Dietary energy requirements of dogs, cats and people.

First, following Okin [7], the dietary energy consumed by US dogs and cats was calculated. Maintenance energy requirements (MERs) describe “the amount of energy an animal needs to support energy equilibrium and accounts for thermoregulation, spontaneous activity and exercise. It also accounts for energy lost as heat during dietary thermogenesis or the metabolism and digestion of foods.” [41].

Domestic cats are relatively uniform in body weight (BW) reflecting their consistent roles as companion animals, notwithstanding exceptions such as cats used in laboratories, who are far fewer in number. In contrast, dog breeds vary dramatically in size from toy, small, medium, large to giant [42]. Canine MERs also vary significantly with husbandry type and activity level. These are greatest for racing dogs, followed by working and hunting dogs, and finally, pet and kennelled dogs. MERs appear equal between sexes, but are lower in neutered compared to sexually intact dogs [43]. Despite these variations, it’s nevertheless possible to determine average body weights and MERs, for both dogs and cats.

Until recently, the most accurate estimations of canine and feline calorific requirements were those supplied by the National Research Council (NRC) of the United States National Academies of Sciences, Engineering, and Medicine. According to the NRC [44], dogs’ average daily energy requirements in kJ are 544 (kg BW)^-0.75 (where kg BW = kg of body weight), and cats’ average daily energy requirements in kJ are 418 (kg BW)^-0.67. The NRC based its recommendations on the published studies available at the time. Since then, however, a range of additional studies have provided further evidence, particularly concerning the energy requirements of pet dogs. These may differ significantly from those of dogs kept in a kennelled environment, which commonly form the basis for controlled studies.

After systematically reviewing 29 published studies (in their final dataset), including 70 treatment groups and a total of 713 dogs, Bermingham et al. ([43], Table 3) found an average body weight for dogs of 20.1 kg, and an average MER of 1,351 kcal/day. Similarly, with respect to cats, after systematically reviewing 42 publications describing studies of cats (with 115 treatment groups included in their final analysis), Bermingham et al. ([45], Table 2) found an average body weight for domestic cats of 4.1 kg, and an average MER of 222·1 kcal/day. These figures were used in conjunction with population totals for dogs and cats, globally and within the US, to calculate total annual dietary energy requirements.

Energy requirements for average men and women were taken from the UK Government Dietary Recommendations [46], using the age bracket with the greatest energy requirements: 19–64 years of age. This yielded the maximum energy requirements of humans compared to dogs and cats, and hence, the most conservative study results. These daily energy requirements were: men– 2,500 kcal, women– 2,000 kcal. These figures were used in conjunction with population totals within the US and globally, to calculate total annual dietary energy requirements for people.

Total energy from animal sources (EA), consumed by dogs, cats and people.

Next, utilising dietary energy proportions attributable to animal sources within pet food [7], the total dietary energy provided by animal sources was calculated for dogs and cats. For humans, the dietary energy attributable to animal sources was calculated using Food and Agriculture Organisation of the United Nations (FAO) data, for the US in 2020, and globally in 2018 [47]. These FAO data reflected food supplied rather than consumed, and the dietary proportion of animal products consumed was presumed to be equal to the dietary proportion supplied. That is, (in the absence of data to the contrary), the proportions of losses, wastage and overconsumption after supply, were assumed to be equal between the animal and non-animal dietary fractions [47]. These dietary energy proportions supplied by animal sources were then applied to the dietary energy required annually by dogs, cats and people, within the US and globally, to quantify these E_A and E_non-A fractions.

Human-consumable (HC) and non-human-consumable (NHC) ingredients within dog and cat food.

The process used to analyse HC and HNC dietary fractions, or ingredient sets, was similar in each case. The HC ingredient set was examined first. The largest HC ingredient groups within dog and cat food were identified, along with their consumption levels compared to other HC ingredient groups. For each of these largest HC ingredient groups, the livestock species used were identified. For each species, the average proportion (by mass) of livestock animals (i.e., carcasses) that provided these ingredients was sourced from scientific literature, to establish the efficiency of these livestock species at providing these HC ingredients. Next, averages were generated, weighted by consumption levels of these different livestock species, to create overall weighted averages for the largest HC ingredient groups within dog and cat food. This represented the proportion of ‘average’ livestock animals that provided these HC ingredients. This indicated the efficiency of providing these HC ingredients. These largest HC ingredient groups were then used as proxies for all other ingredient groups within the HC ingredient sets, for both dog and cat food. ‘Organ meats’ were excluded from the meat groups within dog and cat food, as these derived from multiple species which were not specified. This meant that the ‘organ meat’ contributions within dog and cat food were effectively assigned the same weighted averages attributed to the rest of these meat groups.

This process was then repeated for the NHC ingredient sets within dog and cat food. In the subsequent step this allowed comparison of the overall efficiencies of average livestock animals at supplying the HC and NHC ingredient sets–or dietary fractions–within dog and cat food.

Proportionate livestock consumption by dogs, cats and people.

Collectively, the HC and NHC fractions comprised the animal-sourced ingredients used. The relative consumption of total animal-sourced dietary energy consumed by dogs, cats and people (calculated in an initial step), was combined with the relative consumption of average livestock animals required to collectively produce the animal-sourced (HC + NHC) fractions within these diets (calculated in the prior step), to calculate the proportionate consumption levels for dogs, cats and people.

When calculating proportionate livestock use globally, global averages for NHC and HC consumption proportions within pet food ingredients were used [51] rather than relying on the US pet food ingredients [49] analysis. These global averages differed from US averages, as a significantly higher proportion of NHC ingredients are used within pet food globally, compared to US pet foods.

Number of ‘food animals’ spared from slaughter.

The proportionate consumption of livestock animals by dogs, cats and people was applied to the numbers of terrestrial animals killed for food in the US in 2020, and globally in 2018 [52]. Next, this was applied to the numbers of aquatic animals estimated to have been killed to maintain the U.S. food supply in 2013 [53], and within the US and globally from 2016–2017 (the years available) [54]. This enabled determination of the numbers of animals who would no longer be slaughtered annually, were nutritionally-sound vegan diets instead used to feed dogs, cats and people.

Various environmental impacts.

As noted previously, the calories supplied by pet food are comprised of E_A and E_non-A fractions, and these proportions vary between dog and cat food. Transition to vegan pet food would result in no change in environmental impacts for the existing E_non-A fraction. However, impacts would change for the E_A fraction. To determine relative environmental impacts of animal- vs plant-based ingredients that could be consumed if dogs and cats transitioned on to vegan diets, two data sources were used.

In 2018 Poore and Nemecek provided calculations of a range of environmental impacts associated with the production of 52 plant- and animal-sourced food ingredients, using 2009–2011 averages [55]. They calculated land and water use (freshwater and stressed water–see following), GHG emissions as CO₂ equivalents, acidifying emissions as SO₂ equivalents, and eutrophifying emissions as PO₄^3- equivalents. The components included within these are indicated in S1 Table in S1 File.

For GHGs, IPCC [56] AR5 100-year characterisation factors were used, which are the most commonly-used indicators of the impacts of GHGs on the mid- to long-term climate. These data also included direct and indirect impacts of GHGs, and climate-carbon feedbacks–the effects of climate change on factors affecting CO₂ emission, such as the land and ocean carbon cycles, and radiative forcing. Data on the acidification and eutrophication emissions relied on CML2 baseline method characterisation factors [57]. Scarcity-weighted freshwater withdrawals relied on the WULCA consensus characterization model for water scarcity footprints (AWARE), which quantifies the relative available water remaining per area (water scarcity or stress–Str-Wt), once the demand of humans and aquatic ecosystems has been met. The resulting characterization factor from 0.1–100 indicates the potential to deprive another user (human or ecosystem) when consuming water in an area [58].

Poore and Nemecek’s data [55] quantifying the environmental impacts of these 52 plant- and animal-sourced food ingredients were examined. A small number of ingredients were excluded due to uncertainty about whether these were entirely plant- or animal-based. For dog and cat diets, ingredients unlikely to be used within animal- or plant-based diets for these species, were also excluded, and the remainder were divided into animal- and plant-based ingredients. Production volumes were supplied for all ingredients based on 2009–2011 averages, including amounts for food and food waste. Production volumes including non-food purposes were also supplied but not used, as these included uses such as biofuel and textiles (e.g., leather) production, rather than ingredients that are, or could be, consumed by dogs, cats or people.

Based on the production volumes for food and food waste, weighted averages were derived for these ingredient sets, for all of the above environmental impact categories. Ratios for the relative impacts of plant- versus animal-based ingredient consumption were then calculated (‘W’ in the following). This process was then repeated to determine the same relative environmental impacts, for human diets. In this case, the ingredients unlikely to be used within dog or cat diets, were included, as these are used within human diets.

Additionally, Reijnders and Soret [59] provided the relative impacts of meat protein production compared to plant protein production, for a range of environmental sustainability parameters. Most were superseded by the more recent Poore and Nemecek data [55], but Poore and Nemecek did not provide data for biocide use. Hence Reijnders and Soret’s ratio for biocide use was also included.

When switching to vegan pet food–i.e., replacing all animal-sourced calories, with plant-based ingredients, the impacts due to the E_A fraction decrease–not to 0, but to 1 –which is the relative impact if plant-based ingredients are used instead. Hence the reduction in impact through switching all animal-sourced calories to vegan ingredients (alternatively, the increase in impact accruing through use of animal-based ingredients), is:

W = livestock production impacts due to the E_A fraction

j = environmental impact category: land use, water use, GHG emissions, acidifying emissions, eutrophifying emissions or biocides

E_A = proportion of dietary energy derived from animal sources

The E_A values for dog, cat and humans diets, calculated previously, were then used to calculate these reductions in impact for all categories ‘j’. These were then added to the relative impacts of vegan diets (1) to determine total impacts associated with meat-based diets. The percentage reductions that would be achieved by replacement with vegan diets were then calculated. Finally, these percentage impact reductions within each diet, were multiplied by the proportions of total livestock consumption attributable to the diets of dogs, cats and humans respectively, both within the US and globally, to determine the reductions in total livestock sector impacts that would be expected after transitioning to vegan diets.

Additional people, dogs and cats who could be fed using food energy savings.

As noted, the calories supplied within pet food come from two dietary fractions: E_A and E_non-A. For the existing E_non-A fraction, no excess energy would result from transitioning to a vegan diet, as this fraction would not change. However, excess dietary energy is available within the E_A fraction, because most of the plant calories fed to livestock animals are used to support their bodily growth and maintenance processes, rather than directly producing consumable products [35]. After considering the average American consumption of beef, pork, poultry, other meats including fish, milk and eggs, Pimentel and Pimentel [1] reported that for every 1 kg of high-quality animal protein produced, livestock animals are fed about 6 kg of plant protein, which are produced, in turn, from many additional kg of grain and forage.

When considering the average loss-adjusted feed conversion ratio for beef+lamb, pork, and poultry, weighted by their relative availability in the diets of American people [60], Okin [7] determined that 4.7 J of plant energy were required to produce 1.0 J of meat energy. For the purposes of this study, this was generalised to all HC animal-sourced ingredients. Hence, on average 3.7 J of energy were considered to be lost during conversion from plant to HC animal-sourced ingredients. These 3.7 J of excess dietary energy could instead be freed for direct consumption as plant-sourced ingredients, when using a vegan diet.

As noted previously, the efficiencies of average livestock animals, at providing the HC and NHC dietary fractions, differed by an efficiency factor (EF), which was different for dog and cat food. For the less efficient dietary fraction, correspondingly more livestock animals were required, further reducing the efficiency of conversion from plant energy below 1.0/3.7. The differences in the numbers of livestock animal required, correspond to the EF multiples calculated previously. For example, as noted for dogs, L_dogs = (CF x HC fraction) + (CF x NHC fraction x EF_dogs). Hence, for the less efficient HC or NHC dietary fraction, conversion to plant energy decreases in efficiency, by these EF same multiples. Hence, the excess dietary energy freed via direct consumption of plant ingredients, increases by these factors.

Accordingly, the excess dietary energy that would be available, were plant sources consumed directly instead of converting them to HC and NHC animal-sourced ingredients, for dog, cat and human diets, was calculated as follows. For human diets, the NHC fraction = 0.

These dietary food energy savings were calculated, and then compared to the annual dietary energy requirements of US people in 2020 (calculated in an earlier step), to determine the number of additional Americans who could be fed by consuming this energy directly in the form of plant-based ingredients, i.e., within a vegan diet.

These steps were repeated using E_A consumption for dogs, cats and humans globally. In this case global (rather than US) averages for NHC and HC consumption proportions within pet food ingredients were used as noted previously [51]. These dietary food energy savings were calculated, and then compared to the annual dietary energy requirements of all people globally in 2018, to determine the number of additional people who could be fed by consuming this energy directly, i.e., within a vegan diet.

Finally, the dietary food energy savings within dog and cat food were compared to the annual dietary energy requirements of these species, to determine the numbers of additional animals of the same species who could be fed if dogs and cats were transitioned on to nutritionally-sound vegan diets. These calculations were also performed considering both the 2020 US and 2018 global populations.

Within various data tables, data were often displayed as rounded values, e.g., rounded to one decimal place. However, where these data were used within subsequent calculations, the underlying exact values were often used, to achieve greater accuracy.

Total energy from animal sources (EA), consumed by dogs, cats and people.

Okin [7] analysed premium dog food (n = 102), market-leading dog food (n = 9), premium cat food (n = 163), and market-leading cat food (n = 9) products within the US. He examined the mass of the five ingredients listed first within these pet foods (with each assumed to be virtually 20% by weight), and the energy density of these ingredients. He estimated the total fraction of calories derived from animal-based ingredients (E_A) to be 34% ± 4% for dog food, and 31% ± 4% for cat food. Applying these proportions to the dietary energy required annually by dogs and cats within the US and globally, gave the E_A amounts in Tables 3 and 4. The remaining dietary energy was derived from non-animal sources (E_non-A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Proportion of dietary energy attributable to animal and non-animal sources, in the diets of US dogs, cats and humans in 2020.

https://doi.org/10.1371/journal.pone.0291791.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Proportion of dietary energy attributable to animal and non-animal sources, in the diets of dogs, cats and humans globally, in 2018.

https://doi.org/10.1371/journal.pone.0291791.t004

Considering human diets, within the US in 2020, an average of 3,926 kcal were supplied daily. 1,125 kcal (28.7%) of this came from animal produce. Globally in 2018, an average of 2,961 kcal were supplied daily, of which 553 kcal (18.7%) were from animal produce [47]. These were assumed to reflect the E_A proportions consumed within human diets in the US and globally. Applying these proportions to the dietary energy required annually by people resulted in the E_A and E_non-A amounts in Tables 3 and 4.

Human-consumable (HC) ingredients within dog and cat food.

The HC ingredient groups within dog and cat food were meat (including poultry and organ meats), fats and oils, broth, fishery ingredients, and dairy and eggs (Table 5). Of these, meat was the largest group. For dog food, HC ingredients comprised 47.4% of all animal-sourced ingredients, and meat comprised 66.2% of this HC group. For cat food, HC ingredients comprised 49.2% of all animal-sourced ingredients, and meat comprised 50.0% of this HC group.

The meat used within US pet food was chicken, organ meat, beef, turkey, lamb, poultry, pork, duck, venison and bacon. The proportions normally derived from carcasses of the source species, and their levels of inclusion within dog and cat food, are given in Tables 6 and 7. The proportionate use of these species differed between dog and cat food, resulting in different weighted averages for meat per average carcass. For dog and cat food respectively, these weighted averages were 53.0% and 58.8%. As noted, meat was the largest ingredient group among all HC ingredients used within dog and cat food, and these meat weighted averages were used as proxies for all HC ingredients within these diets.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Meat proportions within carcasses of animal species used within dog food.

https://doi.org/10.1371/journal.pone.0291791.t006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 7. Meat proportions within carcasses of animal species used within cat food.

https://doi.org/10.1371/journal.pone.0291791.t007

Non-human-consumable (NHC) ingredients within dog and cat food.

The NHC ingredient groups within dog and cat food were animal meal (which is derived from ABPs), ABPs and ‘other’ (digest flavourant and animal plasma) (Table 5). Of these, animal meal was the largest group. For dog food, NHC ingredients comprised 52.6% of all animal-sourced ingredients, and animal meal comprised 88.1% of this NHC group. For cat food, NHC ingredients comprised 50.8% of all animal-sourced ingredients, and animal meal comprised 74.0% of this NHC group.

The meat meal was derived from ABPs of production of the following meats: unspecified (meat and bone), chicken, unspecified (poultry), beef and bone, unspecified (fish), lamb, salmon, turkey, pork and tuna. The proportions normally derived from carcasses of the source species, and their levels of inclusion within dog and cat food, are given in Tables 8 and 9. The proportionate use of these species differed between dog and cat food, resulting in different weighted averages for ABPs per average livestock carcass. For dog and cat food respectively, these weighted averages were 39.2% and 31.3%. As noted, animal meal was the largest ingredient group among all NHC ingredients used within dog and cat food, and these weighted averages were used as proxies for all NHC ingredients within these diets.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 8. Meat meals used as ingredients within dog food.

https://doi.org/10.1371/journal.pone.0291791.t008

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 9. Meat meals used as ingredients within cat food.

https://doi.org/10.1371/journal.pone.0291791.t009

Proportionate livestock consumption by dogs, cats and people, within the US in 2020.

The E_A dietary fractions required by dogs (60.6 PJ), cats (6.4 PJ) and humans (324.1 PJ) in the US in 2020 were given in Table 3. For humans, as noted all of these animal-sourced ingredients were HC. Hence, the total number of average livestock animals required to supply the animal-sourced dietary energy within these human diets was:

For dog food, the E_A dietary fraction was comprised of (HC: 47.4% = 28.7 PJ) + (NHC: 52.6% = 31.9 PJ) = 60.6 PJ. Hence, the total number of average livestock animals required to supply the animal-sourced dietary energy within dog food was:

For cat food, the E_A dietary fraction was comprised of (HC: 49.2% = 3.1 PJ) + (NHC: 50.8% = 3.3 PJ) = 6.4 PJ. Hence, the total number of average livestock animals required to supply the animal-sourced dietary energy within cat food was:

Hence, the consumption of average livestock animals to supply the animal-sourced dietary energy required by US dogs, cats and humans in 2020, was 17.7% for dogs, 2.3% for cats, 80.0% for humans, and 20.0% for dogs and cats jointly (Table 10).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 10. Proportionate use of average livestock animals required to meet animal-sourced dietary energy needs, within US dog, cat and human diets in 2020.

https://doi.org/10.1371/journal.pone.0291791.t010

Proportionate livestock consumption by dogs, cats and people, globally in 2018.

Similarly, the E_A dietary fractions required by dogs (330.4 PJ), cats (39.1 PJ) and humans (4,940.8 PJ) globally in 2018 were given in Table 4. For humans, as noted, all of these animal-sourced ingredients were HC. Hence, the total number of average livestock animals required to supply the animal-sourced dietary energy within these human diets was:

As noted, for US dog and cat food, NHC components comprised 52.6% and 50.8% of all animal-sourced ingredients, respectively. In comparison, the global consumption of meat meal, ABP meal and animal digest within pet food (comprising all NHC ingredients) in 2019 (the closest available year to 2018), comprised 16,416.3 kT, or 74.9% of the 21,904.5 kT total meat and meat products consumed within pet food (T = US ton) [51]. Separate figures for dog and cat food were not available; hence this 74.9% average was applied equally to dog and cat food consumed globally in 2018.

For dog food, the E_A dietary fraction was comprised of (HC: 25.1% = 82.9 PJ) + (NHC: 74.9% = 247.5 PJ) = 330.4 PJ. Hence, the total number of average livestock animals required to supply the animal-sourced dietary energy within dog food was:

For cat food, the E_A dietary fraction was comprised of (HC: 25.1% = 9.8 PJ) + (NHC: 74.9% = 29.3 PJ) = 39.1 PJ. Hence, the total number of average livestock animals required to supply the animal-sourced dietary energy within cat food was:

Hence, the consumption of average livestock animals to supply the animal-sourced dietary energy required by dogs, cats and humans globally in 2018, was 7.7% by dogs, 1.2% by cats, 91.1% by humans and 8.9% by dogs and cats jointly (Table 11).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 11. Proportionate use of average livestock animals required to meet animal-sourced dietary energy needs, within dog, cat and human diets globally in 2018.

https://doi.org/10.1371/journal.pone.0291791.t011

Number of ‘food animals’ spared from slaughter.

Terrestrial animals. Transition to nutritionally-sound vegan diets would no longer require the slaughter of livestock animals for food. Given the proportionate consumption of average livestock animals within the diets of dogs, cats and humans, this would spare billions of terrestrial animals from slaughter annually, within the US and globally (Tables 12 and 13).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 12. Terrestrial animals killed for food in 2020, within the US, used within the diets of dogs, cats and humans.

Totals were sourced from FAOSTAT [52].

https://doi.org/10.1371/journal.pone.0291791.t012

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 13. Terrestrial animals killed for food in 2018, globally, used within the diets of dogs, cats and humans.

Totals were sourced from FAOSTAT [52].

https://doi.org/10.1371/journal.pone.0291791.t013

Aquatic animals. Aquatic animal deaths are challenging to calculate because their numbers are provided as tonnages. Harish [53] calculated the numbers of finned fish, shellfish, ‘feedfish’ (used within animal feed, primarily for livestock animals), and bycatch aquatic animals (killed within capture fisheries), that were collectively killed to maintain the U.S. food supply in 2013 (Table 14). Total U.S. fish landings reportedly remained consistent at these levels, at least through 2018. Using FAO and other sources, Fishcount.org provided similar data globally, per nation and per species (Tables 15 and 16). As demonstrated by Harish [53] (Table 14), vast numbers of ‘feedfish’ and bycatch aquatic animals were not reflected within fisheries, aquaculture and decapod numbers (Table 15).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 14. Aquatic animals killed for food in 2013, within the diets of US dogs, cats and humans (billions).

Data: Harish [53].

https://doi.org/10.1371/journal.pone.0291791.t014

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 15. Fish and decapods consumed annually within the diets of US dogs, cats and humans (billions).

Data: fishcount.org [54].

https://doi.org/10.1371/journal.pone.0291791.t015

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 16. Fish and decapods consumed annually within the diets of dogs, cats and humans, globally (billions).

Data: fishcount.org [54].

https://doi.org/10.1371/journal.pone.0291791.t016

The proportions of aquatic species used within US dog and cat food respectively were 2.8% and 15.6% by mass (combining HC and NHC aquatic species and excluding animal-sourced ingredients from unspecified species) [49]. Because actual consumption levels were determined by E_A and by carcass provision of HC:NHC components–and because the latter were not 1:1, true consumption levels cannot be directly discerned from these 2.8% and 15.6% proportions. The proportion of overall consumption would also depend on human consumption levels. Nevertheless, if in excess of just 1% of overall consumption–as appears likely, this would equate to billions of aquatic animals being consumed within dog and cat food annually, in the US alone.

Various environmental impacts.

As described within the Methodology, data on plant- and animal-sourced food ingredients provided by Poore and Nemecek [55] were examined. ‘Oils misc.’ and ‘sweeteners and honey’ were excluded due to uncertainty about whether these were entirely plant- or animal-based. Collectively these totalled only 0.7% by weight of these 52 ingredients.

When considering dog or cat diets, eight plant- and three animal-based ingredients or ingredient groups were excluded from further analysis as they were unlikely to be used within these diets (after considering the ingredients used within dog and cat diets [49]; e.g., sweeteners and spices (S18 Table in S1 File).

Weighted averages (based on production volumes) for the remaining 29 plant- and 12 animal-based ingredients were calculated, for each of the environmental impacts calculated by Poore and Nemecek [55]: land and water use, GHG emissions as CO₂ equivalents, acidifying emissions as SO₂ equivalents, and eutrophifying emissions as PO₄^3- equivalents (S19 Table in S1 File). The ratios for the relative impacts of plant- versus animal-based ingredient consumption, are provided in S20 Table in S1 File. For example, for land use, the relative impact (W) of an animal-based:vegan diet = 18.911:1. As described in the Methodology, Reijnders and Soret’s ratio for biocide use [59] is also included in S20 Table in S1 File.

As noted, when considering human diets, the ingredients unlikely to be used within dog or cat diets (S18 Table in S1 File–‘Excluded’), were included, as these are used within human diets. The corresponding environmental impacts for human diets, are also provided in S20 Table in S1 File.

The relative environmental impacts in all categories (other than biocides–which did not rely on these ingredient calculations) were markedly higher for dog and cat food, compared to human diets (S20 Table in S1 File: ‘Relative impact: dog or cat (W)/human (W)’). And yet, these calculations did not account for different consumption levels of ingredients between dog, cat and human diets (other than exclusion of certain ingredients from dog and cat food, as noted in S18 Table in S1 File). Hence, this significantly underestimates the true differences, because a higher proportion of dog and cat diets (34.0% and 30.9% of calories respectively), are supplied by animal sources (which have greater environmental impacts), compared to human diets (18.7 or 28.7% of calories, globally or in the US, respectively) (Tables 3 and 4). Hence the relative impacts of dog and cat diets, compared to human diets, are actually significantly greater than indicated in S20 Table in S1 File.

Nevertheless, using those very conservative relative impacts, along with E_A values for dog, cat and human diets (Tables 3 and 4), the impact reductions within each diet, achievable through transition to vegan diets, were calculated as described in the Methodology. For dogs + cats, the US figure of 0.337 (33.7% in Table 3) was used rather than the global figure of 0.336 (33.6% in Table 4), as the underlying data for US pet food consumption was most likely to be accurate. For dogs and cats considered individually, the US and global figures were identical (dogs: 34.0%, cats: 30.9%, in Tables 3 and 4). These impact reductions were therefore:

These impact reductions are also provided in S20 Table in S1 File. For example, for biocides, the additional impact within the dog food diet, accrued by animal-based ingredients, is (6.000–1) x 0.340 = 1.700, compared to a vegan diet with an impact of 1, creating a total impact for meat-based dog food of 2.700. The reduction of biocide impacts achieved via vegan dog food is 1.700/2.700 = 63.0%.

Given these impact reductions associated with vegan diets, and considering the relative proportions of livestock consumption required to supply the E_A within the diets of dogs, cats and humans (Tables 10 and 11), the reductions in total livestock sector impacts in each category, achieved through use of vegan diets, were calculated for the US (2020 consumption levels) and globally (2018 consumption levels) (Tables 17 and 18).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 17. Reductions in total livestock sector impacts within the US, achieved through use of vegan diets for dogs, cats or humans, based on 2020 consumption levels.

https://doi.org/10.1371/journal.pone.0291791.t017

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 18. Reductions in total livestock sector impacts globally, achieved through use of vegan diets for dogs, cats or humans, based on 2018 consumption levels.

https://doi.org/10.1371/journal.pone.0291791.t018

The proportions above can be applied to a range of livestock sector impacts, to illustrate the benefits likely to accrue from transitions to vegan diets for dogs, cats and people. Examples follow for land and freshwater use, and GHG emissions.

Land use. In 2006, Steinfeld et al. [2] noted that 78% of the world’s agricultural land, and 33% of the world’s cropland, is used for livestock production. Since then, livestock numbers have increased significantly. Hence, Poore and Nemecek [5] calculated that meat, aquaculture, eggs and dairy production utilised around 83% of the world’s agricultural land. The more conservative 2006 figures alone, indicate that livestock grazing and feed crop production uses 3.9 billion ha (hectares) of land, or 30% of the non-polar terrestrial surface of the Earth. Hence, considering global consumption levels, at least the following land savings would result from vegan diets (in billion ha): dogs– 0.26 (larger than Saudi Arabia or Mexico), cats– 0.04 (larger than countries such as Japan or Germany), dogs and cats– 0.30 (larger than Argentina), humans– 2.36 (larger than Russia–the world’s largest country–combined with India) [65].

Additionally, livestock are often major sources of pollution, releasing large quantities of organic matter, pathogens and drug residues on to soil and in to rivers, lakes and coastal zones [66–68]. The 100+ million cattle produced in the US annually each generate an average of 9,000 kg of solid waste per year [67]. Livestock impacts landscapes, often profoundly diminishing biodiversity. The Amazon rainforest is among the world’s most biodiverse ecosystems. Around 70% of the previously forested Amazonian land has been converted to pastures, with much of the remaining 30% converted to croplands, largely for livestock feed [2]. Vegan diets would free up vast amounts of land, allowing rewilding and biodiversity recovery.

Freshwater use. The water used by the livestock sector exceeds 8% of global human water use [69]. Global animal production requires about 2,422 Gm³ of water per year (87.2% green, 6.2% blue, and 6.6% grey water). The green water footprint derives from precipitation. Blue water is sourced from surface or groundwater, and grey water is fresh water required to assimilate pollutants to meet water quality standards. One third of this volume is consumed by the beef cattle sector, and another 19% by the dairy sector. Almost all (98%) of water consumed is required to grow feed crops. Drinking water for the animals, service water and water for feed mixing, require only 1.1%, 0.8% and 0.03% of these water volumes, respectively [70]. Freshwater is encapsulated by the blue and grey water components. Globally, this freshwater used for animal production comprises 310.01 Gm³. Hence, considering global consumption levels, freshwater use reductions achieved by vegan diets would be (in Gm³): dogs– 7.75 (greater than all renewable water in Denmark), cats– 1.24 (greater than all renewable water in Jordan), dogs and cats– 8.99 (greater than all renewable water in Gambia), and humans– 42.47 (greater than all renewable water in Cuba) [71–73].

Greenhouse gases. Anthropogenic GHGs created by the livestock sector are second only to those created by the energy sector [69]. Livestock-associated GHGs come from deforestation for pasture and feed crops, pasture degradation, and from direct emissions from livestock and their waste products.

The main GHG emissions associated with livestock production are CO₂, CH₄, and N₂O. Of these, 19% of CH₄ emissions come from the livestock sector. Enteric fermentation and manure collectively contribute 80% of the methane emissions [74, 75]. Of next greatest importance is N₂O. Livestock production contributes 15% of N₂O emissions. Finally, livestock production contributes 1.35% of CO₂ emissions [76]. The global warming potential of these gases varies greatly. The IPCC [56] reported a warming potential for CH₄ of 34 CO₂-eq, and for N₂O of 310 CO₂-eq, over a 100 year timeframe. The equivalent figures reported by the UNFCCC [77] for CH₄ were 21 CO₂-eq, and for N₂O were (also) 310 CO₂-eq.

The food system results in 35% of all GHGs globally, and 57% of all food sector emissions come from livestock, resulting in a total of 20% of all GHGs–or 9.8 Gt CO₂-eq–from livestock [3]. Hence, reductions in total anthropogenic GHGs achieved by vegan diets globally would be 20% of the reductions shown in Table 18, given that Table 18 relates only to those impacts attributable to the livestock sector. As percentages of all anthropogenic GHGs, these would represent reductions of: dogs– 1.2%, cats– 0.2%, dogs and cats– 1.3%, and humans– 8.4%.

Considering the 9.8 Gt CO₂-eq from livestock, and the reductions achieved by vegan diets shown in Table 18, these would equate to GHG emissions savings, in Gt CO₂-eq, of: dogs– 0.57 (greater than all emissions from South Africa or the UK), cats– 0.09 (greater than all emissions from Israel or New Zealand), dogs and cats– 0.66 (greater than all emissions from Saudi Arabia or Australia), and humans– 4.13 (greater than all emissions from India or the entire EU). These refer to the total GHG emissions used for the production of all goods and services in these nations or regions, based on 2018 figures [78].

Additional people who could be fed using food energy savings.

Within the US in 2020. As noted in the Methodology, an average of 4.7 J of plant energy were required to produce 1.0 J of energy from HC animal-sourced ingredients. The remaining 3.7 J of dietary energy was lost during conversion from plant- to animal-based ingredients. Within a vegan diet, this lost energy would have been available for direct consumption in the form of plant-based ingredients.

Furthermore, as noted previously, just over half of the animal-sourced ingredients within dog and cat food were supplied by NHC components. For dog food this proportion was 52.6%, and for cat food it was 50.8%. As calculated previously, the numbers of livestock animals required to provide the animal-sourced NHC fractions for dogs and cats, were respectively 1.352 and 1.879 times the numbers required to provide equivalent dietary energy as HC components. Hence, for the NHC dietary fraction, conversion to plant energy decreased in efficiency by these factors, and the excess dietary energy potentially freed via direct consumption of plant ingredients, would have increased by these factors.

Accordingly, the excess dietary energy that would be available, were plant sources used instead of all HC and NHC animal-sourced ingredients, for dog, cat and human diets within the US in 2020, would be as follows. The E_A quantities were provided in Table 3. For human diets, the NHC fraction = 0.

Given this excess dietary energy, and recalling that a million US people can be fed per 3.43 PJ of dietary energy (Table 1), the numbers of additional people that could be fed through consuming this energy directly in the form of plant-based ingredients (i.e., within a vegan diet), are provided in Table 19.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 19. Proportion of the 2020 US human population who could be fed with food energy savings associated with vegan diets.

https://doi.org/10.1371/journal.pone.0291791.t019

Hence, compared to using vegan diets to feed American people, the use of nutritionally-sound vegan dog food would free sufficient food energy to feed 0.22 times as many Americans. Nutritionally-sound vegan cat food would free sufficient food energy to feed 0.03 times as many, and use of vegan dog and cat food combined would free sufficient food energy to feed 0.25 times as many Americans–i.e., one quarter of the number of Americans who could be fed using the food energy saved, if all American people transitioned on to vegan diets.

Globally in 2018. Considering dog and cat food globally in 2018, as noted previously the NHC and HC proportions for both were considered to be 74.9% and 25.1% [51]. Given this, the excess dietary energy that would be available, were plant sources used instead of all HC and NHC animal-sourced ingredients, for dog, cat and human diets, would be as follows. The E_A quantities were provided in Table 4. For human diets, the NHC fraction remains 0.

Given this excess dietary energy, and recalling that a million global citizens could be fed per 3.44 PJ of dietary energy (Table 2), the numbers of additional people that could be fed through consuming this energy directly in the form of plant-based ingredients (i.e., within a vegan diet), are provided in Table 20.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 20. Proportion of the 2018 world human population who could be fed with food energy savings associated with vegan diets.

https://doi.org/10.1371/journal.pone.0291791.t020

Hence, compared to using vegan diets to feed people globally, the use of nutritionally-sound vegan dog food would free sufficient food energy to feed 0.08 times as many people. Nutritionally-sound vegan cat food would free sufficient food energy to feed 0.01 times as many, and use of vegan dog and cat food combined would free sufficient food energy to feed 0.10 times as many people–i.e., one tenth as many people who could be fed using the food energy saved, if all people globally transitioned on to vegan diets.

Dietary energy requirements of dogs, cats and people.

For human populations within the US and globally, energy requirements for average men and women aged 19–64 were applied [46]. These were the male and female categories with the greatest, or equal greatest, energy requirements. Actual requirements among people vary based on demographic differences in age, sex, body weight, climate, exercise level, medical conditions and other factors. On average, actual male and female dietary energy requirements will normally be lower than those used in this study, meaning that human dietary energy needs have been over-estimated, compared to those of dogs and cats. This also means that actual consumption of livestock by dogs and cats will be greater than the proportions conservatively estimated within this study, and that the sustainability benefits of nutritionally-sound vegan canine and feline diets, are greater than those calculated.

The FAO data [47] revealed significant differences in the levels of animal produce consumption within human diets (28.7% in the US, versus 18.7% globally). This is consistent with much higher animal produce consumption within high income nations, compared to lower income regions [83]. This is consistent with the lower proportion of NHC ingredient consumption within US pet food compared to more expensive HC ingredients, than was identified globally.

Animal by-product use within society.

Until recently, accurate information on the level of NHC animal-based ingredients within pet food has been sparse. In 1997, Halpin et al. [63] surveyed large petfood manufacturers. They reported that meat by-products comprised around 25–40% of dog foods, and 35–50% of cat foods. Within the current study using 2018–2019 data sourced from 68.3% of US retail pet food sales, NHC sources (primarily, ABPs), provided 52.6% of dog food ingredients, 50.8% of cat food ingredients, and 52.1% of dog and cat food ingredients overall.

It has often been assumed that the use of ABPs within pet food effectively recycles by-products of the human food production system that would otherwise be wasted (e.g., [21, 84])–i.e., that this is environmentally beneficial. One noteworthy finding of this study, is that this assumption has been incorrect.

This study found that NHC sources were less efficient than HC sources, requiring more livestock animals to produce– 1.352 times as many, for dog food, and 1.879 times as many, for cat food. This is consistent with a study by Rushforth and Moreau [85], who found that using lean meat within dog food was better—in terms of environmental impacts—than using offal.

Rather than being wasted, if not consumed within pet food, all meat ingredients, ABPs and their derivatives, would normally be consumed either directly by people, or within other sectors of society [86, 87] (Fig 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Main social applications of animal by-products.

After Toldrá et al. [88].

https://doi.org/10.1371/journal.pone.0291791.g004

ABPs account for the majority of slaughtered animal carcases for agricultural species such as cattle (66%), pigs (52%) and lambs (68%) [64]. Around two thirds of these ABPs are directly edible by humans [89]. Organs such as the liver, kidney, heart, brain, intestine, tongue and spleen are HC, and are also termed ‘organ meats’ or ‘variety meats’ [64]. The great majority of animal-sourced material is edible if cleaned, handled and processed appropriately. In developing nations, most of the soft tissues are consumed by people. These include livers, hearts, brains, lungs, the thymus and pancreas, testicles, tongue and gizzard, etc. Classification as HC or NHC often depends on cultural factors such as purchasing power and economics, custom, tradition, food habits, hygiene, availability and religious beliefs of consumers.

ABPs usually considered inedible by humans include hides, skins, ears, snouts, gallbladders, foetuses, hoofs, horns, hair and bristles, etc. Some apparently NHC ingredients are actually used within the human food industry (e.g., edible tallow, blood sausages or pudding, sausage skins, gelatin and defatted meat tissue). Hence, such ingredients are actually HC [63]. Furthermore, many initially inedible ABPs may be converted to edible products through technological innovations. For example, poultry feathers and heads, skin trimmings, fish scales, horns and hooves can all be converted into protein hydrolysates through acidic/alkaline/enzymatic hydrolysis. These protein hydrolysates are used as protein fortifying agents for concentrated soups and beverages, and within solid and liquid seasoning [64].

ABPs may also be classified into principal by-products–directly harvested from the animals, e.g., hides and skins, bones, blood, hoofs and horns, and secondary by-products–derived from these, e.g., bone meal, fat, intestinal linings, etc. ABPs may be further converted via rendering into meat meals and fats. Meat meal is the major secondary by-product produced and is an important components of livestock feeds for pigs and poultry [64]. Bone meal is also important within livestock feeds [90].

ABPs may also be used to create a wide variety of industrial, consumer and medical products. These may include clothing, carpets, blankets, upholstery, rubber, adhesives, lubricants, abrasives, paints, pesticides and fertilizers, soaps, other cosmetics and personal care products, shampoos, detergents, foaming agents and musical strings. They also include a variety of medical products such as pharmaceuticals, surgical sutures, prosthetic materials, collagen sheets, burn dressing, dialyzing membrane, heparin, numerous exogenous hormones, and others [63, 64, 91]. Even bones are used to create a wide variety of products, including tallow, dicalcium phosphate, bone meal, glue and gelatine, and bone morphogenic protein for use within human facial, dental and aesthetic surgeries [64].

Even parts which are contaminated and decomposed may be suitable for products such as fertilizers and soil conditioners. Components such as urine, faeces, ruminal contents, blood, meat and fat trimmings, can be used to create biogas, which may then be burnt to help power abattoirs, power stations or other facilities [64]. Such use of animal parts within energy production may be set to increase further, with ABPs potentially being used within sustainable jet fuel. Forthcoming European legislation could require a majority of aviation fuel to be sustainably sourced. Such developments could lead to scarcity of ABPs for use within pet food [91].

In fact, very little of any animal carcass is wasted. Hence the slaughtering industry colloquialism that, “the packer uses everything but the squeal” [64]. The pet food industry is, in fact, a minority user of animal-based ingredients. Halpin et al. [63] estimated that only approximately 25% of all ABPs produced in the US are used within pet foods.

In short, ABPs and their derivatives are used within pet food as protein sources, because they’re considerably cheaper than HC ingredients such as meat. This is not done to ‘recycle’ produce that would otherwise be wasted. Were animal-based ingredients not used within pet foods, they would be consumed in a wide variety of other social sectors. Their consumption within petfood increases overall demand for ABPs–and hence, the number of livestock animals required to provide them.

Various environmental impacts.

The reductions of various environmental impacts associated with the livestock sector, that could be achieved through transition to nutritionally-sound vegan diets for dogs, cats and people, were shown in Tables 17 and 18. Although the relative numbers of livestock animals required to fulfil the E_A needs of humans was much greater than those of dogs and cats (Tables 10 and 11), the diets of dogs and cats have much higher proportions of animal products (Tables 3 and 4), which increases their relative environmental impacts. Accordingly, whilst the greatest reductions in environmental impacts are achievable through transition of humans to vegan diets, the benefits achieved by transitioning dogs in particular, often appear around one quarter to one third of the benefits that could be achieved through human dietary change, at least in the US (Table 17).

At global consumption levels (Table 18), the benefits achieved by vegan pet food reduce, due to lower per capita levels of pet guardianship when compared to the US. This is partly offset due to the higher use of NHC ingredients (74.9%) globally, compared to the US (∼50%). As demonstrated previously, greater NHC use requires more average livestock animals, increasing environmental impacts. Hence, environmental impact reductions achieved by vegan diets for dogs in particular are still significant, compared to reductions achieved by vegan diets for humans. They generally achieve between one fifth and one tenth of the latter effect.

Consistency with prior studies. The environmental impacts of dog and cat food demonstrated within this study were very considerable. This concurs with results of other studies within this field. Okin [7] calculated that pet food was responsible for 25–30% of the environmental impacts of the livestock sector within the US, such as the use of land, water, fossil fuels, eutrophifying phosphates, and biocides. The current study estimated the proportion of livestock consumption–and hence livestock-associated environmental impacts–attributable to the diets of US dogs and cats collectively, to be 20.0%. Key differences between these studies are that Okin did not account for the proportion of NHC ingredients within dog and cat diets, and the inefficiencies of producing these ingredients–which requires more livestock–compared to HC ingredients. Additionally, Okin calculated the E_A of humans was 20%, but used only data for red meat, poultry and fish, published in 2012. This data excluded animal produce such as eggs and cheese, which are included within diets of humans, dogs and cats [49], and milk, which is included within human diets, and is associated with substantial environmental impacts. After analysing the more complete FAOSTAT supply dataset [47], the current study calculated the E_A of US people in 2020, to be 28.7% (Table 3). Furthermore, when apportioning calories between pet and human diets, Okin used consumption data for people, but only energetic needs for dogs and cats. Due to excesses including losses, wastage and overconsumption, actual consumption levels for dogs and cats were therefore underestimated, compared to human levels, lowering estimations of the environmental impacts of pet food. Okin acknowledged this: “An important caveat for the calculations of the relative consumption of pets and humans is that the sources of the data, and mode of calculation, are dramatically different. As a result, their ratios may be systematically biased.” Nevertheless, Okin’s study was an important initial estimation of the environmental impacts associated with dog and cat diets. Okin also concluded that these were very substantial.

Su et al. [8] described the concept of the dietary “Ecological Paw Print” (EPP) for dogs and cats. This is equivalent to the human dietary “Ecological Footprint” (EF), and indicates how much productive land is required for an individual or population to maintain itself, and to process resultant waste. These are distinct from total paw- or footprints, which consider requirements for all activities, rather than just diets.

When considering the 27.4 million companion dogs and 58.1 million companion cats in China in 2015, Su et al. [8] calculated that the dietary EPP for all dogs and cats was 43.6–151.9 million ha. per year, or 0.82–4.19 ha per year for an average sized dog, and 0.36–0.63 ha per year for a cat. This was equivalent to the dietary EF of 5.1% - 17.8% (70.3–245.0 million) of the Chinese human population in 2015. The annual food consumption of all these dogs and cats was responsible for 2.4–7.5 million tons of carbon emissions, and equivalent to the dietary carbon emissions of 2.5% - 7.8% (34.3–107.1 million) of the Chinese population in 2015.

Similarly, when considering the over 20.3 million companion dogs and cats in Japan, Su and Martens [9] found that the dietary EPP of all dogs and cats was 6.6 million—28.3 million ha per year, comparable to the dietary EF of 4.62 million—19.79 million Japanese people. For an average-sized dog this was 0.33–2.19 ha per year–equivalent to one Japanese person’s dietary EF. The dietary EPP of an average-sized cat was lower, at 0.32–0.56 ha per year. The GHG emissions from Japanese dog and cat food consumption were 2.52 million—10.70 million tons, which was equivalent to the dietary GHG emissions of 1.17–4.95 million Japanese people.

With regard to Dutch companion dogs and cats, the dietary EPP of an average-size dog was 0.90–3.66 ha per year, whereas for a cat, it was between 0.40–0.67 ha per year. The dietary EPP of all Dutch companion dogs and cats was 2.9 million—8.7 million ha per year, equivalent to the entire EF of 0.50 million—1.51 million Dutch people. The GHG emissions from Dutch dog and cat food consumption were 1.09–3.28 million tons, equivalent to the total (i.e., not just dietary) emissions of 94,000–284,000 Dutch people [10].

This demonstrates the capacity for national variation. The dietary EPP of an average companion dog relying on commercial dry food in the Netherlands or in China was considerably greater than in Japan, although for companion cats these were similar among all three nations. Even in Japan, however, the dietary EPP of an average companion dog was equivalent to the dietary EF of an average Japanese person. And in all cases, dietary EPPs of companion dogs and cats equalled significant proportions of total human dietary EFs.

Vale and Vale [30] calculated dietary EPPs for small, medium and large dogs of 0.18, 0.27 and 0.36 ha/year, and for cats, of 0.3 ha/year. These were usually slightly lower than calculated by Su et al. [8], Su and Martens [9], and Martens et al. [10]. However Vale and Vale excluded footprints produced by ingredient processing, diet manufacturing, packaging and transporting. Using data from North Western Europe, Leenstra and Vellinga [92] estimated a dog paw print of 0.2 ha, and a cat paw print of 0.1 ha. However, the relatively high crop yields within this region may have lowered paw prints, compared to some other world regions.

The Brazilian canine population of 52.2 million is one of the world’s largest. Pedrinelli et al. [11] studied the diets of 618 Brazilian dogs and 320 Brazilian cats. An average canine diet was responsible for 828.37 kg of CO₂eq annually (dry diets) or 6,541 kg of CO₂eq (wet diets), equivalent to 12.4 or 97.8% respectively of the emissions of a Brazilian person (6.69 t CO₂eq annually). For the entire Brazilian canine population, dog food-associated emissions were 0.04–0.34 Gt CO₂eq annually, or 2.9–24.6% of Brazil’s total estimated emissions (1.38 Gt annually). This study demonstrated the markedly greater impacts of wet diets compared to dry diets.

Alexander et al. [93] estimated the environmental impacts associated with global dry pet food production. This was estimated to create 56–151 Mt CO₂ equivalent emissions (1.1% − 2.9% of global agricultural emissions), and to use 41–58 Mha of agricultural land (0.8–1.2% of global agricultural land), and 5–11 km³ of freshwater (0.2–0.4% of agricultural water extraction). However, they noted that this was based solely on dry food data, which constituted only 79% of US pet sales. Furthermore, they used an economic valuation to consider the impacts of ABPs, thereby substantially underestimating the environmental impacts of ABPs, which have low economic value. As demonstrated in this current study, ABPs require more, not less, average livestock animals, and have greater environmental impacts, than the use of HC ingredients. Alexander and colleagues also did not account for pricing variations globally, but similar pet food may be priced very differently, in different world regions. Finally, they assumed that global pet food volumes were weighted equally according to US dog (78%) and cat (22%) energy consumption [7], although dog and cat populations, and their relative proportions, vary substantially between countries. Hence, their results were impacted by substantial underestimations and uncertainties. Even so, they also estimated very significant environmental impacts, associated with global dry pet food production.

Additional impacts and future trends. It must be acknowledged that although the environmental impacts of dog and cat food revealed by all of these studies and this current study are considerable, they do not capture all impacts. Impacts are associated not only with primary production of animal- and plant-based ingredients, but with their processing, shipping, retail, storage, cooking, dishwashing and waste disposal. Many of these stages also include transportation impacts [59].

Impacts of pet food are also likely to increase in future, due to the rapid increases in the global companion dog and cat populations over decades [93], driven partly by human population growth, and facilitated by the economic development of some nations, which increases disposable incomes, and capacity to support pet guardianship. This is demonstrated by pet food sales trends. From 2022 to 2027, the global market for pet food ingredients is expected to increase from $32.2 - $44.5 billion–a compound annual growth rate of 6.7% [51].

Dietary energy requirements of dogs, cats and people.

The first assumption related to dietary energy requirements of dogs, cats and humans. The energy needs for dogs [43] and cats [45] were calculated using body weight averages published by Bermingham et al. However, dog breeds vary dramatically in size [42], resulting in markedly different MER requirements ([43], Table 3). Energy requirements also vary significantly with husbandry type and activity level, with requirements greatest in racing dogs, followed by working and hunting dogs, and finally, by pet and kennelled dogs. Very young or old dogs, or those who are pregnant, lactating or unwell, may also have significantly different energy requirements [94]. Although MERs appear equal between sexes, they are lower in neutered compared with sexually intact dogs [43]. For the purposes of this study, the average MER of dogs calculated by Bermingham et al.–partly on the basis of BW, was extrapolated to all US dogs. However, this is only an approximation of the true MER of all US dogs. As Bermingham et al. noted, “estimating maintenance energy requirements based on BW alone may not be accurate … predictions that factor in husbandry, neuter status and, possibly, activity level might be superior.” They also noted more information is needed about the nutrient requirements of older dogs, and of giant and toy breeds.

Similarly, the average energy requirements for domestic cats, calculated by Bermingham et al. [45], were extrapolated to all US cats. This was also an approximation. As stated by Bermingham et al., “maintenance energy requirements were significantly affected by weight, sex and neuter status, age and methodology”.

One key consideration is that the average body weight of dogs and cats is increasing over time, due to the increasing prevalence of overweight and obesity in kept dogs and cats. Hence the average body weights of dogs and cats can be expected to have increased significantly since the canine and feline averages were published by Bermingham et al. in 2014 [43] and 2010 [45] respectively. In both species weight gain results in significant increases in daily energy requirements [43, 45]. Hence the energy requirements estimated by Bermingham et al. for US dogs and cats, used within this study, probably significantly underestimated the true energy requirements of dogs and cats today. Updated demographic data for humans, dogs and cats, would allow more precise characterisations of these populations, and more accurate calculations of their dietary energy needs.

Total energy from animal sources (E_A), consumed by dogs, cats and people.

When calculating the proportion of human dietary energy attributable to animal produce (E_A), FAOSTAT data [47] were used to provide separate estimates for the US, and globally. The daily calories provided did not include quantities exported, fed to livestock, used for seed, put to manufacture for non-food uses, or losses during storage and transportation. Nevertheless, they remained substantially higher than daily energetic needs, both within the US, and globally. The excess calories were assumed to have been lost at later stages, e.g., retail, wasted or overconsumed. Comparative data on such excess levels within the diets of dogs, cats and people were not available; hence, these were assumed to occur at equal proportions within all of these dietary groups, allowing them to be discounted when considering the proportional consumption of average livestock animals, among dogs, cats and people. In reality however, such excess proportions may not be equal. Data on actual levels and differences between dietary groups, would allow refinements of the proportional livestock consumption estimates provided by this study.

When calculating the proportion of dog and cat dietary energy attributable to animal produce (E_A), data from Okin [7] were used. However, these E_{A dogs} and E_{A cats} proportions were calculated by Okin using the E_A within US premium and non-premium (‘market leading’) dog and cat foods, weighted by the proportions of US consumers choosing each. E_A was significantly higher within the premium brands. These data allowed accurate prediction of livestock consumption within US pet foods, but would have been less accurate when considering pet food globally. Due to lower average wealth, a higher proportion of consumers globally would have been likely to choose cheaper, non-premium brands, with lower E_As. This factor could decrease the relative impacts of pet food globally, compared to those predicted by US figures. To provide more accurate estimations of global pet food E_A fractions, global data could be sought and utilised if available, concerning the E_A fractions within premium and non-premium brands, and the proportion of consumers choosing to purchase each.

Animal-based ingredients used to feed people, dogs and cats.

When considering the various animal- and non-animal sourced ingredients within dog and cat food, the consumption data analysed for US pets were unusually detailed, but were not perfectly so. The DIS data [49] studied directly represented 68.3% of US retail pet food sales from July 2018 –June 2019. These were extrapolated (multiplied by 1/0.683) to estimate all of US retail pet food sales. Hence, the data used covered just over two thirds of the market. It also included major pet food companies. Accordingly, this extrapolation was probably quite accurate, although data covering the entirety (without extrapolation) of US retail pet food sales would have been even more accurate. Unfortunately, such data were not available within the US, nor globally, meaning that these US results also had to be extrapolated to pet food globally.

Consumption of HC and NHC ingredients within dog and cat food.

For each HC and NHC group, it was necessary to determine the proportion of average livestock animals (carcasses), that produced HC or NHC components. This allowed comparison of the efficiency of average livestock animals, at providing these two ingredient groups. To achieve this, the largest subgroup within each group was used as a proxy for the entire group. As noted, just under half of dog and cat food was provided by HC components, and just over half, by NHC components. Meat was used as a proxy for the HC group (comprising 66.2% of this group, for dog food, and 50.0%, for cat food), and animal meal as a proxy for the NHC group (comprising 88.1% of this group, for dog food, and 74.0%, for cat food). Given the proportionate sizes of these subgroups, extrapolation to cover each entire group seemed reasonable. However, accuracy could be increased by considering the full range of ingredients used. Livestock (carcass) proportions for all species supplying each of those ingredients could be sought where available, and could be included within averages weighted by consumption. This would allow more accurate determination of the proportion of average livestock animals, that produced HC or NHC components.

Attribution of energy consumption to HC and NHC components.

Having calculated the relative efficiencies of average livestock animals at providing HC and NHC dietary components, the E_A dietary fraction was then appropriately apportioned to these HC and NHC components, for dog and cat food. However, this required assuming that the E_A dietary energy was evenly distributed across the animal-sourced ingredients used.

In reality, the energy density of different animal-sourced ingredients is not uniform. However, neither do they seem widely distributed. As noted within the Methodology, the energy density of a variety of meats including poultry and fish, are around 200 kcal/100 g [50]. Hence, this assumption does appear reasonable. It was also hoped that any differences would average out to some degree, across the ingredients used. For dog food, 52 animal-sourced ingredients existed, represented by nine ingredients within the meat proxy group, and 14 ingredients within the animal meal proxy group. For cat food, 47 animal-sourced ingredients existed, represented by seven ingredients within the meat proxy group, and 11 ingredients within the animal meal proxy group.

Whilst this study has provided a reasonable estimation based on averaging, future research accounting for differences between these ingredients is recommended to provide more accurate estimates. Actual energy densities could be sought and used where available, within weighted averages. Energy densities of some ingredients (especially HC) are available via sources such as the USDA Food Data Central database [95].

Environmental sustainability indicators.

Calculation of environmental impacts of plant- versus animal-based ingredients relied on 2009–2011 averages for 52 plant- and animal-sourced food ingredients, using globally-sourced data [55]. There are very few such comprehensive data sets, and this is one of the most recent. These calculations could be updated in future as more recent data sets become available.

From these data, production volumes for food purposes for all ingredients were used to calculate weighted averages. However, whilst this is accurate for society as a whole, within the different dog, cat and human dietary groups, consumption proportions of the various ingredients would vary. Hence, the environmental impact estimates derived could be refined through consideration of actual ingredient consumption proportions, within these different dietary groups.

Finally, the attribution to dog and cat food of specific proportions of global livestock animal consumption–and hence, of global environmental impacts associated with the farming of those animals, relied on analysis of ingredients within the diets of US dogs, cats and humans. In reality, there will be regional and national differences in ingredient consumption, across all dietary groups, and global extrapolation will not be entirely accurate.

Despite such international variations, several factors made it reasonable to use US data as the basis for global extrapolation. Firstly, with over 86 million owned dogs and 61 million owned cats, the US was the country with the largest national populations of these animals. It comprised around 18.3% of the world’s 471 million owned dogs, and 16.4% of the world’s 373 million owned cats (based on US 2020 figures, and global 2018 figures). With respect to ingredients consumed, the US was the only region with very detailed data concerning dog and cat food ingredient consumption levels, predicting national consumption.

Finally, for US dog and cat food, NHC components comprised 52.6% and 50.8% of all animal-sourced ingredients respectively. In comparison, the global consumption of meat meal, ABP meal and animal digest within pet food (comprising all NHC ingredients) in 2020, comprised 17,113.1 kT, or 74.9% of the 22,841.1 kT total meat and meat products consumed within pet food (T = US ton) [51]. Hence, such NHC ingredients comprised a significantly higher proportion of pet food globally, than within the US. This probably occurred because such ingredients are cheaper, and the US is wealthier than most other countries. The global pet food ingredients market was worth $32.2 billion in 2022, with the North American market worth 36.2% of that–the largest regional share ([51], Table 2)–despite including only ∼16–18% of the world’s owned dogs and cats, as noted. Hence, US pet food has a significantly higher HC component, than pet food globally. But as calculated previously, HC ingredient provision is more efficient than NHC provision. It requires fewer average livestock animals to produce, decreasing environmental impacts. Hence, per kg of dog and cat food, environmental impacts would have been significantly lower in the US, than the global average. As noted previously, approximately 8.65 million tons of animal- and plant-based ingredients were used within US dog and cat food annually, from mid 2018–2019 [96]. Globally, 53.49 million tons of ingredients were used in pet food, in 2019 [51]. Hence, around 83.8% of consumption globally, was from regions where environmental impacts were significantly higher, than estimated for the US in this study. Accordingly, despite the various assumptions made–frequently based on the use of averages, the estimates of environmental impacts for dog and cat food, derived in part through extrapolation from US data, are very conservative. The true global environmental impacts of dog and cat food, are probably significantly higher than estimated in this study. More accurate estimations of impact in non-US regions and globally, could be derived through consideration of actual levels of NHC ingredient use within pet food, where these are available.

Additional people who could be fed using food energy savings.

It was noted that for every 1.0 J of animal-sourced HC ingredients consumed, an average of 3.7 J of excess dietary energy was lost during conversion from plant- to animal-sourced ingredients. This 3.7 J was used to calculate additional food energy that would become available, were dogs, cats or people transitioned on to vegan diets. However, this 3.7 J was calculated by considering the average loss-adjusted feed conversion ratio for beef+lamb, pork, and poultry, weighted by their relative availability in the diets of American people [60]. Although these meat products comprise the great majority of meat consumed by people, as well as by dogs and cats, the average diets of American people, dogs and cats all include additional animal-sourced ingredients, and the proportions of these animal-sourced ingredients are not uniform. Accordingly, whilst 3.7 J covers most of the meat consumed, it remains only an approximation for the excess energy inherent within the animal-sourced ingredients within these diets. More accurate estimations could be derived by considering a wider range of animal-based ingredients, and their different consumption levels, within dog, cat and human diets.