Land‐use history as a guide for forest conservation and management

Conservation efforts to protect forested landscapes are challenged by climate projections that suggest substantial restructuring of vegetation and disturbance regimes in the future. In this regard, paleoecological records that describe ecosystem responses to past variations in climate, fire, and human activity offer critical information for assessing present landscape conditions and future landscape vulnerability. We illustrate this point drawing on 8 sites in the northwestern United States, New Zealand, Patagonia, and central and southern Europe that have undergone different levels of climate and land‐use change. These sites fall along a gradient of landscape conditions that range from nearly pristine (i.e., vegetation and disturbance shaped primarily by past climate and biophysical constraints) to highly altered (i.e., landscapes that have been intensely modified by past human activity). Position on this gradient has implications for understanding the role of natural and anthropogenic disturbance in shaping ecosystem dynamics and assessments of present biodiversity, including recognizing missing or overrepresented species. Dramatic vegetation reorganization occurred at all study sites as a result of postglacial climate variations. In nearly pristine landscapes, such as those in Yellowstone National Park, climate has remained the primary driver of ecosystem change up to the present day. In Europe, natural vegetation–climate–fire linkages were broken 6000–8000 years ago with the onset of Neolithic farming, and in New Zealand, natural linkages were first lost about 700 years ago with arrival of the Maori people. In the U.S. Northwest and Patagonia, the greatest landscape alteration occurred in the last 150 years with Euro‐American settlement. Paleoecology is sometimes the best and only tool for evaluating the degree of landscape alteration and the extent to which landscapes retain natural components. Information on landscape‐level history thus helps assess current ecological change, clarify management objectives, and define conservation strategies that seek to protect both natural and cultural elements.


Introduction
Most sustainable forestry initiatives, whether at the international, national, or regional level, are challenged by climate projections that suggest a significant restructuring of vegetation and fire regimes in the future (e.g., Gottfried et al. 2012;Diffenbaugh & Field 2013;Elsen & Tingley 2015). To put these projections into the context of ecosystem variability, many researchers have examined global and regional biotic vulnerability to future climate change in light of what is known about past climatevegetation-fire linkages (e.g., Willis et al. 2007;Gillson et al. 2013;Benito-Garzón et al. 2014). Broad-scale generalizations, however, often have limited application for on-the-ground decision making because they overlook a host of nonclimatic factors that shape present ecosystems at fine spatial scales, including the legacy of disturbance, biotic interactions, and, perhaps most importantly, past land use.
Present landscapes may be categorized along a gradient based on their land-use history, with natural or pristine landscapes forming largely in the absence of people at one end and landscapes subjected to long and intensive human impacts at the other (Vale 2002). Truly pristine places are nonexistent, but some of the large coreprotected U.S. national parks and other nature reserves support vegetation that has experienced only short or minor human impacts (here referred to as nearly pristine landscapes). Altered or humanized landscapes at the other end of the gradient have been modified by land-use activities of different types, duration, and intensity. Most landscapes fall somewhere between the 2 end points and support both cultural and natural elements. These intermediate conditions are shaped by complex interac-tions of changing climate and land use that operate over different temporal and spatial scales. As a result, intermediate landscapes pose a unique conservation challenge: to support natural structure and diversity on the one hand and to maintain cultural or utilitarian attributes on the other (Lindenmayer & Hunter 2010).
We considered the importance of paleoecology for evaluating current landscape status in terms of its naturalness or alteration. We built on a growing body of literature that describes the use of historical and paleoecological data to broaden understanding of long-term perspectives on the historical range of variability (e.g., Swetnam et al. 1999;Whitlock et al. 2010;Gillson 2015); climate-driven changes in species ranges and vegetation composition (e.g., Williams et al. 2004;Tinner et al. 2013;Iglesias et al. 2014); human alteration of native vegetation, biodiversity, and ecosystem services (e.g., Dearing et al. 2012;Conedera et al. 2016); and potential rates of ecological change and no-analog situations (e.g., Williams & Jackson 2007;MacDonald et al. 2008;Willis et al. 2007). We drew on examples from our own research on 3 continents and 2 hemispheres with 2 objectives in mind: illustrate the extent to which present vegetation in our study regions has been shaped by past climate change and human activity and show how landscape-level paleoecological information can be incorporated into conservation strategies. Although multiple factors shape conservation strategies, we suggest knowledge of the past should be given priority consideration.
Our study sites were in 8 regions: northern and southern Switzerland; Tuscany and Sicily, Italy; interior South Island, New Zealand; northern Patagonia, Argentina; and western Washington and the Yellowstone region of the northwestern United States (Fig. 1). Information on past ecological change came from pollen, plant macrofossil, and charcoal records preserved in radiocarbon-dated sediment cores from lakes and wetlands. Land-use history was inferred from archeological and historical records, ethnographic accounts, paleobotanical studies, and models that explicitly considered the impact of different types of human activity on fire, vegetation, and climate (Henne et al. 2013;Pfeiffer et al. 2013).

Past Changes in Climate and Human Activity in Landscape Development
The current vegetation in each study region is an outcome of a particular sequence of events that were caused by climate, human manipulation, and disturbance (Fig. 2). Sites in the U.S. Northwest and northern Patagonia, for example, have undergone relatively little or only recent human alteration and feature large tracts of natural vegetation. In these regions, the most intense land use occurred in the last 150 years with Euro-American settlement. New Zealand and European sites feature both natural and humanized vegetation, reflecting land use over centuries in New Zealand and millennia in Europe (Fig. 2). Among the European examples, the Italian sites have been so heavily altered that they presently support little natural vegetation. The steplike progression of land-use change in the European sites has shifted the vegetation of the study sites to more altered conditions through time, although the pattern is not unidirectional. The extensive deforestation during the Iron Age and Roman Period in Europe was followed by a period of land abandonment and some forest recovery, and recent landscapes in Switzerland are more forested now than they were a century ago (Fig. 2).
In all these regions, the importance of climate change in shaping past vegetation is evident prior to substantial human activity (Table 1). In the northern pre-Alps, boreal species that survived the last glaciation in isolated populations (e.g., Betula pendula, Juniperus communis, and Hippophaë rhamnoides) rapidly expanded their range during a 5-6°C warming at onset of the Bølling Interstadial (approximately 14,650 cal yr BP [years before AD 1950]) (Ammann et al. 2009). A subsequent south-north expansion of temperate trees, including linden (Tilia), elm (Ulmus), oak (Quercus), and hazel (Corylus), was facilitated by additional warming in the early Holocene (11,000-8200 cal yr BP) (Lang 1994;Birks & Tinner 2016). A rapid cooling of -2°C at 8200 cal yr BP led to decline of thermophilous communities and expansion of forests dominated by European beech (Fagus sylvatica) and silver fir (Abies alba) (Tinner & Lotter 2001).
In landscapes less altered by humans, climate has remained the primary driver of vegetation dynamics up to the present day. One of the large-scale but slowly varying  climate forcings is the change in summer solar radiation (insolation) occurring over millennia. Summer insolation reached a maximum from 11,000 to 9000 years ago in the Northern Hemisphere and led to a period of warmer-thanpresent conditions. It steadily declined to present values and resulted in gradual cooling. The opposite trend in summer insolation characterizes Southern Hemisphere records (Fig. 3). At Cygnet Lake in Yellowstone, forests of lodgepole pine (Pinus contorta) were established with Holocene warming beginning 11,000 years ago, and fire activity was higher than at present between 11,000 and 6000 cal yr BP. Fire activity decreased as the climate cooled in the last 6000 years ( Fig. 3a; Table 1) (Millspaugh et al. 2004). In western Washington, high levels of prairie woodland taxa were present in western Washington during a warm period, 9000-4000 cal yr BP (Table 1; Fig. 3b), and this period was followed by an expansion of present forests of mesophytic and xerophytic conifers in the last 4000 years as the climate cooled (Walsh et al. 2008). In northern Patagonia, Chilean cedar (Austrocedrus chilensis) expanded after 3500 cal yr BP as a result of increased moisture and warmer summers (Fig. 3c) (Souto et al. 2015). Climate became less important as an agent of vegetation change in Europe after the Mesolithic-Neolithic transition, approximately 8000-6000 years ago, when small foraging populations were replaced by larger more sedentary cultures that were supported by agriculture and pastoralism. In the course of this cultural change, primary forest was lost and fire activity increased (Figs. 3e-h; Table 1 (Lotter 1999;Tinner et al. 2005;Conedera et al. 2016). From the end of the 19th century through the 20th century, however, declining rural populations and land abandonment reversed this trend and led to expansion and closing of many European mountain forests (Fig. 2) (Conedera et al. 2016). Similarly in the Mediterranean region, agricultural fields abandoned in the 20th century have become flammable shrubland (Figs. 3g & 3h) (garrigue, maquis) and have led to increased fire activity in recent decades (San Miguel-Ayanz et al. 2013).
In contrast to the European sites, Diamond Lake in South Island, New Zealand, has a comparatively short history of land use, and pollen and charcoal data show the vulnerability of mesic podocarp-Lophozonia forests to human-set fire (Wilmshurst et al. 2008;McWethy et al. 2010) (Fig. 3d; Table 1). Prior to Maori arrival, about 700 years ago, natural ignitions were exceedingly rare and the dominant forest species were poorly adapted to fire. People represented a new ignition source, resulting in a loss of approximately 50% of the native forest in a matter of decades (McWethy et al. 2013). Rapid deforestation was facilitated by the postfire expansion of highly flammable shrubs (e.g., Lepidospermum, Kunzea) ( Fig. 3d; Table 1), and a positive feedback was created in which each new fire led to further forest loss (Perry et al. 2012 (Fig. 3d), resulting in the present vegetation mosaic of native and non-native forest and pasturelands (Fig. 2).

Figure 3. Data on pollen percentage and charcoal accumulation rates (CHAR [particles cm −2 ·yr −1 ]) from our study sites over time: (a-d) non-European locations and (e-h) European locations. The CHAR data describe variations in fire activity, and peaks are usually interpreted as individual fire episodes (Whitlock & Larsen 2001). The bar at the base of each diagram shows the changing anomaly (relative to present) of summer insolation (the darker the shading, the more intense the radiation anomaly). Age (cal yr BP) refers to years before AD 1950. The records for (a) Cygnet Lake, (b) Battle Ground Lake, and (c) Lago Mosquito show before (prior to approximately AD 1850) and after Euro-American (after approximately AD 1850) periods. (d) Diamond Lake (South Island, New Zealand) shows no change in summer insolation because the record spans only the last
The history of individual species (i.e., those of special conservation interest) supplements our general vegetation reconstructions. For example, European silver fir was once more widespread than it is today, based on paleoecological records and related model simulations Ruosch et al. 2016). Silver fir grew well under conditions that were warmer than present and relatively humid, but the species is highly sensitive to fire and browsing. As a result of increasing human activity starting in the Neolithic Period, fir forest was replaced by stands of almost pure beech, sweet chestnut (Castanea sativa), and deciduous or evergreen oak (Quercus spp.) in the lowlands and mountains (Figs. 3e-g) (Colombaroli et al. 2007;Tinner et al. 2013). Given its wide distribution prior to anthropogenic disturbance, silver fir would likely occupy a broader range in Europe than it does today, so long as browsing by domestic and wild animals and arson fires are controlled (Ruosch et al. 2016). The current disequilibrium with present climate is recognized only on the basis of paleoecological studies Ruosch et al. 2016), and ecological niche models that neglect such land-use legacies overlook a critical component of the environmental history (e.g., Maiorano et al. 2013).
In recent decades, mortality of whitebark pine (Pinus albicaulis), a keystone species of high-elevation forests in the U.S. Northwest, has been widespread as a result of climate change, fires, non-native pathogens, and insect outbreaks (McKinney & Tomback 2011). Ecological niche models based on present climate suggest whitebark pine will be largely extirpated from its current range with continued warming (Chang et al. 2014). Pollen and charcoal data from the Yellowstone region provide insights into white pine's vulnerability to past climate change and fire. P. albicaulis or Pinus flexilis was apparently more abundant and widely distributed in the region from 11,000 to 7000 cal yr BP when summers were warmer than at present, winters were colder and wetter, and fires were more abundant (Iglesias et al. 2015). White pine was abundant at all elevations in a period when competing species, Engelmann spruce (Picea engelmannii) and lodgepole pine, were poorly represented (Iglesias et al. 2015). Thus, paleoecological data provide important insights for the future: white pines have survived periods of warmer summers and higher fire activity in the past and these factors may not represent critical thresholds, at least in the near future. In contrast, recent threats from non-native white pine blister rust (Cronartium ribicola) and native mountain pine beetle (Dendroctonus ponderosae) (Logan et al. 2010;Smith et al. 2013) have no or unclear precedence and are cause for concern.
Cultivation of sweet chestnut in southern Switzerland began in Roman times as a source of fiber, intensified in the Medieval Period as a food source, and continued through the 1950s to create a widespread monoculture in the southern Swiss lowlands that is well adapted to fire ( Fig. 2f) (Conedera et al. 2004;Morales-Molino et al. 2015). Recent abandonment of chestnut cultivation has led to mixed forests of chestnut and other broad-leaved trees that are often mistaken as natural. In the last 40 years, native (e.g., Ilex aquifolium, Hedera helix, Laurus nobilis) and exotic evergreens (e.g., Trachycarpus fortunei, Cinnamomum camphora) are spreading in the understory of mature, former chestnut groves, and pioneer exotic species (e.g., Ailanthus altissima, Pawlonia spp.) are colonizing forest patches after windthrow and fire (Conedera et al. 2001). Forest encroachment into open areas has also resulted in loss of diverse, human-created meadows of cultural and ecological value . Without the benefit of paleoecological data, the long and intensive management history of chestnut would not be known, and the altered nature of present chestnut forests might be overlooked.

Incorporating Landscape History into Conservation Strategies
We contend that knowing ecological history, including the degree of past landscape alteration, can help clarify management objectives, conservation targets and, to some extent, the intensity of effort required to achieve conservation goals (Fig. 4) (Machado 2004). For example, paleoecological information from the most pristine landscapes provides critical insights about long-term ecological dynamics, which forms a basis for evaluating current biodiversity, disturbance regimes, and structural complexity. Pollen data from Cygnet Lake in Yellowstone suggest that lodgepole pine forests have a long history and a high tolerance for different levels of burning and climate change (Fig. 3a). This information supports handsoff management strategies as long as the frequency of fires does not exceed Holocene levels (e.g., Westerling et al. 2011).
Most of our study sites are neither completely altered nor pristine but represent an intermediate condition that falls within Vale's (2002) categories of "natural/inhabited" with large patches of nearly pristine elements and some areas of alteration, "mosaic" with both natural and humanized components, and "unevenly altered" landscapes dominated largely by a matrix of humanized elements with isolated refugia of native vegetation (Fig. 4). Paleoecological records from intermediate landscapes inform such topics as the vulnerability of the different vegetation components to a range of climate conditions (e.g., rapid warming, severe or prolonged drought in the past); the extent to which present vegetation is maintained by climate-driven or anthropogenic fire regimes and the vulnerability of particular vegetation components and species to changes in fire activity; the identity of species that have benefited or been disadvantaged by past land-use practices, including the appearance of new taxa and elimination of others; and the role of past land-use and management in creating the current landscape pattern of natural and cultural components.
In study sites where both natural and cultural components are valued, complex management strategies are needed to protect native plant communities alongside cultural or managed vegetation. We suggest that intermediate landscapes require the most intensive conservation effort, given objectives to maintain both cultural and natural components (Fig. 4). For example, at Battle Ground Lake in the U.S. Northwest, pollen and charcoal data point to the relatively young age of the current oldgrowth forests, which were first established only about 4000 years ago (Fig. 3b). This insight affirms the importance of late-successional reserves to protect biodiversity and retain structural complexity as the region becomes more developed and conversion to commercial forests increases . At Diamond Lake in New Zealand (Fig. 3d), protecting remnants of native podocarp-Lophozonia forest requires active fire suppression, but this objective must be balanced against the use of fire to maintain culturally important tussock grasslands created by the early Maori. In Switzerland, the high cultural value of chestnut has made it a management priority. Deciding which humanized landscape is the desired condition (i.e., chestnut groves like Roman and Medieval time or closed mixed forests of the last century) can only be informed by paleoecology (Fig. 3e & 3f), and restoration goals must be balanced against new realities including the co-occurrence of non-native forest species and recent chestnut mortality due to drought (Tinner et al. 1999;Conedera et al. 2010).
Altered landscapes, such as represented by our sites in Italy (Figs. 3g & 3h), have been manipulated for millennia to meet cultural values and utilitarian needs, including agriculture, livestock production, and silviculture. Conservation of cultural components and often small remnants of native vegetation often conflicts with present utilitarian needs (Fig. 4). The structural and biotic simplicity of managed forests in all our settings leaves them vulnerable to disturbance (e.g., monospecific beech and spruce forests at Lobsigensee; olive orchards near Lago di Massaciuccoli and Gorgo Basso; non-native pine plantations near Diamond Lake and Lago Mosquito; commercial Pseudotsuga forests near Battle Ground Lake), and contrasts with the natural resilience of native forests in the past (Kulakowski et al. 2016). Altered forests can return to a more natural level of structural complexity and biodiversity in the absence of silvicultural management, but conversion can be slow and unpredictable. Moreover, back-to-nature conservation efforts may be impractical in landscapes where newly established non-native species are well adapted to disturbance and resistant to change. In the Italian sites (Figs. 3g & 3h), for example, high levels of disturbance (e.g., large severe fires, intensive grazing) and flammable shrub expansion (e.g., maquis) have reduced opportunities for native species recovery, including silver fir (Henne et al. 2015;Vannière et al. 2016). The same issues are noted in New Zealand and northern Patagonia where establishment of flammable shrubs and pine after human-set fires has created a positive feedback that leads to more fires and native forest loss (Simberloff et al. 2010;Paritsis et al. 2015).

Future Perspectives
Ecological history is sometimes regarded as interesting background information in conservation efforts but of little practical value. This viewpoint is increasingly voiced in discussions about future climate change on the grounds that restoration to a prior state may be inadequate to address the rapid climate changes and novel conditions that lie ahead (e.g., Loarie et al. 2009;Benito-Garzón et al. 2014;Elsen & Tingley 2015). We argue that even in places where temperatures may soon exceed those of the last 11,000 years, knowledge of the past remains indispensable for the preservation of ecosystems and species of special concern (see also Hunter et al. 1988;Birks 2012). Paleoecology can help identify the levels of management required to meet desired restoration goals as part of a cost-benefit analysis. Restoring native podocarp-Lophozonia forests in New Zealand, for example, will require intense conservation effort with uncertain outcome, given evidence of past forest vulnerability to fire. Replanting silver fir where it once grew, in contrast, has the potential to return an important native species to Italian and southern Swiss lowland forest and help maintain biodiversity and ecosystem services under climate change (Henne et al. 2015;Ruosch et al. 2016).
Beyond serving as a guidepost for restoration, paleoecology can help in the assessment of current conditions in light of the historical range of variability (Landres et al. 1999) by providing a baseline for assessing current precedence. A temporal baseline that is too short, however, will lead to erroneous estimates about the range of conditions necessary to maintain particular species and vegetation types as well as the naturalness of present dis-turbance regimes. It may also overlook important species that are currently missing or overrepresented as a result of active or passive management in the past (Conedera et al. 2016). Lengthening the historical baseline through paleoecology can avoid incorrectly selecting altered vegetation conditions as a back-to-nature restoration goal or adopting fire-management policies that inadvertently increase natural disturbance risk (Gillson & Marchant 2014;Whitlock et al. 2015).
Paleoecological insights from one region can also guide management actions in another (Smith et al. 2016). Responses of species to past warming (i.e., at the onset of the Bølling Interstadial in Europe or during the early Holocene in the U.S. Northwest) in nearly pristine settings can suggest their response to a similar magnitude of change in more altered settings where data may not be available. Similarly, the vulnerability of introduced species to disturbances (e.g., fire, blowdown, insect outbreaks, and avalanches) can be informed by their response to past disturbances in their native range. For example, alteration of fire regimes as a result of pine expansion in New Zealand and Patagonia may best be understood by examining pine responses to past fires in North America.
Paleoecology has been an active discipline for nearly a century, and in recent decades, high-resolution records of vegetation and fire history are stimulating new research questions and applications (Seddon et al. 2014). These data sets have contributed substantially to understanding of how species and communities adapt to changes in climate and land use of varying duration and intensity. Although paleoecology is often motivated by a curiosity about environmental history, the threats of current and future changes in land use and climate have elevated its importance. Our 8 study sites show the uniqueness of the ecological history of each location and the need for landscape-level reconstructions. Inasmuch as management success relies on knowing which ecosystems are most vulnerable and why, the sequence of events leading to a landscape's current position along the pristine-tohumanized gradient is critical information for clarifying conservation objectives and evaluating outcomes.