- Top of page
Individual-based behavioural models are powerful tools that can predict the response of animal populations to environmental change (Pettifor et al. 2000; Schadt et al. 2002). These models rely on fundamental behavioural rules that should remain the same under a variety of circumstances. In order to parameterize these models, there is a need for basic data on the response of animals to anthropogenic structures. Wind farms are such structures whose effects on animals have been particularly neglected. However, the use of wind as a renewable energy source is now increasing in many countries. Wind turbines are often arranged in rows, along coasts or mountain ridges, where soaring birds can use the same air currents that lead to the placement of power facilities. As a result, they present a risk to birds of collision or mortality.
In order to mitigate bird mortality at wind farms, factors associated with collisions must be identified. Several such factors have been suggested, and include structural and design features of farms and their elements. Lattice towers may be more suitable for perching and thus present more risk than tubular towers (Osborn et al. 1998, 2000). Weather may be important too, and fatalities have been related to poor visibility (Winkelman 1985). The number of casualties also tracks temporal fluctuations in bird abundance and activity (Musters, Noordervliet & Ter-Keurs 1996; Osborn et al. 1998). Finally, bird habituation to turbines, or hunting and flying at low height near turbines, all might increase mortality (Winkelman 1985; Orloff & Flannery 1992). Less attention has been paid to habitat use as a factor of risk. Some soaring birds may make intensive use of the mountain ridges where wind turbines are placed, but no explanation has so far been provided for the circumstances in which flying near the ridges increases the risk of collision.
The Straits of Gibraltar are the main point of migratory passage for many soaring birds of the north-west Palaearctic on their journeys between Europe and Africa (Bernis 1962; Moreau 1972; Fernández-Cruz et al. 1990; Finlayson 1992). Several hundred thousand soaring birds cross these straits each year during pre- and post-nuptial migrations (Bernis 1980; SEO/BirdLife 2001). The Straits of Gibraltar are included among the four areas in Spain with the greatest potential for producing wind energy (INM 1988; IDAE 1992). Relief and wind are the two principal factors affecting both the behaviour of soaring birds (Ciconiiformes and Falconiformes; Bernis 1962) and the selection of sites for wind farms in this area.
In this study we analysed the effect on soaring birds of the first two wind energy plants ever to be installed in the Straits of Gibraltar. The specific aims were to determine (i) the bird mortality rate associated with wind energy facilities; (ii) the effect of these facilities on bird behaviour and habitat use; (iii) the factors that lead birds to approach the turbines; and (iv) mitigation measures that may reduce avian mortality.
- Top of page
The distribution of collisions with wind turbines was clearly associated with the frequencies at which soaring birds flew close to rotating blades. Patterns of risky flights and mortality include a temporal component (deaths concentrated in some seasons), a spatial component (deaths aggregated in space), a taxonomic component (a few species suffered most losses) and a migration component (victims were usually species with resident populations rather than species occurring during migrations). This raises the question of what determines whether a bird approaches a turbine close enough to be hit by blades. We suggest that the mechanisms involved must be behavioural.
Bird abundance varies markedly with season. Kestrel density was highest during the post-fledging period, in July and August. Wintering vultures, which may be up to 10 times the number of residents (SEO/BirdLife 2001), sometimes join the local population for months. Therefore, the concentration of kestrel deaths in summer and vulture deaths in autumn–winter might support the hypothesis that the rate of wind turbine casualties increases with bird density (Musters, Noordervliet & Ter-Keurs 1996; Osborn et al. 1998).
The seasonal pattern of vulture deaths may also be explained by flight behaviour. In common with other soaring birds, griffon vultures need vertical air currents to gain height (Bernis 1980; Pennycuick 1989, 1998). The availability and location of lifting currents varies seasonally. In summer, high temperatures allow the formation of thermals from valley bottoms, whereas in winter lower temperatures make thermals scarcer. Birds are thus constrained to gain height with slope updrafts (Pennycuick 1989), whose force on most winter days may be insufficient to lift vultures well above the ridge, thereby exposing them to the turbines. In weak winds, other soaring birds, such as common buzzards Buteo buteo L. and short-toed eagles, often circled together with vultures in slope updrafts but did not closely approach the turbine blades and rarely collided with them. These species have lower wing loadings than vultures, and apparently make a more efficient use of the ascending currents, gaining altitude quicker and farther from the turbines.
Most vulture deaths were recorded at PESUR. At this wind farm, the number of vultures counted was 1·5 times larger, but the overall risk index was 3·4 times higher, than at E3. Therefore, the risk of collision at PESUR was higher than predicted from local differences in the relative abundance of vultures. The gentle and short slopes of PESUR may generate weak updrafts, especially in light winds, exposing circling vultures to the turbines once at the ridge, the point where wind strength is maximum (Pennycuick 1989). Such weak updrafts may have formed at the two turbine rows where both vulture corpses and risk situations aggregated. In strong winds at PESUR, or in light winds at the long and steep slopes of E3, lifts gained sustaining force and risk situations disappeared. Besides circling, straight flights below turbine level were risky because birds had to cross through the narrow space between two operating machines. It appears that vulture risk of mortality was mediated by flight behaviour, which in turn was determined by the interaction between wind and relief at specific locations.
The importance of local variation in wind behaviour as a potential predictor of vulture mortality was indirectly supported by the patterns of kestrel mortality. Common kestrels hover during hunting flights but do not depend upon lifting air currents to reach the mountain ridges. Accordingly, kestrel carcasses were less aggregated at PESUR. Almost all observations of kestrels were made at PESUR, especially in areas with open vegetation suitable for hunting. Open areas were rare at the mountain ridges of E3, where no kestrel casualties were recorded.
Kestrels sometimes perched on lattice towers, and griffon vultures frequently flew at close distance to the blades, or between two adjacent turning turbines. Indeed, the frequency of risk situations was higher than expected for all species, which indicates that turbines or towers were not actively avoided, as has been suggested by Orloff & Flannery (1992) but in contrast with the conclusions of recent studies (Osborn et al. 1998; Guillemette & Larsen 2002).
Lattice towers have been considered more dangerous to birds than tubular towers because many raptors use them for perching and, occasionally, nesting (Howell & Noone 1992; Orloff & Flannery 1992; Osborn et al. 1998). However, we found that deaths in lattice and tubular towers occurred with frequencies proportional to their respective occurrences. Most deaths and risk situations occurred in two rows at PESUR with little space between consecutive turbines. This windwall configuration (Orloff & Flannery 1992) might force birds that cross at the blade level to take a risk greater than in less closely spaced settings. However, little or no risk was recorded for five turbine rows at PESUR having exactly the same windwall spatial arrangement of turbines. Therefore, we conclude that physical structures had little effect on bird mortality unless in combination with other factors.
Low visibility has been suggested as a causal factor of bird collisions at wind facilities. Winkelman (1985) observed that the probability of a collision occurring during a clear day was remote. However, collisions in our study area always occurred at times of good visibility. In addition, it was obvious that vultures adjusted their movements to avoid the revolving rotor blades.
Most wintering vultures were young birds (Griesinger 1996). Young vultures, however, were not especially prone to collisions compared with other age classes. Most common kestrels observed in summer at wind farms, as well as all common kestrels found dead, were juveniles. Aggregation of juveniles during the post-fledging period is not unusual in common kestrels (Bustamante 1994). Higher numbers of juvenile kestrels at wind farms may be the simplest explanation of their vulnerability to turbines.
A fraction of the vultures that collided with wind turbines could be migratory or wintering birds. Other soaring-bird species, chiefly birds that occur in the study region in large numbers but only during their migration periods (e.g. white stork Ciconia ciconia L.), were rarely involved in risk situations and collisions. In 1994, the location of both wind farms was well away from the routes used by migratory birds, which, in addition, flew above the turbines. Such site-specific factors will not necessarily apply to other locations.
conclusions and management implications
Deaths per turbine and year in Tarifa were much larger than those recorded in similar studies elsewhere (Howell & Noone 1992; Orloff & Flannery 1992). Although the effect of turbine mortality on populations cannot be established, all species affected are listed as threatened or vulnerable in Spain (Blanco & González 1992). Mitigation measures are thus necessary in order to minimize mortality. Our results indicate that the most sensible approach is to suspend the operation of the small number of turbines that cause most deaths only under the wind speeds that lead to risk situations.
A more general recommendation is that each new wind power facility project should include a detailed study of bird behaviour at the precise location where construction is proposed in order to identify species that are particularly vulnerable, which sites are intensively used, and hence the optimum turbine location.