31.12.2013 Views

The primate cranial base: ontogeny, function and - Harvard University

The primate cranial base: ontogeny, function and - Harvard University

The primate cranial base: ontogeny, function and - Harvard University

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

D.E. Lieberman et al.]<br />

PRIMATE CRANIAL BASE 133<br />

TABLE 2. Variance components for indices of relative brain size <strong>and</strong> measures of flexion across <strong>primate</strong>s 1<br />

Level<br />

N<br />

IRE1 IRE5 CBA4 CBA1<br />

% N eff % N eff % N eff % N eff<br />

Infraorder 2 12 0.24 48 0.94 0 0 52 1.05<br />

Superfamily 6 22 1.34 10 0.62 34 2.07 0 0.00<br />

Family 15 54 8.16 16 2.44 44 6.56 35 5.19<br />

Genus 60 13 7.81 22 12.99 17 10.26 8 4.74<br />

Species 62 1 0.78 3 2.38 5 3.10 5 3.25<br />

Total N eff 18.33 17.38 21.99 14.23<br />

df 16.00 15.00 20.00 12.00<br />

1 Maximum likelihood variance components were calculated using mainframe SAS (proc varcomp). %, percentage of variance at<br />

each taxonon 1 mic level; N eff , effective N at each level; Total N eff , total effective N for each variable. Bivariate comparisons utilize<br />

lowest N eff of the pair.<br />

TABLE 3. Correlation coefficients for <strong>primate</strong>s <strong>and</strong> haplorhini 1<br />

Variables CBA4 N (P) df eff (P) CBA1 N (P) df eff (P)<br />

Primates<br />

IRE1 0.783 60 (0.001) 14 (0.01) 0.672 60 (0.001) 10 (0.05)<br />

IRE5 0.621 60 (0.001) 13 (0.05) 0.790 62 (0.001) 10 (0.01)<br />

Haplorhini<br />

IRE1 0.813 51 (0.001) 9 (0.01) 0.641 51 (0.001) 9 (0.05)<br />

IRE5 0.636 51 (0.001) 10 (0.05) 0.548 51 (0.001) 14 (0.05)<br />

1 N, total N; d eff ,N eff 2, effective degrees of freedom for each comparison. Bivariate comparisons utilize lowest N eff of the pair.<br />

flexion expected for their relative brain size,<br />

depending on which measures are used.<br />

One problem with the above studies is<br />

that they do not consider the potential role<br />

of phylogenetic effects on these correlations<br />

(Cheverud et al., 1985; Felsenstein, 1985).<br />

Accordingly, the above data were reanalyzed<br />

using the method of Smith (1994) for<br />

adjusting degrees of freedom. 3 Table 2 presents<br />

the percentage of total variance distributed<br />

at each taxonomic level within the<br />

order Primates. Table 3 presents the correlation<br />

coefficients for comparisons of CBA<br />

<strong>and</strong> IRE for Primates <strong>and</strong> Haplorhini, along<br />

with sample sizes, degrees of freedom adjusted<br />

for phylogeny (df eff ), <strong>and</strong> their associated<br />

P-values (for which strepsirrhines<br />

had no significant correlations). Examination<br />

of the variance components in Table 2<br />

shows that the relationship between <strong>cranial</strong><br />

<strong>base</strong> angle <strong>and</strong> relative brain size is subject<br />

to significant phylogenetic effects. However,<br />

3 Unlike methods such as “independent contrasts” (e.g., Purvis<br />

<strong>and</strong> Rambaut, 1995), Smith’s method uses values for variables in<br />

just terminal taxa, therefore avoiding potentially spurious estimates<br />

of these values for ancestral nodes. Smith’s method is also<br />

more robust when five or fewer taxonomic levels are considered<br />

<strong>and</strong> is less affected by arbitrary taxonomic groupings (Nunn,<br />

1995). We also used the maximum likelihood method for calculating<br />

variance components rather than a nested ANOVA<br />

method because maximum likelihood does not generate negative<br />

variance components.<br />

the method of Smith (1994) is conservative<br />

<strong>and</strong> the correlations that survive these corrected<br />

degrees of freedom are robust, although<br />

of relatively low magnitude. Across<br />

<strong>primate</strong>s <strong>and</strong> haplorhines, both CBA4 <strong>and</strong><br />

CBA1 are significantly correlated (P 0.05)<br />

with IRE5 (Fig. 8; Table 2). <strong>The</strong>se results<br />

corroborate the results of Ross <strong>and</strong> Ravosa<br />

(1993), but with a more appropriate measure<br />

of basi<strong>cranial</strong> length incorporated into<br />

the measure of relative brain size. This confirms<br />

that brain size relative to basi<strong>cranial</strong><br />

length is significantly correlated with basi<strong>cranial</strong><br />

flexion, but that the correlations are<br />

not particularly strong, indicating that<br />

there may be other important influences on<br />

the degree of basi<strong>cranial</strong> flexion (see below).<br />

Most hypotheses explaining basi<strong>cranial</strong><br />

flexion have focused on increases in relative<br />

brain size as the variable driving flexion.<br />

However, as Strait (1999) has noted, it is<br />

important to consider the scaling relationships<br />

of basi<strong>cranial</strong> length <strong>and</strong> brain size.<br />

Using interspecific data from Ross <strong>and</strong> Ravosa<br />

(1993) in conjunction with other studies,<br />

Strait (1999) found that basi<strong>cranial</strong><br />

length scales with negative allometry<br />

against body mass <strong>and</strong> telencephalon volume<br />

(results confirmed here using BL2 instead<br />

of BL1; see Table 4), <strong>and</strong> BL2 also

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!