Keystone species

A keystone species is one whose impact on its community or ecosystem is disproportionately large relative to its abundance [8]. The classic example is a starfish (Pisaster ochraceus) in the rocky intertidal of the Pacific Northwest:

• Pisaster ochraceus is an efficient predator of the common mussel, Mytilus californicus.

• It reduces abundance of M. californicus, allowing other macroinvertebrates to persist.

• Experimental removal of P. ochraceus results in near total dominance of intertidal by Mytilus to the exclusion of other intertidal macroinvertebrates.

• Pisaster present $\Longrightarrow$ diverse intertidal community

• Pisaster absent $\Longrightarrow$ depauperate intertidal community dominated by M. californicus.

Kangaroo rats (members of the genus Dipodymys) play a similar role in the Chihuahuan desert of southeastern Arizona [2].

• Kangaroo rats are the largest of the seed-eating desert rodents. James Brown and collaborators excluded them from experimental plots starting in 1977.

• From 1977-1995 energy use by rodents on plots without kangaroo rats averaged only 14% of the energy use by kangaroo rats on matched control plots.

• The density of small seed-eating rodents averaged almost two times higher on plots without kangaroo rats.

• In 1996 a pocket mouse, Chaetodipus baileyi, colonized the plots, and by 1999 energy use increased to 80% of the energy use by kangaroo rats.

If we're going to focus on conserving whole systems, it might make sense to begin with a focus on keystone species. After all, ensuring that they persist and, more importantly, that their ecological role persists would, it seems, guarantee that the system as a whole will persist relatively intact. Although the idea of keystone species is intuitively appealing, it has been much criticized in the ecological literature.¹ Why?

• Even a disproportionately large impact may be small if the species is very rare.

• How ``disproportionately large'' is disproportionate enough for something to be considered a keystone?

• More importantly, investigators tend to identify keystone species as those that have a large and consistent impact on the dynamics of other species. ``Large'' makes sense, but does ``consistent''?
  • Berlow [1] points out that species effects identified as ``weak'' are often extremely variable among replicates, with variation on the order of the strongest mean effects observed (Figure 1).

    Figure 1: Interaction strength versus range of interaction strength for (a) six rock intertidal grazers on kelp, (b) praying mantids feeding on arthropods, and (c) six predators on six benthic invertebrates in the Ythan river estuary, Scotland (from [1]).

    
    1. It is important to distinguish among effects that are strong, but variable (in magnitude and direction), those that are consistently weak, and those that are consistently strong (and in a single direction). Strong effects that differ in direction from one circumstance to the next might, on average, appear to have no effect.

    2. By identifying species with ``strong'' effects, we may simply be identifying those that have low variance.

  • He goes on to illustrate how important ``weak,'' but variable, interactions might be in a simple experimental system.
    • Whelks (Nucella emarginata and N. canaliculata) prey on our old friend Mytilus californicus. By manipulating the density of whelks in three experimental treatments (0, 50, and 150 per square meter), he showed that the Mytilus declines linearly with increasing density of Nucella (Figure 2). In this experiment, Berlow transplanted mussels to experimental plots and used enclosures to manipulate whelk density.

      Figure 2: Direct effect of whelks (Nucella emarginata) on mussels (Mytilus californicua) on the rocky, central coast of Oregon (from [1]).

      

    • In a second experiment, Berlow manipulated whelk density (none, low, and high) and the abundance of an alternative prey (acorn barnacles, Balanus glandula). The experimental plots were scraped bare, mussels and barnacles were allowed to colonize, barnacles were then removed monthly from half of the plots, and whelk density was maintained with enclosures. In the barnacle removal plots, both low- and high-density predation have a negative effect on mussels (Figure 3a,b,c). Unlike the transplant experiment above, the strong negative effect of whelks is seen even at low density in the absence of barnacles. Mussel colonization is low in the absence of barnacles, so whelks essentially eliminated the few that managed to settle.

      Figure 3: Effects of whelks on mussel density when newly settled barnacles are removed monthly (a-c) or allowed to remain (d-f) (from [1].)

      

      When barnacles are present, mussel recruitment is enhanced. So not only do barnacles provide an alternate food source for whelks, they increase the chancels that mussels will become established. When whelks are abundant, however, they reduce the abundance of mussels regardless of the barnacle density - not too surprising. But when whelks are rare, they enhance mussel abundance when barnacles are very common, have no detectable effect when barnacles are only moderately common, and decrease mussel abundance when barnacles are uncommon (Figure 3d,e,f).

    • Clearly whelks have an important influence on the abundance of mussels. It's just that the influence they have depends on the environmental context, i.e., the abundance of mussels. And yet, if we simply took the average effect across years, we'd be tempted to conclude that whelks don't have much effect.

McCann et al. [6] also point out that it may be the weak trophic interactions that are most important in stabilizing food webs. Weak links dampen the oscillations in population size characteristically associated with resource-consumer dynamics, which tends to keep both resource population sizes and consumer population sizes bounded away from zero and decreasing the chance they will become extinct. Thus, strong links may have the greatest influence on overall community dynamics and structure, but it's the abundance of weak links that allow communities to persist.

In spite of its limitations, the keystone species concept has some usefulness in the same way that elasticity coefficients in a Leslie or Lefkovitch matrix are useful. It directs attention to those aspects of the system where management efforts are likely to have the greatest impact. Of course, identifying those aspects is not an easy task.