As with other biotrophic pathogens and pests, rust fungi are presumed to have co-evolved with their host plants (Thompson & Burdon, 1992; Giraud et al., 2008). The following forces are at play: the parasite makes use of the host plant's resources, causing a fitness cost to the plant; the host plant then advances a defence system against the pathogen; irrespective of the nature of this defence (either constitutive or induced), there is a cost to the plant, and a balance between the fitness costs of parasitism and defence ensues (Barrett et al., 2009). Furthermore, pathogenicity, as well as providing access to resistant hosts, can carry costs to pathogens, superfluous virulence being detrimental to their fitness (Laine & Barrès, 2013). Heteroecious rust fungi exemplify plant parasites specialized in sympatric ecological relationships with two unrelated host plant species and normally exerting pressures on both, proportional to the negative impact of each stage; similar parasitisms are known from animal pathology (Otranto & Traversa, 2003; Arora, 2012). This co-evolution is on-going and new rust–host relationships arise frequently (Qi et al., 2011), especially where major gene host plant resistance has been identified, exploited and introduced at large scales in agricultural crops. Flor (1942) famously described the gene-for-gene relationships of Linum (flax) and its rust Melampsora lini. Nevertheless, host specificity still represents a barrier to rust range expansion, and background levels of plant diversity normally contain rust infection (Roscher et al., 2007). Pronounced host specificity is therefore a limiting factor, especially considering the dependence of heteroecious rusts on the timely occurrence of three critical elements: (1) the presence of two unrelated and susceptible host species within proximity; (2) the successful mechanisms for spore dispersal between the host plants at the appropriate time in the respective life cycles of both rust and hosts; and (3) the prevalence of favourable environmental conditions during both infection periods. Mechanisms of co-evolution and host specificity are a key area in the response of rusts to global change, and have strong relevance to biological homogenization of floras and the erosion of biodiversity. As Newcombe & Dugan (2010) report, the problem with plant (and pathogen) introductions is not a new one. From the beginning of agriculture c. 10 000 yr ago, plants have been moved into new habitats, selected for beneficial traits and thus restricted in their natural diversity. The range and movement of plants was initially local and along well-known trade routes (Farrington & Urry, 1985). It has increased strongly since the first global circumnavigations in the 16th Century and the rate and speed of plant movement have again increased dramatically in the last century with modern transport developments (Stukenbrock & McDonald, 2008). The internet and global free trade have brought another step change to the movement of plants and plant products during the past two decades (Meissner et al., 2009). Within the context of floristic change, rusts with narrow host ranges are clearly less prone to be invasive compared with those that exhibit a wide range of hosts (Carnegie & Lidbetter, 2012). However, this is not the case where the hosts are common, susceptible and of low genetic diversity, as in many crop plant species. For instance, the coffee rust Hemileia vastatrix is of relatively low importance in its natural range in Ethiopia. This is probably because of the low rainfall in this part of Africa (McCook, 2006) in addition to established natural resistance in the host. When faced with a densely growing cultivated host of low genetic diversity in a moist and warm climate, the rust spreads explosively, however, and has rendered coffee cultivation uneconomical in several regions of the world (Ridley, 2011). The reproductive success of this rust fungus has in turn led to nonspecific hyperparasitism by another fungus, the dual entomo- and mycopathogenic species Lecanicillium lecanii, aided by invertebrate vectors (Vandermeer et al., 2009). There are indications that the evolution of new pathogenicity traits in a number of agricultural pathogens has accelerated in the recent past (Stukenbrock & McDonald, 2008). Where the fungus exhibits a wide host range, the possibility of invasion of new territories and host species can lead to serious concerns. Puccinia psidii is a native rust of guava (Psidium guajava) in South America, causing only minor damage to the original native host (Coutinho et al., 1998). Introductions of Australian Eucalyptus trees into its native range have provided the pathogen with the opportunity to spread to other members of the Myrtaceae, where it now poses a considerable risk to many host plant species in their native ranges, for which susceptibility has been established in trials (Glen et al., 2007). Puccinia psidii has since been accidentally introduced to Australia and shown to be of major plant health concern, with 107 new host species in Australia (Carnegie & Lidbetter, 2012).
As rust fungi and similar pathogens have short generation times spanning weeks or months, combined with vast numbers of propagules (Helfer, 1986; Newton et al., 1999; Pringle & Taylor, 2002), their adaptive and evolutionary potential is relatively great, when faced with rapid environmental change. Many host plants, in contrast, have generation times of several months (herbaceous plants) to many years (trees) and reproduce far less numerously, giving them fewer opportunities for adaptation or evolution and resulting in loss of overall fitness. Fast anthropogenic dispersals of rusts and similar pathogens to new hosts and territories, combined with rapid climate change, are therefore expected to lead to repeated disease outbreaks and possibly host extinctions, as shown in the global amphibian chytridiomycosis crisis (Kilpatrick et al., 2010).