Niche characteristics explain the reciprocal invasion success of stream salmonids in different continents
- Kai Korsu*,†,
- Ari Huusko‡, and
- Timo Muotka*,§
- *Department of Biology, University of Oulu, P.O. Box 3000, 90014, Oulu, Finland;
- ‡Finnish Game and Fisheries Research Institute, Kainuu Fisheries Research, Manamasalontie 90, 88300, Paltamo, Finland; and
- §Research Program for Biodiversity, Finnish Environment Institute, P.O. Box 413, 90014, Oulu, Finland
-
Edited by Ray Hilborn, University of Washington, Seattle, WA, and accepted by the Editorial Board April 12, 2007 (received for review December 5, 2006)
Abstract
An ability to understand and predict invasions is elemental for controlling the detrimental effects of introduced organisms on native biota. In eastern North America, European brown trout generally dominates over, and eventually replaces, the native brook trout. We show here that in northern Europe the pattern of replacement between these two species is reversed: when transferred to North European streams, brook trout spread extensively and partially replaced the native brown trout. The effect of brook trout on brown trout was habitat-specific: brook trout excluded the native species only in small tributary streams where the reproduction of brown trout was severely reduced, whereas in larger streams brown trout was largely unaffected. Thus, the pattern of coexistence among the two salmonids in our study area is approaching that typically observed in North American streams. In both areas, brook trout ultimately settles in small headwater streams, but the process of replacement differs profoundly: in North Europe, brook trout replaces brown trout in headwater streams, whereas in North America these same streams are the ultimate refuge area for brook trout under the invasion pressure by brown trout. Our results underline the importance of knowing species' niche characteristics to explain and predict biological invasions.
Invasions by alien organisms pose a global threat to biodiversity, ecosystem functioning, and even human health (1). Species transportation beyond their natural ranges has resulted in multiple, and often unpredictable, changes to recipient ecosystems. Thus, an improved ability to predict invasions is one of the key challenges to invasion biology and, indeed, to ecology, in the near future (2, 3). Progress has already been made by, for example, identifying biological traits characteristic of successful invaders and of communities vulnerable to invasion (4). Recently, niche modeling has been applied to predict the risk of invasion by alien organisms in relation to biological and environmental resistance of the host ecosystem (2).
Most studies on species invasions have been reactive by nature; i.e., hypotheses are proposed to explain invasions that have already taken place. For example, the enemy release hypothesis suggests that alien species may be universally superior in their new environments because they have left old enemies behind and the new ones might not detect the invader as a potential threat (or resource) (1). Hence, alien species may have access to surplus resources, resulting in superior ecological performance. However, invasions can be highly context-dependent, making the outcome of a single invasion event unpredictable, at least without firm knowledge of the ecology of the species and the environment considered (5).
Brown trout (Salmo trutta L.) is a European salmonid fish that has been introduced to all major continents, often to the detriment of the native fish fauna (6, 7). In eastern North America, brown trout generally dominates over, and eventually replaces, the native brook trout (Salvelinus fontinalis Mitchill) (7–10). Brown trout is also a dominant competitor in New Zealand and Japanese streams (11, 12), having far-reaching effects on lotic food webs (11), and it has been listed as one of the 100 worst alien species in the world by the Invasive Species Specialist Group. In contrast, there is no rigorous documentation of the impact of brook trout on native salmonids in European streams, although the species has been introduced to a number of streams throughout the continent (13, 14). Based on experience from North America, one might predict that brook trout should be unable to establish populations if the competitively dominant native species, brown trout, is present. Therefore, fisheries managers have assumed that brook trout does not pose a serious threat to brown trout or other native fish in European streams (13). Here we show that biotic interactions among stream salmonids can be strongly context-dependent and that a species inferior in its native range can become an effective invader when transferred to areas where the presumably superior competitor is the native species, provided that the characteristics of the new environment are favorable for the invader.
We collected our data from the headwaters and the main channel of the Upper Kemijoki River system, northeastern Finland [see supporting information (SI) Fig. 4]. This stream system supports abundant populations of brown trout. Brook trout was introduced in the system in the 1970s and 1980s in multiple, nondocumented events, but introductions ceased by the mid 1980s. We sampled fish in first-to-fourth order streams twice, in 1994 and 2004. We assessed fish densities using electroshocking and measured several habitat variables at each sampling site to examine whether brook trout had spread during the intervening years and whether the invasion success was habitat-mediated.
Results
By 1994, brook trout had established mainly in small, slightly acid tributary streams in the middle-reaches of the drainage system, whereas by 2004 it had invaded further toward the headwaters (Fig. 1; see also SI Fig. 4). This upstream expansion was very rapid, because the limit of brook trout distribution had shifted ≈20 km in 10 years, markedly reducing the allopatric refuge area of brown trout. The tributary stream populations probably acted as high-density (up to 60 individuals per 100 m2) (Fig. 2 A) sources for the recent expansion, which was targeted toward habitats formerly occupied by brown trout only (Fig. 1). By 2004, brook trout had colonized 20 new sites in headwater sections of the study system (Table 1 and SI Fig. 4), leaving only six of 30 small (<7 m wide) stream sites uninvaded by 2004. Of these six sites, four were in remote headwaters outside the current distribution of brook trout.
Distribution of the study sites along the first two principal components of the stream habitat data. Variation explained (%) and the highest loadings of the original variables are given for each axis. ●, brook trout allopatric; ■, sympatric; □, brook trout invasion; ○, brown trout allopatric. Site categories differed in PC1 scores (ANOVA, F 3,58 = 17.26, P < 0.001), with brook trout allopatric and sympatric sites having lower scores than other categories (Tukey's test, P < 0.001). No differences were detected for PC2 (F 3,58 = 1.05, P = 0.38).
Total trout densities (means ± 1 SE) (A) and brown trout/brook trout density ratios (100 = brown trout only, 0 = brook trout only; means ± 1 SE) (B) for the two study years. Total trout densities remained constant across years (repeated-measures ANOVA: F 1,58 = 0.22, P = 0.64), whereas sites in different categories supported different fish densities (F 3,58 = 9.25, P < 0.001), with allopatric brook trout sites having the highest and allopatric brown trout sites having the lowest densities (Tukey's test, P < 0.01). The interaction term (year × site category) was not significant (F 3,58 = 0.58, P = 0.63). Brook trout had increased its abundance compared with brown trout in sympatric sites (paired t test: t 10 = −2.99, P = 0.01).
Site categorization based on the presence or absence of the two trout species in both study years
The relative density of brook trout compared with brown trout had increased in both sympatric and newly invaded sites while total trout density (brook trout plus brown trout) remained constant (Fig. 2). Natural reproduction of brown trout, as indicated by the presence of age-0 fish, was severely reduced in the presence of brook trout, but only in small streams (Fig. 3). Of the other native fish species, only bullhead (Cottus gobio L.) provided sufficient data for a meaningful analysis. Bullhead densities remained constant across study years (repeated-measures ANOVA: main effect of time, F 1,39 = 1.80, P = 0.188), whereas sites lacking brook trout in both years supported higher densities of bullhead than those invaded by brook trout by 2004 (main effect of site category, F 1,39 = 6.98, P = 0.012). Importantly, however, the interaction term (time × site category) was insignificant (F 1,39 = 1.30, P = 0.261), indicating that brook trout invasion did not directly modify bullhead densities.
Proportion of study sites (%) in small (<7 m wide, A) and large (>7 m, B) streams that supported age-0 brown trout. ■, brook trout present in both years (small streams, n = 11; large streams, n = 5); □, brook trout invasion sites (n = 5 and 15, respectively); ×, allopatric brown trout sites (n = 6 and 20).
Interestingly, most (75%) of the newly invaded sites were in relatively large streams (see Table 1), but it is doubtful whether these sites will meet the habitat requirements of brook trout in the long term. In fact, all newly invaded sites lacking age-0 brook trout were main-channel sites (mean stream width 16.7 m). Moreover, some main-channel populations had already disappeared (n = 7, mean stream width 18.8 m) between the study years. Thus, occupation of the main-channel sites may represent a transient phase in the ongoing invasion process that is directed toward the upmost headwaters.
Discussion
Our data demonstrate that a subordinate native species can become a harmful invader when transferred to other systems outside its native range. These data also show that the outcome of biotic interactions can be highly context-dependent, and, when trying to understand the mechanisms of a successful invasion, or to identify potential invaders a priori, one should have ample information about the invaded habitat, recipient biota, and niche characteristics of the introduced species.
Successful invaders often have novel ecological traits that make them superior competitors in the new environment (15). For example, brown trout has replaced native galaxids in New Zealand streams because invertebrate prey are much more vulnerable to this evolutionarily novel predator than to native fish (11). However, the trout species studied by us are both visually hunting, drift-feeding fish, and replacing brown trout with brook trout is unlikely to modify profoundly the predation regime of stream invertebrates. Obviously, however, we cannot exclude the possibility that subtle differences in microhabitat selection and foraging tactics (16) might render the two species sufficiently different, conferring a competitive edge for brook trout in North European streams (and, conversely, for brown trout in North American streams).
The fish fauna of North European streams is highly impoverished, as demonstrated by our study sites that supported on average only three fish species (range 1–7). To this end, our data might be interpreted as supporting the conventional notion that systems with low species richness are particularly susceptible to invasions. However, freshwater fish communities do not seem to obey this rule, because there are numerous examples of introduced fish becoming established in diverse assemblages (5). More likely, it appears that brook trout, being an acid-tolerant headwater specialist (17–19), has colonized niche space only marginally used by any salmonid fish in North European streams. Accordingly, we found brook trout mainly in small tributary streams characterized by harsh and variable environmental conditions (20), whereas brown trout was prevalent in larger, more benign downstream sites (see also ref. 19).
In eastern North America, where brown trout has spread extensively, headwater streams often serve as a refuge for brook trout (17, 21). In our study area, these upmost headwaters are now inhabited by brown trout. However, considering the direction and rate of brook trout expansion, it is likely that brook trout will soon invade these streams, eventually replacing the native species, which is less adapted to these extreme conditions. Indeed, this is exactly what has already happened in our sympatric and recently invaded tributary streams, where the natural reproduction of brown trout is reduced to negligible levels. Thus, at the river-wide scale, the spatial pattern of coexistence among the two salmonid species is approaching that typically observed in North American streams (17, 21). In both continents, brook trout ultimately settle in small headwater streams, but the process of replacement differs profoundly: here brook trout replace brown trout in headwater streams, whereas in North America these same streams are the ultimate refuge area for brook trout under the invasion pressure by brown trout. Thus, comparing the species' fundamental niches provides the basis for understanding the seemingly inconsistent pattern of species replacement during invasion on each continent. Importantly, this result implies that if the niche characteristics of an introduced organism are well known, its invasion potential in an area can be predicted and management actions can be targeted accordingly (2).
The reciprocal pattern of invasion demonstrated here underlines the vital importance of autecological information in both native and nonnative ranges for explaining and predicting invasions (see ref. 22). For the widely studied and commercially important salmonid fishes such information is largely available, but there are numerous other instances where even the key ecological traits and niche requirements of the species involved are insufficiently known, making invasions often look entirely unpredictable (23). A combination of carefully designed experiments and long-term monitoring of both the invading and the native species are therefore needed to clarify the population-level mechanisms underlying the idiosyncratic and sometimes even paradoxical patterns of invasion (24, 25).
Genetic stocks of brown trout are considered threatened in many parts of Europe (26, 27). Therefore, the few naturally reproducing populations that still remain in headwater streams have high adaptive and evolutionary significance and, therefore, high conservation value for the species whose populations in lowland rivers are affected by extensive aquaculture releases (28). Ironically, headwaters are the very environment where brown trout is most likely to be replaced by its alien competitor, brook trout. Furthermore, headwater streams comprise a unique biodiversity resource, supporting a number of freshwater taxa absent from larger rivers (29, 30). They have a long history of human exploitation and are therefore considered seriously threatened all over the world (31, 32). Although habitat degradation is the primary cause for changes to headwater ecosystems, invasions by exotic species may add to further loss of biodiversity in these endangered habitats. We therefore strongly urge that any further stocking of brook trout and other nonnative salmonids to North European streams be strictly regulated or, rather, ceased altogether.
Materials and Methods
Study Area.
The study area comprises the upper parts of the River Kemijoki drainage system, northeastern Finland (between 67°15′N and 68°00′N; 28°00′E and 29°10′E) (see SI Fig. 4). Although the lower parts of the river hold several dams, only one is located within our study area, blocking the upstream migration into one of the 69 study sites. The virtual absence of lakes in the drainage system results in high snowmelt-induced floods in May. The main stream is meso-to-oligotrophic [mean and range of total phosphorus in 1973–2004: 15 (2–57) μg·liter−1] and circumneutral (mean pH 6.9, range 6.0–7.8) (Archives of the Regional Environmental Centre of Lapland, Finland). Smaller tributaries tend to be more acid with pH minimum <5.
The study sites are mainly low-gradient (mean slope 0.7 m·km−1) streams, with a pool-riffle sequence characteristic of unmodified forest streams. The sites are scattered across the Upper Kemijoki River drainage system, comprising variable stream habitats from wide, open channels to narrow, heavily shaded headwater streams (SI Table 2). Although most of the study sites are historically impacted by forestry, they have remained practically untouched for at least 40 years and are therefore nearly pristine. Salmonid populations in the study area are mainly resident, and fishing pressure is negligible. Therefore, at present, the only potential human-induced threat to stream integrity in the system is the introduction of the alien fish species brook trout.
Brown trout is the only native trout species in the study system, comprising 18% of total fish density. Other native species, in order of abundance, are bullhead, European minnow (Phoxinus phoxinus L.), Alpine bullhead (Cottus poecilopus Heckel), brook lamprey (Lampetra planeri Bloch), burbot (Lota lota L.), nine-spined stickleback (Pungitus pungitus L.), and European grayling (Thymallus thymallus L.). The mean number of fish species per study site was three in both years (range 1–7).
More than 1.5 million mainly age-0 brook trout were introduced in the middle and southern parts of the study area between 1972 and 1978 (Archives of the Finnish Game and Fisheries Research Institute). Stocking ceased in 1983, and the only fish stocked ever since has been brown trout (annual mean of 14,000 individuals, age 1–5 years). Brook trout now constitutes 15% of the total fish density in the drainage system, and its average density is comparable to that of brown trout.
Sampling Protocol.
The first electrofishing survey was conducted by the Finnish Game and Fisheries Research Institute in 1993 and 1994 (referred here as the 1994 sampling) during the late-summer low-flow conditions (33). In August 2004 we revisited the same sites (n = 69 in 32 streams, mean sample area 289 m2) and collected fish using exactly the same methods as in the previous survey. Fish densities were estimated by the removal method (34, 35). All captured fish were measured, identified, and returned to the stream. Age-0 (young-of-the-year) fish were separated from the older ones based on scale analysis and length-frequency histograms. We consider the presence of age-0 fish to indicate local reproduction (36). Altogether 762 brook trout and 1,270 brown trout were sampled during the two study years.
In 2004 we measured several habitat variables at each site. Measurements were made along randomly placed cross-sectional transects covering the whole study section. The number of measurements varied between 18 and 25, depending on stream width. At each site we assessed the percent cover of instream vegetation (mainly bryophytes), canopy shading (percentage of cover), and dominant substrate size (modified Wentworth scale) (see SI Table 2), and we measured current velocity (at 0.6 × depth with Schiltknecht MiniAir 20), water depth, and stream width. We also measured pH and conductivity (μS·cm−1) for each site using a portable recorder (pH/Cond 340i; Wissenschaftlich-Technische Werkstätten, Weilheim, Germany).
Data Analysis.
We divided the sampling sites into four categories based on the presence or absence of the two trout species (allopatric brook trout sites, sympatric sites, brook trout invasion sites, and allopatric brown trout sites) (see Table 1). To demonstrate possible species replacement, we calculated brown trout/brook trout density ratios (percentages) for each site, with 100 indicating complete domination by brown trout and 0 indicating complete domination by brook trout. Absolute densities were not used because they were highly variable (0.004–0.79 individuals per m2), potentially masking any system-wide effects of brook trout on brown trout. To compare trout densities between time periods and site categories we used two-way repeated measures ANOVA. Seven sites where brook trout occurred in 1994 but not in 2004 were omitted from the analysis. We used a paired t test to test for density ratio changes in sympatric assemblages. To examine the effect of brook trout on the reproduction of brown trout, we calculated the number of sites where age-0 brown trout were found, separately for small (<7 m) and large (>7 m) streams in each of three site categories: (i) sites where brook trout were present in both years, (ii) invaded sites (brook trout present in 2004 but not 1994), and (iii) allopatric brown trout sites (brook trout absent in both years).
Principal component analysis (PCA) was used to examine fish–habitat relations. Environmental variables were log- or arcsin-transformed, if needed, to better meet the assumptions of PCA. The analysis produced four components with eigenvalues >1.0 (SI Table 2). We then performed one-way ANOVA followed by Tukey's pairwise comparisons for each component to examine whether site categories, represented by site scores in PCA, differed in stream habitat structure. We performed all statistical analyses with SPSS for Windows version 12.0.1 (SPSS, Chicago, IL).
Acknowledgments
We are indebted to Pekka Korhonen for collecting the 1994 data set. Tapio Rautiainen and Olli van der Meer led the groups of highly motivated field workers, which made possible the collection of the 2004 data set within a reasonable time period. We also appreciate comments by two anonymous referees on a previous version of the manuscript. Financial support was provided by the Maj and Tor Nessling Foundation and the Academy of Finland.
Footnotes
- †To whom correspondence should be addressed. E-mail: kai.korsu{at}oulu.fi
-
Author contributions: K.K., A.H., and T.M. designed research; K.K. performed research; K.K. analyzed data; and K.K., A.H., and T.M. wrote the paper.
-
The authors declare no conflict of interest.
-
This article is a PNAS Direct Submission. R.H. is a guest editor invited by the Editorial Board.
-
This article contains supporting information online at www.pnas.org/cgi/content/full/0610719104/DC1.
- © 2007 by The National Academy of Sciences of the USA
