Figure 1.
Topographic maps of the ganglion cells in the retina of Rhinochimaera pacifica.
Topographic maps of the ganglion cells in the retina of the Pacific spookfish, Rhinochimaera pacifica. A. Akima interpolation. B. Thin plate spline interpolation. C. Thin plate spline smoother. D. Gaussian kernel smoother. E. Distribution density curves of the 4 models. F. Cumulative distribution functions for the four models; dash lines denote cell density higher of the 95% of the area of the retina. All the densities are in cells mm−2. The maps were predicted with a distance of 200 µmbetween points.
Figure 2.
Topographic maps of the ganglion cells showing area temporalis in Cephalopholis miniatus.
Topographic maps of the ganglion cells in the retina of the coral cod, Cephalopholis miniatus that show an area temporalis specialization. A. Modified map from Collin and Pettigrew, 1988a. B. Thin plate spline interpolation using a lambda value of 0. C. Thin plate spline under-smoothed with the 66% of count numbers for calculated degrees of freedom. D. Thin plate spline over-smoothed using the 33% of count numbers for calculated degrees of freedom. All the densities are ×104 cells mm−2. The maps were predicted with a distance of 200 µm between points.
Figure 3.
Diagnostic analysis of the maps for Cephalopholis miniatus.
Horizontal and vertical transects, and residual analysis of the two levels of smoothness for the coral cod, Cephalopholis miniatus. A. Horizontal transect crossing the highest density area, the dots represent observed points that are in the transect or at a distance not higher than 0.5 mm of the transect line. B. Vertical transect crossing the highest density area, the dots represent observed points that are in the transect or at a distance not higher than 0.5 mm of the transect line. C. Residual map for the under-smoothed model showing percentage of variation between the calculated data and the observed data, the contour lines represent the areas that varied more than 10% and 25%. D. Residual map for the over-smoothed model showing percentage of variation between the calculated data and the observed data, the contour lines represent the areas that varied more than 10% and 25%.
Figure 4.
Topographic maps of the ganglion cells showing horizontal streak and area in Choerodon albigena.
Topographic maps of the ganglion cells in the retina of the blue tusk fish, Choerodon albigena, that show a horizontal streak with a dorso-temporal area specialization. A. Modified map from Collin and Pettigrew, 1988b B. Thin plate spline interpolation using a lambda value of 0. C. Thin plate spline under-smoothed with the 66% of count numbers for calculated degrees of freedom. D. Thin plate spline over-smoothed using the 33% of count numbers for calculated degrees of freedom. All the densities are ×104 cells mm−2.
Figure 5.
Diagnostic analysis of the maps for Choerodon albigena.
Horizontal and vertical transects, and residual analysis of the two levels of smoothness for the blue tusk fish, Choerodon albigena. A. Horizontal transect crossing the highest density area, the dots represent observed points that are in the transect or at a distance not higher than 0.5 mm of the transect line. B. Vertical transect crossing the highest density area, the dots represent observed points that are in the transect or at a distance not higher than 0.5 mm of the transect line. C. Residual map for the under-smoothed model showing percentage of variation between the calculated data and the observed data, the contour lines represent the areas that varied more than 10% and 25%. D. Residual map for the over-smoothed model showing percentage of variation between the calculated data and the observed data, the contour lines represent the areas that varied more than 10% and 25%.
Figure 6.
Topographic maps of the ganglion cells showing multiple areae in Amblyglyphidodon curacao.
Topographic maps of the ganglion cells in the retina of the staghorn damselfish, Amblyglyphidodon curacao, that show multiple areae specializations. A. Modified map from Collin and Pettigrew, 1988a. B. Thin plate spline interpolation using a lambda value of 0. C. Thin plate spline under-smoothed map with the original isodensity contour levels. D. Thin plate spline under-smoothed map with lower number of isodensity contour levels. E. Horizontal transect crossing the highest density area, the dots represent observed data in no more than 0.5 mm to the transect line. F. Residual map for the over-smoothed model showing percentage of variation between the calculated data and the observed data, the contour lines represent the areas that varied more than 10% and 25%. All the densities are ×104 cells mm−2.
Figure 7.
Topographic maps of the ganglion cells showing dorsal area specialization in Gymnocranius bitorquatus.
Topographic maps of the ganglion cells in the retina of the collared sea bream, Gymnocranius bitorquatus, that show a dorsal area specialization. A. Modified map from Collin and Pettigrew, 1988b. B. Thin plate spline interpolation using a lambda value of 0. C. Thin plate spline under-smoothed map. D. Horizontal transect crossing the supposed streak specialization. E. Vertical transect crossing the highest cell density area, the dots represent observed points that are in the transect or at a distance not higher than 0.5 mm of the transect line. F. Residual map for the over-smoothed model showing percentage of variation between the calculated data and the observed data, the contour lines represent the areas that varied more than 10% and 25%. All the densities are ×104 cells mm−2.
Figure 8.
Topographic maps of the ganglion cells showing three areae specializations in Parapercis cylindrica.
Topographic maps of the Ganglion cells in the retina of the sandperch, Parapercis cylindrica, that shows a horizontal streak with three areae specializations. A. Modified map from Collin and Pettigrew, 1988a. B. Thin plate spline interpolation using a lambda value of 0. C. Thin plate spline under-smoothed. D. Horizontal transect crossing the horizontal streak specialization. E. Vertical transect crossing the middle area specialization, the dots represent observed points that are in the transect or at a distance not higher than 0.5 mm of the transect line. F. Residual map for the over-smoothed model showing percentage of variation between the calculated data and the observed data, the contour lines represent the areas that varied more than 10% and 25%. All the densities are ×104 cells mm−2.
Figure 9.
Topographic maps of the ganglion cells showing a fovea in Conocara murrayi.
Topographic maps of the ganglion cells in the retina of the smooth-head fish, Conocara murrayi, which shows a fovea. A. Modified map from Collin and Patridge, 1996. B. Thin plate spline interpolation using a lambda value of 0. C. Thin plate spline under-smoothed. D. Hybrid model with Tps smoothed to GCV in the periphery and cubic interpolation in the fovea region. E. Horizontal transect crossing the fovea, the dots represent observed data 0.5 mm the transect line. F. Detail of the fovea region using Tps interpolation. All the densities are ×103 cells mm−2.