Recall that when referring to non-homology, we are stating that a character present in a lineage has indendently evolved when compared to another lineage, and the most recent common ancestor did not possess the character of interest. For the case of compound eyes within insects and crustaceans, this means that compound eyes evolved multiple times from a separate origin. Keep Dollo's Law in mind when reading this section, which states that detailed similarities found in a structure are unlikely to evolve more than once! (Oakley, 2003).

One study found a possible independent origin of compound eyes in a group of crustaceans called Myodocopid Ostracods. The Ostracoda are an ancient group of bivalved crustaceans with a fossil record dating back 500 million years. With our current fossil records, this dates the appearance of Ostracods significantly after the Cambrian Explosion, which occurred 530-540 million years ago, when compound eyes were already believed to have evolved. Ostracods can be divided into three major clades: Podocopa, Palaeocopa, and Myodocopa. Myodocopa can be further divided into the clades Halocyprida and Myodocopida. Podocopa, Palaeocopa, and the majority of Myodocopa (including Halocyprida) possess median eyes, which are non-image-forming eyes that detect light. However, the Myodocopida are the only Ostracods that possess a pair of lateral compound eyes along with their ancestrally retained median eye. Using molecular data to create a phylogenetic reconstruction of Ostracods, this study finds strong support for the independent origin of compound eyes, separate from other crustacean lineages and thus the insects (Oakley, et. al., 2002).

The study used primers to build a complete sequence for DNA encoding 18S rRNA and a partial sequence of 28S rRNA to run a maximum likelihood phylogenetic analysis of ancestral states and the presence or absence of the median and compound eyes. Two Maxillopods groups with median eyes were chosen as outgroups, which are taxon groups outside of the group of interest, to allow for a basal comparison to the group of interest. The two Maxillopod groups included Argulus sp., Branchyura, and Tigriopus, Copepoda. Both of these outgroups possess median eyes, but Argulus also possesses a compound eye. By comparing the outgroups with the Ostracod lineage using a maximum likelihood ancestral reconstruction, they found significant support for the homology of the median eyes and the independent origin of compound eyes in Myodocopida. The image below (Fig. 1) displays their phylogenetic reconstruction (Oakley, et. al., 2002).

Pie charts to the left (A) depict maximum likelihood support for the presence (purple) and absence of median eyes (white) in Ostracods. The asterisks present at ancestral nodes indicate statistically significant results. Since there is a higher proportion of purple relative to white, there is stronger support for the homology of the median eye. In the case of the most recent common ancestor of Myodocopida, the pie chart is fully purple, and there is undeniable support for homology of the median eye derived from the ancestral nodes. Pie charts to the right (B) depict maximum likelihood support for the presence (yellow) and absence (white) of compound eyes in Ostracods. The numbers displayed at each ancestral node are bootstrap values, indicating stronger support for results the closer the value is to 100. Considering the little amount of yellow present in the pie charts in all groups other than Myodocopida, it is very likely that the ancestors of Myodocopida did not possess a compound eye. Suddenly, at the most recent common ancestor of Myodocopida, the pie chart is fully yellow with a bootstrap value of 100. This means that compound eyes evolved independently from all of the other Ostracod groups while still retaining their ancestral median eye. This can be seen in images C (dorsal view) and D (lateral view), with purple shading representing the median eye and yellow shading representing the compound eye (Oakley, et, al., 2002).

In addition to this ancestral reconstruction, other studies have found some morphological differences in compound eyes when compared to those of most other arthropods. Recall that in the "Evidence for Homology" page, most arthrpods typically have 8 retinular cells and 4 crystalline cone cells making up their ommatida (Oakley, 2003). However, Ostracod compound eyes have been found to contain 6 retinular cells and 2 crystalline cone cells. This makes Ostracod compound eyes morphologically different than the compound eyes of other arthropods. There could be two explanations for this: either the compound eyes of Ostracods evolved independently and are non-homologous to other arthropod compound eyes, or mutations causing the loss of retinular and crystalline cone cells were favored for this group (Oakley, et. al., 2002; Harris, et. al., 1976). Perhaps we do not have the full fossil record of ostracods, implying that there are still many undiscovered species that may change the results found in Figure 1. 

It is evident that more research is needed to fully understand why these differences in ommatidial structure occur. It is especially important that we look into other groups that differ from the 8/4 pattern, as Ostracods are not the only group that deviates in their cell number (Zrzavý, et. al., 1979).