Evaluating a Forest Conservation Plan with Historical Vegetation Data A Transdisciplinary Case Study from the Swiss Lowlands

Forest types shaped by humans host a specific set of plant species that may disappear subsequent to the abandonment of traditional uses of the forest. To improve the effectiveness of conservation management for plant species associated with traditionally managed forests, information on histor ical forest conditions and plant occur rence is vital. In the Swiss Canton Zurich, an action plan to promote light forests has been initiated by the state office for nature protection. This action plan has defined a set of 172 target species for light forests. In order to evaluate in which forest types these species occurred historically, we combined a vegetation survey from 1907 from the wooded mountain range of Lägern (Switzerland) with infor mation on forest structure in the early 20th century based on forest management plans. For 21 target species, 181 locations were determined and linked to historical forest structures. We recommend enhancing the effectiveness of the action plan for light forests by recreating coppice-with-standard forest on productive soils, considering the role of non-timber forest uses for ecosystem devel op ment, and engaging private forest owners to assist in the creation of light forest reserves.

abitat loss is a prime cause of species extinction (Pimm and Raven 2000).Halting or even reverting the decline in biodiversity is only possible if habitats with high conservation value are identified, located, and appropriately managed (Stoll-Kleemann and Job 2008).In ecosystems shaped by humans, nature conserva tion should consider the characteristic set of natural and anthro pogenic disturbances (Pressey et al. 2007).
In many European forests, the vegetation structure and composition developed under a multitude of forest uses and management types (Salbitano 1988, Kirby and Watkins 1998, Bürgi 2008).Consequently, information on human activities is especially important for understanding the current species composition (Hermy andVerheyen 2007, Josefsson et al. 2009).
Due to the introduction of modern forestry and the abandonment of traditional forest uses such as woodland pasturing and litter collecting, many European forests have become denser and darker (McCollin et al. 2000, Van Calster et al. 2008).Several studies have observed a decline in light-demanding plant species for these ecosystems in recent years (Stehlik et al. 2007, Van Calster et al. 2008).
For Switzerland, both the changes in forest use and management and the shift in forest vegetation are well documented, but these two trends have rarely been brought together (but see Gimmi et al. 2010).Changes in forest management have been analysed in the Swiss lowlands (Bürgi 1999, Bürgi andSchuler 2003) as well as in alpine areas (Gimmi et al. 2008).Changes in vegetation over hundred years have been reconstructed on the wooded mountain range of Lägern by studying old herbarium specimens, excursion notes, and publications, and comparing them with the We analysed a detailed forest vegetation survey from 1907 and historical forest manage ment plans to establish links between forest types and the occurrence of plant species.Based on our findings, we formulate recommendations for reserve site selection.Our case study shows how applied historical ecology can help improve conservation measures.
present flora (Egloff 1991).Several light-demanding plant species typical of forests or forest edges have disappeared entirely or declined in population size or in the number of occurrence sites.Chang es in microclimate -especially increased shading -have been considered the predominant cause of species loss.Similarly, a general decline in the average light indicator values over 50 years has been reported based on repeated vegetation sampling on forest plots in Switzerland (Kuhn et al. 1987, Kuhn 1993).
The reported decline in light-demanding forest plant species has caught the attention of nature conservationists.In the Swiss Canton Zurich, an action plan to promote light forests (Abegg et al. 2005) has been initiated by the state office for nature protection (Fachstelle Naturschutz Kanton Zürich).In this action plan, a set of 172 target species for light forests (figure 1) has been compiled taking into account the responsibility of the Canton to pro-tect and foster species specified in the conservation master plan for Zurich (Kuhn et al. 1992), as well as experts' assessments of the species' dependency on light forests.However, reserves are not selected primarily due to ecological criteria.In fact, potential new reserves are proposed by the various forest owners, e. g., mu nicipalities or forest cooperatives.A procedure for prioritising existing and proposed light forest reserves has been established based on the occurrence of target species and the forestry operations needed, such as initial logging to increase light availability and periodic interventions to maintain the open stand characteristics (Bertiller et al. 2006).This procedure makes it possible to select the most valuable forest stands among the proposed sites.
The action plan aims to create and maintain 1,000 hectares of light forests within a total forest area of 47,500 hectares.The list of target species in the action plan to promote light forest species (Abegg et al. 2005) encompasses attractive plants.The yellow foxglove (Digitalis lutea) (left) grows on dry, chalky, stony, and warm soils at sunny places.In the Canton Zurich the species occurs almost only on the Lägern.The peachleaved bellflower (Campanula persicifolia) is one of the more common target species in light forests.It is often found on mowed roadsides or at forest-road slopes (where the crown canopy is not entirely closed).

FORSCHUNG | RESEARCH
ance review conducted in 2007 revealed that 76 of the 172 target species did not occur in any of the reserves delineated.Apparently, the reserves selected do not contain all necessary habitat types to include the entire pool of target species.This triggered interest in the question under what conditions the missing species were once present, and what measures have to be taken to increase the number of target species in the reserves.
Links between historical species occurrence and habitat characteristics can only be established if data on historical vegetation are combined with data on historical forest use and management.
We had the opportunity to test this link on the small wooded mountain range of Lägern (northern Switzerland), as the local vegetation was surveyed over a hundred years ago (Rikli 1907).Forest use, management, and structure at the time can be derived from forest management plans (Bürgi 1999, Axelsson et al. 2002).
Based on these two main sources, we address the following questions: 1. What historical light forest types were present on the Lägern at the beginning of the 20 th century?What did they look like and what area did they cover?2. Which plant species listed today as target species in the action plan to promote light forests (Abegg et al. 2005) occurred once on the Lägern? 3. How do these results translate into recommendations to improve the effectiveness of the action plan?

Study Area
The study area covers about 1,000 hectares of forests on an altitudinal gradient from about 450 to 866 metres above sea level.It is situated on the southern slope of the Lägern, which forms the eastern most part of the Jura Mountains.This limestone mountain range stretches from eastern France to northern Switzerland.The forests belong to five municipalities: Otelfingen, Boppelsen, Buchs, Regensberg, and Dielsdorf, all in Canton Zurich (figure 2).

Data
The historical vegetation data were taken from Rikli's study Das Lägerngebiet (1907).This study is exceptionally detailed and includes descriptions of the vegetation as well as a map showing the exact locations of a subset of the species recorded (scale 1 : 25,000).The data were most probably collected between 1900 and 1907.
The forest structure was reconstructed mainly based on forest management plans.Forest management plans are the most important planning tool in forestry.They include a report on the previous use of the forests, a description of their current state (e. g., tree species composition), and guidelines for future management.These plans have been available since about 1880 for all non-private (i.e., community or cooperative) forests in the study area.For our study, we selected all plans issued in roughly the same period as Rikli's study, resulting in a total of 15 forest management plans from 1872 to 1926 (State Archive in Zurich, signatures in the range Z31 1350 to Z31 1466).To reflect the informal character of the plans, we use the English translations of the German names used in the sources: spruce is Picea abies, larch is Larix decidua, pine is Pinus sylvestris, beech is Fagus sylvatica, fir is Abies alba, but oak, like the German Eiche, can be either Quercus robur or Quercus petraea.
Aerial photographs from 1943/44 and 1952 (source: Swiss topo 1 ) were analysed as an additional means to identify and locate historical forest types, especially for the private forests, which are not covered by the management plans.

Methods
To assess and locate the historical light forests, we defined a set of historical forest types for Canton Zurich, based on adequate classification criteria.For this task, an extensive literature search was conducted and complemented by targeted archival research.
The forest area at the time was first digitised in ArcGIS (ESRI), based on a topographical map from around 1900 (Siegfriedkarte, provided by Swisstopo 1 [DV 033492.2]).The forest management plans contain maps depicting the management compartments, which are often further divided into sub-compartments.As all information on forest structure and tree species composition is structured by compartments and, sometimes, sub-compartments, compartment boundaries were located and digitised within the digitised forest area.Sub-compartments, which were only listed and described in the stand description parts of the forest management plans, were -wherever possible -located with the help of the aerial photographs from 1943/44 and 1952.In cases where sub-compartments could not be located, all information was summarised on the compartment level (details in Gimmi et al. 2008).Each forest patch was assigned to a historical forest type.
From the plant species mentioned in Rikli (1907), we selected all species also listed as target species in the action plan (Ab egg et al. 2005), and from these those we could localise on the map.As most target species occur at different locations, the number of occurrences surpasses the number of target species.We then as sessed the numbers of light for est species and their cumulative occurrences in the different forest types.We could thus determine which forest types bear the highest potential for fostering target species, by a procedure which corresponds to the selection criteria for light forest reserves applied in the Canton Zurich (Bertiller et al. 2006).

Typology of Historical Forests
Based on an extensive literature study (Bürgi 2009), we propose distinguishing seven types of historical light forests: private forests, stands including open areas, floodplain coppice forests, other coppice forests, coppice-with-standard forests (with few standards [< 100 m 3 /ha], medium density of standards [100 to 200 m 3 /ha], and many standards [> 200 m 3 /ha]), pine (high) for ests, and oak high forests.As for the historical non-light forest types, we distinguished between freshly planted stands and high forest stands.
Private forests in Canton Zurich were historically characterised by their special ownership structure and intensive use (Bürgi 1998), mainly focussing on the owner's needs and including forest pasturing and litter collecting (Forststatistik 1880).The average size of private forest plots was far below 0.5 hectares in many municipalities in Canton Zürich.In the district of Andelfingen, for example, 424 hectares of private forests were divided into 2,460 plots (Meister 1875).The intensive use of private forests led to low growing stocks of merely 60 to 120 cubic metres per hectare (Steiner 1954), which justifies classifying them as light forests.For our study, we assumed an average growing stock in private forests of 90 cubic metres per hectare (figure 3).

> Historical Light Forest Types on the Lägern
According to the historical map analysis, at the beginning of the 20 th century forests covered 1,003 hectares in the study area, of which 622 hectares were public forests.This figure corresponds well to the 615 hectares specified in the forest management plans as the area covered by all non-private forests in the study region (figure 3).All stands were assigned to one of the historical forest types.
Figure 3 shows the areas covered by the different forest types.Private forests were mostly located in the vicinity of the settlements (figure 2).About two-thirds of the public forests were covered by one of the types of coppice or coppice-with-standard forests.Coppice forests were mostly located along the ridge top of the Lägern.The light forest types "stands including open areas", "floodplain coppice forests", and "oak high forest" were not found in the study region.In total, about 90 percent of all forests in the study region assessed in the early 20 th century fall within one of the historical light forest categories.

Historical Forest Structure and Tree Species Composition
The average timber volume of coppice forests was characteristically low (50 m 3 /ha, figure 3).A similar timber volume was observed in the understorey of the two types of coppice-with-standard forests with a low or medium density of standards (data not shown).In the class with many standards, where the amount of growing stock in the standards surpassed 200 cubic metres per hectare, the growing stock in the understorey amounted to only 29 cubic metres per hectare.Pine forests, and especially the other high forests, had markedly more growing stock (figure 3).
For most of the historical forest types in public forests (except young stands and coppice forests), the management plans indicate the amount of growing stock for each tree species, which can Area covered and growing stock for all forest types occurring in the early 20 th century on the Lägern (data source: forest management plans, and, for the growing stock in private forests, the general literature -see text).The values given for coppice-with-standard (CWS) forests include the coppice layer and the standards.

FORSCHUNG | RESEARCH
target species) within each forest type area, interesting patterns become apparent (figure 4): The coppice forests were most diverse, including eight of 21 target species (38.1 percent) on only 3.7 percent of the area.The coppice-with-standard forests with a high density of standards, in contrast, included only two target species on 8.0 percent of the forest area.All types of coppice and coppice-with-standard forests included more occurrences of target species than to be expected according to the area covered.

Additional Information on Forest Use and Management
The anthropogenic impact on forest is, of course, not limited to forest management and the corresponding different forest types and their characteristic structure.The forest management plans and published studies document, in the northern part of Canton Zurich, widespread non-timber forest uses, such as woodland pas turing, litter collection, temporary agricultural use of the clearings, collection of oak bark for leather tanning, and grass cutting (Bürgi 1999).In the study area, litter collecting seems to have been of special relevance up to the early 20 th century (Bürgi 1998), and woodland pasturing was also reported up to the beginning of the 20 th century (Egloff 1991).The sources do not allow a spatially and temporally precise picture to be constructed about when and where these practices were performed and abandoned.The ecological effect of their long-term performance should still, however, be considered in the design of reserves and management guidelines.

Discussion
Data Requirements and Quality Studies of changes in forest vegetation have to be based on historical data on flora, with sufficient information on the locality of the plants to enable a repetition of the vegetation samples (Klecak et al. 1997 -for discussion of the methodological problems of comparing historical and modern vegetation data, see Tingley and Beissinger 2009).But even for the last 100 years, spatially precise information on forest vegetation in Central Europe is scarce (Schuler 1998).Literature and historical herbaria for the 19 th century, or even before, may allow the analysis of vegetation changes on a regional or sometimes a local scale (e. g., Shaffer et al. 1998, McCollin et al. 2000, Spillmann and Holderegger 2008),

208
Species composition of selected historical forest types in the Lägern area at the beginning of the 20 th century according to the proportions in growing stock (source: forest management plans).CWS: coppice with standard, BL: broadleaved species.Corresponding to the wording in the management plans, the colloquial names of the tree species are given.The values for CWS forests comprise the coppice layer and the standards.The target species listed in the recent action plan for promoting light forest species (Abegg et al. 2005) were not distributed evenly over the different light forest types around the beginning of the 20 th century.Data sources: Rikli (1907), forest management plans, Swisstopo (DV 033492.2).CWS: coppice with standard.

FIGURE 4:
be interpreted as tree species composition (see table above).Not surprisingly, pine forests were in fact dominated by pine, whereas the oth er high forests, as well as the coppice-with-standard for ests with medium to high densities of standards, were dominat ed by spruce.The highest density of oak is recorded in the coppice-with-standard stands with few standards.As coniferous species do not re-sprout after being cut, the coppice layer generally consisted exclusively of broadleaved species.Consequently, the proportion of "other broadleaved species" is highest in the coppice-with-standard forests with only a few standards, where the coppice layer contributes relatively more to the total species composition.

Light Forest Target Species Listed in Historical Vegetation Data
Out of the 172 target species listed in the action plan to promote light forests (Abegg et al. 2005), 37 species are mentioned by Rikli (1907).For 21of these, 181exact locations were determined from information in Rikli's text or map, and entered into the GIS.
If the relative area covered by the different forest types is compared with the proportion of the number of target species represented and the proportion of their cumulative occurrences (all

FORSCHUNG | RESEARCH
but their spatial precision is usually insufficient to link the information to forest stands and their structural characteristics.
The data used for the study presented (Rikli 1907) do not meet the standards of modern vegetation science.Although there does not seem to be any specific bias towards, e. g., forest types, special plant groups, or topographic or edaphic conditions, a bias in sampling cannot be ruled out.We therefore refrained from performing statistical analyses which could result in over-interpreting the data, and we are cautious about the overall interpretation.

Historical Light Forests
We used information on stand characteristics extracted from historical forest management plans and applied our own typology for historical forests, to classify more than 90 percent of the forest area in the early 20 th century as "light forest".The modern policy to restore the typical flora associated with light forests (figure 5), specified in the action plan to promote light forests (Abegg et al. 2005), only calls for about 2.1 percent of the forest area to be restored to light forest conditions.Given the range and number of modern demands put on the forests, this is probably reasonable.However, the larger proportion once covered by light forests according to our findings indicates it would be wise to choose these 2.1 percent carefully.Otherwise it will not to be possible to re-establish the 172 target species listed in the action plan.The finding that 76 target species did not occur in any of the light for est reserves created so far might reflect the sub-optimal placement of these reserves.
The stand characteristics of the light forests on the Lägern in the early 20 th century show that the proportion of spruce in coppice-with-standard forests, especially in those with a medium or high density of standards, is similar to the proportion of spruce in high forest stands (see table ).This observation contrasts with the common perception that coppice-with-standard forests have to be dominated by oak and hornbeam (e.g., von Hornstein 1951, Hasel 1985), but has been corroborated by other studies as a specialty of north-eastern Switzerland (Surber 1950, Voegeli 1950, Leib undgut 1971, Bürgi 1998).

Importance of Historical Light Forest Types for Target Species
The light availability in forests in the Swiss lowlands has been found to have declined as a result of a general increase in timber volume, changes in tree species composition (i.e., a trend from oak and pine to beech and spruce), and changes in forest management, especially the abandonment of coppice management (Bürgi 1998(Bürgi , 1999)).Our study supports the finding that coppice and coppice-with-standard forests are very relevant for the con->

© René Bertiller
The blue lettuce (Lactuca perennis) grows on sunny dry grasslands or on rocky hillsides on limesoil (such as, e. g., the Lägern).In forests, plants only grow if they receive a lot of sunlight.(2005) even found the highest number of forest plant species in coppice-with-standard stands and suggest distinguishing forest species from "coppice-woodland species".
The high proportion of spruce in the coppice-with-standard for ests illustrates that species-rich coppice-with-standard forests do not necessarily have to be dominated by oak.However, since oaks are important for insects (Southwood 1961, Kennedy andSouthwood 1984), it might well be that the effect looks different for fauna than it does for flora.In any case, if the growing stock surpasses a threshold of about 200 to 250 cubic metres per hectare (figure 3), high proportions of coniferous species seem to go parallel with a lower abundance of light-forest species.
The documented history of non-timber forest uses in the wider study area underlines the importance of considering the effect of multiple anthropogenic disturbances on forest ecosystems.In many of the historical light forest types, woodland pasturing and litter collecting caused most likely a significant nutrient export (see Gimmi et al. 2008).These extractive activities contributed to forest conditions that were not only lighter than today, but al so lower in nutrients (Wohlgemuth et al. 2002).General considerations imply that non-timber forest uses were performed wherever an adequate return for the time invested could be expected.As soon as alternative resources or products were available at affordable costs, traditional practices were abandoned.Forest litter, for example, was collected until ample straw was imported by railroad, and its use was taken up again in some regions during World War One and World War Two, when importing straw became more difficult (Bürgi 1998).

Limitations of the Study
Before specific recommendations can be formulated, some constraints of the study have to be considered.Forest vegetation is not only influenced by forest use and man age ment, but addition ally by a whole range of factors which are only partly and/ or indi rectly linked to human activities, such as climate, density of ungulates, natu ral disturbances, invasive species, and the at mospheric deposition of nutrients or toxic substances.Only those factors can be considered for which ample information is available.Given the widely reported decline in light-demanding species in Euro pe an forests (Stehlik et al. 2007, Van Calster et al. 2008), we assume that the chang es in microclimatic conditions in for est stands were foremost influenced by the increasing density and growing stock in forest stands, which tend to overrule the effects of climate change.An increase in ungulate population is likely, given that during the 19 th century, the numbers were small due to centuries of intensive hunting (Breitenmoser 1998).During the 1930s, foresters in Canton Zurich started complaining more about damage due to browsing (Bürgi 1998).It seems thus likely that the effects of abandoned woodland pasturing were to a certain extent counterweighted by increasing numbers of ungu lates, although their browsing effect differs from the impact by herded cattle, pigs, and goats in intensity, timing, and duration.A completely different development took place when litter collecting was abandoned.The associated reduction in nutrient export was not counterweighted, but instead In coppice-with-standard forests, the understorey is cut every 15 to 25 years.In the first years after a cut, the stands are fairly open, with optimal growing conditions for light forest species.
© WSL Photoarchiv A1081(Rümlang, Mittelwaldbild, zwei Jahre nach dem Schlag, 1924) Nowadays, light forest reserves are often placed on marginal land, where timber production is not profitable.Historically, however, light forests also occurred on productive forest land.FORSCHUNG | RESEARCH further accentuated by an increase in atmospheric deposition of nutrients, especially of nitrogen (Flückiger and Braun 1998).

Recommendations for the Action Plan
We thus propose improving the effectiveness of the action plan to promote light forests for Canton Zurich (Abegg et al. 2005) as follows.

Light forest reserves should be established not only on marginal
land, but also on more productive soils.As the reserve selection is based on suggestions of forest owners and the respective foresters in charge, light forest reserves are often created on marginal land, i.e., steep and stony terrain with shallow soils on which timber production is not profitable (figure 7).Historically, light forests were not restricted to marginal growing conditions, but also occurred on deep soils in flat terrain.We therefore suggest motivating forest owners to create additional reserves on more productive soils, to represent the whole diversity of growing conditions.
Our study shows that especially coppice and coppice-withstandard forests are important for the protection of light forest species.Coppice-with-standard forests do not have to be dominated by broadleaved species.High proportions of spruce as standards do not seem to suppress the presence of lightfor est species, as long as the growing stock does not exceed a certain limit of, e. g., 200 to 250 cubic metres per hectare.

Apart from light conditions, changes in nutrient availability in
forest soils should be considered.The concept of light forests should not only focus on historical light conditions but should also include nutrient availability.Changes in nutrient level in forest soils must be considered in the design of any management measure aiming at recreating past habitat conditions, e. g., by reintroducing woodland pasturing and litter collecting.Experiments have been established to study the ecological effects of litter removal (Bürgi et al. 2006).The action plan might additionally initiate systematic reviews and experiments on the ecological impact of different non-timber forest uses on various forest ecosystems.

An additional programme should be set up to address the specific needs and circumstances of private forest owners.
Forests can be managed sustainably with very different levels of productivity and growing stock.Our study indicates that growing stock in historical light forests was generally less than 150 to 200 cubic metres per hectare, compared to about 357 cubic metres per hectare in public forests today (data for 2005 2 ).In private forests, growing stock has even increased about five times, from merely 90 to 462 cubic metres per hectare, i. e., they changed from being very open, light forests to being especially dense forests.We suggest that an additional programme specifically addressing private forest owners could improve the effectiveness of the action plan, as private forests experienced the most dramatic changes regarding light conditions.Based on studies in other parts of the canton (Bürgi 1998), no major shifts between private and public ownership are to be expected for the last 150 years.We therefore can assume that most of today's private forests experienced this development.
Combining information on historical or contemporary popula tions of underrepresented light forest target species with habitat suitability modelling could be used to delineate high priori ty areas for additional forest reserves on private forest land.
The forest owners affected could then be financially and logisti cally supported and intensive logging and maintenance measures planned.
Changes in forest use, forest structure, and the ecological context of forest ecosystems should be systematically assessed to make conservation planning more efficient and effective in forest ecosys tems strongly shaped by human impact.Attempts to simply re store forests to an earlier historical state are anachronistic (Jackson andHobbs 2009, Hall 2010).But knowledge on historical states will lead to a more informed design and management of ecosystems.