Climate Change and Distribution Shifts in Marine Fishes

See allHide authors and affiliations

Science  24 Jun 2005:
Vol. 308, Issue 5730, pp. 1912-1915
DOI: 10.1126/science.1111322


We show that the distributions of both exploited and nonexploited North Sea fishes have responded markedly to recent increases in sea temperature, with nearly two-thirds of species shifting in mean latitude or depth or both over 25 years. For species with northerly or southerly range margins in the North Sea, half have shown boundary shifts with warming, and all but one shifted northward. Species with shifting distributions have faster life cycles and smaller body sizes than nonshifting species. Further temperature rises are likely to have profound impacts on commercial fisheries through continued shifts in distribution and alterations in community interactions.

Climate change is predicted to drive species ranges toward the poles (1), potentially resulting in widespread extinctions where dispersal capabilities are limited or suitable habitat is unavailable (2). For fishes, climate change may strongly influence distribution and abundance (3, 4) through changes in growth, survival, reproduction, or responses to changes at other trophic levels (5, 6). These changes may have impacts on the nature and value of commercial fisheries. Species-specific responses are likely to vary according to rates of population turnover. Fish species with more rapid turnover of generations may show the most rapid demographic responses to temperature changes, resulting in stronger distributional responses to warming. We tested for large-scale, long-term, climate-related changes in marine fish distributions and examined whether the distributions of species with fast generation times and associated life history characteristics are particularly responsive to temperature changes.

We studied the demersal (bottom-living) fish assemblage in the North Sea. This group is composed of more than 90 species with varied biogeographical origins and distribution patterns. North Sea waters have warmed by an average of 0.6°C between 1962 and 2001, based on four decadal means before 2001, and by 1.05°C from 1977 to 2001 (7), which correspond with our fish survey time series. Survey data were used to calculate catch per unit effort to determine centers of abundance (mean latitudes and depths) for all species and boundary latitudes for those species that have either northerly or southerly range limits in the North Sea (7). No species range was entirely confined to the North Sea. Measures of distribution were regressed against same-year and time-lagged bottom temperatures, and also a composite measure of temperatures, the North Atlantic Oscillation Index, the Gulf Stream Index, and the ratio of abundances of northern and southern calanoid copepod species (7). We also controlled for changes in abundance that may have influenced species distributions (7).

Centers of distribution as measured by mean latitudes shifted in relation to warming for 15 of 36 species (Table 1). These trends were shown by both commercially exploited species [such as Atlantic cod (Gadus morhua) and the common sole (Solea solea)], and by species that are not targeted by fisheries [such as scaldfish (Arnoglossus laterna) and snakeblenny (Lumpenus lampretaeformis)]. Distances moved ranged from 48 to 403 km (average distance = 172.3 ± 98.8 km, n = 15 species) (Fig. 1) and most of these shifts (13 of 15) were northward (Table 1). The spatial temperature gradient of the North Sea is somewhat unusual; water temperatures become colder with increasing latitude in the southern North Sea but become slightly warmer with increasing latitudes in the north (8), where warm North Atlantic Current waters enter the region (9). This temperature pattern may explain one of the two exceptional species that moved south, the Norway pout (Trisopterus esmarkii). Its distribution was centered in the northern North Sea, and its southern movement brought it into cooler waters. The other exception was the common sole. We speculate that the southward shift in its distribution may have been caused by the fact that the cleanup of the Thames estuary led to its emergence as a major sole nursery ground during the study period (10).

Fig. 1.

Examples of North Sea fish distributions that have shifted north with climatic warming. Relationships between mean latitude and 5-year running mean winter bottom temperature for (A) cod, (B) anglerfish, and (C) snake blenny are shown. In (D), ranges of shifts in mean latitude are shown for (A), (B), and (C) within the North Sea. Bars on the map illustrate only shift ranges of mean latitudes, not longitudes. Arrows indicate where shifts have been significant over time, with the direction of movement. Regression details are in Table 1.

Table 1.

Statistically significant multiple regressions of the effects of three measures of North Sea warming on mean latitudes of 36 demersal fishes from 1977 to 2001. PC1, first principal component from principal components analysis (PCA) of eight environmental variables (PC1 generally describes warming). Winter temp. and summer temp. indicate 5-year running mean bottom temperatures for December to March and June to September, respectively.

Species Common name df Mean latitude (°N) SD PC1 r2P Winter temp. r2P Summer temp. r2P
Agonus cataphractus Pogge 22 54.67 0.90
Anarhichus lupus Atlantic wolffish 21 58.14 0.46
Argentina spp. Argentines 24 59.59 0.30
Arnoglossus laterna Scaldfish 15 54.17 0.31 0.456 0.43 0.006
Buglossidium luteum Solenette 23 54.14 0.28
Callionymus lyra Dragonet 23 55.40 0.65 0.265 0.16 0.049 0.937 0.34 0.002
Echiichthys vipera Lesser weever 24 53.30 0.13 0.191 0.39 0.001
Eutrigla gurnardus Grey gurnard 23 56.13 0.35 0.194 0.30 0.006 0.651 0.61 <0.001 0.402 0.17 0.040
Gadiculus argenteus Silvery pout 23 59.83 0.41
Gadus morhua Atlantic cod 23 56.81 0.34 0.256 0.58 <0.001 0.534 0.38View inline <0.001 0.578 0.33View inline <0.001
Glyptocephalus cynoglossus Witch 24 58.22 0.42
Hippoglossoides platessoides Long rough dab 24 57.62 0.21 0.304 0.40 0.001
Lepidorhombus boscii Fourspot megrim 24 60.51 0.37
Leucoraja naevus Cuckoo ray 19 58.06 0.57
Limanda limanda Dab 24 55.86 0.13 0.180 0.35View inline 0.001
Lophius piscatorius Anglerfish 23 57.99 0.58 0.254 0.19 0.032 0.818 0.37 0.001
Lumpenus lampretaeformis Snake blenny 12 56.52 1.15 3.174 0.81 <0.001
Melanogrammus aeglefinus Haddock 24 57.91 0.16
Merlangius merlangus Whiting 23 56.57 0.15 0.066 0.19 0.034
Merluccius merluccius Hake 24 58.84 0.59
Micromesistius poutassou Blue whiting 21 60.13 0.48
Microstomus kitt Lemon sole 24 57.06 0.24
Molva molva Ling 24 59.26 0.74
Myxine glutinosa Hagfish 11 57.51 0.62
Pleuronectes platessa Plaice 24 55.52 0.18
Pollachius virens Saithe 24 59.44 0.20
Psetta maxima Turbot 13 54.73 0.31
Rhinonemus cimbrius Four-bearded rockling 22 56.05 0.68 0.419 0.40 0.001 1.147 0.53 <0.001 0.950 0.28 0.008
Scyliorhinus canicula Small-spotted catshark 20 58.34 0.89
Sebastes spp. Redfish 18 59.89 0.49
Solea solea Common sole 13 53.68 0.66 -0.941 0.38 0.020 -0.963 0.34 0.028
Squalus acanthias Spurdog 19 56.29 0.68
Trigla lucerna Tub gurnard 19 53.89 0.50
Trisopterus esmarkii Norway pout 23 58.59 0.26 -0.190 0.52 <0.001 -0.304 0.25 0.010 -0.429 0.37 0.001
Trisopterus luscus Bib 9 53.29 0.51 0.489View inline 0.45 0.035
Trisopterus minutus Poor cod 23 55.63 0.66 0.334 0.26 0.012 0.877 0.33 0.003 0.753 0.18 0.035
  • View inline* A relationship with annual mean summer or winter temperature.

  • View inline To identify the proportion of variance in distribution accounted for by warming, r2 and P describe the squared semi-partial correlation coefficient, where abundance was also a significant predictor of distribution.

  • Most species that showed climate-related latitudinal changes also shifted in depth, which was unsurprising because North Sea depths are roughly positively correlated with latitude (8). A further six species, including plaice (Pleuronectes platessa) and cuckoo ray (Leucoraja naevus), moved deeper with warming but did not change in latitude, suggesting that they may have responded to climatic variation through local movements offshore or into pockets of deeper water. Considering both latitude and depth, nearly two-thirds of species (n = 21 out of 36) have shown distributional responses to climatic warming (table S1).

    We tested whether species boundaries have also been displaced by warming, by examining those 20 species from our data set with a southern or a northern range limit in the North Sea. The boundaries of half of these fishes moved significantly with warming (Fig. 2 and table S2). Southern boundaries shifted in 6 of 12 cases, and all shifts were northward. Four of eight northern boundaries also moved with warming. All but one of these species shifted north, despite the fact that their northern range limits lay in the relatively intensively fished southern North Sea (11). Shifting species again included both exploited and nonexploited fishes. Boundaries moved over distances ranging from 119 to 816 km ( = 304 ± 196 km, n = 10), with the highest value describing the range of movement of the southern boundary of blue whiting (Micromesistius poutassou), which is the target of the largest fishery in the Atlantic (12). In the case of bib (Trisopterus luscus), the northern boundary shifted by 342 km from 1978 to 2001, a trend that is supported by observations that North Sea catches of this species have been increasing (13).

    Fig. 2.

    Frequency distributions of fish species shift rates in relation to warming and time. (A) Rates of shift for northerly species' (southern) boundaries with climate. (B) Southerly species' (northern) boundaries with climate. (C) All species' mean latitudes with climate. (D) Northerly species'(southern) boundaries over time. (E) Southerly species' (northern) boundaries over time. (F) All species' mean latitudes over time. Rates for shifting species are slopes from regressions.

    To identify shifts that may have been driven by fishing or other nonclimatic influences, we also examined distribution changes over time. Fishing pressure could not be included explicitly in our analyses because reliable fishing effort data on a comparable spatial and temporal scale do not exist for the North Sea. However, during at least the last decade of the 25-year period of analysis, the spatial distribution of effort remained relatively constant (11), and total fishing effort may have declined slightly (14). Temporal trends in distribution suggested that fishing alone could not explain climate-related shifts; despite the general increase in temperature over the study period, warming-related shifts occurred independently of time for centers of distribution in 8 of 36 species and for range limits in 4 of 20 species (table S3). Such shifts may have reflected year-to-year environmental variability, with northward movement during warm years cancelled by southward movement during cool years. If so, long-term distribution shifts could depend strongly on future climatic variability, in addition to longer-term average conditions.

    The examination of temporal trends also allowed for rough comparisons to be drawn with rates of warming-related distribution shifts in other taxa. A recent meta-analysis of climate-change impacts on natural systems estimated the mean annual rate of boundary movement for 99 species of birds, butterflies, and alpine herbs at 0.6 km northward or 0.6 m upward (1). From the current study, the mean rate of movement for the six fish species whose boundaries shifted in relation to both climate and time [bib, blue whiting, lesser weever (Echiichthys vipera), Norway pout, scald-fish, and witch (Glyptocephalus cynoglossus)] was 2.2 km per year. It is perhaps unsurprising that the rate of shift might be higher for marine fishes than for alpine herbs and butterflies, given that marine fish may generally face fewer constraints on movement. However, if such a difference is indicative of more widespread trends in marine fishes, climate change could pose a greater threat to fish populations that are constrained by their dispersal capabilities or habitat requirements.

    If the differences in rates of movement among the taxa documented here result from differential rates of population turnover, we would expect species with life history traits associated with fast population growth to have responded most strongly to climate change. To test this prediction, we compared life history traits between shifting and nonshifting species (7). As predicted, shifting species tend to have faster life histories than do nonshifting species, with significantly smaller body sizes, faster maturation, and smaller sizes at maturity (Fig. 3). Body growth rates did not differ significantly between shifting and non-shifting species (P = 0.19). These relationships therefore provide a starting point for predicting species' responses to future climate change. These predictions could be refined, through detailed studies of the relative sensitivities of different life history stages, to uncover the specific mechanisms driving the patterns.

    Fig. 3.

    Differences in life-history traits between shifting (n = 15) and nonshifting (n = 21) species with respect to centers of distribution (mean latitudes). (A) Maximum body size [t = –2.41, degrees of freedom (df) = 34, P = 0.02]. (B) Age at maturity (t = –2.86, df = 27, P = 0.01). (C) Length at maturity (t = –2.29, df = 29, P = 0.03). Means are shown with standard errors.

    Our study shows that climate change is having detectable impacts on marine fish distributions, and observed rates of boundary movement with warming indicate that future distribution shifts could be pronounced. Mean annual surface temperatures in the North Sea are predicted to increase by 0.5 to 1.0°C by 2020, 1.0 to 2.5°C by 2050, and 1.5 to 4.0°C by 2080 (15). We used the midpoints of these temperature ranges as the basis for a rough approximation, which suggested that two types of commercial fishes, blue whiting and redfishes (Sebastes spp.), may retract completely from the North Sea by 2050, and by 2080, bib may extend its range northward to encompass the entire region. Such changes will clearly also depend on the responses of their predators and prey to increases in bottom temperature and on the availability of suitable habitat.

    These findings may have important impacts on fisheries. For example, species with slower life histories are already more vulnerable to overexploitation (1618) and may also be less able to compensate for warming through rapid demographic responses. A further concern is that differential rates of shift could result in altered spatial overlap among species, thereby disrupting interactions and also potentially compounding the decoupling effects of climate-driven changes in phenology (19). Previous work off the eastern United States has shown that fishes with the most temperature-sensitive distributions included key prey species of nonshifting predators (20). Such changes could have unpredictable effects in an ecosystem already under heavy anthropogenic pressure.

    Supporting Online Material

    Materials and Methods

    Tables S1 to S4


    References and Notes

    Stay Connected to Science

    Navigate This Article