In North America and East Asia, the autumn is dominated by many tree species with red autumn leaves (e.g. Hoch et al., 2001; Lee et al., 2003). We examined the distribution of each of the 290 tree species with red autumn leaves listed in Archetti (2009a) and found that most of them also grow in North America and East Asia. By contrast, yellow autumn leaves dominate the parallel phenomenon of autumn leaf colouration in Finland and most of temperate Northern Europe (Holopainen & Peltonen, 2002). While for many years these colours were considered a by-product of chlorophyll degradation (chlorophyll ceases to mask the colours) it is clear that this is not the case, at least for red autumn leaves that produce anthocyanins shortly before leaf shedding (Sanger, 1971; Matile, 2000; Hoch et al., 2001; Lee, 2002; Lee & Gould, 2002). The proportion of tree species with red autumn leaves in the flora and landscape of Northern Europe is small; for example, four indigenous tree species (Prunus padus, Prunus spinosa, Sorbus aucuparia and Acer platanoides) reaching their northernmost distribution in Northern Europe (Alanko, 2001; Holopainen & Peltonen, 2002), and only 24 such species in the whole of Europe (according to our examination of the distribution of tree species with red autumn leaves listed in Archetti, 2009a), while in eastern North America and East Asia both the proportion in the landscape and the actual number of tree species with red autumn leaves are much greater; at least 89 species in a subset of the woody flora of North America (e.g. Lee et al., 2003; and examination of the distribution of the 290 tree species with red autumn leaves listed in Archetti, 2009a) and at least 152 species in East Asia.
The question of red-dominated autumn leaves received much research attention in the last decade: we found more than 80 papers discussing it. Archetti (2009a) has already described red autumn colouration in 290 temperate tree species belonging to 70 genera, and stated that the full list is probably even longer. Because the physiological, ecological and evolutionary aspects of the hypotheses were discussed in great detail in those numerous studies we will mention them only briefly. These hypotheses are of three types: physiologically oriented, anti-herbivory oriented, and a combination of the two. Recently the topic of the evolution of red autumn colouration as an anti-herbivory defence has been hotly debated, dividing physiologists and ecologists, who generally ignored each other and proposed only physiological or herbivory-related explanations, respectively. At the same time, ecologists have not agreed upon the various anti-herbivory functions of autumn colouration. While the simultaneous physiological and defensive roles of red autumn colouration were recently recognized by many (e.g. Lev-Yadun & Gould, 2007; Archetti et al., 2009), the difficulties in proving some of the anti-herbivory functions remain.
Here we use the prevalence of red autumn colouration of trees in North America (Lee et al., 2003) and East Asia (our examination of the distribution of the 290 tree species with red autumn leaves listed in Archetti, 2009a) versus the prevalence of yellow autumn leaves in Northern Europe (Holopainen & Peltonen, 2002; J. K. Holopainen & S. Lev-Yadun, field notes), along with known patterns of migration and extinction during the drastic climatic changes in the Tertiary and the Pleistocene (e.g. Milne & Abbott, 2002), as the basis for a new hypothesis. We propose that higher extinction rates of both trees and their insect herbivores in Europe as opposed to North America and East Asia seem to explain the difference, indicating that red autumn leaves are probably a relict Tertiary adaptation of temperate floras to past climates and herbivore faunas.
In Europe, northern refugia for broad-leaf trees during glaciations are known for a small number of deciduous species, none of which has red autumn leaves: Alnus glutinosa (green autumn leaves), Betula pendula (yellow), Fagus sylvatica (yellow), Fraxinus excelsior (green and sometimes yellow), Salix sp. (yellow), Corylus avellana (yellow) and Frangula alnus (yellow) (Bhagwat & Willis, 2008).
Mountain ridge direction and severity of ice-age extinctions
While the examples we give of fossil plant adaptations to extinct faunas are from large vertebrates, there is no reason to assume that the same is not true for plant–insect interactions. The fact that fossil insects are less extensively studied, and that it is harder to find a specific connection between a fossil insect and its host plant, does not rule out the probability that many insect species probably became extinct during the Pleistocene. There are solid experimental and field data concerning the sensitivity of aphid and other herbivorous insect eggs and all their life stages to very low temperatures (Niemelä, 1979; Tenow & Nilssen, 1990; Strathdee et al., 1995; Strathdee & Bale, 1998). This sensitivity must have resulted in extinction of many insect species during the drastic climatic changes of the Pleistocene, leaving defaunated floras in all continents, but especially in Northern Europe, similar to the defaunated cacti in North America (e.g. Janzen, 1986).
Further support for our hypothesis of an ancient Tertiary origin of red autumn colouration stems from the fact that dwarf shrubs with red autumn leaves, rather than trees, dominate the northern territories of Scandinavia. For instance, the deciduous species Arctostaphylos alpina with its circumpolar distribution (Hämet-Ahti et al., 1992) is one of the most common dwarf shrub species in mountainous areas of Lapland, and has bright red autumn leaves. In lowlands and forested areas, the deciduous shrubs Vaccinium myrtillus and Vaccinium uliginosum have reddish or darker brown autumn colouration. Several evergreen dwarf arctic shrub species of Northern Europe or Alaska (Andromeda polifolia, Cassiope tetragona, Diapensia lapponicum, Dryas integrifolia, Empetrum nigrum, Ledum palustre, Oxycoccus microcarpus, Pyrola grandiflora, Rhododendron lapponicum and Vaccinium vitis-idaea; Oberbauer & Starr, 2002) have red winter and spring leaves. The autumn landscape in the treeless far northern parts of both Scandinavia and Alaska is conspicuously dominated by red leaf colouration expressed only by very low shrubs.
There is a critical difference in the sensitivity of trees and shrubs to extinction when drastic climatic changes such as glaciation occur (e.g. Milne & Abbott, 2002). Trees are much larger and have a much longer life span and generation time than shrubs, which makes trees at an individual and evolutionary level less flexible and more susceptible to extinction. By contrast, the boreal shrubs we discuss not only have much smaller size and a shorter generation time, but usually have berries so that their seeds can be dispersed over large distances by animals. Shrubs can also manage much better than trees in colder and less productive habitats because their low stature allows them to enjoy an isolating snow cover in winter. Shrubs thus could find more refugia than trees in periods of glaciation. All these differences allowed shrubs with red autumn leaves to escape extinction where trees could not survive. Moreover, if red autumn leaves are at least partially an anti-herbivory adaptation (e.g. Archetti, 2000; Hamilton & Brown, 2001; Archetti & Brown, 2004; Lev-Yadun et al., 2004; Manetas, 2006; Lev-Yadun & Gould, 2007; Archetti et al., 2009), the persistence of shrubs during periods of glaciation in the Pleistocene also allowed their herbivores to find refuge from extinction. The refuge of their specific herbivores is related to the ability not only to feed, but also to be sheltered from extreme low temperatures under the snow cover, like their hosts, continuing their role in selection for red autumn leaves.
The proposed historical origin (in geological era perspective) of red autumn colouration
A broad phylogenetic analysis of the origin of red autumn colouration in 2368 tree species indicated that this character evolved independently in temperate trees at least 25 times (Archetti, 2009a). There are several possible periods when red autumn leaves of deciduous trees could have evolved. First, it could be an ancient, Tertiary adaptation that was selected for during the periods of global cooling that began in the mid-Tertiary. A second period that could have strongly selected for such an adaptation is the Pleistocene, which started approximately 2.6 MY ago and exhibited dramatic and repeated climatic changes. Finally, it may be a recent Holocene (the end of the Pleistocene, which occurred some 11 000 years ago) adaptation. A combination of some of these is also possible. The question is which of these scenarios is most likely. The repeated evolution of the red autumn leaf colour in many tree taxa that have a long generation time and therefore slow evolution is a good indication of an ancient origin. We believe that the conspicuous differences in the distribution of red autumn colouration in eastern North America and East Asia (where many taxa have red autumn leaves) and Northern Europe (which is poor in red autumn colouration) are the key to solving this puzzle. If adaptations for low temperatures per se were the selective agent for red leaf colouration, we would expect that the Scandinavian autumn would have been as red as the North American or East Asian autumn, but it is yellow. Alternatively, yellow autumn leaves would have dominated the autumn landscape of all continents. So, while we agree that anthocyanins provide several physiological solutions under low temperatures, as proposed previously (e.g. Matile, 2000; Hoch et al., 2001, 2003; Lee, 2002; Lee & Gould, 2002; Wilkinson et al., 2002; Schaberg et al., 2003, 2008; Ougham et al., 2005; Lev-Yadun & Gould, 2007; Archetti et al., 2009), there is clearly no inherited physiological problem in functioning successfully with yellow autumn leaves under similar low autumn temperatures, as seen in Betula sp., Populus sp. and Salix sp. and the majority of deciduous temperate tree taxa (e.g. Archetti, 2009a). The possibility that the trees with yellow autumn leaves cannot produce anthocyanins should be dismissed because many temperate taxa with yellow autumn leaves have red pigmentation in various parts of their canopy, for example during spells of cold weather during leaf flush at the beginning of the growing season, or in their reproductive organs.
We propose that the solution to the problem of the origin of red autumn leaves in general, and their limited distribution in Northern Europe in particular, reflects the well-known difference in the extinction histories of trees in eastern North America and East Asia (lower extinction rate) and those in Northern Europe (higher extinction rate). If red autumn leaf colouration is the result of various ancient Tertiary adaptations of temperate floras of tropical and subtropical origins, then, while physiological adaptations are habitat and climate dependent and a geographical shift is enough to allow trees to grow when the climate changes, anti-herbivory adaptations may reflect not only the current herbivore fauna and their predators and parasites, but also many extinct animal species (both vertebrates and insects). As discussed above, anachronistic adaptations of plants to past faunas are well known. Herrera (1985), when discussing the lack of fine-scale adjustments in the coevolution of woody plants and animal seed dispersers, proposed that slower species turnover of woody plants over geological time compared with vertebrate dispersers may have favoured a sort of very diffuse coevolution. We propose a parallel scenario for red autumn leaves and herbivores (vertebrates and insects). Not only did trees with red leaves mainly become extinct in Europe, but when many of their herbivores became extinct, the driving selective agents for red autumn colouration also declined.
Thus, the anti-herbivory component of the character of red autumn leaves partly reflects anachronistic adaptations to past faunas, many of which became extinct. This will significantly reduce the chances of obtaining good theoretical and experimental evidence supporting anti-herbivory hypotheses using the current biota. Our hypothesis may calm the hot debate concerning autumn leaf colouration and coevolutionary hypotheses.
We thank Marco Archetti and two anonymous reviewers for their constructive comments.