The extent and function of ‘food grinding’ in the laboratory mouse ( Mus musculus )

Abstract

Many laboratory rodents grind their food into crumbs that are discarded at the bottom of the cage (sometimes called orts). This can have substantial impacts on measures of food intake and assimilation efficiency. We quantified food grinding in two laboratory mouse strains on eight different diets and distinguished between two hypotheses of why food grinding occurs: a stereotypic behaviour due to a lack of environmental enrichment, or part of an optimal food intake strategy. Orts were quantified when mice were exposed to environmental enrichment and when offered diets of differing energetic quality. Grinding was significantly different between diets, but not between strains, although there was a significant diet by strain interaction. Ort production was lowest on the hardest diets. Not accounting for orts could affect food intake estimates by up to 31.8% and assimilation efficiency by up to 16.7%. Environmental enrichment increased physical activity, but did not reduce grinding. Mice selected the higher energy density components of the food. We suggest a refinement of the current methodology for measuring food intake is essential, primarily because failure to take ort production into account created inaccurate estimates of food intake and assimilation efficiency in mice. Adding environmental enrichment is unlikely to reduce food grinding, but careful choice of diet will reduce the errors.

Keywords

Feeding environmental enrichment refinement rodents body weight

Laboratory mice are usually offered food in a hopper so they must gnaw at food pellets through metal bars. Food is offered this way to reduce contamination by faeces and urine and for easy measurement of food intake; calculated as the mass of food missing from the hopper each day. As mice feed, large chunks of food occasionally fall through the bars, which are usually put back into the hopper by researchers prior to weighing the food to estimate intake.¹ However, some mice also ‘grind’ food, resulting in a fine spillage discarded on the floor of the cage, sometimes called ‘orts’.² This term has been defined in the literature in different ways^3–10 and in many cases the exact definition is not stated. Both captive^11–14 and wild rodents^10,15,16 have been reported to grind food.

Koteja et al. ⁹ suggested that ‘food wastage should not be ignored without justification in calculations of food consumption’. An example of the importance of including orts in measurements of food intake comes from lines of mice selected for high and low food intake.¹⁷ Including or excluding orts can also impact on calculations of assimilation efficiency.¹³ In a literature search using the search terms ‘food intake’ and ‘mouse’ or ‘rat’ in the publications database Web of Science reveals more than 100,000 papers in the last five years alone. Out of 50 of these studies selected at random, only five mentioned correcting the intake for food grinding.^18–22 This may be because ort production is negligible^1,18 or it may be because some researchers are unaware of the extent of the potential problem with ignoring orts.

Two separate hypotheses for the behaviour have been suggested. First, food grinding in captivity may be a stereotypic or compulsive behaviour⁹ resulting from a lack of environmental enrichment. Normal behaviour patterns shown in wild mice, such as exploring and hiding may be lost in laboratory mice with no environmental enrichment.²³ This hypothesis predicts that grinding should be significantly reduced by providing mice with enrichment. Also, grinding should be independent of the quality and availability of the diet. Alternatively, optimal foraging theory predicts that ‘foragers should select a subset from the set of potential food items, to maximize net energy intake per unit time spent foraging’.¹⁵ For example, wild animals often discard the non-digestible parts of the food (such as seed coatings) to select food of the highest energy density.^10,13,15 For mice, it may be beneficial to grind food to extract a more rewarding food component to maximize energy intake.²⁴ Therefore, as the heterogeneity of the diet increases and its quality decreases, it would be predicted that mice grind more to select the more energetically profitable parts of the food. Therefore, orts should have a lower energy content than the total diet as they should reflect poorer components. The impact of greater selectivity should generate a positive correlation between ort production and assimilation efficiency.

Koteja et al. ⁹ found that variation in grinding between individuals was high (from 2% to 40%), but grinding was consistent within individuals. The trait was highly correlated in siblings suggesting it has a genetic component. This suggestion is not independent of the two hypotheses presented above as animals may genetically differ in their responses to lack of enrichment and propensity to forage optimally. The hypothesis that individual variability in grinding behaviour is genetic predicts consistency in grinding within individuals, independent of the diet offered and the availability of the diet.

Food grinding may impact on measurements of food intake if it is not adequately accounted for. We aim to evaluate the extent of the problem of ignoring ort production in food intake measures and to point to conditions under which orts may be particularly significant. In the context of ‘3Rs’, we aim to provide a refinement of the methodology for measuring food intake. Food grinding was quantified in two laboratory mouse strains: MF1, an outbred strain used extensively in studies of food intake during reproduction,^1,25,26 and C57BL/6J, an inbred strain used extensively, particularly in studies of obesity.²⁷ All mice were offered eight different commercially available pelleted rodent diets.

Animals, materials and methods

Experiment 1

Study animals were 16, six-month-old MF1 (HsdOla:MF1) mice. This strain was generated in 1970 from a cross of LACA and CS1 mice, maintained as a closed colony since 1970. Fifteen, three-month-old C57BL/6J (C57BL/6JOla:Hsd) mice were also used. Both strains were purchased from Harlan (Oxon, UK) and naïve to any other procedures. All procedures were licensed under the UK Animals (Scientific Procedures) Act of 1986 and received approval from the University of Aberdeen Ethical Review Committee. All mice were female and individually housed from birth, in cages of dimensions 48 cm × 15 cm × 13 cm (M3 base: North Kent Plastics, Kent, UK). Water was provided ad libitum and sawdust (Aspen B8/20 Standard) and shredded paper (B/PW10 Paper wool DJW 300) given as nesting material (DBM Holdings, Edinburgh, UK). Mice were housed at 20 ± 2°C on a 12 h light:12 h dark photoperiod at a relative humidity of 48 ± 4%. Body mass and food intake were measured Monday to Friday at 09:00 h (±0.01 g; Sartorius top-pan balance). Food was provided ad libitum in a standard metal grid hopper; the bars of the hopper were on average 0.5 cm apart. Each diet was offered to two MF1 and two C57BL/6J mice for seven days and the diet order was randomized. Diets were changed every Friday after measurements so that mice had two full days of acclimatization to the new diet. The diets (n = 8) were formulated by the manufacturer's machine by pelleting the powdered components of the diet (average length 2–3 cm and diameter 1 cm). First were the ‘Research Diets’ (Research Diets Inc, New Brunswick, NJ, USA) commercially available diet-induced obesity ‘DIO series’ consisting of a low-fat control (D12450B: 10% kcal from fat), a medium fat (D12451: 45% kcal from fat) and a high-fat diet (D12492: 60% kcal from fat). Also included was a custom-made high protein diet for use in macronutrient choice experiments (DX04080301, Research Diets Inc). The main source of carbohydrate in these diets was sucrose and the primary source of fat was lard. TestDiet^® (Richmond, IN, USA) created the TestDiet^® 21st Century Western Diet™ Series. This consists of a low-fat control (5TJS: 12% kcal from fat), a medium fat (5TJN: 40% kcal from fat) and a high-fat diet (5TJP: 56% kcal from fat). These diets contained a variety of carbohydrates and fats from different sources. Finally, a standard rodent chow was included (12% kcal from fat) (CRM[P] Special Diet Services; BP Nutrition Ltd, Essex, UK). The diets were stored according to the manufacture's guidelines (diets with high-fat contents were stored at −20°C, but defrosted for 24 h before putting in food hoppers). Fresh food was given at every diet rotation.

Any large pieces of uneaten food found in the cage (that had fallen through the hopper bars) were returned to the hopper, and the mass of food missing from the hopper was recorded. Each day the paper bedding was shaken out and the contents of the cage separated by hand into faeces and orts. Orts and a sample of each diet were dried at 60°C (Gallenkamp, Loughborough, UK) for a minimum of 14 days. Food intake was calculated as the amount of food removed from the hopper, corrected for hydration and dried orts. Ort production was expressed as the mean amount fragmented (g/mouse/day). The percentage of the food fragmented (PF) was calculated as: (ort production/food intake) × 100%.

The hardness of each diet (kg/mm²) was measured using a digital microhardness tester fitted with a Vickers diamond indenter (Buehler, IL, USA). The hardness of each diet was tested from fresh material (as if at the point of feeding). Five replicates of each diet were taken and a mean hardness value calculated for each. Two diets (5TJP and 5TJN) were unsuitable for testing because they were too soft and easily crumbled by hand. The hardness of the softest measurable diet (D12492) was 1.7 kg/mm². In the two cases where hardness could not be measured, it was set at the midpoint between 1.7 and 0 kg/mm².

Experiment 2

Fifteen, 4.5-month-old female MF1 laboratory mice were used to explore the effects of diet quality, environmental enrichment and caloric restriction on food grinding. Housing and daily measurements were as detailed in experiment 1. Mice were sedated and implanted intraperitoneally with a wireless E-mitter (Model PDT-4000 E-Mitter, Mini-Mitter, Bend, OR, USA). Mice were sedated (isoflurane and oxygen mix) and maintained under anaesthesia throughout surgery via a nose tube. Surgical anaesthesia was assessed by checking the eyelid reflexes and limb movements. Mice were situated on a heat blanket throughout and breathing was continually monitored by observation. The E-mitter (size: 23 × 8 mm) was inserted intraperitoneally, and the wound sutured (Ethicon Vicryl; Dunlop's Ltd, Dumfries, UK). The surgical procedure took 20–30 min/mouse. Mice were administered with an analgesic by subcutaneous injection (Vetergesic [buprenorphine]; Dunlop's Ltd) and left to recover on a heat pad for up to one hour. Mice were put back in their cage and monitored for wound healing, but otherwise left alone for four weeks postsurgery. The implant recorded a series of activity counts each minute recorded by the Windows PC-based data acquisition system VitalView™ (Mini-Mitter).

Three independent experiments were carried out to determine if food grinding was affected by diet quality, environmental enrichment and caloric restriction. To test if ort production was affected by diet quality, mice were offered diets of three different energy densities consisting of standard pelleted rodent chow (Purina Mills chow #5001), with 0%, 20% or 40% cellulose added by weight (custom-made by Research Diets Inc). Diets were offered ad libitum in a hopper. Mice were offered the diets for 10 days in turn. To test if ort production was affected by the presence of environmental enrichment, on three separate days during exposure to 0% and 40% cellulose diets, enrichment was added to the cages. The enrichment products were Aspen blocks, Crawl Balls^TM, Des. Res.^TM, Jolly Balls, Mouse Huts^TM, Mouse tinted igloos, Nestlets and Tinted tunnels (Lillico Biotechnology, Surrey, UK). There were always at least two days without enrichment between the days enrichment was added. To test if ort production was affected by caloric restriction, mice were calorically restricted on the 20% cellulose diet for 10 days, during which they were offered 80% of their food intake during the ad libitum period. Each individual received each treatment (diet, enrichment and restriction), and the order of which was randomized across individuals.

Gross energy determinations and apparent energy assimilation efficiency

Faeces were collected from eight individuals during experiment 2 after at least five days on each diet ad libitum feeding (0%, 20% or 40% cellulose). Orts were pooled from each individual, from each day on each diet. The gross energy (GE) content of the faeces, orts, diets and cellulose were measured using adiabatic bomb calorimetry (Parr, Moline, IL, USA). The apparent energy absorption efficiency (AEAE) was calculated as detailed previously, whereby both the actual GE and the GE of only the ingested food were used in the calculation.²⁶ These calculations were repeated using food intake with and without correction for orts.

Statistics

Experiment 1

A one-way analysis of variance (ANOVA) was used to compare ort production, PF, food intake and body mass between each diet for each strain. Food intake, strain, individual and diet were all analysed in a general linear model (GLM) to determine which significantly influenced ort production and PF, including body mass as a covariate. Interactions between all these factors were considered.

Experiment 2

The effects of diet quality (0%, 20% or 40% cellulose) on ort production and PF were analysed using one-way repeated measures ANOVA. GLM was used to first examine if ort production was influenced by food intake or diet, including body mass as a covariate and second to determine if ort production differed within individuals on the different quality diets. To examine the effects of environmental enrichment, mean food intake, ort production, PF, body mass and activity were calculated for each mouse on three days: the day prior to, on the day of and the day following enrichment. One-way repeated measures ANOVA was used to find significant differences between days for each parameter. Data for the 0% and 40% cellulose diets were analysed separately. A paired Student's t-test was used to compare ort production and PF for the periods when the same mice were and were not under caloric restriction on the 20% cellulose diet. One-way ANOVA was used to find differences in GE content for the diets, faeces and orts. Correlations between AEAE and the GE of the diets, and also ort production and diet hardness were made using linear least-squares regression. To determine the primary composition of the orts, single-sample t-tests were used to compare the GE of orts produced on each diet (0%, 20% and 40% cellulose) to a single mean GE value of the diets, cellulose and faeces measured by calorimetry. All statistical analysis was performed with MINITAB version 13.1 and significance values were taken as P ≤ 0.05. Data are presented as the mean ± standard deviation.

Results

Experiment 1

Mean ort production, food intake, PF and body mass for mice from each strain on each diet are shown in Table 1. MF1 mice were on average 41% heavier than C57BL/6J mice (F _(1,246) = 1577.43, P < 0.001). Food intake averaged 4.13 g/day in MF1 and 2.92 g/day in C57BL/6J, which was a 29% difference (F _(1,246) = 230.92, P < 0.001). Ort production across all diets averaged 1.10 g/day in MF1 and 1.70 g/day in C57BL/6J. Ort production was not significantly related to body mass (F _(1,229) = 0.02, P = 0.881) or food intake (F _(1,229) = 2.24, P = 0.136), and did not differ between individuals (F _(1,229) = 0.00, P = 0.947) or between strains (F _(1,229) = 0.55, P = 0.461). However, there was a significant difference in ort production between the diets (F _(7,229) = 4.95, P < 0.001) and a significant diet-by-strain interaction (F _(7,229) = 2.77, P = 0.009). Consequently, the way the two strains of mice reacted to the different diets was not the same. The diet that resulted in the highest ort production averaged 4.30 g/day for MF1 and 8.41 g/day for C57BL/6 mice, which was 13.3% and 43.3% of food intake, respectively. Diets 5TJS and D12450B promoted no grinding in the MF1 (Figure 1a) and C57BL/6 strains (Figure 1b). The same significant effects were found when this analysis was repeated using the PF. Ort production was averaged over the two strains on each diet. There was a significant negative correlation between ort production and diet hardness (F _(1,6) = 7.63, P = 0.033) (Figure 2). The effect was non-linear and ort production in both strains increased enormously once the diet hardness fell below 3 kg/mm².
Figure 1
Mean ort production for each mouse, on each diet in experiment 1. The MF1 strain is shown in (a) (n = 16) and the C57BL/6J strain in (b) (n = 15). Values represent the mean ± standard deviation

Figure 2
Ort production on each diet for mice from both strains combined in experiment 1 (n = 31 mice) plotted against the hardness of the diet. Ort production decreased as the hardness of the diet increased

Table 1
Average values of ort production, food intake, PF and body mass as measured in experiment 1 for n = 16 MF1 (upper panel) and n = 15 C57Bl/6J mice (lower panel) on each experimental diet (n = 8)

Diet Ort production/day (g) Food intake/day (g) PF (%) Body mass (g)

MF1

CRM (P) 1.28 ± 1.64 4.39 ± 0.70 16.99 ± 18.20 37.81 ± 3.36

D12450B 0.01 ± 0.02 4.31 ± 0.57 0.15 ± 0.50 38.75 ± 3.24

D12451 0.37 ± 1.47 4.16 ± 0.47 2.60 ± 10.24 39.08 ± 4.42

D12492 1.55 ± 4.45 4.05 ± 0.53 6.68 ± 7.97 39.66 ± 4.05

DX04080301 0.02 ± 0.04 3.70 ± 0.61 0.41 ± 0.82 36.91 ± 3.12

5TJS 0.00 ± 0.00 3.90 ± 0.54 0.00 ± 0.02 36.89 ± 2.91

5TJN 4.30 ± 10.97 4.52 ± 0.52 13.26 ± 28.52 39.98 ± 3.40

5TJP 1.27 ± 3.11 4.03 ± 0.55 8.55 ± 17.78 39.46 ± 3.87

F _(7,112) value 1.72 3.25 2.84 1.85

P value 0.110 0.003 0.009 0.083

C57BL/6J

CRM (P) 0.16 ± 0.23 3.35 ± 0.38 3.98 ± 5.04 22.52 ± 1.77

D12450B 0.00 ± 0.00 2.76 ± 0.48 0.00 ± 0.00 22.40 ± 2.25

D12451 0.01 ± 0.02 2.53 ± 0.25 0.24 ± 0.87 22.86 ± 2.23

D12492 2.53 ± 4.99 2.82 ± 0.66 18.63 ± 27.85 24.12 ± 3.09

DX04080301 0.00 ± 0.02 2.66 ± 0.37 0.18 ± 0.69 22.16 ± 1.60

5TJS 0.02 ± 0.08 2.70 ± 0.36 1.05 ± 3.93 21.84 ± 1.91

5TJN 2.92 ± 4.19 3.20 ± 0.61 25.55 ± 25.53 23.27 ± 2.72

5TJP 8.41 ± 9.49 3.36 ± 0.95 43.27 ± 28.34 23.86 ± 2.63

F _(7,112) value 7.89 5.42 13.66 1.83

P value < 0.001 < 0.001 < 0.001 0.088

PF: percentage of the food fragmented

Values represent the mean ± standard deviation and significant differences are highlighted in bold

Experiment 2

During the diet quality experiments, ort production was not significantly related to body mass (F _(1,40) = 0.19, P = 0.663) or food intake (F _(1,40) = 0.08, P = 0.785), but there was a significant dietary effect (F _(2,40) = 4.81, P = 0.013). Ort production significantly increased as the proportion of cellulose in the diet increased (F _(2,42) = 13.10, P < 0.001) and as hardness decreased (F _(2,12) = 83.92, P < 0.001). The PF was 11.1%, 12.9% and 31.8% on the 0%, 20% and 40% cellulose diets, respectively (F _(2,42) = 22.99, P < 0.001). When offered the three cellulose diets, ort production was significantly different between individuals (F _(1,384) = 5.13, P = 0.024) and there was a significant individual-by-diet interaction (F _(2,384) = 4.31, P = 0.014). This suggested that while individuals differed in the extent of grinding, these differences were not consistent across the different diets (Figure 3).
Figure 3
Mean ort production for each mouse (n = 15) in experiment 2 on diets of three differing energetic qualities, with 0%, 20% or 40% cellulose. Each mouse is shown to illustrate the significant individual variation

On the days that enrichment was added, mice were observed to be using the products, for example Des. Res.^TM (provided a refuge for nesting) and Nestlets were shredded to make additional bedding and bedding was often dragged into the tunnels and huts. Activity levels increased by 5.8% and 5.5% for the MF1 and C57BL/6 strains respectively, as compared with their average activity. Independent of the diet, activity levels significantly increased on the day enrichment was provided (0% cellulose diet: F _(2,42) = 8.67, P = 0.001; 40% cellulose diet: F _(2,42) = 9.04, P = 0.001) (Table 2). However, there was no significant effect of environmental enrichment on ort production when compared with the day preceding or the day following enrichment for the 0% (P = 0.197) or 40% (P = 0.315) cellulose diets.
Table 2
Data from experiment 2, from n = 15 MF1 mice to examine the effects of adding environmental enrichment

Difference from mean

Mean (calculated over 3 days) Day prior Day of enrichment Day after F _(2,42) value P value

0% added cellulose

Ort production (g) 0.60 ± 0.37 0.05 ± 0.24 0.02 ± 0.24 −0.07 ± 0.19 1.69 0.197

Food intake (g) (corrected for orts) 5.01 ± 0.65 0.02 ± 0.27 0.08 ± 0.27 −0.10 ± 0.39 1.22 0.305

Body mass (g) 37.53 ± 3.23 0.02 ± 0.25 0.03 ± 0.16 −0.05 ± 0.28 2.83 0.070

PF (%) 9.96 ± 5.16 0.94 ± 1.67 −0.18 ± 2.40 −0.76 ± 1.83 0.37 0.695

Activity (arbitrary units) 13.90 ± 4.46 −0.38 ± 1.09 0.80 ± 0.86 −0.42 ± 0.75 8.67 0.001

40% added cellulose

Ort production (g) 4.23 ± 3.99 −0.64 ± 2.89 0.50 ± 1.95 −0.07 ± 0.40 1.19 0.315

Food intake (g) (corrected for orts) 7.09 ± 1.16 −0.05 ± 1.30 −0.17 ± 0.70 0.20 ± 0.68 0.62 0.543

Body mass (g) 36.13 ± 2.78 0.03 ± 0.20 −0.05 ± 0.16 0.03 ± 0.14 0.94 0.398

PF (%) 33.98 ± 16.02 −1.24 ± 7.68 0.01 ± 3.04 1.81 ± 5.20 1.10 0.342

Activity (arbitrary units) 12.05 ± 3.30 −0.36 ± 0.84 0.65 ± 0.84 −0.42 ± 0.63 9.04 0.001

PF: percentage of the food fragmented

Difference from the mean value of ort production, food intake, body mass, PF and activity were calculated on the day prior to, the day of and the day following addition of enrichment to the cages. One-way repeated analysis of variance was used to find significant differences between days for each parameter on each diet (0% and 40% cellulose). Values represent the mean ± standard deviation and significant differences are highlighted in bold

Average ort production on the 20% cellulose diet presented ad libitum was 1.10 ± 0.71 g/day and when the same animals were calorically restricted fell to zero (t = 6.02, P < 0.001). PF also significantly decreased from 13.0 ± 5.8% to zero (t = 8.69, P < 0.001).

GE contents and AEAE

On all three diets of varying cellulose content, the GE of orts was significantly lower than the GE of the food (Table 3). Orts had significantly greater GE content than faeces and cellulose. This shows that the rejected portion of the diet was not pure cellulose and suggested that individuals were selecting higher energy density components of the food for ingestion.
Table 3
Mean GE content of the faeces, orts, diets (0%, 20% and 40% cellulose) and cellulose from n = 8 MF1 mice in experiment 2

Gross energy content (kJ/g) 0% cellulose t value P value 20% cellulose t value P value 40% cellulose t value P value

Orts 16.67 ± 0.33 17.28 ± 0.53 17.25 ± 0.48

Food 18.29 ± 0.09 −14.01 < 0.001 18.19 ± 0.10 −4.89 0.002 17.81 ± 0.12 −3.26 0.014

Faeces 16.05 ± 0.43 5.39 0.001 16.55 ± 0.20 3.94 0.006 16.70 ± 0.16 3.25 0.014

Cellulose 16.67 ± 0.05 0.02 0.983 16.67 ± 0.05 3.27 0.014 16.67 ± 0.05 3.39 0.012

GE: gross energy

Values represent the mean ± standard deviation and significant differences are highlighted in bold

Mean AEAE was 77.5 ± 9.4%, 68.9 ± 6.1% and 45.4 ± 8.8% for the 0%, 20% and 40% cellulose diets, respectively (F _(2,21) = 31.70, P < 0.001). Because the mice rejected the poorer constituents of the diets in the orts, the mean GE of only the ingested food were estimated as 18.5 kJ/g for 0% cellulose, 18.3 kJ/g for 20% cellulose and 18.0 kJ/g for 40% cellulose. This was an average increase of 1.23%, 0.61% and 0.99% compared with the GE of the diets measured directly using calorimetry. Substituting the GE of the ingested food resulted in an average increase in the estimated energy intake of 0.25%, and a mean increase in AEAE of 0.28%. There was no significant correlation between AEAE (calculated using the GE of the diet or the ingested GE) and ort production on any diet. AEAE was calculated as 81.2%, 70.9% and 62.6% for the 0%, 20% and 40% cellulose diets using food intake not corrected for orts. This was an overestimation of 2.6%, 1.9% and 16.7%, respectively.

Discussion

The magnitude of error induced by failing to take orts into account when calculating food intake and AEAE can be large. The present study suggests that the effect varies with diet and with mouse strain, but errors in food intake can be as large as 31.8%, and assimilation efficiency estimates can be in error by up to 16.7%. Peterson and Wunder¹³ found only five out of 30 studies accounted for orts in digestibility calculations. We also found that only five of 50 studies made reference to accounting for food grinding.^18–22 It is unclear whether this failure to mention grinding is because researchers do not recognize its significance and do not account for it. However, our data suggest that on some diets, mice produce significant levels of orts. We therefore suggest that a refinement in experimental designs involving the measurement of food intake and assimilation efficiency in mice is required (by collecting orts) to gain accurate estimates of each.

Grinding was directly proportional to the hardness of the diet, as also reported by Ford.¹⁴ Rodent food manufacturers could strive to minimize the error in estimating food intake by making their pellets harder. Our data indicate that a target hardness of at least 2.5 kg/mm² is desirable to minimize ort production. Currently, commercially available pellets containing high percentages of fat or fat from a number of different sources tended to fall below this critical hardness and were ground more. Failure to account for grinding may seriously compromise studies examining the impact of high-fat diet feeding on food intake. The Research Diets 45% kcal from fat diet (D12451) was the only ‘high-fat’ diet in our sample that combined a high level of fat with a hardness above this critical limit.

The difference between strains detected here supports the hypothesis that food grinding is heritable.⁹ Recently, neurophysiological signalling pathways involved in grinding behaviour have been examined, further suggestive of a genetic component to the behaviour.^28,29 The fact that grinding may have a significant genetic component is particularly important when considering studies that concern genetically manipulated mice, as observed by Hastings et al. ¹⁷

It has been suggested that grinding may be a stereotypic or compulsive behaviour because of a lack of environmental enrichment.⁹ In our experiments, physical activity was stimulated when enrichment was added, indicating mice were responding to the presence of enrichment. However, this did not reduce the level of ort production, as was also reported in laboratory rats.⁶ In fact, the presence of some plastic enrichment items (igloos) resulted in additional grinding of the plastic. The present data suggest that grinding is not due to a lack of environmental enrichment and that adding enrichment to cages as a strategy to reducing grinding behaviour in mice is unlikely to be successful. In our experiments, we only added enrichment for a single day and it might be argued that there would have been a more significant effect on grinding if it was in place for longer. However, this seems unlikely as the novelty aspect of enrichment would be lost over time and its capacity to distract the animals from grinding food probably also reduced.

Alternatively, grinding might be the consequence of an optimal food intake strategy. In support of this hypothesis, the energy content of the orts was lower than that of the food, suggesting mice were selecting the higher energy density parts of the pellets. Second, ort production fell to zero during caloric restriction when there was no advantage to be gained from being selective. Third, the extent of grinding increased as the cellulose content of the food increased indicating mice were being more selective of the best parts of the diet when there was more incentive to do so. However, the energy density of the diet was correlated with hardness and since in experiment 1 we showed that hardness was a key feature influencing the level of grinding, this effect may have been a coincidence. Moreover, if grinding the food served to eliminate the poorest parts of the food, the impact would be an increase in the energy content of the ingested portion of the diet. The impact of selecting out the ort fraction on the energy content of the ingested food was however very minor. Koteja et al. ⁹ also observed no impact of the extent of grinding on the AEAE. Our data suggest that grinding may be part of an optimization strategy, but that its impact on GE intake and assimilation efficiency is only small because modern pelleted diets are extremely homogeneous in their composition. Maintaining this homogeneity in chow manufacture may help to minimize the impact of grinding behaviour.

In conclusion, the present data suggest that food grinding by mice is not due to lack of environmental enrichment, but may reflect an optimization strategy. Grinding was dependent on diet hardness and was particularly high in high-fat diets. Moreover, the significant strain effect on levels of grinding supported the idea that grinding may have a genetic basis. Together, these effects mean that quantifying ort production is paramount in studies of different genotypes of mice and high-fat feeding. Failure to take ort production into account may cause serious errors in estimates of food intake and assimilation efficiency.

Diet	Ort production/day (g)	Food intake/day (g)	PF (%)	Body mass (g)
MF1
CRM (P)	1.28 ± 1.64	4.39 ± 0.70	16.99 ± 18.20	37.81 ± 3.36
D12450B	0.01 ± 0.02	4.31 ± 0.57	0.15 ± 0.50	38.75 ± 3.24
D12451	0.37 ± 1.47	4.16 ± 0.47	2.60 ± 10.24	39.08 ± 4.42
D12492	1.55 ± 4.45	4.05 ± 0.53	6.68 ± 7.97	39.66 ± 4.05
DX04080301	0.02 ± 0.04	3.70 ± 0.61	0.41 ± 0.82	36.91 ± 3.12
5TJS	0.00 ± 0.00	3.90 ± 0.54	0.00 ± 0.02	36.89 ± 2.91
5TJN	4.30 ± 10.97	4.52 ± 0.52	13.26 ± 28.52	39.98 ± 3.40
5TJP	1.27 ± 3.11	4.03 ± 0.55	8.55 ± 17.78	39.46 ± 3.87
F _(7,112) value	1.72	3.25	2.84	1.85
P value	0.110	0.003	0.009	0.083
C57BL/6J
CRM (P)	0.16 ± 0.23	3.35 ± 0.38	3.98 ± 5.04	22.52 ± 1.77
D12450B	0.00 ± 0.00	2.76 ± 0.48	0.00 ± 0.00	22.40 ± 2.25
D12451	0.01 ± 0.02	2.53 ± 0.25	0.24 ± 0.87	22.86 ± 2.23
D12492	2.53 ± 4.99	2.82 ± 0.66	18.63 ± 27.85	24.12 ± 3.09
DX04080301	0.00 ± 0.02	2.66 ± 0.37	0.18 ± 0.69	22.16 ± 1.60
5TJS	0.02 ± 0.08	2.70 ± 0.36	1.05 ± 3.93	21.84 ± 1.91
5TJN	2.92 ± 4.19	3.20 ± 0.61	25.55 ± 25.53	23.27 ± 2.72
5TJP	8.41 ± 9.49	3.36 ± 0.95	43.27 ± 28.34	23.86 ± 2.63
F _(7,112) value	7.89	5.42	13.66	1.83
P value	< 0.001	< 0.001	< 0.001	0.088

		Difference from mean
0% added cellulose
Ort production (g)	0.60 ± 0.37	0.05 ± 0.24	0.02 ± 0.24	−0.07 ± 0.19	1.69	0.197
Food intake (g) (corrected for orts)	5.01 ± 0.65	0.02 ± 0.27	0.08 ± 0.27	−0.10 ± 0.39	1.22	0.305
Body mass (g)	37.53 ± 3.23	0.02 ± 0.25	0.03 ± 0.16	−0.05 ± 0.28	2.83	0.070
PF (%)	9.96 ± 5.16	0.94 ± 1.67	−0.18 ± 2.40	−0.76 ± 1.83	0.37	0.695
Activity (arbitrary units)	13.90 ± 4.46	−0.38 ± 1.09	0.80 ± 0.86	−0.42 ± 0.75	8.67	0.001
40% added cellulose
Ort production (g)	4.23 ± 3.99	−0.64 ± 2.89	0.50 ± 1.95	−0.07 ± 0.40	1.19	0.315
Food intake (g) (corrected for orts)	7.09 ± 1.16	−0.05 ± 1.30	−0.17 ± 0.70	0.20 ± 0.68	0.62	0.543
Body mass (g)	36.13 ± 2.78	0.03 ± 0.20	−0.05 ± 0.16	0.03 ± 0.14	0.94	0.398
PF (%)	33.98 ± 16.02	−1.24 ± 7.68	0.01 ± 3.04	1.81 ± 5.20	1.10	0.342
Activity (arbitrary units)	12.05 ± 3.30	−0.36 ± 0.84	0.65 ± 0.84	−0.42 ± 0.63	9.04	0.001

Gross energy content (kJ/g)	0% cellulose	t value	P value	20% cellulose	t value	P value	40% cellulose	t value	P value
Orts	16.67 ± 0.33			17.28 ± 0.53			17.25 ± 0.48
Food	18.29 ± 0.09	−14.01	< 0.001	18.19 ± 0.10	−4.89	0.002	17.81 ± 0.12	−3.26	0.014
Faeces	16.05 ± 0.43	5.39	0.001	16.55 ± 0.20	3.94	0.006	16.70 ± 0.16	3.25	0.014
Cellulose	16.67 ± 0.05	0.02	0.983	16.67 ± 0.05	3.27	0.014	16.67 ± 0.05	3.39	0.012

Footnotes

Acknowledgements

This work was partly funded by a studentship from the University of Aberdeen. Thanks to Yuko Gamo for help with ort collection. Thanks to Professor Steve Bull, Head of the School of Chemical Engineering and Advanced Materials and Chris Aylott, Design Unit at Newcastle University for assistance with diet hardness measurements.

References

Johnson

, Thomson

, Speakman

. Limits to sustained energy intake. I. Lactation in the laboratory mouse Mus musculus. J Exp Biol 2001;204:1925–35

Van Soest

. Nutritional Ecology of the Ruminant. New York: Cornell University Press, 1994

Garcia-Bojalil

, Staples

, Risco

, Savio

, Thatcher

. Protein degradability and calcium salts of long-chain fatty acids in the diets of lactating dairy cows: productive responses. J Dairy Sci 1998;81:1374–84

Hanley

, Robbins

, Hagerman

, McArthur

. Predicting digestible protein and digestible dry matter in tannin-containing forages consumed by ruminants. Ecology 1992;73:537–41

Loeb

, Schwab

, Demment

. Responses of pocket gophers (Thomomys bottae) to changes in diet quality. Oecologia 1991;86:542–51

Fiala

, Snow

, Greenough

. ‘Impoverished’ rats weigh more than ‘enriched’ rats because they eat more. Dev Psychobiol 1977;10:537–41

Koteja

, Swallow

, Carter

, Garland

Jr . Energy cost of wheel running in house mice: implications for coadaptation of locomotion and energy budgets. Physiol Biochem Zool 1999;72:238–49

Koteja

, Swallow

, Carter

, Garland

Jr . Maximum cold-induced food consumption in mice selected for high locomotor activity: implications for the evolution of endotherm energy budgets. J Exp Biol 2001;204:1177–90

Koteja

, Carter

, Swallow

, Garland

Jr . Food wasting by house mice: variation among individuals, families, and genetic lines. Physiol Behav 2003;80:375–83

10.

Owl

, Batzli

. The integrated processing response of voles to fibre content of natural diets. Funct Ecol 1998;12:4–13

11.

Tagliaferro

, Levitsky

. Spillage behavior and thiamin deficiency in the rat. Physiol Behav 1982;28:933–7

12.

Justice

, Smith

. A model of dietary fiber utilization by small mammalian herbivores, with empirical results for neotoma. Am Nat 1992;139:398–416

13.

Peterson

, Wunder

. Food sorting by collared lemmings (Dicrostonyx groenlandicus) and prairie voles (Microtus ochrogaster): a cautionary note for digestibility studies. Comp Biochem Physiol A Mol Integr Physiol 1997;116:119–24

14.

Ford

. Influence of diet pellet hardness and particle size on food utilization by mice, rats and hamsters. Lab Anim 1977;11:241–6

15.

Kerley

GIH

, Erasmus

. What do mice select for in seeds? Oecologia 1991;86:261–7

16.

Felicetti

, Shipley

, Witmer

, Robbins

. Digestibility, nitrogen excretion, and mean retention time by North American porcupines (Erethizon dorsatum) consuming natural forages. Physiol Biochem Zool 2000;73:772–80

17.

Hastings

, Moruppa

, Bunger

, Hill

. Effects of selection on food intake in the adult mouse. J Anim Breed Genet 1997;114:419–34

18.

Ogilvie

, Saladin

, Nagy

, Urcan

, Heyman

, Leibowitz

. Activation of the retinoid X receptor suppresses appetite in the rat. Endocrinology 2004;145:565–73

19.

Farley

, Cook

, Spar

, Austin

, Kowalski

. Meal pattern analysis of diet-induced obesity in susceptible and resistant rats. Obes Res 2003;11:845–51

20.

Challis

, Coll

, Yeo

GSH

, Mice lacking pro-opiomelanocortin are sensitive to high-fat feeding but respond normally to the acute anorectic effects of peptide-YY3-36. Proc Natl Acad Sci USA 2004;101:4695–700

21.

Vidal-Puig

, Grujic

, Zhang

, Energy metabolism in uncoupling protein 3 gene knockout mice. J Biol Chem 2000;275:16258–66

22.

Chen

, Nyomba

BLG

. Glucose intolerance and resistin expression in rat offspring exposed to ethanol in utero: modulation by postnatal high-fat diet. Endocrinology 2003;144:500–8

23.

Olsson

IAS

, Dahlborn

. Improving housing conditions for laboratory mice: a review of ‘environmental enrichment’. Lab Anim 2002;36:243–70

24.

Speakman

. Apparent absorption efficiencies for redshank (Tringa totanus L.) and oystercatcher (Haematopus ostralegus L.): implications for the predictions of optimal foraging models. Am Nat 1987;130:677–91

25.

Speakman

, Król

. Limits to sustained energy intake IX: a review of hypotheses. J Comp Physiol B 2005;175:375–94

26.

Król

, Murphy

, Speakman

. Limits to sustained energy intake. X. Effects of fur removal on reproductive performance in laboratory mice. J Exp Biol 2007;210:4233–43

27.

Speakman

, Hambly

, Mitchell

, Król

. The contribution of animal models to the study of obesity. Lab Anim 2008;42:413–32

28.

Baldock

, Allison

, Lundberg

, Novel role of Y1 receptors in the coordinated regulation of bone and energy homeostasis. J Biol Chem 2007;282:19092–102

29.

Sainsbury

, Lin

, McNamara

, Dynorphin knockout reduces fat mass and increases weight loss during fasting in mice. Mol Endocrinol 2007;21:1722–35

		Difference from mean
	Mean (calculated over 3 days)	Day prior	Day of enrichment	Day after	F _(2,42) value	P value
0% added cellulose
Ort production (g)	0.60 ± 0.37	0.05 ± 0.24	0.02 ± 0.24	−0.07 ± 0.19	1.69	0.197
Food intake (g) (corrected for orts)	5.01 ± 0.65	0.02 ± 0.27	0.08 ± 0.27	−0.10 ± 0.39	1.22	0.305
Body mass (g)	37.53 ± 3.23	0.02 ± 0.25	0.03 ± 0.16	−0.05 ± 0.28	2.83	0.070
PF (%)	9.96 ± 5.16	0.94 ± 1.67	−0.18 ± 2.40	−0.76 ± 1.83	0.37	0.695
Activity (arbitrary units)	13.90 ± 4.46	−0.38 ± 1.09	0.80 ± 0.86	−0.42 ± 0.75	8.67	0.001
40% added cellulose
Ort production (g)	4.23 ± 3.99	−0.64 ± 2.89	0.50 ± 1.95	−0.07 ± 0.40	1.19	0.315
Food intake (g) (corrected for orts)	7.09 ± 1.16	−0.05 ± 1.30	−0.17 ± 0.70	0.20 ± 0.68	0.62	0.543
Body mass (g)	36.13 ± 2.78	0.03 ± 0.20	−0.05 ± 0.16	0.03 ± 0.14	0.94	0.398
PF (%)	33.98 ± 16.02	−1.24 ± 7.68	0.01 ± 3.04	1.81 ± 5.20	1.10	0.342
Activity (arbitrary units)	12.05 ± 3.30	−0.36 ± 0.84	0.65 ± 0.84	−0.42 ± 0.63	9.04	0.001