The New Zealand falcon and wind farms: a risk assessment framework

Abstract

Internationally, birds of prey are often reported as being relatively prone to collision with wind turbines in comparison to other groups of birds. However, as yet it is unclear to what extent New Zealand's only endemic bird of prey, the New Zealand falcon (Falco novaeseelandiae), is at risk. In this paper we summarise the potential for wind farms to impact New Zealand falcon, evaluate the efficacy of a range of risk assessment and post-consent monitoring practices, and present options for mitigating and/or offsetting any residual effects. We conclude that the lack of knowledge on the effects of wind farms on New Zealand falcon is the result of inconsistency in the assessment methods thus far employed and the absence of a coordinated approach to monitoring methods and the dissemination of results. To remedy this we present a risk assessment framework that, if adopted, will provide the information necessary to ensure alternative energy targets can be met without compromising the conservation of this threatened species.

Keywords:

• risk assessment
• AEE
• wind farm
• collision mortality
• mitigation
• New Zealand falcon
• Falco novaeseelandiae

Introduction

Greenhouse gases are widely acknowledged as the primary cause of anthropogenically driven climate change (Huntley et al. 2006). The development of renewable energy sources has a significant role to play in reducing the emission of these gases. As part of its response to climate change, the New Zealand Government is aiming for 90% of New Zealand's electricity to be produced from renewable sources by 2025 (NZES 2007). Wind electricity generation has the potential to contribute significantly to this target, with current trends tracking towards wind energy providing 20% of New Zealand's electricity by 2030 (NZWEA 2012).

Notwithstanding the environmental benefits of wind energy generation, negative ecological impacts can result in some circumstances. For example, it is now widely accepted that under some situations wind turbines can pose a significant collision risk to birds (De Lucas et al. 2007). Birds of prey have been noted to be particularly prone to collision in comparison to other groups of birds (Madders & Whitfield 2006). These studies raise the question of whether wind farms pose a risk to the New Zealand falcon (Falco novaeseelandiae) (falcon) (Seaton 2007a; Powlesland 2009a), a threatened species (Miskelly et al. 2008) that has the potential to be present at wind farms over much of New Zealand.

In New Zealand, developments that require resource consent under the Resource Management Act 1991 (RMA) call for applicants to illustrate and address the risks of their proposal to the environment. However, where novel and rapidly advancing technological developments are proposed, the knowledge required to adequately assess the potential ecological effects often lags behind the pace at which developments are proposed (Garvin et al. 2011). An example is that of the falcon and wind farm development in New Zealand, where a lack of knowledge on the behaviour of falcons around wind turbines has led to uncertainty in the level of potential effect (Powlesland 2009a). This paper evaluates: 1) the efficacy of a range of risk assessment methods to appraise the risk of a proposed wind farm to falcons; and 2) the ability of different post-construction monitoring approaches to inform future proposals. These results are used to develop a risk assessment framework (a stepwise methodology to establish the effect of a proposed wind farm) to enable more consistent and objective decision making. The paper ends by discussing options available to avoid, remedy, mitigate and/or offset any adverse effects.

A summary of the potential effects of wind farms on New Zealand falcon

The effects of wind farms on birds outside of New Zealand have been reviewed on numerous occasions (e.g. Drewitt & Langston 2006; De Lucas et al. 2007). To determine the potential effects of wind farms on New Zealand's birds, Powlesland (2009a, b) summarised this literature and identified species of concern, of which the falcon was one. It is not the purpose of this paper to repeat this review. Rather, a summary discussion is provided as a background context to the risk assessment framework presented later in this paper. The following summary discussion is partitioned into the three threat categories identified by Powlesland (2009b): habitat loss; displacement/disturbance; and collision mortality.

Habitat loss

The construction of a wind farm inevitably results in the loss of vegetation to turbine pads, substations, roads and other structures associated with its development. The scale of loss is dependent on the size of the development but is considered small on a per turbine basis (Drewitt & Langston 2006). At these scales, we agree with Powlesland (2009b) that it is unlikely that habitat loss associated with wind farm operations will be significant enough to cause a reduction in the occupied range of falcon.

Construction of a wind farm also results in modification or habitat loss for species forming the prey of falcons, largely small passerines (Fox 1977; Barea et al. 1999; Seaton et al. 2008). However, falcons are adaptable hunters selecting prey in proportion to their availability (Barea et al. 1999; Seaton et al. 2008); thus, the loss of habitat for prey species at the small scales generally associated with wind farm development is unlikely to render an area unsuitable as foraging habitat.

Loss of nest sites (e.g. emergent trees or rock outcrops) could affect falcons at local scales. However, because falcons commonly select nest sites at mid slope or in gullies (Fox 1977; Barea et al. 1997) and habitat loss is generally confined to ridges and plateaus where wind farm construction occurs, the effects of habitat loss on falcon nesting habitat are likely to be absent or very low.

Displacement/disturbance

Overseas literature suggests that raptors may be displaced if disturbance prevents the use of nest sites and/or foraging areas. Although most of this research suggests that disturbance to raptors at wind farms is generally low (Madders & Whitfield 2006), some suggests that displacement effects can occur in some circumstances (Pierce-Higgins et al. 2009; Garvin et al. 2011). Powlesland (2009b) suggested that displacement of falcons in New Zealand might be minor because of their ability to temporarily occur at some times of year in rural and urban environments that contain a variety of artificial man-made structures. Our own observations of falcons in urban environments, their bold nature and their ability to coexist alongside anthropogenic structures, concur with this view. However, additional investigation is needed to confirm whether this holds true for wind farms.

Powlesland (2009b) also stated that if construction of a wind farm occurred near a nest then this could lead to nest failure or abandonment, including of traditional nesting areas, and this could be significant for local populations given the low densities at which falcon populations exist.

Collision mortality

Collision mortality may occur if a falcon collides with wind turbines or other associated structures (e.g. power lines; Powlesland 2009b). The high risk of collision reported for raptors relative to other groups of birds overseas (Madders & Whitfield 2006) led Powlesland (2009b) to consider the falcon at potential risk of collision at wind farms. These collisions include species that share behavioural and morphological traits with the New Zealand falcon, such as European sparrowhawk (Accipiter nisus), European goshawk (A. gentilis), prairie falcon (F. mexicanus), lesser kestrel (F. naumanni), American kestrel (F. sparverius), European hobby (F. subbuteo), European merlin (F. columbarius) and peregrine falcon (F. peregrinus) (Kingsley & Whittam 2005). Powlesland (2009a) also emphasised that juvenile raptors, through their naivety and poor flying skills, may be particularly prone to colliding with turbine blades. Of the potential risks posed by wind farms, collision with turbines is clearly the primary concern.

Assessment and monitoring methods currently employed in New Zealand

Background

Most of the environmental risk assessment work on avifauna undertaken to date has involved pre-development risk assessment as part of the resource consent application process required under the RMA. Several countries overseas have developed guidelines or standards for assessing the effects of wind energy developments on avifauna (e.g. Australia [AUSWEA 2005], USA [Strickland et al. 2011]). However, currently there are no recognised standards for assessing the effects of wind farms to avifauna in New Zealand. Contrasting with the structured stepwise approach of some international guidelines, the lack of specific policy or detailed guidance in New Zealand has resulted in a varying approach to addressing effects of wind farms to avifauna and falcon.

Currently, only a few wind farm proposals in New Zealand have attempted to quantify effects to avifauna during the resource consent application process (e.g. Taharoa wind farm and Hauāuru mā raki wind farm). Continuing this trend, to date only one proposed wind farm has attempted to quantitatively assess the effects to falcons pre-consent (Golder Associates 2012) and few have been required to conduct the robust construction and operational monitoring necessary to determine whether there is an effect or not post-construction. The information necessary to understand risk is further limited by the confidential nature of many post-consent monitoring reports.

Evaluation of current pre-construction risk assessment practices

A range of methods has been employed in New Zealand to assess the risk that proposed wind farms pose to falcons. The efficacy of each of these methods to describe risk is evaluated in Table 1 and expanded on below.

Table 1  Methods used for assessing the collision risk of a wind farm to New Zealand falcon pre-consent and pre-construction.

Method	Options	Suitability and constraints for risk assessment
Desktop survey	Search literature and contact organisations collecting falcon distribution data (OSNZ, DOC, Wingspan, RANZ, NZFDS^1)	Useful first step to determine whether the development is within the distributional range of the falcon. However, biased towards observations in areas where there is frequent public access. Describe broad falcon distribution rather than detailed local population abundance. Limited data on breeding activity. Data quality sometimes unreliable.
Onsite habitat suitability assessment	Non-expert assessment	Risks false conclusions of habitat suitability.
	Expert assessment	Provides for confidence in assessment.
	Within wind farm envelope	A measure of the likelihood of falcons breeding within the wind farm envelope and the probability of the presence of non-residents.
	Within and surrounding wind farm envelope	A measure of the likelihood of falcons breeding within and surrounding the wind farm envelope, the probability of the presence of non-residents and the likelihood of the home-ranges of breeding and non-resident falcons overlapping the wind farm envelope.
Presence/absence surveys	Surveys by non-expert	Risk of high non-detection rates. Falcons can be highly cryptic even during the breeding season; surveys by those not experienced in surveying for falcons risk overlooking them.
	Surveys by expert	Provide increased confidence in results.
	Year-round incidental surveys during general bird surveys	The naturally low density of falcons and low detection rates means they can be overlooked during incidental surveys.
	Specific surveys for falcon outside of the breeding season (March–August)	Low detection rates make falcons very difficult to locate outside of the breeding season. Dispersing juveniles may be mistaken for resident birds.
	Surveys during the breeding season (September–February)	Activity can be expected every four hours around an active nest and characteristic calls are made at this time making falcons more easily observed at this time of year.
	Single visit	High probability of non-detection.
	Multiple repeat surveys	Increased probability of detection in survey area.
Productivity assessment	Incidental notes, e.g. record breeding activity only	Limited ability to provide productivity baseline or to assess effects to productivity.
	Detailed description of clutch, brood size and fledging rates	If done before, during and after construction this could contribute to assessing whether the wind farm disturbs breeding or not. However, it is very difficult to prove cause and effect conclusively because a BACI monitoring design is not feasible for such a low-density species.
Flight activity	Incidental observation	Low detection rates mean the sample size will be very low. Highly biased towards locations being frequented by observers or nest site.
	Targeted observation, e.g. at nest site	Low detection rates means the sample size will be very low. Highly biased towards observation stations.
	Tracking of adults using telemetry combined with observations	Allows the proportion of time resident birds spend within the wind farm envelope to be determined. Provides objectively based data for collision risk analysis.
	Tracking of fledglings using telemetry	Allows the proportion of time fledglings spend within the wind farm envelope before they disperse to be determined. Provides objective data for collision risk analysis.
Flight height	Incidental observation	Low detection rates mean the sample size will be very low. Highly biased towards locations being frequented by observers and to observations of low flying birds (typically below RSH^2).
	Targeted observation	Low detection rates mean the sample size will be very low. Highly biased to observation stations and to observations of low flying birds (typically below RSH).
	Tracking adults with telemetry combined with observation	Radio transmitters (and other tracking devices) allow systematic data collection and real-time tracking, reducing observation bias towards low flights and increasing sample size. Provides objectively based data for collision risk analysis. Problems with low sample sizes and accurate height estimation remain, however.
	Tracking fledglings with telemetry combined with observation	As above. Fledglings may differ in behaviour from adults.
Estimating potential collisions	Qualitative assessment	Highly dependent on the level of investigation and expert opinion. Requires consideration of avoidance rates.
	Collision risk modelling	Onsite data for all inputs essential. Allows a more objective estimate of the number and frequency of collisions to be made. Requires consideration of avoidance rates.
Notes: ^1OSNZ=Ornithological Society of New Zealand; DOC=New Zealand Department of Conservation; Wingspan=Wingspan Birds of Prey Trust; RANZ=Raptor Association of New Zealand; NZFDS=New Zealand Falcon Distribution Survey. ^2RSH=rotor swept height.

At a minimum most wind farm developers conduct a desktop search of the literature to assess the potential for the presence of falcons in the area they are proposing to develop a wind farm (R. Seaton and L. Barea, Wingspan, pers. obs. 2012). These generally involve searching for records of falcons within and surrounding the wider wind farm development area. Although a desktop search can be an appropriate first step, the studies from which the data are sourced usually reflect incidental observations rather than targeted surveys and, as such, are unlikely to match the spatial and temporal scales required to adequately inform the likelihood of falcon presence within a proposed site. Consequently, desktop searches for records of falcon presence rarely add information beyond understanding whether a proposal is within the distributional range of the falcon.

Desktop studies may also include the assessment of habitat suitability using aerial photographs. However, because of the very broad habitat use by falcons (see habitat descriptions Fox 1977; Barea 1995; Barea et al. 1997; Stewart & Hyde 2004; Seaton 2007b), habitat suitability assessments using aerial photographs alone only provide for a cursory assessment of habitat and cannot be relied upon to provide a detailed assessment of the likelihood of falcon presence or absence.

When onsite habitat suitability assessments are conducted by an ecologist familiar with falcon habitat, this approach provides increased confidence in the assessment of whether nesting falcons and/or non-resident falcons are likely to be present in the proposed wind farm area. Onsite habitat suitability assessments allow the need for follow-up targeted surveys to be evaluated and resources to be focused into areas of high habitat suitability.

Although falcons can be conspicuous when an observer is in close proximity to an active nest (falcons are very vocal during food passes and when intruders pass close to a nest), their behaviour is often cryptic, rendering their detection by visual means alone low (Fox 1977). Unstructured surveys or those relying on incidental detection during general avifauna surveys, especially during the non-breeding season, have a high probability of failing to record the presence of falcons, despite their presence within the survey area. Greater confidence in an assessment of the likely use of a wind farm development area by breeding falcons is possible when suitable habitat has been identified and ecologists familiar with falcon behaviour and vocalisations conduct multiple, targeted surveys during the breeding season (September to February) employing a range of survey techniques (e.g. Hardey et al. 2006). Recent development and use of call playback methods (Teng 2010) appear useful as a supplementary technique to other survey methods for detecting the presence of falcons, particularly during the breeding season. However, because detection rates have not yet been statistically determined, employing this technique should be limited to reporting presence and not absence.

Pre-construction data on reproductive productivity have provided a baseline from which to assess and monitor the effects of the construction and operation of several wind farms in New Zealand (e.g. Golder Associates 2012). Ideally, to robustly detect an effect, a before–after control–impact (BACI) design should be implemented (Madders & Whitfield 2006; Strickland et al. 2011). However, the low densities at which falcons naturally occur result in sample sizes that are likely to be insufficient to support a BACI design at an individual wind farm. Notwithstanding this, monitoring breeding success prior to wind farm development and again afterwards over multiple sites and years will, if employed in the future, allow for a level of correlative inference regarding effects on falcon productivity.

Quantitative assessments of collision risk need to be informed by data on flight activity, including flight height with respect to the height of wind turbine rotors. However, few wind farm proposals where falcons have been located have quantitatively assessed falcon flight activity or flight heights. In the absence of such data, risk assessments are limited to qualitative statements that can be prone to subjective bias. Bias should also be considered when designing quantitative risk assessment studies. For example, studies aimed at providing a measure of flight activity within the wind farm envelope (wind farm development area—including wind turbines, roads, transmission lines and all associated infrastructure) that are based only on recording incidental observations—often conducted during the course of other field work—are likely to provide few data due to low detection rates and are spatially biased towards the location of observers. Radio telemetry has been used in New Zealand to quantify flight activity and map home ranges relative to the proposed wind farm design (e.g. Golder Associates 2012). Despite such studies having the potential to provide quantitative data to inform estimates of collision risk to date only one wind farm proposal in New Zealand has included a quantitative estimate of collision risk in their resource consent application (Golder Associates 2012). Until the collision risk to falcons is better understood, telemetry studies offer a more robust method for spatially describing flight activity. Scenarios where telemetry should be considered include those where falcons are detected nesting within or in close proximity to a wind farm envelope.

The flight behaviour of falcons under a range of weather conditions and how this might affect collision risk is not well understood. Ideally, flight height assessments should consider error associated with visual estimation of height and its variation over different height ranges (Band et al. 2007).

Collision risk modelling (CRM) is one quantitative approach that has been used in New Zealand to estimate the number of falcons that may suffer collision mortality as a result of a wind farm proposal. CRM is a relatively coarse tool strongly influenced by the rate at which individuals avoid turbines (Chamberlain et al. 2006; Band et al. 2007). As such, the results of CRM should be used to guide decisions regarding the appropriateness of the consent application or what a suitable level of mitigation or biodiversity offsets might be, rather than to attempt to precisely predict effect. Collision models that account for statistical and biological variation allow for a wider range of potential outcomes to be objectively considered. With an understanding of its appropriate use, CRM can be a useful tool when limited information is available to adequately inform collision risk (Madders & Whitfield 2006; Strickland et al. 2011), as is the case for falcons at New Zealand wind farms.

New Zealand falcon risk assessment framework

Based on the evaluations above, we present a risk assessment framework for consideration when assessing the effects of a wind farm development on falcons (Fig. 1). This includes feedback mechanisms suitable to address uncertainty in lieu of more detailed post-construction information being available. The framework is consistent with international approaches to risk assessment, adopting a staged adaptive management approach to assessment and monitoring with decisions regarding further effort based on the results of previous steps.

Figure 1  A framework suitable for assessing the risk of a proposed wind farm to New Zealand falcon.

Pre-consent assessment of effects

Stage one involves a desktop study of published literature to identify whether a proposed development lies within the falcon's distributional range. Stage two builds on a positive stage one result with an analysis of potential nesting habitat both within the wind farm and beyond the wind farm envelope. A preliminary assessment of habitat suitability might use aerial photographs, but onsite assessments of habitat are usually required to rule out potential nesting habitat entirely. Individuals that are not part of the breeding population are likely to range widely, as are juveniles dispersing from the nest (Holland & McCutcheson 2007; Seaton et al. 2009a; Thomas et al. 2010). Accordingly, stages two to five of the framework reflect the potential for non-resident birds (i.e. adults that occasionally visit the site but nest elsewhere and dispersing juveniles) to use the site and require consideration of risk to these individuals.

Stage three flows from stage two when suitable habitat is located within or surrounding the wind farm envelope. When this occurs, an ecologist experienced with falcon ecology and breeding behaviour conducts targeted surveys during at least two breeding seasons.

Stage four involves documenting the location and outcome of falcon nests to provide baseline data and to assist in informing the monitoring of disturbance and displacement post-construction. It also involves the quantitative assessment of flight activity within the wind farm envelope. The intent of this stage is to provide data suitable to make an objective, quantitative estimate of risk. In scenarios where falcons are detected breeding within or close to a proposed wind farm, implementing a telemetry study provides an objective approach to risk assessment and a higher level of quantification than that provided by incidental or direct observational methods alone.

Collision risk is quantitatively assessed in stage five using data collected in the previous stage with the results flowing through to the development of appropriate avoidance, remediation, mitigation and/or biodiversity offset options (Table 2).

Table 2  Potential avoidance, mitigation and offset options for addressing effects to New Zealand falcon associated with wind farm developments.

Effect addressed	Desired outcome	Option
Habitat loss	Retention of foraging and nest habitat	Avoid effect: refine wind farm design to avoid potential nest sites and high use areas
Disturbance to nesting falcons	Breeding productivity unaffected	Avoid effect: construct wind farm outside breeding season
Disturbance to nesting falcons	Breeding productivity unaffected	Avoid effect: establish construction set backs
Collision mortality	Reduced collision risk exposure	Mitigate effect: microsite turbines
Potential displacement	No displacement	Mitigate effect: microsite turbines
Habitat loss	Maintenance of habitat quality	Remediate effect: re-vegetate
Collision mortality	No-net-loss or net gain	Biodiversity offset: establish additional pairs of falcons

Post-consent monitoring

Avoiding disturbance during construction

Stage seven provides recommendations for avoiding disturbance to nesting falcons during the construction of a wind farm. Disturbance to nesting falcons is best avoided by limiting construction to outside the breeding season. If this is not possible, then a 200 m setback should be established around a nest while that nest is active. Because these recommendations are based on those developed to avoid negative effects associated with forestry management (Seaton et al. 2009b), we recommend that the behaviour of adult falcons and the outcome of each nest is monitored to establish the effectiveness of these setbacks in the context of wind farm construction. If negative effects are detected, the risk assessment framework flows to either modified setback distances or a review of mitigation and offsets (Fig. 1).

Post-construction monitoring

Stage eight outlines a process for monitoring the occurrence of displacement and collision mortality during the operational phase of a wind farm. Monitoring starts with the location of all falcon nests within one standard home range of the wind farm envelope (note that home ranges will likely be habitat- or site-dependent). Data on breeding and location are compared with data collected during stages one through four and decisions made about the need for the review of mitigation and/or offset approaches.

Collision mortality is probably best informed over time with a coordinated approach involving monitoring at multiple wind farms in a variety of locations. Collision risk is influenced by site-specific factors (De Lucas et al. 2007), hence is likely to be site-specific, so monitoring at multiple sites is also required to allow for variation of risk in differing situations. Monitoring avian collision rates usually involves systematic searches for carcasses in plots beneath turbines and statistical accounting for imperfect searcher detection and removal of carcasses by scavengers (e.g. Strickland et al. 2011). Although well designed studies following such an approach can be suitable for monitoring avian mortality generally, we do not recommend carcass searches for monitoring falcon collisions because the relatively low densities in which they occur means that the probability of detecting a collision is extremely low.

We recommend an approach whereby the survival of individual falcons is monitored using telemetry. For example, the survival of falcons fitted with radio transmitters incorporating mortality sensors, if monitored at regular and appropriate intervals, can be explicitly determined over time. Greater inference about survival is obtained if falcon use of the wind farm is also quantified, thus allowing conclusions about the survival of individuals relative to actual use of the wind farm to be made. Stage nine provides for mitigation and/or offsets to be reviewed based on the results of stage eight. Monitoring by telemetry is not practical for non-resident falcons that may occasionally visit a wind farm, due to the difficulties of locating such individuals within a wider region. Consequently, monitoring their survival is constrained by the limitations of performing carcass searches for species occurring at low densities.

Options for avoidance, remediation, mitigation and offsets

We recommend that project planning employ an approach reflecting the mitigation hierarchy (Quintero & Mathur 2011) followed by consideration of biodiversity offsets (Quétier & Lavorel 2011) incorporating the principles and standards of the Business and Biodiversity Offsets Programme (2009) to address any significant residual effects. In adopting this approach, efforts should first be focused on avoiding loss of nesting habitat and disturbance to nesting falcons, followed by consideration of collision risk. Mitigation and remediation should follow consideration of avoidance options and should develop approaches to minimise and remediate effects that cannot be avoided.

Table 2 summarises avoidance, mitigation, remediation and biodiversity offset options to address the potential effects of wind farms to falcons. Although falcons are known to be susceptible to predation at the nest site (Seaton 2009b; Kross & Nelson 2011) no predator control prescriptions have been developed for falcons. Therefore, until predator control methods for improving falcon productivity rates are effectively designed, tested and proved to be effective, we do not recommend predator control as a biodiversity offset technique because outcomes cannot be guaranteed.

Conclusion

Assessing the ecological effects of infrastructure projects on biodiversity is arguably best driven by good science responding to effective standards or policy. Currently, New Zealand has no formal standards or policy for assessing the potential effects of wind farms to avifauna. This has led to varying approaches to assessing the risk a proposed wind farm development poses to falcon, inconsistencies in the conclusions reached, and possible shortfalls in the actions put in place to address potential effects to this species. Because falcons are threatened, and raptor mortality has been a problem at some overseas wind farms yet interactions between New Zealand falcons and wind farms are poorly understood, it is important that this lack of standards is addressed. Where risk is thought to be high or very uncertain, assessments should include quantitative risk analyses to provide a more objective basis to support the RMA decision-making process and what action should be taken to avoid, mitigate, remedy or offset any likely effects. The risk assessment framework presented here comprises a series of steps that adaptively build on each other and, in doing so, provide a structured and objective hierarchical approach to assessing the effects of a proposed wind development to falcons.

Currently, much of the information on falcon/wind farm interactions is contained in confidential reports and, in moving our understanding forward, it is important that information from across New Zealand is collated and made publicly available. In the future we anticipate that this framework will be refined to reflect the knowledge gained during the operational life of wind farms, in particular through the explicit monitoring of individual falcons across multiple operational wind farms.

We suggest that, because of inconsistencies in assessment methods and gaps in the knowledge of how wind farms effect native fauna generally in New Zealand, similar risk assessment frameworks should be developed for other notable species. If these frameworks were adopted as national standards, the collective knowledge gained from their implementation would allow for improved decision making and the development of solutions that allow alternative energy targets to be met without compromising biodiversity conservation goals.