Avian collision threat assessment at ‘ Bhambarwadi Wind Farm Plateau ’ in northern Western Ghats , India

1,3,4,5,6,7,8,9 Ela Foundation, C-9, Bhosale Park, Sahakarnagar-2, Pune, Maharashtra 411009, India 2 Department of Zoology, M.E.S.’ Abasaheb Garware College, Pune, Maharashtra 411004, India 1 pande.satish@gmail.com (corresponding author), 2 adpadhye@gmail.com, 3 pramod.deshpande@gmail.com, 4 ponkshe.aditya@gmail.com, 5 pranav.vet@gmail.com, 6 amit.pawashe@gmail.com, 7 bandapednekar@gmail.com, 8 rohanpandit87@gmail.com, 9 detaman@gmail.com


INTRODUCTION
India is facing shortage of power (Singh 2006) and attempts are being made to address this problem through alternate and renewable energy sources.As a result, there is a rapid increase in the number of wind farms at suitable sites all over the country.As projected by the Ministry of Non-conventional Energy Sources, Government of India, 10% of the installed capacity of power requirement by the year 2012 (24,000MW) will come from renewable energy, of which 50% (12,000MW) is likely to come from wind power (Ghose 2006;Krithivasan 2006).India is the fifth largest producer of wind energy in the world with installed capacity of 10,891MW as in October 2009 (Meisen 2006, updated by Avinash & Timbadiya 2010).Some of the key sites where adequate wind velocities are encountered throughout the year are the plateaus on the Western Ghats (Ghosh 2006), which is identified as a global hotspot of biological diversity (Myers et al. 2000).
The rocky plateaus on the Western Ghats are described as terrestrial habitat islands facing extreme micro-environmental conditions, and even though it is documented that rocky outcrops such as inselbergs, barrens and others support rich and threatened floristic endemicity (Porembski et al. 1998), scant information is available on the ecology of these plateaus (Watve 2003).Considering the above scenario, we undertook a twoyear study to document avian diversity and assess the impact of wind farms at Bhambarwadi Plateau on avian populations.To the best of our knowledge this is the first such study in India.

Study area
The study area is situated on the Bhambarwadi Plateau (0.5km 2 area around 17 0 8'90"N & 73 0 54'96"E; 1053m) on the northern Western Ghats or the Sahyadri Mountains, near Gude-Pachgani Village, Satara District (Image 1).There was a proposal to construct 13 wind turbines in the study area, where ten wind turbines were previously constructed.The Chandoli Wildlife Sanctuary is approximately 5km to the west of the study area.The study area is a high level rocky plateau on the Sahyadris.It is composed of ferricrete duricrust, usually described as laterites, capping underlying basalt summits (Ollier & Sheth 2008).The soil cover ranges from few centimeters to less than one meter.The study area falls in the bio-geographic zone of Western Ghats and the agro-climatic zone is Western Plateau and Hills Region Image 1. Sketch Map Showing Point Count Sites in the Study Area.Region included by site marked from 2 to 3 is plains, from 3 to 5 is hill slope and from 5 to 7 is the plateau.The inner two circles comprise the study area where only point counts were taken.(Rodgers & Panwar 1988).There are three seasons in the region.Summer is from March-May; monsoon from June-September; and winter is from October-February.Humidity ranges from almost 100% during the monsoon and around 45% in summer; the climate is monsoonal and the summer temperature rises up to 38 0 C on a few days and the winter temperature dips to 5 0 C; the average temperature is 24 0 C (Lakshminarayana et al. 2001).Visibility is generally very good except during monsoon when there is a thick cloud cover on the plateau.

Data Collection
The data was collected from July 2008 to June 2010.Fortnightly visits were made during the study period.Data was collected for two years including the three seasons, summer, monsoon and winter during the daylight hours.The dimensions of the wind turbines required for further analysis, such as height of the wind turbines; length, width, pitch angle, thickness of the rotor blades; maximum cord width and dimensions of the nestle; were obtained from the wind farm company.The dynamic data regarding wind turbine revolutions per minute (rpm), direction of wind and wind velocity for each visit was obtained from the computerized system installed in the field office of the wind farm (only average values are provided by the wind farm company for the above three parameters for confidential reasons).Actual bird and mammal species found dead in the study area, due to collision with wind turbines, wind mast and overhead power lines, were also recorded.All observations were made during the entire study period by four trained observers.
Point counts were taken for the recording of avian activity in the study area.Point counts were made from the view point of an external observer with a 50m radius around the wind turbines.Each count lasted for the duration of 20 minutes.We recorded the following parameters: (i) avian species, (ii) number of individuals of each species (abundance) flying in the study area, (iii) whether the bird was flying in the risk zone, below it or above it, (iv) total flight time of each species in minutes (flight activity) and (v) the flight activity of birds in the risk zone (risk activity).Risk zone is the region between the lowest and top most points swept by the rotor blades or the aerial height band swept by the rotor blades (Image 2).The band span was 10-100 m above the ground level.
Known length of bird species (from the tip of the beak to the tip of the tail in meters) and known wing chord (from the wrist to the longest primary feather in flexion of the wing in meters) were taken from Ali & Ripley (1969).The standard multiplier (the ratio of wing span to wing chord for that species) was taken from Fergusson-Lees & Christie (2005).The average flight speed for most of the species was taken from Bruderer & Boldt (2001) and Alerstam et al. (2007).For some species, bird length and wing chord were measured from rolled bird specimen in the collection of the Zoological Survey of India, Western Regional Center, Akurdi, Pune.The wing span in meters was calculated by multiplying the wing chord and standard multiplier.For some species, the average flight speed in meter/second and type of flight (0 = Fl.-Flapping, 1 = Gl.-Gliding) were recorded in the field.These parameters were used for calculation of hypothetical collision probability of all the bird species flying in the risk zone.

Calculation of collision risk
The assessment of the collision risk was done by the suitable modification of the Band Model (Anonymous 2000;Band et al. 2007), after taking into consideration the actual wind farm and rotor blade parameters in the study area.
Collision index for a species (CI) = Number of birds flying through rotor x Probability of bird flying through rotor being hit.Therefore, where, n= number of wind turbines; Vr (the combined volume swept out by the wind farm rotors [where, N is the number of wind turbines (N=10), d is the depth of the rotor back to front, l is the length of the bird, π is Pythagoras constant (3.14159) and R is radius of rotor (26m)]; Vw [flight risk volume which is the area of the wind farm (5x10 6 sq.m) multiplied by the risk height of the turbines 90m) = 45000000m 3 ].p(r) is the probability p of collision for a bird at a radius r from hub.
where, α = v/rΩ; β = aspect ratio of bird i.e. l / w; b = number of blades in rotor; Ω = angular velocity of rotor (radians/ sec); v = velocity of bird through rotor; K = 0 for onedimensional model (rotor with no zero chord width); K = 1 for three-dimensional model (rotor with real chord width); c = chord width of blade; γ = pitch angle of blade; w = wingspan of bird; F = 1 for a bird with flapping wings (no dependence on ϕ); F = (2/π) for a gliding bird; l = length of bird; r = radius of point of passage of bird.
Yearly average collision rate = Sum of collision Index for each species / number of turbines.
Several approximations and assumptions were involved in the study.The bird was assumed to be of simple cruciform shape, with the wings at the halfway point between nose and tail.The turbine blade is assumed to have a width and a pitch angle (relative to the plane of the turbine), but to have no thickness.It was also assumed that no avoiding action was taken by the bird.Hence, the calculated collision risks should be held as an indication of the risk (±10%).It was also assumed that bird flight velocity is likely to be the same relative to the ground, both upwind and downwind.We have separately calculated collision indices for upwind and downwind flight speeds as suggested by Band et al. (2007).

A. Point Count
In all, 89 species were recorded during point count, of which 27 were recorded in the risk area.Seasonal flight activity (in minutes) of each species in the study area during Monsoon, winter and summer, irrespective of the number of individuals was recorded.The maximum flight activity of 191 minutes was presented by Redrumped Swallow Hirundo daurica.Red-vented Bulbul Pycnonotus cafer -162 minutes, Wire-tailed Swallow Hirundo smithii -123 minutes and Malabar Lark Galerida malabarica -102 minutes were the other species who showed total flight activity of over 100 minutes (Table 1).
Total avian flight activity, as recorded in the study area, irrespective of the number of species and number of individuals, was 1604 minutes; while total seasonal flight activity was maximum during summer (645 minutes) followed by monsoon (548 minutes).Flight activity was the least in winter which was 411 minutes.Out of 1604 minutes of the total avian flight activity, flight activity in the risk area was 1067 minutes.Seasonal flight activity in the risk area showed the same trend with maximum flight activity in the risk area during summer (449 minutes) followed by monsoon (324 minutes).Flight activity in the risk area was the least in winter which was 294 minutes (Fig. 1).
Analysis of monthwise total avian flight activity and flight activity in the risk area during the entire study period depicted that overall pattern of avian flight activity and the flight activity in risk area corresponded to each other (Fig. 2).Activity was high in July 2008, when there was minimal disturbance in the study area.It peaked again in March 2009, mainly due to a forest    We consider this as the species displacement effect.

b. Bird Collision Indices
The average annual wind velocity was 7.6m/s; the wind direction was variable with average of 261 degrees with respect to North.The average windmill rotor RPM was 23.6.The lowest RPM were seen during the March of each year and the peak was seen in December and July.The monthly variation is shown in Figure 3.
Assuming that the birds do not take any preventive action so as to avoid collision with the rotor blades, the yearly average collision rate was 1.9 birds per turbine.Considering the presence of 13 wind turbines in the study area the total collision rate is 24.9 birds annually.The biometric parameters used for the calculation of hypothetical collision probability of all 27 bird species flying in the risk area are given in Table 2.The hypothetical probability of bird collision and the collision index indicating probable bird hits per year for all 27 bird species is given in Table 3. Season wise bird collision assessment studies revealed that the maximum collision risk was in winter while it was the minimum in monsoon (Fig. 4).Amongst all the species, raptors were at the maximum collision risk.Season wise collision risks for each species is given in Table 4.
During the study period, 19 birds and mammals were found dead due to collision with the rotor blades (n=10) or electrocution (n=9) due to contact with overhead transmission lines or transformers.Asian Palm Civets Paradoxurus hermaphroditus were found dead in the transformers built for transmitting windmill power to the base stations.Maximum collisions of raptors were seen during the monsoon months.Swallows and martins were found dead in post monsoon period.In addition, we also noticed that two Black Kites Milvus migrans and one Changeable Hawk Eagle Spizaetus cirrhatus collided with wind masts.Actual bird and mammal species found dead in the study area and their respective numbers are listed in Table 5.

DISCUSSION
The study area assumes special significance because it lies in the Western Ghats, which are listed as one of the global biodiversity hotspots (Myers et al. 2000).Table 2. Biometric parameters (length of bird from the tip of the beak to the tip of the tail in meters, known wing chord measured from the wrist to the longest primary feather in flexion of the wing in meters, standard multiplier, calculated Wing Span by multiplying the wing chord and Multi-multiplier in meters, average flight speed in m/sec, and type of flight (0 = Flapping, 1 = Gliding) used for calculation of hypothetical collision probability of all the 27 bird species flying in the risk zone swept by the rotor blades.
Being situated at higher altitudes, these areas receive high and year round wind velocities required for wind power generation; hence these plateaus are increasingly utilized for wind farm erections.However, these plateaus with unique geographical features, are poorly studied (Lakshminarayana 2001;Watve 2003).
In the current study, we enlisted the avian diversity and species that are at risk due to collision with turbines, transformers, wind-masts and at risk of electrocution due to power lines, for the first time for this unique biogeographical region.Albeit unintentional, birds die as Table 3.The Hypothetical Probability of Bird Collision (Hypo.Prob.) and the Collision Index indicating probable bird hits per year are given for all the 27 bird species that were recorded flying in the risk zone swept by the rotor blades in the study area.The species are arranged in the decreasing order of their collision index.It is assumed that the birds do not take any preventive action so as to avoid collision with the rotor blades.In fact they actually do so and the actual hits are expected to be less than the calculated index.a result of collisions with wind turbines (Banks 1979;Drewitt & Langston 2008;Rothery et al. 2009;Martin 2011), collisions with power lines (Manville 2005) and subsequent electrocutions can threaten survival of certain avian populations such as juveniles (Schaub & Pradel 2004), migrants (Christensen et al. 2004;Kahlert et al. 2004) or endangered species (ESKOM 2008;Shaw et al. 2010).
Our observation of reduction in the avian activity status in the study area with progression of wind farm erection activity is in accordance with similar bird displacement effect of wind farms reported by others (Anderson et al. 1999).Even after the wind turbines erection and other related human activities had ceased after commissioning of the wind farms, the avian displacement effect was conspicuous.Though the footprint of an individual wind turbine is small, the associated infrastructure development activities like road construction, establishment of power substations, and laying of power cables cause an effectively greater level of habitat destruction and modification, which could explain this displacement effect.
We did not observe the presence of an avian winter migratory corridor in the study area.Our study showed only one seasonal influx of Common Kestrels in winter, in contrast to well known avian migratory movements along coastal areas (Ali & Ripley 1969;Pande et al. 2003;Fox et al. 2006) and a few locations in the northern Western Ghats (Padhye et al. 2007), that are potential wind farm sites.
We recorded 27 bird species flying in the risk zone in the study area out of which 11 were raptors.Out of the 12 birds (belonging to seven species) that were found dead, five were raptors belonging to three species.This indicates that raptors are at a higher risk of collision as compared to other species.Moreover, the seasonal variation in collision index was highest in raptors.The overall risk of collision for all species, including raptors, was highest in winter.Such high risk of raptor collisions with turbine rotors and overhead power lines has also been reported by Madders & Whitfield (2006).Further, out of five Indian avian endemic species observed in the study area, Malabar Crested Lark Galerida malabarica (endemic to the Western Ghats) was recorded in the risk zone.
In addition to the risk zones created by the turbines, the wind masts are supported by very thin steel wires that are not visible from a distance, which lead to avian collisions and subsequent mortality.We strongly recommend that the supporting wires of the wind mast and the mast itself should be marked in bright colours or flags to make the wires and the mast prominently visible from a distance.
Modeling collision risk can help to determine the approximate level of mortality likely to result from particular developments such as wind farms, which enables us to explore the consequences for local and regional populations (Madders & Whitfield 2006).There is a mismatch between theoretical and actual collision risks due to several reasons (Richardson 2000).The theoretical risk can be an overestimate because the birds in practice take active collision avoidance measures.On the other hand, the actual number of birds found dead in the field can be underestimated because these birds can be scavenged before they are recorded by investigators.It is agreed that the reliability of collision models is limited by difficulties in gathering appropriate field data and by the large number of assumptions necessary during the modeling process, notably for the levels of collision avoidance (Madders & Whitfield 2006).Higher wind velocities and subsequent higher RPM of the turbine blades were recorded in July and December that may lead to a higher risk, when the visibility in the study area is low due to clouds and fog.However, the overall flight activity may also be underestimated during this period as a consequence of poor visibility.
We found highest number of dead birds during monsoon (9 out of 12 birds), and this could be due to the carcasses being left for longer time in monsoon due to absence of scavengers in these months, when the weather conditions are harsh in the study area.Carcasses due to collision are more likely to be scavenged immediately in winter and summer months.Further, we would also like to mention that we have collected the field data twice in a month which itself can be a reason of the underestimate of the dead birds; though the efforts were significant considering the remoteness of the study area.Therefore, it is evident that the search for dead birds alone may be inadequate to assess the true effects of wind farms on birds.
There are few published studies describing the activity budgets of upland bird species and potentially influential factors globally (Collopy & Edwards 1989) and Western Ghats is not an exception.Hence our study assumes special importance.The number of hours per day that birds are potentially active, and the influence of factors such as weather, time of year as well as the breeding status are poorly understood (Madders & Whitfield 2006).
It is suggested that due to their unique nature the plateaus of Western Ghats need protection by limiting human activities.None the less, many of the plateaus adjoining the study area are mushrooming with wind farms and associated infrastructure development activities.Such activities can lead to immense loss of local biodiversity (Lakshminarayana et al. 2001).
It is accepted that hydropower and thermal power generation by burning of fossil fuels have their own environmental and biological risks (Huntley et al. 2006), so also, it is increasingly recognized that 'Green Energy' providing wind farms do impact wildlife and environment (Drewitt & Langston 2006).In view of the above avifaunal risks, we feel that wind farm erections in strategic locations such as biodiversity hotspots should be subject to prior strategic environmental assessments (SEA) as well as environmental impact assessment (EIA) studies.The need for such SEA's and EIA's have been emphasized elsewhere (Fox et al. 2006).There is a need for 'site-based approach' for detailed biodiversity assessment studies of the plateaus of Western Ghats that are potential wind farm locations, so as to effectively enforce conservation measures during erection of wind farms in future.
Schematic drawing of a wind turbine showing the risk zone

Figure 1 .Flight
Figure 1.Seasonal flight activity, as recorded in the entire study area on the plateau as well as in the risk area, irrespective of the number of species and number of individuals (Monsoon: June-September; Winter: October-February; Summer: March-May).

Figure 2 .
Figure 2. Month wise total avian flight activity in bird minutes (y-axis) and flight activity in the Risk Area swept by the rotor blades, in the study area on the plateau during the entire study period (July2008-June2010).
cause in the study area when there was a sudden increase in the activity of Black Drongos, Dicrurus macrocercus.It increased again in June 2009 (early monsoon) and then in November-December 2009 (during the winter), when there was an influx of the migratory Common Kestrels Falco tinnunculus.However, by June 2010, there was a definite reduction in overall avian activity in the study area as compared to activity in July 2008, even though the wind turbine erection and road construction activities had ceased and human presence was minimized to maintenance work.

Figure 3 .
Figure 3.The RPM (revolutions per minute) on the day of the study for all the wind turbines (Y-axis) in the study area are plotted for each month during the entire study period (X-axis).(Data from Enercon India Ltd.)

Figure 4 .
Figure 4. Box plot showing seasonal variation in the collision Indices in monsoon, summer and winter irrespective of the species.The minimum and maximum values are shown by crossbars on the vertical line.The frequency distribution within 75 percentile is shown by dark grey box and within 25 percentile is shown by light grey box.The median is indicated by the confluence of the boxes.