Effects of climate on oxygen consumption and energy intake of chromosomally divergent populations of the House Mouse (Mus musculus domesticus) from the island of Madeira (North Atlantic, Portugal)


  • M. L. MATHIAS,

    Corresponding author
    1. Centro de Biologia Ambiental, Departamento de Biologia Animal, Faculdade de Ciências, Universidade de Lisboa, Edifício C 2, 3° Piso, Campo Grande, 1749–016 Lisboa, Portugal,
    Search for more papers by this author
  • A. C. NUNES,

    1. Centro de Biologia Ambiental, Departamento de Biologia Animal, Faculdade de Ciências, Universidade de Lisboa, Edifício C 2, 3° Piso, Campo Grande, 1749–016 Lisboa, Portugal,
    Search for more papers by this author
  • C. C. MARQUES,

    1. Centro de Biologia Ambiental, Departamento de Biologia Animal, Faculdade de Ciências, Universidade de Lisboa, Edifício C 2, 3° Piso, Campo Grande, 1749–016 Lisboa, Portugal,
    Search for more papers by this author
  • J.-C. AUFFRAY,

    1. Laboratoire Génétique et Environnement, Institut des Sciences de l’Evolution, Université Montpellier II, 34095 Montpellier Cedex 5, France,
    Search for more papers by this author

    1. Laboratoire Génétique et Environnement, Institut des Sciences de l’Evolution, Université Montpellier II, 34095 Montpellier Cedex 5, France,
    Search for more papers by this author
  • G. GANEM,

    1. Laboratoire Génétique et Environnement, Institut des Sciences de l’Evolution, Université Montpellier II, 34095 Montpellier Cedex 5, France,
    Search for more papers by this author
  • I. GÜNDÜZ,

    1. Department of Biology, University of York, York YO10 5YW, UK,
    Search for more papers by this author

    1. Centro de Biologia Ambiental, Museu Bocage, Universidade de Lisboa, Rua da Escola Politécnica, 1268–102 Lisboa, Portugal
    Search for more papers by this author
  • J. B. SEARLE,

    1. Department of Biology, University of York, York YO10 5YW, UK,
    Search for more papers by this author

    1. School of Biological Sciences, University of Aberdeen, Aberdeen AB24 2TZ, UK and Rowett Research Institute, Bucksburn, Aberdeen AB21 9SB, UK
    Search for more papers by this author

†Author to whom correspondence should be addressed. E-mail: mlmathias@fc.ul.pt


  • 1We explored the effects of climatic variables (mean altitude, mean daily winter temperature, mean daily summer temperature, mean annual precipitation and days of precipitation per year) on energetic parameters (food intake and resting oxygen consumption) in six Robertsonian chromosomal races and hybrid populations of House Mice (Mus musculus domesticus) collected throughout the island of Madeira in the North Atlantic.
  • 2Food energy intake and resting metabolism (oxygen consumption) were measured, in 59 non-reproducing adult males trapped in April–September 1998 and June 1999 and maintained in captivity for at least 3 weeks prior to measurements.
  • 3Mean daily energy intake of Robertsonian mice varied between 25·3 kJ day−1 in race S. Vicente (2n = 25–27) and 34·6 kJ day−1 in race Achadas da Cruz (2n = 24–27), while in hybrids (2n = 22 × 2n = 40) it was 23·0 kJ day−1. All races exhibited low resting rates of oxygen consumption compared with the predicted basal metabolic rates expected for Muridae, between 49·2% and 66·5% of the expected values.
  • 4The main factor influencing both food energy intake and oxygen consumption was body mass, explaining 41% of the variation in food energy intake and 39% of the variation in resting oxygen consumption. Body mass was significantly related to the mean summer temperature at the sites where the mice were captured, but was unaffected by the chromosomal race or other biogeographical and climatic variables. There were no additional effects of these factors on resting oxygen consumption beyond the effect of body mass.
  • 5Once the effects of body mass were removed, food energy intake was significantly correlated with the chromosomal race. The different environmental conditions experienced by the races may have favoured the physiological adaptation of mice to different habitats.


The standard karyotype of the House Mouse (Mus musculus domesticus Rutty 1772) consists of 40 acrocentric chromosomes, but fixation of chromosomal rearrangements, Robertsonian (Rb) fusions, has led to the formation of many different chromosomal races in Europe and North Africa characterized by a reduction of the ancestral karyotype (e.g. Gropp et al. 1972; Said et al. 1986; Belkhir & Britton-Davidian 1991; Nachman & Searle 1995). Recently, six Rb races, with diploid numbers varying from 2n = 22 to 2n = 30, have been described on the island of Madeira (Mathias & Ramalhinho 1992; Britton-Davidian et al. 2000) (Fig. 1). This is the main island of the Madeira archipelago (30°−33° N/16°−17° W), which also includes the islands of Porto Santo and Desertas, where mice have the standard karyotype. The six Rb races are characterized by five to nine pairs of Rb fusions, involving all chromosomes, except autosome pair 1 and the sex chromosomes (Table 1). Out of the 20 Rb fusions identified in Madeiran mice, 13 have not previously been recorded, and chromosome 19 was found involved in Rb fusions, in the wild for the first time (Nachman & Searle 1995; Britton-Davidian et al. 2000; Gazave et al. 2003).

Figure 1.

Sample sites of Rb races of the house mouse in the island of Madeira (32·75° N, 16·97° W) (see Table 1 for races designation).

Table 1.  Robertsonian (Rb) fusions characterizing Madeiran Rb races of the House Mouse. Chromosomes occurring in an acrocentric state are not mentioned
Rb racesRb fusions
Santana (2n = 22) XX  XX X  X  X X  X
Lugar de Baixo (2n = 24)X  X    X X XX  X  X
Estreito da Calheta (2n = 24–26)X  X   XX X XX  X   
Achadas da Cruz (2n = 24–27)X  X   X XX XX  X   
S. Vicente (2n = 25–27)X X    XX   XX XX   
Ponta Delgada (2n = 28–30)  X X        XX  XX 

The island of Madeira is the most diverse of the archipelago, regarding climate and landscape (e.g. Ribeiro 1985; Pena and Cabral 1997). The extreme topography of the island, in which the most central part is occupied by high mountains separated by deep valleys, defines a clear north–south differentiation in climate parameters. The north coast is more exposed to the wind, less sunny, the average annual precipitation is higher and the average annual maximum air temperatures are lower than in the south, where the precipitation is reduced and temperatures are higher (Ribeiro 1985; Pena & Cabral 1997). Climate heterogeneity is also associated with altitude, as a 3 °C decrease in temperature is estimated for each 500 m increase in altitude (Ribeiro 1985). The whole island is humid and rainy (3200 mm year−1 of maximum precipitation, on average), except in the south coast where climate can be considered semiarid.

Mouse populations on Madeira are highly dependent on the activities of people although they can also live away from human settlements (Mathias 1993). Thus, although energy spent searching for food could be reduced in anthropogenic habitats, energy for maintenance can be very different among populations owing to the characteristics of the occupied areas. Besides, at a microhabitat level, the time spent inside buildings can also be different and of a major importance in regulating energy expenditure, as previously pointed out by Górecki et al. (1990) in other populations of House Mice.

The present study focuses on the Robertsonian system of Madeira. Data indicate that chromosomal differentiation evolved in Madeira most likely favoured by the island's particular topography leading to geographical isolation within valleys. In addition, the chromosomal differences observed between races suggest that they are at least partially if not totally reproductively isolated, setting the stage for the accumulation of further divergence. Considering that the Madeiran Rb races are exposed to different environmental conditions, selection may have favoured different physiological responses in these mice. The main objective was thus to evaluate the effect of climatic factors on the energy metabolism (food energy intake and oxygen consumption) in the chromosomally divergent populations and their distribution.

Materials and methods

sample identification and maintenance of animals

Considering that sex, age and social status of random-trapped wild individuals may represent a source of possible bias in assuming sampled individuals as representative of populations (e.g. Richardson, Dohm & Garland 1994), only non-reproducing adult males, obtained during three different trapping periods (April–September 1998 and June 1999) were analysed. The House Mice used in this study were captured in 17 different sites in Madeira, selected according to the geographical distribution of the six Rb races previously described in Britton-Davidian et al. (2000). Because we sampled few individuals at each site and time the possibility that individuals sampled within each race or locality were closely related is very small.

All 59 mice analysed were associated with human-influenced habitats, i.e. farms, cultivated fields and fallow fields. Collecting sites were characterized by the following environmental variables: mean altitude, summer and winter mean temperatures, mean annual precipitation and days of precipitation per year (Water Regional Planning, Madeira) (Table 2). Experimental mice were transported to Lisbon in mainland Portugal where they were maintained in individual cages, in a quiet room, at natural ambient temperature averaging 24 °C (22–26 °C) and natural photoperiod. Commercial food and water were provided ad libitum. Mice were allowed to adjust to the captive diet and ambient conditions for at least 3 weeks prior to experiments. By keeping the animals in a constant environment during this period we hoped to eliminate any immediate effects of the climate and season on their energy metabolism, thus revealing effects of long-term climatic adaptation (e.g. McDevitt and Speakman 1994).

Table 2.  Total number of mice (N) analysed from each sample and race in feeding and metabolism experiments, date of capture (1 = April 1998, 2 = September 1998, 3 = June 1999) and environmental variables characterizing collecting sites
Chromosomal groups identificationCollecting site/date of captureEnvironmental variablesFeeding exp.Metabolism exp.
Mean altitude (m)Mean summer temp. (°C)Mean winter temp. (°C)Mean annual precipitation (mm)No. days precipitationSample (N)Race (N)Sample (N)Race (N)
Santana (2n = 22)Camacha (3)70019·511·51500125216210
Terreiro da Luta (1,3)68021·512·515001256 4 
Referta (1)25018·510·511001352 1 
Santana (1)47218·511·515001656 4 
Lugar de Baixo (2n = 24)Lugar de Baixo (2) 3622·515·5 900 755 55 5
Estreito da Calheta (2n = 24–26)Prazeres (2)65018·512·517001152 82 5
Estreito da Calheta (2)38018·511·511001053 3 
Chão da Ribeira (1)55017·511·519001353  
Achadas da Cruz (2n = 24–27)Sitío do Lombo (2)58017·511·515001152 61 5
Paul do Mar (2) 3018·511·517001154 4 
S. Vicente (2n = 25–27)S. Vicente (1,3)15018·510·522501356 76 7
Sitío do Rosário (3)32216·5 8·519001251 1 
Ponta Delgada (2n = 28–30)Ponta Delgada (1)12020·514·511001554 4
Hybrids (2n = 22 × 2n = 40)Funchal (1,3) 5022·514·5 900 953133 9
Palheiro (3)25620·514·5 700 953 3 
S. Martinho (1)15020·513·511001092  
S. Roque (1,3)26020·513·511001105 3 

After completing the experiments, the diploid number of each individual was checked before assigning the mice to a chromosomal race for data analysis (Table 2). Chromosomes were prepared from bone marrow cells of femurs and tibias, following the air-drying technique (Lee & Elder 1980); the G-banding method was used for chromosomal arm identification according to the nomenclature of Cowell (1984). Consequently, the mice trapped were grouped into seven chromosomal groups, corresponding to the six Madeiran Rb races plus the wild hybrids between race 2n = 22 (race Santana) and standard mice present in Funchal, the main city in the southern coast, where the main commercial port is located. Races were named as in Ramalhinho et al. (2005).

feeding experiments

Feeding experiments lasted 5 consecutive days and commenced at least 3 weeks after animals had been captured. At the beginning of experiments, mice were weighed (to an accuracy of 0·1 g) and received food ad libitum. Every day, at the same hour, mice were reweighed, uneaten food was collected and weighed (to an accuracy of 0·1 mg) and fresh food was again supplied in excess. Dry mass of uneaten food was obtained after weight had stabilized in an oven at 60 °C and then samples were combusted in a PARR 1425 semimicro bomb calorimeter (Parr instrument company, Moline, Illinois) for calculation of food energy contents. These procedures allowed the estimation of the energy obtained from food intake –I (kJ day−1) = dry mass of food ingested (g day−1) × energy content of the food (kJ g−1).


Approximately 1 month after the feeding experiments, resting oxygen consumption (VO2) was measured separately for each mouse at 28 °C, which is considered the lower critical temperature for M. musculus (Speakman and McQueenie 1996). This was performed in an open-circuit respirometry system using a Servomex Oxygen Analyser (Series 1100), as described elsewhere (e.g. Hayes, Garland & Dohm 1992; Mathias et al. 2004). Mice were placed in cylindrical Perspex chambers (0·19 m length × 0·10 m diameter), restricting their locomotory activity, which were then placed inside a Sanyo MIR-153 cooled incubator (Sanyo Biomedical Europe, B.V., Az Etten Leur, The Netherlands) allowing precise temperature control (±0·1 °C). The inner part of chambers contained a wire-mesh grid to avoid the contact of mice with urine and faeces. A flow of dried air passed through the metabolic chambers at a rate of 500 ml min−1. Analogue signals of the O2 content in the air leaving the chambers were digitized using the ‘Labtech: Data Acquisition and Process Control’ software (Labtech Windows User's Guide, Laboratory Technologies Corp., Andover, USA). Measurements were made at 15 second intervals over a 3-h period.

Each mouse was monitored twice, on two consecutive days, between 8 a.m. and 1 p.m. on one day and between 1 p.m. and 6 p.m. on the following day (or vice versa), to avoid any effects of daily individual metabolic rhythms (e.g. Hayes et al. 1992; Bennett and Spinks 1995). Body masses of mice (with an accuracy of 0·1 g) were registered before and after the VO2 measurements. Oxygen consumption was not measured for the race Ponta Delgada.

A total of 82 measurements were performed on 41 individuals. Oxygen consumption was calculated as the average of the lowest 10 measurements (= 2·5 min) of stable sections of each respirometry measurement period. Animals were not fed while in the respirometry chamber but were not food-deprived prior to being placed in the chamber. We have defined these conditions for measurement of VO2 elsewhere as RMRt (resting metabolic rate at thermoneutrality: Speakman, Krol & Johnson 2004) reflecting a measurement of minimal metabolic rate of an animal at rest in the thermoneutral zone during the resting phase of the activity cycle but not necessarily postabsorptive. VO2 measurements were obtained following Depocas and Hart (1957) as VO2 = V2(F1O2 –F2O2), where V2 is the flow rate measured after the metabolic chamber, and F1O2 and F2O2 are the oxygen concentrations before and after the metabolic chamber. All VO2 measurements were corrected to standard temperature and pressure (STPD). Means and standard deviations (SD) of metabolic rate values (VO2) for each chromosomal group were calculated, using the mean values of the two sessions.

relationships between physiological and environmental variables and rb race

The influence of race and environmental factors at the sites of origin upon food energy intake and oxygen consumption was assessed using generalized linear modelling. We first checked all the data for normality and homogeneity of variances and we transformed traits where necessary. Whenever a significant effect was observed (P < 0·05), post-hoc comparisons were performed by the Tukey HSD for unequal sample sizes. The Wilcoxon matched-pairs test was applied to compare observed values of BMR with their expected values based on body weight and according to the formula for Muridae: BMR = 2·735 × (body weight0·608) (Koteja and Weiner 1993). All statistical analyses were performed using either SPSS Software, version 6·0 (Norusis 1992) or MINITAB 13·31. Values are given as mean ± standard deviation (SD).


Average daily energy intake for all the chromosomal groups is summarized in Table 3. During the 5-day feeding experiments, mice of each race presented almost constant levels of daily energy consumption, although these levels were significantly different among races (F = 3·820, P = 0·003), ranging from 25·3 ± 7·91 kJ day−1 to 34·6 ± 5·11 kJ day−1, respectively, in races S. Vicente and Achadas da Cruz. Race S. Vicente consumed, respectively, 27% and 23% significantly less energy than races Achadas da Cruz and Estreito da Calheta (post-hoc Tukey tests, P < 0·05). Moreover, on average, hybrids ingested 20–30% significantly less energy each day than races Achadas da Cruz, Estreito da Calheta and Santana (post-hoc Tukey tests P < 0·05) (Table 3).

Table 3.  Influence of Rb races on gross energy intake (I) in House Mice from the Madeira island. N– number of mice in each chromosomal group; results given as mean ± standard deviation
Chromosomal groupBody mass (g)Energy intake (I) (kJ day−1)
Race Santana (N = 16)17·51 ± 2·5928·85 ± 4·58
Race Lugar de Baixo (N = 5)16·49 ± 2·2628·96 ± 7·59
Race Estreito da Calheta (N = 8)19·69 ± 2·1732·80 ± 2·64
Race Achadas da Cruz (N = 6)18·96 ± 2·9134·63 ± 5·11
Race S. Vicente (N = 7)18·71 ± 1·9925·27 ± 7·91
Race Ponta Delgada (N = 4)16·68 ± 2·1128·25 ± 1·29
Hybrids (N = 13)15·42 ± 1·8923·01 ± 4·30

Mean resting oxygen consumption of Rb mice and hybrids is shown in Table 4. No significant association was found between the whole animal oxygen consumption and the chromosomal group (F = 0·684, P = 0·639). All the Rb races under analysis consumed significantly less oxygen compared with the expected basal metabolic rate predicted from their body masses, using the predictive equation in Koteja and Weiner (1993) (Wilcoxon matched pairs test). The discrepancy to the prediction amounted to between 50·8% in S. Vicente mice, and 33·5% in Achadas da Cruz and Santana mice. Hybrids also had very low metabolic rates compared with those predicted.

Table 4.  Comparison between observed resting oxygen consumption and predicted basal metabolic rate in Robertsonian races of House Mouse (%– measured VO2/BMR predicted; N– sample size)
Chromosomal groupOxygen consumption (ml O2 h−1)BMR (ml O2 h−1)
  • a

    BMR = 2·735 × (body weight0·608);

  • b

    b Wilcoxon matched-pairs test.

Race Santana (N = 10)25·9 ± 6·6638·9 ± 0·1266·5−2·800·005
Race Lugar de Baixo (N = 5)21·1 ± 8·8736·7 ± 0·1157·5−2·020·043
Race Estreito da Calheta (N = 5)26·2 ± 4·2340·8 ± 0·0564·2−2·020·043
Race Achadas da Cruz (N = 5)27·1 ± 6·1840·8 ± 0·0466·5−2·020·043
Race S. Vicente (N = 7)18·9 ± 10·0338·4 ± 0·0749·2−2·370·018
Hybrids (N = 9)18·7 ± 8·8734·6 ± 0·0954·1−2·670·008
F statistics0·68    
P = 0·639    

There were two measurements of body mass available for these animals: the mass during the food intake measurements and that taken just prior to the respirometry measures. Since these measurements were taken some weeks apart, the two estimates of body mass were correlated (r = 0·688, P < 0·001) but not identical. We explored the factors influencing body mass by analysing each estimate of body mass separately. For mass at food intake, there were no significant correlations with chromosomal race, mean altitude, mean winter temperature and level or days of precipitation at the capture site. However, there was a significant negative relationship to the mean summer temperature at the capture site (Fig. 2) – the least squares fit regression loge body mass (g) = 5·42 − 0·862(loge T°C) explained 22·2% of the variation in body mass (F1,57 = 16·31, P < 0·001). The same effects were recorded for the mass at measurement of oxygen consumption, with chromosomal race, mean altitude, mean winter temperature, level and days of precipitation all having non-significant correlations, but mean summer temperature being significantly negatively associated with body mass (F1,39 = 7·74, P = 0·011). Mice derived from sites that were hotter in the summer had lower body masses.

Figure 2.

Body mass of mice from the island of Madeira, prior to measurements of food intake, plotted against the mean summer temperature at the site where the mice were originally captured. There was a significant negative relationship.

There was a significant effect of body mass on the rate of oxygen consumption (Fig. 3). The least squares fit regression loge VO2 (ml h−1) = −0·227 + 1·137 loge BM (g) explained 40·7% of the variation in oxygen consumption. The residual variance once the influence of mass had been accounted for was not related to the altitude, or weather variables (mean summer and winter temperatures, mean annual precipitation and number of days with precipitation) at the sites where the mice were captured. Residual variance in oxygen consumption was not significantly associated with the chromosomal race (P > 0·05).

Figure 3.

Rate of resting oxygen consumption (ml O2 h−1) at thermoneutral (28 °C) plotted against body mass (g) for mice captured on the island of Madeira. This plot includes mice captured from different sites and of different chromosomal groups (Rb fusion groups 1–7: 1 Santana, 2 Lugar de Baixo, 3 Estreito da Calheta, 4 Achadas da Cruz, 5 S. Vicente, 6 Ponta Delgada, 7 Hybrids 2n = 22 × 2n = 40 – see Materials and methods and Table 1 for details).

Food energy intake was also significantly related to body mass (Fig. 4), loge intake (kJ day−1) = 0·654 + 0·935 loge BM (g), which explained 39·0% of the variation in energy intake (F1,57 = 36·47, P < 0·001). Although both energy intake and resting oxygen consumption were both related to body mass measured immediately prior to the respective measurements, the relationship between energy intake and resting oxygen consumption marginally failed to reach significance (F1,35 = 3·57, P = 0·067). There was no significant relationship between the residual food energy intake and the residual resting oxygen consumption once the effect of body mass was taken into account (F1,35 = 0·2, P = 0·66).

Figure 4.

Daily energy intake (kJ day−1) of mice from the island of Madeira plotted against body mass immediately prior to the food intake measurements. This plot includes mice captured at a range of sites and consisting of several chromosomal groups (Rb fusion groups 1–7: 1 Santana, 2 Lugar de Baixo, 3 Estreito da Calheta, 4 Achadas da Cruz, 5 S. Vicente, 6 Ponta Delgada, 7 Hybrids 2n = 22 × 2n = 40 – see Materials and methods and Table 1 for details).

Residual food energy intake was not significantly associated with any of the environmental variables at the sites of origin of the animals, but there was a significant variation between the chromosomal races (F6,52 = 4·05, P = 0·002) (Fig. 5), with S. Vicente mice having significantly lower daily food consumption than mice from races Santana, Estreito da Calheta and Achadas da Cruz (post-hoc Tukey tests, P < 0·05) (see Table 1 for karyotype identification). All other pairwise comparisons were not significant (P > 0·05).

Figure 5.

Residual loge daily food energy intake with the effects of body mass removed in relation to the chromosomal race of mice (Rb fusion groups 1–7: 1 Santana, 2 Lugar de Baixo, 3 Estreito da Calheta, 4 Achadas da Cruz, 5 S. Vicente, 6 Ponta Delgada, 7 Hybrids 2n = 22 × 2n = 40 – see Materials and methods and Table 1 for details).

The effects of chromosomal (Rb race) and environmental factors are summarized in Fig. 6. There were no direct effects of environmental factors on the energetic parameters. However, body mass was significantly influenced by the mean summer temperature at the site of capture, and body mass was a significant factor influencing both the resting oxygen consumption and the daily energy intake. The Rb race of mice did not influence body mass, but race had a significant independent correlation with food energy intake. This association was mostly influenced by the low residual food intake (with mass effects accounted for) reported for mice from S. Vicente race.

Figure 6.

The significant associations (solid lines) between genetic and environmental factors potentially influencing the body mass, food energy intake and resting oxygen consumption of wild mice on the island of Madeira.


Robertsonian mice from Madeira were characterized by low levels of resting oxygen consumption relative to the expected BMR from allometric predictions for other murids and mammals in general. The fact that mice in the present study may not have been completely postabsorptive prior to the measurements makes this difference all the more significant. Direct comparisons with measurements of resting oxygen consumption in Mus reported in other studies confirm this difference. For example, oxygen consumption estimated in Madeiran mice, at the lower limit of thermoneutrality, was lower than that reported by Hayes et al. (1992) in M. domesticus (1·860 ml O2 g−1 h−1), at 32 °C, by Górecki et al. (1990) in both M. m. musculus, at 30 °C (2·64 ml O2 g−1 h−1) and M. spretus, at 32 °C (2·84 ml O2 g−1 h−1), by Richardson et al. (1994) in female M. m. domesticus, at 32 °C (1·91 ml O2 g−1 h−1), by Speakman and McQueenie (1996), Speakman and Rossi (1999) and Johnson, Thomson & Speakman (2001) in female laboratory MF1 M. musculus (2·27 ml O2 g−1 h−1), at 28 °C, and by Mathias et al. (2004) in M. m. domesticus captured in anthropogenic habitats in mainland Portugal (1·80 ml O2 g−1 h−1). The oxygen consumptions of Rb mice from Madeira were more comparable with resting levels of oxygen consumption previously reported in fossorial rodents or insectivores, especially when inhabiting arid areas (e.g. Nevo and Shkolnik 1974; Lovegrove 1986; McNab 1986; Haim et al. 1991; Bennett and Spinks 1995). These low rates of metabolism are generally interpreted as adaptive mechanisms to reduce energy demands in habitats where energy supply may be a limiting factor. Reduced metabolism is also characteristic of birds and mammals inhabiting islands (McNab 2000, 2002), which is also generally interpreted as an adaptive response to the generally poorer resources available in island habitats.

Two factors indicate that the low levels of oxygen consumption in these mice were not related to the chromosomal differentiation or founder effect. First, the different combinations of Rb fusions (Table 1) did not influence the level of resting oxygen consumption, since there were no significant differences between the chromosomal races, either before or after the effects of body mass had been taken into account. More significantly, the resting oxygen consumptions of mice from Madeira, reported here, were similar to measurements for mice captured on the adjacent island of Porto Santo (Mathias et al. 2004), which have no Rb fusions (2n = 40).

The scaling exponent we observed for oxygen consumption in relation to body mass (1·14) was significantly greater than the interspecific exponents generated through comparisons across the Mammalia in general (Kleiber 1961; exponent 0·752). This was also significantly greater than the exponent derived by comparisons across the Muridae (Koteja and Weiner 1993: exponent 0·603). High intraspecific scaling exponents have previously been reported in various species (e.g. Kestrels, Falco tinnunclus, Daan et al. 1990; Tree Swallows, Tachycineta bicolour, Burness, Ydenberg & Hochachka 1998) including mice (Speakman and Johnston 2000), but the precise causality remains obscure. Attempts to tie individual variation in metabolic rates to differences in body composition have met with differing degrees of success, with some studies reporting excellent agreement, particularly in rats (Greenberg & Boozer 2000; Greenberg et al. 2000; Wang et al. 2000; Selman et al. 2005), while in others, including mice, the results have been disappointing (Speakman & Johnson 2000; Selman et al. 2001) pointing to additional physiological factors that vary between individuals that are not reflected by simply measuring the masses of component tissues.

The positive effect of body mass on gross food energy intake (scaling exponent 0·95) was in line with expectations from previous studies of intraspecific variability in food intake. The significant association with the Robertsonian races was, however, a novel finding in the present study. This depended primarily on the decrease of intake in race S. Vicente relative to the other races. S. Vicente Rb mice include a unique fusion of acrocentric chromosomes 11 and 19 that is not present in any of the other races (Table 1). All the other fusions present in this race also occurred in one or another of the Madeiran chromosomal races (Table 1) which did not show a similar depression in intake (Fig. 5). Chromosome 19 is also involved in the Rb(2·19) fusion in the Santana race, which is physiologically very different from S. Vicente. Given this pattern it is tempting to suggest that the fusion of chromosomes 19 and 11 may have led to the disruption of a gene or genes that are associated with regulation of food intake in the neighbourhood of the fusion point (Sebat et al. 2004). Such variation between races may also arise due to different linkage relationships of genes close to the centromere, as the races are characterized by different metacentrics. In addition, although they have a common colonization origin (Gunduz et al. 2001), recent population history (e.g. in founding and subsequent population size) may differ between the races, and result in genic divergence between the races. All these differences may in turn lead to physiological effects.

The significant negative effect of mean summer ambient temperature at the capture sites on the body sizes (masses) of the mice was the only mechanism we detected that linked energetic adaptation of these animals to environmental factors. Because body mass was the most significant factor influencing both the resting oxygen consumption (Fig. 3) and gross energy intake (Fig. 4), the climatic effect on overall patterns of energy utilization was important, and much greater than the associations with the chromosomal races which were relatively minor and confined to correlations with gross intake and not resting expenditure. This effect of mean summer temperature did not seem to be related in any obvious way to the composition in Rb fusions, suggesting that this chromosomal differentiation did not have clear adaptive significance with respect to climatic variation across the island. The distribution of Robertsonian races (Fig. 1) therefore seems likely to have been predominantly generated and sustained by chance founder effects, lack of significant movement within the island because of its topography and the infertility of hybrid animals (Nachman & Searle 1995; Britton-Davidian et al. 2000). However, the correlations of Rb fusions with food energy intake independent of mass effects may contribute to how well certain races thrive in given habitats and hence the distribution of races may have some adaptive significance as well. In the absence of any information on the historical distributions of different races, whether there is any selection acting on the races to influence their distributions, and thus whether they are energetically adapted to their habitats, remains unknown.

In summary, the main factor influencing the oxygen consumption and food energy intake of mice captured across the island of Madeira was body mass. This in turn was significantly related to the mean summer temperature at the sites where the mice were captured. None of the other biogeographical and climatic factors we examined was directly related to body mass or the energetic parameters. Although there is substantial chromosomal variation on Madeira, chromosomal race was not significantly associated with body mass or oxygen consumption, but did have a significant independent association with food intake – principally because of a depressed intake in one of the races. Overall, the impact of climate on energetics occurred independent of the chromosomal races, but Robertsonian fusions may not be completely benign in their relationships to energetics. This suggests their distribution may in part reflect adaptive effects, but in the absence of additional historical and genomic information this inference remains speculative.


The authors are most grateful to the Director and staff of Laboratório Marítimo da Guia (University of Lisbon, Portugal) for providing laboratory facilities regarding calorimetric determinations. Thanks are also due to the Departamento de Climatologia of Instituto de Geofísica Infante D. Luís for kindly providing the atmospheric pressure readings. Mouse captures at Madeira would have not been possible without the collaboration of Ruben Capela (University of Madeira) and Manuel Biscoito (Museum of Funchal).

This study was supported by a grant (PRAXIS/PCNA/BIA/135/96), involving European FEDER funds, from Fundação para a Ciência e Tecnologia to Centro de Biologia Ambiental (University of Lisbon, Portugal).