Development and evolution of segmentation assessed by geometric morphometrics: The centipede Strigamia maritima as a case study

Highlights

• Morphometrics is an ideal tool for studying the evolution of segmentation.

• Morphometric analyses allow to tease apart several sources of phenotypic variation.

• In a centipede, canalization and developmental stability are possibly associated.

Abstract

Using the centipede model species Strigamia maritima as a subject of study, we illustrate the potential of geometric morphometrics for investigating the development and evolution of segmentation, with a specific focus on post-embryonic segmental patterning. We show how these techniques can contribute detailed descriptive data for comparative purposes, but also precious information on some features of the developmental system that are considered relevant for the evolvability of a segmented body architecture, such as developmental stability and canalization. Morphometric analyses allow to separately investigate several sources of phenotypic variation along a segmented body axis, like constitutive and random segment heteronomy, both within and among individuals. Specifically, in S. maritima, the segmental pattern of ventral sclerite shapes mirrors that of their bilateral fluctuating asymmetry and among-individual variation in associating the most anterior and most posterior segments in diverging from the central ones. Also, among segments, there seems to be a correlation between fluctuating asymmetry and shape variation among individuals, suggesting that canalization and developmental stability are somehow associated. Overall, these associations might stem from a joint influence of the segmental position on the two processes of developmental buffering.

Introduction

Segmentation is a key feature of arthropod body architecture, and as such considerable attention has been paid to its development and evolution, also in relation to tagmosis, the morpho-functional regionalization of the main body axis (reviewed in Fusco and Minelli, 2013).

The evolution of segmentation as a body feature, in connection to the developmental process of segmentation itself, is a favourite subject of evolutionary developmental biology (evo-devo), as it is thought that the developmental mechanism of segmentation, along with the network of genetic interactions that controls it, can have significant influence on the evolution of segmentation as a morphological trait (e.g., Davis and Patel, 1999, Peel et al., 2005, Vroomans et al., 2016). However, most of the more recent studies on the development of segmentation have focused on the embryonic (often, early-embryonic) phase of development. This has been done using techniques like in-situ hybridization to study gene expression, parental and embryonic RNA interference to study gene function, and sequencing of transcriptomes and genomes to reveal the complete gene repertoires of some of these animals (Leite and McGregor, 2016). However, despite remarkable recent progresses in developmental biology, especially in comparative developmental genetics, the study of post-embryonic development still remains insufficiently investigated, if compared to the preceding phase of ontogeny. This is particularly unfortunate for arthropods, where developmental processes of segmentation and tagmosis are not restricted to embryogenesis, but continue prominently through post-embryonic life (Minelli and Fusco, 2013).

Comparative analysis of the evolution of segmentation, including segmental patterning, needs other types of data, typically morphometric data, and different analytical approaches, such as those offered by geometric morphometric (GM) methods, a suite of analytical tools that provide a statistical description of biological forms in terms of their size and geometric shape (e.g., Klingenberg, 2010).

Morphological variation along a segmental series (segment heteronomy) within an individual has at least two different components. The first is constitutive segmental variation in size and shape of the elements of the series, corresponding to the more general concept of target phenotype, i.e. the phenotype specified by the genetic makeup of the organism and the environmental conditions during its development (Nijhout and Davidowitz, 2003, Fusco and Minelli, 2010). The second is random segmental variation, i.e. variation around the target phenotype produced by developmental noise, that can be manifested as deviations from the expected body symmetry, or fluctuating asymmetry (see below) (Fusco and Minelli, 2000a, Savriama et al., 2016). Both constitutive and random segmental variation, metaphorically the “signal” and the “noise”, are of high interest for both developmental studies, across ontogenetic stages within species, and evolutionary studies, comparing species in a phylogenetic context.

At the same time, phenotypic variation observed in a population is the outcome of two opposing sets of influences: on the one hand are the sources of variation, including genetic and environmental differences among individuals and developmental noise; on the other hand is a set of developmental processes that buffer those variations, including canalization and developmental stability (e.g., Debat and David, 2001; see below). Because developmental noise is difficult to set apart from environmental effects, a strategy is to focus on within-individual variation among repeated body parts. These parts are indeed genetically identical and are usually facing the same environmental conditions, and thus differ only by stochastic differences attributable to developmental noise (e.g., Van Valen, 1962).

Within-individual deviations from the expected body symmetry are known as fluctuating asymmetry (FA) (e.g., Debat and David, 2001, Debat and Peronnet, 2013, Klingenberg, 2015). In organisms with bilateral symmetry, random deviations from left-right symmetry, or bilateral FA (Palmer and Strobeck, 1986), are often employed to investigate developmental stability, i.e. the ability of an organism to buffer random perturbations of its developmental process (Nijhout and Davidowitz, 2003, Fusco and Minelli, 2010). However, for segmented animals, or for organisms with a modular body organization in general, other types of body symmetry, specific of their body architecture, can be exploited to study developmental stability by means of FA (Savriama and Klingenberg, 2011). Along their main body axis, arthropods, like other segmented animals (Minelli and Fusco, 2004), present translational symmetry, which can be effectively exploited through the analysis of translational FA for the study of developmental stability (e.g., Savriama et al., 2016).

Among-individual phenotypic variation results from the interplay between genetic and environmental influences and canalization. The term canalization (Waddington, 1942) refers to the ability of the developmental system to buffer such influences, with the effect of limiting phenotypic variation (e.g., Debat and David, 2001, Fusco and Minelli, 2010). Although the observed variation is thus the manifestation of both the sources of variation and canalization, differences in variation among individuals among groups are often interpreted as differences in canalization (e.g., Clarke, 1998, Debat et al., 2009, Breno et al., 2011; Lazić et al., 2015), assuming that the sources of variation are globally constant.

Whether developmental stability and canalization are independent features of a developmental system has been a contentious issue (e.g., Klingenberg and McIntyre, 1998, Debat et al., 2000, Hallgrímsson et al., 2002). The most commonly used approach to investigate the link between canalization and developmental stability has been to compare, among groups, the levels of variation among individuals and FA. The use of GM methods has provided an additional, more subtle criterion: a similarity in the patterns of shape variation among individuals and between sides within individuals would suggest that similar developmental processes are acting at both levels, and that canalization and stability are related. Conversely, if these patterns are different, then canalization and stability might involve different processes. The literature has so far provided contrasting results (see Debat et al., 2009, Klingenberg, 2015 for reviews). In this context, an obviously appealing feature of segmented animals, is that they offer as many traits as their number of segments to assess this relationship, with the additional question of whether it might differ across segments, and be influenced by the overall heteronomy.

Here, we illustrate the potential of geometric morphometrics for investigating development and evolution of segmental patterning, using the model centipede species Strigamia maritima as the subject of study. In particular, we present some general morphometric approaches for the study of size and shape segment differentiation (constitutive heteronomy) and bilateral and translational FA. We focus in particular on the patterns of bilateral shape asymmetry of segments (as a proxy for developmental stability) and shape variation among individuals (as a proxy for canalization), and explore whether they might differ regarding the position of segments along the main body axis.

In the last two decades, together with the pill millipede Glomeris marginata, the centipede S. maritima (Fig. 1A) has become a favourite subject of myriapod developmental biology, with a series of studies on its embryonic (Chipman et al., 2004, Brena and Akam, 2012) and early-postembryonic (Brena, 2014) development, and developmental gene expression (Brena et al., 2006, Chipman and Akam, 2008, Brena and Akam, 2013, Brena, 2015). S. maritima is also the only myriapod with a fully sequenced genome (Chipman et al., 2014). As typical of geophilomorph centipedes (Chilopoda, Geophilomorpha), S. maritima presents a highly polymerous and a rather homonomous (i.e., morphologically little-differentiated) segmental body organization, with respect to other arthropods. Nonetheless, as we will show, detailed quantitative morphometric analyses, that can be equally applied to more complex segmental patterns, reveal a surprising richness in segmental patterning, in the relationship between different aspects of developmental control at the level of segments, and in the connections between segmental patterning and developmental control.

The quantitative analysis of arthropod segmental patterning has repeatedly been an object of interest in morphological studies on the group, although always based on traditional (meristic or distance measurements) morphometric analysis (e.g., Enghoff, 1986, Minelli et al., 1996, Berto et al., 1997, Fusco and Minelli, 2000b, Bonato et al., 2011, Ivanenko et al., 2016) and generally considering only one or very few specimens of the same species. Centipede segmental patterning has also been used as a model for developing an index of morphological complexity (Fusco and Minelli, 2000a). On the contrary, the list of studies dedicated to the random deviations from constitutive heteronomy, i.e. studies on FA in relation to segmentation, or translational FA, reduces to a couple of pioneering studies (Astaurov, 1930, Fusco and Minelli, 2000b) and a very recent one (Savriama et al., 2016), the only making use of geometric morphometry.

The selection of GM analyses that we present here applied to S. maritima trunk segmentation can all be employed in other segmented organisms, and have thus the potential to generate suitable comparative data for the study of the evolution of segmental patterning and its developmental control in wide phylogenetic contexts.