Transcript
Page 1: Moving towards IPM with robust sampling strategies

Cooperative Research Centre for National Plant Biosecurity

Dr  Grant  Hamilton  

Be/er  sampling  strategies  for  post  harvest  grain  in  Australia  

Page 2: Moving towards IPM with robust sampling strategies

Project  Aims  

•  To  review  current  sampling  methodologies    •  develop  a  flexible,  staBsBcally  robust  sampling  system  for  the  detecBon  of  post-­‐harvest  grain  storage  pests  in  the  Australian  grains  industry.  

Page 3: Moving towards IPM with robust sampling strategies

1:  review  of  sampling    

•  Current  sampling  gives  a  number  of  opportuniBes  to  detect  infestaBons  

•  In  the  1950’s  Australia  began  to  develop  a  reputaBon  for  infested  grain  

•  Response  -­‐  Export  grain  regulaBons  (1963)  •  NO  live  insects  •  Grain  needed  to  be  sampled  –  but  how  much?  

– Will  determine  how  effecBve  a  sampling  programme  is  at  detecBng  what  is  there  

   

Page 4: Moving towards IPM with robust sampling strategies

1:  review  of  sampling    •  2.25L  /33  Tonnes  –  based  on  pragmaBc  consideraBons  –  Belt  loading  speeds  –  Smoko  breaks  –  Size  of  storages  and  transport  infrastructure  –  Samplers  capacity  to  sieve  sample  

•  sampling  model  reviewed  by  Hunter  and  Griffiths  (1978)  

•  reasonable  IF  insects  spread  homogeneously    

Page 5: Moving towards IPM with robust sampling strategies

Hunter  and  Griffiths  

Page 6: Moving towards IPM with robust sampling strategies

•  But  they’re  not  – Grain  type  – Behaviour  – Micro-­‐climaBc  condiBons  – Storage  type  

Grant  Hamilton  and  David  Elmouee  (2011).  Insect  distribuBons  and  sampling  protocols  for  stored  commodiBes.  Stewart  Postharvest  Review  

Page 7: Moving towards IPM with robust sampling strategies

2:  New  sampling  model  

•  To  be  more  accurate  sampling  model  needs  to  account  for  heterogeneous  distribuBon  

Page 8: Moving towards IPM with robust sampling strategies

2:  new  sampling  model  

•  New  sampling  model  -­‐  number  of  samples  that  need  to  be  taken  to  detect  (rejecBon  sampling  approach)  

– ProporBon  of  grain  infested  p  – Density  of  infestaBon  λ  – Size  of  sample  unit  

                 

Elmouee,  Kiermeier  and  Hamilton.  (2010).  Pest  management  Science  

Page 9: Moving towards IPM with robust sampling strategies

Advantages  

•  Closer  representaBon  of  biological  system-­‐  greater  capacity  to  detect  infestaBons  

•  Parameters  intuiBve    •  Inform  parameters  from  range  of  informaBon  sources  (expert  opinion,  samples  taken  for  other  reasons)  

Page 10: Moving towards IPM with robust sampling strategies

3:  Assess  the  accuracy    

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

1

LL HH VH HL

Perc

enta

ge m

odel

suc

cess

Type of Infestation

CM

H&G

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

1

LL MM HM ML

Perc

enta

ge m

odel

suc

cess

Type of Infestation

CM

H&G

Rhyzopertha  dominica  

Cryptolestes  ferrugineus  

(Density  of  infestaBon,  ProporBon  infested)  

oversampling  undersampling  

2  bins  –Parameter  esBmates  1,  permute  and  ‘sample’  other  2    10,000  simulaBons    

Page 11: Moving towards IPM with robust sampling strategies

4:  Sampling  for  Integrated  Pest  Management  

•  Sampling  integral  to  IPM  programmes  •  Can  inform  decisions  (to  treat,  treatment  type,  movement  of  product)  

•  Currently  modelling  rejecBon  (decision  to  treat/fumigate)  based  on  detecBon  of  single  insect  

•  Use  model  for  scenario  tesBng–  treat  at  some  higher  acBon  threshold      

Page 12: Moving towards IPM with robust sampling strategies

Other  outcomes    

•  Masters  project  – 3D  analysis  spaBal  locaBon  Rd  –  IntegraBng  with  sampling  model  

Steel,  Elmouee,  Hamilton.  JSPR,  2012  

30°C, 14 days

Horiz. dist from PoI

Hor

iz. d

ist f

rom

PoI

10cm5cm0cm5cm

10cm

10cm 5cm

0cm

5cm

10cm

25°C, 1 gen.

Horiz. dist from PoI

Hor

iz. d

ist f

rom

PoI

10cm5cm0cm5cm

10cm

10cm 5cm

0cm

5cm

10cm

30°C, 1 gen.

Horiz. dist from PoI

Hor

iz. d

ist f

rom

PoI

10cm5cm0cm5cm

10cm

10cm 5cm

0cm

5cm

10cm

35°C, 1 gen.

Horiz. dist from PoI

Hor

iz. d

ist f

rom

PoI

10cm5cm0cm5cm

10cm

10cm 5cm

0cm

5cm

10cm

Horiz. dist from PoI

Vert.

dis

t fro

m P

oI

37.5cm32.5cm27.5cm22.5cm17.5cm12.5cm

7.5cm2.5cm

10cm 5cm

0cm

5cm

10cm

Horiz. dist from PoI

Vert.

dis

t fro

m P

oI

37.5cm32.5cm27.5cm22.5cm17.5cm12.5cm

7.5cm2.5cm

10cm 5cm

0cm

5cm

10cm

Horiz. dist from PoI

Vert.

dis

t fro

m P

oI

37.5cm32.5cm27.5cm22.5cm17.5cm12.5cm

7.5cm2.5cm

10cm 5cm

0cm

5cm

10cm

Horiz. dist from PoI

Vert.

dis

t fro

m P

oI

37.5cm32.5cm27.5cm22.5cm17.5cm12.5cm

7.5cm2.5cm

10cm 5cm

0cm

5cm

10cm

Low

High

Page 13: Moving towards IPM with robust sampling strategies

Outcomes  for  industry  •  Review  •  TheoreBcal  framework  for  further  work      •  Model  can  be  used  to  establish  level  of  confidence  from  number  of  samples  

•  Model  structured  so  that  different  forms  of  informaBon  can  be  used  

•  Sampling  could  base  on  fixed  number  of  samples  rather  than  by  size  of  consignment  

•  StaBsBcal  foundaBon  for  alternaBve  acBon  thresholds        

Page 14: Moving towards IPM with robust sampling strategies

Thanks  •  Dr.  David  Elmouee  •  Peterson  family  (Killarney)  •  Philip  Burrill,  GRDC  •  Pat  Collins,  Greg  Daglish,  Manoj  Nayak  •  Jim  Eldridge  and  Roderic  Steel  (QUT)  •  CBH,  Graincorp,  Viterra,    •  Dr.  Andreas  Kiermeier  –  SARDI    •  Dr.  Paul  Flinn  –  USDA  •  Prof.  Bhadriraju  Subramanyam  &  Prof.  David  Hagstrum  –  

KSU  


Recommended