The evidence for Allee effects
Allee effects are an important dynamic phenomenon believed to be manifested in several population processes, notably extinction and invasion. Though widely cited in these contexts, the evidence for their strength and prevalence has not been critically evaluated. We review results from 91 studies on Allee effects in natural animal populations. We focus on empirical signatures that are used or might be used to detect Allee effects, the types of data in which Allee effects are evident, the empirical support for the occurrence of critical densities in natural populations, and differences among taxa both in the presence of Allee effects and primary causal mechanisms. We find that conclusive examples are known from Mollusca, Arthropoda, and Chordata, including three classes of vertebrates, and are most commonly documented to result from mate limitation in invertebrates and from predator–prey interactions in vertebrates. More than half of studies failed to distinguish component and demographic Allee effects in data, although the distinction is crucial to most of the population-level dynamic implications associated with Allee effects (e.g., the existence of an unstable critical density associated with strong Allee effects). Thus, although we find conclusive evidence for Allee effects due to a variety of mechanisms in natural populations of 59 animal species, we also find that existing data addressing the strength and commonness of Allee effects across species and populations is limited; evidence for a critical density for most populations is lacking. We suggest that current studies, mainly observational in nature, should be supplemented by population-scale experiments and approaches connecting component and demographic effects.
KeywordsCritical density Depensation Inverse density dependence Mate limitation Predator satiation Positive density dependence
Although Allee effects were first described in the 1930s (Allee 1931), interest has recently surged in tandem with high profile conservation problems (Stoner and Ray-Culp 2000; Courchamp and Macdonald 2001; Wittmer et al. 2005; Angulo et al. 2007). Allee effects, defined as positive effects of increasing density on fitness, are a predicted consequence of multiple mechanisms in small or low-density populations, and are expected to affect many population-level processes, including extinction (Dennis 1989, 2002; Boukal and Berec 2002; Allen et al. 2005), population establishment and biological invasion (Drake and Lodge 2006; Liebhold and Tobin 2006; Ackleh et al. 2007), metapopulation fluctuations (Zhou et al. 2004; Martcheva and Bolker 2007), predator–prey interactions (Gascoigne and Lipcius 2004; Morozov et al. 2004), and parasite transmission (Deredec and Courchamp 2006). The putative variety of mechanisms and affected ecological processes suggest that Allee effects may be widespread, with pervasive effects in nature.
What is the evidence for Allee effects in natural populations?
What mechanisms cause Allee effects?
Do these mechanisms give rise to hypothesized strong Allee effects, or are most observed Allee effects weak (see below for the distinction between strong and weak Allee effects)?
How common are Allee effects?
Using these distinctions as a guide, in this paper we survey evidence for Allee effects in animals. Focusing on natural populations, we quantify emerging patterns in which populations or species exhibit Allee effects, what causes them, and whether they are strong or weak. We identify links between study design and the propensity of studies to detect demographic Allee effects and critically evaluate the existing evidence for strong Allee effects in natural populations. We also examine whether existing evidence enables assessment of the rarity or ubiquity of Allee effects in natural populations. To provide context, we include a discussion of evidence for Allee effects in natural plant populations. Our results showed both similarities and differences among taxonomic groups in mechanisms causing Allee effects, and suggest that detection of Allee effects and the ability to link component and demographic Allee effects was influenced by study design. Overall, there is substantial and increasing evidence that Allee effects play an important role in numerous, diverse systems. However, we find existing studies provided only limited direct evidence for strong Allee effects. The pervasiveness of Allee effects in natural populations remains unknown.
Published studies on Allee effects were identified as follows. We first conducted a search of the ISI Web of Science database with each of the search terms “Allee”, “depensation”, and “inverse density dependence”, examining all references available through August 2008. We enlarged the sample of studies by searching references of review papers (Stephens and Sutherland 1999; Gascoigne and Lipcius 2004; Berec et al. 2007) and examining other papers known to the authors. These approaches identified 492 studies. Our goal was not to include all studies considering low-density populations, but rather to assemble a thorough collection of studies which focused on the dynamics of low population densities and the role of Allee effects, a collection that we believe is more complete than any previously assembled.
To understand further the role of Allee effects in natural animal populations and their importance relative to the other population processes occurring in nature, we selected studies including data from field populations (manipulated or unmanipulated), and excluded purely laboratory experiments, reviews, and strictly theoretical papers; 91 studies met this criterion. For context, we discuss an additional 28 studies on Allee effects in plant populations. Our search on “inverse density dependence” also revealed a group of studies that focus on parasitoid attack rate as a function of host density. These 24 papers are distinct from the main body of literature on animal populations, because they differ in their focus on parasitoid behavior and their lack of explicit consideration of overall fitness or population dynamics of the host species. We considered these separately.
Each study was classified by study design, type of evidence, the classes of phenomena in which Allee effects were present, and taxonomic group (S1 in Electronic Supplementary Material (ESM)). We defined the study design as either experimental manipulation (i.e., treatment imposed by researchers), natural experiment (i.e., observations following a known, uncontrolled perturbation, such as species introduction), or observational study. We categorized evidence according to whether it addressed component Allee effects, demographic Allee effects or both. Those studies presenting evidence for or against demographic Allee effects were further classified according to the type of phenomenon in which Allee effects were identified: changes in population size and growth rate (“population dynamics”), population persistence (“colonization/extinction”) or rate of spread (note that individual studies occasionally provided more than one type of data).
Studies were categorized into six taxonomic groups based on the study species: terrestrial arthropods, aquatic invertebrates, mammals, birds, fish, and reptiles. No study included more than one taxonomic group, although 16 tested for Allee effects in more than one species. Papers that compared several species were classified on the basis of the overall finding, i.e., if evidence for Allee effects in some subset of populations was conclusive, the study was counted as evidence for an Allee effect in that taxonomic group as a whole. Of special note are seven meta-analyses which examined data on many populations (11–330) of related species (S1, see “Results and discussion”). They were excluded from analyses which considered the number of species rather than the number of studies exhibiting Allee effects, because of the often specialized methods that these studies used to test for Allee effects and the diluting effect that their large number of species would have on the other studies examined. There were also two studies that used retrospective analysis to argue for Allee effects by examining the current distribution (Jordal et al. 2001) or persistence (Munoz-Duran 2002) of beetles and carnivores respectively, and relating it to species characteristics. We considered this indirect evidence to be inconclusive and classified it as such. The remaining papers considered comparable data on only two or three species, each of which was categorized as described below when the analytical unit was species rather than study.
We critically assessed the support for Allee effects in each study according to the following criteria. If the authors presented no evidence for an Allee effect or evidence contrary to an Allee effect, these were labeled as “No Allee effect.” Studies were labeled “Allee effect” when a positive relationship was demonstrated between population density and any component of individual fitness, population growth rate, the probability of persistence or colonization, or spread rate. The remaining papers were labeled “Unclear” because the data did not conclusively demonstrate such a relationship. Many “Unclear” studies were presented by the authors as evidence for an Allee effect, and were suggestive of Allee effects, but relied on circumstantial evidence or contradictory results.
When data supported an Allee effect, the mechanism(s) causing the component or demographic Allee effect, as represented by the authors, were noted. In some cases the mechanism(s) were directly supported by data, in other studies the conclusion was based on expert knowledge. In a few cases the mechanism was not addressed. Although authors used a variety of terminology, all the mechanisms cited fit into eight broad categories (reviewed in Dennis 1989; Courchamp et al. 1999; Stephens and Sutherland 1999; Boukal and Berec 2002, unless otherwise cited): cooperative breeding, cooperative defense, cooperative feeding, dispersal (Kuussaari et al. 1998), habitat alteration, mate limitation (Gascoigne et al. 2009), predator satiation (Gascoigne and Lipcius 2004), and somatic growth (McCormick 2006).
Each study presenting demographic evidence for an Allee effect was examined for strong vs. weak Allee effects. Because a goal of this review was to assess the evidence for strong Allee effects in nature, we were stringent in our assessment. Strong Allee effects were only recognized if authors reported quantitative support for a critical density in the population (as in Fig. 1a, b for example; see “Results and discussion”). Cases where a critical density was not indicated by the data were classified as weak, and the remaining studies, classified as “uncertain”, lacked the data necessary to address the strength of the Allee effect.
Results and discussion
What are the patterns in the empirical evidence for Allee effects in natural populations?
Taxonomic groups and detection of Allee effects
Study design and detection of Allee effects
Number of species where Allee effects were detected, and type of study design
Type of study
Allee effect present?
The type of data collected varied significantly among taxonomic groups (Fig. 5b, Fisher’s exact test, P = 0.02) with aquatic invertebrates and birds having relatively little information on demographic Allee effects, and a substantial majority of studies on other taxonomic groups addressing demographic Allee effects. Many studies of fish and terrestrial arthropods considered demographic data exclusively and, as a result, determination of the mechanisms underlying Allee effects relied on circumstantial evidence or expert knowledge. Studies of both birds and aquatic insects usually lacked empirical evidence for demographic Allee effects, but for different reasons. For aquatic invertebrates most studies were experimental manipulations that rarely provided demographic evidence (see above). For birds, studies that detected Allee effects were primarily observational, documenting components of fitness such as mortality and hatching success, but not population dynamics (Cuthbert 2002). Thus, the lack of evidence for demographic Allee effects probably does not suggest a lower incidence in these taxonomic groups but rather indicates that additional studies are needed that collect the type of data necessary to assess demographic Allee effects in these groups. On the other hand, the substantial number of studies based only on demographic evidence in terrestrial invertebrates and fish was largely because of the use of existing large datasets on long-term dynamics (Myers et al. 1995; Tobin et al. 2007) for detecting Allee effects. Studies on mammals came closest to the ideal balance of understanding both the mechanism and the population dynamic implications, with six studies (46%) assessing both component and demographic Allee effects (Wittmer et al. 2005; Angulo et al. 2007).
What causes Allee effects?
The collected studies presented empirical evidence that Allee effects in natural populations are caused by six mechanisms: mate limitation, cooperative defense, predator satiation, cooperative feeding, dispersal, and habitat alteration. For each, at least one study provided evidence for a component Allee effect and for a demographic Allee effect (S1). The four most commonly identified mechanisms (Fig. 6b) occurred in studies containing direct evidence for both the component and demographic Allee effect. An additional process, somatic growth, caused a component Allee effect in one study (Hoddle 1991), but was not implicated at the demographic scale.
The contribution of these mechanisms to Allee effects differed among taxonomic groups (Fisher’s exact test, P < 0.0001). Two-thirds of detected Allee effects were caused by mate limitation, cooperative defense, or predator satiation (Fig. 6b). Mate limitation was the most common mechanism, occurring mostly in aquatic invertebrates where it is the dominant cause of detected Allee effects (Fig. 6b). This is expected, given the challenges of fertilization in aquatic environments (Gerritsen 1980; Quinn et al. 1993). In general, one expects that species with limited mobility or passive reproduction, for example plankton, plants, and broadcast spawners, are more susceptible to mate limitation (Gascoigne et al. 2009). Mate limitation was less common—or more unexpected and therefore less studied—in vertebrates. Another mechanism used as a classic example of Allee effects, cooperative defense, seems to be important only for vertebrate species. Predator satiation, whereby an Allee effect results from a Type II functional response by a predator (Dennis and Patil 1984; Gascoigne and Lipcius 2004), was also dominant in vertebrates. Further investigation into whether the smaller amount evidence for predator-mediated Allee effects in invertebrates is merely a result of study bias or suggests a real difference between taxonomic groups is important given the ubiquity of predator–prey interactions.
Terrestrial arthropods were subject to the most different mechanisms, but at least four mechanisms were implicated in all taxa. At the same time, each mechanism occurred in multiple taxonomic groups. Dispersal and habitat alteration, the least frequently examined mechanisms, are also less commonly discussed in the theoretical literature (A. Kramer, unpublished data). An Allee effect caused by dispersal occurs when animals are more likely to emigrate from a low-density population than a high density population, or when animals are more likely to colonize already-occupied habitats, thus reducing population growth rate at low abundances (Menendez et al. 2002). Habitat alteration can cause an Allee effect when the presence of conspecifics increases recruitment to the population by altering the biotic or abiotic conditions in a way that increases fitness of recruits (Kent et al. 2003; Gascoigne et al. 2005).
Six studies implicated two mechanisms as contributing to a demographic Allee effect (S1). In four of these studies, component Allee effects were directly documented, with low-density populations subject to reduced fitness from more than one fitness component—for example, reproduction and survival (Angulo et al. 2007). The population-level implications of multiple Allee effects on a single population have recently been explored theoretically (Berec et al. 2007), and these empirical studies confirm the occurrence of multiple Allee effects in natural populations. However, as argued by Berec et al. (2007), existing data do not directly assess the interaction between multiple Allee effects or the population-level consequences. This problem requires studies measuring component and demographic Allee effects simultaneously in addition to experimental manipulations to isolate the influence of each component Allee effect.
Are Allee effects common?
Given the increasing number of studies presenting evidence for Allee effects, it remains to ask two questions:
What proportion of species is subject to Allee effects? and
Do the population densities at which Allee effects are measurable often occur?
Concerning the first question, the broad patterns of occurrence and diversity of relevant mechanisms uncovered imply that no taxonomic group is invulnerable to Allee effects and that Allee effects should be considered potentially important in low-density dynamics of many species.
Unfortunately, because the only evidence available is from studies in which researchers have already recognized the relevance of Allee effects, the rate of detection in these studies is a poor indicator of the relative frequency of Allee effects in nature overall. One approach to assessing the commonness of Allee effects is to look for evidence in a randomly selected set of populations. The seven meta-analyses identified here adopted that approach, using a collection of datasets on the dynamics of populations of multiple species (within a restricted taxonomic group, see “Introduction”). The number of populations examined varied from eleven (Sæther et al. 1996) to 330 (Walters and Kitchell 2001). Of these, five were based on model fitting. One study found no evidence for an Allee effect in fur-bearing seals (Gerber and Hilborn 2001). Three studies on fish (Myers et al. 1995; Liermann and Hilborn 1997; Barrowman et al. 2003) and one on birds (Sæther et al. 1996) found weak support for Allee effects in very few populations, or inconclusive evidence for Allee effects. At the same time, all acknowledged very low power for detecting Allee effects. Again, the difficulty of sampling low-density populations is a factor that limits the detection of Allee effects, even those that are strong (Stephens et al. 2002). The failure of these studies to obtain conclusive evidence for Allee effects could have resulted from uncommonness across populations of these species, a lack of observations at sufficiently low densities, or that very few species are subject to Allee effects at any realistic density. The remaining two reviews found support for Allee effects by very different methods than the model-fitting used in the above studies (Hopper and Roush 1993; Walters and Kitchell 2001). Walters and Kitchell (2001) used visual analysis by experts to identify stock-recruitment patterns consistent with Allee effects in multiple populations and Hopper and Roush (1993) use data on biological control introductions which supported significant limitation of establishment success at small inoculum size across broad groups of parasitoid species. The different approaches and inconclusive or conflicting findings (especially between Myers et al. (1995), Liermann and Hilborn (1997), and Walters and Kitchell (2001)) illustrate the limitations of drawing any general conclusions about the commonness of Allee effects overall from existing meta-analyses.
Weak vs. strong Allee effects
Studies finding quantitative support for strong Allee effects show no pattern when grouped by mechanism or taxonomic group (Fig. 7a, b) and both natural experiment and observational study designs provided a variety of types of evidence (S2 in ESM). Strong Allee effects occurred in multiple taxonomic groups and were caused by mate limitation, predator satiation, and cooperative defense. However, there were some similarities among the seven studies. Three of the studies demonstrated the presence of strong Allee effects in a single species, gypsy moths (Fig. 8b, Liebhold and Bascompte 2003; Johnson et al. 2006; Tobin et al. 2007). In two cases (Johnson et al. 2006; Tobin et al. 2007), the response variable was the replacement rate, or the proportion of populations that replaced themselves. If less than 50% of populations replaced themselves then the density represented a negative growth rate. A critical density is also revealed by a regression of log-population growth rate on population density (Fig. 8b, Liebhold and Bascompte 2003). The large datasets and the presence of an advancing front of an invasive species resulted in numerous measurements at very low density, enabling statistically significant estimates of the critical densities in various forms. Concurring evidence from multiple datasets and analyses reinforces the potential impact of Allee effects on the spread of invasive species. Notably, Tobin et al. (2007) found evidence for a critical density in some regions of the gypsy moth range but not in others. This extensive work on gypsy moths underscores the potential importance of Allee effects in invasive species, and the feasibility of detecting strong Allee effects in natural populations, while highlighting the large amounts of data that will often be necessary to ensure detection.
Two other studies also compared per capita growth rate to population density. While sample sizes were smaller, population densities were sufficiently low to quantify the critical density (Dulvy et al. 2004, Fig. 8c; Wittmer et al. 2005, Fig. 8d). Clutton-Brock et al. (1999, Fig. 8e) provided evidence for a strong Allee effect by estimating the probability of group loss across various group sizes (an indirect measure of population density), showing that it is lower than expected from a stochastic model and that it extends above 50%, an approach that is similar to the replacement rate for gypsy moths discussed above. Finally, Chen et al. (2002, Fig. 8f) used a modified Ricker model, to argue for a critical density in coho salmon, estimated at 5000 adults (95% CI = (1736, 8686)). All of these studies had data at very low densities, at or below the critical density. Without these data, demonstration that such a density exists must be by extrapolation, severely reducing the power of any statistical test to rule out the null hypothesis that there is no positive critical density. But, the requirement of observing populations below their critical densities is rarely achieved, because a single population that falls below this density is likely to decline to extinction, making detection and monitoring of sufficiently low densities difficult. Accordingly, predicting the critical density for a population of conservation concern before it crosses that density is correspondingly difficult (Lundquist and Botsford 2004).
How do studies on plants compare with those on animals?
Throughout, we have primarily focused on animal populations because they have been studied more thoroughly from both theoretical and empirical perspectives. However, Allee effects can be important in plant populations. Reduced fecundity at low density because of pollen limitation has been shown for animal-pollinated species of conservation concern (Lamont et al. 1993) and for a wind-pollinated invasive species (Davis et al. 2004a, b). Another invasive species, Vincetoxicum rossicum (Asclepiadaceae), attains higher fitness at high density by suppressing background vegetation (Cappuccino 2004). There are clear parallels between these mechanisms and those found in animals, such as between pollen limitation and mate limitation. Are the patterns in empirical evidence for Allee effects in plants similar to the patterns found in animals?
Our ISI search found 28 studies with empirical evidence necessary for examining Allee effects in natural plant populations (S3 in ESM). Of these, twenty-three reported evidence of the Allee effect (Fig. 4b), with twenty focusing on pollen limitation as the mechanism, one on habitat alteration, one on a genetic inbreeding mechanism, and one unknown. The predominance of studies on pollen limitation suggests there may be a more limited set of mechanisms leading to Allee effects in plants. However, the authors often purposely focused on self-incompatible plants as they should be likely to experience pollen limitation (Elam et al. 2007), making it difficult to assess the commonness of Allee effects in plants with other reproduction strategies or the commonness of other mechanisms, a situation which is analogous to mate limitation in aquatic invertebrates. In both cases empirical studies confirm the prior expectation of mate limitation because of passive reproductive behavior, though this expectation probably also biases researchers against investigating other possible mechanisms. The smaller body of research suggests that Allee effects in plants, especially when caused by mechanisms other than pollen limitation, may be less common or less extensively investigated than in animals. Future work should confirm this by looking for additional mechanisms paralleling those found in animals (e.g., Allee effects induced by herbivore satiation) and should take advantage of cases where the sessile nature of plants allows for comprehensive population census, documentation of individual fitness, and replicated manipulation.
Inverse density dependence and parasitoid attack rate
Our bibliographic search revealed a substantial body of work documenting “inverse density dependence” in the attack rate of parasitoids on host insect larvae and eggs (S4 in ESM). These studies included both experimental manipulations and observational data and were often motivated by questions about biological control of pest species. Inverse (positive) density dependence in attack rate should result in an Allee effect on host survival, because a higher proportion of hosts are parasitized at lower host density. Interestingly, these papers did not endeavor to assess fitness or population dynamics of the host species, reflecting the more limited focus in this literature than in the broader Allee effects literature. Three papers on parasitoids (Campbell 1976; Hopper and Roush 1993; Fauvergue et al. 2007) were included in our broader analysis because they assessed the influence of parasitoid population density on the establishment probabilities of the parasitoids themselves, which was also absent from this group of papers on attack rate. We note that the rate of publication in this area is independent of the recent interest in Allee effects that took place in the plant and animal literature (Fig. 3c), highlighting the distinctiveness of this body of work. Coordinating terminology and linking research on parasitoid attack rate with that on Allee effects has the potential to contribute to both fields by improving understanding of host–parasitoid dynamics and testing theory against a very important class of interspecific interactions.
The number of studies presenting empirical evidence for or against Allee effects in natural animal populations has increased in the past decade, supporting the following conclusions. First, Allee effects have been detected in diverse animal taxa, Mollusca, Arthropoda, and Chordata, including three classes of vertebrates; amphibians and reptiles are notable exceptions. Second, there is evidence of Allee effects caused by at least six mechanisms: mate limitation, cooperative defense, predator satiation, cooperative feeding, dispersal, and habitat alteration. The first three mechanisms constitute 66% of studies on Allee effects in animals, and while the importance of mate limitation is not surprising, the prominence of predator satiation is notable, given its absence from several well-cited definitions of the Allee effect (Myers et al. 1995; Amarasekare 1998). This variety of species and mechanisms suggests that Allee effects may be common, notwithstanding existing meta-analyses that seem to suggest otherwise (Sæther et al. 1996; Gerber and Hilborn 2001). Alternatively, populations may seldom experience densities low enough to produce Allee effects, so that Allee effects are rare among natural populations although they have the potential to affect most species. Current evidence is unable to broadly address this dichotomy. Furthermore, because few studies have directly linked component and demographic Allee effects, and even fewer have demonstrated the existence of critical densities in natural populations, the fact that important theoretical models and assumptions rely primarily on a handful of well-studied examples may be under-appreciated.
Our analysis of the evidence for Allee effects across animal populations has highlighted knowledge gaps and points to a need for future studies to definitively address the commonness of Allee effects and critical densities. We found that observational studies and natural experiments are generally able to assess demographic Allee effects, but experimental manipulations have primarily focused solely on component Allee effects. We suggest two types of studies have the greatest potential to advance our understanding of the dynamics of low-density populations. First, strong tests of theory and improved understanding of the importance of critical densities in natural populations will need to rely on field-based experimental manipulations that are able to assess population dynamics or rates of colonization and extinctions. Such experiments should allow for larger amounts of data (via replication) from very low population densities and a comparison of the relative strength of positive and negative density-dependent components of fitness. Suitable experiments would also increase our understanding of the interaction among multiple Allee effects. Second, the seven studies successfully detecting critical densities in natural populations depended on long-term/large-scale observational studies and well-chosen analytical methods. While measuring low-density populations is inherently challenging, these studies prove that it can be done effectively, particularly by taking advantage of population response to perturbations such as invasion or harvest. This combination of experiments, increased focus on effective ways to measure very low densities, and approaches connecting component and demographic effects presents the most promise for figuring out the frequency and importance of Allee effects in natural populations.
We thank Blaine Griffen, Andrea Silletti, Philip Stephens, David Boukal, and three anonymous reviewers for suggestions on an earlier version of this manuscript. Funding for this project was provided by the University of Georgia and a grant from the Great Lakes Protection Fund.
- Allee WC (1931) Animal aggregations. University of Chicago Press, ChicagoGoogle Scholar
- Clark CW (1976) Mathematical bioeconomics: the optimal management of renewable resources. Wiley, New YorkGoogle Scholar
- Dennis B (1989) Allee effects: population growth, critical density, and the chance of extinction. Nat Resour Model 3:481–538Google Scholar
- Gascoigne J, Berec L, Gregory S, Courchamp F (2009) Dangerously few liaisons: a review of mate-finding Allee effects. Popul Ecol. doi:10.1007/s10144-009-0146-4
- Hoddle MS (1991) Lifetable construction for the gorse seed weevil, Apion ulicis (Forster) (Coleoptera, Apionidae) before gorse pod dehiscence, and life-history strategies of the weevil. NZ J Zool 18:399–404Google Scholar
- Munoz-Duran J (2002) Correlates of speciation and extinction rates in the Carnivora. Evol Ecol Res 4:963–991Google Scholar
- Quinn JF, Wing SR, Botsford LW (1993) Harvest refugia in marine invertebrate fisheries: models and applications to the red sea urchin Strongylocentrotus franciscanus. Am Zool 33:537–550Google Scholar