Talk at ISEC 2016, Seatlle, June 2016

1. Slides from https://www.slideshare.net/oharar
Twitter: @bobohara
Full slides link:
https://www.slideshare.net/oharar/what-are-we-
statistical-ecologists-or-ecological-statisticians/

3. ●
Crispin: UNTB
– combine stats & ecology

4. ●
Crispin: UNTB
– combine stats & ecology
Ecological Theory + Data + Statistics = Knowledge

5. What are we?
Statistical Ecologists or
Ecological Statisticians?
Bob O'Hara
Senckenberg BiK-F
Frankfurt am Main, Germany

6. Statistician Ecologist
Source: Wikipedia http://bit.ly/1PLq3rG

7. Ecology isEcology is
ComplicatedComplicated
Data is (or are) messyData is (or are) messy

8. Abundance – Range Size
Meanabundance
Joint work with Johan Kotze
Statistical artefact?

9. Typical data for one species
Need to model abundance carefully

10. The Augmented Truncated Negative
Binomial Distribution
Fit inflation with an EM algorithm

11. All to get this...

12. From the other end: ecological models
Public Domain, https://commons.wikimedia.org/w/index.php?curid=445086

13. A Gompertz Model
nt=nt−1r I −Ant−1t
log Abundances
Growth rates Density Effects
Environmental
shocks
Work by Crispin Mutshinda-Mwanza, and Ian Woiwod

14. Decompose the
variation
Interspecific
variation:
off-diagonal
Intraspecific
variation:
leading diagonal
Environmental
variation
v=BV BT
C

15. Proportions of Environmental Variance
0 . 0
0 . 2
0 . 4
0 . 6
0 . 8
1 . 0
A.exclamationis
D.mendica
X.xanthographa
N.pronuba
O.gothica
H.alsines
M.secalis
C.trapezina
A.macilenta
C.vacinii
H.proboscidalis
H.grisealis
L.holsatus
C.maenas
C.pagarus
P.serratus
P.montagui
C.crangon
P.sivado
L.limanda
L.liparis
M.merlangus
P.flesus
P.minutus
S.solea
S.sprattus
T.luscus
T.minutes
S.ruticilla
E.minimus
V.olivaceus
D.caerulescens
S.aurocapillus
D.ordii
D.spectabilis
P.flavus
ProportionofEnv.Variance
M o t h s C r u s t a c e a F is h B ir d s R o d e n t s
Sampling variation
also estimated

16. Statistics is big and difficult

17. What can ecologists do?

18. What can ecologists do?
Think!

19. Species Distribution Models
Source: Wikimedia, US Govt http://www.ntsb.gov/aviation/CVR_FDR.htm

20. Species Distribution Models
Source: Swedish Accident Investigation Authority

21. A simplified and not
wholly inaccurate
explanation of how
SDMs are developed
by ecologists

22. Step 1: get occurrence
data from GBIF

23. Step 2: get climate
data from WorldClim

24. Step 3: put data
into a black box

25. From Alejandro Ruete's talk at ISEC 2014
Step 4

26. Show that sampling issues are important
What can we do?

27. ● Software
– MEE Applications
– Ecography
unmarked

28. A Typical Discussion of P-values
Source: https://www.flickr.com/photos/madaise/2612487968

29. My Thoughts on P-values
Source: https://www.pexels.com/photo/cars-nature-hd-wallpaper-54277/

30. Missing the bigger issue(s): converting
numbers into knowledge
p-values are
seductive

31. Some Alternatives

32. “Bayes factors ... are not without their own
complications, … [b]ut, if they were more
widely used, rules would evolve.”
http://science.sciencemag.org/content/352/6290/1180.full

33. NO! Bayes Factors are
NOT the answer
“Bayes factors ... are not without their own
complications, … [b]ut, if they were more
widely used, rules would evolve.”
http://science.sciencemag.org/content/352/6290/1180.full

34. Bottom line: p-values are tools
All tools will be abused

35. There are often better
statistics
Try to be less abstract

36. Visualisation and Sponge Bacteria
Work by Johannes Björk
Crappy graphs by me
Positive
Negative

37. Big Picture Stuff

38. The distance between
ecological models and
data is big
A lot of statistics in
ecology seems to
increase it

39. We need to be able
to communicate
with each other
(less equations?)

40. We are both statistical ecologists
and ecological statisticians

41. The End