Trade‐offs in plants and the prospects for breeding using modern biotechnology

Evolutionary ecologists have devoted much effort to identify and test models for the occurrence of trade-offs in plants, particularly within the frame of reproduction costs (Reznick, 1985; Stearns, 1989, 1992; Bazzaz, 1997). Thus, the concept of trade-offs is a major issue in functional and evolutionary ecology, but is perceived sceptically by many researchers involved in plant breeding. However, as trade-offs in many cases are brought about by constraints such as physiological and/or genetic links between traits, they set limits to the possible phenotypes that can evolve. This should have major implications for plant breeding. The evolutionary pattern of traits related to, for example, growth rate and stress tolerance, lends some evidence to the importance of trade-offs for shaping plants. However, the mechanisms behind important trade-offs remain unclear in most cases. Modern biotechnology is likely to offer powerful tools to identify these mechanisms and to separate physiological from genetic trade-offs. However, physiological trade-offs could turn out to form major limitations for the construction of superior plants with the methods of modern biotechnology. In general, trade-offs are considered a special type of constraint which involves links between two or more traits, and cost-benefit considerations (Pease & Bull, 1988; Stearns, 1989, 1992). In the simplest case, a trade-off is defined as a negative association between two or more traits (Southwood, 1988; Stearns, 1992). This definition is often applied by ecologists and also plant breeders (see below), but does not invoke any mechanism behind the anticipated trade-offs at the physiological or genetic level. It is therefore difficult to evaluate the possible implications that the existence of trade-offs (e.g. between fast growth rate under optimum environmental conditions and high stress tolerance; Chapin et al., 1993; Weih, 2001) might have for plant breeding. Where definitions involve any prediction of the underlying mechanisms, these are commonly established on an ecological, physiological or genetic level. Ecological mechanisms (or macro-evolutionary trade-offs (Stearns, 1992)) refer to the interactions among organisms and their biotic or abiotic environment (Magnhagen, 1991) and ultimately must be related to critical physiological and/or genetic processes that result in different phenotypic responses of organisms across an environmental gradient. Physiological mechanisms (or physiological trade-offs (Stearns, 1989)) are invoked by physiological links between traits within an organism, which means that the benefit obtained through one physiological process entails a cost paid in the value for another process. Genetic mechanisms (or microevolutionary trade-offs (Stearns, 1992)) are based on antagonistic pleiotrophy among genes encoding different traits and/or gene linkage (Rose et al., 1987; Stearns, 1989; Armbruster & Schwaegerle, 1996). Numerous examples for trade-offs are reported in the literature, but most often these lack any clear model for the mechanism(s) of the processes causing them. The lack of a mechanistic basis today is a plausible argument for plant breeders to reject the consideration of any trade-offs that are of potential interest for them. For example, Chapin et al. (1993) expect stress resistant plants to develop a suite of traits that aid in conserving nutrients and simultaneously result in slow growth, but the lack of an analysis of the mechanisms involved makes their findings rather useless for plant breeders. An additional complication is that the mechanisms causing trade-offs (including those of interest for plant breeders) often might involve complex interactions between the physiological and genetic constitution of organisms. This is illustrated in a study on different birch ecotypes by Weih & Karlsson (1999), where we attempted to analyse the underlying mechanisms for a hypothesised trade-off between the maintenance of relatively fast growth under low-temperature conditions and an ability to utilise periods of high temperature for rapid growth. Thus, Weih & Karlsson (1999, 2001) suggested a physiological link between relatively high growth rate at low temperature, low growth rate response to temperature and high leaf nitrogen concentration, which is genetically determined. This sort of research on interactions between the environment, physiology and genotype of plants certainly should gain added value, especially for breeders, if it was complemented with a functional analysis of the involved genes (see below). The physiological and genetic structures causing trade-offs have been incorporated into individual organisms through a long array of adaptations and selections during evolution, in response to an increasing exploitation of new environments and permanently changing environmental conditions. The compilation of growth studies on higher plants suggests that there was probably an increase in internal growth rate from phylogenetically old to younger taxa (Cornelissen et al., 1998; Shinamo et al., 2001). In parallel, adaptations to extreme environments seem to have evolved relatively recently (viz., many Angiosperms) and generally were associated with decreased growth rate, such as in the majority of arctic-alpine plant species (Willis & McElwain, 2002). In a peculiar contrast to this general pattern many fast-growing annuals of phylogenetically younger taxa that can survive in extreme environments mostly by the avoidance of the most stressful conditions, rather than the possession of other typical stress resistance traits (e.g. many desert herbs) appear. Thus, the macro-evolutionary pattern of a trade-off between high growth rate and stress resistance emerges, although the mechanistic explanation that urgently is needed by breeders is lacking. Because trade-offs played a major role in shaping plants in the past, they need to be considered increasingly in future plant breeding. Nevertheless, there are important differences between natural evolution and breeding processes: evolution is driven by random processes (e.g. mutations) in combination with survival under the prevailing environmental conditions, that is, natural evolution lacks any ‘ideas’ of desirable organisms. By contrast, breeders using modern biotechnology tools will be able to construct desirable genotypes by excluding many of the random processes involved in natural evolution. In genetic engineering of plants, a foreign piece of DNA is incorporated into the genome of a plant cell, which is then regenerated into a mature transgenic plant. Plant breeders commonly invoke correlations of certain traits (i.e. the existence of a trade-off in its simplest definition) as criteria, whether or not it is possible to breed or genetically engineer a desirable plant. For example, Pilon-Smits & Pilon (2002) state that metal accumulation, tolerance, and plant productivity are not necessarily correlated and that it therefore should be possible to genetically engineer a plant with high metal tolerance and metal accumulation as well as high productivity. However, the noncorrelation of certain traits at the phenotypic level obviously cannot be taken as proof for the absence of a link between the respective traits, because noncorrelation of traits can be caused by various reasons. Instead of relying on simple (non) correlations, we need to establish a network of links between genes, their functions and the phenotype, and thereby search for evidence for genetic and/or physiological links between traits. The separation of physiological links appears crucial for the successful engineering of desirable plants, because genetic links between traits could possibly be eliminated by modern biotechnology, while physiological links may not. Thus, major physiological trade-offs in plants are much more likely to limit the engineering of desirable plants compared to genetic trade-offs. Several successful approaches have indeed been made recently to investigate the genetic basis of single metabolic processes by the study of transgenic plants in which one or a few proteins were drastically reduced or deleted (Stitt & Sonnewald, 1995; Stitt et al., 2002). However, the traditional mutant research is only able to ‘vertically’ study a very limited number of genes, proteins, etc. at a time. Complex trade-offs require the integration of many different processes related to, for example, carbon and nitrogen metabolism, water use and stress tolerance. The new tools offered by functional genomics methodology (Colebatch et al., 2002) enable also the ‘horizontal’ analysis of hundreds of genes, transcripts, proteins and metabolites at a time, and therefore have a great potential to complement the traditional methods with increased power for the investigation of even complex trade-offs. For example, Seki et al. (2001) studied the expression of 1300 genes of Arabidopsis plants grown under different drought and cold treatments and found 44 drought-inducible genes that were not expressed in the control treatment. Unfortunately, Seki et al. (2001) did not investigate any traits that reflect the growth and physiology of the plants. Nevertheless, the more or less simultaneous analysis of 40–50 genes should easily be possible with the methods of functional genomics up to the levels of protein and metabolite pattern. In combination with research on important growth and physiology traits of the plants grown in the different treatments, the functional genomics approach should have the potential to move us a great deal towards a separation of physiological from genetic mechanisms behind complex trade-offs. In addition, when compared to traditional research, such an approach should be less biased by experimenter-specific hypotheses that focus on selected physiological processes (Rounsley & Briggs, 1999). As the methods of functional genomics are developing very rapidly, evolutionary biologists and plant physiologists will soon receive greatly improved possibilities to test their ideas regarding anticipated trade-offs. However, in order to create fruitful co-operation with plant breeders, researchers working in the different disciplines need to agree upon suitable study organisms (apart from the model plant Arabidopsis). Particularly plant breeders should be most interested in the possible limitations for breeding brought about by the trade-offs. Due to the (likely) existence of these limitations, the goal for breeding programmes should not be the construction of few superior genotypes in the first place, but the development towards a wide selection of well-characterised and specialised genotypes that are adapted to a range of different environmental conditions (Weih & Nordh, 2002). Ideally, this work should be done in close co-operation between molecular and evolutionary biologists, plant physiologists and breeders.