vol. 155, no. 6 the american naturalist june 2000 The Logic and Realism of the Hypothesis of Exploitation Ecosystems Lauri Oksanen* and Tarja Oksanen Department of Ecology and Environmental Sciences, UmeaÊ Keywords: herbivory, mammals, primary productivity, trophic cas- University, S-901 87 UmeaÊ, Sweden cades, vegetation, predation. Submitted June 24, 1998; Accepted December 29, 1999 In the current debate on trophic dynamics in terrestrial ecosystems, two main views can be distinguished. One emphasizes defenses themselves or the diversity of defen- abstract: Hypotheses on trophic dynamics in terrestrial ecosystems sive strategies in herbivores and plants and maintains that fall into two major categories: those in which plants are assumed to strong, cascading trophic interactions are uncommon in be invulnerable to their consumers and those in which the build-up species-rich terrestrial ecosystems (Murdoch 1966; White of plant biomass is assumed to require top-down control of folivores. 1978; Hunter and Price 1992; Strong 1992; Polis and The hypothesis of exploitation ecosystems (EEH) belongs to the latter Strong 1996; Polis 1999). If anything, consumers are seen category and focuses particularly on the consequences of the high as factors that might change the behavior of their resources energetic costs of maintenance of endotherms. Carnivorous endo- therms require relatively high prey densities in order to break even. or the composition of resource guilds (Haukioja and Hak- Moreover, they are dependent on folivorous prey during the limiting ala 1975; Rhoades 1985; Leibold 1989, 1996; Abrams 1992, season, at least at high latitudes. The endotherm branch of the grazing 1993, 1996; Pastor and Cohen 1997). In order to structure web is thus predicted to collapse from three-link trophic dynamics the discussion, the above view will be referred to as the (carnivores r folivores r plants r inorganic resources) to two-link defense diversity hypothesis (DDH). A diametrically op- r r dynamics (folivores plants inorganic resources) along gradients posite view of trophic dynamics was outlined by Elton of decreasing primary productivity. Consequently, the vegetation of (1927) and is more rigorously presented in the green world cold and unproductive areas is predicted to be under intense winter grazing pressure, which prevents the accumulation of aboveground hypothesis (HSS) of Hairston et al. (1960). According to plant biomass and excludes erect woody plants. In the most extreme HSS, plants are vulnerable to folivores but are nevertheless habitats (e.g., polar deserts and their high alpine counterparts), even seldom severely defoliated because the collective density folivorous endotherms are predicted to be absent, and the scanty of folivores (grazers, browsers, and folivorous insects) is vegetation is predicted to be structured by preemptive competition. regulated by the collective action of carnivorous and in- Within temperature-determined productivity gradients, EEH is cor- sectivorous animals. The hypothesis of exploitation eco- roborated by biomass patterns, by patterns in the structure and dy- systems of Oksanen et al. (1981; to be referred to as EEH) namics of carnivore, folivore, and plant communities, and by ex- perimental results. The general idea of top-down trophic dynamics converges with HSS with respect to productive areas (for- is supported for other autotroph-based systems, too, but the rele- ests and their successional stages, productive wetlands). vance and suf®ciency of the energy constraint in explaining patterns According to EEH, however, the control of folivorous en- in trophic dynamics appears to be variable. Moreover, critical em- dotherms by predators fails in unproductive ecosystems pirical evidence for or against the capacity of folivorous insects to (tundras, high alpine areas, steppes, and semideserts), regulate plant biomass has not yet been obtained. Another open which are characterized by intense natural folivory. question is the ability of boreal and temperate browsers, evolved in Considering the time elapsed since the formulation of productive environments with intense predation pressure and abun- dance of forage, to prevent the regeneration of the least palatable the two main alternatives, relatively few critical experi- tree species. There are, thus, many open questions waiting to be ments have been conducted. Differences in approach and answered and many exciting experiments waiting to be conducted. terminology have probably contributed to this situation. The DDH is focused on individual populations (e.g., Ehr- * To whom correspondence should be addressed; e-mail: lauri.oksanen@ lich and Birch 1967; Polis and Strong 1996), while HSS eg.umu.se. and EEH deal with guild-level dynamics of plants, foli- Am. Nat. 2000. Vol. 155, pp. 703±723. q 2000 by The University of Chicago. vores, and carnivores. These guilds have been referred to 0003-0147/2000/15506-0001$03.00. All rights reserved. as trophic levels, which is conceptually debatable. Nev- 704 The American Naturalist ertheless, they are identi®able units, created by adapta- our work during the past 2 decades, starting from the 1981 tional constraints and even re¯ected in taxonomy. More- model of EEH. over, the conjecture that individual folivore populations The independent variable of EEH is potential produc- display density-vague dynamics (Strong 1986) is entirely tivity, de®ned as the maximum primary productivity of compatible with the conjecture of Hairston et al. (1960) the habitat, when its nutrient pool is fully available, when that the collective standing crop of folivores is regulated leaf area index has reached its optimal value, but before by the collective action of carnivores. the accumulation of heterotrophic stem tissues. In terres- The purpose of this article is, ®rst, to explain why we trial ecosystems, this value depends primarily on evapo- regard both DDH and HSS as incomplete; second, to rean- transpiration and secondarily on nutrient pool (Lieth alyze the 1981 model of EEH with focus on endotherms, 1975). Belowground production is included because graz- relaxing unnecessarily speci®c assumptions of the original ing can dramatically change community-level allocation model, and replacing graphics with explicit analysis; third, patterns between roots and shoots (Tilman 1988; HambaÈck to provide an up-to-date review of relevant empirical ev- 1998). For herbaceous vegetation and low scrublands, ac- idence; and ®nally, to outline a research strategy on trophic tual net primary production is a good index for potential dynamics. productivity. For woodlands, potential productivity should ideally be measured in the early stages of secondary suc- cession. Values for mature forests must be corrected for The Productivity Connection respiration of stem tissues. In northern ecosystems, pri- mary productivity is even in¯uenced by indirect effects of For us living at high latitudes, the shared point of de secondary succession (SireÂn 1955), which can be prevented parture of DDH and HSSÐrarity of severe foli- (Zimov et al. 1995) or accelerated (Pastor and Naiman voryÐappeared unrealistic. In the early 1970s, we saw 1992; Pastor et al. 1993) by folivory. However, feedbacks Norwegian lemmings destroy the moss cover of the tundra caused by biotic factors cannot be included in the inde- (Kalela 1971; Kalela and Koponen 1971). Simultaneously, pendent variable, which must derive from the properties subarctic forests were dying over enormous areas. The of the physical environment. main trunks had been killed after recurrent defoliation by The intention of the simple model of EEH was to pro- a geometric moth, and the basal recovery shoots were duce a parsimonious hypothesis on trophic dynamics consumed by mammals (Kallio and Lehtonen 1975). along broad biogeographical gradients, where potential Moreover, the profound impact of reindeer and gray-sided productivity varies by orders of magnitude. Unfortunately, voles on the vegetation of the heathlands was easy to see the 1981 model includes two unnecessary assumptions: (Oksanen 1978; Callaghan and Emanuelsson 1985; Ok- logistic growth in plants and Type II functional response sanen and Virtanen 1995). Except for the insect outbreak, in consumers. Moreover, the team was divided on two the strong impacts were not caused by exceptionally high central issues. The American part (S. Fretwell and J. Ar- numbers. Densities of microtine rodents were lower in ruda) wanted to retain the integrity of trophic levels, typical tundra than in productive habitats (willow thick- whereas the Finnish part (L. Oksanen and P. NiemelaÈ) ets), where no dramatic changes in vegetation were ob- preferred to restrict the model to endotherms (Oksanen served (Oksanen and Oksanen 1981). The winter reindeer et al. 1981, p. 257). Moreover, the Finns preferred to model density in the area where forests were changing to tundra secondary carnivory as diffuse intraguild predation (Oks- was about three animals or 200±250 kg km22, excluding anen et al. 1981, p. 250). The formal analyses presented a period of mass starvation in the mid 1970s (L. Oksanen in the paper were based on the American alternative. In et al. 1995). This density is only twice as high as the average the verbal part, the paper glided confusingly between the density of barren-ground caribou in northwestern main- two views. Below, we will reanalyze the model, relaxing land Canada (CreÃte 1999). Higher standing crops of cer- super¯uous assumptions and consistently applying the