Predator-prey dynamics transform the tundra

Arctic fox by Eric Kilby

Editor's Introduction

Introduced Predators Transform Subarctic Islands from Grassland to Tundra

The collapse of the fur trade in the late 19th and early 20th centuries resulted in the introduction of foxes to >400 Alaskan islands as an additional fur source. However, several islands remained fox free from failed introductions or attempts at introductions were never made. As a result, these islands provided a natural setting for a large-scale experiment to evaluate the effects of exotic predators on ecosystems. It was observed that the introduction of foxes had large impact on the seabird populations. The authors of this paper attempt to further this understanding by investigating the role of seabird populations on islands with and without foxes on soil and plant nutrients.

NOTE: This article is part of a Collection of student-annotated papers that are the product of the SitC team’s research into best practices for using primary literature to support STEM education. For this reason, these papers have undergone an alternate review process and may lack educator guides. To learn more, visit the main Collection page: SitC Lab.

Paper Details

Original title
Introduced Predators Transform Subarctic Islands from Grassland to Tundra
Original publication date
Vol. 307, Issue 5717, pp. 1959-1961
Issue name


Top predators often have powerful direct effects on prey populations, but whether these direct effects propagate to the base of terrestrial food webs is debated. There are few examples of trophic cascades strong enough to alter the abundance and composition of entire plant communities. We show that the introduction of arctic foxes (Alopex lagopus) to the Aleutian archipelago induced strong shifts in plant productivity and community structure via a previously unknown pathway. By preying on seabirds, foxes reduced nutrient transport from ocean to land, affecting soil fertility and transforming grasslands to dwarf shrub/forb-dominated ecosystems.


Nearly half a century ago, Hairston et al. (1) proposed that plant productivity and composition were influenced by apex predators through cascading trophic interactions. According to their “Green World” view, the direct effects of predators on herbivore populations transcend multiple trophic levels indirectly to enhance plant community productivity and biomass. Despite great progress in food web ecology, the indirect effects of top predators on vegetation dynamics in terrestrial systems remain unresolved and actively debated (26). Compelling demonstrations of multitrophic predator impacts on entire plant communities are scarce, in part because the spatial and temporal scales necessary to perform the appropriate community-wide experiments are daunting.

The introduction of predators to islands provides an opportunity to explore the indirect effects of predators on vegetation. Introduced predators commonly have devastating direct effects on their prey (7). The histories of these introductions are often well known, and the relative simplicity and isolation of insular systems facilitate the study of whole-community responses. Here we investigate how the introduction of arctic foxes (Alopex lagopus) to the Aleutian archipelago affected terrestrial ecosystems across this 1900-km island chain.

The Aleutian archipelago is a remote series of physically similar volcanic islands extending westward from the Alaska Peninsula (Fig. 1). The archipelago currently supports 29 species of breeding seabirds, together numbering >10 million individuals (8). Seabirds deliver nutrient-rich guano from productive ocean waters (9) to the nutrient-limited plant communities (1011). Historically, seabirds inhabited most islands along the Aleutian chain. Following the collapse of the maritime fur trade in the late 19th and early 20th centuries, foxes were introduced to >400 Alaskan islands as an additional fur source (12). The introduced foxes severely reduced local avifaunas, especially seabirds (13). However, several islands remained fox free, either because introductions failed or were not undertaken (1214). Hence, a large-scale natural experiment to evaluate the effects of exotic predators on insular ecosystems was unwittingly initiated more than a century ago. We use this experiment to show how differing seabird densities on islands with and without foxes affect soil and plant nutrients; plant abundance, composition, and productivity; and nutrient flow to higher trophic levels. These determinations were based on contrasts among 18 islands (9 with foxes and 9 fox free) (Fig. 1) that were matched as carefully as possible for size and location in the archipelago (12).

Fig. 1. The Aleutian archipelago with sample islands indicated in red (fox-infested) and blue (fox-free). Adak Island, the location of fertilization experiments, is indicated with a yellow dot.
Fox-free islands

The blue circles represent fox-free islands. The islands have remained fox free because either foxes were not introduced to this particular island or the attempt at introduction had failed.

Fox-infested islands

The red circles represent fox-infested islands. These islands had foxes introduced during the late 19th and early 20th centuries following the collapse of the fur trade.

Adak island

This island is represented by the yellow circle on the map. This was where the authors performed their fertilization experiments later described in the paper.

A geographical information system (GIS) was used to superimpose spatially explicit grids over maps of each island. All islands were sampled at the completion of the growing season (August) between 2001 and 2003. We established a 30 m by 30 m plot at each of the grid crosspoints (12 to 32 per island, depending on island size) (12), within which we sampled plant species presence and cover; aboveground plant biomass; total soil N, P, and δ15N; and %N and δ15N from a common grass (in most cases Leymus mollis but in some instances Calamagrostis nutkanensis) and forb (Achillea borealis) (12). At each island, we also haphazardly sampled δ15N in at least five individuals from a diverse group of terrestrial consumers, including a mollusk (Deroceras laeve), arachnid (Cybaeus reticulates), dipteran (Scathophaga impudicum), and passerine bird (Lapland longspurs, Calcarius lapponicus, and song sparrows, Melospiza melodia). δ15N was measured to determine the degree to which nitrogen-based nutrients were marine derived. Soils and organisms that obtain their N from locally fixed sources have lower δ15N values than those that obtain their N from higher trophic levels, such as marine fish and zooplankton (1516).

Breeding seabird densities were almost two orders of magnitude higher on fox-free than on fox-infested islands (Fig. 2B) (Mann-Whitney rank sum, T = 36, P < 0.001) (12). We estimate that this reduction in seabird abundance translates to a decline in annual guano input from 361.9 to 5.7 g m–2 (median values; T = 42, P = 0.005) (12). The resulting difference in marine nutrient input is reflected in soil fertility. Total soil phosphorus on fox-free islands was more than three times that on fox-infested islands (Fig. 2C) (F1,16 = 8.01, P = 0.012) (12). Although seabird colonies are often concentrated on the perimeter of islands, guano-derived nutrients can be broadly redistributed across islands and not solely concentrated within the colonies (17).

Fig. 2. Mean (±SE) values for parameters sampled on fox-infested (red) and fox-free (blue) islands in the Aleutian archipelago. (A) Photographs of typical plant communities on fox-infested (Ogangan Island) versus fox-free (Buldir Island) islands. (B) Logarithm of the density (birds m–2) of breeding seabirds estimated from population counts made by the U.S. Fish and Wildlife Service (26). (C) Soil Bray phosphorous (%). (D) Composition of island plant community from point contact counts of 1-m2 photo quadrats. (E) Grass biomass (g dry weight m–2). (F) Percent nitrogen composition of the dominant grasses (Leymus molis or Calamagrostis nutkanensis). (G) Percent nitrogen composition of a common forb (Achillea borealis).
Fig. 2A

In figure 2A the fox infested islands have a lower incidence of vegetation and is overall more barren looking than the fox free ones which boast lots of grasses.

Fig. 2B

In figure 2B it shows that the log density of seabirds is greater in fox free islands than fox infested ones by almost double.

Fig. 2C

Figure 2C shows that the total phosphate is higher again in fox free islands vs. fox-infested ones.

Fig. 2D

Figure 2D shows the percent (%) composition of different vegetation. For the fox infested island the vegetation ranked from highest to lowest is: grasses, forbs, shrubs, mosses and other. For the fox free island the vegetation ranked highest to lowest is: grasses, forbs, mosses being N/A (not available) and shrubs/other being very similar. Between the graphs the fox free island had higher amounts of grass than the fox infested but the fox infested islands had higher values for all other types of vegetation.

Fig. 2E-G

For panels E-G the fox free island had higher values for biomass, grass % nitrogen and forb % nitrogen respectively compared to the fox infested islands.

The different soil fertilities between fox-free and fox-infested islands corresponded with strong shifts among island types in the biomass and nutrient status of terrestrial plants, as well as overall composition of the plant community (Figs. 2, A to D, and 3A). Grass biomass was almost a factor of 3 higher (Fig. 2E) (F1,15 = 10.58, P = 0.005), shrub biomass was a factor of 10 lower (0.48 ± 0.31 versus 4.95 ± 0.84 g m–2F1,15 = 19.97, P < 0.001), and the nitrogen content in grasses and forbs was significantly greater on fox-free versus fox-infested islands (Fig. 2, F and G) (F1,16 = 8.28, P = 0.01 and F1,13 = 12.51, P = 0.004 for grasses and forbs, respectively). Plant communities on fox-free islands were graminoid dominated, whereas those on fox-infested islands had a more equitable distribution of graminoids, shrubs, and forbs (Fig. 2D).

Fig 3
Fig. 3.(A) Detrended correspondence analysis (DCA) comparing plant assemblages (based on the presence or absence of species) on fox-infested (red triangles) and fox-free (blue circles) islands. Analysis was conducted on species presence data from one of the three 1-m2 quadrats within 30 m by 30 m plots on the two island types. Each small point represents a census of plant species occurrence within a 1-m2 quadrat. Large points represent the mean ± 99% confidence interval axis scores from all samples taken across each island type. (B) Stable nitrogen isotope (†15N mean ± SE) analyses of soils and a suite of common species across trophic levels on fox-infested (red) and fox-free (blue) islands.
Graph set up

In part A of figure 3 the same colors as figure 2 are used to indicate which type of island; red for fox infested and blue for fox free. Each triangle represents the presence of plant species within a meter quadrat across the islands. For most of the measures the number of species falls between 150 and 250 with some outliers below 150, usually in the fox free islands. The outliers above 250 are concentrated from fox-infested islands and only 4 measures above 250 were from fox-free islands.

Stable nitrogen isotope

In part B the amount of a stable isotope of nitrogen in soil, grass, forbs, mollusks, passerines, dipterans and arachnids is shown. For every species and soil tested the amount of nitrogen isotope is higher in the fox free islands opposed to the fox infested ones where the amount of nitrogen is lower across all categories. The measure of nitrogen goes below zero in the grasses and forbs for the fox infested islands. This shows that the introduction of foxes has large effects on the nutrients available in ecosystems leading to fox free islands being more favorable for higher organisms than the fox infested islands which are mostly made up of grasses.

Detrended correspondence analysis

Part A in figure 3 is a Detrended Correspondence Analysis (DCA). DCA is a multivariable ordination method used to order sets of data points. The relative distance determines the correlation between each relative data point. The farther a point is away from another, the more different the points are compared to one another. The large circles represent the mean of similar island types, fox infested (red) and fox-free (blue).

15N measures from soils, plants, and consumers all indicate that fox introductions reduced nitrogen input from sea to land. δ15N was significantly greater in soils from fox-free islands compared with fox-infested islands (Fig. 3B) (F1,16 = 14.07, P = 0.002). Similar patterns in δ15N between fox-free and fox-infested islands were evident in grasses, forbs, mollusks, passerines, dipterans, and arachnids (Fig. 3B). These findings demonstrate that fox-free islands are strongly subsidized by marine-derived nutrients, which in turn assist in fueling the ecosystem at higher trophic levels.

To test whether differences in the magnitude of nutrient subsidies transported by seabirds onto fox-free versus fox-infested islands could have produced the observed differences in plant communities, we conducted a fertilization experiment on a large fox-infested island. Experimental nutrient additions to a community representative of fox-infested islands over 3 years caused a 24-fold increase in grass biomass (24.33 ± 6.05 g m–2) compared with control plots (0.51 ± 0.38 g m–2 increase; two-factor analysis of variance, F1,20 = 23.96, P < 0.001) and a rapid shift in the plant community to a grass-dominated state. In fertilized plots, grass increased from 22 (±2.7%) to 96 (±17.3%) of total plant biomass, whereas grass biomass in control plots was relatively unchanged (11.4 ± 3.0% and 12.1 ± 1.2% of total biomass at the start and end of the experiment, respectively) (12). In a parallel experiment (18), we disturbed and fertilized plots to mimic the effects of both seabird burrowing and guano addition. Here we found that disturbance negatively rather than positively affected grass biomass; the effects of fertilization alone were far greater than the joint effects of disturbance and fertilization. These results confirm the importance of nutrient limitation in these ecosystems and establish that nutrient delivery in the form of seabird guano is sufficient to explain observed differences in terrestrial plant communities between islands with and without foxes.

In total, our results show that the introduction of foxes to the Aleutian archipelago transformed the islands from grasslands to maritime tundra. Fox predation reduced seabird abundance and distribution, in turn reducing nutrient transport from sea to land. The more nutrient-impoverished ecosystem that resulted favored less productive forbs and shrubs over more productive grasses and sedges.

These findings have several broad implications. First, they show that strong direct effects of introduced predators on their na“ve prey can ultimately have dramatic indirect effects on entire ecosystems and that these effects may occur over large areas—in this case across an entire archipelago. Second, they bolster growing evidence that the flow of nutrients, energy, and material from one ecosystem to another can subsidize populations and, importantly, influence the structure of food webs (1921). Finally, they show that the mechanisms by which predators exert ecosystem-level effects extend beyond both the original conceptual model provided by Hairston et al. and its more recent elaborations (22). Trophic cascades (2324) have traditionally been thought to involve a series of strictly top-down interactions, where predators, by affecting herbivore populations and altering the intensity of herbivory, ultimately influence plant production at the base of food webs (25). Our work illustrates that predators, by thwarting the transport of nutrients between systems, can have powerful indirect effects on systems via a route different from that of classic trophic cascades. The impact of highly mobile predators and their prey on the transport of materials between ecosystems remains poorly understood. Because few ecosystems support food webs that are undisturbed either through introductions or extirpations, it may be that the all-too-common addition or deletion of predators from systems have had substantial but largely unexplored effects.

Supporting Online Material

SOM Text

SOM References and Notes

References and Notes

  1. N. G. Hairston Jr., F. E. Smith, L. B. Slobodkin, Am. Nat. 94, 421 (1960).
  2. J. Halaj, D. H. Wise, Ecology 83, 3141 (2002).
  3. R. T. Paine, J. Mammal81, 637 (2000).
  4. G. A. Polis, A. L. W. Sears, G. R. Huxel, D. R. Strong, J. Maron, Trends Ecol. Evol. 15, 473 (2000).
  5. M. L. Pace, J. J. Cole, S. R. Carpenter, J. F. Kitchell, Trends Ecol. Evol. 14, 483 (1999).
  6. O. J. Schmitz, P. A. Hamback, A. P. Beckerman, Am. Nat. 155, 141 (2000).
  7. M. Williamson, Island Populations (Oxford Univ. Press, Oxford, 1981).
  8. There are several reasons why the Aleutian archipelago provides an ideal large-scale experimental system to study the effects of introduced predators: (i) Most islands are small; (ii) high-latitude floral diversity is relatively low; (iii) there are no native vertebrate herbivores; (iv) the islands are geologically and climatologically homogeneous with similar overall soil properties; (v) the large number of islands in the archipelago provides the opportunity for meaningful replication; (vi) the islands are isolated from anthropogenic nutrient inputs; and (vii) fox introductions were not targeted for particular island types, and the history of introductions is reasonably well known.
  9. N. R. Council, The Bering Sea Ecosystem (National Academy Press, Washington, DC, 1996).
  10. F. S. Chapin III, G. R. Shaver, A. E. Giblin, Ecology 76, 694 (1995).
  11. G. R. Shaver, F. S. Chapin III, Arct. Alp. Res. 18, 261 (1986).
  12. Materials and methods are available as supporting material on Science Online.
  13. E. P. Bailey, U.S. Fish Wildl. Serv. Resour. Publ. 193, 1 (1993).
  14. G. V. Byrd, J. L. Trapp, C. F. Zeillemaker, Trans. 59th N. Am. Wildl. Nat. Resour. Conf., 317 (1994).
  15. T. E. Dawson, S. Mambelli, A. H. Plamboeck, P. H. Templer, K. P. Tu, Annu. Rev. Ecol. Syst. 33, 507(2002).
  16. D. E. Schneider, S. C. Lubetkin, in Food Webs at the Landscape Level, G. A. Polis, M. E. Power, G. R. Huxel, Eds. (Univ. Chicago Press, Chicago, 2004), pp. 12–24.
  17. P. D. Erskine et al.Oecologia 117, 187 (1998).
  18. J. L. Maron et al., data not shown.
  19. G. A. Polis, S. D. Hurd, Am. Nat. 147, 396 (1996).
  20. G. A. Polis, M. E. Power, G. R. Huxel, Food Webs at the Landscape Level (Univ. of Chicago Press, Chicago,2004).
  21. J. F. Kitchell et al.Limnol. Oceanogr44, 828 (1999).
  22. S. D. Fretwell, Oikos 50, 291 (1987).
  23. S. R. Carpenter, J. F. Kitchell, Bull. Ecol. Soc. Am. 74, 186 (1993).
  24. R. T. Paine, J. Anim. Ecol. 49, 667 (1980).
  25. J. Terborgh et al.Science 294, 1923 (2001).
  26. U.S. Fish and Wildlife Service, “Beringian Seabird Colony Catalog Computer Database” (U.S. Fish and Wildlife Service, Homer, AK, 2004).
  27. We thank S. Buckelew, S. Elmendorf, and the Fox/Seabird field teams for field assistance; K. Bell and crew of the M/V Tiglax for ship support; and S. Talbot, J. Williams, and the Alaska Maritime National Wildlife Refuge for advice and logistical assistance. R. Ostfelt, J. Kitchell, T. Martin, D. Pearson, R. Callaway, and R. Holt provided important comments on the manuscript. Supported by NSF OPP-9985814 (J.A.E. and D.A.C.) and NSF OPP-0296208 (J.L.M.).