Species pools in cultural landscapes – niche construction, ecological opportunity and niche shifts


O. Eriksson, Dept of Botany, Stockholm Univ., SE-106 91 Stockholm, Sweden. E-mail: ove.eriksson@botan.su.se


This paper discusses the ecology of species that were favoured by the development of the cultural landscape in central and NW Europe beginning in the Neolithic and the Bronze Age, with a focus on mechanisms behind species responses to this landscape transformation. A fraction of species may have maintained their realized niches from the pre- agricultural landscape and utilized similar niches created by the landscape transformation. However, I suggest that many species responded by altering their niche relationships, and a conceptual model is proposed for this response, based on niche construction, ecological opportunity and niche shifts. Human-mediated niche construction, associated with clearing of forests and creation of pastures and fields promoted niche shifts towards open habitats, and species exploited the ecological opportunity provided by these created environments. This process was initially purely ecological, i.e. the new habitats must have been included in the original fundamental niche of the species. Two other features of human-mediated niche construction, increased interconnectivity and increased spatial stability of open habitats, resulted in species accumulating in the habitats of the constructed landscape. As a consequence, selection processes were initiated favouring traits promoting fitness in the constructed landscape. This process implied a feed-back to niche shifts, but now also including evolutionary changes in fundamental niches. I briefly discuss whether this model can be applied also to present-day anthropogenic impact on landscapes. A general conclusion is that ecological and evolutionary changes in species niches should be more explicitly considered in modeling and predictions of species response to present-day landscape and land-use changes.

Almost all parts of the world are affected by anthropogenic impact. Human land-use transforms habitats, species distributions are altered by intentional and non-intentional movement of species, and environmental conditions are altered chemically and physically through eutrophication, pollution and climate change. It is a well established notion that these impacts are negative for countless species, suggesting that we may even be entering a sixth major mass extinction (Barnosky et al. 2011). The scale of these impacts is unprecedented historically, resulting from industrialization, economic and societal development, and human population growth during the last centuries. However, large-scale transformations of ‘natural systems’ (in the meaning ‘not strongly influenced by humans’) are not new historically, and although previous historic changes mediated by human culture may not be strictly comparable to the present impacts, it may be informative to consider how they affected species and species communities. In this paper I discuss one of the greater previous landscape transformations, the vast forest clearing and opening of the landscape that took place along with development of pastoralism and agriculture from the Late Neolithic and the Bronze Age in central and north western Europe. My focus is an issue that has for long been controversial, namely how the pre-agricultural landscape may have harboured the species-pool that later became associated with the open to semi-open cultural landscape. This issue has received attention because some habitats in cultural landscapes, such as meadows, pastures, and open woodlands, are exceptionally species-rich (Emanuelsson 2009). The often used term ‘traditional landscape’ referring to landscapes which include these species-rich habitats, may reflect a culturally biased, romanticized view, of landscapes (Widgren 2012). Therefore, henceforth in this paper, ‘cultural landscape’ means any landscape shaped by humans. Even if avoiding the term traditional landscape, it is relevant to note that some of these cultural landscape elements that are very species-rich have a management continuity rooted in pre-industrial history. Although the time depth of this mana ge ment certainly varies, we know that there are some areas, e.g. in England, southern Scandinavia and central Europe, where management continuity at specific sites goes back several thousand years (Fyfe et al. 2008, Hannon et al. 2008, French 2010, Poschlod and Baumann 2010).

Perhaps paradoxically, viewed from the current concern that human impacts wherever they occur are considered mostly negative for many species, the initial large-scale landscape transformation that shaped the cultural landscape in central and NW Europe obviously created conditions that were favourable for numerous species. Many of these species are presently declining due to abandonment of mowing and grazing management, or small-scale farming. As an example, 68% of red-listed flowering plants (n = 402) in Sweden is associated with the agricultural landscape, and the corresponding figure for red-listed butterflies and moths is 72% (n = 504) (Gärdenfors 2010). Such an association between abandonment of agricultural management and declining plant and insect species has also been found in many other studies (Fischer and Stöcklin 1997, van Swaay 2002, van Swaay et al. 2006, Kivinen et al. 2008, Emanuelsson 2009, Berg et al. 2011, Dover et al. 2011).

My objectives are to try answering two questions: 1) where (ecologically) did the species favoured by the development of the cultural landscape during the Neolithic and Bronze Age occur in the pre-agricultural landscape? 2) What were the mechanisms behind species´ response to the major landscape transformation occurring at that time? My main argument is that species are more flexible with regard to their niche use than is commonly thought, and that a combination of human-mediated niche construction, altering key features of the environment, and niche shifts by responding species, favoured many species. I will conclude with a brief discussion of whether similar processes may apply to species responses in present-day changing landscapes.

The landscape transformation

While agriculture developed gradually in parts of NW Europe possibly as early as during the sixth millennium BCE (Jarman et al. 1982), more dramatic changes with large-scale deforestation and creation of wide-spread open pastures and local areas with crop fields occurred during the late Neolithic and the following Bronze Age, i.e. from the mid 3rd and during the 2nd millennia BCE (Berglund 1991, Kristiansen 1998, Odgaard and Rasmussen 2000, Kristiansen and Larsson 2005, French 2010). Although there are discrepancies in results from different modeling approaches, results from recent land cover models analysed for selected areas (Gaillard et al. 2010) suggest a drastic increase in landscape openness in for example southern Scandinavia during the 3rd millennium BCE, and during the period 1200–700 BCE in the Czech Republic. Several case studies indicate that the 2nd millennium BCE (i.e. during the Bronze Age) was a particularly decisive period in the formation of open grassland systems, e.g. in the French Pyrenees (Bal et al. 2010), southeastern France (Henry et al. 2010), southern UK (Fyfe et al. 2008), southern Germany (Poschlod and Baumann 2010), and in southern Scandinavia (Hannon et al. 2008). During the 2nd millennium BCE there was also a large expansion of transport and communication across Europe, as a result of increased productivity, human population density and trade (Anthony 2007, French 2010). Bronze is an alloy of tin and copper, and bronze production was dependent of sites where these metals could be exploited. While copper has a relatively wide-spread occurrence, tin is found in only few places in Europe, so trading systems based on tin are likely to have stimulated a geographically extensive inte gration of Europe. As stated by Kristiansen (1998, p. 56): ‘The Bronze Age is in certain aspects historically unique. For nearly two millennia it unified Europe within a common framework of interacting exchange networks…. Initially, settlements ranged from mostly small individual farmsteads to more aggregated but still rather small villages (Earle and Kolb 2010), but in some areas the increasing population density also promoted development of larger settlements. Some Tripolye settlements (from the 3rd millennium BCE) in Ukraine were huge, covering > 400 ha, and it is estimated that feeding the population would have necessitated an area of 3300–4600 ha cultivated fields in the vicinity of the settlement (Anthony 2007). In central Europe some fortified settlements from the 8th century BCE are estimated to have harboured a population of around 1000 people (Kristiansen 1998). From the 7th century BCE, along with a cooler climate (van Geel et al. 1996), there are signs of decline in grassland productivity and soil degradation, and stabling of cattle increased in importance (Kristiansen 1998, French 2010). This in turn created an increased need for production of winter fodder, from hay-meadows or by pollarding and coppicing of trees, and organization of field-systems close to farms and small villages (Kristiansen 1998).

While these general landscape changes are supported by lots of evidence, we have a less clear picture of the actual species richness in the created open to semi-open grassland systems. Fyfe et al. (2008) suggested that there was a shift to species-rich grasslands in Dartmoor, UK, around 1480 BCE. Based on pollen data, Berglund et al. (2008) described a steady increase in floristic diversity in southern Scandinavia during the 3rd and 2nd millennia BCE, with a peak around 500 BCE. A positive effect on local species richness associated with the expansion of agricultural management is also indirectly implied by present-day patterns of diversity. Remnants of grazed or mowed grasslands with a long continuity of management today harbour exceptionally high local species richness (Kull and Zobel 1991, Fischer and Stöcklin 1997, Eriksson et al. 2002, 2005, van Swaay 2002), and plant species diversity in rural landscapes has been found strongly positively related to the length and continuity of management (Lindborg and Eriksson 2004, Pärtel et al. 2007a, Cousins 2009a).

There has for long been a general opinion that the pre-agricultural landscape in central and NW Europe was covered by extensive forests (see Vera 2000, for a review of background literature). As a consequence, the high species richness associated with cultural landscapes, initially created by humans from the late Neolithic and onwards, may be seen as intriguing. Where did these species live before the expansion of open and semi-open habitats? Vera (2000) challenged the prevailing view of the pre-agricultural forest landscape, by suggesting that the impact of large wild herbivores maintained forest-grassland mosaics, even before humans had any appreciable influence on vegetation. He suggested that the closest modern analog for this landscape would be the wood pasture, i.e. systems where, most commonly, deciduous trees grow in low density and with a field layer of herbaceous and graminoid species. Several authors have questioned Vera's hypothesis. For example, Van Vuure (2005) argued that large herbivores (he focused on the now extinct auroch) cannot possibly maintain such a vege tation mosaic of open and semi-open habitats. Bradshaw et al. (2003) compared the ‘wood pasture hypothesis’ by Vera (2000) with the ‘high forest hypothesis’ (i.e. a dominance of dense forest in the pre-agricultural landscape), and concluded that a combination of fire and browsing by large herbivores may have maintained some open habitats, but overall the landscape was dominated by closed forest. Mitchell (2005) reached a similar conclusion, and suggested that open canopy forest could only have been maintained by human exploitation. Svenning (2002) also concluded that closed forests were predominating before human intervention, but suggested that conditions allowing for open grasslands occurred on for example floodplains, chalklands and infertile soils. For Scandinavia, open habitats are also likely to have occurred along coasts due to land-upheaval (Cousins et al. 2002).

This brief overview leads to a few conclusions. Most evidence support the hypothesis that forests dominated the landscape in central and NW Europe before the expansion of agriculture. During a period from the late 3rd millennium BCE and about two thousand years onwards a major landscape transformation took place. This landscape transformation consisted of extensive forest clearing creating pastures and fields. Human population density, trade, and communication increased. Permanent settlements expanded and interconnectivity between settlements increased in spatial scale, creating networks of human contacts over large parts of Europe. Although detailed estimates of its extent and magnitude may vary (Gaillard et al. 2010), this landscape change must, by any standards, be considered as a major transformation of the habitat conditions for wild animals and plants. Nevertheless, few species, as far as known, went extinct in association with these landscape changes. Some large herbivores such as the auroch are likely to have declined, although it did not finally go extinct until the 17th century (Van Vuure 2005). In contrast, it seems that the new landscape constructed by humans favoured numerous species, for example among plants and insects.

Natural community analogs and niche shifts

The suggestions by Vera (2000), Svenning (2002) and Bradshaw et al. (2003) are, despite disagreements on the prevalence and structure of forests in the pre-agricultural landscape, similar in one respect. They all imply that we have to look for natural habitat (community) analogs before the human cultural expansion, in order to explain where open habitat species came from. For example, based on evidence from pollen and macrofossils, Svenning (2002) suggested that open habitats were sufficiently common before the Neolithic for the existence of species that later occupied traditional cultural landscapes. The factor in focus of the discussion of the original habitat for the species in the cultural landscape is canopy openness. The analog habitats looked for are thus ‘naturally’ open habitats, that, implicitly, are assumed to have harboured the same set of species that later, after the landscape transformation, instead invaded the open pastures and meadows created by humans. If these species did prevail in small open areas in the pre-agricultural landscape, this implies that there should have been community analogs existing at that time, just as Vera (2000) proposed in his ‘wood pasture hypothesis’. However, despite existence of open habitats in the pre- agricultural landscape, there are reasons to believe that it is unlikely that these habitats harboured communities that were analogs to those developed in pastures and meadows of the transformed landscape.

Most current reasoning on this issue relates to another kind of environmental change, climate change, and concern plants, but the arguments should be applicable also to environmental changes due to human land use, and hold also for animals. Although the idea is by no means new (Gleason 1926), it is now firmly established that plant species respond individualistically to Holocene climate change, that community composition changes continuously, and that communities are no-analogs on time-scales of 103–104 yr (Overpeck et al. 1992, Jackson 2006, Williams et al. 2004). A tendency to overstate stability of present- day communities may have a phychological background, and it may affect scientific analyses. Nogués-Bravo (2009) critically reviewed climate envelope models used for hindcasting past species’ distributions, and concluded that many studies make too simplified assumptions on the stability of species´ relationship to their environment. Most ecological models are at least partially parameterized from modern observations, and may fail to predict ecological responses both to past and future environmental change (Williams and Jackson 2007). ‘Novel ecosystems’ is a recent concept that formalizes the view that communities are constantly shifting, and novel ecosystems are expected to appear in response to presently ongoing environmental change (Williams and Jackson 2007, Hobbs et al. 2009).

Given the insights from the studies mentioned above, it seems unlikely that there were analogs to open pasture and meadow communities in the pre-agricultural landscape. If we accept, as an a priori assumption, that the species pool that came to inhabit the traditional cultural landscape, have not existed together in natural analog communities, species (or at least some of them) must have shifted their relationship to niche factors as a result of the landscape transformation. There has been much discussion about the niche concept; see Chase and Leibold (2003) for an overview, and Araújo and Guisan (2006) for a useful discussion about the application of the niche concept for modeling of geographic ranges. Here I use the conventional distinction (from Hutchinson 1957) between the fundamental niche (the environmental conditions that allow a species to satisfy its minimum requirements for population persistence), and the realized niche (the part of the fundamental niche that is used in a specific situation). The factors constraining the realized niche are extended to incorporate not only competition, as originally suggested by Hutchinson, but all factors that hinder a species to use its whole fundamental niche. The realized niche is thus a subset of the fundamental niche. A niche shift implies either a shift in the position of the realized niche within the fundamental niche, or a shift in both fundamental and realized niches (Fig. 1). While the former may be a purely ecological process, the latter involves also an evolutionary process mediated by selection.

Figure 1.

A schematic illustration of niche shifts. The realized niche of a species is a part of fundamental niche space. Niche shifts result from either (A) realized niche change within an unchanged fun damental niche, or (B) a shift also in the fundamental niche. While (A) may be purely ecological, (B) also includes evolutionary change.

There is some controversy on the question of how common niche shifts are, in contrast to niche conservatism (reviewed by Wiens and Graham 2005). One view is that ecological niche characteristics are strongly conserved (Peterson et al. 1999, Martínez-Meyer et al. 2004, Peterson 2011). Others (Davis and Shaw 2001, Davis et al. 2005) have challenged the view that niches are static over time scales of 102–104 yr, and argued that evolutionary responses to short-term environmental change have been generally overlooked. Holt (2009) suggested that there is a spectrum of rates of change in species’ niches, from rapid niche evolution to profound niche conservatism. Moreover, some aspects of niches may be conserved, while other features, more related to fine-tuning of species requirements, are flexible. Wiens and Graham (2005) suggested that there may actually be no necessary contradiction between niche conservatism and niche shifts, since the manifestation of these processes may act on widely different scales. As an example, Peterson et al. (1999) made comparisons between sister-species (phylogenetic pairs) with regard to coarse climate related data (annual mean temperature, annual mean precipitation, elevation and potential vegetation), and concluded that pairs of species were overall similar, suggesting that their niches have been stable since the evolutionary lines diverged (when this occurred is unknown but may be in the magnitude of 104–106 yr). It is however obvious that if such a comparison would be made for species shifting in some aspects of niche, e.g. resource use, before and after the Neolithic-Bronze Age landscape transformation, but being more or less persistent geographically, this would probably also have been perceived as niche conservatism. The species remain in the same geographical context, and are thus ‘stable’ with regard to annual mean temperature, annual mean precipitation, elevation and potential vegetation.

Although there are still quite few empirical studies, the idea that niches may shift also on a short ecological time scale has gained empirical support (Levin 2003, Pearman et al. 2008). In a study of insect pest species on cultivated cacao at different locations, Strong (1974) found that most species were recruited from the local fauna, indicating that pest species have shifted their realized host plant niche over the last centuries. Chuine (2010) suggested that phenological traits have a strong potential to evolve rapidly in response to climate change. The same may hold as a response to land-use. Early flowering genotypes of meadow plants such as Euphrasia spp., Rhinanthus spp. and Gentianella spp. (Karlsson 1984, Lennartsson 1997, Svensson and Carlsson 2005), thought to have responded evolutionarily to hay-making, are illustrative examples. Rehfeldt et al. (1999) made an experimental study of Pinus contorta over a range of different climates and concluded that density dependent selection produced narrow niches that differed across the geographical range. In an experimental study of the annual plant Brassica rapa, Franks et al. (2007) found evolutionary response to increasing drought and onset of flowering (obtained in just a few generations). Broennimann et al. (2007) found niche shift with regard to climatic factors between native and non-native Centaurea maculosa. Weber and Schmid (1998) found a considerable differentiation related to resource use and phenology in two North American Solidago species after only ca 250 yr since their introduction to Europe. Hybridization is common in certain groups of plants and is a well-known mechanism for rapid evolution in plants (Ellstrand et al. 1996). There is growing evidence that hybridization increases in landscapes under anthropogenic influence, and that it promotes invasiveness (Schierenbeck and Ellstrand 2009, Sloop et al. 2009, Ridley and Ellstrand 2010), indirectly implying that niche relationships have been altered. There are also a few examples for insects. Due to climate warming, the butterfly Aricia agestis has expanded its host plant niche to incorporate new species (possibly included in the former fundamental, but not realized niche), and by doing so it has been able to expand its range in the UK northwards (Thomas et al. 2001). Based partly on this example, Parmesan (2006) suggested that although evolutionary shifts are not generally expected due to climate change, features related to resource use and dispersal may evolve rapidly at expanding range margins. In a particularly informative study, Friberg et al. (2008) analysed niche relationships in two closely related butterflies Leptidea reali and L. sinapis across their geographic range in Europe. They found that the species’ niches are different in different parts of their sympatric distribution, and that both species may appear as habitat specialists or generalists, but in different areas. In Sweden, L. reali is specialized on open meadows whereas L. sinapis is a generalist also inhabiting forests. This is in contrast to its habitat use in other parts of Europe, where L. sinapis is a specialized open meadow species. Although both species under laboratory conditions prefer open meadow legumes as host plant, in the field L. sinapis used 26% forest legumes (compared to L. reali which used 6% forest legumes). Moreover, L. sinapis reached its flight optimum at a lower temperature than L. reali, which was interpreted as a secondary effect of habitat choice.

Some conclusions can be drawn from this brief overview of natural community analogs and niche shifts. Although there is no data presently that can falsify claims that there existed natural analogs to the environments created by human pastoralism and agriculture from the Neolithic and onwards, evidence from other systems suggest that species respond individualistically to environmental change, and that community composition is continually shifting over time spans of a few thousand years. Thus, it seems unlikely that the species pool now associated with the cultural landscape was previously associated with community analogs in the pre-agricultural landscape. This means that many species may have responded to the landscape transformation by shifts in relation to niche factors. Evidence, although still limited, support that such niche shifts, related to habitat choice, resource use, phenology and behaviour do occur. Following these conclusions, the next issue is to ask in what way the human-mediated landscape transformation actually changed niche factors for the wild species.

Human-mediated niche construction

Odling-Smee et al. (2003) examined the ecological and evolutionary implications of species’ effects on their own and other species niches, processes they termed niche construction. Through their activities, species may partly create and partly destroy their own niches. In addition, the activities of species may indirectly create or destroy the niches of other species. This latter aspect of niche construction has been termed ecosystem engineering (Jones et al. 1994), and it is this aspect that is in focus here; the impact of human culture on wild species. While Odling-Smee et al. (2003) maintained ecosystem engineering as a component of niche construction, a concept that also have been used in some theoretical analyses of niche construction (Kylafis and Loreau 2008, 2011), other authors (Erwin 2008) use niche construction strictly when the same species is both causing the effects, and is influenced by them. However, for the present discussion it makes sense using niche construction in the original sense, both because it allows for feed-back interactions between species, and because humans literally ‘constructed’ a new landscape for many wild species.

Which were the key elements of the niche construction associated with the great landscape transformation from the late Neolithic and onwards? The most obvious ecological impact, which also is in focus of previous treatments (Vera 2000, Svenning 2002, Bradshaw et al. 2003, Mitchell 2005) is increased openness, that forests were cleared and large tracts of land were used as pastures and meadows (some of them wooded) and fields, thus primarily altering light and microclimatic conditions. But also soil conditions were probably affected, due to burning, and deposition of manure. However, two other aspects of change are also important. As described above, human interconnectivity (due to trade and other societal interactions) increased. Trade and movement of people and livestock have large potential impact on dispersal, particularly of plants (Auffret 2011). Poschlod and Bonn (1998) highlighted the importance of such interconnections for the realized dispersal of plants associated with the traditional cultural landscape, and suggested that the disappearance of such connections is one major reason for the present decline of many species. Changing environmental conditions may also affect life- history components related to plant dispersal which through niche construction feed-back to cause novel dispersal features (Donohue 2005). Insect species and communities associated with tree species may also be affected by landscape changes that influence the spatial patterns of tree populations, and thereby dispersal processes of insects. A striking example is a study by Tack and Roslin (2010) and Tack et al. (2010) on oak Quercus robur. Both the composition of the insect community, and local adaptations among leaf miners and gallers, were affected by the spatial landscape context of their host trees. Oak populations are likely to have been strongly influenced by the human-mediated landscape transformation, for example affecting trees living in open and semi-open landscapes. Thus, human-mediated niche construction may indirectly influence niche con struction processes involving also other key species, such as large trees.

A third impact was due to the development of an increasing density and size of permanent settlements, and their associated cultivated areas for pastures and fields. Even if we imagine that there were open habitats in the pre-agricultural landscape due to for example herbivory, fire, flooding or other disturbances, these areas are likely to have been small and transient. The creation of an open landscape, where the structure was maintained by human culture, resulted in a spatial stabilization of all the habitats associated with this landscape.

Thus, there are three major elements of the niche construction emanating from the landscape transformation, increasing openness, interconnectivity and spatial stability, and all three are likely to have had an impact on the niche dimensions for wild species. The first of these impacts, the transformation from a mostly forested to a mostly open or semi-open landscape relates to the potential of species to respond by niche shifts, associated with the presence/absence of a tree cover. There is indirect evidence supporting the idea that species considered ‘typical’ for open cultural landscapes may include forested habitats in their fundamental niche. Among the Swedish red-listed species of butterflies and moths associated with agricultural landscapes (n = 364), 34% are also listed as occurring in forest habitats (Gärdenfors 2010). For the species pool of plants occurring in semi-natural grasslands in Sweden, approximately 70% also have occurrences in forest (Eriksson unpubl.). Some of these species have, perhaps mistakenly, been interpreted as forming declining remnant populations in forests (Johansson et al. 2011). Thus it may be more common than previously recognized that both forest habitats and open or semi-open habitats were included in the fundamental niches of many species that were favoured by the landscape transformation.

The two other impacts, increasing interconnectivity and spatial stability of habitats, influence the spatial dyna mics of species. Primarily, these changes are not related to niche shifts, but instead affect the occupancy of species at available sites. Although far from all species exhibit meta population dynamics (Hanski 1999) in a strict sense, one may use a metapopulation model argument to illustrate the outcomes of increasing dispersal and local site persistence. Occupancy, i.e. the fraction of occupied sites among those available for a species to inhabit, is a balance between colonization and local extinction. Increasing connectivity among sites promotes dispersal. Increasing site area has a positive effect on both the abundance of potential source populations, and the probability of finding suitable sites. Holt (2009) suggested that the evolutionary effect on niches when species colonize habitats that are outside the present fundamental niche (‘bad’ or ‘sink’ habitats in Holt's conception) is dependent on the propagule pressure (cf. Simberloff 2009). If propagule pressure is high it is likely to promote niche evolution. We have no possibility to obtain information about the propagule pressure experienced by the created pasture and meadow habitats. Only few studies have estimated aspects of propagule pressure in relation to abundance of source populations, but those that have, suggest a positive relationship (Jakobsson et al. 2006). It thus seems likely that propagule pressure increased as a result of increasing in grassland area and connectivity associated with the landscape transformation. At least for plants, there is evidence suggesting that deterministic local extinctions predominate, i.e. population extinctions that are driven by local habitat change (Lienert et al. 2002, Lindborg and Ehrlén 2002). Spatial stability of habitat conditions due to management thus reduces local extinction rates. Taken together, these effects increase occupancy. This is to say that species will tend to accumulate at the available sites, resulting in high local species richness. This ‘species accumulation process’ has been suggested as a mechanism behind the exceptionally high small-scale richness of plant species in grassland habitats of the traditional cultural landscape (Eriksson et al. 2002, 2005). There are theoretical arguments that species niche characteristics adapt to the conditions where most individuals occur (Holt and Gaines 1992). Thus, a process where species, and thereby individuals, accumulate in open and semi-open grassland sites domi nating the landscapes promote evolutionary niche shifts towards the conditions prevailing at those sites.

A model of species response to the agricultural landscape transformation

The response of species that were favoured by the major landscape transformation initiated during the Neolithic and Bronze Age can be summarized by a conceptual model (Fig. 2). The first feature of human-mediated niche construction, associated with clearing of forests and creation of pastures and fields resulted in two alternative responses. Species that used naturally occurring niche space in the pre-agricultural landscape similar to the niches created by the landscape transformation could expand into the new habitats while maintaining their realized niches. Other species responded by niche shifts towards open habitats. This response was initially purely ecological, i.e. the new habitats must have been included in the original fundamental niche of the species. As a result of the two other features of human-mediated niche construction, increased interconnectivity and increased spatial stability of open habitats, species began to accumulate in the habitats of the constructed landscape. As a consequence, selection processes were initiated favouring traits promoting fitness in the constructed landscape. This process implied a feed-back to niche shifts, but now also including evolutionary changes in fundamental niches.

Figure 2.

A conceptual model of species response to ecological opportunity caused by human-mediated niche construction along with landscape transformation during the Neolithic and the Bronze Age, and onwards. See the text for explanation.

Levin (2003) suggested that niche shifts occur by establishment of ill-adapted populations where ecological opportunity allows genetic refinement of populations, and that this process was promoted by disturbances. Based on an original idea by Simpson (1944, 1953), ecological opportunity, defined as a relaxation of selection or environmental constraints, is considered as a first step in the process that may ultimately lead to adaptive radiation (Yoder et al. 2010). One of the possible initial mechanisms behind ecological opportunity is the entry into new habitats. The landscape transformation during the Neolithic and the Bronze Age can be seen as large scale disturbance, and it undoubtedly created new habitats, in which species entered. The suggested model (Fig. 2) can thus be viewed as included in a more general model of adaptive radiation (Schluter 2000, Yoder et al. 2010), but only concerned with the absolute first stage in the process. This suggestion is not intended to imply that the landscape transformation a few thousand years ago resulted in a burst of speciation (obviously it did not), only that the initial stages in these processes are similar. In line with Thompson (1998), I suggest that ecological opportunity, and the response by species following from this, shall be seen as constituting a link between theory of short-term ecological processes and long-term evolution. The model fits into the concept of eco-evolutionary dyna mics, implying a reciprocal association between short-term environmental changes that influence evolutionary processes, in turn causing a feed-back to the environment experienced by organisms (Kinnison and Hairston 2007, Pelletier et al. 2009, Post and Palkovacs 2009).

Two questions raised by the model

The suggestion that species responded to the landscape transformation by niche shifts, either as a purely ecological expansion into previously unexploited fundamental niche space or by evolutionary changes in their fundamental niche, raises two questions that should be commented on.

1. How can the suggested niche shifts be reconciled with a notion that niche conservatism prevails?

The model in Fig. 2 is seemingly in conflict with an opinion by many authors (Peterson 2011) that niche conservatism prevails, and should be regarded as the norm for predicting species response to environmental change (Wiens et al. 2010). Contrary to the long-standing and well supported notion that evolutionary rates are mostly slow on long- time scales (Stanley 1979, Evans et al. 2012), there are numerous examples of very rapid evolutionary changes on ecological time-scales (Thompson 1998, Hendry and Kinnison 1999, Hairston et al. 2005, Carroll et al. 2007), and, as suggested by the latter authors, particularly in response to anthropogenic impact. Several of the studies on niche shifts mentioned above (Weber and Schmid 1998, Friberg et al. 2008) provide examples of such rapid evolutionary response. Gingerich (1983) recognized that the seemingly paradoxical combination of rapid short-term and slow long-term evolutionary change can be resolved by acknowledging an effect of time-averaging over different time-scales. Thus, niche shifts (ecological or evolutionary) occurring over time-scales of 102–103 yr may, if fluctuating and averaged over longer time-scales, for example those considered by Peterson et al. (1999) which are in the range 104–106 yr, still be manifested as niche conservatism. As implied by the argument of time-averaging (Gingerich 1983, Carroll et al. 2007), if fluctuating over time, short-term evolutionary changes in niche-related features will not be manifested as adaptive radiations or long-term evolutionary trends. It is likely that in most cases ecological and evolutionary response following from ecological opportunity fluctuate and cancel out, thus producing the commonly observed stasis in evolutionary lineages. Actually, even in systems where rapid adaptive radiations are known to have occurred, for example in cichlids, most ‘attempts’ to enter new habitats (or niche space) do not lead to rapid speciation (Seehausen 2006). Thus, viewed over long time-scales species niches appear conserved. Nevertheless, acknowledging short-term niche shifts may be crucial for understanding species response to environmental change. Thus, it would be a mistake to disregard ecological or evolutionary responses affecting niche shifts following environmental change.

2. If many species used forested habitats in the pre-agricultural landscape, what would prevent them from using present-day forests?

Part of the answer to this question is that many species in the species-pool of the open cultural landscape do indeed inhabit forests. As mentioned above, a considerable fraction of plants and insects associated with this landscape also utilize forested systems. However, the main answer is probably that most present-day forests are not similar to the forests occurring in the pre-agricultural landscape. There are several reasons for this. Firstly, most forests in central and NW Europe are managed, meaning that they are more homogeneous and probably denser than natural forests. Secondly, along with the modernization of the agricultural landscape during the last century, that has caused such drastic decline of many species due to habitat destruction, there have been other environmental impacts that are likely to have been negative as well. One such impact is nitrogen deposition, which has increased drastically during the last century (Gilliam 2006), and has caused large changes in vegetation composition (Bobbink et al. 2010). Probably as a result of long-term adaptation to generally more nutrient-poor soils, the natural species pools of at least plants (and thereby indirectly probably insects) are larger for low-productive terrestrial ecosystems in temperate regions (Pärtel et al. 2007b). However, we presently live in a literally speaking fertilized world, which may exclude parts of the original species pool. Indeed, changes in the composition of understory plant communities in deciduous forest over decadal time-scales suggest shifts towards more shade-tolerant and more nutrient demanding species (Verheyen et al. 2012). In rapidly changing landscapes, where semi-natural grasslands have declined drastically, the source populations may also have become lost due to the impact of fragmentation (Cousins 2009b), further reducing the species pool available for re-colonization of forests. Even apart from human impact, forest ecosystems change in a systematic way during interglacials, for example with regard to soil phosphorous content (Wardle et al. 2004, Kuneš et al. 2011). Thus, the forests that functioned as source habitats for many of the species that ultimately built up the species pool in cultural landscapes, were very different from the forest system that are presently occurring, and may thus presently not (or at least, not yet) be suitable as habitats for the open habitat species, as we now perceive them.

Concluding remarks

The suggested model (Fig. 2) for the mechanisms guiding species response to the landscape transformation initiated during the late Neolithic and the Bronze Age implies that species are more capable to cope with drastic landscape and habitat changes than would be expected from a static view of niches. Accordingly, if most species survived, and were even favoured by, the landscape transformation during the millennia BCE, what would prevent them from surviving the present human impact? The model does not include any principal constraints for species to survive by adapting also to current land-use and landscape change. The ongoing impact of human society will create new niche space that provides arenas for both ecological and evolutionary niche shifts. Thus, there may be strong selection for such shifts, as suggested by recently evolved adaptations to for example industrial pollution (Cook 2003), heavy metal pollution (Antonovics et al. 1971), and urban habitats (Partecke and Gwinner 2007). An important condition for the model is that the ecological processes causing niche construction, the ecological and evolutionary response to niche construction, and the feed-back to ecological processes, all occur at similar time-scales (Thompson 1998, Hairston et al. 2005, Post and Palkovacs 2009). It is thus important to ask whether there are any important differences between present-day landscape changes and those that took place during the Neolithic and the Bronze Age. Most likely, the present-day rate of change is at least an order of magnitude faster. The most drastic changes in land-use in central and NW Europe have occurred after WWII (< 102 yr) whereas the previous major transformation of the landscape was extended over more than 103 yr. This difference may however to some extent be alleviated by a slow response of many plant species to ongoing landscape change (Lindborg and Eriksson 2004, Herben et al. 2006). Furthermore, the spatial scale of the changes also differs, probably by several orders of magnitude. The size of fields, pastures and settlements were considerably smaller previously, making the landscape much more heterogeneous at smaller spatial scales. It is possible that there is a lower temporal and spatial scale boundary for the suggested model, and that the ongoing changes may have passed beyond that boundary. The question of whether such a lower boundary of spatial and temporal scales exists will be one important issue for developing the model in Fig. 2.

The model specifically concerns changes in the ecological conditions and their associated effects on niche relationships that took place several thousand years ago. As with all historical hypotheses, tests can only be done indirectly, and considering that the proposed mechanism of niche shifts has not been discussed before in the context of the Neolithic landscape transformation, it comes as no surprise that there is presently limited evidence to test the model. The evidence supporting the model (as have been described above) are all indirect, and focus on the fact that niche shifts, both ecological and evolutionary, are possible within a relatively short time-frame. The most powerful examination of the model would be by investigations of niche relationships, including both the realized niche and the fundamental niche, across the geographic range of species which at some parts of their range are associated with the cultural open landscape. Such studies would necessarily include both observational data and field experiments. The effects of anthropogenic landscape change on species should however be general, and although the model was developed for effects of the Neolithic/Bronze Age transformation of landscapes, it could as well be applied to present-day landscape change. A key issue is the mechanisms for interactions and feed-back between ecological and evolutionary processes, and the time scale for these interactions. Hairston et al. (2005) outlined different approaches to examine relationships between rates of ecological and evolutionary processes, including analysis of observational data, process-based modeling, and experimental manipulation. Their suggested approach should be fully adequate for testing the proposed model.

A general conclusion is that the static view of species niches presently dominating research and interpretations of species response to ongoing changes in land-use and climate should be confronted with hypotheses that species are capable of niche shifts, also including evolutionary shifts in fundamental niches. One of the major transformations occurring today is urbanization, and although cities are not historically new, the enormous expansions of urban habitats implies that extensive areas are subjected to change creating a range of ecologically new conditions (Ramalho and Hobbs 2012), undoubtedly provoking response by many species. The suggested model based on human- mediated niche construction providing ecological opportunity for niche shifts, should provide one useful framework for research on species response to urban development and other anthropogenic impact.


This paper was presented at the conference ‘Frontiers in Historical Ecology’ at Birmersdorf, Switzerland 2011. I would like to thank M. Bürgi for organizing the conference, and S. A. O. Cousins, R. Lindborg, M. Widgren, J. Ehrlén and G. Kylafis for comments on the manuscript.