Evolution of pest-induced defenses in Brassica plants: tests of theory

Ecology, March, 1998 by David H. Siemens, Thomas Mitchell-Olds

INTRODUCTION

Plant resistance may increase as a result of prior feeding by herbivores (Karban and Myers 1989) or infection by microbial pathogens (Karban et al. 1987, Ajlan and Potter 1990). This plastic increase in resistance is termed induction, and often involves elevated levels of certain secondary metabolites (Green and Ryan 1972, Waterman and Mole 1989). Induced resistance may be an evolved defense strategy (Brody and Karban 1992, and references therein); plants that are attacked infrequently in habitats where resources are limited should not maintain high constitutive levels of costly secondary metabolites (Berenbaum et al. 1986, Simms 1992, Han and Lincoln 1994). The combination of resource availability and incidence of attack should therefore dictate the evolution of constitutive or inducible defensive mechanisms, an idea with obvious underpinnings in optimal-defense theory (McKey 1974, Rhoades 1979, Zangerl and Bazzaz 1992). However, there may also be genetic interactions between deployment mechanisms that prevent independent evolution of basal and induced resistance mechanisms. Regulatory genes that affect constitutive levels may also influence induced levels of some secondary metabolites. Thus, studies that examine genetic correlations or trade-offs between these different deployment mechanisms of secondary metabolites are needed (Zangerl and Berenbaum 1990).

Genetic correlations between characters may be evolutionary responses to ecological and physiological factors (Dorn and Mitchell-Olds 1991). For instance, if we predict an inverse phenotypic relationship between induced and constitutive levels of secondary metabolites (e.g., Lewinsohn et al. 1991) because of allocation costs and frequency of attack (Brody and Karban 1992), then we should also expect a negative genetic correlation between these defense deployment mechanisms. Similarly, if induction is ineffective in the presence of high constitutive levels of defenses (Mattson et al. 1988, Herms and Mattson 1992:305), and if higher constitutive levels are selected because of frequent attack, then a negative genetic correlation is again predicted. We should also expect a genetic cost to defense production, a negative genetic correlation between secondary metabolite production and fitness (e.g., seed production) (Simms 1992).

We tested predictions regarding pest-induced defenses in the annual mustard Brassica rapa (syn. campestris). Important secondary metabolism in brassicas includes glucosinolates and their breakdown products (Chew 1988). Glucosinolates are low molecular mass nitrogen- and sulfur-containing compounds that are hydrolyzed by myrosinase(thioglucoside glucohydrolase, EC 3.2.3.1) producing D-glucose, sulfate, isothiocyanates (volatile mustard oils), and other cyanide compounds (Larsen 1981, Poulton and Moller 1993). Glucosinolates and myrosinase occur together throughout the plant, but are separated cellularly or subcellularly, and mix and react when cells are damaged by pests (Chew 1988). Concentrations of glucosinolates vary with environmental factors such as nutrients (e.g., available soil nitrogen), water, light intensity, and infection by pathogens (Wolfson 1982, Louda and Rodman 1983, Gershenzon 1984).

Since it has been suggested that soil nutrient availability may also affect the evolution of plant defense chemistry (Janzen 1974, McKey 1979, Bryant et al. 1983, Coley et al. 1985, Bazzaz et al. 1987, Herms and Mattson 1992, Tuomi 1992), we were also interested in the interaction between soil nutrient availability and genetic changes in defense physiology. We asked if this interaction was important to resistance and allocation costs. Because nitrogen is a main constituent of glucosinolates and proteins, we added nitrogen fertilizer as a treatment in field experiments designed to estimate allocation costs and resistance. The carbon/nutrient balance hypothesis (Bryant et al. 1983) states that under low-nutrient conditions, growth may be reduced to a greater extent than photosynthesis, and carbohydrates accumulate in excess of those needed for growth. Consequently, plants growing in poorer nutrient conditions may produce relatively more carbon-based defensive compounds and lower levels of nitrogen-based compounds. Evidence suggests that nutrient stress does indeed lower levels of some nitrogen-based allelochemicals (Gershenzon 1984) as predicted. The effects of genetic changes in levels of nitrogen-based defensive compounds should therefore be contingent on nitrogen availability. For instance, genetic increases in nitrogen-based secondary metabolites like glucosinolates and myrosinase may be more likely to affect herbivory and costs if nitrogen availability is not limited. Sulfur is another constituent of glucosinolates and its availability in soils may also be important in glucosinolate levels (Wolfson 1982). Moreover, in resource-poor environments, other chemical resistance factors in brassicas (Chew 1988) that are carbon-based, may obscure effects of genetic changes in nitrogen-based secondary metabolites. This implies that costs and resistance effects of genetic changes in secondary metabolites may be contingent on resource availability.