Factors controlling the post-glacial patterns of beech spread
The logistic increase in the post-glacial distribution of Fagus adds new insights to our understanding of the primary factors controlling the spread of beech in Europe, a matter that has long been the subject of debate among palaeoecologists. Three main explanations are generally considered: (1) the spread of beech is related to changes in climatic conditions, which became cooler and wetter in the second half of the Holocene (Huntley et al., 1989); (2) Fagus expansion is mainly a response to farming and other human activities, which may have contributed significantly to openings in the natural forest vegetation established at the beginning of the post-glacial, favouring colonization by beech (Andersen, 1984; Aaby, 1986; Latałowa, 1992; Reille & Andrieu, 1994; Björkman, 1997; Küster, 1997); or (3) the expansion of beech is an entirely natural phenomenon the timing of which, often coincident with anthropogenic interference, would be the result of slow migration and establishment rates, typical of the internal dynamics of beech forests (Gardner & Willis, 1999).
Giesecke et al. (2007), comparing modelled patterns of climate parameters 6000 bp with Fagus distribution at a European scale, have shown that climatic factors are the likely major determinants of the potential range of F. sylvatica. However, climatic factors are regionally moderated by competition, disturbance effects and the intrinsically slow growth rate of beech. Tinner & Lotter (2006) suggest that climatic change was the main forcing factor for the Holocene expansion of beech in central Europe. However, in the landscapes of northern–central Europe, it is likely that human activities influenced expansion dynamics, as Fagus populations expanded only after the beginning of the Neolithic. According to Bradshaw (2004), anthropogenic activity catalysed the spread of F. sylvatica through increasing the rates of forest disturbance, but it had a minimal impact on the current genetic structure and diversity of beech populations.
The conformity of the increase in beech distribution to the classical logistic model of population growth (Figs 4–6) strongly supports the hypothesis that, given non-limiting climate conditions for Fagus, a biological process was the main factor shaping the pattern of the post-glacial beech expansion in Europe. However, possible effects of climate fluctuations and human activity in determining the rate of change of the past distribution of F. sylvatica at a continental scale cannot be excluded.
The modern distribution of F. sylvatica is clearly limited by unfavourable climatic conditions at its southern and northern boundaries. It is therefore very likely that unfavourable climate conditions could have regionally limited the distribution of beech also during the late-glacial and post-glacial. In Figs 4 and 5, a very light departure of the number of Fagus sites from the logistic model is observed at 12,400 cal. yr bp (mid-point of the time window 10,000–11,000 yr 14C bp) and 8300 cal. yr bp (mid-point of the time window 7000–8000 yr 14C bp). The hypothesis that these two gentle slow-downs in beech expansion may be correlated with the Younger Dryas and the 8.2-kyr events, respectively, is attractive and highlights the need for more detailed data to reach a conclusive assessment. A data set of high-resolution records would probably produce a curve with many more fluctuations, in relation to decadal- to centennial-scale processes, which are smoothed in Figs 4 and 5. On the other hand, the fact that the coarse procedure used in this paper was able to detect such a significant pattern of logistic increase indicates that this phenomenon is strong enough that secondary patterns do not blur it. Broad-scale interpretations are therefore needed.
The progressive decrease of summer insolation during the past 12 kyr (Berger, 1978) may have been a favouring factor for the expansion of beech, which avoids hot and dry summers. However, the insolation curve is sinusoidal, and as such it does not parallel the exponential increase of the beech sites. Numerically, the two curves refer to different functions, which are not related to each other. Therefore the insolation trend, determining increasingly cooler summer temperatures, may have favoured the expansion of beech by influencing its rate of spread, but it cannot have controlled its exponential pattern.
Exponential patterns are typically induced by multiplicative biological factors. The pattern of spread of beech populations might therefore have been induced by other biological populations, increasing or decreasing exponentially. It may be tempting to hypothesize a positive link between a possible exponential increase in Neolithic human populations and the exponential spread of beech. However, recent estimates of human population fluctuations indicate a complex history from the Mesolithic to the Neolithic in central and northern Europe, with rapid rises and crashes of enormous magnitude (Shennan & Edinborough, 2007). In particular, both Germany and Poland show a remarkable decline in population after 7000–6700 cal. yr bp, lasting until after 5500 cal. yr bp, in the course of the Neolithic. In Poland, a sudden increase in human populations at 5500 cal. yr bp is followed by a decline to very low levels until 4000 yr bp. These fluctuations are by no means perfectly in phase with the fluctuation history of the Danish and German populations. At the same time (7000–4000 cal. yr bp), a steady exponential increase in the number of beech sites is observed, corresponding to the steep part of the population growth (Figs 4 & 5). This discrepancy suggests that the exponential pattern of beech spread is not a response to the increase in human populations.
Competition with other tree taxa, which were established in Europe well before beech, does not appear to have been strongly influential on the increase of Fagus populations. In particular, pine and mixed oak forests were widely spread over most of Europe at the time of the exponential increase of Fagus, but they did not prevent its diffusion, which is still continuing at present, although at a slower rate compared to the early- and middle-Holocene.
If we exclude climate, human activity and competition as determinants of the exponential increase of F. sylvatica in Europe, the multiplicative biological process of population increase of beech remains the most likely explanation for the observed pattern.
It is important to note that this result is not in disagreement with the hypotheses of climate or human influence and competitive influences on F. sylvatica, as discussed by previous authors. Human activity and climate may well have affected the frequencies of Fagus in pollen diagrams, which were very variable through time, without extirpating beech populations from the landscape and therefore without causing a decrease below the 2% level.
Although considerable problems are encountered when reconstructing the spread of a plant population from the pollen record, as the initial phases of spread may have occurred at population densities too low to be detected by pollen analysis (Bennett, 1986), the increase in the number of Fagus sites with time is so impressive as to justify further discussion.
It is well known that a population increase conforms to an exponential function when there is a progressive multiplication of the number of individuals, depending on the dimension of the population as it increases with time. An increase of surface area occupied by a population moving with a closed front would not follow an exponential model, as the increase rate would be proportional to the linear dimension of the front. Instead, a population increase starting from scattered nuclei with a low density may produce an exponential increase, as shown in Fig. 7, where in three successive time windows the number of dots, representing small stands of a taxon, increases exponentially both within and outside the spreading front.
Figure 7. Schematic representation of three successive time windows of a diffuse moving front, formed by scattered nuclei of a tree population growing exponentially both within and outside the front.
Download figure to PowerPoint
This model is in agreement with the distribution of pollen records, showing that after the initial colonization of a region, the density of Fagus sites became increasingly higher during the following millennia (Fig. 2). This pattern can be observed, for example, in the southern Balkan Peninsula and in the outer Alpine chain, where after the initial colonization, from 8000 to 3000 cal. yr bp, the number of Fagus sites increased progressively. This model conforms very well with the ‘diffuse spread’ suggested for Picea abies in Fennoscandia by Giesecke (2005).
The exponential diffuse spread model is also confirmed by the patterns of increase in pollen series with high temporal resolution and precision. For example, in three pollen sites located within a distance of 150 km in Switzerland (Soppensee: Lotter, 1999; Bibersee: Beckmann, 2004) and southern Germany (Schleinsee: Clark et al., 1989), a synchronous exponential increase of Fagus is recorded, starting at 8200 cal. yr bp (Tinner & Lotter, 2006). However, the 2% threshold was reached at the three sites at 7900, 7450 and 7500 cal. yr bp, respectively, indicating that the beech stands had local differences of density. This situation may be exemplified by the difference in density of dots on the left and right sides of the diffuse spread in Fig. 7.
In the past few thousand years, the number of Fagus sites appears to approximate an equilibrium distribution (Figs 4 & 5). It is difficult to establish whether this pattern may depend on changing climate and/or increasing human impact. An alternative, but more likely hypothesis is that beech is progressively approaching its carrying capacity, due to limiting environmental conditions far north in Europe. In the Mediterranean regions, a moderate decline in beech range observed over the past two millennia may also be due to the progressive aridification of climate, documented by pollen analysis at many sites (Follieri et al., 2000; Jalut et al., 2000; Pérez-Obiol & Sadori, 2007).
Assessing the extent of the last glacial refugia of Fagus sylvatica
Based on the assumption that the post-glacial increase of beech sites corresponds to an increase of the surface area where beech was locally present, the logistic model of the past distribution of Fagus populations can be used to estimate the extension of the glacial refugia for beech, by applying the logistic function fitted to the fossil data to the value of the modern surface area covered by beech.
Based on the map of the natural potential vegetation of Europe by Bohn et al. (2000), the modern surface of beech forests is estimated as about 931,575 km2 (UNEP, 2000). Setting this value at the present time in the logistic function fit for the increase of Fagus sites, the surface area covered by beech at the beginning of the Holocene is estimated to have been around 31,000 km2 using the normalized data (Fig. 8). Pollen records, however, indicate that in southern Italy F. sylvatica had already started its population increase at the beginning of the late-glacial (Huntley et al., 1999), and in fact the logistic increase of Fagus sites is clearly observed since the late-glacial (Figs 4–6). At 15,000 cal. yr bp, when Fagus is found with continuity in southern Italy, the surface area covered by F. sylvatica would have been c. 5000 km2. The estimates obtained from the non-normalized Fagus sites are somewhat lower (2000 km2), but in any case the final result is that the refuge area for beech at the end of the last glacial was likely to be of two orders of magnitude less extensive than at present (Fig. 8).
Figure 8. Modelled logistic increase of the surface area occupied by Fagus sylvatica during the past 15,000 years, based on the functions best fitting the increase of Fagus sites [dotted line, y = 962,000/(1 + 0.0285 × e0.0006x), from Fig. 4] and of normalized Fagus sites [solid line, y = 1,010,000/(1 + 0.0797 × e0.0005x), from Fig. 5].
Download figure to PowerPoint
In Fig. 9, a tentative reconstruction of the distribution of F. sylvatica at 15,000 cal. yr bp is advanced, based on the location of glacial refugia deduced from palaeobotanical and genetic data (Fig. 3), and on the quantitative assessment of the surface area occupied by beech (Fig. 8). This reconstruction suggests that the Pleistocene refugia of F. sylvatica were not a limited number of extensive areas with closed beech forests. Instead, they appear as very sparse stands of small populations of beech scattered in multiple regions. Starting in the late-glacial and during the Holocene, the southern European populations increased their density at a very low growth rate, without any important displacements. The populations that survived in the rest of Europe, especially those in the eastern Alps–Slovenia, and possibly Moravia, not only became increasingly dense locally, but expanded far to the north and are still expanding, but at a more moderate rate.
Figure 9. Black dots represent a tentative quantitative reconstruction (5000 km2) of refuge areas for Fagus sylvatica 15,000 cal. yr bp. Grey area corresponds to the modern distribution (c. 931,500 km2).
Download figure to PowerPoint