Changes in the range of dragonflies in the Netherlands and the possible role of temperature change

Th e trends of 60 Dutch dragonfl y species were calculated for three diff erent periods (1980–1993, 1994– 1998 and 1999–2003). Comparing period 1 and period 3 shows that 39 of these species have increased, 16 have remained stable and 5 have decreased. Th ese results show a revival of the Dutch dragonfl y fauna, after decades of ongoing decline. Th e species were categorized in diff erent species groups: species with a southern distribution range, species with a northern distribution range, species of running waters, species of fenlands and species of mesotrophic lakes and bogs. Th e trends of these diff erent species groups were compared with the all-species control group. As expected, a signifi cantly higher proportion of the southern species show a positive trend than the all-species group. In the northern species group on the contrary, a signifi cantly higher proportion of the species show a negative trend than the all-species group. Diff erent explanations for these results are discussed, such as climate change, improved quality of certain habitats and degradation of other habitats. It is likely that the observed increase of southern species is at least partly caused by the increasing temperatures. Th e less positive picture of the northern species group is probably more infl uenced by other environmental factor than directly by climate change. Th ree out of six southern species which have become established since 1990 have done so during the aftermath of large invasions. It is concluded that dragonfl ies are well capable of using changing climate circumstances to colonise new habitats.


Introduction
During the last century, the Dutch dragonfl y fauna has shown large changes.Destruction of habitats, canalisation of streams and rivers, desiccation, eutrophication, acidifi cation and pollution led to an often strong decline of many species.Th is started in the fi rst half of the 20 th century, but was especially severe in the sixties and seventies of that century (Kalkman et al. 2002).Most aff ected were species of running waters and species of mesotrophic lakes and bogs (Wasscher 1994(Wasscher , 1999)), some of which even disappeared from the Netherlands (Coenagrion mercuriale, Nehalennia speciosa, Gomphus fl avipes, Ophiogomphus cecilia, Oxygastra curtisii, Leucorrhinia caudalis).Th e degradation of the Dutch dragonfl y fauna reached a maximum in the 1980's.Since the start of the 1990's, many species have increased.Th is is very obvious for some species of running water and ubiquistic species for which the Netherlands lie on the northern limit of their distribution range.Th ese species seem to have profi ted from the improving water quality (RIVM 2003) and the recent warm summer seasons (KNMI 2006).However, a number of species of other habitats, such as mesotrophic lakes and bogs, have also increased during last decade.
In this article we describe the revival of the Dutch dragonfl y fauna, which seems to be happening.Special attention is given to the role of temperature change.

Database
Th e database used for this article is build and maintained by the Dutch Society for Dragonfl ies, Butterfl y Conservation and the European Invertebrate Survey -the Netherlands.It contains over 307,000 records of 71 dragonfl y species up to and including 2003, mainly submitted by volunteers.Each record constitutes a species on a date on a locality.Th e records are checked for mistakes by a committee of experts, based on the known distribution and fl ight period of the species.For records of rare species further documentation like a picture or a description is required.
More than 279,000 records are available from the period 1980-2003.By far the largest number of these records was collected from 1994 onwards, but the number of records prior to this period is large enough to give a good impression of the distribution of the species in that period.
Th e database gives good information on the distribution of species.However it is subject to infl uences of the diff erences between fi eldwork done by the volunteers and large-scale professional projects.Th erefore, results based on the database can only be interpreted correctly with a good knowledge of the database itself.

Calculation of trends
Th e data set was divided in three periods : 1980-1993 (period 1), 1994-1998 (period 2) and 1999-2003 (period 3).Relatively few records are available from each year in period 1. Th erefore, this period includes fourteen years while periods 2 and 3 only include fi ve years.Th e 5×5 kilometre squares which had been visited at least in three diff erent months in the period May till August were selected for each year (table 1).Only records from these squares were used for the analysis.For eleven of the 71 Dutch species this resulted in usable records for only one or none of the three periods.Th erefore these species, all extinct or very rare, were not included in the trend calculation.
Presence or absence of dragonfl y species in the selected 5×5 kilometre squares was used, instead of the recorded number of individuals, as the latter is more prone to diff erences in recording behaviour.Th e consequence of this method is that a decrease or increase in observed numbers or in localities within a 5×5 kilometre square will go unnoticed.
For each species and period the relative abundance (RA) was calculated as follows: RA= (Number of squares in which a species is recorded)/(number of investigated squares) × 100%.
Th e RA's for each year were summed for each period and divided by the number of years.Th e relative change of a species was calculated as follows: Trend= (RA in recent period -RA in historical period)/(RA historical period × 100% Th e trends were divided in fi ve trend categories (table 2).

Southern and northern species group
Th e Dutch dragonfl y species were categorized as southern species, northern species or species without a typical southern or northern distribution pattern.Th is categorization was based on distribution maps of Northwest Europe (NVL, 2002).A southern species was defi ned as a species of which the northern limit of its range runs through the southern tip of Sweden or more southwards.A northern species was defi ned as a species of which the southern border of its range runs through the Netherlands or Belgium and which is further south only found at higher elevations or in small, scattered populations.

Habitat groups
Next to the southern and northern species groups, three ecological species groups were selected: species of running water habitats (rheophilic species), species of mesotrophic lakes and bogs and species of fenlands.For this categorization the habitat preference of Dutch dragonfl ies was used, as given in NVL (2002).Table 3 lists the species of the four selected species groups.Note that some species are appointed to more than one species group.

Statistics
χ 2 -tests were conducted to test the diff erences between the all-species group and the selected distribution and habitat species groups.Th is was done by using Microsoft Excel 2000 software.Species with increasing (>20 %) and strong increasing trends (>100 %) were lumped together and tested as increasing species.

Results
Th e relative abundance for each period and the trend between the periods is given for each species in table 4. Table 2 gives the number of species showing a certain trend between the diff erent periods.Trends between the fi rst and the third period could be calculated for 60 species.39 species (65%) show a positive trend, 16 species (27%) remained stable and 5 species (8%) show a negative trend.Most increasing species show the strongest positive trend between the fi rst and second period (see fi gure 1).
Th e results of the χ 2 -tests are given in table 5.

Species with a southern distribution pattern
Within the southern species group, signifi cantly more species show a positive trend than the all-species group, when period 1 is compared to period 2 and when period 1 is compared to period 3. Furthermore, a signifi cantly lower proportion of the southern species remained stable, when period 1 is compared to period 3 (fi gure 2).

Species with a northern distribution
Within the northern species group, signifi cantly more species show a negative trend than the all-species group, when period 1 is compared to period 2. Furthermore, a signifi cantly lower proportion of the northern species remained stable, when period 2 is compared tot period 3 (fi gure 3).

Diff erences in trends between habitats
Within the species group of mesotrophic lakes and bogs, signifi cantly less species show a positive trend than the all species group and signifi cantly more species show a stable trend, when period 1 is compared to period 3 (fi gure 4).Within the ecological species groups of running waters and fenlands, no signifi cant diff erences are found for the three trend categories.

Discussion
Th e results show that the Dutch dragonfl y fauna has recovered since the start of the 1990's, which is in sharp contrast with some other groups of invertebrates as but- terfl ies and bees (Peeters and Reemer 2003;Swaay and Groenendijk 2005).Only 5 dragonfl y species have declined, while a majority of 39 species has increased and 16 species remained stable.Out of the 27 species placed on the red list in 1999 (Wasscher 1999) 17 show an increase, 4 a decrease, 1 remained stable and 3 are still extinct.For the remaining 2 red-listed species (Coenagrion armatum and Leucorrhinia albifrons) no trend was calculated, as they were only recorded in one period.Populations of both species have recently been rediscovered (Van der Heijden 2001; De Boer and Wasscher 2006) in the Netherlands and although they are extremely rare, there is no evidence for an actual decline.Two diff erent causes can be pointed out for the increase or decrease of the diff erent species.Th e fi rst is climate change, the second is changes in the quality of habitats.

Climate change
Th e average temperature in the Netherlands in the last twenty years of the 20 th century was 0,7 degree higher than the average temperature of the fi rst twenty years of the 20 th century (KNMI 2006).Especially the spring temperature has shown a strong increase.Th is increase in temperature caused several southern species to expand their range northwards, becoming more common in the Netherlands.Th is is at least the case for Lestes barbarus, Aeshna affi nis, Anax parthenope, Crocothemis erytraea, Orthetrum brunneum and Sympetrum fonscolombii.Coenagrion scitulum expanded its range in northern France and Belgium and was fi rst found in the Netherlands in 2003 (Goudsmits 2003).Also for more common southern species like Lestes virens and Ceriagrion tenellum a positive eff ect of increasing temperatures is expected.
Whether or not higher temperatures also play a role in the negative trend shown by some northern species is diffi cult to say, because the habitats of northern species are Th e northern distribution of many southern species seems to be directly limited by the summer temperatures, resulting in a direct expansion of their range when temperature permits (Appendix1).Th e southern border of northern species on the other hand does not seem to be limited directly by temperatures, but seems to be determined by habitats being absent more southerly and by competition with other species prevailing in warmer climates.
Th e decrease of northern species as a result of increasing temperatures would in that case be caused by degradation of habitats and by increasing competition from southern species.Th is would result in a slow decline, which is far more diffi cult to detect than the rapid increase shown by southern species.
Another negative eff ect of increasing summer temperatures is increasing evaporation, resulting in lower surface and ground water tables.Th is can lead to desiccation of important vegetation structures in the riparian zone of lakes and the upstream stretches of streams.Th is happens especially in late summer, when the fi rst and most vulnerable larval instars of most species are present in the water.Furthermore, desiccation leads to the stagnation of ground water in seepage fed lakes and streams, causing acidifi cation.Also the turn-over rate of organic matter increases when lake shores dry out, causing nutrient enrichment.
Coenagrion hastulatum, Cordulegaster boltonii and Somatochlora arctica are examples of threatened species which are known to react negatively on desiccation caused by human influences (e.g.intensive drainage in agricultural areas and drinking water collection) (Groenendijk 2002;Groenendijk 2005;Ketelaar 2001a;Ketelaar 2001b;Wasscher 1999).It is expected that hot summers contribute to this problem.On the other hand, temporary water specialist like Lestes barbarus and Sympetrum flaveolum might have profited from waters becoming shallower.

Changes in quality of habitats
Th e test failed to show that the species group of running water contains a significantly higher portion of increasing species than the all-species group.However, this is probably due to the low number of species included in the group, making it diffi cult to fi nd signifi cant results.Of the ten included species fi ve show a strong increase, two a moderate increase, one is stable and two show a decrease when the fi rst period is compared with the third.Most striking is the comeback of Gomphus fl avipes, which from 1996 onwards reoccupied all large river systems in the Netherlands (fi gure 5), after an absence of more than 90 years (Kleukers and Reemer 1998;  1980-1993, period 2 = 1994-1998, period 3 = 1999-2003.1980-1993, period 2 = 1994-1998, period 3 = 1999-2003.1980-1993, period 2 = 1994-1998, period 3 = 1999-2003. and Groenendijk 2005)).In our opinion, these fi ndings leave no doubt that species of running water have increased strongly since 1980.Water quality improvement and restoration of the natural morphology of streams and rivers are likely to be the important causes for it.Some species probably profi ted from the higher summer temperatures as well.Th is is at least very likely for Orthetrum brunneum, O. coerulescens and Sympetrum pedemontanum.
Whereas the quality of running water habitats has improved, the threats for stagnant water habitats such as mesotrophic lakes and bogs are still present.Eutrophication, dessication and habitat fragmentation are still factors which explain why relatively few species in this species group show a positive trend.Th e intensity of eutrophication has reduced in recent years (RIVM 2003), but in many cases this has not lead to the recovery of lakes and bogs that have already been spoiled.Th e results of our analyses suggest that the negative trend of the species group of mesotrophic lakes and bogs stopped, but that they fail to recover.Especially Coengrion hastulatum, a species of mesotrophic lakes and bogs, is still declining in the Netherlands and is becoming increasingly endangered (Termaat 2006).

Conclusions
Th e analyses of the trends in the period 1980 to 2003 shows that the 55 Dutch dragonfl y species for which a trend could be calculated remained stable or increased during  1980-1993, period 2 = 1994-1998, period 3 = 1999-2003.that time period and that only 5 species have declined.Habitat degradation during the larger part of the 20 th century resulted in a degradation of the dragonfl y fauna in the eighties of that century.Improved water quality and increasing summer temperatures in the last two decades resulted in a revival of the Dutch dragonfl y fauna.
Although our analyses failed to show that the species group of running water contains a signifi cantly higher portion of increasing species than the all-species group, it is clear that especially species of running water have increased since 1980.Th is is probably largely due to the improved water quality of running waters and the restoration of the natural morphology of these systems.
Th e average temperature in the last twenty years of the 20 th century was 0,7 °C higher than those of the fi rst twenty years of the 20 th century.As a result signifi cantly more species with a southern distribution show a positive trend when compared with the all-species group.
Seven species very rare or absent prior to 1990 became established in the Netherlands, probably due to the increase in temperature.Th ree of these established themselves by means of large invasions.Th ese invasions were very eff ective, showing once more that dragonfl ies are highly capable of colonising new areas.No evidence could be provided to state that species with a northern distribution are decreasing due to the higher temperatures.Th e habitats where these species live (mostly mesotrophic lakes and bogs) have been strongly infl uenced by eutrophication, acidifi cation and desiccation in the 1960 th and 1970 th resulting in a decline of most of these species.Th is decline might have masked the infl uence of climate change.

Southern species and invasions
Six southern species rare or absent in the 1980's are now well recognised members of the Dutch odonate fauna: Lestes barbarus, Erythromma lindenii, Aeshna affi nis, Crocothemis erytraea, Orthetrum brunneum and Sympetrum fonscolombii.Anax parthenope is expected to become established in the coming years, as it recently became a regular guest and has reproduced successfully.Th e way in which southern species became established in the Netherlands diff ers among the species.E. lindenii, C. erytraea (fi gure 6) and to a lesser extent O. brunneum gradually expanded the northern border of their range.Th e other three species L. barbarus, A. affi nis, and S. fonscolombii became established after invasions, being rare in the years preceding these invasions (see table 6).Th e invasion of Lestes barbarus started in July 1994 (Ketelaar 1994) During the invasion records were made in most areas of the country with a strong emphasises on the dunes and the Pleistocene areas.At the majority of these localities several (up to 40) individuals were found.Almost all records were made at shallow, warm waters such as dried-out bogs and smaller dune lakes.In many cases the species established itself at these localities.Probably several smaller invasions occurred since 1994 but these went largely unnoticed as the species was already established.In the period since 1994 the species is found yearly in suitable habitat all over the Netherlands.Preceding the 1994 invasion the northern border of the distribution of L. barbarus was situated to the south of the Netherlands.Th e invasion in 1994 therefore resulted in a northwards expansion of its range of well over 300 km.
Th e invasion of A. affi nis started mid July 1995.All 39 records from 1995 came from the southern part of the Netherlands, most of them from the coastal dunes or from the Pleistocene areas.Almost all individuals were found at drying or dried-out waters, with low reeds or bulrushes.Of the 81 sexed specimens only four were females.Th is might be partly due to the inconspicuous behaviour of the females.Since the 1995 Figure 6.Th e distribution of Crocothemis erythraea in the periods 1980-1993, 1994-1998 and 1999-2003, showing its gradual northwards expansion.invasion the species is found several times a year in all parts of the Netherlands.Th e fi rst proof of successful reproduction was found in 2005 (Wasscher 2005) although it is likely that small (temporary) populations have existed since 1995.
In end May and begin June of 1996 a massive invasion of Sympetrum fonscolombii reached Northwestern Europe (Lempert 1997;Dijkstra and Van der Weide 1997).As with Lestes barbarus the species was recorded all over the country with a strong emphasises on the dunes and the Pleistocene areas.Most records were made at unshaded, standing waters with sparse vegetation and often sandy banks.Th e species managed to establish itself at many of these localities.Since 1996 the species is found every year at numerous localities across the country, although it has become less abundant than in 1996.
Th e invasions of L. barbarus, A. affi nis, and S. fonscolombii have two things in common: 1 During the invasion almost all specimens were found at suitable habitats and not seldom successful reproduction was noticed in later years; 2 Most records during the years of the invasions referred to more than one specimen.Th e three species which invaded Th e Netherland in 1994, 1995 and 1996 were rarely seen at unsuitable sites.Th is stresses the fact that these species are highly capable of localising suitable habitats.Th is is further emphasised by the fact that in most cases more than one individual was found at a locality.Th ese species do not fl y in clustered groups making it likely that the individuals from one locality all located the habitat on their own.Probably these species used their ability to recognise polarized light combined with visual cues on vegetation structure to detect suitable habitat from some height as has been shown for some species of dragonfl ies (Corbett 1999).Th is makes that a relatively high portion of the individuals taking part in the invasion is able to reproduce at a potentially suitable location.Th ese examples show that at least these species are capable of taking advantage of favourable circumstances in an extremely eff ective way.

Figure 5 .
Figure 5. Th e distribution of Gomphus fl avipes in the period 1996-2005.Th e species was not found in the Netherlands from 1902 to 1995.

Table 1 .
Th e number of well investigated 5×5 kilometre squares

Table 2 .
Categories of trend and the number of species showing this trend between periods

Table 3 .
Categorisation of the species in fi ve diff erent species groups.

Table 4 .
Relative abundance (RA) and trends for each species.
more prone to negative infl uences of other environmental factors.Five out of seven northern species occur in mesotrophic lake and bog habitats, while there are no northern species occurring in running waters.It is clear that habitat degradation is an important factor to explain the results of the northern species group, possibly climate change makes this decrease more severe.

Table 5 .
Results of χ2-tests of the observed proportions of trend categories within the diff erent species groups.*

Table 6 .
Southern species rare during the eighties which have become established since 1990.Th e second column states whether or not the species became established during a large invasion or gradually expanded northwards.(1)Anax parthenope is not yet established but has become a regular guest and is likely to become established in the future.