Abundance and diversity of dragon ies four years after the construction of a reservoir

Abundance and diversity of dragon ies four years after the construction of a reservoir Few studies have investigated the impacts of river impoundments on reservoir constructions. Reservoir construction deeply changes dragon ies’ habitat structures, especially in relation to shoreline vegetation. This study investigated the effects of the impoundment of the Guadiana River and its tributaries on dragon ies four years after the construction of a reservoir. A total of 17 dragon y species (11 Zygoptera and ten Anisoptera), representing six families, were recorded in 21 sites in the years 1999 and 2003. Aeshna mixta, Coenagrion caerulescens, Coenagrion scitulum, Sympetrum foscolombei, Sympetrum meridionale and Sympetrum striolatum were sampled just before the impoundment took place, and Anax parthenope, Onychogomphus forcipatus, Orthetrum coerulescens, Trithemis annulata, Platycnemis acutipennis and Platycnemis latipes were recorded only after the construction of the reservoir. We concluded that the construction of the Alqueva Reservoir four years earlier did not change the dragon y species richness, possibly because of species overlap, but that the species composition was modi ed. Changes in marginal vegetation may have been important to new species compositions.


INTRODUCTION
Reservoirs are hybrid artiſcial systems of water storage (Thornton et al., 1990;Gopal et al., 1981). The formation of a hybrid system, according to some authors, alters the physical and chemical characteristics of the water as the current speed and oxygen concentration modify the marginal vegetation on the river's shoreline. These changes in the habitat directly affect the aquatic organisms. Due to this, many studies have been carried out to assess reservoir impacts on biota such as the dragonƀy.
The dragonfly is a key predator among insects, recorded frequently in aquatic ecosystems, such as streams, rivers, ponds and lakes (Corbet, 1999). In recent years, it has been used as an ecological tool to assess the integrity of aquatic ecosystems because of its high sensitivity to human disturbances, such as reservoir constructions (Samways & Steytler, 1996;Chovanec & Raab, 1997;Chovanec, 2000;Chovanec & Waringer, 2001;Clausnitzer, 2003). Many authors showed that lotic damming modified the marginal vegetation and thus the composition of dragonfly species (Cannings & Cannings, 1998;Suh & Samways, 2005). Marginal vegetation provides shelter and especially foraging sites for dragonflies (Corbet, 1999;Corbet, 2006). Some species with a longer pre-reproductive period can be negatively affected because they need a higher quantity of resources to conclude their life cycle (Kadoya et al., 2008). However, there are reports of highly tolerant species, such as Brachythemis leucosticte (Samways & Grant, 2007) and Crocothemis erythraea (Sato & Riddiford, 2007). Therefore, the construction of a reservoir can favour more resistant species and reduce dragonƀy species with higher sensitivity to the impacts of their habitat.
The objective of the present study was to assess 21 sites in a reservoir that was created four years earlier to determine the changes that had taken place, document the marginal vegetation changes and show how these have affected the assemblage of dragonƀies over time.

Study area
The study site was the Guadiana River and its tributaries in southern Portugal, 38 • 08 N, 7 • 35 E (Fig. 1a). The impoundment of the Guadiana River and its tributaries created the Alqueva Reservoir (Fig. 1b). The Alqueva Reservoir has a Mediterranean climate with hot dry summers and humid cold winters (annual rainfall of 400 mm). The sites were visited in 1999 and 2003. After sampling in 1999, a strong humandisturbance was noted by the removal of the vegetal cover (shrubby, shrubby and arboreous, and riparian cover) for the construction of the Alqueva Reservoir.

Dragonƀy sampling
Twenty-one sampling sites were used during the springs of 1999 and 2003, making three replicates during the peak ƀight period of the species known to occur in the area (Suh & Samways, 2001). Adult male Odonata were recorded by walking for 20 m around the shoreline. Only males were counted because females and prereproductive-age individuals are not always in close association with the water, and their identiſcation can be uncertain and unreliable. When the identiſcation of males was difſcult, the individual was captured, identiſed with a hand lens and then released. The following keys were used for identifying males: Askew (2004) andD'Aguilar &Dommanget (1998). The collected specimens were stored in ethyl-alcohol (70 %).

Environmental variables
Twenty-one circle areas (400 m in diameter) were delimited. The diameter of 400 m was used to avoid the superposition of the sampling sites. The centre of the circle was the sampling site. Six environmental variables were measured (hectare) in each area: arboreous, shrubby, herbaceous, shrubby and arboreous, riparian and water, according to the following criteria: 1. Arboreous: cork oak, olive and holm oak.
4. Shrubby and arboreous: cork oak, olive and holm oak in association with bushes.
To assess the vegetal cover, Idrisi32, module CROSSTAB software was used. The CROSSTAB module allows a comparison between the picture and the vegetal cover by the number of pixels. The result is a table with a number of pixels and many combinations with pictures and measures of association.

Statistical analyses
Canonical correspondence analyses (CCA) (Ter Braak, 1986), using CANOCO for Windows, were used to relate the species abundance data to environmental variables. Data on dragonƀy species were subjected to an analysis of diversity using the species richness Shannon-Wiener Information Index (Shannon & Weaver, 1963).

RESULTS
A total of 17 dragonƀy species (11 Zygoptera and ten Anisoptera) representing six families were recorded at 21 sites during the years 1999 and  (Table 2).

Ordination with environmental variables
The results of the Monte Carlo permutation test based on the ſrst eigenvalue and the sum of all of the canonical eigenvalues showed the signiſcance of the ſrst ordination axis and that of all of the canonical axes together (F = 0.388, P < 0.01; F = 2.849, P < 0.01, respectively), showing that the relationship between the species and the environmental variables was statistically signiſcant. The eigenvalues were 0.702 for the ſrst axis and 0.366 for the second axis.
A. mixta, C. scitulum, I. graellsi, L. viridis, S. fonscolombei, S. meridionale and S. striolatum were located near the centre of the ordination diagram, showing characteristics of typically ubiquitous species. The occurrence of C. lindeni was positively inƀuenced by the quantity of water and negatively inƀuenced by the presence of stones in the shoreline. C. caerulescens were found in conditions with high stone covers and low quantities Table 2. Summary of the shoreline characteristics measured for each of the 21 studied sites included in the Canonical Correspondence Analysis (CCA). Bodies of water were categorised based on size: large (15 m or more in width); medium (from 5 m but less than 15 m in width) and small (less than 5 m in width). Shade, stones and submerged macrophytes were measured in cover percentages. Resumen de las caracterǼsticas del litoral medidos en cada una de las 21 localidades estudiadas, que se han incluido en el Análisis Canónico de Correspondencias (CCA). Cada localidad fue categorizada por su tamaño: grandes (igual o mayores de 15 de anchura); medias (de 5 m pero menos de 15 m de anchura) y pequeñas (menos de 5 m de anchura). El sombreado, las piedras y los macróſtos sumergidos se midieron como el tanto por ciento de cobertura.

Shoreline characteristic Sites
Shade of water. All dragonƀy species recorded in this study showed no correlation with shade percentage, suggesting an afſnity for sunshine (Fig. 2).

DISCUSSION
After the impoundment of the North Anna River in Virginia, USA, Voshell & Simmons (1978) recorded a reduction dragonƀy species from 24 to 13. In the present study, after the impoundment of the Guadiana River in Portugal, the species richness was not modiſed. However, the composition was different. A. mixta, C. caerulescens, C. scitulum, S. foscolombei, S. meridionale and S. striolatum were no longer sampled. The preference of dragonƀy species for lotic or lentic ecosystems may have inƀuenced this new composition. The impoundment of a river creates an ecosystem that is similar to a lake, modifying chemical and physical characteristics of the water, such as dissolved oxygen, conductivity and current velocity (Hecky et al., 1984;Thornton et al., 1990;Avakyan & Podol'skii, 2002). Askew (2004) reported that dragonƀy species such as A. parthenope and O. coerulescens can show preferences for lentic ecosystems, whereas others, such as C. caerulescens, can prefer lotic ecosystems. A. parthenope and O. coerulescens were recorded exclusively after the construction of the reservoir when the river was like a larger lake; C. caerulescens was not sampled at all after the reservoir was constructed.
Apparently, there was a species overlap that occurred in rivers and streams by species that generally occur in ponds or lakes. Beyond modiſcations in the characteristics of the water, the construction of the Alqueva Reservoir also modiſed the structure of the vegetation. Before the impoundment of the river in 1999, the riparian vegetation was mechanically removed; this strongly altered the shoreline characteristics of the river.
Habitat structure, such as marginal vegetation, is very important for all dragonƀy species (Niba & Samways, 2006). However, some species, such as Sympetrum spp., can be more affected. This species has a longer pre-reproductive phase and needs high quantities of resources, such as food obtained mainly from riparian areas (Corbet, 1999;Kadoya et al., 2008). At site 36, where we recorded S. fonscolombei, S. meridionale and S. striolatum, there was an marked reduction in the shoreline vegetation, and these species were no longer recorded. A strong reduction in the abundance of Sympetrum spp. also was recorded after the construction of a dike (Cannings & Cannings 1998). According to the authors, modifications in the structure of the habitat negatively affected the abundance of this species. On other hand, C. erythraea exhibited high tolerance in the present study.
C. erythraea's abundance increased at the sites with reduced shrubby and arboreous vegetation. The positive correlation between human disturbances and C. erythraea has already been recorded in the S'Albufera Natural Park in Mallorca, Spain (Sato & Riddiford, 2007) and the Kogelberg Biosphere Reserve in the Western Cape Province of South Africa (Grant & Samways, 2007). This species exhibits rapid development, which contributes to its survival in disturbed sites . T. annulata was also tolerant, but only in the sites with reduced riparian vegetation. Males of this species show territoriality. and degraded sites decrease the risk of disruption by rival males .
We concluded that, from 1999 to 2003, after the impoundment of the Guadiana River, the dragonƀy species richness did not change, pos-sibly because of species overlap, yet the species composition was modiſed. Changes in marginal vegetation after the construction of the reservoir may have had an important inƀuence on the change in dragonƀy species composition.