Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
†Present address and correspondence: Y. Yom-Tov, Department of Zoology, Tel Aviv University, Tel Aviv 69978, Israel. E-mail: email@example.com
1It was recently shown that body size of Palearctic shrews decreases with increasing latitude, thus contradicting Bergmann's rule, and this trend was explained by food shortage during the cold northern winter. In Alaska, global warming has resulted in milder winters that may improve food supply. In this study we tested the hypothesis that body size of Alaskan shrews increased during the second half of the twentieth century, in response to global warming.
2Data on body weight and length of body, tail, hind foot and ear of museum specimens of the masked shrew Sorex cinereus Kerr from Alaska were used in order to examine the effects of latitude, longitude, mean ambient temperature in January and July, and year of collection, on these parameters.
3We found that variation in body size of the masked shrew in Alaska appears to contradict to the prediction of Bergmann's rule, decreasing in high latitudes and in areas cold January temperature.
4Body size of shrews in Alaska increased significantly during the second half of the twentieth century, apparently due to the higher food availability in winter as a result of improved weather conditions for its prey.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Bergmann's rule is probably the best-known rule in zoogeography. It states that, ‘In warm blooded animals, races from warm regions are smaller than races from cold regions’ (Mayr 1970). Bergmann's rule was interpreted as an adaptation to ambient temperature: the larger body surface areas relative to volume of the smaller races serve as efficient heat dissipaters in warm climates, while small body surface area relative to volume may help in heat conservation in cold climates. Shrews are among the smallest extant mammals, and there are conflicting reports on trends in body size among populations and species. Jones & Findlay (1954) and Jones & Glass (1960) showed trends conforming to Bergmann's rule in North American Blarina Gray & Huggins & Kennedy (1989) have shown that in North America smaller body size was characteristic of Sorex cinereus Kerr in southern latitudes, thus conforming to Bergmann's rule. On the other hand, Kirkland & van Deusen (1979) reported that tundra forms of North American Sorex are smaller than those from forest habitats. Trends opposite to those predicted by Bergmann's rule have been reported in Palearctic shrews. Mezhzherin (1964) found that Sorex from colder regions were smaller than those from warmer ones, and Ochocinska & Taylor (2003) reported that in three of five species of Sorex examined in the Palearctic region, condylobasal length of the skull (CBL, a reliable indicator of body size) was negatively correlated with latitude, and the same trend, although not statistically significant, was also found in the fourth species. Ochocinska & Taylor (2003) suggested that food scarcity in winter is a major factor selecting for smaller body size in shrews; these authors also showed that longitude has a significant positive effect on skull size of Sorex araneus Linnaeus and a negative effect on skull size of S. isodon Turov. These effects were explained as a result of character displacement.
Predation may also affect body size, as smaller body size may have an advantage because it enables easier access to burrows and cover. Accordingly, gigantism on islands was attributed to lack of predation (Brown 1995). On the other hand, high food availability may increase body size, and a recent increase in body size among commensal carnivores in Israel was attributed to this factor (Yom-Tov 2003).
Global mean surface temperatures increased on average by 0·6 °C from the late nineteenth century to 1994 (Houghton et al. 2001). In high latitudes, the increase was greater than average, and minimum temperatures increased at about twice the rate of maximum temperatures (Houghton et al. 2001). In north-western America, and particularly in Alaska, average warming since 1950 has been about 2 °C and almost twice as much in the interior in winter, and the growing season has lengthened by more than 14 days (Parker et al. 1994; http://www.usgcrp.gov/usgcrp/Library/nationalassessment/overview). Global warming may affect the physiology, distribution, phenology and adaptation of plants and animals (for a review, see Hughes 2000). Morphological changes due to global warming have also been reported. For example, decreases in body weight in one rodent and four species of passerines during the second half of the twentieth century have been attributed to global warming (Smith, Browning & Shepherd 1998; Yom-Tov 2001). Gosler (2002) has shown that the rate of afternoon fat reserve accumulation in the great tit Parus major Linnaeus has declined during the last 20 years in correspondence with elevated temperatures. Although none of these studies represents a controlled experiment, these trends are explained most parsimoniously by a correlation with recent climate change (Hughes 2000).
The masked shrew S. cinereus is a widespread species that inhabits most of northern North America, including much of Alaska. In this state it occurs in a variety of habitats, including areas where minimum winter temperature reaches −20 °C. In this study we examined the hypothesis that body size of the Alaskan masked shrews increased during the second half of the twentieth century, due to global warming, which produced milder winters and thus improved food supply. Consequently, we examined whether body size of the masked shrew in Alaska contradicts Bergmann's rule, as reported for its Palearctic congeners.
For this study we used data on body weight as well as the length of body, tail, hind foot and ear of 650 specimens of S. cinereus in the University of Alaska Museum of the North (available on-line in April 2004 at http://arctos.database.museum). These specimens were collected between 1950 and 2003; 88% of them were collected between May and August Their body weight was taken to the nearest 0·1 g, and other body parameters were measured to the nearest millimetre.
The specimens of our sample were collected by 51 recorded collectors and an unknown number of others. However, two individuals collected 324 (222 + 102) specimens (50% of total) that comprise between 53% and 62% of measurements, depending on parameter. The other 49 collectors collected between 1 and 39 specimens each (mean = 6·0 9·3 SD).
Continental Alaska is a vast state (about 1·5 million km2) stretching between 54·5°N−71·5°N and 130°W−168°W, comprising Arctic tundra in the north, conifer and boreal forest further south and south-east and rainforest along the west Pacific coast. North of the Arctic Circle, areas experience constant daylight in the summer, but remain in darkness for much of the winter months. As a result, the north–south gradient in temperatures is large. For example, mean monthly temperatures of the coldest (January) and warmest (July) month in the extreme south of Alaska at Ketchikan (55°20′N 131°38′W) are 0·9 °C and 14·3 °C, respectively, and on the coast of the Arctic Ocean at Barrow (71°18′N 156°43′W) they are −25·4°C and 4·7°C, respectively (The Alaska Climate Research Center: http://climate.gi.alaska.edu).
With approximately 10 600 km of coastline, a significant portion of Alaska is influenced by ocean waters and the seasonal distribution of sea ice. Locations that are under the predominant influence of the sea are characterized by relatively small seasonal temperature variability. Conversely, locations that are inland and cut off from the moderating influence of the waters experience a continental climate. This type of climate is characterized by large daily and annual temperature range. For example, mean monthly temperatures of the coldest (January) and warmest (July) month on the Pacific coast at Nome (64°32′N 165°28′W) are −14·6 °C and 11·4 °C, respectively, and near the Canadian border at Eagle (64°49′N 141°09′W) they are −24·1 °C and 16·0 °C, respectively.
In high latitudes soricine shrews are born in late spring and summer and breed in the following summer, after which most die and rarely enter a second winter (Churchfield 1990). Juveniles attain almost adult size by the time they leave their nests at the age of about 3 weeks, and linear body dimensions of weaned shrews as well as skull size do not differ from those of adults (Pucek 1970; Ochocinska & Taylor 2003). However, body mass of weaned shrews is smaller than that of adults, and goes through considerable changes during the year: it alters little throughout the summer, greatly decreases during autumn and winter due to decreases in various parts of the body (Dehnel effect), and then increases considerably in spring as the animals put on weight and achieve sexual maturity (Pucek 1970).
Climate data (mean minimum and maximum temperatures for January and July) were extracted from isotherm maps for the period 1931–52 (Ruffner 1978). The isotherms in these maps are generally represented in 2 °F (sometimes 4 °F) intervals for January and July, respectively. From these data we calculated mean monthly temperatures in °C for each 1 degree square for which we had specimens.
In order to account for annual climate change in Alaska during the study period (1950–2003) we used character means for the year of collection corrected for mean annual temperature at a site (Talkeetna 62·3°N 150·1°W) that is close to the distribution centre of our data (62·5°N and 148·5°W; range 55·8–68·7°N, 130·7–166·2°W). Temperature data were downloaded from http://www.giss.nasa.gov/data/update/gistemp/station_data
When testing Bergmann's rule, latitude is often used as a proxy for ambient temperature (see Ashton, Tracy & de Queiroz 2000; Meiri & Dayan 2003). We used multiple regressions to test the effect of month and year of collection, latitude and longitude on body weight and length of body, tail, hind foot and ear. As body parameters fluctuated during the year we accounted for the monthly variation in all of them by using the sinusoidal component Sin(2 * π * I/12), where I = month, and May (the month with highest values for all parameters but ear) = 1, June = 2, etc. The sinusoidal component was found to fit well the monthly effect. Across North America latitude and longitude are significantly correlated with temperature (see Results), and we used a second set of regressions where latitude and longitude were replaced by mean monthly temperatures in January and July.
All specimens were sorted to localities, and the number of specimens collected at each locality varied between 1 and 121. In order to avoid the problem of pseudoreplication we did not use individual specimens but rather locality means or yearly means corrected for the various factors that may affect body character size.
In all parameters there was no significant difference between females and males (weight: t228,169 = 0·202, P = 0·8397; body length: t253,181 = −0·617, P = 0·5375; tail length: t259,185 = −0·680, P = 0·4970; hind foot length: t253,178 = 0·627, P = 0·3144; ear length: t192,115 = 1·316, P = 0·1891). Hence, in further analyses all specimens were treated as one group, thus increasing sample size considerably.
The coefficient of variation (CV) varied between body parameters (27·5%, 8·2%, 12·2%, 7·8% and 24·5% for body weight and the length of body, tail, hind foot and ear, respectively). The relatively large CV of body weight is mostly due to its being influenced by many factors, including stomach contents, body and reproductive conditions and others, such as rain that may soak the fur. The large CV of ear length is probably a result of measurement accuracy: mean ear length was 6·7 mm, but measurements were taken at 1 mm accuracy.
In order to test for differences in method of measurement between the two main collectors, we conducted unpaired t-test on the residuals of five body parameters (after correcting for latitude, longitude, year and month of collection and mean January temperature). There were no significant differences between the residuals of weight and the length of body, tail and hind foot (P = 0·0778, 0·6224, 0·6142 and 0·6587 for body weight, and the length of the body, tail and hind foot, respectively), but there was a significant difference in ear length (P < 0·0001). This is not surprising, as this parameter is the shortest of all (mean 6·7 mm), and because measurements were taken to 1 mm accuracy, it was most likely to differ between collectors. Hence, in further analysis we did not use ear length.
Latitude was significantly related to mean monthly temperature in January (January temperature = 140·46 – 2·46 * Latitude, F1,73 = 587·80, P < 0·0001, R2 = 0·690) but not to July temperature (F1,73 = 0·459, P = 0·5000). Longitude was significantly related to mean January temperature (January temperature = 72·69 – 0·588 * Longitude, F1,73 = 33·522, P < 0·0001, R2 = 0·315), and to July temperature (July temperature = 23·50 – 0·069 * Longitude, F1,73 = 8·979, P = 0·0037, R2 = 0·110). Thus, in Alaska inland and northern regions were colder in winter than coastal and southern ones, while in July coastal regions were colder.
Latitude was negatively and significantly related to body weight and the length of body, tail and hind foot, while longitude had the opposite effect on body and hind foot length but not on the other parameters (Table 1).
Table 1. The effects of latitude and longitude on body weight, and the length of body, tail, and hind foot (controlled for month [Sin(2 * π * I/12 where I = month, and May = 1, June = 2, etc.] and year of collection) of masked shrews. Coefficients are marked c. Significant results are in bold. Body parameters are locality means
Hind foot length
Latitude c. (P)
Longitude c. (P)
In the second set of regressions that examined the effect of monthly temperatures of the various localities on body parameters (controlled for month and year of collection), mean January temperature was positively related to body weight and the length of the body, tail and hind foot (Table 2, Fig. 1). However, mean July temperature was not significantly related to any of the body parameters (body weight: F1,51 = 0·07, P = 0·7918; body length: F1,68 = 0·380, P = 0·5398; tail length: F1,65 = 0·072, P = 0·7894; hind foot length: F1,64 = 0·270, P = 0·6049).
Table 2. The effect of mean January temperature (T) on body weight, and the length of body, tail, and hind foot (corrected for year of collection and month [Sin(2 * π * I/12) where I = month, and May = 1, June = 2, etc.]) of masked shrews. Coefficients are marked c. Significant results are in bold. Body parameters are locality means
Hind foot length
January T. c
It is also interesting to note that there was a highly significant relationship between the proportion of variation explained by latitude and longitude for each parameter (controlled for month of collection; Table 1) and the proportion of variation explained by January temperatures (Table 3; R2 = 0·964, P = 0·0182).
Table 3. The effects of year of collection (corrected for mean annual temperature of that year in Talkeetna) on body weight and the length of body, tail and hind foot corrected for latitude, longitude and month of collection [Sin(2 * π * I/12) where I = month, and May = 1, June = 2, etc.] of masked shrews. Coefficients are marked c. Significant results are in bold
Hind foot length
We examined the effect of year of collection (corrected for mean annual temperature in the year of collection in Talkeetna) on body size by relating it to body parameters corrected for latitude, longitude and month of collection. Year of collection was significantly and positively related to the length of the body and tail, but not to body weight, or the length of the hind foot (corrected for latitude, longitude and month of collection; Table 3, Fig. 2).
Soricine shrews inhabiting high latitudes are born in late spring and summer, reach adult size by the time they leave their nests and complete their life cycle within a year (Churchfield 1990). They are active throughout the year and in Alaska they are exposed to extreme weather conditions. Hence, one would expect that their body size would be greatly influenced by ambient temperature and food availability.
Our results show that body size of the masked shrew in Alaska declined with latitude and increased with January temperature, thus behaving opposite to the prediction of Bergmann's rule. These effects were strongest in body parameters whose coefficient of variation was smallest. The almost perfect (R2 = 0·964) relationship between the proportion of variation explained by latitude and longitude for each parameter and the same proportions explained by January temperatures, may indicate that the factor determining body size is winter temperature. Hence, this shrew reacts to ambient temperatures similarly to other Sorex species in the northern Palearctic region (Mezhzherin 1964; Ochocinska & Taylor 2003). These results contradict those of Huggins & Kennedy (1989), who found that in southern latitudes body size of S. cinereus conforms to Bergmann's rule, and support those of Kirkland & van Deusen (1979), who found that tundra forms of this species are smaller than those from forest habitats. Further support for our conclusions comes from the finding that all body parameters showed an opposite trend to longitude, increasing from east to west. This is explained by the trend for colder winter ambient temperatures inland and milder ones close to the southern and western Alaskan coasts.
The length of the body and tail were positively correlated to year of collection, indicating that body size of S. cinereus increased during the second half of the twentieth century. These results too support the observation that Alaskan shrews behave in a way that contradicts Bergmann's rule. Mean ambient temperatures in Alaska have increased by about 2 °C during the last 50 years, and almost twice this in the interior in winter, precipitation has increased by 10% and the growing season has lengthened by more than 14 days (http://www.usgcrp.gov/usgcrp/Library/nationalassessment/overview). It seems conceivable to assume that these changes have resulted in higher food availability in winter, enabling the shrews to increase body size accordingly. A similar increase in body size in the large Japanese field mouse Apodemus speciosus Temminck during the second half of the twentieth century was attributed to elevated ambient minimal temperatures, which increased food availability and thus conserved energy (Yom-Tov & Yom-Tov 2004).
Mezhzherin (1964) suggested that the decrease in body size of shrews in winter (Dehnel's effect) as well as the positive relationship between body size and latitude are a strategy to reduce food requirements. Shrews consume small invertebrates that are sensitive to cold and whose availability decreases in cold winters. Mezhzherin (1964) also suggested that the northern limit of shrews is determined by the January isotherms. Our finding that January temperature has a strong effect on body size while July temperature does not have any effect supports this hypothesis. The positive relationships between body size and January temperature, which in turn may reduce food availability in the cold winters, may result in a consequential decline in body size. Shrews are very common in the cool and damp forests of North America, apparently due to the abundant, diverse and dependable food resources found in the form of soil invertebrates (Churchfield 1990), but too cold temperatures may reduce these resources.
In Alaska, global climate change is manifested in higher temperatures (particularly in winter during which temperatures have increased by an average of 2 °C; Parker et al. 1994), higher precipitation and extended growing seasons. These conditions have enabled both masked shrews (this study) and Japanese field mice (Yom-Tov & Yom-Tov 2004) to show opposite trends to those predicted by Bergmann's rule, and to increase their body size during the second half of the twentieth century.
In conclusion, we found that changes in the body size of the masked shrew in Alaska appear contrary to the prediction of Bergmann's rule, decreasing in high latitudes and in areas of cold January temperatures. We suggest that these shrews are affected by lower food availability in winter. During the second half of the twentieth century overall body size increased, apparently due to increased ambient temperatures that resulted in higher food availability in winter. However, we would like to emphasize that our data are correlational and as such our interpretation has to be taken with caution.
We are grateful to Gordon Jarrell, Link Olson and Kevin Winker (University of Alaska Museum of the North) for their encouragement and help in downloading the data, and Brian Hartmann (NOAA-CIRES Climate Diagnostic Center) for helping in obtaining temperature data. This study would not have been possible without the efforts of many (at least 51) collectors who deposited shrew specimens at the museum. We are very grateful to them all and especially to Timothy O. Osborne and Stephen O. MacDonald who collected about half of the shrews included in this study. We thank Sara Churchfield, Robin McCleery and Dorota Ochocinska for advice, Jon Wright and two anonymous referees for useful comments and Naomi Paz for editing the manuscript. Y.Y.T. thanks Nick Davies for his warm hospitality during his sabbatical in Cambridge. This work was partially financed by the Israel Cohen Chair for Environmental Zoology to Y.Y.T.