Consistent proportional macronutrient intake selected by adult domestic cats (Felis catus) despite variations in macronutrient and moisture content of foods offered
- 4.2k Downloads
We investigated the ability of domestic cats to regulate the macronutrient composition of their diet when provided with foods that differed not only in macronutrient content but also in texture and moisture content, as typically found in the main forms of commercially manufactured cat foods. Cats were provided with foods in different combinations (1 wet + 3 dry; 1 dry + 3 wet; 3 wet + 3 dry) in three separate experiments. Within each experiment cats were offered the wet and dry food combinations in two (naïve and experienced) diet selection phases where all the foods were offered simultaneously, separated by a phase in which the foods were offered sequentially in 3-day cycles in pairs (1 wet with 1 dry). Using nutritional geometry we demonstrate convergence upon the same dietary macronutrient composition in the naïve and experienced self-selection phases of each experiment as well as over the course of the 3-day cycles in the pair-wise choice phase of each experiment. Furthermore, even though the dietary options were very different in each of these experiments the macronutrient composition of the diets achieved across all experiments were remarkably similar. These results indicate that a mammalian obligate carnivore, the domestic cat, is able to regulate food selection and intake to balance macronutrient intake despite differences in moisture content and textural properties of the foods provided.
KeywordsMacronutrient regulation Nutritional geometry Right-angled mixture triangles Carnivore nutrition Domestic cat
In order to meet its nutrient requirements, an animal is faced with the seemingly simple task of eating food. But foods are not simply parcels of nutriment; they are complex mixtures of nutrients, water and other chemical components. Some of these components add bulk to the food and change its physical characteristics; others are ‘anti-nutritional’, being toxic or interfering with the palatability of the food or the availability of nutrients to digestion (Rosenthal and Berenbaum 1991; Provenza et al. 2003); and yet others can be medicinal (Huffman 2001, 2003; Villalba et al. 2006; Raubenheimer and Simpson 2009). In addition, animals in their natural environment may be faced with a number of food sources which differ in quality (i.e. nutritional and non-nutritional content) as well as quantity (availability) leaving the animal with the problem of deciding ‘what’ and ‘how much’ to eat.
It would appear that natural selection has been successful at solving this problem from the animals’ perspective with extensive evidence showing that animals across a variety of taxa including insects, fish and mammals regulate and balance their intake of nutrients by adjusting their choice of foods and the amounts eaten (Raubenheimer and Simpson 1993, 1997, 2003; Simpson and Raubenheimer 2000, 2012; Rubio et al. 2003; Sánchez-Vázquez et al. 1999; Felton et al. 2009). Even under artificial selection (domestication), where the diet of the animal today is largely determined by humans, the evidence is unequivocal that when provided with a choice of foods with different nutritional profiles both companion animals (e.g. cats and dogs) and livestock (e.g. pigs, poultry and mink) consume different amounts of the foods to balance their nutrient intake (Kyriazakis et al. 1991; Hewson-Hughes et al. 2011; Raubenheimer and Simpson 1997; Romsos and Ferguson 1983).
Domestic cats are often fed manufactured pet foods which are produced in two main formats, dry (i.e. kibbles/biscuits; ~7–10 % moisture) and wet (i.e. in cans or pouches; ~75–85 % moisture). We previously investigated the ability of cats to regulate macronutrient intake when provided with a choice of dry foods or wet foods and demonstrated that cats have a ‘target’ intake of approximately 52 % of total energy as protein, 36 % as fat and 12 % as carbohydrate (Hewson-Hughes et al. 2011). This target was only attainable by cats offered the wet foods since the macronutrient compositions of the dry foods did not span this region of nutrient space (the foods contained a minimum of 26 % energy from carbohydrate), although cats did mix diet compositions from the dry foods provided that approached as closely as possible the target selected by cats offered wet foods.
As well as differing in water content and texture, the typical macronutrient compositions of these formats are also quite different with dry foods typically having a higher carbohydrate content compared to wet foods as complex carbohydrates, mainly starches, are widely used as binding agents in the manufacture of dry feeds for domestic animals. Whereas herbivores are adapted to deal with complex carbohydrates in plants, there is evidence that excessive starch content in manufactured feeds for carnivores can compromise the attainment of a balanced complement of other nutrients—as reported, for example, in salmonid fish under aquaculture (Ruohonen et al. 2007) and suggested for domestic cats (Hewson-Hughes et al. 2011); our data indicated that cats have a limit to the amount of carbohydrate they will ingest (~300 kJ per day) which limited further food intake and which we termed the ‘carbohydrate ceiling’ (Hewson-Hughes et al. 2011). Accordingly, cats confined to high-carbohydrate foods (>50 % energy from carbohydrate) were left with a shortfall in protein and fat intake (relative to the target), potentially encouraging them to seek those nutrients elsewhere in the urban environment (Hewson-Hughes et al. 2011).
Although these results clearly demonstrated that cats were able to regulate their macronutrient intake when provided with a choice of foods of the same format (i.e. wet or dry) it is not known to what extent differences in physical properties (e.g. texture/hardness) and water content may influence food selection and macronutrient intake when these formats are offered together. Here, we describe a series of experiments in which cats were offered different combinations of wet and dry foods representing an overlapping series of nutritional compositions in order to investigate this.
Materials and methods
Animal housing and welfare
Adult, neutered domestic short hair cats (Felis catus) of both sexes bred and housed at the WALTHAM® Centre for Pet Nutrition (WCPN), Melton Mowbray, Leicestershire, UK, participated in these diet selection experiments. Throughout each study the cats were housed and fed individually in purpose-built, behaviourally enriched lodges (w × d × h: 1.1 m × 2.5 m × 2.1 m) and were socialised as a group for approximately 1 h each day and had access to drinking water at all times. The cats were housed in social groups when not participating in experiments. The studies were approved by the WALTHAM® Ethical Review Committee.
Diets and general protocols
In this series of experiments cats were offered different combinations of wet and dry foods together with the aim being to determine the macronutrient balance selected by cats when offered foods not only with different macronutrient content but also different textures and moisture levels.
Nutrient compositions of wet and dry foods used in the experiments
Nutrient (g/100 g)
Detailed experimental designs are given below, but in general, the cats received 150 g of each dry food from 10:30 to 08:30 h the following morning and for wet foods, 190 g of each diet was offered from 10:30 to 15:00 h and was replaced with a fresh aliquot (190 g) from 15:00 to 08:30 h the next day. Food intake for each cat was determined at the end of each feeding period (i.e. at 15:00 and 08:30 h for the wet foods and at 08:30 h for dry foods) as the difference between the mass of food offered (g) and the mass of food remaining (g). Each experiment consisted of three phases.
Phase 1: naïve simultaneous self-selection (NSS)
For 7 days, cats were exposed to all of the experimental foods simultaneously. The aim of this phase was to measure nutrient self-selection by the cats when naïve to the experimental foods. To avoid positional bias, the position of each food was rotated daily.
Phase 2: pair-wise self-selection
Cats were cycled through eight 3-day periods during which they were confined to a different wet and dry food-pair on each of the 3 days. The aim of this phase was to determine the nutrient balance selected within the various ‘restricted’ food-pair choices available to them each day. This phase also served as a conditioning phase in which the cats gained further experience of the foods.
Phase 3: experienced simultaneous self-selection (ESS)
In this phase, the regimen of phase 1 was repeated on the now ‘experienced’ cats.
Experiment 1: one wet and three dry foods
Eighteen neutered adult cats (9 male, 9 female), aged 2.0–9.1 years (mean ± SEM, 4.3 ± 0.4 years) and weighing 5.49 ± 0.24 kg were used in this experiment. The cats were offered one wet food (W, Table 1) with three dry foods (Da–c, Table 1) simultaneously in separate bowls during the NSS and ESS phases (phases 1 and 3). For the pair-wise selection (phase 2), the wet food was paired with each one of the dry foods over the course of each 3-day cycle (e.g. cycle 1, day 1, W + Da (pair A); day 2, W + Db (pair B); day 3, W + Dc (pair C); repeated 8 times in total).
Experiment 2: one dry and three wet foods
Seventeen neutered adult cats (9 male, 8 female; different cats to those used in experiment 1) aged 2.1 – 9.1 years (4.3 ± 0.4 years) and weighing 5.27 ± 0.26 kg were used in this experiment. The design of this experiment was the same as experiment 1 except here the cats were offered a single dry food (D, Table 1) together with three wet foods (Wa–c, Table 1) in the NSS and ESS phases and food D paired with each of the wet foods during each cycle of the pair-wise selection phase (D + Wa (pair A); D + Wb (pair B); D + Wc (pair C).
Experiment 3: three wet and three dry foods
Ten cats (4 male, 6 female), aged 2.9–9.8 years (5.27 ± 0.6 years) and weighing 4.61 ± 0.27 kg, that had previously taken part in experiment 1 (4 cats) or 2 (6 cats) were used in this experiment. For the NSS and ESS phases of this experiment cats were simultaneously offered 3 wet foods (Wa–c, Table 1) and 3 dry foods (Da–c, Table 1) in six separate bowls. The pair-wise choices offered during phase 2 were Wa + Da (pair A), Wb + Db (pair B) and Wc + Dc (pair C).
Experiment 4: one wet and one dry food
Having investigated the ability of cats to balance macronutrient intake when provided with combinations of wet and dry foods offered in differing ratios (i.e. relative number of bowls of each, 1 wet:3 dry; 3 dry:1 wet; 3 wet:3 dry) in experiments 1–3; here cats were offered a combination of foods more likely to be faced by cats in a domestic setting—one wet and one dry food. The foods were nutritionally complementary relative to the ‘intake target’ previously described in cats offered only wet foods (Hewson-Hughes et al. 2011) to determine whether providing different formats of food affected the ability of cats to achieve their target intake.
Mean macronutrient intake (g/day) in the naïve and experienced self-selection phases across experiments 1–3
The outcomes analysed were total energy consumed and the % energy from each macronutrient as a proportion of total energy intake [i.e. protein: energy ratio (PER), fat: energy ratio (FER) and carbohydrate: energy ratio (CER)]. Mixed model analyses were used to analyse these outcomes to take into account the repeated measures on an individual cat when estimating the variance structure.
Experiments 1–3 were analysed individually and collectively; for individual experiments phase nested in cat were defined as random effects and phase defined as the fixed effect; for combined analysis of all three experiments, phase nested in cat nested in experiment were defined as random effects and experiment defined as the fixed effect. Data from the pair-wise selection phase were defined differently in the models for the PER/FER/CER and total energy intake analyses. There was no sense in determining whether the PER/FER/CER selected in each diet pairing was statistically different from each other as the compositions of the foods in each pairing would have made this the case. Instead, the average PER/FER/CER (Pair Average) over the three pair-wise choices was compared statistically to the average PER/FER/CER selected during the NSS and ESS phases. In contrast, it was of interest to know whether total energy intake was different depending on the diet pair offered and compared to the energy intake in the NSS or ESS phases and so for these analyses the average energy intakes for each pair-wise choice were compared separately to the values for NSS and ESS.
Experiment 4 was analysed by mixed models with cat as a random effect, to form summaries of overall means (i.e. no fixed effect was fitted).
Differences between levels within the fixed effects were tested at the overall 5 % level using Tukey honestly significant difference tests to adjust for multiple comparisons. All analyses were performed in R v2.13 statistical software (http://www.R-project.org/), with ‘lme4’ and ‘multcomp’ packages (R Development Core Team 2010).
In addition, simulation analyses were performed to determine whether the macronutrient profiles determined in experiments 1–3, pooled over all phases, were significantly different from profiles that would have resulted from random intake of the foods provided. Thus, food intakes (g) were first simulated assuming a total average food intake of 400 g (sd 100 g). The proportion of food eaten from each bowl was also simulated assuming an equal intake from each bowl on average (e.g. where four bowls were offered simultaneously the proportion of intake was simulated to be 25 % on average). From these simulations, the relative amount eaten (g) and the resulting PER, FER and CER of the diet composed were calculated for each meal in the design of each experiment. These simulations were performed 1,000 times and for each simulation the average PER, FER and CER were calculated using the previously described mixed model analyses. The proportion of simulated PER, FER or CER averages that were greater than or less than the experimental PER, FER or CER averages actually selected provided a significance test (GenStat® v14 statistical software, GenStat VSN International, Hemel Hempstead, UK.).
Experiment 1: one wet and three dry foods
Experiment 2: one dry and three wet foods
Experiment 3: three wet and three dry foods
Compilation of experiments 1–3
Experiment 4: one wet and one dry food
This series of experiments examined the ability of cats to regulate macronutrient intake when provided with foods that not only differed in macronutrient composition, but also in moisture content and consequently in texture and energy density. When the results of experiments 1–3 are superimposed on a single RMT (Fig. 7), where overlapping regions in diet composition space across these three experiments are covered, it can be seen that self-selecting cats in all three experiments achieved remarkably similar diet compositions in terms of the proportions of protein, fat and carbohydrate selected when offered very different combinations of wet and dry foods. Whilst not identical, these profiles accord well with the target composition reported previously (52/36/12) for adult domestic cats offered choices of wet foods (Hewson-Hughes et al. 2011; yellow dot in Fig. 7) and provide further evidence of the cats’ ability to regulate their macronutrient intake, even when provided with foods of very different macronutrient and moisture content simultaneously. Hence, achieving this regulatory outcome involved cats eating different amounts and proportions of foods according to nutrient content, not whether wet or dry. This conclusion is supported by simulations which indicated that had the cats eaten a fixed amount from each bowl of food offered, the macronutrient composition of the resulting diet would have been significantly different to the actual compositions selected and the target macronutrient profile.
Interestingly, the macronutrient profile of the diets composed by domestic cats in the present experiments and previously (Hewson-Hughes et al., 2011) are similar to that reported for free-ranging feral cats (52/46/2; Plantinga et al., 2011), indicating that domestic cats have retained the capacity to regulate macronutrient intake to closely match the ‘natural’ diet of their wild ancestors, even though the manufactured foods provided to domestic cats bear little resemblance to the natural foods (e.g. small vertebrate prey). The macronutrient with the biggest discrepancy between our studies and the reported natural diet of feral cats is carbohydrate. Previously we reported that in achieving their target macronutrient intake cats consumed ~8 g/day carbohydrate (Hewson-Hughes et al. 2011), while in the present experiments cats consumed ~13–20 g/day (Table 2). Eisert (2011) calculated that the maximal amount of digestible carbohydrate (from glycogen and gut contents) a cat could derive from consuming (carbohydrate-loaded) rodent prey is ~2.1 g per day.
The prey-based natural diet of a hypercarnivore such as the cat supplies insufficient carbohydrate to meet the metabolic demands for glucose required, for example, by the brain, and this demand for glucose is met by a high capacity for gluconeogenesis from amino acids (Eisert 2011). Since the cat appears to be metabolically adapted to meet its glucose requirements on a very low carbohydrate diet, it seems unlikely that the higher carbohydrate intakes seen in the present experiments is the result of cats actively seeking higher carbohydrate intake, although this cannot be completely discounted. Thus, having a brain that metabolises glucose like any other animal (Eisert 2011) might have encouraged evolution of broader metabolic use of glucose after generations of access to a higher carbohydrate diet through association with humans. This possibility aside, the cats in the naïve and experienced self-selection phases of these experiments were faced with two or more foods that contained at least 24 % of energy from carbohydrate and intake of only relatively small amounts of these foods would obviously lead to increased carbohydrate intake. Of course, it could be argued that if cats do not ‘need’ dietary carbohydrate then they could have completely avoided these foods, but this does not allow for sampling errors or that animals may have an adaptive strategy of actively sampling available foods to assess their nutritional value or potentially toxic nature (Day et al. 1998). Under either scenario (sampling errors or adaptive sampling), ingesting any of a high-carbohydrate diet will boost intake beyond that possible on a rodent-based natural diet. Although no food type was avoided completely, nonetheless, it is interesting to note that the intake of food Da (the dry food with the highest carbohydrate content, 52 % CER) was very low, particularly in the ESS phase, suggesting that cats had learnt to avoid eating this food. Given the availability of a number of relatively high carbohydrate foods the cats could have consumed much greater amounts of carbohydrate, but in fact ingested ~20 g on average, which is entirely consistent with our previous finding of a 300 kJ/day carbohydrate ceiling (~20 g/day) limiting further food intake (Hewson-Hughes et al. 2011). Furthermore, in situations where cats were offered only two foods (one wet and one dry) that were nutritionally complementary to the previously identified target (i.e. Pair B in experiment 1 and the foods in experiment 4), the cats composed diets that were lower in CER and similar to the target (53/34/13 in experiment 1 and 48/41/11 in experiment 4 compared to target of 52/36/12, Hewson-Hughes et al. 2011).
A particularly notable finding was the convergence upon the same diet composition in the naïve and experienced self-selection phases (where cats were offered all foods simultaneously on the same day) as well as over the course of the 3-day cycles in the pair-wise choice phases of each experiment. We observed this phenomenon previously in sequentially (i.e. 3 foods offered over 3 days—one food/day) and simultaneously (i.e. 3 foods offered on same day) self-selecting cats offered 3 wet foods, but not in experiments where cats were offered 3 dry foods (Hewson-Hughes et al. 2011). This difference between wet and dry foods was explained in terms of cats being unable to avoid the carbohydrate ceiling when forced to switch between dry foods each day over a 3-day period and as a result they under-ate protein relative to simultaneous self-selecting cats (i.e. this was a consequence of the macronutrient composition of the dry foods rather than a property of dry food per se). This phenomenon of regulating to a target macronutrient intake when provided with nutritionally complementary foods simultaneously or at fixed time intervals is not a peculiarity of the cat since it has also been observed in migratory locust (Locusta migratoria) nymphs (Chambers et al. 1998). Obviously, it would be interesting to establish if this nutritional regulatory capacity is more widely held across other species.
These studies clearly demonstrate that cats regulate their macronutrient intake even when provided with foods that differ not only in macronutrient composition, but also in their physical characteristics, such as texture and water content. Furthermore, our present results highlight that providing nutritionally complementary wet and dry foods offers cats the opportunity to mix a diet that meets their macronutrient target.
This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.
- R Development Core Team (2010) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, http://www.R-project.org/
- Rosenthal GA, Berenbaum M (1991) Herbivores: their interactions with secondary metabolites, 2nd edn. Academic Press, San DiegoGoogle Scholar
- Simpson SJ, Raubenheimer D (2012) The nature of nutrition: a unifying framework from animal adaptation to human obesity. Princeton University Press, PrincetonGoogle Scholar