262 Ulrich Krohs
Since modularity includes a dynamic dimension, the concept is applicable to develop-
mental processes as well. Characters of adult organisms do not pop up out of nothing; they
develop during the ontogenesis of the organism. It was an interesting fi nding that devel-
opmental processes show near-decomposability as well. An example is the development
of the hind limb of tetrapods. The hind limb is both a discrete structure and a develop-
mental module with a unique and intrinsic set of patterning mechanisms (Raff 1996;
Franz-Odendaal and Hall 2006). The character of the adult organisms is used to delineate
the developmental module—the set of developmental pathways and resources that brings
about this very character. This is again a functional delineation, and nothing guarantees
that the interactions within this functional module are stronger than the interactions with
processes involved in the development of other characters. But the structural delineation
of developmental processes indicates that functional and structural modules seem to coin-
cide in this case. However, biological evolution has not always delineated modules as
nicely as engineers tend to do. Biological modularity (but also its technological counter-
part) comes in degrees and nature’s joints are sometimes fuzzy. So if structural modular-
ization is the aim, developmental modules must be delineated not by their products
but—like the modules of metabolic networks—by unbiased methods that identify semi-
autonomous developmental pathways (Raff 1996), as is in fact successfully performed
(e.g., Davidson et al. 2002; Davidson and Erwin 2006).^6
15.3 Explanations for the Evolution of Modularity
Modular systems can evolve from different starting points by changes going in opposite
directions: by parcellation of a highly integrated system, or by integration of existing
systems (Callebaut 2005: 9). The mechanisms producing modularity are usually described
as specialization of existent structures in the case of parcellation, and assembly processes
in the case of integration (Simon 1969: 193). Since not every specialization needs to end
up in parcellation, and components integrated by assembly may simultaneously specialize,
evolutionary modularization processes need not belong to only one of the kinds. However,
it is clear that both ways of modularization are relevant for biological evolution: 1) The
eukaryotic cell evolved by integration of prokaryotic cells of different species. In particu-
lar, incorporated bacteria that already possessed a respiratory chain became mitochondria,
and photosynthetic bacteria were modifi ed into plastids (Sagan [Margulis] 1967; Margulis
1970, 1981; for the history of this idea see Khakhina 1992). 2) Parcellation by specializa-
tion of eukaryotic cells into different tissues and organs occurred during the further evolu-
tion of higher metazoa (animals and plants).
The mere description of the different ways by which modularity may or did evolve is
not all that biologists aim for. The evolutionary processes also require causal explanation.
Different explanations were proposed and discussed in the literature. I focus here on the