on May 22, 2018http://rsif.royalsocietypublishing.org/Downloaded from

rsif.royalsocietypublishing.org

ResearchCite this article: Parle E, Dirks J-H, Taylor D.

2016 Bridging the gap: wound healing in

insects restores mechanical strength

by targeted cuticle deposition. J. R. Soc.

Interface 13: 20150984.

http://dx.doi.org/10.1098/rsif.2015.0984

Received: 12 November 2015

Accepted: 15 March 2016

Subject Category:Life Sciences – Engineering interface

Subject Areas:biomechanics, bioengineering

Keywords:insects, cuticle, wound, repair, strength,

toughness

Author for correspondence:David Taylor

e-mail: dtaylor@tcd.ie

& 2016 The Author(s) Published by the Royal Society. All rights reserved.

Bridging the gap: wound healing ininsects restores mechanical strengthby targeted cuticle deposition

Eoin Parle1, Jan-Henning Dirks2 and David Taylor1

1Trinity Centre for Bioengineering, Trinity College the University of Dublin, Dublin, Ireland2Max Planck Institute for Intelligent Systems, Stuttgart, Germany

If an insect is injured, can it repair its skeleton in a manner which is mechani-

cally strong and viable? Previous work has described the biological

processes that occur during repair of insect cuticle, but until now, there

has been no biomechanical assessment of the repaired area. We analysed

the biomechanics of the injury repair process in the desert locust (Schistocercagregaria). We show that after an incision, a healing process occurred which

almost doubled the mechanical strength of locust tibial cuticle, restoring it

to 66% of the original, intact strength. This repair process occurred by tar-

geted cuticle deposition, stimulated by the presence of the injury. The cut

surfaces remained unrepaired, but a patch of endocuticle was deposited,

reinforcing the area and thus increasing the effective fracture toughness.

The deposition rate of endocuticle inside the tibia increased fourfold com-

pared with uninjured controls, but only on the dorsal side, where the

incision was placed. The limb is highly loaded during jumping, so this par-

tial restoration of strength will have a profound effect on the fitness of the

insect. A finite-element model provided insights into the mechanics of the

repair, predicting that the patch material reaches its ultimate strength

before the fracture toughness of the existing cuticle is exceeded.

1. IntroductionThe exoskeletal body parts of arthropods are made from cuticle: an acellular

chitin/protein composite material, mainly comprising two histological layers:

an outer layer of hard, stiff exocuticle and an inner layer of softer endocuticle.

Cuticle is secreted by a cellular epidermis which is attached to a basal

membrane. After moulting, during the normal growth process, the new, soft

exoskeleton quickly sclerotizes and hardens. From this point on, the overall

thickness of the cuticle gradually increases for several weeks by deposition of

further layers on the inner surface, orchestrated by the epidermal cells.

Insects and other arthropods respond to injury by a process not unlike that

found in mammals. First, coagulation occurs and melanin is formed, giving rise

to a clot that seals the wound and provides a scaffold for cell migration. Then,

cells in the underlying epidermis migrate to restore epidermal continuity across

the damaged area. Subsequently, these cells secrete new endocuticle on the

inner surface [1–7]. Many mammalian tissue types, such as skin and bone,

can be fully restored by tissue secretion and remodelling [8,9]. In adult insects,

however, damage in sclerotized parts of the cuticle exoskeleton is not comple-

tely healed: the new endocuticle creates a patch and the clot remains, creating a

kind of scar tissue at the wounded area [1,2]. No previous work has been done

to investigate the mechanical strength of this construct. Our aims for the present

work were

(a) To introduce damage in a form of an incision in the tibiae of adult locusts

and measure the reduction in bending strength thus caused.

(b) To characterize the effectiveness of the repair process by measuring bending

strength and toughness as a function of time after incision.

T1

L

F

dIX

T3T3

T2

0.7

0.6

0.5

0.4

forc

e (N

)

0.3

0.2

0.1

00 0.5 1.0

deflection (mm)1.5 2.0 2.5

(a)

(c)

(b)

Figure 1. (a) Cross section of the locust tibia: we made an incision of length L on the upper (dorsal) side. Cuticle thickness was measured on all sides: dorsal (T1);ventral (T2) and lateral/medial (T3). (b) Tibiae were loaded in cantilever bending applying a force F at a distance l from the fixed end; the incision was placed on the topsurface a distance x from the loading point such that (l – x) was typically 2 mm. Failure occurred at the incision (see right-hand photo). (c) Typical force/deflection results.(Online version in colour.)

rsif.royalsocietypublishing.orgJ.R.Soc.Interface

13:20150984

2

on May 22, 2018http://rsif.royalsocietypublishing.org/Downloaded from

(c) To measure cuticle deposition as a function of time, near

the injury site and remote from it.

(d) To investigate the mechanics of the repair process using a

finite-element modelling approach.

2. Material and methodsFemale adult desert locusts (Schistocerca gregaria) were kept in a

controlled 12 h (358C)/12 h (208C) day/night cycle and fed

with fresh plants and dried cereals ad libitum. Thirty-two insects

were randomly selected to receive injuries; a further 64 were used

as controls. Injuries were applied 30 days after the final moult.

Using a scalpel a transverse incision was made, perpendicular

to the tibial axis on the dorsal side of the hind tibia (approx.

5 mm distal to the tibiofemoral joint) such that the epidermis

was breached (figure 1a). The depth of the cut was controlled

by using a purpose-built slotted guide. All subjects received

the same size injury: average length L ¼ 780 mm (s.d. ¼ 60 mm,

n ¼ 32; figure 1a). Trial and error was used to find an ‘ideal’

cut length—that is, an incision deep enough to breach the epider-

mis but shallow enough, so that it would not extend further

during normal locomotion.

After incision, the locusts were separated into individual small

tubs (approx. 20� 1 � 10 cm), maintaining the same environ-

mental and feeding conditions. These tubs were large enough to

comfortably allow the insect to be fully mobile, and while jumping

was not eliminated entirely, separating the locusts from one

another discouraged the excessive use of the hind tibia (for jump-

ing, kicking, and fighting) often observed when they are housed

together. After a recovery period of between 1 and 50 days, the

injured legs were removed from the insects under sedation by

cutting just below the tibiofemoral joint. Immediately after

removal, legs were set in a well of polymer cement (Howmedicaw

Surgical Simplex P Bone Cement mixed at a 1 : 1 ratio for rapid

hardening). Cantilever bending tests were performed by applying

a position-controlled load to failure. Displacement was gradually

increased at a rate of 5 mm min21 using a mechanical testing

machine (Zwick/Roell Z005, Ulm, Germany), with the purpose-

built testing rig shown in figure 1b. Failure was characterized by

a sudden drop in load (figure 1c) and visual fracture at the incision

point (figure 1b). After testing, the samples were stored in a 3.7%

glutaraldehyde fixative for 24 h, after which they were preserved

in 70% ethanol solution.

Images were taken of the fracture surfaces of each sample

using a Zeiss Ultra Plus scanning electron microscope (5 kV;

Oberkochen, Germany). Cuticle dimensions (radius and thick-

ness at various locations around the circumference, cut length

and cut depth) were measured using FIJI software (an open

source image processing package based on IMAGEJ). Control sub-

jects received no incision but were otherwise treated in the same

manner; they were tested at times varying from 3 to 63 days

post-moult.

Standard engineering formulae were used to calculate

strength and toughness, as follows. The nominal stress to failure

sf was calculated using

sf ¼Fxr

I, ð2:1Þ

where F is the applied force at failure, x is the distance from the

failure location to the loading point, r is the average outer radius

of the tibia and I is the second moment of area of the cross sec-

tion, which is given (assuming a circular section of radius r

ANSYSR14.5

academic

0

1.500

3.000 (mm)

0

0.250

0.500 (mm)

A: static structuralmaximum principal stress 3type: maximum principal stressunit: MPatime: 115 February 2016 10.04 AM

94.277 max

–0.24945 min

83.77473.27162.76852.26541.76231.25920.75610.253

0

0.500

1.000 (mm)

Y

XZ

Y

XZ

ANSYSR14.5

academic

(b)(a)

(c)

Figure 2. (a) The finite-element model. (b) A close-up view showing the slot and also the layer of new endocuticle (in this case of thickness 20 mm). (c) Typical resultsfrom the model: maximum principal stress plotted on a cross section through the cut, showing high stresses in the layer of new cuticle. (Online version in colour.)

rsif.royalsocietypublishing.orgJ.R.Soc.Interface

13:20150984

3

on May 22, 2018http://rsif.royalsocietypublishing.org/Downloaded from

and thickness t) by

I ¼ p

4(r4 � ðr� tÞ4): ð2:2Þ

Fracture toughness Kc was calculated using

Kc ¼ Qsf

ffiffiffiffiffiffipap

, ð2:3Þ

where a is the half-length of the cut measured around the circum-

ference and Q is a factor which depends on a, r and t (given by

Takahashi [10]). Individual measurements of radius, thickness

and cut length were taken for each sample. These were used to

calculate the failure strength for each test, a unique shape

factor Q for each injured leg, and hence the fracture toughness

for each sample.

2.1. Finite-element modellingWe created computer models using ANSYS finite-element soft-

ware. The aim was not to reproduce the exact geometry of the

tibia, but rather to gain insights through a simplified model in

the form of a circular tube (figure 2) having dimensions similar

to the average values found experimentally: radius 440 mm,

thickness 100 mm with a slot of length L ¼ 780 mm located a dis-

tance 2 mm from the fixed end. The width of the slot was varied

from 10 to 100 mm, and we also created a slot which more closely

modelled the wedge-like shape of the scalpel used to make the

cut. Repair was simulated by depositing a layer of material, vary-

ing from 20 to 100 mm, on the inside surface of the tube.

All material was assumed to be linear-elastic and isotropic,

with a Young’s modulus of 8 GPa and a Poisson’s ratio of 0.3.

Testing was simulated by applying a force to create cantilever

bending. An appropriate size for the mesh elements was

chosen using critical distance theory [11] that allows one to esti-

mate the appropriate length scale on which to study material

fracture processes. Given a material tensile strength so and

fracture toughness Kc, the critical distance G is found by

G ¼ 1

p

Kc

so

� �2

: ð2:4Þ

Using the experimental values for fracture toughness and nom-

inal strength reported below gives G ¼ 0.101 mm, so we used a

mesh element size of 0.1 mm throughout the model. Stress inten-

sity (K ) values were calculated using a standard approach

(described in [12]) in which the length of the crack is extended

by a small amount da and the change in deformation of the

sample under load is used to calculate the change in stored

0.25

T1

T1

injured samples control

p = 0.0061

p = 0.077

p = 0.216T2

T2

T3

T3 <20days days

>20

0.15

0.20

0.10

cutic

le th

ickn

ess

(mm

)

grow

th r

ate

(mm

per

day

)

0.05

2.0

1.6

1.2

0.8

0

0.4

00 10 20

days since injury30 40 50 60

(a) (b)

Figure 3. (a) Cuticle thickness as a function of time post-injury for the 15 subjects in which repair occurred; (b) cuticle deposition rate post-injury at the injurylocation (T1) and elsewhere (T2, T3), compared with uninjured controls in the immediate post-moult modelling period (less than 20 days) and the later dormantperiod (more than 20 days). (Online version in colour.)

rsif.royalsocietypublishing.orgJ.R.Soc.Interface

13:20150984

4

on May 22, 2018http://rsif.royalsocietypublishing.org/Downloaded from

strain energy in the body, dW. K can then be found by

K ¼ Et

dWda

� �� �1=2

: ð2:5Þ

3. Results and discussionOur tests on control subjects that had not received any

incisions allowed us to establish normal baseline rates for

cuticle formation in the absence of injury. We found that, in

adults following their final moult, cuticle was deposited at

an average rate of 1.8 mm d21 (s.d. ¼ 2.4, n ¼ 24) in an initial

modelling phase which lasted on average 21 days. This rate

is in agreement with the results of previous work [13],

which examined growth up to 14 days. After 21 days, we

found that the deposition rate decreased considerably, to

0.4 mm d21 (s.d. ¼ 0.4, n ¼ 40). Therefore, for our injury

experiments, we used subjects which were 30 days post-

moult and thus in a relatively dormant state as regards cuticle

deposition.

Figure 3a shows results of cuticle thickness as a function

of time post-injury. Cuticle thickness was measured from

the fracture surfaces, at the following locations (figure 1a):

T1 (dorsal side, at the incision); T2 (ventral side, opposite

the incision); and T3 (medial and lateral sides, for which

results were averaged as no differences were found). Cuticle

thickness increased considerably with time at the injury site

(T1) but only slightly elsewhere. Regression analysis

showed that the increase in thickness over time at T1 is stat-

istically significant ( p ¼ 0.0061), whereas the changes in

T2 and T3 are not significant ( p ¼ 0.077 and 0.216, respect-

ively). As figure 3b shows, the rate of deposition near the

injury (1.6 mm d21; s.d. ¼ 1.3, n ¼ 13) was similar to that

found in uninjured control subjects immediately post-

moult, but much greater than in controls of the same age,

i.e. adults more than 20 days after the moult. However, the

measured deposition rates remote from the injury site, on

the other surfaces (T2 and T3) remained low after injury, com-

parable to uninjured controls of the same age (average

0.39 mm d21; s.d. ¼ 0.41, n ¼ 40). As a result of 20–50 days

of repair, the cuticle thickness on the dorsal side increased

to 146 mm (s.d. ¼ 36 mm, n ¼ 9), becoming significantly

thicker than the ventral side (109 mm, s.d. ¼ 11 mm, n ¼ 9,

t-test p ¼ 0.003). This is in marked contrast to uninjured

controls in which the dorsal and ventral sides developed

the same thickness (T1 ¼ 90 mm (s.d. ¼ 10 mm, n ¼ 40),

T2 ¼ 97 mm (s.d. ¼ 10 mm, n ¼ 40)). The lateral/medial thick-

ness T3 was similar in all cases (injured insects T3 ¼ 82 mm

(s.d. ¼ 16 mm, n ¼ 18); controls T3 ¼ 73 mm (s.d. ¼ 9 mm,

n ¼ 80)).

The above results imply the existence of a repair process

of cuticle deposition which is targeted to the damaged area.

However, this repair process did not always occur. From a

total of 32 subjects tested, repair (which we defined as visible

deposition of cuticle at the incision site—figure 4) occurred in

only 15 subjects. Of the others, there was no deposition pre-

sent in 12 cases; in the remaining four cases, it was unclear

from the scanning electron microscopy (SEM) observations

whether deposition had occurred or not so these samples

were discarded. A possible explanation for non-repair in

these subjects is that our scalpel cut may sometimes have

caused a relatively wide incision, and/or a relative displace-

ment of the cut surfaces, preventing the recovery of a

continuous epidermal layer (figure 4). As mentioned above,

the recovery of the epidermis is an essential first step, without

which the cascade of repair activities culminating in cuticle

secretion cannot occur. The results shown in figure 3 refer

only to the cases where repair occurred.

To compare mechanical behaviour, we followed pro-

cedures established in our previous work [14]. Bending

strength was defined as the nominal stress to failure, which

is the maximum tensile stress that would occur, at the failure

location, in an undamaged circular tube having the same

average diameter and thickness as measured for a given

sample. We also calculated an apparent fracture toughness.

Fracture toughness, Kc, is a measure of the ease with which

a crack can propagate in a material. Here, we calculated Kc

from the bending strength, assuming the incision to be a

sharp crack of the same size in a circular tube [10,14]. As

figure 5 shows, both strength and toughness approximately

doubled as a result of repair. Uninjured controls of

the same age had a strength of 172.4 MPa (s.d. ¼ 31.5 MPa,

n ¼ 40). In subjects where repair did not occur, the incision

lowered the average strength to 54.3 MPa (s.d. ¼ 19.1 MPa,

n ¼ 12) and this was unchanged over time. In subjects

where repair occurred, however, strength rose over the first

10 days and then remained approximately constant at an

average of 113.7 MPa (s.d. ¼ 16.4 MPa, n ¼ 15), which is

66% of the original, uninjured strength. This increase in the

nominal stress to failure was reflected in a similar increase

in the bending moment to failure, from 2.50 N mm (s.d. ¼

200 mm

50 mm

50 mm100 mm

100 mm 100 mm

(b)(a)

(c) (d )

( f )(e)

Figure 4. SEM photos of repaired and non-repaired cuticle. Panels (a) and (b) show the fracture surface of a tibia that has received an incision (dashed line) andsubsequently repaired the area by depositing cuticle (enhanced on the photo by red shading). At high magnification (b), there is a clear difference between thescalpel-cut surface and the fracture surface of the new cuticle. Panels (c) and (d ) show the fracture surface of a tibia in which repair did not occur. Panels (e) and ( f )are longitudinal sections of tibiae 30 days after receiving incisions. In (e), the incision (dashed line, arrowed) caused a relative displacement of the two sides ofthe cuticle, which are indicated by white lines. No repair occurred. In ( f ), there was no such displacement and new cuticle formed (red area) to repair the injury.(Online version in colour.)

rsif.royalsocietypublishing.orgJ.R.Soc.Interface

13:20150984

5

on May 22, 2018http://rsif.royalsocietypublishing.org/Downloaded from

250 9

8

7

6

5

4

control

injured(no repair)

injured(repaired)

3

2

1

0

200

150

100

failu

re s

tren

gth

(MPa

)

appa

rent

fra

ctur

eto

ughn

ess

(MPa

÷m)

50

0

Figure 5. Results for strength and fracture toughness of three groups: injured tibiae for which no repair occurred (tested 20 – 50 days after incision), those for whichrepair did occur (also tested after 20 – 50 days) and uninjured controls (tested 20 – 63 days post-moult). Error bars indicate standard deviation.

rsif.royalsocietypublishing.orgJ.R.Soc.Interface

13:20150984

6

on May 22, 2018http://rsif.royalsocietypublishing.org/Downloaded from

1.63 N mm, n ¼ 12) to 4.71 N mm (s.d. ¼ 0.52 mm, n ¼ 15).

The tibia will experience high bending moments in vivowhile walking, and is loaded almost exclusively in bending

during activities such as kicking and jumping [15,16].

In previous work [14], we measured a fracture toughness

for uninjured cuticle using the same protocol, finding a value

of 4.12 MPap

m (s.d. ¼ 0.4, n ¼ 9) for fresh, fully hydrated

cuticle and 2.06 MPap

m (s.d. ¼ 0.6, n ¼ 9) for dried cuticle.

In this study, we found a value of 3.07 MPap

m (s.d. ¼ 0.99,

n ¼ 12) in cases where no repair occurred (figure 5), which

falls between the above two values, probably because a cer-

tain amount of dehydration took place in the area around

the incision. Repair increased the measured toughness more

than twofold, to 7.01 MPap

m (s.d. ¼ 1.32, n ¼ 15). This was

not due to an increase in the toughness of the material

itself: it indicates the effect of the newly deposited cuticle in

hindering crack propagation by bridging the crack, thus

acting as a patch that reduces the local concentration of

stress along the edge of the incision. All of these strength

and toughness values displayed statistically significant

differences from each other (t-test, p , 0.05).

The original cut surfaces were still clearly visible after fail-

ure (figure 4a–d ), demonstrating that cuticle is not being

deposited directly across the incision. This must inevitably

limit the strength of the repaired area, because the original

cut will still remain and act as a stress concentrator. However,

it is clear that the deposition of the cuticle patch goes far

beyond what would be needed to simply seal the wound

and prevent ingress or egress of fluid. Continuing for more

than 20 days post-injury and considerably increasing the

local thickness, this cuticle clearly plays a mechanical role.

This can be quantified by comparing the resulting strength

of the limb to the bending stress that arises in it during jump-

ing, which was estimated [15] to be 42 MPa. Defining a safety

factor as the ratio of strength to applied stress, our incision

reduced the safety factor to 1.3, making fracture during jump-

ing highly probable. After repair, the safety factor rose to 2.7,

so even though this is less than the original intact strength

(safety factor 4.1), it is clear that it will greatly reduce the

probability of failure in vivo.

Our finite-element model gave a result for the fracture

toughness of unrepaired cuticle (using equation (2.5)) which

was comparable to that found above using the formula

from Takahashi (in equation (2.3)). The finite-element result

was 26% lower: this difference may have been due to a differ-

ent assumption about the crack front, which Takahashi

assumed to be parallel to the tube radius rather than

horizontal. This result was not affected by the choice of slot

width. Using our model, we were able to estimate the

increase in effective fracture toughness as a result of the

newly deposited endocuticle. For a new layer thickness of

20 mm, which is comparable to the average measured value

(19 mm; s.d.¼ 1.6 mm, n ¼ 21), we estimated that measured

toughness would increase by a factor of �3.3, somewhat

larger than the experimental increase which was �2.3.

Using a thicker endocuticle layer of 100 mm, comparable to

the maximum value measured experimentally (110 mm)

gave a predicted toughness increase of �7.7. One reason

why these predicted increases are higher than the experimen-

tal value is that the stress in the cuticle patch may become

very large before the critical stress intensity is reached

(figure 2c). In our model, the maximum stress in the patch

was not affected by the thickness of the patch material but

was found to be inversely proportional to the width of the

slot. This is because the patch material that spans the slot is

loaded by the opening displacement of the slot faces: if we

make the slot wider this will not affect the amount by

which the slot opens under load, but it will affect the

amount of material spanning the slot, and thus the strain

that arises in this material. In our most accurate model of

the slot, based on measurements of the scalpel used, we

found that for an applied stress intensity equal to

3.07 MPap

m (the material’s fracture toughness), the maxi-

mum stress in the patch material was 99 MPa, whereas at

the higher stress intensity of 7.01 MPap

m (when failure

occurred in the repaired samples), this stress was 226 MPa.

These values are comparable to the measured strength of

cuticle (172.4 MPa), implying that failure in the repaired

samples is occurring not because a critical stress intensity

has been reached but, rather, because we have reached the

strength of the material in the patch, causing the patch to

break and thus no longer protect the cut area.

This is the first ever biomechanical study of injury repair

in an arthropod. By contrast, the literature on this subject for

mammalian tissues is extensive, demonstrating the impor-

tance to mammals of the ability to create repairs which are

mechanically strong. For example, cracked or broken bones

can be completely repaired and remodelled to the point

where they are morphologically and mechanically identical

to uninjured bones [9]. However, in some mammalian tis-

sues, the result of healing may be a material that is

structurally different but functionally satisfactory [8], and in

tissues such as tooth enamel and articular cartilage, repair

is absent or limited, presenting major problems for the

rsif.royalsocietypublishing.orgJ.R.So

7

on May 22, 2018http://rsif.royalsocietypublishing.org/Downloaded from

continued functionality of these body parts. We have shown

that an effective system exists to repair the locust leg in a tar-

geted manner. An incision large enough to breach the

epidermis can stimulate the resurrection of a dormant model-

ling process, which might be the same as that used for cuticle

growth immediately post-moult. However, following injury,

this process is now confined to the injured area, placing

new cuticle where it can be most effective to improve

strength, significantly increasing the ratio between strength

and in vivo stress. However, we also found that this only

occurred in about half of the injured subjects, and we suspect

that this is related to the type and severity of the injury,

providing a stimulus for further work.

Data accessibility. All original data can be made available on request:please contact David Taylor at dtaylor@tcd.ie.

Authors’ contributions. E.P. carried out the experimental work and analy-sis. J.-H.D. advised on experimental protocols and biologicalinterpretations. D.T. supervised the overall project, assisted in dataanalysis and prepared the manuscript. All authors discussed theresults and commented on the manuscript.

Competing interests. We declare we have no competing interests.

Funding. We received no funding for this study.

Acknowledgements. We thank the Zoology Department, Trinity CollegeDublin for providing animal housing facilities and the Centre forMicroscopy and Analysis, Trinity College Dublin for its contributionto the electron microscopy work.

c.Interface

References

13:20150984

1. Wigglesworth VB. 1937 Wound healing in an insect(Rhodnius proxilus hemiptera). J. Exp. Biol. 14,364 – 381.

2. Lai-Fook J. 1968 The fine structure of wound repairin an insect. J. Morphol. 124, 37 – 78. (doi:10.1002/jmor.1051240104)

3. Lai-Fook J. 1970 Haemocytes in the repair ofwounds in an insect. J. Morphol. 130, 297 – 313.(doi:10.1002/jmor.1051300304)

4. Galko MJ, Krasnow MA. 2004 Cellular and geneticanalysis of wound healing in Drosophila larvae.PLoS Biol. 2, e239. (doi:10.1371/journal.pbio.0020239)

5. Rowley AF, Ratcliffe NA. 1976 The granular cells ofGalleria mellonella during clotting and phagocyticreactions in vitro. Tissue Cell 8, 437 – 446. (doi:10.1016/0040-8166(76)90004-5)

6. Rowley AF, Ratcliffe NA. 1978 A histological study ofwound healing and hemocyte function in the wax-

moth Galleria mellonella. J. Morphol. 157, 181 –200. (doi:10.1002/jmor.1051570206)

7. Belacortu Y, Paricio N. 2011 Drosophila as a modelof wound healing and tissue regeneration invertebrates. Dev. Dyn. 240, 2379 – 2404. (doi:10.1002/dvdy.22753)

8. Li J, Chen J, Kirsner R. 2007 Pathophysiologyof acute wound healing. Clin. Dermatol. 25, 9 – 18.(doi:10.1016/j.clindermatol.2006.09.007)

9. Taylor D, Hazenberg JG, Lee TC. 2007 Livingwith cracks: damage and repair in humanbone. Nat. Mater. 2, 263 – 268. (doi:10.1038/nmat1866)

10. Takahashi Y. 2002 Evaluation of leak-before-breakassessment methodology for pipes with acircumferential through-wall crack. Part I: stressintensity factor and limit load solutions.Int. J. Pressure Vessels Piping 79, 385 – 392.(doi:10.1016/S0308-0161(02)00036-4)

11. Taylor D. 2007 The theory of critical distances: a newperspective in fracture mechanics. Oxford, UK:Elsevier.

12. Janssen M, Zuidema J, Wanhill R. 2002 Fracturemechanics. London, UK: Spon Press.

13. Hepburn HR, Joffe I. 1974 Locust solid cuticle—atime sequence of mechanical properties. J. InsectPhysiol. 20, 497 – 506. (doi:10.1016/0022-1910(74)90158-9)

14. Dirks JH, Taylor D. 2012 Fracture toughness of locustcuticle. J. Exp. Biol. 215, 1502 – 1508. (doi:10.1242/jeb.068221)

15. Taylor D, Dirks JH. 2012 Shape optimization inexoskeletons and endoskeletons: a biomechanicsanalysis. J. R. Soc Interface 9, 3480 – 3489. (doi:10.1098/rsif.2012.0567)

16. Sutton GP, Burrows M. 2008. The mechanics ofelevation control in locust jumping. J. Comp. Physiol.194, 557 – 563. (doi:10.1007/s00359-008-0329-z)