Page 34

Proceedings of International Conference on Climate Change, Biodiversity and Ecosystem Services for the Sustainable Development Goals : Policy and Practice 27-29 June 2016 at the Sirindhorn International Environmental Park, Cha-am, Phetchaburi, Thailand

Where  Z,  is  the  linear  combination  model  of  species  I  (Water  Onion)  as  follows: Z=  ß0  +  ß1X1  +  ß2X2  +…+  ßnXn ß1  =  coefficient Xi  =  independent  variables  (environment  factors)  The  probability  values  derived  from  the regression  models  range  from  0.0-0.1.  The  higher  the  value,  the  greater  the  likelihood  of  occupancy  of  the Water  Onion  and  the  lower  value  suggests  an  unlikely  occurrence.  A  cut-off  value  of  0.5  was  used  for  binary  classification. Thus,  any  pixel  containing  the  probability  values  equal  or  greater  than  0.5  was  categorized  as  presence,  otherwise  classified  as  absence. Besides   co n tingency  t a b l e ,   the  performance  of  the  logistic  regression  model  was  assessed  by  using  the  area  under  curve  (AUC)  of  a  receiver  operating  characteristic  curve  (ROC)  (Hosmer  and  Lemeshow,  2000). 2.  Maximum  entropy  (MaxEnt)  which  operates  by  establishing  a  relationship  between a  presence-only  data  and  ecological  variables  within  that  region  and  then  identifying  other  suitable  areas  (Phillips  et.  al.,  2006).  Default  setting  in  MaxEnt  as  outlined  in  Phillips  and  Dudik  (2008)  were  used  and  a  25%  random  test  percentage  and  f ive  cross  validate  replicated  runs  were  applied.  The  performance  and  accuracy  of  MaxEnt  were  evaluated  similar  to  the  logistic  regression  model. 2.4  Assessment  of  conservation  status The  current  conservation  status  of  Water  Onion  in  Thailand  was  evaluated  using  the  IUCN  Red  List  categories  and  criteria  version  3.1  (criterion  Geographic  range  in  the  form  of  B1-predicted  extent  of  occurrence  (IUCN,  2001). III.  RESULTS  3.1  Distribution  of Water  Onion The  results  of  the  logistic  regressions  indicated  that  elevation,  NDVI,  annual  mean  Temperature,  max  Temperature  of  warmest  month,  annual  precipitation,  precipitation  of  wettest  quarter,  precipitation  of  driest  quarter,  dredging,  distance  to  village  and  distance  to  road  were  significantly  related  to  the  distribution  of Water  Onion  as  shown  below. Z =  -56.42  -  0.02  (dem) +  15.74 (NDVI)  -  0.30  (bio_1) +  0.42  (bio_5) +  0.02  (bio_12)  -  0.04  (bio_16)  -  0.12  (bio_17)  -  0.00007  (dred)  -  0.00032  (village_dist)  -  0.00121  (road_dist) Where  : dem  =  elevation  (meter) NDVI  =  normalized  difference  vegetation  Index bio_1  =  annual  mean  temperature bio_5  =  max  temperature  of  warmest  month bio_12  =  annual  precipitation bio_16  =  precipitation  of  wettest  quarter Fig.1  The  area  under  curve  (AUC)  for Water  Onion  from  logistic  regression 32 Proceedings  of  the  International  Conference  on  Climate  Change,  Biodiversity  and  Ecosystem  Services  for  the Sustainable  Development  Goals  (SDGs):  Policy  and  Practice  27-29  June  2016,  Cha-am,  Phetchaburi,  Thailand


Proceedings of International Conference on Climate Change, Biodiversity and Ecosystem Services for the Sustainable Development Goals : Policy and Practice 27-29 June 2016 at the Sirindhorn International Environmental Park, Cha-am, Phetchaburi, Thailand
To see the actual publication please follow the link above