Problems with Structural Definitions of a Segment
If segments are defined according to a set of substructures that appear together, this immediately leads to the problem that only in the rarest cases are all these criteria are truly satisfied. Despite of the importance of serial coelomic cavities for the definition of segments (see earlier), adult arthropods are considered as segmented animals, although they lack these. Yet, the problem with a structural definition of segments pertains not only when different species are compared, but also within organisms, in particular when body sections would not be viewed as segments although they belong to the same series. If, for example, both nerve ganglia and extremities are absent, as is the case in the abdomen of many crustaceans, are the body sections still considered segments just based on the external rings? Other cases raise the same question: wasps lack external rings in the thorax and have no legs in the abdomen, and in many annelids, the genital organs are limited to a few specialized body sections.
Problems with Ontogenetic Definitions of a Segment
If formation through a posterior budding zone distinguishes segments from other serial structures, then the body sections that are not formed in this way are not segments. Classically, this also applies for the anteriormost (prostomium in annelids, acron in arthropods) and posteriormost (pygidium in annelids, telson in arthropods) terminal body sections of segmented animals, as they are already present prior to the differentiation of the budding zone, which then emerges between them (see Westheide and Rieger 2007). As previously discussed, however, segments are not always formed through a posterior budding zone. Frequently, several of the anterior segments are already differentiated before this budding zone becomes active. In animals such as Drosophila, all segments emerge almost simultaneously and independent of a budding zone. In a close reading of the classical definition, the anterior serial body sections would not be segments and Drosophila would not be regarded as segmented at all.
Similar problems arise if the expression of segment polarity genes is used as the basis for determining segmentation criteria. For instance, the engrailed gene is expressed in the forming segments of arthropods, annelids, and chordates (e.g., Scholtz and Dohle 1996; Holland et al. 1997; Prud’homme et al. 2003). Thus, this expression has been used to homologize segments and to discriminate them from other serially repeated structures. However, in arthropods, annelids, and chordates, engrailed and other segmentation genes are expressed in different regions of the segment anlage with respect to germ layers and segment boundaries, and sometimes they are not segmentally expressed at all (e.g., Holland et al. 2000). And differences like this can occur even within one of the segmented groups, as has been exemplified by comparative gene expression studies in Annelida (Seaver and Kaneshige 2006). Genes involved in segmentation are also expressed in the terminal regions of annelids and arthropods, and thus the anterior and posterior terminal regions were classified as segments (e.g., Starunov et al. 2015), although they have been interpreted as non-segmental body units based on structural and developmental grounds (for discussion, see Scholtz and Edgecombe 2006). Moreover, most of these segmentation genes are also expressed in the nervous system (e.g., Scholtz and Dohle 1996; Seaver and Kaneshige 2006). Thus, not every structure that is formed in which these genes are active will definitely become a segment and not all segments express corresponding genes.
There are also regions with transitory expression that do not lead to the formation of segments.
A recently published study now shows, however, that also unsegmented animals use these genes to distinguish very different morphological boundaries such as coelom spaces, brain sections, external body rings, or dorsal shell plates (Vellutini and Hejnol 2016; see Chapter 8). The serial arrangement of gene expression might not determine segment formation as such, but rather the establishment of morphogenetic boundaries as a starting point for the differentiation of additional serially arranged structures.
-  Traditionally, this absence was considered as being of minor importance, since serial coelom anlagenwere described during early development. Yet, the homology of the ontogenetic anlage of coeloms isincreasingly viewed critically (Koch et al. 2014). With that, the classical essential quality of segmentsis also lost.