HERBIVORY MEDIATES GRASS–ENDOPHYTE RELATIONSHIPS: COMMENT

Abstract

In the recent paper by Koh and Hik (2007), the pattern of Neotyphodium spp. endophyte infection frequency is shown to be related with the gradient of grazing pressure by two native herbivores, pikas (Ochotona collaris) and hoary marmots (Marmota caligata) in Yukon Territory (Canada). The grazing gradient is determined by the behavior of herbivores, being high ‘‘on’’ boulder fields and low ‘‘far’’ from there. The mechanism invoked to explain the pattern of infection frequency is the relative fitness of endophyteinfected and uninfected plants along this gradient. According to the defensive mutualism and optimal defense theories (Clay 1993, Zangerl and Rutledge 1996), infected hosts with alkaloid-based deterrents would be favored where grazing was consistently high, while uninfected plants or infected plants without deterrents would be favored where herbivory was low or absent. To reinforce the hypothesis, a trial was carried out to evaluate pikas selectivity for endophyteinfected and uninfected plants sampled from ‘‘on þ near’’ and ‘‘far’’ from boulder fields. Herbivores were able to discriminate endophyte-infected and uninfected Festuca altaica plants from boulder fields, but they did not discriminate when plants came from meadows. In summary, the authors conclude that grazing pressure and herbivory selectivity is the major factor generating the pattern in endophyte infection frequency. Hence, endophyte infection is viewed as an adaptive trait in the environment with high grazing pressure. In our opinion, Koh and Hik (2007) present little information supporting that differences in endophyte infection frequency and herbivore avoidance of infected grasses were mainly a consequence of the distinct longer term grazing history of the grasses mediated by the deterrence fungus effect. We propose that the effect of herbivores on infection frequency might as well be mediated through selection of phenotypic traits not related to antiherbivory (e.g., compensatory growth ability [Karban and Baldwin 1997, Sullivan et al. 2007]), or through an interaction with other environmental factors that can affect both partners to a different extent (Hill et al. 2005, Rasmussen et al. 2007). A usual observational approach to study broad ecological patterns is to find out simple relationships between the distribution and abundance of organisms and any environmental factor. However, this procedure may result in oversimplifications leading to wrong conclusions (see the case of shared enemies discussed by Holt and Lawton 1994). For the specific case of Neotyphodium endophytes, ecological mechanisms underlying patterns of endophyte infection frequency are quite complex (Faeth 2002). In fact, Koh and Hik (2007) acknowledge that ‘‘variations at the population level in the deterrence ability of infected tillers, frequency of infection, and hyphal densities are thought to be produced by multiple factors (genetic and environmental)’’ among which they cite ‘‘differences in grazing pressure.’’ Different theoretical and modeling approaches have indicated that the endophyte infection frequency in local populations may result from some of the following mechanisms: (1) relative fitness of endophyte-infected and uninfected plants, (2) vertical transmission efficiency of endophyte in infected plants, and (3) migration of infected and uninfected individuals among connected local populations (Clay 1993, Ravel et al. 1997, Saikkonen et al. 2002, Gundel et al. 2008). Here, we discuss each of those mechanisms and confront them with the arguments presented by Koh and Hik (2007). Our aim is to show that relative fitness may or may not be related to herbivory deterrence and that the other two mechanisms have been almost completely neglected. 1. The relative fitness of endophyte-infected and uninfected related plants has been always used to explain natural variations in endophyte infection frequency in grass populations (Clay 1993, Clay and Schardl 2002).— By this mechanism, it is predicted that the endophyte infection frequency tends to be high if endophyte enhances the host plant fitness or, alternatively, the endophyte goes to extinction if it depresses the fitness of the plants (Clay 1993). An ecological trait can be considered adaptive if it has a direct impact on fitness in natural environments. In the case of the endophyte infection, to consider it as adaptive based solely on the observed correlation between infection frequency and grazing level may be problematic. Even when the correlation is observed, it is difficult to determine whether the ecological trait (endophyte infection) per Manuscript received 14 December 2007 ; revised 11 March 2008; accepted 18 April 2008. Corresponding Editor: J. N. Klironomos. 1 IFEVA-Facultad de Agronomia UBA/CONICET, Catedra de Ecologia, Av. San Martin 4453 (C1417DSE) Buenos Aires, Argentina. 2 E-mail: gundel@agro.uba.ar