Journal of Arid Environments (2002) 51: 21–34 doi:10.1006/jare.2001.0912, available online at http://www.idealibrary.com on

Alternative refuge strategies and their relation to thermophysiology in two sympatric , brantsii and unisulcatus

T.P. Jackson*w, T.J. Roperz, L. Conradt}, M.J. Jacksonz N.C. Bennett*

*Department of Zoology and Entomology, Research Institute, University of Pretoria, Pretoria 0002, South Africa zSchool of Biological Sciences, University of Sussex, Brighton BN1 9QG, U.K. }School of Biology, University of Leeds, Leeds LS2 9JT, U.K. zNorthern Cape Nature Conservation Services, Private Bag X1, Springbok 8240, South Africa

(Received 28 April 2000, accepted 9 August 2000)

Temperatures were recorded in burrows of Brants’ whistling , Parotomys brantsii, and lodges of the surface-nesting Karoo bush rat, Otomys unisulcatus, in an attempt to relate within-refuge microclimate to known differences in the ’ thermophysiology. The two species were studied simultaneously at the same field site and, hence, under identical climatic conditions. The refuge microclimate of P. brantsii burrows was well buffered from ambient temperature variation, with a narrow average diel temperature range compared with the surface temperature. Maximum burrow temperature was independent of maximum ambient shade temperature. By comparison, the lodges of O. unisulcatus acted as poor buffers from ambient temperature fluctuations, exhibiting greater temperature range. Daily maximum tem- perature within the lodge was closely correlated with maximum ambient shade temperature. We argue that between-species differences in the thermophysiology of these and other otomyine rodents reflect differences in the thermal properties of their refuges, rather than differences in the degree to which the species in question are adapted to surface conditions in arid or mesic environments. # 2002 Elsevier Science Ltd.

Keywords: refuge; thermophysiology; Parotomys; Otomys; burrow; season; Karoo; arid

wCorresponding author. E-mail: [email protected]

0140-1963/02/010021 + 14 $35.00/0 # 2002 Elsevier Science Ltd. 22 T.P. JACKSON ET AL.

Introduction

While the principles underlying the concept of the fundamental niche (Hutchinson, 1957) may remain questionable (Austin, 1999), the idea of a set of environmental variables that define the occurrence of a species in the absence of competition remains a useful framework for explaining its distribution. Arid environments, because they are often associated with particularly challenging conditions, provide an especially useful context in which to make predictions regarding this model. Here, species are largely constrained by physiological limitations, though these may be partially surmounted through aspects of their behaviour. In this respect, activity patterns are important for searching effectively for important resources including food and mates. For example, by carefully regulating their activity periods, are able to control, to a certain extent, the microenvironment to which they are exposed and therefore the periods over which they may remain active (e.g. lizards, Louw & Holm (1972); larks, Williams et al. (1999); ants, Whitford (1999)). For smaller animals, retreating to refuges that provide a more favourable microenvironment, with a higher humidity and buffered from extreme temperatures, is an important means of escaping from adverse external conditions (Louw & Seely, 1982). Amongst rodents, underground burrow systems are widely used as escape refuges, offering protection from both the climate and predators (Bolwig, 1958; Bennett et al., 1988; Downs & Perrin, 1989; Reichman & Smith, 1990; Kinlaw, 1999; Roper et al., 2001). Some rodents, however, construct other types of refuge, particularly in environments in which burrowing would be impossible. Thus, species such as the Namaqua rock mouse, Aethomys namaquensis, and the dassie rat, Petromys typicus, which live in rocky areas, nest within rock crevices (Skinner & Smithers, 1990). Similarly, rodents inhabiting wetland habitats, in which burrowing is precluded by the danger of waterlogging, construct surface refuges such as the lodges of beavers, Castor canadensis (Jenkins, 1980) and the nest cups of the vlei rat, Otomys irroratus (Skinner & Smithers, 1990). Arboreal species build refuges that are elevated above the ground, such as the stick nests of the tree rat, Thallomys paedulcus (Skinner & Smithers, 1990) and the dreys of grey squirrels, Sciurus carolinensis (Gurnell, 1987). While the microclimates of burrow systems have been relatively well documented (for a recent review see Roper et al., 2001), only a few studies have investigated the internal environment of refuges constructed on or above the surface of the ground (Lee, 1963; MacArthur, 1984; du Plessis et al., 1992; Dyke & MacArthur, 1993). Furthermore, since most investigations have been single-species studies, it is impossible to directly compare the thermal buffering properties of different types of refuges. For example, Buffenstein (1984) argues that differences in the thermo- physiological characteristics of two sympatric rodent species, the crevice-dwelling Aethomys namaquensis and the burrow-dwelling Gerbillus paeba, reflect differences in their refuge strategy. However, this hypothesis remains untested since data on the thermal insulating properties of the refuges in question are only available for G. paeba. In the present study, we examined within-refuge temperature of two closely related, sympatric otomyine rodent species that construct markedly different types of refuges. Brants’ whistling rat, Parotomys brantsii, excavates extensive, multi-entranced burrow systems penetrating to a depth of about 40 cm below ground, whereas the Karoo bush rat, Otomys unisulcatus, constructs large lodges made of closely interwoven sticks and sited on the surface (Skinner & Smithers, 1990). Otomys unisulcatus exhibits a lower thermal conductance (i.e. is better insulated) and has a lower metabolic rate than P. À1 À1 À1 brantsii (thermal conductance: O. unisulcatus 0?070 ml O2 g 1C h , P. brantsii À1 À1 À1 À1 À1 0?091 ml O2g 1C h ; resting metabolic rate: O. unisulcatus 0?90 ml O2 g h ; P. À1 À1 brantsii 1?15 ml O2 g h ; Jackson et al., in review), suggesting that temperature may influence refuge selection and even species fitness. However, since the two species are largely sympatric and are therefore exposed to similar climatic conditions throughout REFUGE STRATEGY AND THERMOPHYSIOLOGY OF OTOMYINE RODENTS 23 much of their range, it has been suggested that these differences in thermophysiology reflect differences in the insulating properties of their respective refuges rather than differences in the external environment (Jackson et al., in review). Our aim was to examine one important aspect of this hypothesis by comparing variation in the within- refuge temperature of burrows and lodges subjected to identical external climatic conditions. We predicted that the stick lodges of O. unisulcatus would be less buffered against ambient temperature changes than the burrows of P. brantsii.

Methods

Study area

Fieldwork was conducted on a 12-ha site in the Goegap Nature Reserve (291 370 S, 171590 E), 10 km north-east of Springbok in the Northern Cape Province, South Africa, where P. brantsii and O. unisulcatus occur sympatrically. The area is semi-arid, with an average annual rainfall of 160 mm (Northern Cape Nature Conservation, unpubl. data), and is classed as upland succulent Karoo (Low & Rebelo, 1996). Succulent vegetation and winter ephemerals dominate the sandy habitats in which the two study species occur. In addition, temperature may vary considerably from day to day owing to the passage of cold fronts across the area. Our study was conducted over two periods, referred to as the winter/spring period and summer period. During the winter/spring period, which ran from mid-August to mid-October 1999, the maximum shade temperature ranged from 34?51Cto6?51C, and minimum shade temperature from 20?91Cto1?11C. Over the summer period, extending from mid- December 1999 to mid-January 2000, the maximum shade temperature ranged from 39?81Cto21?31C, and minimum shade temperature from 24?11Cto10?21C.

Rodent refuges

Parotomys brantsii constructs large, multi-entranced burrow systems (De Graaff & Nel, 1965; Du Plessis & Kerley 1991; Coetzee & Jackson, 1999; Plate 1) covering an average area of 72?9m2 and spaced on average 5?1 m apart (Jackson, 2000). Approximately, 50% of burrows at the study site are constructed in the open with the remainder being built under bushes of either Zygophyllum retrofractum or Lyceum spp. (Coetzee & Jackson, 1999; Jackson, 2000). Tunnel systems, which interconnect below ground, are relatively shallow, extending to a maximum depth of 37 cm in the study area and each burrow usually contains several nest chambers (Jackson, 2000). Nests are lined with shredded plant material from nearby grasses and perennial shrubs. A single individual normally occupies a burrow system (Jackson, 1999). In contrast, O. unisulcatus constructs elaborate, dome-shaped lodges made of intricately interwoven sticks and twigs (Vermeulen & Nel, 1988; Plate 2). Lodges may be up to 0?45 m high and 1?3 m in diameter (du Plessis & Kerley, 1991). Each lodge is usually built within the base of a shrub (Z. retrofractum or Lyceum spp. in this study) and has a number of entrances (6–11) opening at or above ground level. Internally, the lodge is criss-crossed by passages and tunnels leading to nests, latrines and entrances. Nests are lined with shredded vegetation. Beneath the lodge may be one or two tunnels, extending to 30 cm below the ground surface and apparently used to escape from predators (Vermeulen & Nel, 1988) but we found no tunnels beneath either of two lodges that we dismantled at our study site. Several O. unisulcatus may share a stick lodge (Skinner & Smithers, 1990; Jackson, pers. obs.) but the exact nature of the relationship between individuals is not understood. 24 T.P. JACKSON ET AL.

Plate 1. Parotomys brantsii constructs large, multi-entranced burrow systems in the sandy habitats to which they are restricted. Burrow systems are often built in open areas away from the protective cover of bushes or other vegetation.

Plate 2. An O. unisulcatus feeds while sitting on its stick lodge, built into a Z. retrofractum bush. Lodges are normally built within the base of such shrubs, allowing them to be constructed in both sandy and rocky habitats. REFUGE STRATEGY AND THERMOPHYSIOLOGY OF OTOMYINE RODENTS 25

Temperature monitoring

Temperatures were recorded to the nearest 0?11C using Orion TinytalkIIt data- loggers (Gemini Data Loggers, Chichester, U.K.) housed in clear plastic containers (34 mm diameter  52 mm long) and set to record temperature at 30-min intervals. Shade temperature was recorded at ground level using a single logger placed in a permanently shaded position, while black-bulb temperature was recorded by a logger in an opaque black plastic container, placed in the sun at ground level. Temperature was recorded in recently abandoned refuges in order to avoid any effects of refuge occupants. Sites were randomly chosen for P. brantsii within a long- term study area. All burrows had been occupied the previous year and were unoccupied at the time of this study due to a reduction in the population density. As several were subsequently reoccupied, we contend that the burrows from which we collected data were not sub-optimal refuges. Suitable sites were selected within nine P. brantsii burrows during the winter/spring by probing into the burrow system from the surface with a metal rod in order to locate a tunnel or nest chamber, then making a vertical hole, 3?5 cm in diameter, into which the logger could be inserted. In another six burrows over winter/spring and ten burrows during summer, nest chambers were located using information about the underground sleeping locations of animals that had been fitted with radio transmitters (Roper & Jackson, in prep.) and that had subsequently been predated. Once a logger had been inserted into a burrow (at depths ranging from 19 to 50 cm), the access hole was back-filled with soil. The O. unisulcatus population appeared to undergo a similar population decline to P. brantsii, due to adverse rainfall, though neither lodge occupancy nor population density was quantified on an ongoing basis. However, the presence of the dried remains of food plants around several of these lodges suggested they were abandoned in the previous year during a short-term population decline. Temperature loggers were placed in 25 O. unisulcatus lodges (15 winter/spring, 10 summer) by inserting a hollow plastic pipe (internal diameter 3?5 cm) into the centre of the lodge at ground level, pushing a logger to the end of the pipe and then removing the pipe. In both burrows and lodges, loggers were left in position for 5–10 days before being retrieved and the data downloaded onto a computer.

Statistical analysis

In order to determine the effect of external temperature on the diurnal temperature fluctuation within refuges we analysed data from two separate 24-h periods, starting at 00:30 h and ending at 00:00 h, for each individual refuge. One of these periods was a ‘cool’ day for the given season, defined as a day on which the maximum ambient shade temperature did not exceed 151C during winter/spring or 321C during summer. The other was a ‘warm’ day, defined as a day on which the maximum ambient shade temperature exceeded 251C during winter/spring or 321C during summer. As cool days were not experienced during recording at three refuges in each species over winter/spring, the sample size for this part of the analysis was reduced from 15 to 12 refuges for each species. Multiple linear regression was used to analyse the relationship between maximum ambient shade temperature and maximum within-refuge temperature. Multiple regression (Zar, 1996) was also used to determine which factors were important determinants of within-refuge temperature. For P. brantsii, we analysed the effects of (1) maximum shade temperature; (2) average diel windspeed; (3) burrow depth; (4) surface area of burrow and (5) number of entrance holes to the burrow, on maximum within-refuge temperature. For O. unisulcatus, we analysed the effects of (1) maximum shade temperature; (2) average diel windspeed; (3) lodge volume and (4) 26 T.P. JACKSON ET AL. bush volume. Since the volume of a lodge or bush approximates to that of a paraboloid of revolution (Vermeulen & Nel, 1988), we calculated volume according to the formula 1/2 pr2h, where r is the lodge radius and h is the lodge height (Spiegel, 1968). For this regression analysis all values were used from each individual on consecutive days. Alhough this did not allow for complete independence of variables within the data set, large differences between temperatures on consecutive days ensured a degree of independence between observations. All data were analysed using the Statistica software package (Statsoft, 1996).

Results

Diurnal temperature fluctuation

On warm winter/spring and summer days, the temperature in both P. brantsii burrows and O. unisulcatus lodges remained below surface shade temperature throughout the day and exceeded surface temperature at night (Fig. 1). The temperature range in both burrows and lodges was less than that of surface shade temperature (Table 1), demonstrating that both burrows and lodges provide some degree of protection from external temperature extremes. Mean diel temperature in refuges did not differ between lodges and burrows during winter/spring (Student’s t-test; winter/spring: t =1?29 df. = 28 NS; Table 1) though the mean diel temperature in lodges was significantly cooler than burrows in summer (Student’s t-test; summer: t =5?10, df. = 18, po0?0001; Table 1). Furthermore, temperature range was significantly greater in lodges than in burrows (Student’s t-test; winter/spring: t =5?05, df. = 28, po0?00005; summer: t =6?14, df. = 18, po0?00001; Table 1). Temperature range did not differ significantly within refuges for either rodent between summer and winter/ spring (Student’s t-test; P. brantsii: t =0?22, df. = 23, NS; O. unisulcatus: t =1?81, df. = 23, NS; Table 1), though mean refuge temperature increased significantly from winter/spring to summer (Student’s t-test; P. brantsii: t = 17.68, df. = 23, po0?0001; O. unisulcatus: t =9?57, df. = 23, po0?0001; Table 1). Over the diurnal cycle, the time of minimum and maximum temperatures within lodges coincided with that of minimum and maximum surface shade temperature, during both summer and winter. However, the minimum temperature within burrows lagged approximately 4 h (winter/spring) and 3?5 h (summer) behind surface minimum; and similarly maximum temperature within burrows lagged behind maximum surface temperature by about 6?5 h (winter/spring) and 8 h (summer). These results show that burrows were better buffered against ambient temperature changes than were lodges. On relatively cool winter/spring days, the temperature in P. brantsii burrows remained at least a few degrees above ambient surface shade temperature throughout the 24-hour cycle, whilst in summer surface shade temperature only rose above refuge temperature for 3?5 h during the afternoon (Fig. 2). In contrast, O. unisulcatus stick lodges differed o0?51C from ambient temperature between 11:30 and 18:30 during the winter/spring period and was only 0À1?21C less than surface shade temperature from 09:30 h to 17:00 h during the summer. Parotomys brantsii burrows were always warmer inside than O. unisulcatus lodges on cool winter/spring days, though during summer afternoons temperatures were within 0?51C of each other. The temperature range in burrows (winter/spring: 2?71C, from 13?91Cto11?21C; summer: 4?01C, from 25?81Cto29?81C) was significantly less than in lodges (winter/spring: 4?11C, from 10?81Cto6?71C; summer: 9?71C, from 18?11C &27?81C) (Student’s t-test; winter/ spring: t =2?48, df. = 22, po0?05; summer: t =3?58, df. = 18, po0?005). These results further confirm that burrows were better insulated than lodges. REFUGE STRATEGY AND THERMOPHYSIOLOGY OF OTOMYINE RODENTS 27

Warm days Winter/spring 50

40 C) ° 30

20

10

0 00:30 02:00 03:30 05:00 06:30 08:00 09:30 11:00 12:30 14:00 15:30 17:00 18:30 20:00 21:30 23:00

Summer 50

40 C) Temperature ( ° 30

20

Temperature ( 10

0 00:30 02:00 03:30 05:00 06:30 08:00 09:30 11:00 12:30 14:00 15:30 17:00 18:30 20:00 21:30 23:00 Time (24:00 hour) Figure 1. Comparison of diurnal temperatures within P. brantsii burrows and O. unisulcatus lodges, as well as with ambient sun and shade temperatures on warm days (maximum shade temperature 4251C in spring/winter; 4321C in summer): ( ) O. unisulcatus;( ) P. brantsi;( ) Sun; ( ) shade.

Maximum refuge temperatures

During winter/spring, maximum within-refuge temperature was lower in lodges than in burrows at ambient temperatures of o191C and higher in lodges than in burrows at ambient temperatures exceeding 241C (see Fig. 3). An analogous pattern was observed during the summer, with maximum within-refuge temperatures being lower in lodges at ambient temperatures of o301C and higher in lodges at ambient temperatures exceeding 361C. Thus, P.brantsii burrows were more efficiently buffered against more extreme ambient temperatures, at both ends of the temperature range, than were the lodges of O. unisulcatus. Multiple linear regression analysis, showed a significant interaction between maximum ambient shade temperature and species both in winter (F1, 229 = 144?0, po0?0001) and in summer (F1, 138 = 246?5, po0?0001). Diel temperature range in lodges was closely correlated to ambient diel temperature range ( y =0?80xÀ1?76, r =0?71, n = 202, po0?0001; Fig. 4), whereas diel tempe- rature range in burrows was independent of ambient diel temperature range ( y0?05xþ 3?06, r =0?11, n = 186, NS; Fig. 4). Temperature ranges were used for all 8TP JACKSON T.P. 28

Table 1. Seasonal variation in average maximum, minimum and mean diel temperatures (1C) as well as temperature range within P. brantsii burrows, O. unisulcatus lodges and surface shade temperatures. All values shown are 7S.D. Summer Winter/spring Mean Max Min Range (n) Mean Max. Min Range (n)

Parotomys brantsii 28?271?230?272?126?470?73?872?01016?371?918?171?714?572?53?671?615 AL. ET Otomys unisulcatus 24?671?930?771?719?273?211?673?41015?272?719?673?310?873?18?973?615 Ambient shade 24?573?433?174?017?073?616?473?13015?175?122?277?08?874?113?674?645 REFUGE STRATEGY AND THERMOPHYSIOLOGY OF OTOMYINE RODENTS 29

Cool days Winter/spring 50

40 C) ° 30

20 Temperature ( 10

0 00:30 02:00 03:30 05:00 06:30 08:00 09:30 11:00 12:30 14:00 15:30 17:00 18:30 20:00 21:30 23:00

Summer 50

40 C) ° 30

20

Temperature ( 10

0 00:30 02:00 03:30 05:00 06:30 08:00 09:30 11:00 12:30 14:00 15:30 17:00 18:30 20:00 21:30 23:00 Time (24:00 hour) Figure 2. Comparison of diurnal temperatures within P. brantsii burrows and O. unisulcatus lodges, as well as with ambient sun and shade temperatures on cool days (maximum shade temperature o151C in spring/winter; o321C in summer): ( ) O. unisulcatus;( ) P. brantsi;( ) Sun; ( ) shade. seasons, as there were no seasonal differences in temperature ranges within refuges (see Table 1). Multiple regression showed that factors affecting the maximum daily temperature within refuges differed with refuge type and season (Tables 2 and 3) and differences were significant for both refugia in both seasons (P. brantsii, summer: R2 =0?63, 2 F5, 58 =19?6, po0?0001; winter: R =0?60, F5, 101 =30?6, po0?0001; O. unisulcatus, 2 2 summer: R =0?98, F4, 67 = 788?0, po0?0001; winter: R =0?78, F4, 106 =93?0, po0?0001). Over winter/spring burrows were influenced by maximum ambient shade temperature but not by wind speed, burrow depth, area or the number of entrances, whilst wind speed and burrow depth also influenced burrow temperature in summer (Table 2). At this time higher burrow temperatures were associated with greater shade temperatures, stronger winds and greater burrow depths. Similarly, lodges were only influenced by maximum ambient shade temperature during winter/spring, whilst wind speed also influenced burrow temperature in summer (Table 3). Unlike P. brantsii burrows, the convective action of stronger winds in summer served to reduce within- lodge temperatures. 30 T.P. JACKSON ET AL.

Winter/spring 45 C) ° 40 35 30 25 20 15 10

Maximum refuge temperature ( 5 0 0 5 10 15 20 25 30 35 40

Summer 45 C) ° 40 35 30 25 20 15 10 5

Maximum refuge temperature ( 0 0 5 10 15 20 25 30 35 40 Maximum ambient temperature in shade (°C) Figure 3. Maximum daily temperature in P. brantsii burrows and O. unisulcatus lodges as a function of maximum daily shade temperature: ( ) O. unisulcatus;( ) P. brantsii.

25

20 C) ° 15

10

in refuge ( 5 Diel temperature range 0 0 5 10 15 20 25 Diel temperature range in shade (°C) Figure 4. Maximum diel temperature range in P.brantsii burrows and O. unisulcatus lodges as a function of maximum diel shade temperature range. ( ) O. unisulcatus;( ) P. brantsii. REFUGE STRATEGY AND THERMOPHYSIOLOGY OF OTOMYINE RODENTS 31

Table 2. Multiple regression analysis, showing the effect of various physical parameters on the maximum and minimum daily temperatures within P. brantsii burrows, during both summer and winter/spring periods Variable Summer Winter /spring

b t(58) p b t(101) p Maximum burrow temperature Maximum shade 0?22 2?28 o0?05 0?77 11?0 o0?0001 temperature Average diel windspeed 0?31 2?95 o0?005 0?00 0?05 NS Burrow Area 0?43 1?14 NS 0?10 1?38 NS Entrances À0?19 À0?53 NS À0?07 À0.94 NS Depth 0?60 3?70 o0?0005 0?03 0?46 NS

Minimum burrow temperature Minimum shade 0?69 5?78 o0?0001 0?62 11?79 o0?0001 temperature Average diel windspeed 0?29 2?33 o0?05 À0?22 À4?51 o0?0001 Burrow Area 0?21 0?43 NS À0?03 À0?46 NS Entrances À0?15 À0?31 NS À0?06 À0.95 NS Depth À0?20 À1?00 NS 0?37 À6?97 o0?0001

Discussion

The results demonstrate that P. brantsii and O. unisulcatus experience markedly different temperature regimes within their respective refuges. While the stick lodges of O. unisulcatus do provide some protection against ambient temperature extremes, they are considerably less well insulated than the burrows of P.brantsii. Thus, temperatures within lodges differed by only a few degrees from ambient surface shade temperature, while temperatures within burrows were largely independent of ambient shade temperature. We suggest that differences in the thermophysiology of the two species reflect this difference in the thermal properties of the refuges that they construct. Otomys unisulcatus has a relatively low thermal conductance (high insulation), a low resting metabolic rate and a lower critical temperature of 251C (Jackson et al., in review). Our results suggest that within-refuge temperatures regularly drop below this lower critical temperature, in which case the efficient insulating ability of individuals would allow them to minimize the extra energy expenditure needed to keep warm. Further protection against low temperatures is no doubt provided by the insulating properties of the nests of shredded vegetation that the species constructs within its refuges (Vermeulen & Nel, 1988) and possibly also by huddling, since O. unisulcatus is known to be gregarious (Brown, 1987; Skinner & Smithers 1990). At the other end of the temperature range, O. unisulcatus is able to withstand relatively high ambient temperatures by becoming hyperthermic (Jackson et al., in review). The burrow systems of P.brantsii provide a much better buffered microclimate. This may explain why this species has a higher thermal conductance than O. unisulcatus 32 T.P. JACKSON ET AL.

Table 3. Multiple regression analysis showing the effect of various physical parameters on the maximum and minimum daily temperatures within O. unisulcatus lodges, during both summer and winter/spring periods Variable Summer Winter/spring

b t(67) p b t(101) p Maximum lodge temperature Maximum shade 0?97 16?61 o0?0001 0?89 18?32 o0?0001 temperature Average diel windspeed À0?21 À4?24 o0?0001 À0?01 À0?13 NS Bush volume 0?53 1?63 NS À0?01 À0?29 NS Lodge volume À0?31 À0?85 NS À0?05 À1?18 NS

Minimum lodge temperature Minimum shade 0?89 19?58 o0?0001 0?87 19?12 o0?0001 temperature Average diel windspeed 0?03 0?95 NS À0?10 À2?31 o0?05 Bush volume À0?17 À3?30 o0?0001 À0?02 À0?51 NS Lodge volume 0?17 4?19 o0?005 À0?06 À1?42 NS

(Jackson et al., in review), since animals that are subject to a more conservative within- refuge temperature regime do not need to be as well insulated. In addition, the lower thermal insulation of P. brantsii may allow individuals to dissipate heat within the cooler confines of their burrows during hot days, following periods of activity above- ground. Thus, individuals typically take refuge underground during the summer between approximately 09:00 and 16:00 hs, retreating below ground at about the time when burrows are at their coolest due to the lag between surface and burrow temperatures (Jackson, 1998). Temperature certainly plays an important role in the summer activity of this diurnal rodent, and a positive correlation has been demonstrated between maximum ambient summer temperatures and the amount of time individuals spend either resting on the surface (as opposed to foraging or feeding) or underground in their refuges (Jackson, 1998). During winter, however, no relationship exists between maximum ambient temperature and surface activity patterns, suggesting that ambient temperatures do not limit the surface activity of individuals during winter (Jackson, 1998). Given that the internal microclimates of burrows are better buffered against surface temperature extremes, it is not obvious why O. unisulcatus should construct stick lodges. However, Du Plessis & Kerley (1991) demonstrated habitat segregation between P. brantsii and O. unisulcatus, the former being associated with areas characterized by deep soils and larger soil particle size than O. unisulcatus, which was restricted to areas with a higher percentage of plant cover. The presence of a suitable burrowing habitat (see also Graaff & Nel, 1965) may restrict the distribution of P. brantsii, whilst O. unisulcatus can extend its range into areas containing very soft sand such as those associated with dry riverbeds or coastal dunes (Vermeulen & Nel, 1988; Du Plessis & Kerley, 1991), or rocky substrates (Jackson, pers obs.). Therefore, differentiation of refuge strategy and their associated microhabitats may allow for species co-existence between these two rodents, a phenomenon well documented amongst desert rodents (M’Closkey, 1976; Price, 1978; Christian, 1980). REFUGE STRATEGY AND THERMOPHYSIOLOGY OF OTOMYINE RODENTS 33

This association between the thermophysiology of the two species and their refuges applies also to other otomyine rodents. Thus, the three burrowing species, P. brantsii (arid-occurring), P. littledalei (arid-occurring) and O. sloggetti (mesic-occurring) are all characterized by thermal conductance and resting metabolic rate typical of cricetid rodents of about the same body size (Richter et al., 1997; Jackson et al., in review). In contrast, the two species constructing refuges above-ground, O. unisulcatus (arid- occurring) and O. irroratus (mesic-occurring) both show relatively low thermal conductance and reduced resting metabolic rate compared with cricetid rodents as a whole (Haim & Fairall, 1987; Richter et al., 1997; Jackson et al., in review). This implies that refuge strategy is more important than external climatic conditions in determining the thermophysiological characteristics of otomyine, and possibly other rodents. At the very least, the environmental conditions experienced within a species’ refuge, as well as those that it experiences on the surface, must be taken into account when seeking a functional explanation of its thermophysiology.

We thank Northern Cape Nature Conservation for their invaluable assistance with this work, particularly Mr Julius Koen (Assistant Director: Research) and Messrs Klaas van Zyl and Enrico Oosthuisen and their staff at Goegap Nature Reserve. Prof. N.C. Bennett was awarded (1) a University of Pretoria post-doctoral bursary, supporting TPJ and; (2) a National Research Foundation travel grant aiding TJR. A. Haim and two anonymous reviewers provided valuable comments on earlier versions of the manuscript.

References

Austin, M.P. (1999). A silent clash of paradigms: some inconsistencies in community ecology. Oikos, 86: 170–178. Bennett, N.C., Jarvis, J.U.M. Davies, K.C. (1988). Daily and seasonal temperatures in the burrows of African rodent moles. South African Journal of Zoology, 23: 189–195. Bolwig, N. (1958). Aspects of ecology in the Kalahari. Koedoe, 1: 115–135. Brown, E.D. (1987). Comparative socio-ecology of Otomys irroratus and Otomys unisulcatus. Unpubl. M.Sc. thesis, University of Fort Hare, South Africa. 184 pp. Buffenstein, R. (1984). The importance of microhabitat in thermoregulation and thermal conductance in two Namib rodentsFa crevice dweller, Aethomys namaquensis, and a burrow dweller, Gerbillurus paeba. Journal of Thermal Biology, 9: 235–241. Christian, D.P. (1980). Vegetative cover, water resources and microdistributional patterns in a desert rodent community. Journal of Animal Ecology, 49: 807–816. Coetzee, C.G. Jackson, T.P. (1999). The comparative behaviour and ecology of the two species of Parotomys (Mammalia, Rodentia, Otomyinae) found in the arid areas of southern Africa. Journal of the Namibian Science Society, 47: 87–106. De Graaff, G. & Nel, J.A.J. (1965). On the tunnel system of Brants’ Karroo rat, Parotomys brantsi in the Kalahari Gemsbok National Park. Koedoe, 8: 136–139. Downs, C.T. & Perrin, M.R. (1989). An investigation of the macro- and micro-environments of four Gerbillurus species. Cimbebasia, 11: 41–54. Du Plessis, A. & Kerley, G.I.H. (1991). Refuge strategies and habitat segregation in two sympatric rodents, Otomys unisulcatus and Parotomys brantsii. Journal of Zoology (London), 224: 1–10. Du Plessis, A., Kerley, G.I.H. & Winter, P.E.D. (1992). Refuge microclimates of rodents: a surface nesting Otomys unisulcatus and a burrowing Parotomys brantsii. Acta Theriologica, 37: 351–358. Dyke, A.P. & MacArthur, R.A. (1993). Seasonal variation in the microclimate and gas composition of beaver lodges in a boreal environment. Journal of Mammalogy, 74: 180–188. Graaff, G. & Nel, J.A.J. (1965). On the tunnel system of Brants’ karroo rat, Parotomys brantsii,in the Kalahari Gemsbok National Park. Koedoe, 8: 136–139. Gurnell, J. (1987). The Natural History of Squirrels. London: Christopher Helm. Haim, A. & Fairall, N. (1987). Bioenergetics of an herbivorous rodent, Otomys irroratus. Physiological Zoology, 60: 305–309. 34 T.P. JACKSON ET AL.

Hutchinson, G.E. (1957). Concluding remarks. Cold Spring Harbour symposia on quantitative biology. 22: 415–427. Jackson, T.P. (1998). The diurnal activity of Brants’ Whistling Rat (Parotomys brantsii): the effect of seasonal and physical conditions. Transactions of the Royal Society of South Africa, 53: 227–236. Jackson, T.P. (1999). The social organisation and breeding system of Brants’ Whistling Rat, Parotomys brantsii. Journal of Zoology (London), 247: 323–331. Jackson, T.P. (2000). Adaptation to living in an open arid environment; lessons from the burrow structure of the two southern African whistling , Parotomys brantsii and P.littledalei. Journal of Arid Environments, 46: 345–355. Jenkins, S.H. (1980). A size–distance relationship in food selection by beavers. Ecology, 61: 740–746. Kinlaw, A. (1999). A review of burrowing by semi-fossorial vertebrates in arid environments. Journal of Arid Environments, 41: 127–145. Lee, A.K. (1963). The adaptations to arid environments in wood rats of the genus Neotoma. University of California Publications in Zoology, 64: 57–96. Louw, G.N. & Seely, M.K. (1982). Ecology of Desert Organisms. London: Longman. 194 pp. Louw, G.N. & Holm, E. (1972). Physiological morphological and behavioural adaptations in the ultrapsammophilous Namib Desert lizar, Aporosaura anchietae. Madoqua, 1: 67–85. Low, A.B. & Rebelo, A.G. (1996). Vegetation of South Africa, Lesotho and Swaziland. Pretoria: Department of Environmental Affairs and Tourism. MacArthur, R.A. (1984). Microenvironment gas concentrations and tolerance to hypercapnia in the muskrat Ondatra zibethicus. Physiological Zoology, 57: 85–98. M’Closkey, R.T. (1976). Community structure in sympatric rodents. Ecology, 57: 728–739. Price, M.V. (1978). The role of microhabitat in structuring desert rodent communities. Ecology, 59: 910–921. Reichman,O.J. & Smith, S.C. (1990). Burrows and burrowing behaviour by . In: Genoways, H.H. (Ed.), Current Mammalogy, Vol. 3, pp. 197–243. New York: Plenum Press. Richter, T.A., Webb, P.I. & Skinner, J.D. (1997). Limits to the distribution of the southern African ice rat (Otomys sloggetti): thermal physiology or competitive exclusion? Functional Ecology, 11: 240–246. Roper, T.J., Bennett, N.C., Conradt, L. & Molteno, A.J. (2001). Environmental conditions in burrows of two species of African mole-rat, Georhychus capensis and Cryptomys damarensis. Journal of Zoology , 254: 101–107. Skinner, J.D. & Smithers, R.H.N. (1990). The Mammals of the Southern African Subregion. South Africa: University of Pretoria. Spiegel, M.R. (1968). Mathematical Handbook of Formulas and Tables. New York: McGraw-Hill. Statsoft, Inc (1996). Statistica for Windows. Oklahoma: Tulsa. Vermeulen, H.C. & Nel, J.A.J. (1988). The bush Karoo rat Otomys unisulcatus on the Cape west coast. South African Journal of Zoology, 23: 103–111. Whitford, W.G.(1999). Seasonal and diurnal activity patterns in ant communities in a vegetation transition region of southeastern New-Mexico (hymenoptera, formicidae). Sociobiology, 34: 477–491. Williams, J.B., Tieleman, B.I. & Shobrak, M. (1999). Lizard burrows provide thermal refugia for larks in the Arabian Desert. Condor, 101: 714–717. Zar, J.H. (1996). Biostatistical Analysis (3rd Edn). Engelwood cliffs, NJ: Prentice-Hall.