SUPPLEMENT 3 – TESTING THE INFLUENCE OF PHYLOGENY ON SURVIVAL
Due to their shared evolutionary history, two species that are closely related will exhibit more
similar traits than two species which are distantly related. Hence, if traits influence survival as we
are proposing here, phylogeny has the potential to have an impact on our inference about the
relative importance of these traits. Observed survival rates will partially be driven by the immediate
ecological factors operating on a species, and partially by evolutionary factors [1]. Teasing these
apart is difficult because of the correlated nature of traits and phylogeny.
The models presented in the main body of this paper focus solely on readily quantifiable traits, as
we are concerned with the potential functional link between these and demographic processes, and
they are relatively easy to observe and measure. In contrast, bat phylogenetic relationships are still
hotly contested. Until very recently, our study species Austronomus australis was part of the
Tadarida genus found outside of Australia [2]. It is also debated whether the horseshoe Rhinolophid
bats and false vampire Megadermatid bats should be lumped into a new
“Yinpterochiroptera” suborder with the fruit bats [3]. A model based solely on traits as we have
presented in the Table 1 is not vulnerable to such changes in accepted phylogeny. However, to
check whether we needed to accounted for phylogeny, we conducted an exploratory analysis where
Superfamily was treated as a fixed effect. We tested two sets of models; the first set was the 26
models currently listed Table S3, and the second set of eleven was exactly the same but including
Superfamily instead of guild as a fixed effect. Models did not converge if both Superfamily and
guild were included because of the number of levels included both of these factors. We found that
the models which included guild instead of Superfamily were more parsimonious, and at the end of
the model selection process we ultimately ended up with the same ‘best’ model that included the
fixed terms guild, number of young, mass, age, sex, season, and method, and the species random
effect (Table S4).
Table S6 - Trait models tested in the analyses presented in the Methods section (1-26), and an
additional eleven (27-37) which include the fixed term Superfamily instead of Guild. All models
included the covariates age, sex, season and method, as well as bat species treated as a random
effect on the intercept.
Model
no.
7
1
6
11
4
10
3
31
17
5
25
28
37
21
13
26
18
2
9
8
29
22
15
23
19
14
16
24
20
34
36
27
12
35
32
33
30
Number of
young
X
X
X
X
X
X
X
X
Additional fixed effects
Time to
Mass Guild Roost
maturity
X
X
X
X
X
Superfamily
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
DIC
387
387.5
387.5
388.4
388.5
388.7
388.8
388.8
389.2
389.4
389.4
389.5
389.7
389.8
389.9
389.9
390.1
390.4
390.4
390.6
390.6
390.8
390.9
391.1
391.6
391.8
392
392
392.6
392.6
393
393.6
393.7
393.8
394.7
394.8
396.4
The survival estimates extracted from the literature were biased towards one Superfamily, the
Vespertilionoidea (165/193 estimates). If these Vesper bats consistently experience higher or
lower annual survival as a result of evolutionary factors which cannot be accounted for by the
traits, this would bias our informative priors. When we look at the best model which included
Superfamily as a fixed effect however (Model 31, Table S4), the coefficient estimates
indicate that this is not the case and that survival rates for this Superfamily are relatively
average once traits have been considered:
Table S7. Posterior parameter estimates for the best trait model which included the
Superfamily term (Model 31 Table S3), which takes the form of logit(ϕ) ~ α + β1(number of
young - µnumber of young) + βSuperfamily + βsex + βage + βmethod + βseason + εspecies
Parameter
Intercept (Sex - Female, Superfamily Emballonuroidea, Season - Summer, Age Adult, Method - CJS)
Number of young
Superfamily - Molossoidea
Superfamily - Noctilionoidea
Superfamily - Rhinolophoidea
Superfamily - Vespertilionoidea
Sex - Male
Sex - Sexes pooled
Season - Winter
Season - Both seasons
Age - Juvenile
Age - Ages pooled
Method - Other
Method – IR/LT/SC
Method - Bezem's approach
Method - Percent recapture
Precision
Posterior
mean
95% credible
interval
0.212
[-0.728, 1.149]
-0.619
0.387
0.273
1.275
0.491
-0.349
-0.098
-0.198
-0.411
0.575
0.303
-0.066
-0.109
-0.078
-0.319
2.715
[-1.066, -0.196]
[-0.648, 1.409]
[-1.406, 1.96]
[0.219, 2.35]
[-0.422, 1.413]
[-0.592, -0.107]
[-0.423, 0.228]
[-0.557, 0.164]
[-0.813, -0.006]
[0.367, 0.785]
[-0.053, 0.658]
[-0.497, 0.367]
[-0.412, 0.186]
[-0.671, 0.507]
[-0.775, 0.14]
[2.126, 3.389]
Hence, we are confident that the model which best explains the variation in survival estimates
between species is that presented in Table 4.
References
1.
Freckleton, R. P., Harvey, P. H. & Pagel, M. 2002 Phylogenetic analysis and
comparative data: a test and review of evidence. Am. Nat. 160, 712–726.
2.
Reardon, T. B., McKenzie, N. L., Cooper, S. J. B., Appleton, B., Carthew, S. &
Adams, M. 2014 A molecular and morphological investigation of species boundaries
and phylogenetic relationships in Australian free-tailed bats Mormopterus (Chiroptera :
Molossidae). Aust. J. Zool. 62, 109.
3.
Teeling, E. C., Springer, M. S., Madsen, O., Bates, P., O’Brien, S. J. & Murphy, W. J.
2005 A molecular phylogeny for bats illuminates biogeography and the fossil record.
Science 307, 580–4.