Introduction

Electrocution has been considered one of the most significant causes of mortality among raptors [1]–[4]. This type of mortality affects raptor population dynamics by regulating the density of the birds [5], targeting specific age classes and changing populations [6]. Thus, this non-natural cause of mortality is capable of destabilising populations [7] and could potentially cause local extinctions [8].

The Iberian Peninsula has one of the highest raptor densities in the western Paleartic, particularly of large eagles (genus Aquila) [9]. Although electrocution is one of the main causes of mortality for these species [2], [3], few broad studies have been undertaken on the factors influencing electrocution on eagles (although see [10]). Many existing studies evaluate birds' interaction with power lines in local areas [11], or focus on a single type of pylon [12]. However, because each pylon has structural characteristics, such as the kind of insulators present and the function of the pylon, these characteristics could influence mortality rates [13]. In this sense, it has been documented than both the insulators (pin-type or suspended, [10], [11]) and the number of phases above the crossarm [14], have an impact on the probability of birds being killed by electrocution.

For raptors, it has been said that this mortality corresponds to ‘contagious’ phenomena, in other words, concentrated in terms of space [15], [16], although there is little proof of these phenomena [11], [16]. There is a need to understand the spatial distribution of the mortality across extensive areas, in order to develop strategies that can reduce this phenomenon.

In Spain, in recent years, more than 25 million Euros have been spent to reduce the impact of power lines on raptors [17]–[19]. Those measures where mainly directed toward the recovery of the Spanish imperial eagle (Aquila adalberti). Where corrections have been implemented, the recovery of the Spanish imperial eagle has been remarkable, [7], [17], although this might be due to other factors [2]. However, other species highly susceptible to electrocution, such as the threatened Bonelli's eagles (Aquila fasciata) or golden eagles (Aquila chrysaetos), have slightly decreased in population or have maintained their numbers [20], [21]. In addition, the long-term efficiency of these measures is unknown [12]. Therefore, it is important to determine the efficiency of these measures in a pylon-per-pylon approach.

This study addresses the abovementioned research gap by focusing on mitigation measures, environmental, spatial and structural factors that influence the electrocution of specific bird groups, such as large eagles, over a large geographic area. Our goals are to describe raptor mortality caused by electrocution in a large area in central Spain, being the objectives of this study: 1) determine whether mortality events are distributed evenly over all power lines or are concentrated around certain lines; 2) analyse structural and environmental characteristics to determine what influences raptor electrocution rates, particularly for Aquila genus (henceforth eagles); and 3) examine the efficiency of mitigation measures implemented in this area prior to this study.

Materials and Methods

Power line survey

Between October 2004 and December 2009, 333 power lines (12–66 kV) and 6304 pylons were surveyed on foot, representing 10% of all power line length in this area (Figure 1). Pylons were only surveyed once and the power lines were chosen according to their potential to impact local birds of prey, following the criteria of [10], [11]. We selected preferably power lines with pin-type insulators, pylons with phases over the crossarm or with short strings of insulators (see Supporting Information Figures S1, S2, S3, S4 and S5). Moreover, lines were selected that passed through an environment with open natural vegetation or a scrubland-crop interface.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Map of the study area with the pylons surveyed (a), raptor (b) and eagle (c) mortality per pylon.

For mortality, black circles indicate 1 or 2 carcasses and grey squares 3 or more.

https://doi.org/10.1371/journal.pone.0028212.g001

Almost all power lines were completely surveyed. In some cases, pylons located less than 300 m from human settlements were not surveyed according to [13] (but see [26]–[28]). Electrocuted birds were collected in a 25 m radius around each pylon [11] and identified by species. A record was made for each pylon that included its characteristics and location (UTM) (Table 1) according to methods used in previous studies [10], [11], [13], [29]. As most of the pylon crossarms were made of metal (n  = 6231) and only a few of concrete (n  = 73) this variable was excluded from analysis. All the pylons were ground wired in the study area.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristic variables and pylon structure and line recorded for each pylon and line reviewed (for more details see Supporting information).

https://doi.org/10.1371/journal.pone.0028212.t001

Based on previous studies in nearby areas [13], the civil service responsible for oversight has implemented mitigation measures such as the use of insulating rubber or silicone cable covers, changing ceramic insulators to glass ones and string of insulator extensions [19]. The insulation of cables, both with rubber and silicone covers, consisted of the installation of wire covers approximately 1 m to either side of the pylon, in addition to on the strained wire when it was present. This avoids any bird perching on the crossarm and coming into contact with the cable, causing a difference in the electrical charge. Changing ceramic insulators to glass ones consisted of changing the type of insulators, so that greater distances are usually achieved between the crossarm and the wire. Finally, the string of insulator extension consisted in the installation of a non-conductive steel extension used to elongate the length of the string of insulators in anchor or special pylons. Most of these measures were implemented in the 1990s [19]. Therefore, in order to try to assess their efficiency after over 10 years, these measures will be considered another factor to be included in the analyses.

Environmental variables

To obtain environmental variables surrounding each pylon, the UTM location of each was recorded on a digital map scaled at 1∶50.000 using ArcView 3.1 GIS. Vegetation maps were obtained from the Spanish Ministry of the Environment [30], [31]. Environmental variables were chosen according to whether they influenced electrocution rates (Table 2) based on methods used in previous studies [11], [32]. Topography can affect mortality, if we take into account the fact that raptors prefer exposed high perches. Thus, it likely that the pylons that stand out most on the land will cause the highest mortality rates [4], [11].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Description of chosen environmental variables.

https://doi.org/10.1371/journal.pone.0028212.t002

The “distance” variables were obtained from the Nearest Feature V.3.8. extension for ArcView 3.1 [33]. We considered distances to three elements (roads, paths and urban settlements). The remaining parameters were assessed within a 25 m area surrounding each pylon.

Prey abundance (wild rabbit and red-legged partridge) were characterised according to [23]. Thus, five abundance categories were used, based on direct observations and the abundance of tracks observed when looking for carcasses in a 25-m radius around each pole (see Table 2).

After preliminary analyses, we chose to use categorical distance variables (distance to road, paths and human settlements) instead of continuous variables because much of the data were grouped according to certain distances. The three variables that involved human influence (distance to roads, paths, and inhabited places) were re-coded into two factors: short distance (<1500 m for roads, <1000 m for paths, <4000 m for settlements) or long distance (values greater than those listed above for each variable).

Results

Mortality rate and distribution

A total of 952 electrocuted raptors were found, representing 14 different species. Of these, 929 (97.6%) were identified. We found that 16.6% (n  = 158) of all dead birds belonged to the genus Aquila (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Number of dead specimens by species and their corresponding threat level (Madroño et al. 2004).

https://doi.org/10.1371/journal.pone.0028212.t003

Raptor mortality was caused by 610 pylons (10% of total). For these, the average number of electrocuted birds was 1.5±1 (1–7 range of electrocuted birds per pylon, n  = 610). For eagles, the average was 1.2±0.2 (1–6 range of electrocuted birds per pylon, n  = 133). Incidences of mortality for raptors were more homogenously distributed compared to eagles (Figure 1).

The K_mm(d) function shows whether the processes (mortality) tend to be spatially grouped (points above the 95% confidence interval) or occur randomly (within the confidence interval). The size of the areas in which mortality tends to cluster is defined approximately by the value of d when the values of the K_mm(d) function are above the 95% confidence interval (Figure 2). Thus, in several areas, mortality does not follow a random pattern, but rather is spatially concentrated within those areas. In the case of eagles, (continuous line) incidences of mortality are concentrated within relatively small areas (7–10 km). For raptors in general, this phenomenon occurs in larger areas (20–40 km). This deviation compared to standard distribution reveals a ‘contagious’ mortality pattern for both groups.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Values of K_mm(d) for eagles and whole raptor species.

On the x-axis distance (d) where function K_mm(d) takes a value. If K_mm(d) adopts values above the random distribution it implies mortality clustering phenomena within a area described through d.

https://doi.org/10.1371/journal.pone.0028212.g002

Factors related to electrocution

Raptor mortality caused by electrocution is a result of both environmental characteristics and the structure of the power lines (Table 4). When considering environmental factors, the number of electrocuted raptors increased as the number of prey animals increased (0.620±sd  = 0.060, z  = 10.384, p <0.001). A similar trend emerged for increasing bush cover (0.014±0.002; z  = 6.091; p <0.0001). Electrocutions decreased when distance to roads was above 1500 m (−0.340±0.112, z  = −3.027, p  = 0.002). Results for structural characteristics of the power lines indicated that electrocution rates increased when the number of insulators per phase decreased (-0.266±0.06, z  = −4.150, p <0.001). Electrocution rates also increased as the number of phases above the crossarm grew (0.238±0.070, z  = 3.418, p <0.001). When considering pylon function, results indicated that there was a significant difference in the number of electrocutions among the three types. Anchor-type pylons caused the highest number of electrocutions (1.446±0.171, z  = 8.477, p <0.001), followed by the special-type pylons (0.601±0.167, z  = 3.605, p <0.001). For crossarm more electrocutions were caused by pylons with cross-shaped and flat crossarms compared to other models (−0.641±0.191, p <0.001).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Model selection for raptor and eagle mortality rates.

https://doi.org/10.1371/journal.pone.0028212.t004

The best-fitting model describing eagle mortality included structural and environment variables (Table 4). Eagle mortality rates differed among all pylon types. Anchor-type pylons caused the largest number of electrocutions (2.508±0.280, z  = 8.953, p <0.001) followed by special-type pylons (1.585±0.339, z  = 4.681, p <0.001). Mortality rates also increased depending on the number of insulators present (−0.515±0.110, z  = −4.659; p <0.0001), the slope near the pylons (0.044±0.021, z  = 2.029, p  = 0.043) and prey abundance (1.074±0.162, z  = 6.635, p <0.0001).

Comparison of mortality between improved and unimproved power lines

Only pylons employing insulation extensions showed a lower raptor mortality rate (−1.195±0.519, z  = −2.303, p  = 0.0212), but not for eagles. No differences in mortality rate compared to uncorrected ones were observed for power lines employing other mitigation measures (Figure 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Electrocution rate of raptors (left column) and eagles (right column).

Carcasses per 100 surveyed pylons for each pylon function (a), prey availability category (b) and type of mitigation measure (c).

https://doi.org/10.1371/journal.pone.0028212.g003

Discussion

These results indicate that power lines cause a large number of deaths among many of the most threatened raptor species in Spain. Mortality rates found in this study are higher than others reported previously in different areas (2.6 eagle and 15.1 raptor carcasses per 100 pylons reviewed) [10], [11], [13], [16], [41], [42], but lower than results obtained in a nearby study area in 1998 (28.2/100 pylons reviewed) [13]. Three non-mutually exclusive hypotheses may explain the differences between these results. First, the types of power lines chosen for this study could potentially be more dangerous to raptors than those studied previously. This may be because, unlike the previous studies, 10% of the lines in the area a priori considered to be the most dangerous were sampled. A second possibility that our study area has a high density of raptors, being an important area for large eagle breeding [7], [20], [21] and in particular as a dispersal zone [19], [43], [44]. Since immature birds are more prone to electrocution [3], [10], [14], the electrocution rate rises when compared with the rates measured in the breeding areas. Finally, as a third possibility, the different carcass disappearance rates can modify the mortality rates found [45], [46]. Given the characteristics of this study in which the pylons were only checked once, in order to obtain a global mortality estimate, the data cannot be corrected for mortality rates by locality. When compared to the results obtained by [13] in a nearby area, differences may stem from the previously mentioned power line correction program [19]. This program has modified several pylons within this area. This hypothesis is supported by the lower electrocution rates of the eagles (2.6 electrocuted eagles per 100 pylons surveyed in our study compared to 3.3/100 pylons reviewed in [13]), which are prone to electrocution [42]. This is especially striking as Spanish imperial eagles and golden eagles have respectively increased [3], [47] and maintained their populations within the study area [21]. Meanwhile, Bonelli's eagle has suffered a slight decrease [20]. Thus, source-sink dynamics might be playing an important role, as has been documented in other species [48].

Analysis of the distribution of raptor electrocutions indicates that mortality is not constant across the entire study area. Mortality values, obtained using the K_mm(d) function, suggests that incidences of electrocution tend to be concentrated in specific areas. Thus, raptor mortality rates are not evenly distributed throughout the study area, instead occurring within broad areas (with 20–40 km radius). This fact supports the idea that electrocutions are likely to be spatially related [11], [49]. This clustering effect might be due to concentrations of prey [32], as raptors' main prey species tend to gather in certain places [22], which may create areas with higher raptor densities. As a result, we should be careful to avoid making broad generalisations about bird mortality rates over large geographic areas [50].

For eagles, mortality rates occurred within smaller areas than those considered for raptors (with 7–10 km radius). A possible explanation is that, for eagles, deaths occur along a few closely-spaced power lines, possibly stemming from the fact that power lines in close proximity to each other are likely to be very similar [10], [11]. This cluster effect occurring for eagle electrocutions has been documented previously [3], [11], but this study expanded upon past research to determine that these deaths are linked to factors that make their deaths likely (e.g. habitat, slope, prey abundance, technical design) and which are particularly correlated. This fact is important when attempting to improve existing power lines to prevent future electrocutions.

For the two groups, both the pylon function and number of insulators affected mortality [11]. Moreover, in the case of the raptors, the number of phases above the crossarm affected mortality [14]. Furthermore, the design of the crossarm emerged as one of the most important factors affecting raptor electrocutions. In the case of the eagles, no type of crossarm was discovered to cause differences in mortality rates. This may stem from the larger wingspan of these birds, which is thought to contribute to their electrocution [42] and which facilitates their electrocution, irrespective of the design of the crossarm. Other structural characteristics of the pylons seem less important.

For both, eagles and raptors in general, prey abundance, among other factors, determined mortality rate, perhaps thus contributing to the grouping of deaths. For raptors in general, vegetation coverage also increased the chances of electrocution [10], [41], possibly because vegetation structure may affect prey availability and the predator foraging performance [5], [51]. Similarly, dominant pylons (i.e. tall structures in open areas) have been also shown to boost mortality [5]. In areas with greater human habitation and road banks, higher prey abundance was observed [22]. This suggests that more electrocutions occur in more humanized landscapes, despite the generally observed pattern [52], [53]. However, according our results we cannot state categorically the effect on different distance ranges, as suggested by other authors [52]. In addition, unlike raptors, the mortality rates for eagles increased with slope, possibly due to the habit of hunting from perches [54]. Previous studies have demonstrated that pylons located in dominant sites, surrounded by high slopes tend to produce higher electrocution rates [4], [11].

Many of the lines examined in the study area have similar designs, especially in the construction of the crossarm, likely related to standardisation by the power supplier. If we consider geographically both issues influencing electrocution rates, abundance of prey and pylon design, we are able to obtain the locations of highest mortality for raptors. Thus, the design of mortality monitoring programmes should take these factors into account, particularly in the case of eagles.

This study suggests that not all power line mitigation measures implemented have permanent effects in reducing raptor electrocutions. Here, we illustrated that mortality rates are higher in pylons that have only been insulated, in comparison with similar pylons. This may be provoked by the original insulation of the deadliest pylons. In this sense, after more than 10 years, the degradation process of insulation provokes higher electrocution rates compared with non-corrected pylons. Thus, our conclusion is that structural changes are also required [16]. This result contrast with those previously published [17], [10], [55], [56], which may be due to the conditions of the cable insulation carried out in the study area. Structural changes should focus on eliminating phases above the crossarm and increasing the distance between perch sites and wires, both of which influenced mortality for both groups of birds. For eagles, mortality rates were not influenced by any mitigation measure, so the only advisable strategy is the implementation of structural modifications (changing crossarm and increasing the length of the string of insulators) and not only including extensions.

Importantly, results from this study suggests that the insulation of exposed conductors in ground-wired pylons, as it has been developed in this area, is a practice that is inefficient in long-term raptor electrocution rate reduction (see [55], [56] in contrast). This low efficiency rate may in part be due to the time that has elapsed (on average 15 years) since this insulation was installed. Since then, very little or no maintenance has been carried out, despite this being an area with a very harsh climate.

Eagle deaths seemed to concentrate around a small group of power lines that were located near a large rabbit population. As crossarm design did not affect these electrocution incidences (all caused similar mortality rates, but see [10], [11]), when designing monitoring programmes, it would be useful to check all the power lines in the area. Thus, a new priority could be to concentrate mitigation measures on power lines causing the highest numbers of electrocutions and those in the immediate proximity. Our results suggests that mitigation measures must be implemented along the entire line. Other authors suggested, for certain circumstances, a “preferred pylon” approach [11], [49]. However, although more research is needed and solutions must be developed case-by-case, we consider our results might be applicable to any other ground-wired power network.

Supporting Information

Acknowledgments

Elena Crespo, Elena de Prada and Silvia Villaverde helped with carcass identification. We are also thankful to Sandra Agudin, Nuria El Khadir, Roberto Sánchez-Mateos and Juan Francisco Leiva for their cooperation in the fieldwork and to Alfonso San Miguel for his continuous advice. We are thankful to M^a del Mar Celada and Sarah Young for their help with the English translation and Robert N. Lehman, Santi Mañosa and Fabrizio Sergio for their constructive comments to improve this manuscript.