THE IMPACT OF FORESTRY ON DRAGONFLY DIVERSITY IN CENTRAL SWEDEN

Abstract


Introduction
Wetlands such as lakes, ponds and, perhaps the most important component, moors, constitute a significant part of the Swedish boreal forests.The post-glacial forest landscape in Scandinavia is an immense mosaic where numerous water bodies are integrated with large areas of swampy forest, containing a high level of biodiversity (e.g.Hamberg eta!., 1995).Since the boreal forests are nothing but numerous patches (wetlands and drier areas) it is apparent that the wetlands with their special flora and fauna must not be treated separately from the surrounding forest but as an integrated part of it.Changes in the wetlands affect the surrounding forest and vice versa.
Many studies concerning the colonisation of dragonflies into different wetlands (mostly man-made ponds) have been made, both in temperate and tropical areas (e.g.Wildermuth & Krebs, 1983;Moore, 1991;Steytler & Samways, 1995;Samways, 1998).The opposite aspect, the decline of species due to human activities, is also a very popular topic.The cause of this decline is often change of land use, pollution or increased farming and various other human activities (Moore, 1980(Moore, , 1986;;Wildermuth & Krebs, 1983;Ruddek, 1990;Adomssent, 1994).In Europe, human land use has made several species of dragonflies threatened (van To! & Verdonk, 1988), which has prompted many governments to enact measures of protection (e.g.Pedersen & Holmen, 1994).
In the boreal forest, timber management is important and undoubtedly has effects on the dragonfly fauna inhabiting the wetlands in the forest mosaic.Rith-Najarian ( 1998) studied the dragonfly fauna along the upper Mississippi River in Minnesota and found that recently cleared areas contained the lowest number of species and the lowest species diversity.The highest diversity were found in old-growth forest areas.If timber management affects the fauna of running waters, the same must apply for standing waters.In this paper I ask the question: What will happen to a species-rich dragonfly fauna in a small lake when the hydrological conditions are rapidly changed due to clear-cutting of the surrounding forest?Are all species affected similarly or do some species-groups have a greater tolerance to change?Since the dragonfly fauna is in many ways dependent on the composition of water plants in the lake (Buchwald, 1992(Buchwald, , 1994)), the plant communities at the water's edge are also included in this study.

Materials and methods
During the summers of 1996 and 1997, 32 lakes near the Baltic Sea Coast in the province of Uppland, Central Sweden (approx.60°30'N, l8°E) were sampled for dragonfly larvae.Only small lakes with a surface area of <0.25 km 2 were chosen since larger lakes tend to have a more heterogeneous shoreline, which makes it difficult to sample all microhabitats present.Sampling was carried out using a standard water net with a mesh size of 1.5 mm and sweeping it through the aquatic vegetation close to the shoreline down to a depth of approx.0.5 m.Depending on the size of the lake, all larvae found in a section of up to 30 m of the shoreline were collected.In smaller lakes this section was reduced in order to avoid negative impact on species survival at the site.Between 150 and 200 larvae from each site were considered to mirror the actual species composition in the lake.All larvae were fixed in 80% ethanol with an additive of 4% formaldehyde to preserve pigment patterns and determined according to Norling & Sahlen (1997).The two species, Coenagrion puella and C. pulchellum, could not be separated and were treated as one species.
Since only larvae were used in the study, actual breeding of the species was confirmed at each site and vagrants automatically excluded.For further clarity, the dragonfly species were separated into two sub-groups, i.e. univoltine and partivoltine species.It was assumed that species belonging to the first sub-group would be less selective with regard to the quality of the breeding-water than those in the second group due to them often overwintering as eggs or young larvae followed by a short period of larval growth in spring (the genera Lestes and Sympetrum in the area; cf.Table 1).
The emerged and submerged vegetation was sampled at each site by randomly laying out 6 squares with an area of 0.25 m 2 and noting the plant species enclosed.
A brief survey of the surroundings in a circle of 1 km 2 centred at each lake was performed and notes on any type of forestry activities in the area were taken.Forestry measures now mainly consist of clear-cutting or, rarely, a more selective logging, but in the past (> 15 years ago) ditching in order to reduce the water level was sometimes performed.Based on data from the local forestry company (Korsnas AB) and my own observations, the lakes were sorted into 3 classes: (1) No forestry in the area or forestry measures that took place more than 15 years ago; (2) Recent forestry activities, i.e. 0-5 years ago; and (3) Forestry measures undertaken 6-15 years ago.All forested areas are planted with new trees during the first five years after clear-cutting, and allowing another ten years for the growth of the new trees was estimated to be sufficient to re-establish stable hydrological conditions in the area.These new conditions, however, need not be the same as the original ones.In spite of that, lakes where forestry measures were carried out more than 15 years ago were pooled with those without any forestry into class 1 in this study due to few observations and similar species-numbers.
A presence-absence matrix, ranking the lakes from high to low according to number of species present and ranking the species according to number of lakes they occur in were set up sensu Atmar & Patterson (1993).Species more likely to be present in only species-rich lakes will accumulate to one side of this matrix.
Statistical analyses were carried out using the programme Minitab for Windows™ release 11.12 (Minitab, 1996).
Nomeaclature of Odonata follows Norling & Sahlen (1997) according to which the species are systematically arranges in Tables 1 and 2.

Results
A total of 30 dragonfly species was encountered in the area as a whole (Table 1).The number of species breeding in a single lake ranged from 2 to 14, with an average of7.5.The number of water plant species in the lakes varied between 1 and 15, the average being 10.Altogether 47 plant species were found.The number of dragonfly species was correlated to that of the plants (Regression=0.35;r 2 =0.23;F=8.89; P<0.01) although the correlation is only moderately strong.
One way analyses of variance revealed significant differences between the number of dragonfly species present in the 3 classes of forestry (df=2; F=3.92; P<0.05) but for aquatic plants this was not the case (df=2; F=0.43; P>0.65; Fig. 1), although a certain decrease in the number of plant species was observed.Partivoltine species demonstrated an even more significant difference (F=5.22;P<0.02), whereas the distribution of univoltine species did not differ significantly among classes (F=0.77;P>0.45) (Fig. 2).There was no difference in species number between class I and 2, either for all species (F=0.20;P>0.65) or counting just partivoltine species (F=0.10;P>0.75).After pooling classes 1 and 2, the difference to class 3 becomes even more pronounced (all species: F=7.77; P<0.01; univoltine species: F=10.65;P<0.004).
In the presence-absence-matrix, 8 species were distinguished to occur in relatively species-rich lakes only, 2 univoltine and 6 partivoltine species (Table 2).Most of them occur only in southern Sweden, some of them reaching further to the north along the Baltic coast.1).Only partivoltine species decrease significantly (P<0.004) in class 3 whereas the small reduction in univoltine species is non-significant.

Discussion
Table 2. Dragonfly species more likely to be found in species-rich lakes only and their current distribution in Sweden.The fictive border separating "South" and "North" is at about 61 oN.

Southern
In previous studies it has been shown that certain dragonfly species prefer to live in an environment with a specific composition of aquatic plants (House, 1991;Buchwald, 1992Buchwald, , 1994)).A well-known example in Europe is the association of Aeshna viridis with lakes and ponds containing the water-lily, Stratiotes aloiaes (e.g.Robert, 1958;Norling, 1971Norling, , 1975)).However, this association between dragonflies and plants is never absolute and may vary among different areas.This 'water plant-dragonfly species' association is generally evident in that the lakes with the highest number of dragonfly species also had the highest number of aquatic plants.This connection is hardly surprising since lakes with a high number of plant species can be expected to offer suitable habitats for more dragonfly species.It is likely that even a species which is selective in its choice of breeding water would be able to find its favoured plant composition at least somewhere in a lake rich in plant species.In some species displaying a strong plant association, selectivity in the choice of breeding waters may lead to a reluctance to disperse after emergence, as in the central European species Ceriagrion tenellum (Buchwald, 1994).It may also result in a small chance to spread due to lack of suitable habitats in spite of a good ability to disperse as in Orthetrum coerulescens in central Europe (Buchwald & Schmidt, 1990).As species with these kinds of demands tend to become rarer in Europe today, they will eventually be confined to their isolated localities indefinitely, with no possibility of spreading (or surviving).
The lakes with the highest number of dragonflies (and plants) were all situated in old woodland.There had been forestry activities in the areas, but only a long time ago.Rith-Najarian (1998) found the same link between species diversity and old forests in riverine dragonflies in Minnesota.Apparently undisturbed forest is a component necessary to keep high diversity of dragonflies in boreal forest.Given the assumption that selective, "rare" species will find only lakes and rivers in old growth areas acceptable for breeding, the implications of forestry impact on these habitats are severe.The number of sites with a wide array of microhabitats for dragonflies have decreased significantly and this will make dispersal difficult for species dependent on particular plant-compositions or other factors necessary for egg or larval development.Lack of suitable larval habitats has been observed in the stream-dwelling Cordulegaster boltoni in Wales, where conifer plantations have eroded the normal stream margins of moorland and deciduous forest (Ormerod et al., 1990) Norling (1984a and pers.comm.) has studied the life cycles of several dragonfly species in Central Sweden.He reports that when the forest surrounding one of his study areas was logged the larval development in some species was shortened from the normal 3-4 years to 2-3 years, probably resulting from increased solar heating of the lake.Another effect was a significant increase in population size due to the short-term release of nutrients into the water caused by the logging.In the following year the population returned to a longer development time.The long-term effects on the life cycles in an area with clear-cuts has not yet been thoroughly investigated, but it is tempting to propose that dragonfly life cycles in any given undisturbed lake, provided that it is located in a closed forest that limits solar radiation, are "slow" in comparison to lakes in recently logged areas where changing conditions will induce the species to develop faster.High water temperatures are, in general, favourable to many insect taxa (e.g.Nordlie & Arthur, 1981), including many dragonflies (Waringer, 1983;Baker & Feltmate, 1987;Pritchard, 1989;Mathavan, 1990;Krishnaraj & Pritchard, 1995), but temperature needs to be combined with other factors, e.g.food availability.
Some species (cf.Norling, 1984a;also 1976, 1984band 1984c) thus seem to be able to adapt their life cycles, apparently without any obvious problems.This plasticity with regard to life cycles is probably what keeps most of the rarer species alive in a changing environment such as a lake after logging.The life-cycle changes of dragonflies in this study seem to occur gradually over a period of years.Other insect populations have also shown similar "slow" responses to disturbance, e.g.carabid beetle diversity after forest fires and clear-cutting (Lenski, 1982;Niemela et a!., 1993;Raila eta!., 1994;Wikars, 1995).
But what causes the decrease in the number of species?In the present study, the number of partivoltine dragonflies are clearly diminished by forestry while the number of plants and the univoltine dragonflies are not.The plants seem to react more slowly to imposed hydrological changes than the dragonflies and but a slight decrease is noted.Over a longer time span, the plants would perhaps also show a significant decrease.Possibly the wider temperature range in a lake exposed by logging is one of the key factors in the selective decrease of dragonflies.Temperature is an important limiting factor for many animals in the northern range of their distribution.Sjogren Gulve (1994) found the average water temperature in May to regulate whether the Pool frog (Rana lessonae Camerano) was present in small lakes or not.The same sensitivity to temperature extremes may be used to explain the absence of some dragonfly species in lakes surrounded by clear-cuts.Many of the dragonfly species in the area are on the northern limits of their distribution; so the similarity to the conditions for the Pool frog is obvious (cf.distribution maps for Sweden in Sahh~n, 1996 or Sandhall, 1986).If the selective species of dragonflies are sensitive to changes in temperature, they would quickly become eliminated by the harsh environment resulting when a small lake is surrounded by clear-cuts.All selective species occurring in this area of Sweden should thus be ones with a southern distribution.More northern species, selective or not, have in some cases the ability to endure dry periods as well as freezing within their 3-4 years of development (Johansson & Nilsson, 1991).Water temperature may therefore be of less importance further to the north.I found that the species which disappeared from clear-cut areas all have a southern distribution in Sweden (Table 2).Three of them (Coenagrion lunulatum, Erythromma najas and Leucorrhinia albifrons) also occur along the Baltic Sea coast further to the north where winters are somewhat (if not much) less severe than in the inland.Water temperature, at least in this part of Sweden, seems to play a crucial role in dragonfly distribution.
The univoltine species are not affected and, being opportunists, needing open water and food for the larvae only from April to July in Central Sweden.In general, such species are often present in a wide range of habitats, including temporary waters, where some traits, like a highly flexible life cycle and temperature-linked development, are needed for their survival (cf.e.g., Williams, 1996).
Presence or absence of a species may also depend on the requirements of the other life stages, the eggs and the adults.In any case, water temperature and the age of surrounding forest are not the only factors affecting dragonfly diversity in the forest mosaics of Scandinavia.A question to pursue is if the re-established species-rich fauna present in areas logged long ago indeed contains the same species as the primeval fauna in undisturbed areas.Considering the monoculture of secondary forest plantations one may fear that the seemingly recovered biodiversity turns out to consist of trivial (i.e."common") species only.I hope further research will prove this to be wrong.

Figure 2 .
Figure2.Mean number of dragonfly species (partivoltine species = striped; univoltine species = white) in the 3 forestry classes (defined in Fig.1).Only partivoltine species decrease significantly (P<0.004) in class 3 whereas the small reduction in univoltine species is non-significant.