Transportation Research Interdisciplinary Perspectives xxx (xxxx) xxx

TRIP-100096; No of Pages 11

Contents lists available at ScienceDirect

Transportation Research Interdisciplinary Perspectivesj ourna l homepage: ht tps : / /www. journa ls .e lsev ie r .com/t ranspor tat ion- research-

in te rd isc ip l ina ry -perspect ives

Insects and aviation safety: The case of the keyhole wasp Pachodynerusnasidens (Hymenoptera: Vespidae) in Australia

Alan P.N. House a,⁎, Jackson G. Ring b, Matthew J. Hill c, Phillip P. Shaw d

a Eco Logical Australia, 12 Creek St, Brisbane, Qld 4001, Australiab Brisbane Airport Corporation, 11 The Circuit, Brisbane Airport, Qld 4008, Australiac Qantas Engineering, Brisbane Domestic Satellite, Brisbane Airport, Qld 4008, Australiad Ecosure Pty Ltd, 10/37 Connor Street, Burleigh Heads, Qld 4220, Australia

⁎ Corresponding author.E-mail address: alan.house@ecoaus.com.au. (A.P.N. Hou

http://dx.doi.org/10.1016/j.trip.2020.1000962590-1982/© 2020 The Author(s). Published by Elsev4.0/).

Please cite this article as: A.P.N. House, J.G.(Hymenoptera: Vespidae) in Australia..., Trans

A B S T R A C T

A R T I C L E I N F O

Article history:Received 19 December 2019Received in revised form 22 January 2020Accepted 14 February 2020Available online xxxx

While birds and other vertebrates are well known hazards to aviation at airports, the threat posed by invertebrates isless well understood. Here we present an example of a serious risk to flight safety from the mud-nesting keyhole wasp(Pachodynerus nasidens) which views aircraft pitot probes as an attractive nesting opportunity at Brisbane Airport. Pitotprobesmeasure airspeed, and obstructions can rendermeasurements inaccurate, leading to serious and potentially cat-astrophic consequences. We undertook experiments over 39 months to determine rates of nesting in pitot probes andthe associated risk of blocked probes. We also examined how this risk was reduced by covering probes. A bow-tie riskanalysis was completed to assess the safety, reputation, one-off financial loss, and injury and illness costs of a range ofincidents of increasing severity, and climate modelling was used to show the potential spread of the keyhole wasp inAustralia. The reduction in risk from covering 33% to 75% of all probes on arrival is substantial and made more sig-nificant when the costs of incidents are set against those of wasp management. The prospects for eradicating the key-hole wasp and the prospect of it spreading to other parts of Australia with suitable climatic conditions are discussed inview of the substantial risk the species presents.

Keywords:WaspAviationRisk managementPitot probeEradicationPachodynerus nasidens

1. Introduction

“They build very remarkable cells or nests of earth,finely tempered, andformed in layers of tiny mud-pats, like a swallow's nest. Many of thesewere placed in a small wooden outhouse, between the upright studsand the boarding of the wall; several were formed on a shelf in theporch, where some small pieces of wood lying heaped together offeredconvenient nooks; and one wasp, resolving to have a more costly lodg-ment than his friends, took possession of a meerschaum pipe-bowlwhich lay on the same shelf, and very snugly laid out his house in its in-terior.”

[(Meredith, 1849)]

The risks to air travel posed by wildlife are well known. In Australia,aircraft-wildlife encounters increased steadily between 2008 and 2017,with 16,626 confirmed birdstrikes with a high of 1921 in 2017(Australian Transport Safety Bureau, 2018) equating to 8.7 strikes per10,000 aircraft movements. This rate is higher than that in other countries(e.g. UK 5.0 strikes/10,000 movements (Civil Aviation Authority, 2017);

se).

ier Ltd. This is an open access artic

Ring, M.J. Hill, et al., Insects aportation Research Interdiscip

New Zealand 2.5 strikes/10,000 movements (Civil Aviation Authority ofNew Zealand, 2017); USA 1.7 strikes/10,000 movements (FederalAviation Authority, 2019); Canada 4.0 strikes/10,000 movements (Trans-port Canada, 2017), but similar to those in climatically equivalent regions(e.g. Brazil, mean of three airports 6.6 strikes/10,000 movements(Mendonca et al., 2018). Costs of wildlife strikes are very substantial glob-ally – up to US$2 billion annually 2009 (Weller, 2014), and an annual costof US$72 million to civil aviation in Australia (McKee and Shaw, 2016).Costs are not confined to equipment repair or replacement: significant ex-pense may be incurred through delays in services, damage to reputation,loss of life and property, financial penalties and complex lawsuits (Dale,2009).

While birdstrikes, and strikes of other wildlife such as bats and ground-dwelling animals, make headlines (e.g. US 1549, Marra et al., 2009;AWACS Boeing 707, Flight Safety Foundation, 1996), threats to flightsafety from insects are less well understood. Our interest in the relation-ships between insects and aircraft is limited to what they can tell us aboutthe mechanics of flight (e.g. Takahashi et al., 2018), how aircraft can aidin insect detection for pest control or other purposes (e.g. Amendt et al.,2017; Stone andMohammed, 2017) and disinsection to manage insects po-tentially carrying disease (Rayman, 2006; Mier-y-Teran-Romero et al.,2017). For wildlife management at airports, insects are generally seen asan attractant for birds that might pose a risk to flight safety (Cleary andDolbeer, 2005; Hauptfleisch and Dalton, 2015), rather than a risk in their

le under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/

nd aviation safety: The case of the keyhole wasp Pachodynerus nasidenslinary Perspectives, http://dx.doi.org/10.1016/j.trip.2020.100096

A.P.N. House et al. Transportation Research Interdisciplinary Perspectives xxx (xxxx) xxx

own right. However, aviation authorities, airport managers, commercialcarriers and private aviators are well aware of the threats posed by insects,and especially mudwasps, to aircraft operations, although to date there hasbeen no formal assessment of risk from these agents.

In Australia, the recent discovery of an introduced mud wasp(Pachodynerus nasidens Latreille – the keyhole wasp) has resulted in anovel hazard to aviation in Australia. This species is native to South andCentral America and the Caribbean region (Carpenter, 1986; House et al.,2020), and typically uses man-made structures (or natural cavities) to con-struct nests provisioned with butterfly, moth or other insect larvae andcapped with mud. At airports, ideally-sized cavities are abundant in airsidemachinery, including aircraft, where a number of suitable nesting opportu-nities are present: elevator feel differential probe (on tail), vent line to fuelpressure gauge (under wing), total air temperature probe (front fuselage),engine tail pipe, engine temperature/pressure probe (rear of engine) and,particularly, pitot probes.

Here we report on the risk the keyhole wasp poses to aviation safety inAustralia, using experimental data on nesting to model the probability of ablocked probe; we assess the safety, financial, reputational, and operationalcosts of an escalating scale in severity of incidents caused by wasps nestingin pitot probes, and model the relationship between given acceptable levelsof risk and costs of over- and under-managing the problem.

1.1. The hazard

Pitot probes are hollow instruments projecting forwards from the air-craft fuselage which measure airspeed by registering differential pressuresbetween total pressure (from air entering the probe) and air in a staticchamber (which may be incorporated into the pitot probe or be separate),using Bernoulli's equation (Morris and Langari, 2016). In large passengeraircraft pitot probes are normally mounted at the front of the aircraft onboth captain's and first officer's sides but a third may also be located onthe empennage (tail).

Pitot probes are critical instruments in all aircraft. It is vital that pilotsknow their airspeed, especially for take-off and landing, and anomalies be-tween airspeed indications between probes can lead to costly and hazard-ous rejected take-offs or air turn backs (e.g. (Australian Transport SafetyBureau, 2016, 2006); or significant loss of life (Bureau d’Enquêtes etd’Analyses pour la sécurité de l’aviation civile, 2012; Boeing, 2003;National Transport Safety Bureau, 1981). Obstruction by insects isrecognised as a threat to pitot probe function (Civil Aviation SafetyAuthority, 2018; Australian Transport Safety Bureau, 2018b, 2016, 2006.The maintenance manual for the Boeing 737–800 states: “A pitot probe orstatic port system blocked by foreign objects such as insects may causelarge errors in airspeed-sensing and altitude-sensing signals, which maylead to loss of safe flight”.

Amongst instances of mud wasps causing problems for aircraft, a num-ber of tragic accidents involving pitot probes most likely blocked by waspsare well known, including a Boeing 757 that crashed shortly after take-offfrom the Dominican Republic in February 1996, killing all 189 passengersand crew (Bureau d’Enquêtes et d’Analyses pour la sécurité de l’aviationcivile, 2012). The pilots misjudged the aircraft speed because of anomalousreadings from the pitot probes. In an earlier incident, a Douglas DC-3crashed into the Atlantic Ocean near the Bahamas after take-off fromWest Palm Beach International Airport, Florida in 1980 (NationalTransport Safety Bureau, 1981) with 34 lives lost. In both cases, mud-dauber wasp nest material was found in the pitot heads and consideredthe most likely cause of the pitot system malfunction.

1.2. Wasp incidents at Brisbane Airport

In January–March 2006 five incidents were reported at Brisbane Air-port in which pitot probes gave inconsistent airspeed readings in the cock-pit and flights were either rejected during take-off roll or proceeded to theirdestination with an erroneous airspeed indication on one of the three pitotsystems. All aircraft were Airbus A330s. Of these, one led to a dangerous

2

high-energy stop causing brake overheating and tyre deflation when thetake-off was rejected, resulting in the deployment of airport rescue andfire-fighting services (Australian Transport Safety Bureau, 2008). On in-spection, it was found that one pitot probe contained “wasp-related debris”concluded to have caused anomalous speed readings between the pilot fly-ing and the pilot-not-flying instruments. Material retrieved from this probeby the AustralianMuseum included fragments of an insect body resemblingthe head of a wasp.

An incident in November 2013 initiated a review of the impact of mud-nesting wasps on aircraft operations at Brisbane and a framework forreporting wasp-related incidents. In one of the most significant incidentsat Brisbane to date, an Airbus A330 performed a rejected takeoff andreturned after airspeed discrepancies between probes were identified dur-ing the takeoff roll (Australian Transport Safety Bureau, 2016). Inspectionand subsequent maintenance cleared the aircraft for a second take-off, butonce airborne another airspeed discrepancy caused a mayday situationand the aircraft returned under emergency procedures. Sand and mud con-sistent with a mud wasp nest was discovered in one probe. Since this inci-dent, more detailed records of wasp-related issues have been made atBrisbane and a total of 26 incidents were entered into the database betweenNovember 2013 and April 2019.

2. Materials and methods

2.1. Study site and probe experiment

The study site (Brisbane Airport) is located on the eastern seaboard ofAustralia (27.3911°S, 153.1197°E), on the shores of Moreton Bay, Queens-land at approximately 4 m above sea level. It is flanked on the east by theBrisbane River, on the west by ecologically significant coastal wetlands(mangroves and saltmarshes) and to the south by the city of Brisbane.The climate is sub-tropical, with a long-term mean annual rainfall of1190 mm, 50% of which falls between December and March. The annualmean maximum temperature is 25.4 °C, and annual mean minimum is15.7 °C, with fewer than two frost days per year (Bureau of Meteorology,2019).

In 2016, amonitoring programwas established to identify thewasp spe-cies responsible for pitot probe blockages, the seasonal pattern of nesting,the type of aircraft most likely to be impacted, and the location most atrisk (House et al., 2020). Replica pitot tubes were fixed to three sets ofpanels erected at four locations at Brisbane airport including both domesticand international terminals (Fig. 1). The 3D-printed pitot tubes representedairframes with specific movement frequency at the airport and a range ofaperture dimensions (2.5–8 mm) (Boeing 737-400, 737-800, 747-400, Air-bus A330, De Haviland Dash-8 and Embraer 190). Panels were monitoredweekly (during main breeding season) to monthly, and blocked probes re-moved for incubation were replaced by empty probes of the same type.Blocked probes were incubated in mesh bags, and emerging adult insectsretrieved and identified.

To test whether female wasps could be deterred from nesting in pitotprobes, a series of intercept traps were deployed at the airport in bothairside (16 traps) and landside (11 traps) locations in November 2018.These were intended to provide wasps with alternative nesting sites awayfrom stationary aircraft at gates. Each trap consisted of 68 cardboardtubes 8 mm in diameter and 153 mm long, with one open end. Blockedtubes were removed and replaced by clean ones weekly.

2.2. Incidence and prevalence rates

Standard methods of measuring prevalence and incidence used in epi-demiological studies (e.g. (Brinks and Landwehr, 2015) have been appliedhere as measures of the risk of pitot probes being blocked. Incidence is theoccurrence of new cases (i.e. blocked pitot probes) within a specified timeperiod. Prevalence is the proportion of subjects (pitot probes) blocked atany one time.

Fig. 1. Left: Pitot probe panel established at Brisbane Airport, with 3D-printed probes of (from top) 747-400, 737-400, A330, 737-800, DHC-8. Panels were fixed so thatprobes were facing the same direction as aircraft probes when an aircraft was parked at the gate, and at a roughly equivalent height. Right: Female keyhole wasp(Pachodynerus nasidens) on 3D-printed DHC8 probe, Brisbane Airport, May 2016.

A.P.N. House et al. Transportation Research Interdisciplinary Perspectives xxx (xxxx) xxx

2.3. Failure rates

A classic engineering approach to estimate failure rate was applied (e.g.Finkelstein, 2008). This method assumes that there is no learning and adap-tation by the wasps over time, but it provides a simple method to test theimpact of management measures.

Using annual movement statistics for the airframes used in the monitor-ing study (= total number ofmovements/2 identifying each arrival and de-parture as one movement or one opportunity for nesting to occur), andassuming there are three probes per aircraft, we can calculate the numberof probes potentially “available” for nesting in each year for the selected air-frames, and the number per hour for a range of turn-around scenarios. Wecan calculate the failure rate as:

λ ¼ NbNp

=t

where Nb = failure count (number of probes blocked), Np = number ofprobes available, and t = total run time (hours).

Failure rates were modelled for four protection scenarios (0%, 33%,66% and 75% of probes covered) usingmean time before failure principles.The probability of failure Pf that x probes (or fewer) will be blocked in ayear with n probes available for nesting is given as the complement of:

b x; n;Psð Þ ¼ n!x! n−xð Þ!

� �Psð Þx 1−Pð Þn−x

where the probability of success (no blocked probes) Ps = (1 − Pf)(National Aeronautic and Space Administration, 2019). The binomial dis-tribution is used as there are only two possible outcomes for each probe(blocked or not blocked), and the Poisson distribution approximates tothe binomial when values of n (probes) are large, and x (blockages) aresmall, as we have in this instance. The complement simply means that thesum of probabilities of a probe becoming blocked and of not becomingblocked is 1 (i.e. the sum of all possible outcomes).

Rates were based on the number of blocked probes in the experiment,scaled to reflect the number of probes potentially available for nesting atthe airport in a year: number of arrival/departure cycles of the airframeschosen for the experiment, multiplied by three probes per aircraft. Wealso ran themodels assuming that probes were available for nesting (on av-erage) 24, 16, 10, 6, 3 and 1 h to examine the effect of turn-around times onprobe blockage risk. We assumed that all available probes in each scenariohad an equal chance of being chosen for nesting.

3

2.4. Risk assessments

To examine the potential risks to human life, financial performance andreputation, a standard risk assessment (consequence, likelihood), was un-dertaken, based on the Qantas Group Risk Assessment Guide (internalQantas document), which is typical of processes used in the aviation indus-try to assess and manage risk. Only the consequences relevant to the key-hole wasp threat were included (Table 1). This allowed quantification ofthe costs of each type of incident, and a detailed breakdown of issues relat-ing to damage and liability. The scenario modelled was of a fully ladenBoeing 737–800 (carrying 174 passengers and 7 crew): this represents thecurrent most common configuration for aircraft arriving and departingthe airport. A likelihood/consequence matrix (Table 2) was used to rateoverall risk. The consequence levels are determined by the separate assess-ment of consequence in Table 1. This assessment was done in a dedicatedworkshop involving airport management, the Civil Aviation Safety Author-ity, a major airline and researchers.

Secondly, a qualitative bow-tie risk assessment model was constructed(de Ruijter and Guldenmund, 2016). This allows identification of controlsin place as well as control failures (escalation risks) and in turn how theseare managed (escalation factor controls). The bow-tie framework consistsof four elements: a top or critical event (i.e. the issue that causes a break-down in system function), threats that increase the likelihood of the eventtaking place (equivalent to fault tree analysis), barriers that hindermanage-ment of the system to prevent the event (event tree analysis), and conse-quences (i.e. outcomes of the event once barriers are addressed) (deRuijter and Guldenmund, 2016). Bow-tie diagrams are used extensivelyin aviation risk management (Cacciabue et al., 2015; Kontogiannis et al.,2017). The critical event applied was a pitot probe blocked by a mudwasp nest causing an airspeed reading anomaly on the fight deck.

To assess how the acceptable level of risk (ALOR) changes the cost ofmanagement, we used the method described in Davidson et al. (2015) forinvasive species. This takes the cost of not acting (or under-management:Type II errors) and compares them with the cost of action (or over-management: Type I errors). Type II errors are often assumed to be irrevers-ible. The intention is for these considerations to bemade in the early stagesof an invasion, when the relative benefit of over management is high, andthe chances of success are greatest. We examined the relationship betweenALOR and indicative costs of action in two scenarios: Type I error ratesfixed at 0.05 and 0.01, with a budget based on real costs of current manage-ment (pest control, this pitot probe study and the intercept traps) and theannual mean of indicative costs between November 2013 and April 2019,which we equate here with the cost of management to remove all risk.

Table 1Consequence table used in standard risk assessment (from Qantas Group Risk Assessment Guide).

Consequencetype

Level 1 Level 2 Level 3 Level 4 Level 5 Level 6

Safety andsecurity offlight

Deviationmanagedusingstandardprocedures

Deviation with minimalpotential to impact safeoperation. Managedusing normalprocedures.

Deviation with somepotential to impact safeoperation. Managed usingnon-normal procedures.

Deviation impacting continuedsafe operation. Managed usingmultiple non-normal checklistsand procedures

Condition impacting theimmediate safe operation.Management requiresemergency actions to preserveaircraft.

Total loss of aircraft

Reputation Isolated,short-termchange tocustomerexperience ofa customer.

Isolated short-termchange to customerexperience for a numberof customers, or < aweek of local mediacoverage.

Minor change to customerexperience and/oradvocacy for a number ofcustomers, or extendedlocal media coverage(week +).

Moderate change to customerexperience, and/or advocacyfor our brand for a number ofstakeholders for < a year, orinternational media coverage (aday).

Major change in customeradvocacy and brandpreference for manystakeholders for 1–2 years, orsustained (<week)international media coverage.

Significant change in brandpreference for manystakeholders for a period of>2 years, or prolonged(week+) internationalmedia coverage

One-offfinancialbenefit orloss(revenue orcash)

One-offbenefit or loss< $2 million.

One-off benefit or loss ofbetween $2–$20million.

One-off benefit or loss ofbetween $20–$200million.

One-off benefit or loss ofbetween $200–$500 million.

One-off benefit or loss ofbetween $500 million– $2.5billion.

One-off benefit or loss >$2.5 billion.

Injury andillness

Recoverableinjury withno first aid ormedicaltreatmentrequired.

Recoverable injuryrequiring first aid,physiotherapy ormedical assessmentwith no follow-uptreatment required.

Treatment by a registeredmedical practitionerrequiring ongoingtreatment with nopermanent disability/lossof capacity.

Immediate admission tohosp1ital as an inpatient and/orpermanent partialdisability/loss of capacity

One fatality and/orpermanent totaldisability/loss of capacity tobetween one to ten people.

Two or more fatalitiesand/or permanent totaldisability/loss of capacity tomore than ten people.

Table 2Likelihood table used in risk assessment.

Consequence

Likelihood Level 1 Level 2 Level 3 Level 4 Level 5 Level 6

Almost certainLOW MEDIUM HIGH EXTREME EXTREME EXTREME

Expected to occur in most circumstances, inevitable, could occur within days to weeks. ALikely

LOW MEDIUM HIGH HIGH EXTREME EXTREMECould occur in most circumstances, not surprised if it happens, could occur within months. BPossible

VERYLOW

LOW MEDIUM HIGH HIGH EXTREMEMight occur in some circumstances, has occurred in the business before, could occur within one to twoyears. C

Unlikely VERYLOW

LOW LOW MEDIUM HIGH HIGHCould occur in some circumstances, has occurred in a similar organisation, could occur within years. DRare VERY

LOWVERYLOW

LOW LOW MEDIUM HIGHMay occur but only in exceptional circumstances, has occurred elsewhere, could occur within decades. EVery rare VERY

LOWVERYLOW

VERYLOW

LOW MEDIUM MEDIUMTheoretically plausible but not expected to occur, could occur once in every 100 years. F

A.P.N. House et al. Transportation Research Interdisciplinary Perspectives xxx (xxxx) xxx

Indicative costs were set for the two incident types recorded at Brisbane be-tween those dates: low speed rejected takeoff (AUD$99,000 per incident)and return to land (AUD$434,360 per incident). All incidents were costedassuming a Boeing 737–800 with 174 passengers and 7 crew.

2.5. Risk of spread

The potential spread of the keyhole wasp in Australia was determinedusing the CGIAR Climate Analogues tool (Climate Change Agricultureand Food Security, 2019). We used a coastal site near Florianópolis insouthern Brazil (27° 41′ 22”S, 48° 38′ 27”W) as a source climate in the spe-cies' native range and which broadly matches latitude, distance from theocean and altitude with those of Brisbane Airport, and ran three models:current climates in both source (Brazil) and target (Australia), currentsource and future target, and future source and future target. Currentdata are based on historical climatic averages (1960–1990) and futuredata were forecast to 2020–2049, using the SRES A1B emissions scenario(Intergovernmental Panel on Climate Change, 2000).

3. Results

The results of the experimental work to April 2019 are reported inHouse et al. (2020). Only one species of mud wasp (P. nasidens) used pitot

4

probes to nest at Brisbane Airport. Preferred probes had an aperture≥4 mm diameter (A330, Boeing 737-400, 737-800, 747, and De HavilandDash 8). Both domestic and international terminals are affected, with a con-centration of nesting activity at one end of the domestic terminal, andnesting was recorded in 9 months of the year (September–June) a seasonalpattern similar to that in wild populations in the species' native range,peaking in warmer and wetter months (November–May). However, key-hole wasp activity has been reported at the airport in all months, indicatingsome level of threat thewhole year round. Over the 39-month experimentalperiod, 60 probes were deployed and 93 blockages by P. nasidens wererecorded.

3.1. Incidence and prevalence rates

Using data from House et al. (2020), we calculated the incidence rate ofpitot probe nesting. Sixty probes were available for nesting over a period of39 months. In this time, 93 probe blockages were recorded, with 38 adultwasps emerging (23 males and 15 females) resulting in an average of2.38 blockages per month, and 0.0397 blockages per probe per month.

Prevalence, calculated as the proportion (expressed as a percentage ofthe total population) of pitot probes blocked, was 3.97% over 39 months.Incidence and prevalence vary together, as in our system there is no “mor-tality” to reduce the population of probes available for nesting.

A.P.N. House et al. Transportation Research Interdisciplinary Perspectives xxx (xxxx) xxx

The rate of accumulation of new blockages is not a smooth progressionas there are within and between year influences of environment (House etal., 2020) (Fig. 2), and no asymptote has been reached, indicating furtherpotential for both incidence rate and prevalence (and therefore risk) to in-crease over time and that intervention is likely to be beneficial.

3.2. Failure rate

There is a dramatic reduction in risk, from approximately 26 blockagesper year if no probes are covered, to 10 if 33% of probes are covered, to 5if 66% are covered and to 3 if 75% are covered (Fig. 3). The probabilitythat zero probes will be blocked in a year is 0.3 if 66% of probes are covered,and 0.5 if 75% of probes are covered, assuming that all probes are availablefor nesting for 24 h/day. When the length of time an aircraft is present at agate is also reduced (Fig. 3) the number of blockages and the probability ofblockage are further reduced. However, there is still a 3% chance (i.e. resid-ual risk) that a probewill be blocked if 75%of all probes are covered (Psmin=0.9698), and the aircraft is on the ground for 1 h (Fig. 4), and the benefit ofapplying covers accrues at a faster rate as turn-around times increase.

We can further examine the risk by including the likelihood of nestingbased on the effect of the intercept traps on the population of waspsattempting to nest in probes. The average rate of probe blockage was0.442 blockages/year before the intercept traps were installed, and 0.72blockages/year after installation. However, the number of successful nests(i.e. those that hatched adult wasps) fell from 0.48 wasps/blockage to0.06 wasps/blockage, an 8-fold decline. Although the number of nestingevents increased, the impact on the wasp population appears to be dra-matic, and further monitoring is required to see if this effect is carried for-ward. As the act of nesting is primarily responsible for the risk and notsuccessful nesting, we do not include any potential impacts of the intercepttraps in this analysis.

3.3. Risk assessments

The standard risk assessment included a review of individual risks tosafety of flight, brand reputation, one-off financial loss, and injury and ill-ness applied to a range of possible outcomes of increasing severity: lowspeed rejection of take-off, high speed rejected takeoff, runway excursion,air return and total loss of aircraft. These possible outcomes all stem fromairspeed anomalies detected by the flight crew: the phase of take-off inwhich the decision is made (to proceed or abort) determines the outcome.

As expected, the risks increase dramatically with increasing severity ofincident, but some aspects are high risk even at relatively low levels of phys-ical peril (Table 3).

0

10

20

30

40

50

60

70

80

90

100

1 3 5 7 9 11 13 15 17 1

blockages

m

Fig. 2. Cumulative experimental pitot probe blockages (≡ nesting

5

At the lower end of consequence, delays to flights that have returned togate after a low speed rejection can still incur a significant cost (Table 4),whereas the loss of the aircraft and life is substantial. This assessment didnot include indirect impacts on other parts of airline businesses or associ-ated businesses, or financial impacts of reputational brand damage as thisis carrier-specific, or the loss of infrastructure and life if the aircraft is lostover land.

In the bow-tie analysis (Fig. 5), the top event is a pitot probe blockedby a keyhole wasp. This can be mitigated by addressing the root causes(i.e. fault tree). In our case, of the three groups of root causes, the con-trols of one (keyhole wasp ecology and behaviour) have been addressedentirely, those of another (maintenance regime) is partly addressed bythe local policy of covering probes by some carriers, whilst the controlspossible for pitot probe design and function have not been addressed.There are a number of reasons for this, including the difficulty in re-designing probes to exclude wasps but not impair airflow, and the sub-stantial costs of retro-fitting new probes to all aircraft in the Australianfleet.

On the event tree side of the bow-tie diagram, the responses avail-able to pilots depend on the stage of flight the airspeed anomaly is reg-istered, and the functioning of flight instruments and the training theyhave received. If we assume that all pilots receive adequate training,the instruments are operating correctly and the response to any givenscenario is appropriate, then the only corrective actions can be madeon the fault tree side.

The results of the over- or under-management scenarios on real costs ofcontrol of the keyhole wasp are shown in Fig. 6. Clearly, the greater ALOR,the lower the costs of control, and the more certainty that management willwork (i.e. Type I error rate = 0.05) and the more costs can be reduced. Atpresent the ratio of cost of incidents to cost of management is 19.74, sug-gesting that operators are nearly 20 times more willing to take the risk ofan incident than to spend adequately on managing the problem.

3.4. Risk of spread

Climate futures modelling (Fig. 7) shows that much of the eastern sea-board of Australia is potentially suitable for the persistence of the keyholewasp, under current and future climate scenarios, although other criteriamust also be met (e.g. suitable prey availability). Within the zone of>60% climate similarity in 2020–2049 there are 11 certified airportsAustralian (in addition to Brisbane) that could be at risk if the wasp istransported there. Four of these (Sunshine Coast, Gold Coast, Sydney andPerth) are international ports.

9 21 23 25 27 29 31 33 35 37 39

onths

) over 39 months. Total of 60 probes available, 93 blockages.

0

0.2

0.4

0.6

0.8

1

0 1 2 3 4

Prob

ability

0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6 7

Prob

ability

3 h

0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6 7 8 9 10

Prob

ablity

6 h

0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10 12 14

Prob

ability

10 h

0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10 12 14 16 18

Prob

ability

Number of blockages

16 h

0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10 12 14 16 18 20 22 24 26

Prob

ability

Number of blockages

24 h

1 h

Fig. 3. Probability of failure (i.e. x pitot blockages or fewer in one year) with 0%, 33%, 66% and 75% of probes covered and average time at gate for aircraft of 1, 3, 6, 10, 16and 24 h.

A.P.N. House et al. Transportation Research Interdisciplinary Perspectives xxx (xxxx) xxx

4. Discussion

The keyhole wasp is a known hazard to aviation in many tropical andsub-tropical areas in Central and South America and the Caribbean. Itshighly plastic nesting behaviour allows it to take advantage of man-madecavities, including those in aircraft. The recent discovery of the species inAustralia, and a series of serious incidents at Brisbane Airport, has elevatedthe issue to a significant and potentially catastrophic threat.

4.1. Airport, carrier and aviation authority responses

Routine pest management at Brisbane Airport has included the removalof mud wasp nests (of species other than P. nasidens) from terminal build-ings and other structures since 2006. The number of nests treated has fallendramatically over time: however, this has not reduced the threat from key-hole wasps, which do not build complete nests and very sparingly use thenests of other species (Freeman and Jayasingh, 1975). Deployment of inter-cept traps in late 2018 appears to have reduced the number of wasps

6

hatching from blocked probes, but not the number of blockages which isthe critical safety issue for aircraft.

A NOTAM (notice to airmen) warning of the potential risk to aircraftwas issued at Brisbane in February 2014 and awarning aboutmudwasp ac-tivity at Brisbane was included in the ERSA (En Route SupplementAustralia) to the Aeronautical Information Package for airmen, issued byAir Services Australia, in the same year.

Currently, a number of carriers at Brisbane have local policies of cover-ing pitot probes on arrival, but this is not mandatory. The Civil AviationSafety Authority issued an airworthiness bulletin listing several recommen-dations to reduce risk (Civil Aviation Safety Authority, 2018), including ap-plying pitot probe covers as soon as possible on arrival (after necessarycool-down time), checking these covers for damage, and closely inspectingaircraft after long-term storage in the open. The Civil Aviation SafetyAuthority (2018) noted that standard pre-flight air data module checkswith built-in testing equipment will not detect probe blockages in station-ary aircraft. Also, standard pre-flight walk-around inspections may not de-tect a compromised, uncovered pitot probe, as the first material inserted(of either mud or prey) cannot be seen without close inspection using a

0

0.2

0.4

0.6

0.8

1

0 4 8 12 16 20 24

P smin

Hours uncovered

0% 33% 66% 75%

Fig. 4. Probability of 0 probes blocked in one year (Psmin) with time probes are uncovered, for 0%, 33%, 66% and 75% of probes covered.

Table 3Risk assessment results.

Outcome Consequencelevel

Likelihood Risk

Low speed rejected takeoffSafety and security of flight 2 LIKELY MEDIUMReputation 2 LIKELY MEDIUMOne-off financial benefit or loss (revenue orcash)

2 LIKELY LOW

Injury and illness 1 UNLIKELY VERYLOW

High speed rejected takeoffSafety and security of flight 2 LIKELY MEDIUMReputation 3 LIKELY HIGHOne-off financial benefit or loss (revenue orcash)

1 LIKELY LOW

Injury and illness 1 UNLIKELY VERYLOW

Return to landSafety and security of flight 4 POSSIBLE HIGHReputation 3 LIKELY HIGHOne-off financial benefit or loss (revenue orcash)

2 LIKELY LOW

Injury and illness 2 POSSIBLE LOWRunway overrunSafety and security of flight 5 POSSIBLE HIGHReputation 4 POSSIBLE HIGHOne-off financial benefit or loss (revenue orcash)

POSSIBLE MEDIUM

Injury and illness 4 POSSIBLE HIGHLoss of aircraftSafety and security of flight 6 RARE HIGHReputation 6 RARE HIGHOne-off financial benefit or loss (revenue orcash)

6 RARE HIGH

Injury and illness 6 RARE HIGH

A.P.N. House et al. Transportation Research Interdisciplinary Perspectives xxx (xxxx) xxx

borescope or similar inspection tool: therefore, covers are the preferredmanagement control.

The simple procedure of applying pitot covers to aircraft at gates or instorage, as recommended by manufacturers (Boeing, 2003; Duquesneet al., 2016) and aviation authorities (Civil Aviation Safety Authority,2018) incurs its own risks: failure to remove covers. In July 2018, anA330 bound for Kuala Lumpur issued a pan-pan distress call (an urgent sit-uation requiring air traffic control attention and emergency services

7

mobilisation) after take-off from Brisbane and made a heavy landing. Itwas discovered that probe covers applied on arrival had not been removed(Australian Transport Safety Bureau, 2018b). A second risk identified withprobe covers is that the pitot probe head can be contaminated by fragmentsfrayed from the frequent use of Kevlar covers.

Carriers are justifiably concerned that the risk of leaving covers on maybe greater than that presented by the keyhole wasp, although it is arguablethat the consequence is less as the worst outcome of leaving covers onwould be a high-speed rejected take-off. However, the failure rate analysesdo show that applying covers reduces the risk of nesting substantially, evenif covering every probe on aircraft at the gates is not practical; this is espe-cially beneficial if turn-around times are short. The time taken to approachps = 1 is reduced dramatically by covering probes and shorter gate times,but there is still the potential for four blockages per year even if 75% ofprobes are covered and gate times are limited to 1 h. Aircraft are stationaryat gates for a range of times (from under 1 h to overnight), but this analysisallows authorities to examine the residual risk of blockage given a range ofpolicies and turn-around times.

Engineering solutions to prevent pitot blockages are generally incom-patible with the function of the instrument, although Jackson (2015) hasproposed a design that uses optical sensors to detect debris or ice. This so-lution has not yet been incorporated into operational aircraft, but this ap-proach (or variant) and LiDAR-based airspeed measurement systems (e.g.(Verbeek and Jentink, 2012) will possibly replace conventional pitotprobe systems in time.

4.2. Pest control options

As a relatively recent arrival in Australia, the chances of eradicating thekeyhole waspwould appear to be good. As far as is known the species is stillrestricted to the Brisbane area, centred on the airport precinct: survey andsurveillance in the wider Brisbane area has not detected the species (R.Wylie pers. comm.). However, standard cost/benefit analyses of eradica-tion versus other responses (i.e. containment, control) do not apply wherethere is not an immediate and demonstrable import or export benefit. Inthe case of the keyhole wasp, the benefit of action is the reduction in likeli-hood of a serious (and expensive) outcome. Aircraft will continue to arriveand depart whether a loss has been incurred or not, so decisions arounderadication therefore must be taken with a future loss expected, despitethe actual incidents that have been reported and the (rare) catastrophicconsequences experienced. The risk analysis has shown that the likelihood

Table 4Indicative direct costs of increasingly serious outcomes due to an undetected keyhole wasp nest in a pitot probe. Case aircraft is a Boeing 737-800 with 174 passengers and 7crew.

Outcome Type of cost Estimated itemised cost (A$ 2019) Total cost (A$ 2019)

low speed rejection • replacement and testing of probe• time out of service

• $8500• $1000 / min. for 90 min.

$98,500

high speed rejection • replacement and testing of probe, single wheel brake replacement, tyre replacement• passenger care• time out of service

• $35,000

• $5220a–$43,500b

• $1000 / min. for 120 min.

$160,220-198,500

return to land • replacement and testing of probe, fuel burn/dump, landing fees, incidentals• passenger care• time out of service

• $25,000

• $5220a–$43,500b

• $1000 / min. for 180 min.

$210,220-248,500

runway excursion • aircraft recovery, replacement and testing of probe, brake packs, tyres, engine repair/replacement• passenger care• time out of service

• $10,000,000

• $5220a–$43,500b

• $1000 / min.

>$15 M

loss of aircraft • replace aircraft, time out of service, life insurance pay outs • > $100 M > $4100 M

a Assumes A$30 meal voucher per passenger and re-booked same day.b Assumes A$250 per passenger overnight and re-booked next day.

A.P.N. House et al. Transportation Research Interdisciplinary Perspectives xxx (xxxx) xxx

of an adverse incident occurring as a result of keyhole wasp activity wasrated no higher than likely, but the consequence could be catastrophic.

Early intervention in pest management relies on early detection(Hulme, 2009), but recognising the risk is also critical. Fraser et al.(2006) show that eradicating an invasive species is most likely the pre-ferred option if there is an immediate financial benefit to be derivedfrom the eradication, and where uncertainty about the success of erad-ication is low. Eradicating pests from man-made habitats (such as air-ports) depends on the spatial extent of infestation and is clearly morepossible at a local scale (Pluess et al., 2012b). Although mud nestingwasp species are prone to parasitism and predation by range of otherorganisms, biological control has had limited success (Beggs et al.,2011), and in a solitary species such as the keyhole wasp the low den-sity and widely dispersed nature of the population would presumablylimit success (Zaviezo and Mills, 2000; Samaková et al., 2019). Severaltechnologies are available for detecting and eradicating social wasps(Turchi and Derijard, 2018), but many do not apply to solitary nestingspecies.

Eradication programs can be evaluated using four potential successfactors: reaction time, size of infested area, biological knowledge andpreparedness to act, and insularity. An analysis of 136 pest eradicationprograms (Pluess et al., 2012a) showed that only one of these potentialfactors (extent of infestation) was important for a successful outcome.For wasps (Vespidae) in general there are very few cases of targeted con-trol programs once the species has become established (Beggs et al.,2011), and although P. nasidens is a solitary species the particular threatit poses to aviation theoretically makes it an important target for con-trol. The relative costs of pre-emptive, over-management action (i.e.the current management strategy, 2019 cost approx. AUD$80,000)can be kept low relative to the cost of under-management if this isequated with the cost of incidents (mean annual cost AUD$1.6 M). In re-ality, managers are reluctant to spend on prevention, as the returns oninvestment are less certain than the gain realised from control (Finnoffet al., 2007).

At Brisbane Airport, removing nests from blocked dummy pitotprobes and intercept traps is effectively a breeding disruption techniqueif it operates more efficiently than the population taking advantage ofthe additional available nesting sites. This is a common practice in inte-grated pest insect management and is used as part of integrated strate-gies (Rodriguez-Saona and Stelinski, 2009; Suckling et al., 2013).There is some evidence (House et al., 2020) that intercept traps are re-ducing the available adult breeding population since breeding success(i.e. successful emergence of adult wasps from the replica pitot probes)

8

has declined since their introduction (House et al., 2020). Introducing atargeted keyhole wasp control plan early (i.e. before the population ex-pands to off-airport areas or to other airports) may also take advantageof any Allee effects that may occur (reduction in fitness due to limitedpopulation size), although we do not know at what size of populationthese effects would be manifest (if at all). Allee effects are known tobe an effective containing mechanism in some insect pests (e.g.European gypsy moth in the USA, Tobin et al., 2009), and data fromthis study suggests that this may be playing a part in reducing successfulnesting in pitot probes. As we have no reason to expect that there willnot be further arrivals of the keyhole wasp in Australia, the use of inter-cept traps will ensure a strong Allee effect is maintained, thus reducingthe risk of population growth and dispersal (McDermott and Finnoff,2016).

Given the relatively recent arrival of the keyhole wasp in Australia, thelimited extent of the known infestation, and our knowledge of nesting pref-erences and risk factors, we believe that eradication of the species fromAustralia is possible.

4.3. Risk of spread in Australia

The keyhole wasp's ability to survive long-distance transport hasallowed it to spread from its native range in southern and central Americato the USA (Carpenter, 1986) and various locations in the Pacific Ocean(Yamane et al., 1996), and now toAustralia. As far as we are aware, it is cur-rently confined to the Brisbane area, mainly in the Brisbane Airport pre-cinct but has also been recorded at the nearby Port of Brisbane (R. Wyliepers. comm.). There is also an unconfirmed record of the species in Emerald(central Queensland).

This spread is most likely to have been through shipping, althoughwasps are possibly carried on aircraft as much of the spread is postWorld War 2 as air traffic increased in the region. The appearance ofP. nasidens in Hawaii as early as 1912 confirms that it arrived by boat,as the first crossing of the Pacific to Hawaii (O'ahu) by airplane fromthe west coast of USA (Oakland) did not take place until 1927 (Ryan,2018).

Tramp species such as the keyhole wasp are most likely to betransported as larvae or pupae (rather than adults) in nests attached to ship-ping crates or other cargo, or to the ships themselves. However, aircraft lug-gage bay temperatures may be sufficiently high (7 °C to over 25 °C: Emondet al., 1999) to allow an adult wasp to survive, and evenwheel-wellsmay besuitable for shorter and lower altitude flights (18 °C: Perry, 2002) or evenlong-distance high altitude (12,000 m) flights (8–25 °C: Russell, 1987).

Fig. 5.Bow-tie diagram for keyholewasp risk analysis. Under outcomes, the severity of consequences is ordered (left to right): safety and security offlight, reputation, one-offfinancial loss (revenue or cash), injury and illness.

A.P.N. House et al. Transportation Research Interdisciplinary Perspectives xxx (xxxx) xxx

Invertebrates are especially good stowaways, accounting for over 90%of allstowaway introductions of pests to South Africa (Faulkner et al., 2016).However, to spread further in Australia from Brisbane Airport the speciesneed not take such risks: as an import hub Brisbane Airport receives around52,600 t of freight per year,most ofwhich is transported from the airport byroad, providing the keyhole wasp with significant opportunities to dispersewhere climatic conditions are generally suitable. Similar opportunitiespresent at the Port of Brisbane. That it may not yet have dispersed outside

0

2

4

6

8

10

12

14

16

0 0.02 0.04

ALORmeanannu

alcost(AUD$

M)

Type II erro

Type I error = 0.05

Fig. 6. Relationship between indicative cost of management and

9

the region would be a matter of fortune rather than management, as it isclearly an adaptable, long-distance traveller, and is not dependent onspecialised nesting sites or food resources. Local control measures may bereducing the risk at Brisbane Airport, but may not be enough to preventthe keyhole wasp hitching a ride on road, sea or even air transport toother ports. Surveillance programs (e.g. Wylie et al., 2008) have not in-cluded the keyhole wasp to date: it is not yet classed as an agricultural, so-cial or environmental pest in Australia. If eradication is not attempted, then

0.06 0.08 0.1

r rate (ALOR)

Type I error = 0.1

acceptable level of risk, for Type I error rates of 0.05 and 0.1.

Fig. 7. Similarity between source climate (Florianopólis, Brazil) and target climate(Australia) for 2050 as an indicator of potentially suitable environments for the key-hole wasp in Australia. Dark blue represents themost similar climates and thereforepotential areas of spread. (For interpretation of the references to colour in thisfigurelegend, the reader is referred to the web version of this article.)

A.P.N. House et al. Transportation Research Interdisciplinary Perspectives xxx (xxxx) xxx

10

surveillance and precautionary measures such as pitot probe covers are theminimum response required to ensure that Australia does not suffer a re-peat of the 1996 Caribbean tragedy.

Acknowledgements

We are extremely grateful to Brian Ford (Qantas) for making the printedpitot probes used in the experiment. Our thanks also to Dave Selby and Pe-ter Dunlop (BAC) for early enthusiasm for this project; Chris Burwell(Queensland Museum) for help with wasp identification; Chuck Daviesand Matt Molin (CASA) for supervising the risk and bow-tie analyses; RossWylie (Biosecurity Queensland) for discussions on biosecurity and waspecology; and Ailsa Kerswell, Jeff McKee, Susan House and two anonymousreviewers for helpful comments on earlier drafts of the manuscript. Thiswork was funded by BAC.

References

Amendt, J., Rodner, S., Schuch, C.-P., Sprenger, H., Weidlich, L., Reckel, F., 2017. Helicopterthermal imaging for detecting insect infested cadavers. Sci. Justice 57, 366–372. https://doi.org/10.1016/j.scijus.2017.04.008.

Australian Transport Safety Bureau, 2006. Rejected Takeoff, Brisbane Airport, Qld 19 March2006 VH-QPB Airbus A330–303 (Canberra).

Australian Transport Safety Bureau, 2008. Rejected take-off, Brisbane Airport, Qld, 19 March2006. Aviation Occurrence Investigation Report 200601453. Canberra.

Australian Transport Safety Bureau, 2016. Air data system failure involving Airbus A330–243,A6-EYJ, near Brisbane Airport, Qld on 21 November 2013. Aviation Occurrence Investi-gation AO-2013-212. Canberra.

Australian Transport Safety Bureau, 2018. Australian aviation wildlife strike statistics, 2008 –2017. Canberra.

Australian Transport Safety Bureau, 2018b. Airspeed indication failure on take-off involvingAirbus A330, 9M-MTK, Brisbane Airport, Queensland, on 18 July 2018. Preliminary in-vestigation report AO-2918-053. Canberra.

Beggs, J.R., Brockerhoff, E.G., Corley, J.C., Kenis, M., Masciocchi, M., Muller, F., Rome, Q.,Villemant, C., 2011. Ecological effects and management of invasive alien Vespidae. Bio-Control 56, 505–526. https://doi.org/10.1007/s10526-011-9389-z.

Boeing, 2003. Erroneous: flight instrument information – safety and guidance. Aero 29.Brinks, R., Landwehr, S., 2015. A new relation between prevalence and incidence of a chronic

disease. Math. Med. Biol. 32, 425–435. https://doi.org/10.1093/imammb/dqu024.Bureau d’Enquêtes et d’Analyses pour la sécurité de l’aviation civile, 2012. Final Report on the

accident on 1st June 2009 to the Airbus A330-203 registered F-GZCP operated by AirFrance flight AF 447 Rio de Janeiro – Paris. Le Bourget, France. p. 219.

Bureau of Meteorology, 2019. Climate data [WWW Document]. URL http://www.bom.gov.au/climate/data/ (accessed 9.20.19).

Cacciabue, P.C., Cassani, M., Licata, V., Oddone, I., Ottomaniello, A., 2015. A practical ap-proach to assess risk in aviation domains for safety management systems. Cogn. Technol.Work 17, 249–267. https://doi.org/10.1007/s10111-014-0294-y.

Carpenter, J.M., 1986. The genus Pachodynerus in North America (Hymenoptera: Vespidae:Eumeninae). Proc. Entomol. Soc. Washingt. 88, 575–577.

Civil Aviation Authority, 2017. Wildlife Hazard Management at Aerodromes: ReportedBirdstrikes 2012–2016. Gatwick, UK.

Civil Aviation Authority of New Zealand, 2017. Bird Incident Rate Report, January–March2017. Auckland, New Zealand.

Civil Aviation Safety Authority, 2018. Airworthiness bulletin 02-052 issue 4 – 3 may 2018,wasp nest infestation – Alert. Canberra.

Cleary, E.C., Dolbeer, R.A., 2005. Wildlife Hazard Management at Airports: A Manual for Air-port Personnel. 2nd edition. (Federal Aviation Authority/US Dept. of Agriculture).

Climate Change Agriculture and Food Security, 2019. Climate Change, Agriculture and FoodSecurity Climate Analogues Tool [WWW Document]. URL http://analogues.ciat.cgiar.org/index.html (accessed 6.20.19).

Dale, L.A., 2009. Personal and corporate liability in the aftermath of bird strikes: a costly con-sideration. Human Wildl. Conflicts 3, 216–225.

Davidson, A.D., Hewitt, C.L., Kashian, D.R., 2015. Understanding acceptable level of risk: in-corporating the economic cost of under-managing invasive species. PLoS One 10,e0141958.

Duquesne, B., Jacquot, A., Cote, S., 2016. Pitot probe performance covered on the ground. Saf.First 22, 6–13.

Emond, J.-P., Mercier, F., Nunes, M.C.N., 1999. In-Flight Temperature Conditions in the Holdsof a Widebody Aircraft, in: Refrigeration into the Third Millenium: 20th InternationalCongress of Refrigeration. IIR/IIF, Sydney.

Faulkner, K.T., Robertson, M.P., Rouget, M., Wilson, J.R.U., 2016. Understanding and manag-ing the introduction pathways of alien taxa: South Africa as a case study. Biol. Invasions18, 73–87.

Federal Aviation Authority, 2019. Wildlife strikes to civil aircraft in the United States 1990–2017. Federal Aviation Administration National Wildlife Strike Database Serial ReportNumber 24. Washington DC.

Finkelstein, M., 2008. Failure Rate Modelling for Reliability and Risk. Springer-Verlag, London.Finnoff, D., Shogren, J.F., Leung, B., Lodge, D., 2007. Take a risk: preferring prevention over

control of biological invaders. Ecol. Econ. 62, 216–222. https://doi.org/10.1016/j.ecolecon.2006.03.025.

A.P.N. House et al. Transportation Research Interdisciplinary Perspectives xxx (xxxx) xxx

Flight Safety Foundation, 1996. Military Boeing 707 strikes birds after liftoff; damage to en-gines no. 1 and no. 2 results in loss of power and impact with terrain. Accid. Prev. 53, 1–8.

Fraser, R.W., Cook, D.C., Mumford, J.D., Wilby, A., Waage, J.K., 2006. Managing outbreaks ofinvasive species: eradication versus suppression. Int. J. Pest Manag. 52, 261–268.

Freeman, B.E., Jayasingh, D.B., 1975. Population dynamics of Pachodynerus nasidens (Hyme-noptera) in Jamaica. Oikos 26, 86–91.

Hauptfleisch, M.L., Dalton, C., 2015. Arthropod phototaxis and its possible effect on birdstrike risk at two Namibian airports. Appl. Ecol. Environ. Res. 13, 957–965.

House, A.P.N., Ring, J.G., Shaw, P.P., 2020. Inventive nesting behaviour in the keyhole waspPachodynerus nasidens Latreille (Hymenoptera: Vespidae) in Australia, and the risk toaviation safety. https://doi.org/10.1101/2019.12.15.877274 Submitted for publication.

Hulme, P.E., 2009. Trade, transport and trouble: managing invasive species pathways in anera of globalization. J. Appl. Ecol. 46, 10–18. https://doi.org/10.1111/j.1365-2664.2008.01600.x.

Intergovernmental Panel on Climate Change, 2000. Special Report on Emissions Scenarios.Working Group III of Intergovernmental Panel on Climate Change, Cambridge.

Jackson, D.A., 2015. Concept of a Pitot tube able to detect blockage by ice, volcanic ash, sandand insects, and to clear the tube. Photonic Sensors 5, 298–303. https://doi.org/10.1007/s13320-015-0272-x.

Kontogiannis, T., Malakis, S., McDonald, N., 2017. Integrating operational and risk infor-mation with system risk models in air traffic control. Cogn. Technol. Work 19,345–361.

Marra, P.P., Dove, C.J., Dolbeer, R., Dahlan, N.F., Haecker, M., Whatton, J., Diggs, N.E.,France, C., Henkes, G.A., 2009. Migratory Canada Geese cause crash of US Airways Flight1549. Front. Ecol. Environ. 7, 297–301.

McDermott, S.M., Finnoff, D.C., 2016. Impact of repeated human introductions and the Alleeeffect on invasive species spread. Ecol. Model. 329, 100–111. https://doi.org/10.1016/j.ecolmodel.2016.03.001.

McKee, J., Shaw, P.P., 2016. The costs of wildlife strike to the aviation industry. An updatebased on airline data. Proc. World Bird Strike Science and Engineering Meeting, Tokyo.

Mendonca, F.A., Huang, C., Carney, T.Q., Johnson, M.E., 2018. Assessing the risks: an analysisof wildlife-strike data at the three busiest Brazilian airports (2011-2016). Int. J. Aviat.Aeronaut. Aerosp. 5.

Meredith, L.A., 1849. Sketch of some members of the wasp family in Van Diemen’s land.Chambers’ Edinburgh J. 265, 51–52.

Mier-y-Teran-Romero, L., Tatem, A.J., Johansson, M.A., 2017. Mosquitoes on a plane:disinsection will not stop the spread of vector-borne pathogens, a simulation study.PLoS Negl. Trop. Dis. 11.

Morris, A.S., Langari, R., 2016. Measurement and Instrumentation: Theory and Application.Academic Press.

National Aeronautic and Space Administration, 2019. Kennedy Space Center Reliability[WWW Document]. URL https://kscddms.ksc.nasa.gov/Reliability/ (accessed 9.20.10).

National Transport Safety Bureau, 1981. Florida Commuter Airlines Inc. Douglas DC-3,N75KW Grand Bahama Island, Bahamas September 12, 1980: Accident Report.Washington DC.

Perry, G., 2002. Wheel-well and cargo compartment temperatures of large aircraft in flight:implications for stowaways. Aviat. Sp. Environ. Med. 73, 673–676.

11

Pluess, T., Cannon, R., Jarošik, V., Pergl, J., Pyšek, P., Bacher, S., 2012a. When are erad-ication campaigns successful? A test of common assumptions. Biol. Invasions 14,1365–1378.

Pluess, T., Jarošik, V., Pyšek, P., Cannon, R., Pergl, J., Breukers, A., Bacher, S., 2012b. Whichfactors affect the success or failure of eradication campaigns against alien species? PLoSOne 7, e48157.

Rayman, R., 2006. Aircraft disinsection. Aviat. Sp. Environ. Med. 77, 733–736.Rodriguez-Saona, C., Stelinski, L., 2009. Behavior-modifying strategies in IPM: Theory

and practice. In: Peshin, R., Dhawan, A.K. (Eds.), Integrated Pest Management: Inno-vation-Development Process. Springer Science and Business Media BV, Berlin,pp. 263–315.

de Ruijter, A., Guldenmund, F., 2016. The bow-tie method: a review. Saf. Sci. 88,211–218.

Russell, R.C., 1987. Survival of insects in the wheel bays of a Boeing 747B aircraft on flightsbetween tropical and temperate airports. Bull. World Health Org. 65, 659–662.

Ryan, J., 2018. Race to Hawaii: The 1927 Dole Air Derby and the Thrilling First Flights thatOpened the Pacific. Chicago Review Press, Chicago.

Samaková, A., Hadrava, J., Skuhrovec, J., Jansta, P., 2019. Host population density and pres-ence of predators as key factors influencing the number of gregarious parasitoid Anaphesflavipes offspring. Sci. Reports 9. https://doi.org/10.1038/s41598-019-42503-4.

Stone, C., Mohammed, C., 2017. Application of remote sensing technologies for assessingplanted forests damaged by insect pests and fungal pathogens: a review. Curr. For. Re-ports 3, 75–92.

Suckling, D.M., Stringer, L.D., Stephens, A.E.A., Woods, B., Williams, D.G., Baker, G., El-Sayed, A.M., 2013. From integrated pest management to integrated pest eradication:technologies and future needs. Pest Manag. Sci. 70, 179–189.

Takahashi, H., Abe, K., Takahata, T., Shimoyama, I., 2018. Experimental study of the aerody-namic interaction between the forewing and hindwing of a beetle-type ornithopter. Aero-space 5. https://doi.org/10.3390/aerospace5030083.

Tobin, P.C., Robinet, C., Johnson, D.M., Whitmire, S.L., Bjornstad, O., Liebhold, A.M., 2009.The role of Allee effects in gypsy moth, Lymantria dispar (L.), invasions. Popul. Ecol. 51,373–384.

Transport Canada, 2017. Wildlife Strikes at Canadian Airports: 2017 Annual Report(Ottawa).

Turchi, L., Derijard, B., 2018. Options for the biological and physical control of Vespa velutinanigrithorax (Hym.: Vespidae) in Europe: a review. J. Appl. Entomol. 142, 553–562.

Verbeek, M., Jentink, H., 2012. Optical Air Data System Flight Testing. Amsterdam.Weller, J.R., 2014. Wildlife hazards at airports: regultory oversight, program evaluation and

strike reporting. Proceedings of the Wildlife and Foreign Object Debris (FOD) Workshop.International Civil Aviation Organization, Cairo, Egypt.

Wylie, F.R., Griffiths, M., King, J., 2008. Development of hazard site surveillance pro-grams for forest invasive species: a case study from Brisbane. Australia. Aust. For.71, 229–235.

Yamane, S., Gusenleitner, J., Menke, A.S., 1996. Pachodynerus nasidens (Latreille) (Hymenop-tera, Vespidae), an adventive potter wasp new to Japan. Species Divers. 1, 93–97.

Zaviezo, T., Mills, N., 2000. Factors influencing the evolution of clutch size in a gregarious in-sect parasitoid. J. Anim. Ecol. 69, 1047–1057.