Spe c ie s Distrib utio n Mo de ling L E CT URE 25 SPE CI E - - PDF document

1

Spe c ie s Distrib utio n Mo de ling

L E CT URE 25 SPE CI E S DI ST RI BUT I ON MODE L I NG UNI T 3: ST UF F

Ob je c tive s

 At the e nd o f this se rie s o f le c ture s, yo u sho uld b e a b le to :

 De fine te rms  E

xplain the c o nc e pt o f nic he .

 Disting uish be twe e n me c hanistic and c o rre lative mo de ls.  De sc ribe the ba sic s flo w o f info rma tio n in the de ve lo pme nt o f a SDM.  E

xplain the re latio nship be twe e n g e o g raphic al spac e and e nviro nme ntal spac e .

2

Ob je c tive s

 De sc ribe ho w e quilibrium a nd sa mpling a de qua c y influe nc e the

de ve lo pme nt o f SDM.

 E

xpla in c ro ss valida tio n.

 E

xplain the e rro rs that a SDM c an g e ne rate .

 De sc ribe ho w SDM c a n be use d in c o nse rvatio n bio lo g y.

Mo de l Appro a c he s

 Me c ha nistic Appro a c h

 Do no t re ly o n o b se rve d o c c urre nc e re c o rds  Re quire de taile d physio lo g ic al data

 Co rre la tive Appro a c h

 Assume c urre nt distributio n g ive s a g o o d indic a to r o f e c o lo g ic a l

re quire me nts

3

F lo w dia g ra m o f the ma in ste ps o f spe c ie s distrib utio n mo de l

Processing to generate variables of importance in defining species’ distributions (e.g. maximum daily temperature, frost days, soil water balance) Observed species’ distribution (a list of localities where the species has been observed, and sometimes also localities where the species is known to be absent) Database of ‘raw’ environmental variables (e.g. temperature, precipitation, soil type). Data usually stored in a GIS Modeling algorithm (e.g. Maxent, artificial neural network, generalized linear model, regression tree) Model testing (statistical assessment

• f predictive ability,

using test such as AUC or Kappa) Predicted species’ distribution. Prediction may be for a different region (e.g. for an invasive species) or for a different time period (e.g. under future climate change)

F a c to rs tha t I nflue nc e the L imits o f the Ge o g ra phic Ra ng e

 Ab io tic e nviro nme nt

 T

e mpe rature

 Pre c ipitatio n  So il type



Bio tic inte ra c tio ns

 Pre da tio n  Pa tho g e ns  Mutualisms



Histo ry a nd g e o g ra phy



Dispe rsa l

4

Re la tio nship b e twe e n g e o g ra phic a l spa c e a nd e nviro nme nta l spa c e

5

E q uilib rium a nd Sa mpling Ade q ua c y

 E

q uilib rium: A spe c ie s is in e q uilib rium with c urre nt e nviro nme nta l c o nditio ns if it o c c urs in a ll suita b le a re a s a nd is a b se nt fro m a ll unsuita b le a re a s.

 De pe nds bo th o n bio tic inte ra c tio ns (e .g . c o mpe titive e xc lusio n fro m a n

a re a ) a nd dispe rsa l a bility.

 Sa mpling a de q ua c y: T

he e xte nt to whic h the o b se rve d o c c urre nc e re c o rds pro vide a sa mple o f the e nviro nme nta l spa c e Hig h e q uilib rium a nd e xc e lle nt sa mpling

6

Hig h e q uilib rium b ut po o r sa mpling

Hig h e quilibrium a nd po o r sa mpling in g e o g ra phic a l spa c e , but g o o d sampling in e nviro nme ntal spac e

7

L

• w e q uilib rium b ut g o o d sa mpling

Cro ss Va lida tio n

 T

he spe c ie s o c c urre nc e da ta is divide d ra ndo mly into two pa rts:

 T

raining data

 T

e st data

 T

he tra ining da ta is use d to c o nstruc t the mo de l.

 T

ypic ally a mo de l fits training data fairly we ll, but that is to be e xpe c te d a nd do e s no t te ll us ho w the mo de l will do o ve ra ll.

 T

he mo de l is the n a pplie d to the te st da ta .

 I

f the mo de l do e s an e ffe c tive jo b o f pre dic ting the te st data, the mo de l is pro bably fairly e ffe c tive .

8

E rro rs

 SDM mo de ls c a n g e ne ra te two type s o f e rro rs.

 E

rro r o f o missio n

 Sa ys the spe c ie s do e s no t o c c ur so me whe re it a c tua lly do e s.

 E

rro r o f c o mmissio n

 Sa ys the spe c ie s o c c urs so me whe re it a c tua lly do e s no t.

Type of use Example reference(s) Guiding field surveys to accelerate detection of unknown distributional areas and undiscovered species Raxworthy et al. 2003, Bourg et al. 2005, Guisan et al. 2006 Projecting potential impacts of climate change Iverson and Prasad 1998, Berry et al. 2002, Hannah et al. 2005; for review see Pearson and Dawson 2003 Predicting species’ invasion Higgins et al. 1999, Thuiller et al. 2005; for review see Peterson 2003 Supporting conservation prioritization and reserve selection Araujo and Williams 2000, Ferrier et al. 2002 Assessing the impacts of land cover change

• n species’ distributions

Pearson et al. 2004 Guiding reintroduction of endangered species Pearce and Lindenmayer 1998

Use s o f spe c ie s distrib utio n mo de ls in c o nse rva tio n b io lo g y