Portal de Periódicos da CAPES

Comment on D. M. Wilkinson (1997). ‘Plant colonization: are wind dispersed seeds really dispersed by birds at larger spatial and temporal scales?’

1999; Wiley; Volume: 26; Issue: 2 Linguagem: Inglês

10.1046/j.1365-2699.1999.00277.x

1365-2699

Frank M. Chambers,

Botany and Plant Ecology Studies

Wilkinson (1997: 61) proposes that ‘. . . many plant species often thought of as wind dispersed may in fact be largely dispersed by animals, mostly birds, at larger spatial and temporal scales’. His evidence and analysis largely concentrate on the late-Quaternary (post-pleniglacial) ‘migration’ (sensuHuntley, 1993; cf. ‘spread’: Bennett, 1986a) of temporate trees in Europe, with some reference also to North America. In these contexts, and perhaps also in others (cf. Pole, 1994; Macphail, 1997), the hypothesis bears closer examination. Whilst not dissenting from Wilkinson's general hypothesis – that of the ‘possibility’ that birds may be involved in long-distance dispersal of tree propagules – I would contend that there are other aspects that need to be considered more fully when evaluating the strength of the case. I confine comments here to three aspects: (i) the nature of the evidence for ‘rate’ of migration; (ii) the effectiveness of particular dispersal mechanisms; (iii) vectors of spread and agencies of establishment. These considerations apply generally; but they will also vary markedly in significance between taxa. First, the data tabulated by Wilkinson (1997) are confined to calculated rates of tree ‘migration’ in Europe for the time period between the Weichselian Late-glacial and early (-mid) Holocene. Wilkinson compares the mean maximum dispersal rates between those taxa normally considered to be wind-dispersed and those known to be animal dispersed: his claim that there is little difference between them is based on means of 750±288 m yr−1 and 766±642 m yr−1. The very large standard deviations demonstrate that it is not the coincidence of the mean values that is significant here but (a) the selection of only a small number of taxa for the two groups (comprising just four and three taxa respectively!), and (b) the variance in the range of values. The coincidence of the means is mere accident: had Alnus been included, the means would be further apart (it was probably rightly excluded, owing to ‘its water dispersal strategy’ (Wilkinson, 1997: 62; but see Chambers & Elliott, 1989)); whereas had Corylus been excluded (on the grounds of incomplete separation from Myrica), the mean for animal-dispersed taxa would have been only 400 m yr−1. The exclusion from the analysis of Pinus– a major wind-dispersed taxon with estimated migration rates of at least 1500 m yr−1 for the late-glacial (Huntley & Birks, 1983) – should admit a further note of caution. The migration-rate data are principally derived from pollen records, and so are dependant upon the reliability of both the pollen data and the inferences drawn therefrom. It is well known amongst palynologists that pollen spectra can have a component that is long-distance transported, which, if not recognized as being extraneous, might give a false impression of the regional or local presence of particular taxa. To be sure of a taxon's presence, corroboration is required from other, less easily transported (or less abundantly produced) plant micro- or macrofossils. Whereas some tree macroÍfossils can provide unequivocal evidence for local presence – such as well-dated sub-fossil stumps found rooted in situ, or large macrofossils for which long-distance (e.g. aeolian, fluvial or glacial) transport can be discounted – in contrast, percentage pollen data, of themselves, may not guarantee local presence. In eastern North America, where there has been greater use of absolute pollen frequency measures to ascertain presence and to reconstruct rates of spread, figures suggest range extension of some 100–400 m yr−1 for the major forest trees (Davis, 1976, 1981). In Europe, relative (percentage) pollen data have principally been used for investigating patterns of tree spreading (Birks, 1989) and calculating migration rates (Huntley, 1988), giving rise to the figures quoted by Wilkinson (1997). In the absence of corroborative evidence it has been normal for palynologists in Europe to infer the regional or local presence of a taxon from some notional critical pollen value: examples are the ‘empirical’ and ‘rational’ limits of Smith & Pilcher (1973). However, as Bennett (1986a) has pointed out, the empirical limit (defined as the first consistent trace of pollen of a taxon) will be dependant on the pollen sum (the total number of pollen grains counted in each spectrum), and so is inherently as inconsistent measure; whereas the ‘rational limit’ (the rise to higher percentage values) is still a relative, rather than an absolute, measure of pollen frequency. It has nevertheless become part of the folklore of northwest European Quaternary palynology that some critical threshold value needs to be crossed for the pollen data to indicate local presence, and that this value will vary between taxa. For example, the threshold value originally adopted for Pinus in Europe by Huntley & Birks (1983) was 25% of total land pollen (TLP), and in the British Isles by Bennett (1984) it was 20%. Lageard (1992) subsequently showed the Pinus threshold values were much too high. Fossit (1994) later reported pine pollen at values of 5% TLP in horizons containing pine stomata; whereas, on their own, such low pollen values would not necessarily have been taken to indicate local presence, the stomata were taken as conclusive evidence of the local presence of pine trees (see also Bennett, 1995; cf. Hall & Pilcher & McCormac, 1996). (Contrarily, absence of pine stomata in any horizon does not mean the absence of pine trees!) It therefore remains part of the art of the palynologist to ascertain which pollen are long-distance transported, and which produced from regional or local sources, and so to infer which taxa have a high probability of regional or local presence. For individual pollen diagrams this becomes part science and part intuition; but the use for a taxon of critical (i.e. identical) threshold values for all sites, as has been practised for pollen databases and vegetational reconstructions on a continental scale (e.g. Huntley & Birks, 1983; Huntley, 1988, 1990), inevitably leads to simplification and gives rise to maximum migration-rate figures that have to be wrong to some degree: it is just that we do not known how far wrong or how close to being right they are. Attempting to isolate (and, often, so to discount) the long-distance transported component of pollen spectra is only one of the tasks for palynologists. Modern pollen analogues show that not only can there be pollen evidence when a tree or shrub taxon is demonstrably not present (such as on islands that currently have no trees, yet have tree pollen values of several percent), but also it is possible for pollen spectra not to contain any records of a taxon, yet that taxon can be growing very successfully in the locality, albeit at low densities. Particular species, especially those that might be characteristics of ‘prehistoric wildwood’ (Rackham, 1976) albeit low pollen producers, could be present in a region for several hundred years, yet remain either undetected by pollen analysis, or detected at such low levels that for a palynologist their status would be at best speculative and ultimately unknowable. In consequence, quoted migration rates are at best approximations; and for some taxa they could be misleading. A case in point in Europe is Acer, one of the taxa listed by Wilkinson, but which within its natural range in northwest Europe is nevertheless often undetected at sites by pollen analysis (Huntley & Birks, 1983; see also Chambers, 1995). For shrubs, Bennett (1986b) cites Rhamnus as a ‘palynologically silent’ taxon; modern pollen data show that Rhododendron too can remain undetected, yet be widespread in a region (Chambers, 1995). If it is accepted that in temperate stages of the late Quaternary the status of taxa such as these is often uncertain from palynology, then how certain can one really be of detecting even the prolific pollen dispersers in their early period of ‘presence but minimum abundance’ in the landscape? Indeed, Bennett (1986a, 1988) has cautioned that it is not likely that pollen data would record the first presence of a taxon in a region – even a taxon that produces abundant, well-dispersed pollen – because the first presence may be at such low densities that its pollen would be swamped by that of others. He questions ‘how it is possible palynologically, to distinguish invasion by a population of a species new to the area, and [indigenous] population increase from levels too low to have been detected’ (Bennett, 1997: 115). Part of the argument (in Wilkinson, 1997) for post-glacial dispersal of tree propagules by birds, rather than dispersal by wind or water, rests on the rapidity with which migrations are claimed to occur; but the question remains as to whether the rapidity is more apparent than real. If taxa remain undetected at low densities, then not only can we not tell when a taxon has first reached a region, but we might not know where it is coming from or how far it has to travel: we might not know the real proximity of the starting points (location of refugia); we also might not be able to detect the dispersal routes. We can only really infer that, between some time, presumably after the pleniglacial, and some unknown point in time later, it would seem from isopoll maps (for example, in Huntley & Birks, 1983; Huntley & Webb, 1989) – and from proxy-climate data that seem to require peripheral southern refugia for temporate trees – that very long distances have been traversed by taxa. In truth, despite attempts at interpreting the data (see Bennett, Tzedakis & Willis, 1991), we know not where from, nor when; and so neither how far nor at what rate. Second, in considering the effectiveness of particular dispersal mechanisms, modern analogues show that amongst herbaceous species, seed dispersal mechanisms associated with animals (endo- and exoÍzoochory) do facilitate naturalization and rapid spread of alien species (Malo & Suarez, 1997). However, Wilkinson (1997: 62) steps outside the conventional classification of mechanisms for dispersal and argues instead that tree species not obviously pre-adapted for animal dispersal nevertheless achieve rapid long-distance migration principally ‘due to the acitons of birds’. This hypothesis of [post-pleniglacial] dispersal by birds is not new: indeed there was a previous claim made in this journal for just such a mechanism for Alnus. Dispersal by birds was invoked by Chambers & Elliott (1989) to explain what appear to be markedly disjunct patterns in the early-Holocene appearance of Alnus in the British Isles, because the conventional mechanisms of dispersal (local, by wind; long-distance downstream by water) seemed incapable of explaining both long-distance inter-catchment and lowland-to-upland dispersal. Wilkinson too argues the case persuasively that propagules can be transported long distances by birds, although he envisages seed cacheing or transport in their crops, rather than the accidental mechanisms we implied earlier for Alnus. Third, even if a tree species is capable of being dispersed long distances by birds, there are considerable obstacles to be overcome before that species can establish a founder population in its new environment. In this context I have found it helpful to distinguish between ‘vectors in spread’– those that effect the long-distance transfer of propagules – and ‘agencies of establishment’– those that affect or alter local substrate and light conditions, making them suitable for the taxon's germination and growth (Chambers & Elliott, 1989; Chambers, 1994). (Perhaps we might recognize a minor role for ‘freak’, but nevertheless relatively frequent, violent weather events: for example, mini-tornadoes could (would?) have been more powerful and have covered longer distances in northwest Europe in the early Holocene than today, and theoretically they could be both vector and agent: both small and large seeds may be carried, and clearings created, by hurricane-force winds.) Whichever vector of dispersal is involved, getting the propagules to the new environment is less than half the battle: even in open environments, where there might appear to be little overt competition, establishment can be problematical. Attention should be given both to agencies of establishment (climatic change; and particularly disturbance factors, sensuGrime, 1979), and to ecological tolerances in the pioneer phase (sensuGrime, 1979) of propagules of particular taxa (Sauer, 1988). It is of just as much interest as to why a particular taxon spread rapidly, as to why others, with not-so-dissimilar climatic requirements, apparently (so far as we can tell) did not. The case for long-distance bird dispersal is intriguing and attractive; but the vectors of dispersal are only part of the story: other aspects need to be considered before we can arrive at a full interpretation and explanation of the migration-rate data. The persuasive case for bird dispersal of anemochorous tree species in the late-Quaternary ought not to rest on calculated migration rates because these data must themselves be considered suspect.