When south goes north : Mediterranean dragonflies ( Odonata ) conquer Flanders ( North-Belgium )

Since 1980, eight southern dragonfl y species have been regularly recorded in Flanders. Th ey show a signifi cant increase in relative abundance, relative area as well as indications of reproduction since the beginning of the nineties, with peak occurrence mainly in the 1995–1999 period. Since 2000, numbers are lower but more species were simultaneously present. Th ree species, Lestes barbarus, Crocothemis erythraea and Sympetrum fonscolombii, show a combination of earlier arrival, earlier reproduction with a higher frequency and higher maximum ranges and can be considered as having stable populations in Flanders. All other southern species show in general a later arrival, only one confi rmed or probable reproduction and have much lower maximum ranges. Two other species, reaching their northern limit of distribution in Flanders, Erythromma viridulum and E. lindenii have clearly expanded their relative area since the eighties. Th eir relative abundance also increased although this shows more fl uctuations.


Introduction
Climate change on Earth is causing a growing concern.It is still not fully understood whether this is part of normal global climate fl uctuations or mainly the result of human activities.Whatever the cause, it is clear that the rate of warming during the last 30 years has been greater than at any other time during the last 1000 years and atmos-pheric CO2 concentrations have never been so high (IPCC 2001).Increasing concern over the implications of this "global warming" for biodiversity have resulted in a large amount of studies on a wide array of taxa, communities and ecosystems (Parmesan and Yohe 2003).Ecological responses are already clearly visible on diff erent levels, although in many regions there is an asymmetry in the warming and in the precipitation regimes that undoubtedly will contribute to heterogeneity in ecological dynamics across the system (Hughes 2000;Peňuelas et al. 2002;Walther et al. 2002).As a result of the predicted future increase in global temperatures, between 1.4 and 5.8°C (IPCC 2001;Stainforth et al. 2005), a number of species are expected to shift their ranges in response, rather than adapt to warmer temperatures in situ (Huntley 1991).
Among insects, some European butterfl ies have shifted or expanded their ranges northwards (Parmesan et al. 1999).It is expected that other winged insect species will show similar responses to climate change.However, few studies have focussed on other groups of insects.In particular, there is not much information available on taxa with aquatic larval stages such as dragonfl ies, and data are often lacking or confl icting (Hickling et al. 2005).
In Flanders (Northern Belgium) the distribution and abundance of dragonfl ies is well known over a long time (De Knijf et al. 2006).Flanders is situated in Northwest-Europe, somewhat halfway between the Mediterranean region and Fennoscandia.Moreover, it lays in the smallest part of the (Central) transition zone between the so-called Northern and Southern climatic zones, determined by the 18°C mean July temperature and the 2500 degree day (www.worldclimate.com).Th erefore, it seems an interesting region to check for changes in fauna composition which could be due to increasing temperatures.Like most insects, dragonfl ies have short life cycles and often a high reproduction rate and dispersal capacity.Th ey have the ability to react relatively quickly to changes in climate.
Th e aim of this contribution is to analyse in general to what extend southern dragonfl y species have expanded their range into Flanders.In another paper (in preparation) we will threat in more detail the relations between these distribution patterns and climatic and other environmental variables.Th is will allow us comparisons with the recent analysis of Goff art (2006) for the adjacent Walloon region.

Data sources
Data are derived from the Gomphus Dragonfl ies Working Group's distribution database of Flanders.At the end of 2005 it contained about 55.000 records on 66 species, including historical data going back to the 19 th century (Selys 1888).All records up to 2004 were used for the analysis.Th e Odonata were mapped using Universal Transverse Mercator (UTM) 5×5 km grid squares as units.Special eff orts have been made to achieve a good coverage of the territory during the last 15 years.For the more recent records the database includes information on precise locality, date and number of observed individuals and life-cycle stage (larva, exuvium, teneral, male, female, adult-copulation, egg-laying) thus providing indications of reproduction and/or permanent populations.For reproduction, three categories were considered: confi rmed (exuvium, larva or tenerals), probable (tandem or egg-laying) and possible (population with high number of individuals).For historical records however, population and life-cycle information is often lacking or diffi cult to interpret, in particular from literature sources.

Selection of species
For our analysis we selected 10 species which have their main distribution area in the Mediterranean part of Europe (Askew 1988;d'Aguilar and Dommanget 1998) but are actually present in Flanders (De Knijf et al. 2006).We distinguished two groups: -group A: 8 species for which Flanders is no part of their historical distribution area (< 1980): Lestes barbarus (Lb), Aeshna affi nis (Aa), Anax parthenope (Ap), Orthetrum brunneum (Ob), Crocothemis erythraea (Ce), Sympetrum fonscolombii (Sf ) and S. meridionale (Sm), and Coenagrion scitulum (Cs) although the status of the latter is less clear-cut than for the other 7 species due to its much more fragmented southern distribution range and the probability of its historical distribution area having reached Flanders.
-group B: 2 species that reach in Flanders the northern limit of their distribution range: Erythromma lindenii (El) and E. viridulum (Ev).

Analysis
To detect general changes in range and numbers, we used Relative Area and Relative Abundance in time per Group (combined), and per species.We defi ne Relative Area in a period as the percentage of diff erent 5×5 km UTM squares occupied by the selected species (or by a group) compared to the total number of diff erent squares occupied by all dragonfl y species in that period.A higher percentage does not a priori mean a greater "range" (enlargement of occupied territory) as squares can be clustered without really "enlarging" the range.However, an analysis of the distribution maps per species for the diff erent periods (years) showed that higher relative percentages resulted in a real increase of occupied territory.Th erefore, we decided to use this as a measure for broad changes in range.Relative Abundance is the percentage of the combined record numbers of a group or species compared to the total number of records.We compared fi rst the data for 8 time periods.To obtain suffi cient records, we used two broader categories <1900 and 1900-1949 for the "historical data".From 1950 on, data were grouped in fi ve decades (and one pentade 2000-2004).To better visualize the recent evolution in the last 25 years , fi gures are given per year.To determine permanent and non-permanent populations, we used reproduction frequencies and categories as well as presence patterns of distribution and abundance during the last 25 years.

Relative area, relative abundance and reproduction per time period (Table 1)
For each species group, relative area and abundance are highly correlated (Spearman-Rank A:0.88, B:0.89, p < 0.05).Before 1900 species of Group A were found in 13.8% of the investigated squares.From 1900 onwards the proportion is much lower but in the 1990ties we see a remarkable increase (27%).In the pentad 2000-2004 the proportion reaches even 34.6%.For the species of Group B, proportions increase from the 1980ties but vary in the earlier periods.Th e relative abundance of Group A and B show a similar pattern as the relative area.Both species of Group B have been present in all but one time period.Group A species have been recorded in each time period but numbers are higher before 1900 and since 1990.Indications of reproduction are also higher since that decade.For Group B, reproduction can not always be confi rmed in the older data.

Relative area and abundance by groups per year in the recent period (1980-2004)
For each species group, relative area and abundance are again highly correlated (Spearman-Rank A:0.97, B:0.87, p < 0.05).Th e cumulative relative area for both groups during the last 25 years (Fig. 1) gives for almost all years a higher percentage for Group B than for Group A, except in 1996 and 2003.Compared to the fi rst 15 years, Group A shows a clear increase in relative distribution during the last decade.Th ere is a prominent peak in 1996.Th e cumulative relative abundance for both groups (Fig. 2) gives a very similar pattern to the cumulative relative area (see Fig. 1).
Relative area and distribution by species per year in the recent period  Th e relative area and abundance per year for each species of Group A separately shows again a clear increase in presence and number of occupied squares in the last 10 years (Fig. 3, 4).Four species (Coenagrion scitulum, Anax parthenope, Orthetrum brunneum and Sympetrum meridionale) remain at a very low percentage and have an irregular presence pattern.Aeshna affi nis has slightly higher numbers but is also frequently absent.Lestes barbarus, Crocothemis erythraea and Sympetrum fonscolombii show a rather stable presence although numbers fl uctuate.In Lestes barbarus the area is smaller during the last fi ve years.All three species have a greater or maximum range in 1996, with Sympetrum fonscolombii showing an extreme peak.
In Fig. 4 the recent cumulative distribution of Crocothemis erythraea for four consecutive time periods (a = < 1990, b = 1990-1994, c = 1995-1999 and d = 2000-2004) is given.We also indicate the investigated squares (dotted) for each time period.Th e species shows since 1990 a real increase in distribution area in Flanders.

First presence, confi rmed reproduction and maximum range
Table 2 summarises a number of presence and reproduction data.Th ree species (Lestes barbarus, Crocothemis erythraea and Sympetrum fonscolombii ) show a combination of earlier arrival, earlier confi rmed reproduction with a higher frequency and higher maximum ranges.All others (Coenagrion scitulum, Aeshna affi nis, Anax parthenope, Orthetrum brunneum and Sympetrum meridionale) show only one (more recent) confi rmed (or probable) reproduction and have much lower maximum ranges.

Discussion
In the 19th century, six southern species have been observed in Flanders.However, little is known about their distribution area and population abundance.In historical references like Selys (1859, 1888) and Bamps and Claes ( 1893), comprehensive descriptions often lack and there is certainly a bias resulting from heterogeneous sampling, low visiting eff ort and 'collectioning'.In the following periods, southern species were very rarely observed.Towards the last decades of the 20th century, this changed.Since 1980, eight southern dragonfl y species have been recorded in Flanders.Th e fi rst in row were Anax parthenope (1983), Lestes barbarus and Sympetrum fonscolombii (1984) and Crocothemis erythraea (1987).Th ey were followed by Aeshna affi nis (1991), Orthetrum brunneum (1994), Coenagrion scitulum (1999) and fi nally S. meridionale (2000).In general the present data of this group shows a clear increase in relative abundance as well as relative area since the beginning of the 1990ties, with "peak" occurrence mainly in the 1995-1999 period.Since 2000, numbers are lower but more species were simultaneously present.However, not all the members of the group have achieved a similar level of 'expansion' succes and permanence.Only three of them can actually be considered as having a "permanent" population: Lestes barbarus, Crocothemis erythraea and Sympetrum fonscolombii.Until 1980, only 6 records of Lestes barbarus are known from Flanders, three of them dating from the hot summer of 1976, without proof of permanent populations.In 1984 and 1985, the species was seen in several locations, and at one site there was a very small "population" present but reproduction could not be confi rmed.In 1994 and in particular in 1995 there was an invasion of the species in North West Europe (Monnerat 2002;Parr 2003).Since then, the species maintains permanent populations in Flanders (De Knijf 1994;Stoks 1994;Stoks and De Block 1997).Sympetrum fonscolombii has been recorded 10 times before 1980, and although only isolated or very few individuals have been seen, the species might have reproduced in the past (Selys 1859).In the eighties the species had been observed in two sites, but since the nineties it occurs yearly and has been reproducing.Populations of more then 50 individuals have been sighted.Since a peak in 1996 during an invasion in North Western Europe (Dijkstra and van der Weide 1997;Lempert 1997), S. fonscolombii now is present in several tens of squares and maintains populations on several sites, also in the southern part of Belgium (Goff art 1999;Paternoster 2000).Crocothemis eryth-  (a =< 1990, b = 1990-1994, c = 1995-1999 and d = 2000-2004) in Flanders, based on 5 × 5 km squares.For each time period the investigated squares are given as dotted squares in the background.reae is without doubt the most successful of all southern species.Th e species has been observed in the past (Selys 1878), but permanent populations or reproduction have never been recorded.Apart from observations in 1963 (Cammaerts 1967;Dumont 1967), the species was only seen more regularly since the eighties when fi rst reproduction was proved but is now present in more then 20% of all squares (De Knijf 1989, 1995;Tailly 1991).Anax parthenope has only once been observed in the surroundings of Brussels in 1884 but after this, the species was not recorded anymore until the eighties when there are several observations (all males) (Lerner 1984, Michiels 1984).In the nineties however, Anax parthenope has been recorded in 6 localities in Flanders, mostly individual males, with a small invasion in 1999 (De Knijf 1999).Probable reproduction was only recorded in 2003.It is clear that the species has still no permanent populations in Flanders.Aeshna affi nis has never been observed in Flanders before 1900 and there is only one record (1969) before 1980.Th e species was again observed in 1991 and present in almost every year since 1994.An invasion in 1995 with higher numbers probably resulted in the fi rst proof of reproduction in 1996 (Van de Meutter 1995;Andries 1997;Van den Berghe 1999).At several localities the species has been seen in consecutive years, but it cannot yet be considered as having permanent populations in Flanders.From Orthetrum brunneum there exist only a very few historical records although it might have reproduced in the past (Bamps and Claes 1893).Since 1980, the species was fi rst recorded in 1994 (Verstraeten 1996) and has been present regularly but in very small numbers (Van de Meutter 2004).Reproduction occurred probably in 2005.Orthetrum brunneum has to be considered as an irregular and very rare species in Flanders.Coenagrion scitulum has been mentioned a few times by Selys (1868) but in contrary with the Walloon region without proof of reproduction in Flanders.Later only a few records are known from 1949 and 1973.Since 1980, the species has only been observed in 1998, in 2003 and in 2004, and although most probably there are now some small local populations present where reproduction may occur, the species can not be considered as permanently established yet (De Knijf 2004).Th e only known record of Sympetrum meridionale from the past dates from 1886 (Bamps and Claes 1893).In 2000 the species was seen again at two diff erent localities, with even a young male at one site (Versonnen et al. 2002).In 2001 only one female was seen (Versonnen et al. 2002) and in 2002 no records are know.In 2003 reproduction took place at least at one site, with the observation of several tenerals.Since then, the species has not been observed again.Two species reaching their northern distribution range in the region, Erythromma viridulum and E. lindenii have permanently maintained it and even expanded their area (see also De Knijf 1995), although there are fl uctuations in distribution and abundance.Th e presented patterns of range extension of the southern species in Flanders are a part of a more general long-term expansion of these species into Northwest Europe (Reder 1993, Drees et al. 1996, Ott 1996, 2000, 2001, Dijkstra and van der Weide 1997;Lempert 1997;Vanderhaeghe 1999;Gonseth and Monnerat 2001;Guerold et al. 2001;Ketelaar 2002;Mauersberger 2003;Parr et al. 2004).However, whether in our region the observed patterns are only (or mostly) a result of increasing temperatures as is the case in Wallonia (Goff art 2006), or whether other environmental variables play also an important role, has still to be confi rmed.Th is will be investigated in the next future.

Figure 1 .
Figure 1.Cumulative relative area per year from 1981-2004 for the 2 species of Group B (left: dotted bars) and the 8 species of Group A (right: barred, checkered and black bars indicating respectively no reproduction, probable and confi rmed reproduction of at least one species of the group).

Figure 2 .
Figure 2. Cumulative relative abundance (%) per year from 1981-2004 for the 2 species of Group B (left: dotted bars) and the 8 species of Group A (right: barred, checkered and black bars indicating respectively no reproduction, probable and confi rmed reproduction of at least one species of the group).

Figure 4 .
Figure 4. Cumulative distribution of Crocothemis erythraea in four consecutive time periods(a =< 1990,  b = 1990-1994, c = 1995-1999 and d = 2000-2004)  in Flanders, based on 5 × 5 km squares.For each time period the investigated squares are given as dotted squares in the background.

Table 1 .
Relative area, relative abundance, number of species and reproduction confi rmation for Group A (8 spec.) and Group B (2 spec.) in the 8 time periods (** = confi rmed reproduction, * = probable reproduction, see also Material & Methods).