In most instances, these concepts are taken from economic analogies, yet, like economics, the concept of cost is meaningless without a context. The two defining elements that must be considered are value and affordability.
In general economic terms we recognize that many items may have costs, but without assessing their value it is impossible to determine how to evaluate the cost of something. Similarly, while something may have a high basic cost, this may be irrelevant if the item is readily affordable. Therefore, it is quite difficult to arrive at any meaningful interpretation with which to assess something as vague as "efficiency".
So it is in biology. We must understand that the entire basis of life is that everything "costs" something. The simplest organism "costs" more to exist than not to exist.
Therefore to assess these two main cost areas, let's consider that metabolic costs are incurred as a byproduct of having to acquire the energy to maintain a particular trait. That a particular trait "costs" something is not in dispute, but the first question is whether such a trait has value and to what degree. If a trait provides a useful quality, then it may have value, but this must be measured against the affordability aspect. If it takes a tremendous amount of effort [energy] to sustain, then it becomes a delicately balanced objective which can quickly become a liability if conditions change. Correspondingly if the energy requirements are readily met, then the value of such a trait becomes relatively easy to attain and maintain.
Fitness costs also fall into this category, but they have an additional interpretative difficulty which includes assessing what the maximum fitness for a particular organism is. Furthermore this presumes that the trait in question is highly heritable. Since heritability represents the amount of variation in a population due to genetic influences, then the fundamental requirement to assess fitness costs requires that there be some variation in a population to begin with. More importantly, this variation must be directly linked to genetics. It isn't enough to simply argue that one animal may be more efficient operating on a reduced calorie diet, since even if it results in a fitness advantage to that animal, unless it is capable of being inherited by offspring and persists for future generations, such variations are simply minor oscillations in a species' existence.
Using the common example of the peacock's tail, we invariably hear about the biological cost of such a structure, however, this is meaningless without establishing the actual difficulty presented to the peacock in maintaining it. If resources are plentiful, then it isn't likely to be much of a "cost" factor, regardless of how much metabolic energy it takes.
Similarly when considering fitness costs, the general assumption is that the tail presents a liability in dealing with predators, but again, for this to be a valid consideration one must be able to demonstrate that there is a distinct difference in survival rates between peacocks that possess such a tail versus those that don't.
Even further, if there is no demonstrable difference in the tails [or their implied liability] in a population of peacocks, then it cannot be argued that there is a heritability factor that is selectable. In such creatures, should the tail prove to be a liability, then it may be that the species goes extinct, long before individuals can be selected for the absence of a tail.
Therefore, fitness costs can only be considered when there are two traits offering different fitness criteria that may be selected for within a population. As a result, one will tend to dominate as the conditions favoring its existence take hold. If there is no variation, then the trait is not heritable, therefore it is not selectable, and there is nothing meaningful to be said about fitness costs.
Similarly when considering metabolic "costs" it becomes convenient to think that efficiency is a significant factor, but we already have enough evidence that such efficiency is also a liability. The simple reality is that too much resource translates into fat animals. One point which is rarely considered is that some traits may serve as a regulatory mechanism to ensure proper caloric balance. In other words, if resources are readily available, then we already know that a creature that is too indulgent becomes fat thereby incurring a survival risk. However, it may make sense that certain phenotypic structures exist precisely to ensure that the creature's caloric requirements can be safely met, with the additional physical traits requiring just enough calories to ensure a better balance. This might readily be an extension of epigenetic functions, where the abundance or availability of food can determine the size or development of particular traits. This could be a simple regulatory mechanism to ensure that animals optimal physical size is balanced against resource availability. Fewer resources results in smaller animals, while greater resources results in larger animals. This ensures that such creatures don't have to have elaborate metabolic controls to avoid obesity. It is their size which helps approximate their optimal caloric state. While this can also become a liability such traits provide a high degree of variability without any need for specific selection.
Often metabolic costs are associated with the loss of traits [such as eyesight in cave-dwelling creatures] and rationalized as a quest for greater efficiency, as if natural selection was simply waiting for an opportunity to dump this trait because it was expensive. However, to make such a claim one would have to demonstrate that there is an actual metabolic difference. This is especially relevant if one considers the requirements for dietary changes that might accompany such a transition. In addition, one has to consider whether the loss of a trait is actually a loss, versus simply a "reassignment" where such a loss is offset by utilizing the existing capabilities in other capacities.
In another reproductive scenario, many creatures simply rely on probability by producing a sufficiently large enough number of offspring, so that there is no direct relationship between the offspring and any particular survival trait. Clearly larger numbers of offspring can be a selectable trait, but this does not provide a link to any trait to promote survivability in offspring.
These are some of the interpretive risks associated with overly simplistic considerations regarding the issue of such "costs".
We know that creatures have adapted to a variety of environmental conditions, therefore it is reasonable to conclude that there are no inherent "costs" associated with survival in those environments that translates into much significance for individual populations (2). Some may do slightly better than others, but it isn't sufficient to have an effect unless there is a specific environmental pressure that is capable of acting on a highly heritable trait.
Competition between species can certainly raise the stakes and is generally dependent on the rate of reproduction being a primary factor that can overwhelm more slowly reproducing species. In this situation, one finds that the competition will either force one species into extinction or into another environmental niche. In turn, this will drive other selection pressures.
It is very difficult to assess these types of changes purely on the basis of our interpretation of "costs". Clearly the more extreme the environment the more volatile the survival of individuals of any species. This helps illustrate the point that the more resource rich the environment is, the less impact any concept of "costs" has on a population. The harsher the environment, then the more specialized each species becomes, focusing their "costs" on the specific niches they occupy. Using our economic analogy further, if conditions change radically, then the species essentially loses its "investment" and goes "bankrupt" [extinct].
(1) The distinction between fitness versus metabolic costs is largely arbitrary, so there should be no special significance assigned to what they are called. Instead it is just a convenient way of trying to distinguish different aspects of the organism's "energy budget".
(2) It is important to remember that traits develop because they are beneficial. Therefore if "costs" were a significant factor in evolution, then one would have to question how such traits can arise in the first place. Instead, it is much more likely that such traits develop precisely because they are already well within the "energy budget" of such an organism.