Arguments against resource-based definitions and models
The idea that resource dynamics should be part of most models of the dynamics of competing species seemed to be more widely accepted 40 years ago than it is today. At that time I thought that the Lotka-Volterra model was likely to be largely abandoned in the future, as its form had been shown to be a very special case that was inconsistent with almost all consumer-resource models.
This abandonment never happened; Chapter 4 examines a number of influential recent theoretical articles that either use the LV as their primary model or fail to include resource dynamics. It is useful to ask why methods of defining and measuring interactions have not been reformed.There are two main arguments against requiring resource dynamics. The first is that inclusion of resources is just one of many ways one could make the LV model more realistic. Representing resources is no more important than representing age or size structure or intraspecific variability in resource use or evolutionary change, or a number of other factors left out of most models in community ecology. The second argument asserts that some forms of competition do not include resources, so the necessary inclusion of resources would rule them out.
Bolker’s (2008) book is often cited in support of the first argument above. He pointed out that there was no clear division between ‘mechanistic’ and ‘phenomenological’ models; there was no obvious point at which you should begin or end with the addition of real-world features that were not represented in simpler models. The difference between resources and the other omitted details, however, is the nearuniversal presence of resources, even under the negative effects definition, and the completely universal presence of resources in the historical definition of competition used here. There is no other class of indirect interactions in which the intervening species/entity is left out of the definition or left out of models.
This is largely because the dynamics of intervening species/entities plays a major role in determining theMeasuring competition: a consumer-resource framework • 41 nature of the interaction. The only difference between competition and other indirect interactions involving 3 dynamic entities is that the intermediate entity need not be a biological species under competition; it can be detritus or a nutrient or sunlight or space (or one of many other entities that are not self-reproducing). Although some of these intermediate entities can reach equilibrium with respect to the consumer more rapidly than most biological species, they are equally likely to have dynamics that are slow enough that they cannot be ignored in systems where the system does not come to rest at an equilibrium (i.e., has the significant fluctuations in abundance that characterize most natural populations (Pimm 1991)). This result will be justified in Chapters 8 and 9 on competition in variable environments. This is not to say that age or size structure, inter-individual variation in other traits, time lags between resource intake and reproduction, or other features lacking in the LV model are not in need of exploration. However, they are important in all aspects of ecology, and are not part of the defining properties of the competitive process.
The second argument against universal inclusion of resource dynamics is the idea that competition often does not involve resources. For example, the textbook, Community Ecology, by G. G. Mittelbach and B. J. McGill (2nd ed., 2019) argues (p. 146) that ‘consumer-resource models poorly represent some common types of competition (e.g. interference competition and apparent competition...). Even for exploitation competition, consumer-resource models depend on the notion of distinct resource types.' The first sentence in this quotation is countered by points made above; ‘interference' usually produces effects on resources, and, in most, if not all, cases, it would not exist if interference had no effect on subsequent resource abundance.
The second point about exploitation is certainly true, and neither I nor any other ecologist that I know of has suggested a restriction of the definition of competition to systems with only a single resource. It is true that interference can come in many forms, at least some of which need to be distinguished to be properly represented. Impacts on mortality and those on feeding rates are different forms of interference and have different consequences, both in reality and in consumer-resource models. However, there is no particular difficulty in representing both in a consumerresource model; feeding interference makes the resource capture rate a decreasing function of the competitor's abundance, rather than having competitor abundance increase the mortality rate. In contrast, models that lack resources do not have a way of representing this difference. If the competitor causes mortality directly it can in theorybe modelled equally well under frameworks with and without resources. However, in such a case, the consumer species that can be killed by a competing consumer is likely to exhibit behavioural avoidance of that second consumer, and this is likely to alter its resource consumption behaviour as well. Such behavioural effects cannot be modelled without explicit consideration of the resource. Moreover, direct mortality or other ‘interference' between consumers that does not entail consumption (i.e. that is not ‘mutual predation') is very unlikely to occur unless it increases subsequent resource availability. This requires some spatial localization of resource items, and it therefore means that an understanding of the competitive interaction requires models that explicitly represent the spatially distinct populations of both consumers and resources.The conclusion from this section is that understanding the dynamics of the resources that underlie competitive interactions is usually necessary to describe the interaction of two or more consumer species. The next section examines MacArthur’s ‘all-linear’ consumer-resource model, which has been used as the main logical justification for using and/or believing results from the LV model.