Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jul 15.
Published in final edited form as: Methods Cell Biol. 2011;101:225–248. doi: 10.1016/B978-0-12-387036-0.00011-6

Morphogenesis of the Zebrafish Jaw: Development Beyond the Embryo

Kevin J Parsons ★, Viktoria Andreeva †, W James Cooper ★, Pamela C Yelick †, R Craig Albertson ★
PMCID: PMC13367376  NIHMSID: NIHMS2192004  PMID: 21550447

Abstract

The zebrafish has emerged as an important model for vertebrate development as it relates to human health and disease. Work in this system has provided significant insights into the variety of genetic signals that direct the cellular activities and tissue interactions necessary for proper assembly of the pharyngeal skeleton. Unfortunately our understanding of craniofacial development beyond embryonic stages is far less complete. Stated another way, we know a great deal about the early patterning of the skull, but we know comparatively little about how mature craniofacial shape is determined and maintained over time. Here we propose ways to expand the current molecular genetic paradigm beyond the embryo to gain an understanding of the processes and mechanisms that guide growth and remodeling of mineralized craniofacial, skeletal, and dental tissues. First, we discuss sources of adult mutant phenotypes that can be used to study of postembryonic development. Next, we review salient quantitative methods that are necessary to define complex adult phenotypes. We also discuss how other organismal systems can be used to inform and complement studies in zebrafish. We conclude by discussing the implications for such studies within the context of furthering an understanding of the etiology and pathophysiology of human craniofacial malformations, as well as informing an understanding of adaptive craniofacial variation among natural populations.

I. Postembryonic Development – Framing the Questions and Understanding the Challenges

In stark contrast to the extensive body of literature focused on understanding the basis of early craniofacial patterning, relatively little is known about the mechanisms that underlie skeletal development beyond embryonic stages. Craniofacial development can be broadly divided into chronological stages based on the appearance of tissues, the formation of structures, and the initiation of various morphogenic processes. The cranial neural crest (CNC) cells are the first to arise during embryonic development that directly contribute to the craniofacial skeleton. A PubMed literature search (May 12, 2010) for CNC identified nearly 1400 peer-reviewed research articles. By early larval stages of development the CNC cells condense and differentiate to form craniofacial cartilages, which soon thereafter develop into pharyngeal bones and other components of the skull (Hall, 1999). A search for craniofacial “cartilage development” returned over 600 articles, whereas craniofacial “bone development” identified 200 references. Bone growth and remodeling are initiated during larval development, and continue throughout juvenile and adult stages. A PubMed search for craniofacial “bone growth” and “bone remodeling” each identified fewer than 100 articles.

This brief survey of the literature reveals a dearth of peer-reviewed articles that explore craniofacial development beyond embryonic stages. It also reveals an inverse relationship between developmental chronology and our understanding of the mechanisms that regulate this process. The main purpose of this chapter is to describe ways in which the experimentally tractable zebrafish can be used to promote a more robust understanding of craniofacial development beyond the embryo.

The zebrafish is an excellent vertebrate model with which we study skeletogenesis at all stages of development. They are easily bred and maintained, have short generation times, and large numbers of progeny can be obtained from a single mating (~200–500/week). The zebrafish genome has been sequenced and zebrafish developmental genetics are easily manipulated through functional analyses, transplantation, and transgenics. Zebrafish are also touted for having transparent embryos, allowing various developmental processes to be observed in the living embryo with minimal processing or manipulation. This attribute also holds true for postembryonic zebrafish, particularly for the skeletal system. Most craniofacial elements lie just underneath a relatively thin layer of tissue and are easily observed in live or fixed animals with minimal processing (Fig. 1).

Fig. 1.

Fig. 1

Open in a new tab

Skeletal preparations in two adult zebrafish. (A) A living fish stained with the fluorescent compound, Calcein (Sigma), which binds to free calcium (note the cranial sutures). (B) An adult AB zebrafish that was fixed, enzymatically cleared, and stained with Alizarin red that also binds to calcium (Sigma, St. Louis, MO). In both specimens detailed skeletal anatomy is apparent.

Another relevant attribute of the zebrafish is that, like most teleosts, they exhibit dynamic patterns of allometric growth, whereby anatomical changes accompany their increase in size (Fuiman, 1983; Hernandez, 2000; Osse, 1990; Weatherly, 1990). The life cycles of bony fishes are often characterized by dramatic shifts in ecology and behavior, accompanied by concomitant changes in the body shape (Loy et al., 2001; Zelditch and Fink, 1995). Allometric growth is even considered to be a defining characteristic of fish early life histories (Hernandez, 2000). The use of a model that exhibits pronounced size-related shape change will likely facilitate our efforts to understand the mechanisms that underlie skeletal growth and remodeling.

Zebrafish craniofacial mutants have traditionally been characterized using qualitative methods (i.e., the gross presence or absence of structures) up to, and rarely past, 5 days postfertilization (dpf). We have argued previously (Albertson and Yelick, 2004, 2007; Cooper and Albertson, 2008) that a shift from qualitative to quantitative descriptions of phenotype will facilitate a more comprehensive understanding of craniofacial development, particularly as it pertains to development beyond larval stages. The impetus for this assertion stems, in part, from the fact that the adult skeleton is significantly more complex than that of the larvae. Fig. 2A depicts a 5-day larval skeleton, while Fig. 2B shows the adult complex. The most obvious difference between the two structures is that the larval skeleton is minimally ossified. For the most part, only cartilages are present, and thus only a fraction of the elements that comprise the adult complex is evident. In fact, entire functional units are absent from the larval form, including the upper jaws and portions of the cranium (Harris et al., 2008). So while significant progress has been made toward an understanding of the molecular/genetic factors that regulate development of larval pharyngeal cartilage morphology, we know virtually nothing about major functional complexes of the skeleton that appear later in development (i.e., mineralized jaws).

Fig. 2.

Fig. 2

Open in a new tab

Ontogenetic changes in craniofacial bone in AB zebrafish. (A) depicts a 5-day larval skeleton, while (B) shows the adult complex. The most obvious difference between the two structures is that the larval skeleton is only barely ossified as indicated by the predominance of alcian blue staining.

We contend that a major challenge facing developmental biologists is to extend the current developmental genetic paradigm beyond the embryo. This challenge can be framed by asking two basic questions: (1) What are the molecular determinants of craniofacial development beyond embryonic stages? and (2) What is the molecular basis for quantitative shifts in craniofacial shape? The answer to these and related questions will inform a broad range of scientific disciplines, as we seek a better understanding of both normal and abnormal variation in craniofacial form.

II. Obtaining Phenotypes

A. Postembryonic Mutagenesis Screens in Zebrafish

Over the years, the zebrafish has proven to be an excellent subject for forward-genetic mutagenesis screens. By screening for early-lethal phenotypes, a large number of mutants relevant to human development and disease have been discovered, including a variety of craniofacial mutants (Neuhauss et al., 1996; Piotrowski et al., 1996; Schilling et al., 1996). These studies have shed significant light on our understanding of molecular events regulating early craniofacial and tooth development. Our current focus is to move beyond the embryonic stages in order to identify and characterize genes and signaling pathways that regulate the growth and remodeling of adult mineralized tissues. Studies of viable, adult mineralized craniofacial tissue phenotypes are particularly relevant to determining the genetic basis of human craniofacial pathophysiology and disease. Several recent publications have demonstrated the strengths of the zebrafish model for adult craniofacial and skeletal morphogenesis. For example, Fisher et al. (2003) utilized a radiographic analysis of adult ENU-mutagenized F1 fish to identify the dominant skeletal dysplasia chihuahua (chi) mutant. The chi mutant zebrafish harbors a dominant mutation in the type I collagen gene (col1a1) which leads to uneven mineralization, fragility and defects in bone growth (Fisher et al., 2003), features that are associated with human skeletal dysplasia osteogenesis imperfecta (OI). OI is a dominant heritable disorder of connective tissues in humans, caused in approximately 90% of individuals by mutations in type I collagen genes (col1a1 and col1a2), for which there currently is no effective clinical treatment (reviewed in Basel and Steiner, 2009). Thus, the zebrafish chi mutant is a valuable tool and model with which we can extend our knowledge of OI, and hopefully identify effective clinical treatments.

Another large-scale mutagenesis screen for defects in adult structures led to the discovery of two similar mutants finless (fls) and Nackt (Nkt), both of which exhibited defects in the skeletal elements of the skull, fins, scales, and teeth (Harris et al., 2008). It has been shown that Nkt harbors a mutation in the ectodysplasin (eda) gene,and that mutations of the ectodysplasin receptor (edar) gene were found in fls mutants (Harris et al., 2008). Remarkably, mutations of eda and edar genes are responsible for the majority of hypohydrotic ectodermal dysplasia (HED) cases in humans (Mikkola, 2009). HED is a hereditary disorder characterized by hair and teeth defects, as well as defects in a number of other ectodermal organs (Mikkola, 2009).

In the Yelick laboratory, we have identified several mutants that exhibit craniofacial and bone defects in adult fish, by applying high-throughput Alizarin red (AR) staining of ENU-mutagenized F3 generations at 6–8 weeks postfertilization (Stewart-Swift et al., 2010; 9th International Zebrafish Development and Genetics Conference, Madison, WI, June16–20, 2010). One of them, homozygous recessive mutant belkal (38N), exhibits a midface hypoplasia phenotype, and reduced and fused maxilla (Fig. 3A, B). In addition, approximately 50% of sqr mutants develop scoliosis. A second mutant, which seems to be heterozygous semidominant, knjaz’ (78N), displays craniofacial shape defects including the shape of preorbital region possibly due to a caudal displacement of the upper jaw in addition to scoliosis in the tail region, as detected by AR staining, and as confirmed by geometric morphometric (GM) analyses, the methodology for this will be explained in the following (Fig. 3C, D). Efforts to map each of these mutations, and to identify the genetic mutations responsible for these phenotypes, are currently underway.

Fig. 3.

Fig. 3

Open in a new tab

Analysis of bka and knz mutants. (A) AR staining of bka mutant and wild-type control. (B) GM analysis of bka mutant and wild-type control. The arrows in (A) and (B) point to reduced upper jaw in bka mutant. (C) AR staining of knz mutant and wild-type control. (D) GM analysis of knz mutant and wild-type control. The arrows in (C) and (D) point to changes in upper jaw morphology of knz mutant.

B. Heterozygous Craniofacial Phenotypes in Previously Identified Homozygous Recessive Early Lethal Zebrafish Mutants

Another approach to identify adult craniofacial and bone mutants is to use AR staining and GM analyses to look for heterozygous phenotypes previously described early-lethal homozygous recessive zebrafish mutants – particularly those harboring gene mutations in craniofacial and tooth-expressed genes. One example is the acerebellar (ace) mutant, caused by a mutation in the fibroblast growth factor (fgf8) gene. The ace/fgf8 homozyougous mutants exhibit an asymmetric craniofacial pharyngeal skeleton, among other defects (Albertson and Yelick, 2005), and die at approximately 7 dpf (Brand et al., 1996). Our study of heterozygous ace/fgf8 zebrafish allowed investigations of craniofacial development to be extended beyond the early embryonic stages, and identified roles for fgf8 in adult craniofacial form and function (Albertson and Yelick, 2007). Our geometric shape GM analyses revealed that fgf8 haploinsufficency leads to craniofacial asymmetries and defects in cranial sutures, and staining for alkaline phosphatase and tartrate resistant acid phosphatase activities showed aberrant osteoblast and osteoclast activities, respectively, which contributed to abnormal bone formation and remodeling in the lower jaw (Albertson and Yelick, 2007). Thus, these studies demonstrated that subtle but highly informative phenotypes can be revealed by investigations of viable adult heterozygous mutant zebrafish. Analyses of the viable adult heterozygous zebrafish mutants have high clinical relevance, since the majority of human hereditary disorders are induced by mutations in a single copy of the gene. For instance, single allele mutations in fibroblast growth factor receptors (fgfr)types 1–3, have been found in several human syndromes that are characterized by craniofacial defects including craniosynostosis, or premature fusion of cranial sutures, as well as defects in bone formation and growth (Morriss-Kay and Wilkie, 2005).

C. Rescue of Homozygous Recessive Lethal Mutations to Examine Postembryonic Defects

An additional way to extend our knowledge of craniofacial morphogenesis beyond embryonic stages is by rescuing homozygous recessive mutants from early lethality using wild-type mRNA injections. This approach was successfully used by LeClair et al. (2009) to discover the effects of glypican 4 (gpc4) on craniofacial organization in adult zebrafish (LeClair et al. (2009). gpc4 belongs to the family of extracellular glycoproteins that can modulate Wnt, bmp, and fgf signaling (Filmus and Selleck, 2001; Fransson, 2003). The gpc4/knypek homozygous mutant embryos normally die around 5–7 dpf, and exhibit defects in convergence and extension movements in the ectoderm and mesoderm (Topczewski et al., 2001). However, when investigators rescued gpc4/knypek homozygous mutant embryos by single cell injections of gpc4 mRNA, and then allowed the rescued mutants to develop for up to 1 year, they discovered the loss or rearrangement of several adult craniofacial bones, defects in craniofacial cartilages, and changes in the shape of neurocranium (LeClair et al., 2009). These elegant studies have shown that adult craniofacial phenotypes can be identified and examined in previously identified homozygous recessive lethal mutants, if they are allowed to survive beyond critical early developmental stages.

III. Quantitative Methods for Studying Adult Phenotypes

A. Geometric Morphometrics

Shape is a fundamentally important feature of organisms but has historically been very challenging for biologists to quantify. Until recently morphometric analyses have largely involved measuring sets of linear distances on a form, applying a chosen method of size correction to these distances, and finally applying a multivariate statistical test to these data to discern patterns. While these methods are still used, they are fraught with disagreements over what size corrections to apply, and difficulties in determining the biological meaning of statistical results (Parsons et al., 2003).

The field of morphometrics is undergoing a radical transformation through the introduction and expansion of GM (Adams et al., 2004; Rohlf and Marcus, 1993). These powerful techniques have now become the standard method to quantify the shape and offer researchers a greater ability to discern and interpret trends in shape variation.

Simply put, we contend that morphometrics is a quantitative approach to compare shape differences that zebrafish biologists have been describing qualitatively for years. Moreover, these newer methods are powerful enough that they can extract meaningful information from situations where only subtle or continuous variation occurs, even when these patterns are not apparent to the naked eye.

Although the use of GM in zebrafish research is still in its infancy, there is enormous potential for biologists to adopt these methods to study development beyond the embryo (Albertson et al., 2005; Albertson and Yelick, 2007; Cooper and Albertson, 2008; LeClair et al., 2009). We recognize that GM can seem mathematically complex, abstract, or even obscure, and that this perception could easily lead to discouragement. Accordingly, we have written the following sections as a primer for zebrafish biologists, with largely heuristic explanations of the mathematics and statistics being applied. We will also illustrate how data from GM can be collected and analyzed in a developmental context using samples from the commonly available AB and TL lines of zebrafish. Specifically, we will describe techniques that are useful for assessing developmental robustness (i.e., canalization), the trajectory of morphological development (i.e., allometric repatterning), the rate of shape change over development (i.e., heterochrony), and morphological integration.

B. Data Collection

The beginning of any GM analysis involves the selection and photographic imaging of samples, and the collection of x,y Cartesian landmark coordinates. Samples can be chosen based on their membership to a particular group of interest (i.e., wild-type vs. mutant zebrafish), and at multiple stages of ontogeny. Imaging is a relatively straightforward but extremely important process whereby samples are photographed in a reproducibly standard position in the presence of a scale bar. These images are then used as a source from which x, y landmark data are collected. Currently the most popular software for x, y data collection is TPSdig2 (Rohlf, 2010), which is used in coordination with TPSutil (Rohlf, 2010). TPSutil creates a file that allows images to be read sequentially into TPSdig2, using the “build tps files from images” function so that the coordinate locations of homologous landmarks can be recorded on images (available at http://life.bio.sunysb.edu/morph/). It is extremely important that x, y coordinates are collected in the same sequence at homologous points on each individual so that they are recognized correctly. Ultimately, these data will be used to statistically test for shape differences between groups or ontogenetic stages.

The quality of any statistical test depends upon the quality and amount of data collected. We therefore recommend the use of multiple individuals per group, and the collection of as many landmarks as are reasonable for any GM study. Here we use a total of 146 specimens and seven landmarks across three stages of ontogeny (10 dpf n = 24 AB, 20 TL; 30 dpf n = 30 AB, 26 TL; and adult n = 25 AB, 21TL) (Fig. 4).

Fig. 4.

Fig. 4

Open in a new tab

Three stages of ontogeny in AB and TL lines of zebrafish analyzed for morphometric variation. Note that the seven landmarks used in this study are in homologous positions throughout ontogeny.

C. General Data Analysis

A first step toward analyzing x, y coordinate data is to eliminate variance that is due to differences in the size of specimens, their orientation, and position. Several superimposition algorithms have been developed but the most common approach is a generalized procrustes analysis (GPA), which is available seperately or automatically implemented in a number of morphometric programs (Zelditch et al., 2004). This method superimposes landmark configurations through a process of translation, rotation, and scaling for size (see Fig. 5), in order to minimize the sum of squared distances between corresponding landmarks across specimens (Adams et al., 2004). Once performed, this analysis opens the gateway for a variety of further statistical methods. However, it is important that researchers comparing multiple groups use a single GPA across all specimens, as the comparison of data from separate GPAs can create misleading results.

Fig. 5.

Fig. 5

Open in a new tab

Processes carried out during a generalized Procrustes analysis.

D. Analysis of Shape

A landmark-based approach to shape analysis is far more powerful than traditional methods used to discern and describe patterns of shape variation (Parsons et al., 2003). In the simplest terms a geometric approach can double the number of traits examined as compared to traditional linear methods. For example, consider three distances on the head of a zebrafish: the depth of the eye, the length of the lower jaw, and the overall length of the skull. These measurements represent three traits, but if one were simply to place landmarks at the ends of each linear distances, the number of traits would double to six. Moreover, in a landmark-based analysis variation at each landmark is evaluated relative to every other landmark, leading to a more comprehensive picture of how the geometry of a given shape varies between experimental manipulations or over time.

After GPA, GM shape analysis typically involves performing a thin-plate spline (TPS) analysis. The analogy that is typically used in the field is that TPS analysis models a starting form (e.g., that of a wild-type zebrafish) as an infinitely thin metal plate that is constrained at some combination of points (i.e., landmarks), but is otherwise free to adopt a target form (e.g., that of a mutant zebrafish) in a way that minimizes bending energy (Bookstein, 1991). Bending energy can be thought of as the degree of deformation that is required to bend the Cartesian coordinate (x, y) system such that two landmark configurations can be superimposed on top of one another (Bookstein, 1991). The total deformation of the TPS can be decomposed into geometrically orthogonal components based on scale/bending energy such that less energy is required to bend a metal plate while holding it at distantly positioned points, whereas much more energy is required to bend the plate to the same vertical extend while holding two points in close proximity (Rohlf and Marcus, 1993). These components (partial warps) can be localized to describe precisely what aspects of shape are different. Because the number of partial warps can be large, as they scale with the number of landmarks, a data reduction analysis is typically performed, principal component analysis (PCA) being the most common. Performing a PCA on partial warp scores is formally referred to as relative warp analysis (Zelditch et al., 2004), and can be an effective method to identify the major axes of shape variation.

By applying these methods, we are able to show that adult AB and TL strains of zebrafish significantly differ in craniofacial geometry along PC1 (one-way ANOVA: F-ratio = 35.32, p < 0.001). Differences are largely restricted to the anterior head, and show that TL zebrafish on average have longer upper (yellow region) and lower (purple) jaws than AB wild-type fish, whereas the retroarticular process is longer in ABs than it is in TLs (Fig. 6). TL fish harbor a mutation that results in the expression of their characteristic long tails. It is interesting that, while a craniofacial phenotype has never been noted for this stain, TL fish (at least in our lab) also possess elongated dermal bones in the jaw. These results demonstrate the power of GM to discriminate subtle but discrete differences in craniofacial morphology.

Fig. 6.

Fig. 6

Open in a new tab

Results of a relative warps analysis on craniofacial shape in adult AB and TL lines. Morphological variation between lines along RW1 (the horizontal axis) and RW 2 (the vertical axis). The scatter plot suggests that RW1, the major axis of variation, distinguishes AB and TL lines. Deformation grids along this axis show that AB fish have a relatively smaller maxillae (yellow) and shorter mandible (purple). However, the length of the retroarticular process is larger in ABs. Note that this analysis utilizes 15 landmarks in contrast to the seven used in our ontogenetic analysis because more homologous regions were apparent at adult stages. However, despite this reduction in the amount of shape variation collected AB and TL lines are still distinguishable with seven landmarks. (See Plate no. 12 in the Color Plate Section.)

E. Analysis of Shape Variation

Zebrafish biologists may also be interested in the effects elicited by an experimental treatment or mutation with respect to shape variation. Magnitudes of shape variation may also change over the course of ontogeny (Zelditch et al., 2004). With GM it is possible to quantify the degree to which a sample varies and statistically test for differences in the magnitude of variation among groups. While many biologists are familiar with how variance is calculated for univariate traits, shape is a more complex multivariate trait that is not compatible with these more traditional methods. Fortunately, a metric for calculating variance in multivariate traits exists; instead of “variance” the term “disparity” is used. Disparity is calculated using the approach of Foote (1993) as follows:

D=∑di2/N−1

Here di represents the Procrustes distance of the centroid of individual i from the centroid of all N individuals. Procrustes distance is normally calculated by comparing an individual specimen to the average shape of a group (which is called the consensus) to provide a standard metric of shape difference. Here we test for differences in measures of disparity across ontogenetic stages between our AB and TL lines.

To test for differences in disparity, 1000 permutations of the difference in the disparities between stages within AB and TL lines, as well as between these lines at the same stage were performed. For the permutation tests, residuals of the Procrustes fit were randomly assigned within each treatment group, and then recombined with the mean form of that group to calculate the disparity of the permuted group. The observed results were then compared with the range of values obtained via permutation to determine if the observed values were significant. These tests were performed using the program DisparityBox6 (available at www3.canisius.edu/~sheets/morphsoft.html).

Our analyses show that overall disparity decreases over time in both lines of zebrafish. However, in ABs this decrease was statistically significant between 10 and 30 dpf, whereas in TLs there was a significant reduction in disparity that occurred between the 30 dpf and adult stages (Table I). This suggests that although reductions in shape disparity take place in both lines, these changes are initiated at different times between lines. Comparisons between lines over ontogeny show that initially the AB line is approximately twice as disparate than the TL line after 10 dpf (p < 0.001, Table I), but no significant difference in disparity was observed at 30 dpf or in adults (all p > 0.16). In fact, levels of disparity are nearly equal at adult stages suggesting that an equally canalized phenotype has been achieved in both lines by adulthood (Table I, Fig. 7).

Table I.

Changes in craniofacial shape disparity over ontogeny in AB and TL lines

Disparity Ontogenetic comparisons p-values from 1000 permutations
Line 10 dpf 30 dpf Adult 10 to 30 dpf 30 dpf to adult
AB 0.011 0.005 0.004 < 0.0001 0.108
TL 0.005 0.008 0.004 0.388 0.047
Open in a new tab

Fig. 7.

Fig. 7

Open in a new tab

Plotted GPA superimposed landmarks for each measured stage (10 dpf, 30 dpf, and adult stages) of craniofacial development in AB and TL lines of zebrafish. Note that the scatter of landmarks tends to decrease over ontogeny, happening earlier in the AB line during the period between 10 and 30 dpf, while for TLs a reduction in scatter does not occur until the period between 30 dpf and adult stages.

F. Rates of Shape Change

Heterochrony has most often been studied in the context of evolutionary biology and is usually defined as a change in the developmental rate or timing of the appearance of features that create parallels between ontogeny and phylogeny (Gould, 1977). More general definitions also exist that do not consider this parallelism, and only require that a developmental event occurs at a different time or rate in a descendant relative to an ancestor (McKinney and McNamara, 1991; McNamara, 1995, 1997). In other words, heterochrony causes a decoupling of shape from developmental time and may also be a theory of interest to developmental biologists comparing wild-type (ancestor) to mutant (descendant) phenotypes. Since any developmental process is temporal in nature, thereby making it susceptible to heterochronic modifications, it is likely that heterochrony is a widespread phenomenon for zebrafish mutants relative to their respective wild-type ancestors.

The rate of morphological development over ontogeny can be determined empirically in the absence of trajectories, but we recommend that researchers use the approaches for quantifying trajectories discussed below in a complementary fashion to gain a comprehensive idea of morphological development. Determination of the ontogenetic rate of morphological development is based on the morphometric distance between shapes at a given ontogenetic stage in relation to the average specimen (of a treatment group at one stage of the experiment), and this is calculated as the partial Procrustes distance (Bookstein, 1996; Dryden and Mardia, 1998; Zelditch et al., 2004). We tested whether rates of morphological development differed over ontogeny among AB and TL zebrafish lines (Table II). Ontogenetic tests for differences in the amount of shape change in morphometric distance between 10 and 30 dpf, 30 dpf and adult, and 10 dpf and adult specimens were performed between lines using 900 bootstrap replicates while stage specific tests between lines were conducted by performing Goodall’s F-test in the program TwoGroup6h (available at http://www3.canisius.edu/~sheets/morphsoft.html). Goodall’s F-test compares the difference in the mean shape between two samples relative to the shape variation found within the samples.

Table II.

The rate of morphological development as measured by Partial Procrustes distancesa

Partial Procrustes distances, F, and p-value from comparisons between ontogenetic stages
Line 10 to 30 dpf 30 dpf to adult 10 dpf to adult
AB 0.125, F = 28.24, p = 0.001 0.129, F = 53.21, p = 0.001 0.218, F = 82.44, p = 0.001
TL 0.138, F = 31.08, p = 0.001 0.140, F = 37.10, p = 0.001 0.239, F = 132.11, p = 0.001
Open in a new tab
a

The Fand p-values from a test performed between developmental stages within lines is presented. Note that partial Procrustes distances are similar between AB and TL lines and did not differ significantly.

Our analysis showed that rates of morphological development did not differ between our zebrafish lines at any of the ontogenetic intervals tested. However, stage specific tests did reveal that significant differences in the position of AB and TL lines in shape space (distance) were present at all three stages of the experiment, indicating that lines did differ in shape at each stage. Taken together these data suggest that while rates of shape change do not differ between lines at any stage, both lines are characterized by marked shifts in rates of morphological development over time, which may be a common feature of zebrafish craniofacial development.

G. Trajectories of Shape in Zebrafish Lines

Mutations may also modify the direction of ontogenetic shape change in zebrafish. This aspect of morphological development is distinct from changes in shape due to alterations in the rate or timing of development (i.e., heterochrony). Whereas heterochrony refers to changes in the timing of developmental events, allometric repatterning refers to changes in the trajectory of development. Allometric repatterning can gradually develop over long intervals of ontogeny whereas heterochrony may be sudden. Since both heterochrony and allometric repatterning can affect shape, it is useful to consider both phenomena in relation to each other.

We can quantitatively test whether allometric repatterning has occurred between AB and TL zebrafish lines by measuring their shape trajectories in relation to a continuous variable (ln centroid size in this case). We accomplish this here by measuring ontogenetic growth angles, which are the arc cosines of vector correlations (Zelditch et al., 2000), between AB and TL zebrafish at 10 dpf, 30 dpf, and adult stages. Similar to the analysis of disparity described above the calculation of angle is simply a metric and there is no true statistical model. We therefore compared the observed angle between ontogenetic vectors to bootstrapped estimates of these angles. The null hypothesis was an angle of zero, which is biologically equivalent to the hypothesis of a conserved ontogenetic trajectory of shape between AB and TL lines. To test this null hypothesis, 95% confidence intervals for the angles were created through a procedure involving 900 bootstrapped replicates. The angle between the vectors would be considered statistically significant if they exceeded those drawn from a bootstrapped distribution within a group. In other words the differences between groups of interest would have to exceed those generated from bootstrapping within both groups. This explains how smaller overall differences in trajectories may sometimes be significant, whereas larger differences are not. Growth angles and bootstrap estimates of 95% confidence intervals were calculated in VecCompar6c (available at http://www3.canisius.edu/~sheets/morphsoft.html).

Our analysis shows that ontogenetic trajectory differs between lines at the 10 and 30 dpf stages (110.1° and 94.7° respectively). Trajectories also differ between stages within lines. This is observed for each measured stage within our AB line, and for all but one comparison in out TL line (Table III). However, the trajectory for TLs at 10 dpf is only marginally different from the one present in adult TLs. This suggests that ontogenetic trajectories are relatively dynamic in ABs, while in TLs the trajectory is set early in development and persists with little change to adulthood. Considered in the context of data reported above, these results are consistent with the hypothesis that the unique TL phenotype is due, in part, to an altered growth trajectory established early in development.

Table III.

The angles between ontogenetic trajectories of craniofacial shape in AB and TL linesa

Angle between ontogenetic stages
Line 10 to 30 dpf 30 dpf to adult 10 dpf to adult
AB 84.1°* 127.0°* 101.5°*
TL 140.5°* 76.5° 101.5°
Open in a new tab
a

Trajectories are measured between all stages including 10 dpf, 30 dpf, and adults.

H. Morphological Integration

Morphological integration refers to the degree to which functionally related traits are correlated (Cheverud et al., 1996; Olson and Miller, 1958). Implicit to this theory is the concept of modularity. A module refers to a complex of traits that are inherited together, and which are independent of other character complexes (Wagner, 1996). Developmental architecture figures prominently in questions of integration and modularity, because genetic and functional modules are mediated by developmental processes (Atchley and Hall, 1991; Cheverud, 1996; Klingenberg, 2004; Klingenberg, 2003). For example, the integration of component parts may result from any combination of the following phenomena: common function, shared developmental pathways, pleiotropy, or heritable epigenetic effects.

One way to quantitatively assess general levels of morphological integration is through the combined use of GM and PCA. High levels of positional covariation among landmarks will skew the distribution of shape variation among the PC axes such that most of the variation will be explained by only a few axes (Wagner, 1984, 1990; Young, 2006). This would be reflected by the existence of one or a few axes that explains a high amount of variation followed by a “distinct” drop in explanatory power in subsequent PCs. Several methods exist for determining whether differences in the degree of integration exist between groups of interest. For example, a Chi-squared test can be used to determine if and when there was a strong drop in the levels of variation explained by subsequent PC axes. This method determines whether the total shape variation in a dataset is spread out among many PC axes (a low level of integration) or concentrated within a small number of the initial PC axes (a higher level of integration). This method is incorporated into the program PCAgen (available at http://www3.canisius.edu/~sheets/morphsoft.html). For a detailed explanation of this method, see p. 211–254 in Morrison (2004).

Alternatively, it is possible to simply compare the variance of eigenvalues (the explanatory power of each PC) produced from a PCA on each group by using an F-test. This should be available in most statistical packages, and it is even manageable to perform this test by hand. An increased level of variation in eigenvalues is indicative of increased integration.

Our analysis of integration revealed differences both between AB and TL lines, and between ontogenetic stages within lines. At 30 dpf the TL line showed a significantly higher degree of integration than ABs (variance = 1.32−006 vs. 2.45−007, respectively, F-ratio9,9 = 3.21, p = 0.02). The comparisons between 10 and 30 dpf in ABs (variance = 2.45−006 to 2.41−007 respectively, F-ratio9,9 = 0.0982, p-value = 0.002) showed a reduction in integration had occurred. This period of ontogeny in ABs likely accounts for the majority of change in levels of integration as differences were not detected between 30 dpf and adult stages, but a similar degree of difference was present between 10 dpf and adult stages (variance = 2.45−006 to 1.94−007, respectively, F-ratio9,9 = 0.079, p < 0.001). For TLs a difference in integration did not appear until 30 dpf and adult stages were compared (variance = 1.32−006 vs. 1.64−007, respectively, F-ratio9,9, p = 0.005), and this period of ontogeny is also likely to be responsible for the differences observed between 10 dpf and adult stages (7.62−007 to 1.64−007, respectively, F-ratio9,9 = 0.215, p = 0.032). Thus, within both lines we observed a decrease in integration over ontogeny, but this drop occurred earlier and more precipitously in ABs, leading to the observed difference in integration between lines at 30 dpf. Taken together these analyses suggest that integration is an ontogenetically dynamic property of phenotypes. It is also noteworthy that the timing of changes in integration seems to coincide with the changes in growth trajectory that were observed earlier. In this case it may be that patterns of integration are a surrogate for trajectories in that interactions among anatomical features also determine the spatial direction of development.

I. Synthesis of Quantitative Data

We have demonstrated here that GM can be used to identify differences in craniofacial shape. More importantly we have shown that with the proper experimental design GM can be used to quantify the dynamic processes involved in development and provide an explicit quantitative link between development within and beyond the embryo. We feel this provides a major opportunity to understand when and how key changes occur to produce clinically and evolutionary relevant variation. For clinicians the approaches demonstrated here may be especially useful for the identification of progressive syndromes at very early stages when therapeutic approaches may have their greatest impact. They also enable the identification of life stages when mutational effects are most pronounced. For evolutionary biologists, these techniques may lead to the identification of adaptively important ontogenetic processes. Opposite to developmental biology, the traditional paradigm in evolutionary biology has been to study adult phenotypes. We suggest that the use of GM can be a means to link embryonic and adult life history stages, and is thus equally valuable to both fields of study (see Fig. 8 for an outline of our approach). Indeed, this approach has recently gained traction in evolutionary biology, and such studies have fallen under the title of “evo-devo” (Carroll et al., 2005; Gilbert and Epel, 2009). Critical to both of these biological disciplines is a consideration of developmental processes that occur after embryonic but before adult stages, and GM can provide an empirical avenue for this view.

Fig. 8.

Fig. 8

Open in a new tab

An outline of our approach that incorporates the use of GM and quantitative methods to study development. GM offers an ability to quantify and assess variation at an intermediate stage of discovery. This means that GM can provide a way of testing hypotheses about potential mutant phenotypes and their development, as well as providing data for further explorations into the genetic basis of mutant phenotypes.

IV. A Complementary Approach: The Use of Natural Variation to Complement that Generated in the Lab for Understanding Jaw Morphogenesis

While the use of mutagenic analyses in a select group of model organisms, including the zebrafish, has provided a powerful approach for understanding the genetic basis of traits that emerge very early in development, they have been less effective for understanding complex traits that manifest themselves later in development. This may be due to an ascertainment bias introduced by researchers focusing on embryonic stages, but there are likely other complications. The most confounding effects of induced mutations concern pleiotropy. Most mutations isolated in mutant screens occur in the coding regions of genes and lead to the severe attenuation or complete knockdown of gene function. As a result genetic screens typically recover mutant animals with defects at early developmental stages when the gene first provides an essential function, generally in early embryonic development. Such mutations often result in early lethality, which in turn masks the ascertainment of their effects on adult phenotypes. Thus, phenotype-driven mutagenesis screens often preclude analyses at later developmental stages, and while we have discussed different methods for mitigating this limitation in zebrafish, in the following section we will review another, complementary approach that can be used to identify the genetic mechanisms that contribute to craniofacial development – the analysis of variation among natural populations. The details of this approach have been elaborated on elsewhere (Albertson et al., 2009), but a brief overview is worth repeating here.

Closely related cichlid fishes from lakes in the East African Rift Valley have undergone extensive evolutionary modifications of their oral jaws and faces, providing an array of “evolutionary mutant” models for medically important human craniofacial variation that may not always be present in model organisms like the zebrafish (e.g., micrognathia, midface hypoplasia, facial asymmetries). The utility of integrating models organisms with emerging models was illustrated by Albertson et al. (2005), where a role for bmp4 in directing the development of mandibular shape was assessed through a combination of quantitative trait loci analysis and experimental embryology. Both the expression of and allelic variation in bmp4 was shown to be associated with quantitative differences in the shape of the lower jaws of cichlid fishes. However, because cichlids were less amenable to genetic manipulation zebrafish were used as a surrogate to investigate how altering levels of bmp4 would affect the development of the craniofacial shape. Manipulations of bmp4 levels in zebrafish lead to variable effects on the jaw, but this variation mirrored the natural, quantitative variation in jaw shape seen among cichlid species. These data suggested a novel role for levels of bmp4 in modulating the jaw shape, and underscored the utility of integrating work in both model and nonmodel organisms to paint a more comprehensive picture of how morphology develops and evolves. The results of recent research in Darwin’s finches bears striking similarity to that in cichlids. Here, investigators identified roles for bmp4 and calm1 in regulating differences in beak width/height and beak length, respectively, which were confirmed independently using genetic manipulations in chicken and duck models (Abzhanov et al., 2004, 2006; Wu et al., 2004, 2006). These studies underscore the utility of integrating work in the laboratory with studies of evolutionary model systems, as these approaches have highly complementary attributes; laboratory models bring experimental tractability to the table, whereas evolutionary models provide extensive amounts of novel phenotypic variation to bear on questions related to the morphogenesis of the craniofacial shape. Fortunately, whereas postembryonic screens are hard to come by, for researchers interested in understanding craniofacial shape there is no shortage of examples where adaptive divergence has created a plethora of variation in adult head and jaw characteristics.

Space limitations preclude a thorough review of other natural systems that embody both the experimental and phenotypic potential to advance our understanding of craniofacial development, but these would most certainly include the blind cavefish (Astyanax mexicanus), which has been used to study roles for SHH signaling in jaw development (Yamamoto et al., 2009), the Antarctic notothenioid species flock used to study craniofacial patterning and skeletogenesis (Albertson et al., 2009, 2010), as well as the threespine stickleback (Gasterosteus aculeatus) to analyze an array of mineralized tissue phenotypes (Cresko et al., 2004; Colosimo et al., 2005; Kimmel et al., 2005; Shapiro et al., 2004; 2006). With technological advances in molecular biology and genomics, the once distinct line that separated model from nonmodel organisms is becoming increasingly blurred (Albertson et al., 2009). While the tools and resources that are available to zebrafish researchers far exceed those that can be applied in nontraditional fish systems, we contend that a niche also exists for these nonmodel systems in investigations of the morphogenesis of craniofacial shape.

V. Implications and Conclusions

The zebrafish has become an invaluable model system that has provided detailed knowledge of the molecules that regulate early embryonic development. However, due to the early lethal phenotypes exhibited by most zebrafish mutants, significantly less is known about craniofacial development beyond embryonic stages. There is a significant gap in our understanding of later developmental events including tooth replacement, as well as bone development, growth, and remodeling. We argue that an expansion of the current paradigm is needed, and that this will include (1) postembryonic mutagenesis screens, (2) the application of quantitative shape analyses to evaluate and statistically compare patterns of phenotypic variation among wild-type and mutant lines over extended periods of development, and (3) comparative analyses in evolutionary models that exhibit relevant patterns of phenotypic variation (Albertson and Yelick, 2009).

From a biomedical perspective, the study of the zebrafish beyond the embryological stages may provide insights into the developmental and genetic basis of progressive craniofacial syndromes. Over 70% of all birth defects are associated with craniofacial malformations (Hall, 1999), and while surgery can correct many of these (e.g., orofacial clefting), syndromes associated with oral, dental, and craniofacial defects continue to pose major therapeutic challenges for clinicians (NIDCR, 2007). The NIDRC initiative, FaceBase, states that “much research is needed to achieve a molecular and cellular understanding of the mechanisms by which genes and gene products interact to generate complex phenotypes,” and argues for broadening support for projects “that merge genetics, developmental biology, and modeling expertise and are aimed at achieving a systems biology comprehensive understanding of the mechanisms that underlie complex craniofacial phenotypes.” Many of the statistical frameworks presented here could be folded into genetic and developmental studies to contribute to this initiative. Indeed, by modeling the craniofacial skeleton as a dynamic multivariate trait and analyzing accordingly, we will gain a more accurate understanding of phenotype. Since mutational analysis are, by definition, based on phenotype, these insights will enable a far more comprehensive understanding of the number, type, mode of action and interaction of genes that underlie craniofacial form at all stages of development.

From an evolutionary perspective, the alteration of developmental processes in craniofacial traits over ontogeny often plays a key role in producing adaptive diversity (Carroll et al., 2005). Adaptive phenotypes in the wild are most often studied in adult animals, and while there is mounting evidence to suggest that adaptive variation studied in adults has an early embryonic origin (e.g., Abzhanov et al., 2004, 2006; Albertson et al., 2005), there are undoubtedly myriad adaptive phenotypes that also arise postembryonically (West-Eberhard, 2003). In the sections above, we have presented ways in which natural variation in the wild can compliment that produced by induced mutations, but this may well be a two-way street. Which prompts us to ask, can craniofacial variation induced via a chemical mutagen inform an understanding of the molecular nature of evolutionary change? Since many of the most spectacular adaptive radiations among vertebrates are associated with extensive modifications of the craniofacial skeleton, and given that a proximate genetic basis for natural variation in craniofacial form remains unknown, this question is well worth thinking about.

References

  1. Abzhanov A, Protas M, Grant BR, Grant PR, and Tabin CJ (2004). Bmp4 and morphological variation of beaks in Darwin’s finches. Science 305, 1462–1465. [DOI] [PubMed] [Google Scholar]
  2. Abzhanov A, Kuo WP, Hartmann C, Grant BR, Grant PR, Tabin CJ (2006). The calmodulin pathway and evolution of elongated beak morphology in Darwin’s finches. Nature 442, 563–567. [DOI] [PubMed] [Google Scholar]
  3. Adams DC, Rohlf FJ, and Slice DE (2004). Geometric morphometrics: ten years of progress following the ‘revolution’. Ital. J. Zool 71, 5–16. [Google Scholar]
  4. Albertson RC, and Yelick PC (2004). Morphogenesis of the jaw: development beyond the embryo. Methods Cell Biol. 76, 437–454. [DOI] [PubMed] [Google Scholar]
  5. Albertson RC, and Yelick PC (2007). Fgf8 haploinsufficiency results in distinct craniofacial defects in adult zebrafish. Dev. Biol 306, 505–515. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Albertson RC, and Yelick PC (2005). Roles for fgf8 signaling in left-right patterning of the visceral organs and craniofacial skeleton. Dev. Biol 283, 310–321. [DOI] [PubMed] [Google Scholar]
  7. Albertson RC, Streelman JT, Kocher TD, and Yelick PC (2005). Integration and evolution of the cichlid mandible: Molecular basis of alternate feeding strategies. Proc. Natl. Acad. Sci. U.S.A 102, 16287–16292. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Albertson RC, Cresko W, Detrich HW, and Postlethwait JH (2009). Evolutionary mutant models for human disease. Trends Gen. 25, 74–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Albertson RC, and Yelick PC (2009). Morphogenesis of the jaw: development beyond the embryo. Methods Cell Biol. 91, 453–473. [DOI] [PubMed] [Google Scholar]
  10. Albertson RC, Yan Y-L, Titus TA, Pisano E, Vacchi M, Yelick PC, Detrich WH, Postlethwait JH (2010). Molecular pedomorphism underlies craniofacial skeletal evolution in Antarctic notothenioid fishes. BMC Evol. Biol 10, 4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Atchley WR, and Hall BK (1991). A model for the development and evolution of complex morphological structures. Biol. Rev 66, 101–157. [DOI] [PubMed] [Google Scholar]
  12. Basel D, and Steiner RD (2009). Osteogenesis imperfecta: recent findings shed new light on this once well-understood condition. Genet. Med 11, 375–385. [DOI] [PubMed] [Google Scholar]
  13. Bookstein FL (1991). Morphometric Tools for Landmark Data: Geometry and Biology. Cambridge University Press, Cambridge. [Google Scholar]
  14. Bookstein FL (1996). Biometrics, biomathematics and the morphometric synthesis. Bull. Math. Biol 58, 313–365. [DOI] [PubMed] [Google Scholar]
  15. Brand M, Heisenberg C-P, Jiang Y-J, Beuchle D, Lun K, van Eeden FJM, Furutani-Seiki M, Granato M, Haffter P, Hammerschmidt M, Kane DA, Kelsh RN, Mullins MC, Odenthal J, Nüsslein-Volhard C (1996). Mutations in zebrafish genes affecting the formation of the boundary between midbrain and hindbrain. Development 123, 179–190. [DOI] [PubMed] [Google Scholar]
  16. Carroll SB, Grenier JK, and Weatherbee SD (2005). From DNA to diversity: molecular genetics and the evolution of animal design, 2nd edn. Blackwell, Oxford. [Google Scholar]
  17. Cheverud JM (1996). Developmental integration and the evolution of pleiotropy. Am. Zool 36, 44–50. [Google Scholar]
  18. Cooper WJ, and Albertson RC (2008). Quantification and variation in experimental studies of morphogenesis. Dev. Biol 321, 295–302. [DOI] [PubMed] [Google Scholar]
  19. Cresko WA, Amores A, Wilson C, Murphy J, and Currey M, et al. (2004). Parallel genetic basis for repeated evolution of armor loss in Alaskan threespine stickleback populations. Proc. Natl. Acad. Sci. U.S.A 101, 6050–6055. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Dryden IL, and Mardia KV (1998). Statistical shape analysis. Wiley, Chichester. [Google Scholar]
  21. Filmus J, and Selleck SB (2001). Glypicans: proteoglycans with a surprise. J. Clin. Invest 108, 497–501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Fisher S, Jagadeeswaran P, and Halpern ME (2003). Radiographic analysis of zebrafish skeletal defects. Dev. Biol 264, 64–76. [DOI] [PubMed] [Google Scholar]
  23. Foote M (1993). Contributions of individual taxa to overall morphological disparity. Paleobiology 19, 40n3–419. [Google Scholar]
  24. Fransson LA (2003). Glypicans. Int. J. Biochem. Cell. Biol 35, 125–129. [DOI] [PubMed] [Google Scholar]
  25. Fuiman LA (1983). Growth gradients in fish larvae. J. Fish Biol 23, 117–123. [Google Scholar]
  26. Gilbert SF, and Epel D (2009). Ecological Developmental Biology: Integrating Epigenetics, Medicine, and Evolution. Sinauer Associates, Sunderland, MA. [Google Scholar]
  27. Gould SJ (1977). Ontogeny and Phylogeny. Belknap Press, Cambridge. [Google Scholar]
  28. Hall BK (1999). The Neural Crest in Development and Evolution. Springer-Verlag, New York. [Google Scholar]
  29. Harris MP, Rohner N, Schwarz H, Perathoner S, Konstantinidis P, Nüsslein-Volhard C (2008). Zebrafish eda and edar mutants reveal conserved and ancestral roles of ectodysplasin signaling in vertebrates. PLoS Genet. 4, e1000206. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Kimmel CB, Ullmann B, Walker C, Wilson C, and Currey M, et al. (2005). Evolution and development of facial bone morphology in threespine sticklebacks. Proc. Natl. Acad. Sci. U.S.A 102, 5791–5796. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Klingenberg CP (2004). Integration, modules and development: molecules to morphology to evolution. In “Phenotypic Integration: Studying the Ecology and Evolution of Complex Phenotypes,” (Pigliucci M, and Preston K, eds.), pp. 213–230. Oxford University Press, New York. [Google Scholar]
  32. Klingenberg CP (2003). Developmental instability as a research tool: using patterns of fluctuating asymmetry to infer the developmental origins of morphological integration. In, “Developmental instability: causes and consequences.” Oxford University Press, New York. [Google Scholar]
  33. LeClair EE, Mui SR, Huang A, Topczewska JM, and Topczewski J (2009). Craniofacial skeletal defects of adult zebrafish Glypican 4 (knypek) mutants. Dev. Dyn 238, 2550–2563. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Loy A, Bertelletti M, Costa C, Ferlin L, and Cataudella S (2001). Shape changes and growth trajectories in the early stages of three species of the genus Diplodus (Perciformes Sparidae). J. Morph 250, 24–33. [DOI] [PubMed] [Google Scholar]
  35. McKinney ML, and McNamara KJ (1991). Heterochrony: the evolution of ontogeny. Plenum, New York. [Google Scholar]
  36. McNamara KJ (1995). Evolutionary change and heterochrony. Wiley, Chichester, UK. [Google Scholar]
  37. McNamara KJ (1997). Shapes of time: the evolution of growth and development. Johns Hopkins University Press, Baltimore. [Google Scholar]
  38. Mikkola ML (2009). Molecular aspects of hypohidrotic ectodermal dysplasia. Am. J. Med. Genet. A 149, 2031–2036. [DOI] [PubMed] [Google Scholar]
  39. Morrison DF (2004). Multivariate Statistical Methods. Duxbury Press, Pacific Grove, CA, p. 480. [Google Scholar]
  40. Morriss-Kay GM, and Wilkie A (2005). Growth of the normal skull vault and its alteration in craniosynostosis: insights from human genetics and experimental studies. J. Anat 207, 637–653. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Neuhauss SC, Solnica-Krezel L, Schier AF, Zwartkruis F, Stemple DL, Malicki J, Abdelilah S, Stainier DY, Driever W (1996). Mutations affecting craniofacial development in zebrafish. Development 123, 357–367. [DOI] [PubMed] [Google Scholar]
  42. Olson EC, and Miller RL (1958). Morphological Integration. University of Chicago Press, Chicago. [Google Scholar]
  43. Osse JWM (1990). Form changes in fish larvae in relation to changing demands of function. Neth. J. Zool 40, 362–385. [Google Scholar]
  44. Parsons KJ, Robinson BW, and Hrbek T (2003). Getting into shape: an empirical comparison of traditional truss-based morphometric methods with a newer geometric method applied to New World cichlids. Environ. Biol. Fishes 67, 417–431. [Google Scholar]
  45. Piotrowski T, Schilling TF, Brand M, Jiang YJ, Heisenberg CP, Beuchle D, Grandel H, van Eeden FJ, Furutani-Seiki M, Granato M, Haffter P, Hammerschmidt M, Kane DA, Kelsh RN, Mullins MC, Odenthal J, Warga RM, Nüsslein-Volhard C (1996). Jaw and branchial arch mutants in zebrafish II: anterior arches and cartilage differentiation. Development 123, 345–356. [DOI] [PubMed] [Google Scholar]
  46. Rohlf FJ, and Marcus LF (1993). A revolution in morphometrics. Trends Ecol. Evol 8, 129–132. [DOI] [PubMed] [Google Scholar]
  47. Schilling TF, Piotrowski T, Grandel H, Brand M, Heisenberg CP, Jiang YJ, Beuchle D, Hammerschmidt M, Kane DA, Mullins MC, van Eeden FJ, Kelsh RN, Furutani-Seiki M, Granato M, Haffter P, Odenthal J, Warga RM, Trowe T, Nüsslein-Volhard C (1996). Jaw and branchial arch mutants in zebrafish I: branchial arches. Development 123, 329–344. [DOI] [PubMed] [Google Scholar]
  48. Shapiro MD, Marks ME, Peichel CL, Blackman BK, and Nereng KS, et al. (2004). Genetic and developmental basis of evolutionary pelvic reduction in threespine sticklebacks. Nature 428, 717–723. [DOI] [PubMed] [Google Scholar]
  49. Shapiro MD, Bell MA, and Kingsley DM. (2006). Parallel genetic origins of pelvic reduction in vertebrates. Proc. Natl. Acad. Sci. USA 103, 13753–13758. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Stewart-Swift C, Andreeva V, Gibert Y, Connolly M, Lee Y, Fraher D, Burt J, Cardarelli J, Lattanzi V, Sullivan D, Albertson RC, and Yelick PC (2010). Forward Genetic ENU Chemical Mutagenesis Screen for Mineralized Craniofacial, Skeletal and Tooth Mutants in Zebrafish. 9th International Zebrafish Development and Genetics Conference, Madison, WI, June16–20, 2010. [Google Scholar]
  51. Topczewski J, Sepich DS, Myers DC, Walker C, Amores A, Lele Z, Hammerschmidt M, Postlethwait J, Solnica-Krezel L (2001). The zebrafish glypican knypek controls cell polarity during gastrulation movements of convergent extension. Dev. Cell 1, 251–264. [DOI] [PubMed] [Google Scholar]
  52. Wagner GP (1984). Coevolution of functionally constrained characters: prerequisites for adaptive versatility. Biosystems 17, 51–55. [DOI] [PubMed] [Google Scholar]
  53. Wagner GP (1990). A comparative study of morphological integration in Apis mellifera. Zeitschrift fŸr zoologische. Systematik und Evolutionsforschung 28, 48–61. [Google Scholar]
  54. Wagner GP (1996). Homologues, natural kinds and the evolution of modularity. Am. Zool 36, 36–43. [Google Scholar]
  55. Weatherly AH (1990). Approaches to understanding fish growth. Trans. Am. Fish. Soc 119, 662–672. [Google Scholar]
  56. West-Eberhard MJ (2003). Developmental Plasticity and Evolution. Oxford University Press, New York. [Google Scholar]
  57. Wu P, Jiang TX, Shen JY, Widelitz RB, and Chuong CM (2006). Morphoregulation of avian beaks: comparative mapping of growth zone activities and morphological evolution. Dev. Dyn 235, 1400–1412. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Wu P, Jiang TX, Suksaweang S, Widelitz RB, and Chuong CM (2004). Molecular shaping of the beak. Science 305, 1465–1556. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Yamamoto Y, Byerly MS, Jackman WR, and Jeffery WR (2009). Pleitropic functions of embryonic sonic hedgehog expression link jaw and taste bud amplification with eye loss during cavefish evolution. Dev. Biol 330, 200–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Young NM (2006). Function, ontogeny and canalization of shape variance in the primate scapula. J. Anat 209, 623–636. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Hernandez LP (2000). Intraspecific scaling of feeding mechanics in an ontogenetic series of zebrafish, Danio rerio. J. Exp. Biol 203, 3033–3043. [DOI] [PubMed] [Google Scholar]
  62. Zelditch ML, and Fink WL (1995). Allometry and developmental integration of body growth in a piranha. Pygocentrus nattereri (Teleostei: Ostariophysi). J. Morph 223, 341–355. [DOI] [PubMed] [Google Scholar]
  63. Zelditch ML, Sheets HD, and Fink WL (2000). Spatiotemporal reorganization of growth rates in the evolution of ontogeny. Evolution 54, 1363–1371. [DOI] [PubMed] [Google Scholar]
  64. Zelditch M, Swiderski D, and Sheets D (2004). Geometric morphometrics for biologists: a primer. Elsevier, San Diego, CA. [Google Scholar]

RESOURCES