Modelling the ecological–functional diversification of marine Metazoa on geological time scales

Abstract

The ecological traits and functional capabilities of marine animals have changed significantly since their origin in the late Precambrian. These changes can be analysed quantitatively using multi-dimensional parameter spaces in which the ecological lifestyles of species are represented by particular combinations of parameter values. Here, we present models that describe the filling of this multi-dimensional ‘ecospace’ by ecological lifestyles during metazoan diversification. These models reflect varying assumptions about the processes that drove ecological diversification; they contrast diffusive expansion with driven expansion and niche conservatism with niche partitioning. Some models highlight the importance of interactions among organisms (ecosystem engineering and predator–prey escalation) in promoting new lifestyles or eliminating existing ones. These models reflect processes that were not mutually exclusive; rigorous analyses will continue to reveal their applicability to episodes in metazoan history.

1. Introduction

The first animals were probably sponges, and they formed ecological communities that were considerably simpler than later animal communities. Ecological complexity increased during the evolutionary radiations of the Ediacaran and Phanerozoic [1–7] as animals began to interact with the environment and each other in diverse ways. Here, we discuss several models of ecological–functional diversification that may describe the history of the marine Metazoa. These models invoke various ecological and macroevolutionary processes and can be applied to fossilized animals on geological time scales.

2. Multi-dimensional representations of ecology

To model the ecological diversification of marine Metazoa, organismal ecology/function must be quantified in a way that is applicable to disparate taxa. The approach outlined here focuses directly on the inferred ecological and functional attributes of organisms; these models thus complement more common methods like phylogenetics or morphometrics. Ecology is quantified by scoring fossil species or genera on a set of independent ecological parameters that define a multi-dimensional space (‘ecospace’; [1]). The parameters encompass important ecological features that can be inferred from fossil material through functional morphologic analysis and phylogenetic comparison; such traits often can be inferred despite the information loss inherent in fossilization [4].

Parameters typically include aspects of feeding, habitat, mobility and size (figure 1 and table 1). Parameter values are fairly general, which facilitates comparisons of diverse animals throughout the Metazoa. Thus, these ecospaces are designed to permit analyses of long-term, global patterns (e.g. ecological diversification during the Phanerozoic). However, the methods could also be applied to finer scale questions, although the causative processes might differ [8] and more refined ecological assignments could be used [2]. Once a set of fossil species is classified ecologically, one can examine the temporal dynamics of ecological–functional diversification quantitatively in the resulting multi-dimensional space [2–6].

Figure 1.

Ecological–functional parameters used to quantify multi-dimensional ecospaces. (a) Ecospace of Bush et al. [2], which classifies animals according to tiering (position with respect to the sediment-water interface), method of feeding, and degree of motility and attachment. There are 216 unique combinations of characters, or theoretically possible ecological lifestyles. Modified from Bush et al. [2] and used with permission from the Paleontological Society. (b) Thirteen parameters from Novack-Gottshall's [5] ecospace. There are more than three dimensions, so each dimension is drawn separately; five-ordered, multi-state traits are not illustrated. Boxes represent possible character states; see table 1 or [5] for details. Trillions of ecological lifestyles (unique combinations) can be coded with this ecospace.

Table 1.

Some characters in Novack-Gottshall's [5] theoretical ecospace (figure 1b). (For each character, possible character states are shown in parentheses. For another approach to the ecological classification of fossils see figure 1a [2,3].)

reproduction: reproduction (sexual and asexual)

mobility: mobility (sedentary, passively mobile, facultatively mobile, intermittently mobile and habitually mobile)

substrate: substrate/medium composition (biotic, lithic and fluidic), substrate consistency (hard, soft and insubstantial) and substrate relationship (attached and free-living)

tiering: primary microhabitat (above primary substrate and within primary substrate), immediate microhabitat (above intermediate substrate and within intermediate substrate) and support (supported and self-supported)

feeding: primary feeding microhabitat (above primary substrate and within primary substrate), immediate feeding microhabitat (above intermediate substrate and within intermediate substrate), diet (autotroph, microbivore, carnivore, herbivore and fungivore), physical condition of food (incorporeal, solution, particle and bulk) and feeding habit (ambient, filter, attachment, mass and raptorial)

size: skeletal body volume

3. Models of ecospace diversification

Palaeontologists have discussed aspects of ecological diversification, often one parameter at a time [9–11], but analytical modelling in a unified framework (as recommended above) may reveal important processes, feedbacks and interrelationships. Although the scale and causative processes differ, ecological diversification can be envisioned as analogous to assembly within ecological communities [12,13]. The following models (figure 2) are not exhaustive, but they represent major perspectives through which palaeoecological diversification can be interpreted [14]. Most models describe the pattern of ecospace filling as diversification proceeds according to specific assembly rules; others describe the emptying of ecospace.

Figure 2.

Models of change in ecospace utilization from time 1 to time 3; different macroevolutionary processes produce different distributions of species in functional ecospace. Each point represents an ecological lifestyle characterized by a unique combination of functional traits (figure 1), and proximity represents degree of similarity in functional traits. The two-dimensional space is a simplified representation of a multi-dimensional space; analytical testing could proceed by a variety of means. In (b), the widening range of viable lifestyles (dotted boxes) indicates expansion. See text for details of models.

(a). Diffusion (neutral, null and passive) model

A process-neutral model lacking assembly rules; ecospace is filled at random via Brownian diffusion as new species arise (figure 2a). Biotic interactions and environmental change play little role. This model is more common in ecology (e.g. [15,16]) than in palaeoecology (but see [17,18]).

(b). Expansion model

The range of potentially used lifestyles increases through time as new species occupy progressively more novel lifestyles (figure 2b). Unlike the diffusion model, expansion is a driven process [2,6]. Abiotic or biotic factors could drive expansion; for example, rising oxygen levels [19,20] or primary production [21,22] could permit larger animals and more energetic lifestyles. Other causes could include shifts in habitat availability [23,24], advent of key innovations [25] and adaptive radiations [26]. This model has been widely invoked in metazoan diversification.

(c). Contraction model

Ecological lifestyles are eliminated through time, emptying areas of ecospace and counteracting ecological diversification (figure 2c). As examples, some mass extinctions selectively extirpated certain ecologies [27,28].

(d). Positive feedback model

In this variant on the expansion model, interactions among metazoans increase the range of potentially occupied lifestyles. Specifically, the origin of one set of lifestyles permits or drives the evolution of another set (figure 2d). For example, ecosystem engineers like corals modify the environment [29,30], providing new opportunities for other clades [31]. Shelly taxa create hard grounds for endolithic and encrusting taxa, and every metazoan provides opportunities for parasites. By altering the adaptive landscape, predators likewise promote new ecologic strategies among prey taxa, e.g. evolution of infaunal lifestyles [11].

(e). Negative feedback model

In this combination of the expansion and contraction models, biotic interactions cause some ecologic lifestyles to become untenable through time; the origin or expansion of one set of lifestyles eliminates or restricts existing lifestyles (figure 2e). Exclusion can result from ecological engineering such as seafloor bioturbation (biological sediment mixing), which increased through time and eliminated lifestyles adapted to a stable sediment–water interface [10,32]. Increased predation also made some lifestyles untenable or limited them to refugia [11].

(f). Redundancy model

Ecospace is occupied by clusters of lifestyles, and successive species occupy lifestyles similar to existing ones (figure 2f), such that overall occupation of ecospace is relatively unchanged. This model, once considered ecologically untenable owing to competitive exclusion, has been reinvigorated in discussions of adaptive peaks, the lack of novelty after the Cambrian radiation [33,34], niche conservatism [35], and the role of predation and disturbances in limiting competitive interactions [18].

(g). Partitioning model

New species occupy lifestyles intermediate to existing ones (figure 2g), perhaps owing to niche partitioning and resource specialization. Like the redundancy model, this model predicts that the range of ecospace utilization changes little, but it predicts progressively greater packing of lifestyles. Partitioning is implied in discussions of tiering [9]; also see Valentine [36].

4. Applications

Ecospace utilization clearly has changed through time, but the exact pattern of change and the processes responsible are not fully understood. The models outlined above provide starting points for testing these patterns, and multiple analytic methods are available to evaluate them (e.g. [3,6]). Different models, or combinations of models, are expected to apply to different intervals of life's history.

To close, we highlight several interesting questions that deserve further testing.

• — Ecospace occupation has increased greatly through time, particularly during the early history of the Metazoa [3,4,6], but both the null and expansion models predict such an increase. Was this change passive or driven?

• — Positive and negative feedbacks have affected the occupation of ecospace through ecospace engineering, predation and other interactions. Compared with other processes, how much did these feedbacks affect the amount of ecospace occupied?

• — How much ecospace was evacuated during various mass extinctions, and how does this relate to the magnitude and selectivity of extinctions? What were the dynamics of ecospace filling during the recoveries from mass extinctions?

• — What was the relative importance of the redundancy and partitioning models (i.e. niche conservatism versus subdivision) in the filling of ecospace?

We hope that the methods and models summarized above will spur further testing of these and other questions.