Cytoarchitecture of the oculomotor nuclei
The morphology of nIII, nIV, and nVI was similar across all species examined. For every species examined, we were able to distinguish dm-nIII, dl-nIII, and v-nIII (for some examples see Fig. 2). EW could be identified in almost all species, except for some owls. Of the eight owl species examined, EW could only be identified in the three largest species: great horned owl (Bubo virginianus), snowy owl (Bubo scandiacus), and great grey owl (Strix nebulosa) (Fig. 2l). For the other five, smaller owl species, we were not able to identify any structure as EW in our Nissl stained sections (Fig. 2h). In these smaller owls, a diffuse collection of neurons was observed dorsal to dl-nIII, where EW is normally present, but it lacked the typical cytoarchitectural features of EW and was therefore not distinguishable from surrounding tissue.
Oculomotor nucleus (nIII)
The allometric relationship between the total volume of nIII and brainstem volume was close to isometry and explained 78% of the variation in nIII volume (Fig. 3a, Table 3), however, at larger brainstem volumes there was more variability around the regression line. As shown in Fig. 3a, owls (black dots) had relatively smaller nIII volumes, relative to brainstem volume, than other birds as measured by a change in the intercept (-14%, Table 4). Although the hawks and falcons tended to have relatively larger nIII volumes, the differences were not significant (Table 4).
Table 3
Details of the phylogenetic generalized least squares regression analyses of brain region volume against brainstem volume and number of neurons within each brain region plotted against brain region volumes across 71 bird species. l refers to the phylogenetic signal (Pagel, 1999) of each regression line. All data were log10-transformed prior to analysis. The brain regions are as follows: dl-nIII – dorsolateral oculomotor nucleus; dm-nIII – dorsomedial oculomotor nucleus; v-nIII – ventral oculomotor nucleus; nIV – trochlear nucleus; nVI –abducens nucleus, EW – Edinger-Westphal nucleus.
x-axis | y-axis | λ | F | p | r2 | Intercept | Slope (95% CI) |
brainstem volume | total nIII volume | 0.842 | 245.1 | < 0.01 | 0.777 | -2.445 | 0.904 (0.846–0.962) |
| dm-nIII volume | 0.906 | 223.8 | < 0.01 | 0.761 | -3.015 | 0.897 (0.837–0.957) |
| dl-nIII volume | 0.721 | 217.8 | < 0.01 | 0.756 | -3.092 | 0.905 (0.844–0.966) |
| v-nIII volume | 0.865 | 176.1 | < 0.01 | 0.714 | -2.779 | 0.918 (0.849–0.987) |
| nIV volume | 0.753 | 222.3 | < 0.01 | 0.760 | -2.929 | 0.925 (0.863–0.987) |
| nVI volume | 0.618 | 189.1 | < 0.01 | 0.732 | -2.888 | 0.902 (0.837–0.967) |
| EW volume | 0.830 | 79.9 | < 0.01 | 0.548 | -3.123 | 0.821 (0.729–0.913) |
total nIII volume | total nIII #neurons | 0.146 | 444.1 | < 0.01 | 0.864 | 4.350 | 0.454 (0.433–0.475) |
dm-nIII volume | dm-nIII #neurons | 0.238 | 348.4 | < 0.01 | 0.832 | 4.034 | 0.447 (0.423–0.471) |
dl-nIII volume | dl-nIII #neurons | 0.013 | 620.9 | < 0.01 | 0.898 | 4.108 | 0.469 (0.450–0.488) |
v-nIII volume | v-nIII #neurons | 0.056 | 303.3 | < 0.01 | 0.812 | 4.128 | 0.438 (0.413–0.463) |
nIV volume | nIV #neurons | 0.268 | 179.9 | < 0.01 | 0.719 | 4.053 | 0.465 (0.430–0.500) |
nVI volume | nVI #neurons | 0.267 | 160.6 | < 0.01 | 0.698 | 4.025 | 0.483 (0.445–0.521) |
EW volume | EW #neurons | 0.393 | 200.0 | < 0.01 | 0.754 | 4.315 | 0.524 (0.487–0.561) |
Table 4
p-values from phylogenetic analyses of covariance (pANCOVAs) testing for differences in intercept (int) and slope for volume and number of neurons in each nucleus relative to brainstem volume across nine order for which we had three or more species in our dataset. The brain regions are as follows: dl-nIII – dorsolateral oculomotor nucleus; dm-nIII – dorsomedial oculomotor nucleus; v-nIII – ventral oculomotor nucleus; nIV – trochlear nucleus; nVI –abducens nucleus, EW – Edinger-Westphal nucleus.
| Accipitriformes | Anseriformes | Apodiformes | Columbiformes | Falconiformes | Galliformes | Passeriformes | Psittaciformes | Strigiformes |
Measurement | int | slope | int | slope | int | slope | int | slope | int | slope | int | slope | int | slope | int | slope | int | slope |
nIII volume | 0.10 | 0.90 | 0.49 | 0.85 | 0.32 | 0.36 | 0.92 | 0.71 | 0.27 | 0.71 | 0.64 | 0.35 | 0.65 | 0.08 | 0.59 | 0.16 | < 0.01 | 0.06 |
nIII #neurons | 0.50 | 0.06 | 0.06 | 0.88 | 0.06 | 0.21 | 0.43 | 0.98 | 0.48 | 0.05 | 0.67 | 0.31 | 0.19 | 0.81 | 0.77 | 0.92 | 0.96 | 0.92 |
dm-nIII volume | 0.11 | 0.43 | 0.57 | 0.89 | 0.91 | 0.11 | 0.96 | 0.10 | 0.25 | 0.43 | 0.50 | 0.50 | 0.59 | 0.20 | 0.78 | 0.34 | < 0.01 | 0.58 |
dl-nIII volume | 0.01 | 0.53 | 0.49 | 0.51 | 0.93 | 0.90 | 0.72 | 0.67 | 0.17 | 0.25 | 0.58 | 0.80 | 0.67 | 0.30 | 0.91 | 0.16 | < 0.01 | 0.63 |
v-nIII volume | 0.30 | 0.92 | 0.55 | 0.96 | 0.13 | 0.41 | 0.92 | 0.79 | 0.44 | 0.86 | 0.79 | 0.25 | 0.67 | 0.06 | 0.44 | 0.18 | < 0.01 | 0.01 |
dm-nIII #neurons | 0.74 | 0.09 | 0.01 | 0.21 | 0.09 | 0.52 | 0.76 | 0.83 | 0.33 | 0.05 | 0.72 | 0.58 | 0.57 | 0.44 | 0.15 | 0.65 | 0.77 | 0.08 |
dl-nIII #neurons | 0.46 | 0.14 | 0.06 | 0.72 | 0.32 | 0.40 | 0.21 | 0.92 | 0.29 | 0.32 | 0.59 | 0.57 | 0.45 | 0.82 | 0.70 | 0.39 | 0.52 | 0.77 |
v-nIII #neurons | 0.25 | 0.22 | 0.08 | 0.81 | 0.06 | 0.21 | 0.48 | 0.86 | 0.71 | 0.11 | 0.47 | 0.32 | 0.06 | 0.36 | 0.60 | 0.84 | 0.72 | 0.32 |
nIV volume | 0.65 | 0.98 | 0.86 | 0.07 | 0.05 | 0.32 | 0.47 | 0.61 | 0.22 | 0.05 | 0.79 | 0.84 | 0.73 | 0.17 | 0.56 | 0.73 | < 0.01 | 0.40 |
nIV #neurons | < 0.01 | 0.97 | 0.86 | 0.57 | 0.43 | 0.15 | 0.25 | 0.76 | 0.48 | 0.16 | 0.47 | 0.08 | 0.78 | 0.84 | 0.82 | 0.69 | < 0.01 | 0.56 |
nVI volume | < 0.01 | 0.16 | 0.30 | 0.43 | 0.87 | 0.77 | 0.85 | 0.48 | 0.02 | 0.26 | 0.91 | 0.12 | 0.81 | 0.31 | 0.61 | 0.07 | 0.02 | 0.27 |
nVI #neurons | 0.47 | 0.90 | 0.37 | 0.08 | 0.16 | 0.26 | 0.53 | 0.38 | 0.27 | 0.53 | 0.16 | 0.21 | 0.01 | 0.76 | 0.75 | 0.65 | 0.33 | 0.67 |
EW volume | 0.70 | 0.10 | 0.45 | 0.55 | 0.08 | 0.88 | 0.48 | 0.61 | 0.43 | 0.71 | 0.39 | 0.83 | < 0.01 | 0.19 | 0.28 | 0.53 | < 0.01 | 0.59 |
EW #neurons | 0.44 | 0.62 | 0.86 | 0.49 | 0.54 | 0.91 | 0.97 | 0.26 | 0.34 | 0.13 | 0.72 | 0.16 | < 0.01 | 0.34 | 0.11 | 0.34 | 0.36 | 0.75 |
Neurons numbers and nIII volume were strongly correlated with one another (Fig. 3b, Table 3) and there was far less scatter around the regression line than that observed for volumes. Although some clades appeared to have relatively fewer neurons than others (Fig. 3b), we found no significant differences across orders in our pANCOVAs (Table 4).
When we focused our comparisons to the allometric scaling of the subdivisions of nIII, the results were similar. All three subdivisions shared similar slopes and correlation coefficients when regressed against brainstem volume (Figs. 3c,e,g, Table 3). As with the entire volume of nIII (Fig. 3a), there was considerable scatter around the regression lines for all three subdivisions. Our pANCOVAs revealed that owls have significantly smaller dm-nIII (-11%, Fig. 3c, Table 4), dl-nIII (-11%, Fig. 3e, Table 4), and v-nIII (-14%; Fig. 3g, Table 4) volumes, relative to brainstem volume, than other birds. For the relative size of v-nIII, owls also had a significantly higher slope for this relationship when compared with other clades (+ 113%, Fig. 3g, Table 4). At the other end of the spectrum, the relative size of dl-nIII was significantly larger in hawks (+ 9%, Fig. 3e, Table 4) and the falcons did not differ significantly from other birds for any of the subdivision volumes (Table 4).
As with the total number of neurons within nIII (Fig. 3b), the volumes and neuron numbers of the subdivisions were strongly correlated with one another (Figs. 3d,f,h, Table 3), albeit with shallower slopes than that of subdivision volume versus brainstem volume (Table 3). Across all three pANCOVAs, only one significant difference was detected: waterfowl had relatively fewer neurons in dm-nIII compared to other orders (-2%, Fig. 3d, Table 4). This difference in waterfowl appeared to be largely due to the relatively low number of dm-nIII neurons in the blue-winged teal (Spatula discors).
Trochlear nucleus (nIV)
Similar to nIII, the allometric relationship between nIV and brainstem volume was close to isometry and explained 76% of the variation in nIV volume (Fig. 4a, Table 3). Owls differed from other clades in having smaller nIV volumes for their brainstem volume (-12%, Table 4).
Most of the variation in nIV neurons numbers was due to nIV volume (72%, Fig. 4b, Table 4), but species were scattered widely around the regression line compared to the data for nIII (Figs. 3b,d,f,h). Unlike nIII, owls were above the regression line (Fig. 4b) and had significantly more neurons, relative to nIV volume, than other clades (+ 3%, Table 4). Conversely, hawks and New World vultures have relatively fewer neurons (Fig. 4b). We only have data for two vulture species, so they were not analyzed in our pANCOVAs, but the hawks have proportionately fewer neurons than other groups (-3%, Fig. 4b, Table 4). Hummingbirds also appeared to have relatively large nIV volumes (Fig. 4a), but did not quite reach statistical significance in our pANCOVA (Table 4).
Abducens nucleus (nVI)
Similar to nIII and nIV, the allometric relationship between nVI and brainstem volume was also close to isometry and explained 73% of the variation in nVI volume (Fig. 4c, Table 3). Once more, owls were below the regression line and had significantly smaller nVI volumes, relative to the brainstem, than other birds (-8%, Table 4). Above the regression line, hawks, New World vultures, and falcons stood out as having enlarged nVI volumes. Indeed, both hawks (+ 9%) and falcons (+ 10%) have significantly larger VI volumes than other birds (Table 4). However, as mentioned previously, the New World vultures were not included in these pANCOVAs due to only being able to sample two species.
As with the other nuclei examined, most of the variation in the number of nVI neurons can be attributed to nVI volume (70%, Fig. 4d, Table 4). Although the correlation coefficient for the relationship between nVI volume and neuron numbers is similar to that of nIII and nIV, the scatter around the regression line appeared to be greater, with several songbirds possessing more neurons than expected (i.e., higher density). Our pANCOVA showed that songbirds have relatively more nVI neurons than other birds (+ 3%, Fig. 4c, Table 4), but no other differences across clades were detected (Table 4).
Edinger-Westphal nucleus (EW)
The allometric relationship between EW and brainstem volume was close to isometry (Fig. 5a, Table 3), but only 55% of the variation in EW volume was explained by brainstem volume (Table 3). As noted above, EW was only recognizable in 3/7 owl species. Nevertheless, these three, larger owl species had significantly smaller EW volumes, relative to brainstem volume, than other birds (-14%, Table 4). Located close to these three owls were two other species: the tawny frogmouth (inverted triangle) and Eurasian woodcock (Scolopax rusticola, asterisk). It is worth noting that the position of the woodcock in this space is vastly different from that of the only other charadriiform species sampled, Bonaparte’s gull (Chroicocephalus philadelphia). Although the gull and woodcock share similar brainstem volumes, the former is above the regression line, whereas the latter is below it. At the other end of the spectrum, and somewhat unexpectedly, songbirds had relatively larger EW volumes (+ 13%, Table 4).
In a similar fashion to the other nuclei examined, most of the variation in the number of EW neurons can be attributed to EW volume (Fig. 5b, Table 3). With respect to differences across clades, songbirds emerged as having significantly more neurons, relative to EW volume (+ 3%, Table 4). Thus, songbirds tend to have higher EW neuronal density than other birds, as well as proportionately larger EW volumes.
One of our predictions was that pursuit diving species would have a relatively large EW because of their higher accommodative power (Sivak 1980; Katzir and Howland 2003; Machovsky-Capuska et al. 2012). We were only able to secure a single pursuit diving species for this study, the red-breasted merganser (Mergus serrator, Fig. 2e). The merganser did not stand out in the comparisons across all species, so we examined its EW volume and neuron numbers with respect to other waterfowl. As shown in Fig. 5c, the merganser has a much larger EW than that predicted by brainstem volume across a sample of nine waterfowl species. The enlargement of EW in the merganser appears to be due primarily to the addition of neurons, without a change in neuronal density (Fig. 5d).