General hypotheses in invasion ecology

Every scientific discipline needs concepts and hypotheses – these are the backbones of science. Their development over time also reflects the growth or decline, respectively, of scientific disciplines. Here, I focus on general hypotheses in invasion ecology, a rapidly growing discipline (Richardson & Ricciardi, 2013). The publication of the book The ecology of invasions by animals and plants by Charles Elton (1958) is usually seen as the start of modern invasion ecology. Stressing modern is important here, as earlier researchers also investigated this topic, for example Darwin (1859; Cadotte, 2006). Kowarik & Pyšek (2012) offer insights into the history of invasion ecology with a focus on the Swiss botanist Albert Thellung (1881–1928) who took the first steps towards unifying concepts and hypotheses in invasion ecology. Fifty years after the publication of Elton's book, a special issue of Diversity and Distributions was dedicated to Elton's legacy. An editorial in this special issue outlined how influential Elton and his work remains today (Richardson & Pyšek, 2008). Elton's 1958 book is the most frequently cited source in the field, and several of invasion ecology's current major hypotheses can be found there, for example the biotic resistance hypothesis (Richardson & Pyšek, 2007). In the following, I will focus on this hypothesis and three other influential invasion hypotheses, and studies that have tested them. These hypotheses were arbitrarily selected (see Catford et al., 2009; Lamarque et al., 2011; Jeschke et al., 2012a,b; Lockwood et al., 2013; Lowry et al., 2013; and references therein for other influential invasion hypotheses). I give a definition (but be aware that different framings exist for each hypothesis; Heger et al., 2013) and summarize two selected articles published in Diversity and Distributions on each hypothesis (these articles appear in bold). The hypotheses are ordered chronologically, from oldest to youngest. I will also outline an approach to synthesize invasion hypotheses, and end with the question Where to from here? The biotic resistance hypothesis, which I use synonymously with the diversity-invasibility hypothesis here, posits that ecosystems with high biodiversity are more resistant to invaders than ecosystems with low biodiversity (Elton, 1958; Levine & D'Antonio, 1999; Lonsdale, 1999; Jeschke et al., 2012a,b). It is a classic hypothesis of the field that is, like most other invasion hypotheses, strongly connected with general ecological hypotheses and concepts (Box 1). The resistance hypothesis has been empirically tested in multiple ways. The most common test is to compare native species richness with alien species richness, where the latter is used as an inverse measure of biotic resistance: high alien species richness indicates low resistance. Although this approach does not control for the number of introduced species, it was used by 57 of 129 tests (44%) of the hypothesis identified by Jeschke et al. (2012a). The resistance hypothesis predicts that native and alien species richness are negatively correlated, but a number of studies published at the turn of the century and thereafter reported the opposite pattern. A positive correlation was, for example, reported by McKinney (2002) for alien plants in 77 protected areas in the USA. Findings like these have led to the formulation of the rich-get-richer pattern and the acceptance hypothesis, positing that the relationship between native and alien species richness is typically positive, particularly at large spatial scales (Stohlgren et al., 2003, 2006). In some cases, complex patterns can be found. For example, Espinosa-García et al. (2004) reported for plants in Mexico a negative relationship between alien and native weed species richness (in line with the resistance hypothesis and contrary to the acceptance hypothesis); on the other hand, they reported a positive relationship between alien weed species richness and native non-weed species richness (in line with the acceptance hypothesis and contrary to the resistance hypothesis). In general, most tests of the resistance hypothesis focused on terrestrial plants and reported low empirical support (Jeschke et al., 2012a,b). The latter is particularly true for large spatial scales, as studies conducted at small scales have more frequently reported supporting evidence (Fridley et al., 2007). It is sometimes argued that invasion ecology is isola-ted from the rest of ecology (e.g. Davis et al., 2001). Although stronger connections between researchers working on biological invasions and those focusing on other ecological topics would certainly be beneficial, a look at the concepts and hypotheses in invasion ecology reveals that these are deeply connected with those in general ecology. Let us look at the biotic resistance hypothesis (see also main text) as an example. This hypothesis is a special formulation of a more general ecological hypothesis, namely the diversity-stability hypothesis which states that ecosystems with high biodiversity are more stable than ecosystems with low biodiversity (Ives & Carpenter, 2007; Jeschke et al., 2012a, 2013). In the biotic resistance hypothesis (treated synonymously here with the diversity-invasibility hypothesis), stability is replaced by resistance against invaders. The resistance hypothesis is also related to the concept of the ecological niche, one of ecology's most fundamental ideas. The ecological niche has been defined differently by different authors, as outlined by Pulliam (2000), p. 351: ‘Hutchinson (1957) used the word niche to refer to the environmental requirements of a species, whereas earlier authors, especially Elton (1927) and Grinnell (1917), had used the term niche to refer to a place or ‘recess’ in the environment that has the potential to support a species … According to Hutchinson, species, not environments, have niches.’ And for Elton, environments have niches. These can be vacant, and the probability that a niche is vacant is higher if there are fewer species, genera, life-forms etc. in an ecosystem as compared to an ecosystem with many species, genera, life-forms etc. where most niches are filled. Consequently, ecosystems with high biodiversity should offer fewer niche opportunities for invaders (i.e. they should be more resistant against invaders) than ecosystems with low biodiversity. This brief treatment of the biotic resistance hypothesis shows its strong connection to general ecology. As a consequence, studies testing this and many other invasion hypotheses can reveal insights that are not only relevant for invasion ecology but for ecology in general and other disciplines as well (Sax et al., 2005, 2007). Taking this perspective, invasion ecology is not isolated at all but well connected with other (sub)disciplines. Another influential invasion hypothesis is the tens rule which posits that approximately 10% of species successfully take consecutive steps of the invasion process: about 10% of species transported beyond their native range will be released or escape in the wild (they are called introduced species or casuals); about 10% of these introduced species will be able to establish themselves in the wild (they are called naturalized species); and about 10% of species established will become invasive/pest species (Williamson & Brown, 1986; Williamson, 1996; Jeschke et al., 2012a,b). This rule from the late 1980s became popular in the 1990s which is when total numbers of publications on biological invasions started to rise (Richardson & Pyšek, 2008). The tens rule is of high practical importance, for example for cost-benefit models (Keller et al., 2007) and has also been applied to genetically modified organisms (Regal, 1993; Williamson, 1993). Caley et al. (2008) outlined a Bayesian approach for estimating the proportion of introduced species that will establish themselves. Their approach considers the important issue that if simply taking the ratio of established to introduced species for testing the tens rule, one is assuming that all introduced species have had enough time to actually establish themselves. This may not be the case (see also Kowarik, 1995; Jeschke & Strayer, 2005). They applied their approach to data from woody plants introduced to South Australia and estimated the probability of naturalization to be 18.6%, with the 95% confidence interval ranging from 15.5 to 23.4%. While the confidence interval does not overlap with the 10% predicted by the tens rule, the 18.6% estimate is inside the 5–20% interval that was suggested by Williamson (1996) for comparison of the tens rule with empirical data (see also Jeschke et al., 2012a). Křivánek & Pyšek (2006) also looked at woody species, but in this case those introduced to the Czech Republic. Of 180 species commonly planted there for at least 60 years and well adapted to the conditions of the target region, 57 (31.7%) have taken the first step of the invasion process (i.e. they have become casuals at least), 28 (49.1% of 57 species) have naturalized and 17 (60.7% of 28 species) have become invasive. In their Discussion section, Křivánek & Pyšek (2006) also gave figures for the larger pool of woody plants introduced to the Czech Republic (not restricted to species commonly planted there for at least 60 years and well adapted to the conditions of the target region). In this larger sample, out of 127 species that have taken the first step of the invasion process, 54 (42.5%) have naturalized and 11 (20.4% of 54 species) have become pests. These figures are higher than the 10% predicted by the tens rule and are outside the 5–20% range. Most studies in the field of invasion ecology have focused on plants, but most studies on the tens rule have actually focused on animals, particularly vertebrates (Jeschke et al., 2012a,b). The underlying reason is most likely that reliable data for testing the rule are mainly available for vertebrates, particularly birds and mammals. Yet, empirical support for the rule is particularly low for vertebrates which often show much higher transition rates than predicted by the tens rule: they are typically around 50% rather than 10%, although the data are highly variable (Jeschke & Strayer, 2005; Jeschke, 2008). Empirical support for the rule is significantly higher for plants and invertebrates than for vertebrates, but even for plants and invertebrates studies often report results that are beyond the 5–20% range (Jeschke et al., 2012a). According to the enemy release hypothesis, the absence of enemies in the exotic range is a cause of invasion success (Maron & Vilà, 2001; Keane & Crawley, 2002; Jeschke et al., 2012a,b; Heger & Jeschke, 2014). Although Albert Thellung already implicitly used the enemy release hypothesis in 1915 (Kowarik & Pyšek, 2012), the hypothesis has only truly ‘come alive’ in the last two decades. It is now very influential in the field. Also, it is strongly connected with other invasion hypotheses, for instance the EICA hypothesis (evolution of increased competitive ability; Blossey & Nötzold, 1995) or the shifting defence hypothesis (Doorduin & Vrieling, 2011). Like other hypotheses, many different variants and specific formulations of the enemy release hypothesis have been tested. For instance, some studies looked at a potentially reduced infestation of the alien species with enemies in the exotic range, whereas other studies looked at a potentially reduced damage by enemies. An example for the latter is the study by Ebeling et al. (2008) who compared 10 native Chinese and 10 invasive German populations of the shrub Buddleja davidii, finding reduced damage by leaf herbivores in the invasive populations. At the same time, the plants in the invasive populations were taller, had thicker stems, larger inflorescences and heavier seeds than the plants in the native populations. These results are in line with the enemy release hypothesis. Ashton & Lerdau (2008) did not compare native and alien populations of a given species, but compared alien with native species in the same range. They focused on vines in the US, finding no support for the prediction that alien species would suffer less damage from herbivores. What they did find, however, was that alien species are more tolerant of herbivory. These mixed results are similar to other empirical tests of the enemy release hypothesis (Colautti et al., 2004; Liu & Stiling, 2006; Hawkes, 2007; Chun et al., 2010; Jeschke et al., 2012a; Heger & Jeschke, 2014). The propagule pressure hypothesis posits that high propagule pressure is a cause of invasion success (Lockwood et al., 2005). As with other invasion hypotheses, this one has been interpreted differently by different authors; in particular, the term propagule pressure has been defined and measured in various ways. Typically, propagule pressure is synonymously treated with introduction effort and considered to consist of: (1) propagule size, which is the average number of individuals involved in a single release event; and (2) propagule number or frequency, which is the average number of release events per unit time (Lockwood et al., 2005, 2009, 2013; Blackburn et al., 2009; Simberloff, 2009). The propagule pressure hypothesis has been hugely influential in 21st century invasion ecology – Richardson (2004), p. 317, wrote: ‘Understanding propagule pressure is the new frontier in invasion ecology.’ This underscores the growing realization that (a) event-level characteristics such as propagule pressure and (b) human activities (e.g. intentional or unintentional transporting and release of alien species) play a key role for biological invasions. Twentieth century invasion ecology rather focused on general characteristics of invaders or invaded ecosystems (e.g. Drake et al., 1989). Foxcroft et al. (2004) showed that propagule pressure is highly important for understanding the invasion of South Africa's Kruger National Park by the cactus Opuntia stricta. Bacon et al. (2014) combined agricultural trade databases with European quarantine interception records to estimate propagule pressure for quarantine arthropods introduced to Europe. They found that propagule pressure, climate suitability and host availability of arthropods together are strongly related to their establishment patterns. Overall, the propagule pressure hypothesis seems to be quite well supported by empirical evidence (Lockwood et al., 2005, 2013; Colautti et al., 2006; Blackburn et al., 2009; Simberloff, 2009; Lamarque et al., 2011). Although some critical tones have been raised here too (Moulton et al., 2011), there seems to be more consensus on this hypothesis than on most others. The growth of invasion ecology comes with an increasing number of invasion hypotheses. Several authors have offered tools and frameworks for synthesizing these hypotheses. For instance, Catford et al. (2009) listed 29 invasion hypotheses and outlined a framework to connect them. Key parts of this framework are propagule pressure (P), abiotic characteristics (A), biotic characteristics (B) and how humans influence P, A and B. They also note similarities among the 29 hypotheses. Other tools and frameworks for synthesizing invasion hypotheses have been suggested by Gurevitch et al. (2011), Heger et al. (2013) and Saul et al. (2013). After this review of key publications on invasion hypotheses and their synthesis, I now offer some thoughts on (1) how existing invasion hypotheses might be revised or replaced, and (2) how their synthesis can be moved forward. First, many invasion hypotheses are empirically not well supported (see above and Jeschke et al., 2012a; Moles et al., 2012). It thus seems that the field's progress is hampered by holding on to concepts and ideas that simply do not work, at least not under many circumstances (see also Davis, 2011). As many of these ideas are intuitively appealing, invasion ecologists hesitate to revise or replace them. Another reason is that in invasion ecology – as in other ecological subdisciplines – data are rarely crystal clear and can be criticized from many angles, for example laboratory data can be criticized for being unnatural, or observational field data can be criticized for not revealing cause–effect relationships. As every research method can be criticized in one way or the other, a synthesis of studies with different methods can reveal a clearer and more robust picture. A novel approach to synthesize studies on a given major hypothesis is the hierarchy-of-hypotheses (HoH) approach suggested by Jeschke et al. (2012a) and Heger et al. (2013); see also Heger & Jeschke (2014). This approach may serve the field in revealing which hypotheses work under which circumstances, and which hypotheses should be revised or replaced. Second, what I consider a promising path forward for synthesizing multiple invasion hypotheses is to apply network analysis. A network of invasion hypotheses is a powerful visualization, illustrating how the major concepts of the field are connected. A first preliminary example of such a network is shown in Fig. 1. This network includes the 29 invasion hypotheses featured in Catford et al. (2009) and was created based on their Table 2 that gives information on similarities between the hypotheses. Networks could also be constructed based on citation frequencies of key papers on each invasion hypothesis, or based on ‘traits’ of invasion hypotheses. When applying the HoH approach to a network of hypotheses, each hypothesis in the network can be hierarchically divided into more specific subhypotheses. Such a hierarchical network of hypotheses not only represents the combined theory of the field but can be easily connected with empirical studies on biological invasions that have tested any (sub)hypothesis. As invasion hypotheses are strongly connected with general ecological hypotheses (Box 1), it will be straightforward to extend the network of hypotheses and empirical studies testing them, so that the whole field of ecology is covered. I thus believe this approach has much potential to move forward both invasion ecology and general ecology. I warmly thank David Richardson for the invitation to write this editorial. I am also very grateful to Martin Enders for creating Fig. 1, and I highly appreciate comments on the manuscript by him, Tina Heger and David Richardson.