More burning, more warts:
Frequent burning favours cane toads
Diana Virkki (Griffith University)
Cuong Tran (Ten Rivers)
Tom Lewis (DAF)
Guy Castley (Griffith University)
 Disturbance processes and invasive species
 Synergistic effects on native species
(Anson et al. 2013; McGregor et al. 2014; McGregor et al. 2015)
Open landscapes
Habitat generalists
Invasive species
Improve dispersal
Increase access to prey
Introduction
(McGregor et al. 2014)
 Frequent burning – commonplace as land management tool
Fire
Reduces
vegetation cover
and homogenises
landscape
Habitat
structure
Favour invasive
species
Fire as a landscape disturbance process
 Rhinella marina
 Impacts on native fauna
 Native to S America
 Introduced to NQ in 1935
 Fire  commonly used
 Relationships with fire, native anurans and cane toads unknown
 Ecologically appropriate and sustainable fire management regimes ?
Cane toads
Native anurans and cane toads
 Anurans  Habitat heterogeneity
(Mac Nally et al. 2001, Williams and Hero 2001, Price et al. 2010)
 Structural features, i.e. leaf litter, CWD
 Habitat features reduced by fire
 E.g. Burrowing frog (Myobatrachus gouldii) and wood frogs
(Rana sylvatica) require thick litter affected by fuel reduction
burning (Bamford 1992; Rittenhouse and Semlitsch 2007)
 Cane toads  Open habitats (Zug and Zug 1979, Brown et al. 2006)
 Influence of disturbance, adding to pressure of invasive
cane toads
 SEQ – impacted by cane toads
 Do differing fire regimes (i.e. repeated fires, variable fire
histories and wildfire) influence native ground-dwelling
anuran communities and cane toad abundances in dry
sclerophyll forests of southeast Queensland?
 Hypothesis:
Aims
Burning
Habitat
Heterogeneity
= + Native
anurans
+ Cane
toads
Study sites
 Subtropical climate
mar approx. 1000 mm
St Mary
Tiaro
Bauple
Queensland
Study
region
Australia
Dry sclerophyll forest
60 Km
Fire history
 Variable
 Long-term fire
experiment
Legend
"
³
Study plots
Fire Treatment
Bauple Treatments
AB
LU
TB
WF
Comparative Treatments
SM01
SMWF
T01
T03
Bruce Highway
Forest boundary
0 4 82 Kilometers
 Four surveys at 35 plots
 2 × winter, 2 × spring-summer
(2010 – 2013)
 4 trap nights
 Predictor variables:
 Habitat structure
 Structural features
 Heterogeneity
 Fire - spatial
 Rainfall (BOM)
Data Collection
 Changes in vegetation communities along transects
 Shannon-Wiener diversity index; H = ∑(ni/N) ln(ni/N)
Habitat Heterogeneity
 ni = species richness or length per
patch
 N = total species or number of
patches
Results summary
Dependent variable
Treatment
favoured
Treatment
× season
Native abundance TB
Richness
AB, TB, WF
Cane toad - Rhinella marina* AB
Striped marsh frog – Limnodynastes peronii* TB, LU
Northern banjo frog – L. terraereginae* No preference
Spotted marsh frog - L. tasmaniensis* No preference Spring
Total significant (P<0.05) 3 1
Great barred frog - Mixophyes fasciolatus LU
Tusked frog - Adelotus brevis LU
Ornate Burrowing frog - Platyplectrum ornatum TB
Copper-backed brood frog - Pseudophryne raveni TB
Green-thighed frog - Litoria brevipalmata AB
Smooth toadlet - Uperoleia laevigata Mixed
Longer
unburned
10 year -
unburned
Annually
burned
Wildfire
Key
Triennially
burned
 Figure. Anuran variables with a
significant (P<0.05) treatment
effect, showing mean ± S.E.
a) native anuran abundance,
b) anuran richness and
c) Limnodynastes peronii
abundance.
Effects of fire treatment on anurans
AB
TB
WF
SMWF
SM01
T03
T01
LU
0
2
4
6
8
10
12
Nativeanuranabundance
a
ab
ab
bc
bc
bc
c
c
AB
TB
WF
SMWF
SM01
T03
T01
LU
Treatment
0
1
2
3
4
5
6
Anuranrichness
a
a
a
b
b
b
b
b
Treatment
AB
TB
WF
SMWF
SM01
T03
T01
LU
Treatment
0
1
2
3
4
5
6
Limnodynastesperonii
a a
b b b b
a) b)
c)
Nativeanuran
abundance
Anuranrichness
Stripedmarsh
frogabundance
 Figure. Anuran variables with
significant (P<0.05) interactive
effects of treatment × season,
with seasons separate, showing
mean ± S.E.
a) total anuran abundance and
b) Rhinella marina abundance.
Spring Winter
AB
TB
WF
SMWF
SM01
T03
T01
LU
Treatment
0
10
20
30
40
50
60
70
Totalanuranabundance
a
b
bcbc
c cc
c
AB
TB
WF
SMWF
SM01
T03
T01
LU
Treatment
0
2
4
6
8
10
12
Totalanuranabundance
AB
TB
WF
SMWF
SM01
T03
T01
LU
Treatment
0
1
2
3
4
Rhinellamarina
AB
TB
WF
SMWF
SM01
T03
T01
LU
Treatment
0
10
20
30
40
50
60
Rhinellamarina
a
b
bc
bc bc bc
bc
c
Effects of fire treatment × seasonality on anurans
Spring Winter
a)
b)
Totalanuran
abundance
Canetoad
abundance
 Figure. Scatter plot of Limnodynastes terraereginae abundance with
time since fire (back transformed ln scale), showing line of best fit
(linear) and r2 for significantly correlated (P<0.05) abundances.
Effects of time since fire
0 0.7 1.7 3.5 6.4 11 19 32 53.5 89
Time since fire (ln scale)
0
2
4
6
8
10
12Limnodynastesterraereginae r2
= 0.115
-
Northern banjo frog
Limnodynastes terraereginae
Northernbanjofrog
abundance
Figure. Scatter plots of significant
(P<0.05) anuran relationships with
number of fires (back transformed ln
scale), showing line of best fit (linear
or quadratic) and r2 for a) anuran
abundance, b) anuran richness, c)
Rhinella marina abundance and d)
Limnodynastes tasmaniensis
abundance.
0 0.7 1.7 3.5 6.4 11 19 32
0
1
2
3
4
5
6
7
8
9
Anuranrichness
r2
= 0.066
b)
0 0.7 1.7 3.5 6.4 11 19 32
Fire frequency (ln scale)
0
20
40
60
80
100
Rhinellamarina
r2
= 0.254
c)
Number of fires (ln scale)
0 0.7 1.7 3.5 6.4 11 19 32
Fire frequency (ln scale)
0
2
4
6
8
10
Limnodynastestasmaniensis
r2
= 0.094
d)
Number of fires (ln scale)
0 0.7 1.7 3.5 6.4 11 19 32
0
20
40
60
80
100
120
Totalanuranabundance
r2
= 0.248
a)
Effects of number of fires
+
a) Total anuran abundance
b) Richness
c) Cane Toad
Rhinella marina
d) Spotted Marsh Frog
Limnodynastes tasmaniensis
Totalanuran
abundance
Anuranrichness
Canetoad
abundance
Spottedmarsh
frogabundance
 * represents species modelled using AICc when over
dispersion was minimal.
 Table 5. Model-averaged coefficients ± confidence
intervals of explanatory habitat, fire and weather
variables from negative binomial Generalised Linear
Mixed Models on anuran abundance, richness and
species-specific abundances. Bold values indicate
where coefficient confidence intervals do not overlap
zero.
Determining the most important variables for anurans
Dependent variable
Predictor variables
Bare CWD
Number of
fires Canopy H’ Ground H’ Litter Rainfall R2
glmm(m)
Total abundance 0.07 ± 0.38 -0.27 ± 0.29 -0.16 ± 0.37 -0.09 ± 0.26 0.25 ± 0.28 0.09 ± 0.33 0.20 ± 0.34 0.02
Native abundance 0.13 ± 0.43 -0.01 ± 0.34
Number of fires
0.52 ± 0.37
0.06 ± 0.28
Ground H’
0.66 ± 0.38
0.16 ± 0.32 0.25 ± 0.38 0.08
Richness 0.16 ± 0.17 -0.08 ± 0.17 0.15 ± 0.19 0.07 ± 0.17 0.14 ± 0.18 -0.03 ± 0.19
Rainfall
0.18 ± 0.17
0.07
Limnodynastes
peronii
0.13 ± 0.56 -0.03 ± 0.38 0.35 ± 0.48 -0.06 ± 0.30
Ground H’
0.67 ± 0.43
0.23 ± 0.42 0.39 ± 0.41 0.03
L. tasmaniensis* 0.42 ± 0.93 -0.82 ± 1.19 0.58 ± 1.15 0.76 ± 1.05 0.24 ± 1.04 -0.96 ± 1.01 0.10 ± 0.97 0.01
L. terraereginae* 0.02 ± 0.52 0.30 ± 0.52
Number of fires
0.70 ± 0.47
Canopy H’
0.55 ± 0.49
Ground H’
0.81 ± 0.49
0.43 ± 0.44 0.12 ± 0.49 0.52
Rhinella marina 0.34 ± 0.52 -0.16 ± 0.37 0.16 ± 0.60 -0.09 ± 0.33 0.06 ± 0.39
Litter
-0.47 ± 0.36
Rainfall
0.52 ± 0.38
0.08
Total Significant 2 1 3 1 2
Distance from watercourse
R² = 0.0155
0
5
10
15
20
25
0 200 400 600 800 1000 1200
R² = 4E-05
0
1
2
3
4
5
6
7
8
9
0 200 400 600 800 1000 1200
R² = 0.0182
0
10
20
30
40
50
60
70
80
90
100
0 200 400 600 800 1000 1200
Anuranrichness
Canetoad
abundance
Nativeanuran
abundance
Figure. Scatter plots anuran
relationships with distance
to nearest creekline
(moderate to major stream
based on q-spatial data),
showing line of best fit
(linear) and r2 for a) native
anuran abundance, b)
anuran richness, and c)
Rhinella marina abundance.
Distance to nearest creekline (m)
(Moderate – Major stream, Q-spatial data)
Distance to nearest creekline (m)
(Moderate – Major stream, Q-spatial data)
a)
b) c)
Discussion
 Anurans persisted with frequent fire events
 Refuted hypothesis: frequent burning reduces habitat H’ and
negatively impacts on anurans
 Long-term fire experiment, favoured anurans
 Regular (small-scale) low intensity fires
 Fine-scale patchiness playing a role
 Scale of patchiness
 Fine-scale patchiness or large unburnt patches?
Annually burnt plot – Bauple State Forest
 Anurans often unaffected or positively affected by fire
(Hannah et al. 1998, Keyser et al. 2004, Perry et al. 2009, Lowe et al. 2013)
 Avoid mortality
 Moist microhabitat
 Retreat underground or into water
(Perry et al. 2009, Lowe et al. 2013)
 Congruent with results  low intensity burns
 Particularly for cane toads
Anurans and fire
Toad preferences
 Favoured very frequently burned sites  Annually burned and triennially burned sites
 Past research (USA)
 Cane toads  negatively associated with litter
preferred frequently disturbed areas
 More abundant in open grassland and disturbed habitats
Species Trends with fire References
Cane toad Not described
Dwarf American toad
Bufo americanus charlessmithi
More abundant 1 yr post-fire Perry et al. 2009
American toad
B. americanus
Higher numbers in burnt forest Kirkland et al. 1996,
Greenberg and Waldrop
2008
Boreal toad
Anaxyrus boreas
Abundance tripled in the 3yrs after fire Hossack et al. 2013
Habitat Heterogeneity
 Anuran richness related to habitat heterogeneity
(Dupuis et al. 1995, Delis et al. 1996, Pearman 1997, Williams and Hero 2001).
 Canopy and ground H’  key predictors
 Important for native species
 Repeated burning may reduce fine-scale habitat H’
 Patchy, mosaic burns important
 Guidelines for DSF in SEQ: 40-80% mosaic
 Ideal scale of patchiness unknown
Summary
 Species not excluded from frequent burns
 Positive outcomes:
 Anurans resilient to fires
 Negative outcomes:
 Cane toads favoured at frequently burnt areas
 Potential impacts on natives through:
 Competition or predation (Boland 2004, Greenlees et al. 2006)
 Ingestion of toxins (Phillips et al. 2003)
Land managers need to consider this in fire
management if planning for biodiversity outcomes
Important in high risk areas, i.e. vulnerable
wetlands or areas with threatened anurans
Dr Diana Virkki – virkkid@tenrivers.com.au
Dr Cuong Tran
Dr Tom Lewis
Dr Guy Castley

BushfireConf2017 - 28. More burning, more warts: Frequent burning favours cane toads

  • 1.
    More burning, morewarts: Frequent burning favours cane toads Diana Virkki (Griffith University) Cuong Tran (Ten Rivers) Tom Lewis (DAF) Guy Castley (Griffith University)
  • 2.
     Disturbance processesand invasive species  Synergistic effects on native species (Anson et al. 2013; McGregor et al. 2014; McGregor et al. 2015) Open landscapes Habitat generalists Invasive species Improve dispersal Increase access to prey Introduction (McGregor et al. 2014)
  • 3.
     Frequent burning– commonplace as land management tool Fire Reduces vegetation cover and homogenises landscape Habitat structure Favour invasive species Fire as a landscape disturbance process
  • 4.
     Rhinella marina Impacts on native fauna  Native to S America  Introduced to NQ in 1935  Fire  commonly used  Relationships with fire, native anurans and cane toads unknown  Ecologically appropriate and sustainable fire management regimes ? Cane toads
  • 5.
    Native anurans andcane toads  Anurans  Habitat heterogeneity (Mac Nally et al. 2001, Williams and Hero 2001, Price et al. 2010)  Structural features, i.e. leaf litter, CWD  Habitat features reduced by fire  E.g. Burrowing frog (Myobatrachus gouldii) and wood frogs (Rana sylvatica) require thick litter affected by fuel reduction burning (Bamford 1992; Rittenhouse and Semlitsch 2007)  Cane toads  Open habitats (Zug and Zug 1979, Brown et al. 2006)  Influence of disturbance, adding to pressure of invasive cane toads
  • 6.
     SEQ –impacted by cane toads  Do differing fire regimes (i.e. repeated fires, variable fire histories and wildfire) influence native ground-dwelling anuran communities and cane toad abundances in dry sclerophyll forests of southeast Queensland?  Hypothesis: Aims Burning Habitat Heterogeneity = + Native anurans + Cane toads
  • 7.
    Study sites  Subtropicalclimate mar approx. 1000 mm St Mary Tiaro Bauple Queensland Study region Australia Dry sclerophyll forest 60 Km Fire history  Variable  Long-term fire experiment
  • 8.
    Legend " ³ Study plots Fire Treatment BaupleTreatments AB LU TB WF Comparative Treatments SM01 SMWF T01 T03 Bruce Highway Forest boundary 0 4 82 Kilometers
  • 9.
     Four surveysat 35 plots  2 × winter, 2 × spring-summer (2010 – 2013)  4 trap nights  Predictor variables:  Habitat structure  Structural features  Heterogeneity  Fire - spatial  Rainfall (BOM) Data Collection
  • 10.
     Changes invegetation communities along transects  Shannon-Wiener diversity index; H = ∑(ni/N) ln(ni/N) Habitat Heterogeneity  ni = species richness or length per patch  N = total species or number of patches
  • 11.
    Results summary Dependent variable Treatment favoured Treatment ×season Native abundance TB Richness AB, TB, WF Cane toad - Rhinella marina* AB Striped marsh frog – Limnodynastes peronii* TB, LU Northern banjo frog – L. terraereginae* No preference Spotted marsh frog - L. tasmaniensis* No preference Spring Total significant (P<0.05) 3 1 Great barred frog - Mixophyes fasciolatus LU Tusked frog - Adelotus brevis LU Ornate Burrowing frog - Platyplectrum ornatum TB Copper-backed brood frog - Pseudophryne raveni TB Green-thighed frog - Litoria brevipalmata AB Smooth toadlet - Uperoleia laevigata Mixed Longer unburned 10 year - unburned Annually burned Wildfire Key Triennially burned
  • 12.
     Figure. Anuranvariables with a significant (P<0.05) treatment effect, showing mean ± S.E. a) native anuran abundance, b) anuran richness and c) Limnodynastes peronii abundance. Effects of fire treatment on anurans AB TB WF SMWF SM01 T03 T01 LU 0 2 4 6 8 10 12 Nativeanuranabundance a ab ab bc bc bc c c AB TB WF SMWF SM01 T03 T01 LU Treatment 0 1 2 3 4 5 6 Anuranrichness a a a b b b b b Treatment AB TB WF SMWF SM01 T03 T01 LU Treatment 0 1 2 3 4 5 6 Limnodynastesperonii a a b b b b a) b) c) Nativeanuran abundance Anuranrichness Stripedmarsh frogabundance
  • 13.
     Figure. Anuranvariables with significant (P<0.05) interactive effects of treatment × season, with seasons separate, showing mean ± S.E. a) total anuran abundance and b) Rhinella marina abundance. Spring Winter AB TB WF SMWF SM01 T03 T01 LU Treatment 0 10 20 30 40 50 60 70 Totalanuranabundance a b bcbc c cc c AB TB WF SMWF SM01 T03 T01 LU Treatment 0 2 4 6 8 10 12 Totalanuranabundance AB TB WF SMWF SM01 T03 T01 LU Treatment 0 1 2 3 4 Rhinellamarina AB TB WF SMWF SM01 T03 T01 LU Treatment 0 10 20 30 40 50 60 Rhinellamarina a b bc bc bc bc bc c Effects of fire treatment × seasonality on anurans Spring Winter a) b) Totalanuran abundance Canetoad abundance
  • 14.
     Figure. Scatterplot of Limnodynastes terraereginae abundance with time since fire (back transformed ln scale), showing line of best fit (linear) and r2 for significantly correlated (P<0.05) abundances. Effects of time since fire 0 0.7 1.7 3.5 6.4 11 19 32 53.5 89 Time since fire (ln scale) 0 2 4 6 8 10 12Limnodynastesterraereginae r2 = 0.115 - Northern banjo frog Limnodynastes terraereginae Northernbanjofrog abundance
  • 15.
    Figure. Scatter plotsof significant (P<0.05) anuran relationships with number of fires (back transformed ln scale), showing line of best fit (linear or quadratic) and r2 for a) anuran abundance, b) anuran richness, c) Rhinella marina abundance and d) Limnodynastes tasmaniensis abundance. 0 0.7 1.7 3.5 6.4 11 19 32 0 1 2 3 4 5 6 7 8 9 Anuranrichness r2 = 0.066 b) 0 0.7 1.7 3.5 6.4 11 19 32 Fire frequency (ln scale) 0 20 40 60 80 100 Rhinellamarina r2 = 0.254 c) Number of fires (ln scale) 0 0.7 1.7 3.5 6.4 11 19 32 Fire frequency (ln scale) 0 2 4 6 8 10 Limnodynastestasmaniensis r2 = 0.094 d) Number of fires (ln scale) 0 0.7 1.7 3.5 6.4 11 19 32 0 20 40 60 80 100 120 Totalanuranabundance r2 = 0.248 a) Effects of number of fires + a) Total anuran abundance b) Richness c) Cane Toad Rhinella marina d) Spotted Marsh Frog Limnodynastes tasmaniensis Totalanuran abundance Anuranrichness Canetoad abundance Spottedmarsh frogabundance
  • 16.
     * representsspecies modelled using AICc when over dispersion was minimal.  Table 5. Model-averaged coefficients ± confidence intervals of explanatory habitat, fire and weather variables from negative binomial Generalised Linear Mixed Models on anuran abundance, richness and species-specific abundances. Bold values indicate where coefficient confidence intervals do not overlap zero. Determining the most important variables for anurans Dependent variable Predictor variables Bare CWD Number of fires Canopy H’ Ground H’ Litter Rainfall R2 glmm(m) Total abundance 0.07 ± 0.38 -0.27 ± 0.29 -0.16 ± 0.37 -0.09 ± 0.26 0.25 ± 0.28 0.09 ± 0.33 0.20 ± 0.34 0.02 Native abundance 0.13 ± 0.43 -0.01 ± 0.34 Number of fires 0.52 ± 0.37 0.06 ± 0.28 Ground H’ 0.66 ± 0.38 0.16 ± 0.32 0.25 ± 0.38 0.08 Richness 0.16 ± 0.17 -0.08 ± 0.17 0.15 ± 0.19 0.07 ± 0.17 0.14 ± 0.18 -0.03 ± 0.19 Rainfall 0.18 ± 0.17 0.07 Limnodynastes peronii 0.13 ± 0.56 -0.03 ± 0.38 0.35 ± 0.48 -0.06 ± 0.30 Ground H’ 0.67 ± 0.43 0.23 ± 0.42 0.39 ± 0.41 0.03 L. tasmaniensis* 0.42 ± 0.93 -0.82 ± 1.19 0.58 ± 1.15 0.76 ± 1.05 0.24 ± 1.04 -0.96 ± 1.01 0.10 ± 0.97 0.01 L. terraereginae* 0.02 ± 0.52 0.30 ± 0.52 Number of fires 0.70 ± 0.47 Canopy H’ 0.55 ± 0.49 Ground H’ 0.81 ± 0.49 0.43 ± 0.44 0.12 ± 0.49 0.52 Rhinella marina 0.34 ± 0.52 -0.16 ± 0.37 0.16 ± 0.60 -0.09 ± 0.33 0.06 ± 0.39 Litter -0.47 ± 0.36 Rainfall 0.52 ± 0.38 0.08 Total Significant 2 1 3 1 2
  • 17.
    Distance from watercourse R²= 0.0155 0 5 10 15 20 25 0 200 400 600 800 1000 1200 R² = 4E-05 0 1 2 3 4 5 6 7 8 9 0 200 400 600 800 1000 1200 R² = 0.0182 0 10 20 30 40 50 60 70 80 90 100 0 200 400 600 800 1000 1200 Anuranrichness Canetoad abundance Nativeanuran abundance Figure. Scatter plots anuran relationships with distance to nearest creekline (moderate to major stream based on q-spatial data), showing line of best fit (linear) and r2 for a) native anuran abundance, b) anuran richness, and c) Rhinella marina abundance. Distance to nearest creekline (m) (Moderate – Major stream, Q-spatial data) Distance to nearest creekline (m) (Moderate – Major stream, Q-spatial data) a) b) c)
  • 18.
    Discussion  Anurans persistedwith frequent fire events  Refuted hypothesis: frequent burning reduces habitat H’ and negatively impacts on anurans  Long-term fire experiment, favoured anurans  Regular (small-scale) low intensity fires  Fine-scale patchiness playing a role  Scale of patchiness  Fine-scale patchiness or large unburnt patches? Annually burnt plot – Bauple State Forest
  • 19.
     Anurans oftenunaffected or positively affected by fire (Hannah et al. 1998, Keyser et al. 2004, Perry et al. 2009, Lowe et al. 2013)  Avoid mortality  Moist microhabitat  Retreat underground or into water (Perry et al. 2009, Lowe et al. 2013)  Congruent with results  low intensity burns  Particularly for cane toads Anurans and fire
  • 20.
    Toad preferences  Favouredvery frequently burned sites  Annually burned and triennially burned sites  Past research (USA)  Cane toads  negatively associated with litter preferred frequently disturbed areas  More abundant in open grassland and disturbed habitats Species Trends with fire References Cane toad Not described Dwarf American toad Bufo americanus charlessmithi More abundant 1 yr post-fire Perry et al. 2009 American toad B. americanus Higher numbers in burnt forest Kirkland et al. 1996, Greenberg and Waldrop 2008 Boreal toad Anaxyrus boreas Abundance tripled in the 3yrs after fire Hossack et al. 2013
  • 21.
    Habitat Heterogeneity  Anuranrichness related to habitat heterogeneity (Dupuis et al. 1995, Delis et al. 1996, Pearman 1997, Williams and Hero 2001).  Canopy and ground H’  key predictors  Important for native species  Repeated burning may reduce fine-scale habitat H’  Patchy, mosaic burns important  Guidelines for DSF in SEQ: 40-80% mosaic  Ideal scale of patchiness unknown
  • 22.
    Summary  Species notexcluded from frequent burns  Positive outcomes:  Anurans resilient to fires  Negative outcomes:  Cane toads favoured at frequently burnt areas  Potential impacts on natives through:  Competition or predation (Boland 2004, Greenlees et al. 2006)  Ingestion of toxins (Phillips et al. 2003) Land managers need to consider this in fire management if planning for biodiversity outcomes Important in high risk areas, i.e. vulnerable wetlands or areas with threatened anurans
  • 23.
    Dr Diana Virkki– virkkid@tenrivers.com.au Dr Cuong Tran Dr Tom Lewis Dr Guy Castley