Journal of Biological Chemistry Sign up for PNAS Online eTocs
Link: Info for AuthorsLink: SubscribeLink: AboutLink: Editorial BoardLink: ContactLink: Site Map Link: PNAS Home
Proceedings of the National Academy of Sciences
Link: Current Issue "" Link: Archives "" Link: Online Submission ""  Link: Advanced Search

Institution: NIH Library Sign In as Member / Individual


This Article
Abstract
Full Text
Services
Alert me to new issues of the journal
Request Copyright Permission

Espinoza et al. 10.1073/pnas.0401226101.

Supporting Information

Files in this Data Supplement:

Supporting Data Set 1
Supporting Data Set 2
Supporting Data Set 3
Supporting Methods
Supporting Table 1
Supporting Table 2
Supporting Data Set 4
Supporting References




Supporting Data Set 1

Data Set 1. Data on diet, maximum body size [snout–vent lengths (SVLs)], geographic range, and body temperatures were obtained for liolaemid species from the literature (sources in ref. 36 plus subsequent publications), examinations of museum specimens, or recorded from lizards collected in nature. For diets, each species was assigned a diet (insectivore, omnivore, or herbivore) based on references given in this data set or volumetric estimates of gut contents or feces (averaged across individuals within a species) by using the character states described in Materials and Methods. Freshly deposited feces were used when stomach contents could not be obtained (31, 37, 38). Using these criteria and methods, assignment of diet categories was unambiguous for most taxa included in this study. We obtained field body temperatures for 49 species (n = 1–150 individuals per taxon) of liolaemids in four different years (1995, 1996, 1998, and 2001) during typical activity periods (late spring to early fall) depending on species and locality. Lizards were collected under conditions favorable for thermoregulation (i.e., the intensity of radiation or convection did not appear to impose an obvious challenge to attaining body temperatures well above or below those of the ambient environment). Body temperatures were recorded by inserting a 24-gauge type-K thermocouple » 1–3 cm (depending on body size) into the cloaca and were recorded with a digital thermocouple reader within 15 s of capture (by hand or noose). To minimize heat exchange between the collector and the animal, lizards were shaded and held by a limb or the head during this procedure. Data collected on lizards that were inactive, that required more than 20 s to capture, or were captured before 1000 or after 1800 local time were excluded from the analyses.





 

Table 1. Different modeling strategies used for Bayesian analysis of the combined data

 

Partitions

 

 

Model

ND2

12S

Morphology

No. free parameters

Mean likelihood

MS1

GTR + I + G

GTR + I + G

Mk + G

21

–37,705.31

MS2

L-GTR + I + G

L-GTR + I + G

Mk + G

11

–37,819.11

MS3

GTR

GTR

Mk

16

–42,698.02

MS4

HKY + I + G

HKY + I + G

Mk + G

13

–37,804.22

MS5

JC + I + G

JC + I + G

Mk + G

5

–40,835.15

The table includes the number of free parameters forthat strategy and the harmonic mean of the log likelihoods of the postburn in trees from an analysis of each data set using that strategy. The harmonic mean was obtained by using the sump command in MRBAYES. Modeling strategy 1 (MS1) has the highest likelihood. ND2, NADH dehydrogenase subunit 2; L, parameters are linked between the partitions; GTR, general time-reversible model; HKY, Hasegawa–Kishino–Yano model; JC, Jukes–Cantor model; I, proportion of sites assumed to be invariable; Mk, Markov K model; G , among-site rate variation included, estimated assuming a gamma distribution of rates among sites.





 

Table 2. Comparison among all model strategies (MS1 to MS5) for the combined data by using Bayes factors

 

MS1

MS2

MS3

MS4

MS5

MS1

0

—

—

—

—

MS2

–227.60

0

—

—

—

MS3

–9985.42

–9757.82

0

—

—

MS4

–197.82

29.78

978.76

0

—

MS5

–6259.68

–6032.08

3725.74

–6061.86

0

Each value represents 2loge(B10), with negative values indicating support for the column model over the row model, with absolute values >10 considered to be very strong evidence favoring one model over the other. MS1 is strongly favored overall. Values of 2loge(B10) were calculated as the difference in the harmonic means of the log likelihoods of the two models multiplied by 2 (14).