Behavioral Ecology and Sociobiology

, Volume 68, Issue 3, pp 457–465

Switching to Plan B: changes in the escape tactics of two grasshopper species (Acrididae: Orthoptera) in response to repeated predatory approaches

Original Paper

DOI: 10.1007/s00265-013-1660-0

Cite this article as:
Bateman, P.W. & Fleming, P.A. Behav Ecol Sociobiol (2014) 68: 457. doi:10.1007/s00265-013-1660-0

Abstract

Most studies examining escape behaviour have considered single approaches and single fleeing responses; few have considered how organisms’ response is influenced by persistent pursuit. We explored fleeing behaviour of two grasshopper species to test whether they modified escape behaviour when approached repeatedly. Schistocerca alutacea did not increase flight initiation distance (FID) upon repeated approach but fled farther. Psinidia fenestralis increased its FID on the second approach but decreased its flight distance over successive escapes. Both species showed a bimodal pattern of flight direction, either flying directly away or flying perpendicular to the direction of the observer’s approach. Neither species showed a significant pattern of flight direction or change in flight direction with successive escapes. Most (88 %) P. fenestralis initially landed on sand, but after repeated approaches an increasing proportion landed in grass and hid. Both species therefore changed escape behaviour with persistent pursuit but used different tactics, suiting their flight ability or camouflage, and optimised habitat use. Three grasshopper species have now been examined for responses to repeated approach by predators and all show different tactics supporting escape decision theory. Our results emphasise the variety of escape responses across species and how the dynamic nature of escape responses vary according to an animal’s situation. Rather than single optimum escape options, each grasshopper species shows a range of responses, which vary with risk from persistent predators. Although grasshoppers provide an excellent model, it would be profitable to examine responses of a range of species according to levels of predation risk.

Keywords

Predation Risk-sensitivity hypothesis Repeated pursuit 

Introduction

When an animal is approached by a potential predator, it has options to remain (especially if the species is camouflaged) or it can flee that location. A predator may either shift its attention elsewhere after unsuccessfully chasing potential prey or it may persistently pursue the prey, making repeated attempts to catch it. Prey that are repeatedly pursued are likely to be monitoring the predator’s behaviour during the pursuit sequence. With each successive attempt, the pursued animal should update their risk assessment based on information either relating to the predator’s persistence (i.e. by comparison with previous experience of approaches by predators) or because they themselves have a finite and decreasing amount of energy to invest in the whole series of escape attempts. If the approaching predator persists in pursuing the individual, this might be an indication that tactics used are no longer effective. At some stage during repeated pursuit, it may, therefore, be adaptive for the animal to modify its behaviour.

This plasticity in escape responses by the same individuals over time according to varying risk perception has rarely been examined under the economic escape models initiated by Ydenberg and Dill’s seminal paper. These models (Ydenberg and Dill 1986 and more recent iterations thereof, e.g. Cooper and Frederick 2007) have proved useful in predicting how organisms across a range of taxa make economic ‘decisions’ of when to flee by assessing costs of predation risk against costs of fleeing (e.g. interrupting foraging, energy expended, exposing location). However, as Cooper (2006a) pointed out, most empirical studies testing these models treat each approach and fleeing event by a predator and prey as ‘static,’ assuming that the risk assessed by the prey did not change as the predator approached. Approach and flight events are not static, however, and prey can dynamically ‘update’ risk assessment even during a pursuit (Cooper 2006a). This adaptive shift in escape behaviour has been demonstrated with non-sequential attacks or situations that differ in risk (e.g. Tikkanen et al. 1996; Kotiaho et al. 1998; Bateman and Fleming 2006, 2013b), but few studies have made a point of examining repeated escape attempts by individuals. Carrete and Tella (2010) examined flight initiation distances (FID) in burrowing owls that were approached repeatedly, and found consistency in individual responses over three to five approaches that were separated in time by 3 days. Apart from the potential for the birds to become habituated to human approach -although while most individuals did not change their responses, there were as many birds decreasing as increasing their FID across successive trials-, these discrete approaches represented the same level of risk. If we can demonstrate similar responses to constant levels of risk, the question remains how these individuals will respond when the level of risk is increased.

Using the dynamic situation of repeated pursuit as a framework within which to examine escape behaviour (Fig. 1), it is predicted that an animal being persistently pursued (i.e. experiencing an increased level of risk exposure):
Fig. 1

Potential options for altering behaviour in response to approach from a persistent predator (e.g. human observer)

  1. 1.

    Will demonstrate a longer FID when fleeing (Cooper 2006b; Cooper et al. 2009; Cooper and Avalos 2010);

     
  2. 2.

    Will flee a longer distance (Martín and López 2003; Cooper 2006b; Cooper et al. 2009) or move away at greater speed;

     
  3. 3.

    Is more likely to demonstrate protean fleeing behaviour, i.e. ‘jinking’: unpredictably changing direction of flight path when repeatedly pursued. In his study of grasshopper Dissosteira carolina escape, Cooper (2006b) refers to this pattern of escape behaviour but does not explicitly measure it. Protean behaviour in grouped prey is likely to reduce incidence of successful predation (Jones et al. 2011), but few studies have considered unpredictability of individual prey response between attacks, whether immediately sequential or not (Domenici et al. 2008).

     
  4. 4.

    Is more likely to resort to cover (Cooper 1997; Martín and López 2003; Cooper et al. 2009), will use more protective cover (Martín and López 2003) and will stay longer under cover (Cooper 1998; Martín and López 2003; Rodríguez-Prieto and Fernández-Juricic 2005; Cooper and Avalos 2010).

     
The question of how animals respond to repeated approach has only been previously investigated by a handful of studies (Table 1 and references therein). Partly, this may be because many species do not flee as a series of discrete events (e.g. two studies, Stankowich (2008) and Reimers et al. (2009), report variation in FID response of ungulate species exposed to ‘sequential’ approaches by humans, which are not necessarily repeated pursuit but a function of human density). Discrete escape events are, however, typical of lizards, frogs, many invertebrates and some small mammals, that flee in a series of leaps or short runs. Using grasshoppers as model organisms, we predicted that in response to repeated approach, animals would demonstrate:
Table 1

Review of studies that have examined the response of animals to approach from a persistent predator (human observer)

Species

‘Plan A’: Initial response to human observer approach

No. repeated approaches

‘Plan B’: Change in behaviour with repeated approach

Reference

↑ FID

↑ Flight distance

Change direction of escape

↑ Use of refuge/cover

Use more protective refuge/cover

↑ Latency to emerge from cover or resume activity

Lizards

Lacerta monticola

Flee to refuge

10 (\( \overline{\mathrm{x}} \) =8.9)a

     

Yb

Martín and López (1999)

Acanthodactylus erythrurus

Flee to cover

3

 

Yc

 

Y

Y

 

(Martín and López 2003)

Podarcis muralis

Flee to refuge

3

   

Y

  

Martín and López (2003)

Eumeces laticeps

Flee to cover

2

   

Y

  

Cooper (1997)

2

     

Y

Cooper (1998)

Podarcis lilfordi

Flee to cover

2

Y

Y

 

Y

  

Cooper et al. (2009)

Sceloporus jarrovi

Flee to refuge

2

Y

    

Y

Cooper and Avalos (2010)

Sceloporus virgatus

Flee to refuge

2

     

Y

Cooper (2012)

Frogs

Rana ibericad

Flee to refuge (water)

3

N

    

Y

Rodríguez-Prieto and Fernández-Juricic (2005)

Grasshoppers

Dissosteira carolina

Flee

2

Y

Y

    

Cooper (2006b)

Schistocerca alutacea

Flee

2–6 (\( \overline{\mathrm{x}} \) =2.9)e

N

Y

N

N

N

 

This study

Psinidia fenestralis

Flee

2–11 (\( \overline{\mathrm{x}} \) =5.9)e

Y

N(↓)

N

N

Y

 

This study

↑ indicates an increase and ↓ indicates a decrease; Y indicates a behavioural difference between initial and subsequent escape attempts; N indicates there is no difference; blank indicates factors that have not been tested/measured

aAuthors aimed for 10 approaches, but some individuals could not be approached 10 times, mean (\( \overline{\mathrm{x}} \)) given

bUnless thermal costs of using refuge is high (then = or ↓ time)

cBetween first and third approach

dObserved areas near a stream and recorded return to this site over time

eRange and mean given

  1. 1.

    A longer FID in response to perceived higher risk (Ydenberg and Dill 1986; Cooper and Frederick 2007);

     
  2. 2.

    Longer distances fled in response to perceived higher risk (Cooper et al. 2006);

     
  3. 3.

    Changes in direction between escape attempts, i.e. protean escape tactics (Humphries and Driver 1970); and

     
  4. 4.

    Are more likely to resort to cover when being pursued.

     

We tested these predictions using Orthoptera as model organisms as they demonstrate a wide range of potential escape responses (Cooper 2006b; Bateman and Fleming 2009, 2011a, 2013b), including variation in FID (prediction 1), alter the distance they invest in flight to escape the approaching predator (prediction 2), can fly either directly away from the approaching predator or at an angle lateral to the predator’s approach (prediction 3) and are able to take cover and therefore evade pursuit (prediction 4).

Methods

In this study, we examined the escape behaviour of two grasshopper species that were approached multiple successive times by a ‘persistent predator’ (simulated by a human observer). Although humans do use grasshoppers as food worldwide (e.g. Ruddle 1973; Banjo et al. 2006), Orthoptera also fall prey to many other species of tall, bipedal, stalking predators such as herons (Ardeidae), cranes (Gruidae) and storks (Ciconidae) (e.g. Fogarty and Hetrick 1973; Scott 1984; Petersen et al. 2008). Orthoptera are, therefore, likely to have adaptations to escape from predators of such an appearance.

We collected data on escape behaviour of fully winged, adult specimens of two grasshopper species that were common at the study sites and readily identifiable:
  1. 1.

    Schistocerca alutacea is a medium to large (30–54 mm body length) cyrtacanthridine acridid found in open woods and pastures on woodland edges (Capinera et al. 1997). Escape behaviour data (n = 35 individuals) were collected in a grassy clearing (200 × 200 m) (mean height of grass 95 ± 14 cm) in a lightly wooded (pines and oak scrub) wildlife management area in central Florida, USA (27°31′ N 81°24′ W). S. alutacea tends to fly in a straight, sometimes quite high, parabola.

     
  2. 2.

    Psinidia fenestralis is a small (20–33 mm body length) oedipodine acridid found in sandy areas, on which it is cryptic (Capinera et al. 2001). Escape behaviour data (n = 32 individuals) were collected in a fallow field bordered by orange groves and suburban development in central Florida, USA (27°36′ N 81°29′ W). The site was divided into approximately equal strips of bare sand and vegetation: the field was a cleared orange grove (approx. 85 × 85 m) made up of over 80 strips of bare sand (72.3 ± 6.2 cm wide) separated by lines of growing and standing grass (68.7 ± 9.3 cm wide) running north–south. P. fenestralis tends to fly low and quite fast and rarely changes direction in flight; any changes tend to be when it tumbles on landing (Kral 2010). Based on preliminary walks through the field to flush grasshoppers, P. fenestralis appeared to preferentially rest on the furrows of bare sand on which they are cryptic, as has been previously observed in this species (Capinera et al. 2001) and other cryptic acridids (Eterovick et al. 1997; Kral 2010).

     

For both species, we collected escape behaviour data in the same way: one observer (PWB) walked slowly (approx. 1 m/s for S. alutacea and 0.5 m/s for P. fenestralis, reflecting the difference in distances fled by the two species) through the area until a grasshopper flushed from directly in front of the observer (grasshoppers that flushed tangential to the observer’s path were not used). Because the predation risks simulated would differ for each species (due to approach speed), we do not make a direct comparison of the data between species but do contrast their overall responses. To minimise the chances of re-sampling the same individuals, we did not retrace our steps and moved to a new part of each field over successive recordings. Both study sites were grassed areas that were clear of tall vegetation that would influence distance to cover.

Individuals of both species flew to escape our approach (the first flush was done blind to where the grasshopper was). The angle of departure of the grasshopper relative to the approach by the observer was estimated on a 180° scale (0° was directly toward the observer, 180° directly away; left and right angles were pooled), and the FID and flight distance were measured from practised paces of different lengths (±1 m for S. alutacea, ±0.3 m for P. fenestralis, reflecting the differences in size and average flight distance between the two species) as is typical of similar studies (e.g. Cooper 2006b; Bateman and Fleming 2011a, b). It was usually obvious where the grasshopper had landed, even if it was not immediately visible due to landing on a cryptic surface. If there was any doubt that it was the focal grasshopper, that trial was terminated. Without stopping and at the same speed, we then recorded the same data for subsequent approaches; i.e. the grasshopper was flushed again, producing another FID and flight distance measure. This was repeated until either the grasshopper had flown so far that on reaching that point it could not be found and flushed (in the case of the strong flyer S. alutacea), or when it did not flush because it had moved deep into a grass clump (in the case of P. fenestralis).

Both species generally escaped away from the approaching observer in straight line flights rather than towards him (i.e. angles >90°), with animals tending to take a path that was either in the same direction (away from his direction of approach), or at right angles (perpendicular to his direction of approach). These escape directions were post hoc classified as either directly away from the observer (>130°) or perpendicular to his approach (<130°) due to a natural break in the bimodally distributed data at this point. We also calculated whether an individual switched direction between successive escape attempts (yes) or maintained flight in the same direction as previous escape attempts (no: did not change direction).

Analyses

Data for each species were analysed independently. Distance measures were log-transformed to meet the assumptions of normality (Shapiro–Wilk W Test). FID (log-m), flight distance (log-m), direction of escape (away or perpendicular), change in direction between escape attempts (yes or no) and substrate selected (P. fenestralis only; landed either on bare ground or in grass) were tested using a Type III mixed-model ANOVA (each as separate continuous or bimodal dependent variables) (SPSS Statistics 21.0). We included animal ID as a random effect (to take into account repeated measures on each individual) and escape attempt as an ordinal fixed effect. We used a random slope model which would allow calculation for each individual of the changes in data over escape attempts. We compared substrate selection with that available by χ2comparing the actual substrate selected with expected values calculated according to the proportion of bare sand (51 % area) and grass (49 % area) at the site.

Values are presented as means ± 1SD.

Results

The 35 individual S. alutacea were approached an average of 2.9 ± 1.4 (range 1–6) times before being lost. FID (overall average: 2.3 ± 1.3 m) was not influenced by how many times the individual had been approached (F5, 56 = 0.62, p = 0.684; Fig. 2a) but flight distance was significantly affected by escape attempt (F5, 61 = 10.10, p < 0.001; Fig. 2b). Because these grasshoppers increased their flight distance with repeated approaches, they were generally lost when they flew too far away to relocate the individual.
Fig. 2

Flight initiation distance (FID) and flight distance in two grasshopper species where individuals were approached multiple times (each approach termed an ‘escape attempt’) by a human observer. Total numbers of individuals for each escape attempt are given under the x-axis. Box plots show the mean, standard deviation (box) and range of the raw data (statistical analyses were performed on log-transformed values)

Overall, almost equal proportions of S. alutacea escapes were perpendicular to (56 %) or directly away (44 %) from the observer’s direction of approach (Fig. 3a, c). The majority (63 %) of S. alutacea moved away perpendicularly to the experimenter’s direction of approach on their first escape attempt. There was a trend for a change to moving directly away over subsequent escape attempts (F5, 67 = 2.19, p = 0.066 Fig. 3c). For 29 % of the 62 successive escape attempts followed, S. alutacea changed direction between escape attempts, but there was no effect of escape number (F4, 42 = 2.03, p = 0.108) on whether individuals changed direction or not.
Fig. 3

a, b Angle of flight taken by two species of grasshopper when they were repeatedly approached by an observer (position shown by the binoculars symbol). A total of a 102 escape attempts for 35 S. alutacea and b 189 escape attempts for 32 individual P. fenestralis are shown. Rings on the circle plots represent groups of 10 individuals, and the data were pooled for every 5° increment. Escapes to the left and right were pooled within each species. c, d Direction of escape for each escape attempt. Values are the percentage of individuals followed for each escape attempt (total numbers under the x-axis)

The 32 P. fenestralis were approached an average of 5.9 ± 1.9 (range 1–11) times. FID (overall average: 0.9 ± 0.4 m) was affected by escape attempt number (F10, 149 = 4.70, p < 0.001), with animals having longer FID on the second approach than other escape attempts (Fig. 2c). Flight distance was also significantly affected by escape attempt number (F10, 153 = 2.05, p = 0.032), decreasing over repeated escape attempts (Fig. 2d).

Almost equal proportions of escape attempts for P. fenestralis were perpendicular to the approaching observer (47 %) or directly away (53 %) from his approach, and escape direction did not change significantly over successive escape attempts (F10, 176 = 1.66, p = 0.093; Fig. 3b, d). Forty-one percent of 155 successive escape attempts involved a change in direction, but there was no indication that P. fenestralis were more likely to switch between moving directly away or to the side over successive escapes (F9, 145 = 1.138, p = 0.340).

Of the 32 first escape attempts by P. fenestralis followed, 88 % were to bare sand and the remainder to grass. As the field was divided into nearly equal rows of bare sand (51 % area) and grass (49 % area), the data suggested that the grasshoppers were not landing randomly, but initially selecting the sand substrate to land (χ21 = 17.06, p < 0.001). The observer did not preferentially walk on either substrate but crossed over them depending on the direction fled by the focal grasshopper. Over successive escape attempts, an increasing number of animals moved towards grass refuge (F10, 178 = 3.46, p < 0.001, Fig. 4) where they would move into the grass clump and become impossible to locate or flush.
Fig. 4

Substrate P. fenestralis landed on when approached multiple times by an observer

Discussion

These results broadly supported our predictions regarding the responses of Orthoptera prey to persistent pursuit. Both S. alutacea and P. fenestralis showed quantitative changes in their escape behaviour with repeated pursuit. While P. fenestralis showed an increase in FID for its second escape attempt, S. alutacea showed no change in FID with persistent pursuit. Instead, S. alutacea flew farther with persistent pursuit but P. fenestralis showed shorter distances fled over successive escape attempts.

Our first prediction based on persistent pursuit representing increased risk, that grasshopper FID would increase with repeated approaches, was supported for P. fenestralis (on the second approach, but not further subsequent approaches) but not S. alutacea. This overall lack of increase in FID is surprising as it is generally assumed across taxa that a trend of increasing FID would occur with a higher perception of risk, which should increase with persistent approach (Ydenberg and Dill 1986; Cooper and Frederick 2007). However, we showed that these grasshoppers had a range of other antipredator responses which may have accounted for this overall lack of response in FID. Additionally, FID may be determined to some degree by visual acuity or speed of cognitive processing (e.g. Dukas 1998; Bateman and Fleming 2013a; Lee et al. 2013). FID may simply reflect a response generated by a threshold size and/or speed of a ‘looming’ image on the retina (Rind and Simmons 1992; Javůrková et al. 2012). For example, hiding responses (squirrelling to the other side of a branch from the approaching object) in locusts Locusta migratoria stimulated by expanding shapes occur only when the expanding image has exceeded a threshold visual angle of 8–9.5° (Hassenstein and Hustert 1999). No difference in FID between situations with differing levels of apparent ‘risk’ may, therefore, simply result from the lack of differences in visual stimulus.

Our second prediction that grasshoppers would increase flight distance in response to repeated approaches differed between the two species. P. fenestralis showed a decrease in flight distances for successive escape attempts which could reflect decreasing energetic resources for each successive flight. By contrast, S. alutacea—a strong flyer—fled farther in later escapes, such that the pursuer failed to keep up with them. Increased flight distance may reflect increased risk perception due to persistent pursuit and the animals investing more in each successive escape attempt (Cooper et al. 2006). Fleeing farther with repeated approaches has also been observed in another grasshopper D. carolina (Cooper 2006b), and various lizard species (Martín and López 2003; Cooper et al. 2009), although most previous studies only tested two or three approaches (Table 1) and therefore may not be exhausting the animal’s energetic resources with these escape attempts.

Oedipodine grasshoppers (such as P. fenestralis and D. carolina) have coloured hind wings and males often display by flying up, fluttering these wings and crepitating (Capinera et al. 2001). This advertising flight attracts the attention of potential mates but also makes the animals more visible to visually hunting predators. Tellingly, Belovsky et al. (1990) report dramatically reduced predation rates for oedipodine grasshoppers tethered out in the field if their wings are glued shut, and that, of several grasshopper species, flightless grasshopper species are generally less vulnerable to predation. Furthermore, many orthopteran species have flightless forms, with populations only producing winged forms when crowding induces dispersal (e.g. Masaki and Walker 1987; Higaki and Ando 2003), suggesting flight is not vital except in particular circumstances. Increasing flight distance would mean that a grasshopper spends longer in the air which can make it more obvious to a visually hunting predator. Grasshoppers therefore have another potential reason to escape a persistent terrestrial predator: to reduce the time spent in the air exposed to other opportunistic predators and airborne flight may not be the preferred escape option for all species.

Like other oedipodine grasshoppers, P. fenestralis is a cryptic, arenicolous, species that appears to be adapted for flat, open habitats (Kral 2010). The optimal strategy for cryptic prey appears to be either to run immediately on perceiving the predator, or to run only when the predator initiates a chase (Broom and Ruxton 2005). As also observed for D. carolina (Cooper 2006b), P. fenestralis can seem to vanish immediately upon landing on sand due to their crypsis. This crypsis can be enhanced by disorientation of the predator by the flashing and then disappearance of the coloured underwings upon landing, particularly when it occurs laterally to the predator’s path and is only seen by peripheral vision. Flight perpendicular to the predator may therefore be an adaptive tactic by oedipodine grasshoppers that enhances crypsis. Persistent pursuit, however, would indicate that the ‘predator’ has at least approximate information on where the grasshopper has landed, and crypsis is no longer a safe strategy. Under persistent pursuit, P. fenestralis showed increased reliance on escape to vegetation cover.

We recorded a distinct bimodal distribution in escape directions for both species tested, with animals moving either directly away from or perpendicularly to the observer’s direction of approach. Unpredictable, ‘jinking’ manoeuvres in flight, or protean escape behaviour, where the prey does not react predictably to each attack, preventing the predator from learning to out-manoeuvre the prey (Domenici et al. 2011a, b) would also increase the difficulty of a predator pursuing its prey. While P. fenestralis tended to shift to more perpendicular flight paths, generally, the cyrtacanthridine S. alutacea (with non-contrasting, translucent yellow-tinted underwings) abandoned their initial strategy of moving perpendicularly, instead moving greater distances, directly away from the persistent observer. However, although we showed trends in the direction of escape with escape attempts, these patterns were not statistically significant for either species (S. alutacea: p = 0.066 and P. fenestralis: p = 0.093). We therefore do not have any support for our third prediction that changes in direction in fleeing from a predator are more likely to occur with persistent pursuit.

Supporting our fourth prediction, we found that persistent pursuit would eventually result in P. fenestralis landing on clumps of growing grass and climbing down the stems to hide. Using grass may be the least-preferred option for these animals since there are costs of hiding in grass. Although oedipodine grasshoppers feed on grass, they prefer not to stay in grass clumps (which may compromise their camouflage) but instead remove pieces from grass stems to eat and then return to open areas (Belovsky et al. 1990). Pitt (1999) found that when bird predators were excluded, grass-using grasshoppers perched higher in grass, and when ground predators were excluded, they perched lower in the grass. When all predators were present, they used intermediate height perches. Grass, then, is not likely to be a ‘safe haven’ for P. fenestralis, just safer than being exposed to a persistent predator at that particular time. Similarly, Lacerta monticola lizards run to cover in rock crevices, although there are thermal costs for the lizard of using this refuge (Martín and López 1999), while Podarcis muralis lizards risk exposure to predators that also use the same refuges (Amo et al. 2004).

In this study, we found that in response to persistent pursuit by a putative predator, two grasshopper species altered their behaviour from their initial tactics (switching from ‘Plan A’ to ‘Plan B’; Table 1), presumably reflecting their perception of higher risk due to repeated approaches. When they failed to deter a persistent predator by preferred escape methods, they shifted to potentially riskier behaviour (e.g. increasing FID, flying farther). These shifts in escape tactics are likely to also occur in other taxa (as indicated by results for lizards; see Table 1), and we encourage further research in other taxa on the economic escape decisions of prey approached multiple sequential times.

Jones et al. (2011) point out that we are undergoing a paradigm shift in how we think about escape behaviour. We are shifting from regarding organisms as converging on an optimum strategy, with variance representing noise about this optimum strategy, to now recognising the value of this variance, in that it is fuel for selection and could also represent different strategies for different individuals (Carrete and Tella 2010) and for escaping from different predators. The importance of looking at repeated pursuit is that it shows how individuals of model species can adaptively shift between potential escape strategies over time according to the dynamic conditions they encounter. This approach may allow us to separate individual variation in response from the differences due only to varying levels of risk.

Copyright information

© Springer-Verlag Berlin Heidelberg 2013

Authors and Affiliations

  1. 1.Department of Environment and AgricultureCurtin UniversityPerthAustralia
  2. 2.Veterinary and Life SciencesMurdoch UniversityPerthAustralia

Personalised recommendations