of combinations of environmental influences on plant health.

18.2.3 Curve-fitting approaches

According to Krupa et al. (1998a), at their simplest, statistical methods of detecting interactions use one exposure dose of each of the independent variables. However, a more complete picture is desirable, and methods to describe the dose-response surface of two agents acting together on some measure of plant growth or productivity are outlined in Kreeb and Chen (1991). If the dose-response relationship of one or both agents is non-linear, the surface will not be planar. A non-planar surface does not automatically signify an interaction. The clearest indication of an interaction is probably the existence of a complex surface or landscape in which specific areas are significantly above or below the expected surface, based on mathematical or statistical curve fitting, and expected additivity. For suspected peaks or valleys, it should be possible to reduce the problem to a simple test of whether the actual effect on plant growth is significantly different from the expected effect from additivity of the two dose-response curves. In this sense, a response surface is valuable because it may reveal interactions of two agents at levels that might otherwise have been undetected if only one or a few levels were tested (Myers, 1971). Beyond the few mathematical relationships discussed by Kreeb and Chen (1991), there is desktop computer software (e.g. TableCurve 3D.19, by Jandel Scientific, San Rafael, California, USA) that enables the exploration and representation of response surfaces for data on three factors from among 243 polynomial equations, 260 rational equations and 172 nonlinear models, fitting up to 36,582 equations, and then sorting the best equations according to goodness of fit (Krupa et al., 1998a).

One of the most serious problems with showing interactions is the choice of response measurement and associated scale issues (Falconer, 1960). Reality of the level, duration, etc., of each effect is discussed in a later section of this chapter in relation to individual studies. Researchers have used many different measurements of plant responses to climatic change effects, and results are not always consistent; interactions detected with one type of plant measurement often fail to appear with another type of measurement in the same studies. There is no single best type of measurement to use, and the choice may be based on many things, including ease of measurement (e.g. pollen tube length), perceived importance (e.g. biomass), or economic significance (e.g. agricultural yield). The relationships between different measurements that might be used are often non-linear. Transformation of some measurements may be called for, because the data are not normally distributed or because means are not independent of variances. In reality, very few of the studies of possible interactions explore these statistical issues, and so it may be difficult to determine whether the interactions are genuine - and, if they are genuine, are the measurements that show them meaningful? Data transformation may make statistical sense, but does it make biological sense? At this stage of investigation any measurement can be defended as being informative. Eventually, it may be better to concentrate on measurements that are bound to translate into yield or quality loss in food and fibre crops.

18.2.4 Potential combinations of interactions

Even if stress factors or stimuli are taken singly, their effects on plant growth are complex. Some are inherently complex, such as plant diseases, which affect plants in a myriad of ways and can interact among themselves. Stress factors should not be taken singly, because they almost never occur singly, and when they occur together their interactions should be understood. For simplicity, assume that there are ten stress factors. The total number of combinations of n parameters taken as some number at a given time is 2n - 1. For a set of ten potential stress agents, this means 1023 potential interactions. Subtracting the ten single factorial relationships to a dependent response variable leaves a total of 1013 potential interactions of the combinations of two or more stress agents affecting a single crop response. Clearly, this is a daunting challenge for crop ecologists. It is easy to display the interactions in a graphic form for any three of the 11 dimensions (including the dependent response variable), but a larger number of interactions cannot be clearly and completely illustrated (Krupa et at, 1998a).

18.2.5 Types of interactions

Synergistic change occurs when two or more environmental or ecological processes jointly interact, either simultaneously or sequentially, and the result is not a simple sum of the otherwise individual responses, but instead is multiplicative. This response appears, as an amplification of effects and a compounding of the impacts, to be consistent. Sometimes the tolerance of a species to one source of stress becomes reduced when other stresses are experienced simultaneously (Myers, 1992). Very little is known about such ecological or physiological synergistic interactions and the mechanisms responsible for them.

According to Krupa et at. (1998a), biologically, the value of harvestable crop parts for food, fuel, shelter or fibre is a result of stored energy (carbon, C), along with the nutrient content. For the purposes of national and international policy, initial considerations are probably given to the production of edible crop parts, since food is the most urgently needed commodity. A simple question is: 'What might happen to the crop productivity - namely, the amount of carbon stored?' Although there are elaborate models of crop growth processes, when viewed at the most fundamental and general level, the growth of any crop is a balance between carbon uptake and subsequent carbon storage, after the necessary carbon losses. This is similar to a financial budget: income is disbursed into savings after losses through expenses. The same line of reasoning can be used for nutrients such as nitrogen, but that aspect will be considered only in a limited way in the following discussion. Plants absorb carbon from the atmospheric carbon dioxide in the daytime (photosynthesis) and they normally lose varying portions of that carbon both in the day (photorespiration) and at night (dark respiration):

In addition to the normal maintenance costs, plant stress from unexpected increases in the environmental contaminants causes additional carbon losses within the crop as plants attempt to repair the stress effect.

If carbon uptake increases due to increased [CO2] and possibly due to climatic warming, and if carbon losses also increase because of episodes of increased UV-B radiation interspersed between episodes of tropospheric [O3], it is conceivable that the net effect on carbon storage might only be a negligible change (increase) in the balance:

C-uptake (increase) = [C-storage (increase)] - [C-losses (increase)] (Eqn 18.2)

Even with a direct fertilization effect of increased ambient [CO2], if carbon losses increase substantially because of increases in atmospheric contaminants and from changes due to less optimal levels of meteorological resources needed for crop growth, then carbon storage in plants could decrease, i.e. there will be decreased crop production:

C-uptake (increase) = [C-storage (decrease)] - [C-losses (increase)] (Eqn 18.3)

This idealized concept would vary according to the differences in the sensitivities of crop species and cultivars to multiple, simultaneous or sequential exposures to different stress factors. Within an optimal range for crop growth, light, temperature, moisture and nutrients are necessary resources. The soil nutrient status is usually managed by fertilizer usage practices, but if decreases in seasonal light levels (e.g. increases in cloud cover), or changes in temperature and/or moisture occur in a given region, those changes coupled with any increases in UV-B radiation, tropospheric [O3], or increases in 'biotic factors' such as weeds, diseases or insect pests might be environmentally undesirable. The questions that require attention are: 'How can increases in the various undesirable parameters jointly affect carbon losses from changes in carbon storage in plant parts?' and 'What is the nature of the experimental and/or observational evidence?'

There is a difference between the potential joint effects of two or more physicochemical factors (e.g. elevated UV-B and elevated [CO2]) on individual species and such effects on species competition. Examples of the latter are discussed by Billick and Case (1994). Interactions between changes in the physicochemical processes and their resulting effects on plant host and pathogen and/or insect pests, and the interactions between the organismal species themselves in that system, comprise one of the most complex issues, because it can involve both types of interactions. However, relative to increased [CO2], UV-B and [O3] singly, current evidence suggests the general features presented in Table 18.1, although the emphasis of this chapter is on crop yield.

Guide to Alternative Fuels

Guide to Alternative Fuels

Your Alternative Fuel Solution for Saving Money, Reducing Oil Dependency, and Helping the Planet. Ethanol is an alternative to gasoline. The use of ethanol has been demonstrated to reduce greenhouse emissions slightly as compared to gasoline. Through this ebook, you are going to learn what you will need to know why choosing an alternative fuel may benefit you and your future.

Get My Free Ebook


Post a comment