Calculating Food Production in the Subsistence Harvest of Birds and Eggs

Subsistence harvest studies use number-to-mass conversion factors (CFn-m) to transform numbers of animals harvested into food production (CFn-m = body mass × recovery rate; where recovery rate is the percentage of the body mass represented by the processed carcass). Also, if egg harvest was reported as volume (e.g., a bucket), volume-to-number conversion factors (CFv-n) are needed to calculate the number of eggs taken. Conversion factors (CF) for subsistence harvest of birds and eggs have been based on unclear assumptions. We calculated a mean recovery rate (65%) by weighing and processing wild birds, compiled data on bird and egg mass, developed an egg CFv-n equation, and presented CF for 88 bird species, 13 subspecies or populations, and 25 species categories likely to be harvested in Alaska. We also made recommendations on how to apply and adjust CF according to study objectives. We recommend that subsistence harvest studies (1) collect egg harvest data as egg numbers (not volume); (2) clearly explain considerations and assumptions used in CF; (3) report recovery rates and mass of birds and eggs; and (4) cite original sources when referring to CF from previous studies. Attention to these points of method will improve the accuracy of food production estimates and the validity of food production comparisons across time and geographic areas.


Number-to-Mass Conversion Factors for Birds and Eggs
Studies of subsistence uses of wild resources report harvest as the number of animals taken and as the amount of food produced (edible mass). Estimates of the number of animals taken are used to document subsistence activities, to assess harvest impacts on fish and wildlife populations, and to allocate harvestable amounts among user groups (Usher and Wenzel, 1987). Food production data are used to depict the relative importance of resources (e.g., moose, salmon, geese) and their role in subsistence economies (Brown and Burch, 1992), to assess exposure to contaminants derived from wild foods (Usher, 2000), and to estimate the monetary (replacement) value of harvest. Food production estimates do not account for the nutritional and cultural importance of different resources (Usher, 1976;Behnke, 1982). But these data are also important in adjudicating disputes among stakeholders, quantifying ecological services provided by resources and ecosystems, assessing food security, and prioritizing human activities on the basis of their socioeconomic contribution to communities' well-being (Jones, 1997;Magdanz et al., 2011;Hoover et al., 2013).
Food production is calculated by multiplying the number of animals taken by a number-to-mass conversion factor (CFn-m). A CFn-m integrates two variables: the live (whole, round) body mass of individual species or multi-species categories and the recovery (yield) rate, which is the percentage of the live mass represented by the processed carcass. Studies have commonly failed to explain assumptions used in CFn-m and to report body mass and recovery rates (Table 1). Over the decades, CFn-m developed in earlier studies have often been reused without critical evaluation or clear reference to original sources. Thus, it is difficult to evaluate discrepancies in CFn-m and food production estimates across studies. For instance, conversion factors (CF) used for Sandhill Crane Grus canadensis harvested in Alaska (CFn-m = 10 -15 lb/bird) (4536 -6804 g/bird) (Patterson, 1974;Wolfe, 1981;Ikuta et al., 2014) appear overestimated when compared to body mass of the subspecies occurring in Alaska (8.17 -9.82 lb) (3705 -4455 g) (Rodewald, 2015).
To inform stakeholders who rely on accurate food production data, studies need to clearly report considerations and variables used to derive CF. From subsistence users' perspectives, biased-low recovery rates conflict with the non-waste principle that is intrinsic to their cultural values (Zavaleta, 1999). Biased-low recovery rates also lead to underestimating the importance of wild resources in subsistence economies. On the other hand, biased-high recovery rates can discredit food production assessments and their use in mitigation and litigation.
Birds and eggs are a small proportion of the subsistence harvest, but data on their food production help address complex management issues (Fienup-Riordan, 1999;Zavaleta, 1999). In Alaska, the subsistence harvest (about 34 million edible pounds per year) is composed of fish (53%), land and marine mammals (23% and 14%), plants (4%), shellfish (3%), and birds and eggs (3%) (Fall, 2016). Developing CF for each of these resource categories involves diverse challenges. Previous efforts to consolidate information and clarify CF have addressed all resource categories (JBNQHRC, 1982;Usher, 2000;Ashley, 2002), and because this task is immense, some issues pertaining to bird and egg CF remained unresolved.
For wild birds, body mass depends on species, subspecies, population, sex, and age. Within these categories, body mass also varies because of ecological conditions and breeding, migration, and feather molting cycles (Piersma and Lindström, 1997). Because of difficulties in species identification, harvest surveys use species categories, which also complicate CFn-m because the species within a category may differ in body mass. Some studies have defined CFn-m for categories that include species with considerable size difference (Patterson, 1974). For instance, surveys have used one category for gull eggs, but eggs of large gulls are at least twice the size of those of small gulls (Rodewald, 2015). Social science researchers and other staff working on harvest surveys are often unfamiliar with the identification (including size), distribution ranges, and relative abundance of the dozens of bird species that may be harvested in a region. Thus, it is

Volume-to-Number Conversion Factors for Eggs
To facilitate accurate recall of harvest events and minimize burden on respondents, harvest surveys may use reporting units that are meaningful to harvesters (fish tub, truckload of wood, bucket of eggs) (Tobias and Kay, 1994). Even when respondents are asked to report their harvest in number of eggs, some values may instead be recorded in volume. For such cases, volume-to-number conversion factors (CFv-n) allow calculating the number of eggs taken. A method to estimate egg CFv-n involves comparing the mass (as a proxy for volume) of wild bird eggs to that of chicken eggs (J. Magdanz, pers. comm. in Naves, 2010). But CFv-n estimated in this way seemed high compared to CFv-n based on researcher or key respondent information (Burch, 1985;Fall et al., 1995). Use of padding material (e.g., grass, moss) to protect eggs reduces the total volume of eggs in a given container (Hunn et al., 2003). To clarify assumptions and refine this estimation method, we considered the use of padding material and the fact that full containers may not be filled to the brim to prevent egg damage during harvesting and transporting.

Study Objectives
It is impractical to account precisely for all sources of variation in recovery rate and wild bird body and egg mass that may affect CF (Usher, 1976;Burch, 1985). Thus, rather than defining highly precise and detailed CF, the objectives of this study were (a) to develop CFn-m and CFv-n equations based on clear variables and assumptions that can be easily adjusted depending on the study objectives and context and (b) to provide recommendations on the development and use of CF that will increase the accuracy of food production estimates and the validity of food production comparisons across time and geographic areas.
To achieve these objectives, we first collected ethnographic information from key respondents on subsistence practices in bird processing and egg harvesting. Information on bird parts usually consumed was needed to clarify which parts should be included as edible mass when calculating a recovery rate reflecting subsistence practices. Information on use of containers and padding material in egg harvesting was needed to refine the CFv-n estimation method. Second, we processed and weighed wild birds to calculate a recovery rate. Third, we compiled data on bird and egg mass, as well as distribution ranges and population sizes, for species likely to be harvested in Alaska. We integrated these social science and biological data to develop CFn-m and CFv-n equations and calculated CF for use in harvest studies. Although we addressed bird species composition and subsistence practices in Alaska, our approach and recommendations also apply to food production studies of other resources and regions.

Ethnographic Information on Bird Processing and Egg Harvesting
To calculate the bird recovery rate, the first step was to determine which bird parts should be included as edible mass, depending on how birds are processed and which parts are usually consumed. Also, refining the CFv-n equation required information on egg harvesting (see below). To gather this ethnographic information, we designed 14 questions on bird and egg harvesting and processing related to CF (online Appendix S1). The questions asked for information about general harvesting and processing in a region, as opposed to individual practices. We identified 27 Alaska Native people as key respondents who could provide information on the subsistence harvest and culture in their region of origin. Rather than a random selection of individuals within a sampling universe, key respondents are particularly knowledgeable people who can provide expert opinion on a domain (Huntington, 1998;Bernard, 2011). Participation in the Alaska Migratory Bird Co-Management Council (AMBCC, 2016) and in the U.S. Fish and Wildlife Service Refuge Information Technician Program were the main criteria used to identify such individuals. Participation in these programs served as an index of key respondents' extensive experience as subsistence resource users and community leaders, including their engagement in harvest management.
In April -May 2015, printed copies of the questionnaire were distributed in person or via postal mail to the potential respondents. Pre-stamped, pre-addressed envelopes were provided for return of completed questionnaires. We followed ethical principles for social science research, including informed consent and voluntary participation (ARCUS, 1999). One month after the questionnaire was first distributed, a reminder was mailed to people who had not yet returned responses. We received 16 completed questionnaires (a 59% response rate). In results pertaining to the questionnaire, "n" refers to the number of responses to individual questions or the number of times respondents indicated a categorical answer (yes, no, sometimes).
Respondents were instructed to leave fields for answers blank if they did not know the answer or if some species categories did not occur or were not used in their region. Most responses referred to species categories commonly harvested across Alaska (ducks, geese, grouse, and ptarmigan). Fewer responses were obtained to questions related to egg harvest than to those about bird harvest.

Definition of Edible Mass
On the basis of key respondent information (see Results), we defined the edible mass as including the carcass mass (meat, bones, skin, fat, and other tissues remaining after removal of feathers, feet, head, and viscera), as well as the heart and gizzard mass because these parts were also usually consumed by subsistence users. Although some responses indicated that other parts are sometimes consumed (e.g., liver, blood, intestine, stomach; see Results), these responses were infrequent, and the exclusion of these parts from the edible mass was inconsequential for the calculation of the recovery rate. While the exclusion of these parts may have resulted in a minor underestimation of the recovery rate, such underestimation was likely offset by the inclusion of skin and bones, which are sometimes not consumed. In Alaska, wild birds and eggs cannot be bought or sold and therefore have no defined market value. To facilitate assessments of the monetary value of wild foods, the definition of edible mass must be comparable to likely replacement products available in grocery stores. For bird eggs, we used a recovery rate of 100% (whole egg including the shell). Although the shell is not consumed, chicken eggs are a likely replacement product and are sold whole and by the dozen (not directly by weight).

Processing and Weighing Wild Birds to Calculate Bird Recovery Rate
To calculate the mean bird recovery rate according to our definition of edible mass, we weighed and processed wild birds harvested for home use in September -October 2015 and September 2016. This sample included ducks (n = 18), geese (n = 9), and ptarmigan (n = 2) harvested at several locations in south-central Alaska and the Alaska Peninsula. Mass measurements were obtained using an electronic scale with precision of one gram. We weighed the whole body mass of freshly killed birds. We plucked and singed the birds, removed the head, wing tips (cut at the metacarpus and ulna-radius joint), feet (cut at the tarsometatarsus and tibia-fibula joint), and all viscera. The skin-on, bone-in mass of birds thus processed constituted the carcass mass. After the carcass mass was recorded, we cut out and weighed the breast fillets (boneless, skin-on outer and inner fillets, or pectoralis and supracoracoideus muscles) and the whole leg (bone-in, skin-on thigh and drumstick, or tibia-fibula and tarsus sections). We also weighed the heart and clean gizzard (opened to remove food remains and its tough inner lining) to be included as edible mass. Weights were presented as arithmetic means of proportions of the live mass including all species. The bird recovery rate was calculated as the mean proportion of the carcass, heart, and gizzard mass relative to the live mass.
Breast fillets and whole legs are common cuts in sport hunting and poultry processing. Mass data for these cuts are useful to gauge recovery rates used in previous subsistence harvest studies, to generate alternative recovery rates based on different processing methods, and to allow comparisons with potential replacement poultry products.

Bird and Egg Mass Data
We compiled body and egg mass data for bird species, subspecies, and populations occurring in Alaska from Rodewald (2015) unless otherwise noted. Data on sex and age composition of Alaska subsistence bird harvest were unavailable. Thus, it was impossible to account for variation in body mass among sex and age categories in the harvest. We calculated the arithmetic mean body mass including data for males, females, adults, and immatures (as available) to represent sex and age categories potentially harvested. Mass data referred to Alaska-breeding populations in spring (as available) because at least 51% of the annual bird subsistence harvest occurs in spring (Paige and Wolfe, 1997). We used mass of freshly laid eggs because water loss during incubation decreases egg mass.
Mean body and egg mass were calculated based on all items (species, subspecies, populations) within categories. Population size data were used to weight mass means. Population size data were sometimes unavailable because (a) populations have not been monitored; (b) surveys have not differentiated among species (goldeneyes, mergansers, scoters, scaups) (Stehn et al., 2013;Platte and Stehn, 2015); and (c) estimates of abundance were not directly comparable among items within categories. If mass data were unavailable for one or more items within a category, means or reference values were defined by considering data for similar species and species distributions (online Appendix S2).
Body and egg mass data were reported both in grams (as rounded numbers with no decimal places) and in pounds (body mass data with two decimal places and egg mass data with three decimal places). Rather than displaying excessive precision, this level of detail when dealing with mass data in pounds was needed to properly represent mass of small birds and eggs.

Number-to-Mass Conversion Factors for Birds and Eggs
Using the recovery rates defined in this study and the bird and egg mass data compiled, we calculated CFn-m as: Bird CFn-m = 0.65 × body mass (see Results). (1) Egg CFn-m = 1.00 × egg mass (shell included). (2)

Volume-to-Number Conversion Factors for Eggs
To refine the CFv-n estimation method based on comparison of chicken and wild bird egg mass, we considered how the volume of eggs in a container is affected by (1) use of padding material and (2) not filling the container to the brim. First, to assess whether these considerations ref lect egg harvesting practices, the key respondent questionnaire included questions on characteristics of containers used, frequency of use of padding material, and whether containers are only partially filled to avoid egg loss and damage during transport (questions 8 -11, online Appendix S1). On the basis of information from key respondents (see Results), we assumed that padding material is always used and that full containers are filled to 80% of their capacity.
To quantify the reduction of the volume of eggs in a container resulting from use of padding material, we packed large chicken eggs (24 ounces or 680 g per dozen) (U.S. Department of Agriculture, 2000) in a one-gallon (3.8 L) bucket, filling it without any padding material and then adding dry grass between egg layers. We repeated packing with and without grass three times and counted the number of eggs needed to fill the bucket to the brim. With grass, the number of eggs needed to fill the bucket in the three measurements was 37, 36, and 33 eggs (mean = 35.3). Without grass, in each of the three measurements, 48 large chicken eggs were needed to fill the bucket. We

Number-to-Mass Conversion Factors for Birds and Eggs
The following ethnographic information was used to identify bird parts that should be included as edible mass when calculating the recovery rate. Key respondents reported that Alaska subsistence users consumed wild birds as bone-in, skin-on preparations, usually as a roast or soup (see also Mishler, 2003;Unger, 2014). Birds were consumed fresh or preserved by freezing, drying, or canning. Bird processing involved plucking, singeing, and gutting birds.
Meat from the breast, legs, neck, head, back, and wings was usually consumed, as well as skin, fat, heart, and gizzard ( Table 2). The liver was indicated as consumed in more than half of responses. Other parts were identified as not usually consumed, but some respondents indicated consumption of blood, intestine (ptarmigan, ducks, and geese), stomach (ducks and geese), kidney, and tongue. Bones were boiled to render broth, and bone marrow was sometimes consumed. We did not ask about non-food uses, but respondents reported that sometimes goose down was used and that the viscera and wings of harvested birds were used as bait in traps for fur animals.
Plucking seemed a preferred processing method among subsistence users, although skinning was sometimes used as a quicker option. To facilitate plucking, birds whose feathers are difficult to remove (swan, crane, seabirds, sea ducks) may be dipped in hot water. Such birds were sometimes skinned. The thin skin of grouse and ptarmigan often tears off during plucking, so these birds were commonly skinned. Plucking allowed consumption of the skin and associated fat. We asked respondents what proportion of the bird's body weight they thought is usually consumed when birds are plucked or skinned (the recovery rate) (questions 5 and 6, online Appendix S1). Responses ranged from 50% to 100%, but some seemed to refer to total mass after processing [recovery rate = 100% (n = 2) and "90% minus guts and bones"]. Because this question seemed unclear to respondents, we based the recovery rate solely on the data from wild birds processed in this study.
Using a recovery rate of 65% for birds and 100% for eggs and the mass data compiled, we calculated bird and egg CFn-m for 88 bird species, 13 subspecies or populations, and 25 species categories (Table 4 and online Appendix S2).

Volume-to-Number Conversion Factors for Eggs
The key respondent questionnaire indicated that fivegallon (19 L; n = 7) and one-gallon (3.8 L; n = 5) buckets were commonly used for egg harvesting, but that any available container may be used (basket, tea pot, bag, cooler; n = 8). In areas where eggs were commonly harvested, padding material was almost always used (question 9.a, online Appendix S1: "every time" n = 7, "three out of four times" n = 2). Padding was sometimes not used in murre egg harvesting because murre eggs are sturdy. Responses indicating infrequent use of padding material occurred for regions where eggs are harvested occasionally and in small numbers ("two out of four times" n = 2, "one out of four times" n = 1, "do not use moss, grass" n = 1).
Some responses to the question on whether containers are only partially filled to avoid egg loss and damage during transport considered (1) the volume of padding material separately from the volume of eggs; (2) whether  Values are number of key respondents that indicated bird parts as "usually," "sometimes," and "usually not" consumed in this study's questionnaire (online Appendix S1).
-: Consumption not indicated by key respondents. eggs were abundant enough to fill containers; or (3) the number of eggs that people needed and intended to harvest (question 10.a: "yes" n = 7, "no" n = 2; question 10.b: "yes" n = 6, "no" n = 7; question 10.c: mean = 69%, range = 30% -100%). Although these questions may have been understood differently by some respondents, responses indicated that, even if reports refer to full containers (e.g., two buckets), these were often not filled to the brim. We asked the number of eggs packed in a five-gallon bucket (question 11, online Appendix S1). Only one respondent provided direct information on the number of eggs per gallon, and the answer indicated a range (36 -60 large gull eggs in a five-gallon bucket). Three respondents indicated proportions of volume, which suggested that this question was not clear for them. Two respondents specified that they did not know the answer (e.g., "I never count them"). This question was left blank in the remaining 10 completed questionnaires.

Bird Recovery Rate
Although the composition of our wild bird sample reflected species availability at a limited set of locations and time of the year, our results were consistent with diverse data sources, including previous subsistence harvest studies, poultry production, and biological data on the relative mass of bird body parts. Considering the range of recovery rates used in previous subsistence harvest studies (40% -75%; Table 1), 40% was likely an underestimate, because it was little more than the percentage (32%) that we measured for only the breast fillets and legs. A recovery rate of 75% was likely an overestimate, because (1) it would involve including as edible mass bird parts other than those identified in this study as commonly consumed, and (2) it is higher than our recommended recovery rate (65%), which included the skin and bones, although the skin is sometimes removed during processing. Our recommended recovery rate (65%) agreed with several subsistence harvest studies in which the rate was based on assumptions by researchers. Relative mass is the percent difference in body mass (g) between a given species and the heaviest species, subspecies, or population within the same category. That is, relative mass = [1 -(focal item ÷ heaviest item in category)] × 100.
Even if not explained, recovery rates in some studies were likely based on information from local, Native experts and from researchers with wide experience in ethnographic work involving participant observation and residency in subsistence communities. Selective breeding and commercial production conditions may result in differences in body composition between poultry and wild birds, but recovery rates in wild birds and poultry were similar. For poultry, the recovery rate for a carcass processed for removal of blood, feathers, head, feet, and all viscera was 65% (range = 58% -72%) of the body mass (Watt and Merrill, 1963;Fanatico, 2003;Lessler et al., 2007;Połtowicz and Doktor, 2011). The breast and legs were about 38% of the body mass (Solomon et al., 2006;Haslinger et al., 2007). In wild birds, the breast and legs were 32% of the body mass in our sample and 28% in other sources (Raveling, 1979;Thompson and Drobney, 1996;Jacobs et al., 2011).
Both our study and other sources reported the heart as 1% of the body mass of wild birds (Thompson and Drobney, 1996;Piersma and Gill, 1998;Jacobs et al., 2011). The gizzard was 1% -2% of the body mass in seabirds and shorebirds (Piersma and Gill, 1998;Jacobs et al., 2011), 5% -7% in geese (Raveling, 1979;Barnes and Thomas, 1987), and 2% -5% in ducks (Barnes and Thomas, 1987;Goudie and Ryan, 1991;Thompson and Drobney, 1996). The relative gizzard mass we obtained (4%) was at the mean for ducks and geese. Using this mean was appropriate because it was in accordance with the overall species composition of subsistence harvest in Alaska (ducks were 58% and geese were 31% of the number of birds annually taken; Paige and Wolfe, 1997).
Using the allometric equation of Prange et al. (1979), bone mass for wild bird species likely harvested in Alaska accounted for 7% -9% of the body mass (results not presented here). Because some bone mass is removed during processing (head, feet, wing tips), the lower end of this range could be used to adjust the recovery rate when exclusion of bones is appropriate.
Recovery rates must reflect prevailing processing practices, which may differ among hunting traditions. A characterization of bird processing by sport users was beyond the scope of this study. In Alaska, bird sport hunting generally applies to harvest in non-subsistence areas as defined by the Joint Board of Fisheries and Game, which are primarily urban areas (State of Alaska, 2015:5 AAC 99.015). Sport hunters pluck birds for bone-in, skin-on preparations, and a recovery rate of 60% is likely adequate for this use (if the heart and gizzard are not usually consumed). But sport hunters also commonly skin birds, and only the breast (recovery rate = 22% for skin-on processing) or the breast and legs (recovery rate = 32%) may be consumed (Shaw, 2013). These three values could be combined to generate a recovery rate for sport hunting. In contrast to subsistence harvest studies, sport hunting economic valuations have focused on hunting activities and expenditures rather than food production (Gan and Luzar, 1993;ECONorthwest, 2014). A better understanding of food production in bird sport hunting as well as other differences and similarities between sport and subsistence bird hunting traditions could help alleviate conflict between user groups and promote positive outcomes in management and conservation issues.

Egg Recovery Rate
Studies have often assumed an egg recovery rate of 100% (e.g., Georgette and Loon, 1993), although this assumption may not be clearly stated. The eggshell is 8% -14% of the total egg mass (Williams et al., 1982). Across species, larger eggs have proportionally thicker shells and higher shell mass (Rahn and Paganelli, 1989). Murre eggs are an important subsistence resource, and their shells are about 14% of the total egg mass (Williams et al., 1982). Whether to include eggshell mass within edible mass depends on the study objectives. In replacement cost evaluation, eggshell should be included as edible mass (recovery rate = 100%) because a likely store-bought replacement product (chicken eggs) would include shells. When assessing exposure to contaminants, eggshells should be excluded from the edible mass because they are not consumed. If discounting shell mass, we recommend a recovery rate of 90% for all egg harvest.

Volume-to-Number Conversion Factors for Eggs
It is possible that the previous attempt to calculate CFv-n based on 48 chicken eggs/gallon (12.6 eggs/L) (J. Magdanz, pers. comm. in Naves, 2010) assumed that padding material was not used and that containers were filled to the brim. Estimates based on these assumptions were likely too high and resulted in numbers of eggs about 40% higher than ours. For murre eggs, the CFv-n calculated using our equation (16.0 eggs/gallon) (4.0 eggs/L) was half of that estimated by Burch (1985) (32 eggs/gallon) (8.4 eggs/L). Considerations used by Burch (1985) were unknown, but this difference may be related to the fact that we assumed use of padding material. For large gull eggs, our CFv-n (17.5 eggs/gallon) (4.3 eggs/L) was higher than (a) empirical data in Hunn et al. (2003) (12 eggs/gallon) (3.2 eggs/L); (b) the range provided by a key respondent in this study (7 -12 eggs/gallon) (1.8 -3.2 eggs/L); and (c) the value for "gull (unidentified)" provided by a key respondent in Fall et al. (1995) (10 eggs/gallon) (2.6 eggs/L). Although our CFv-n equation relied on simple assumptions, these were clearly stated and their variables can be easily adjusted to suit different study objectives and contexts. For example, if it is known that padding was not used, the equation could consider 35 chicken eggs/gallon (9.2 eggs/L).
In harvest survey interviews, considering individual harvest events, respondents can provide the best data on the number of eggs harvested. If respondents report eggs as volume, surveyors can assist respondents by sequentially asking (1) the kind of eggs harvested (species); (2) the size of containers used; and (3) whether padding material was used. Then, respondents may be asked to estimate how many eggs were harvested. For egg harvest reported as volume, the unit used in the original report must be reported so that standard CFv-n can be applied. Undocumented conversions of egg volume to number make it difficult to compare results among studies.

Species Categories, Regional, and Seasonal Conversion Factors
Mean body and egg mass used in CFn-m should approximate the harvest composition in a given geographic area and season of the year. In this study, we calculated mean mass for species categories weighted by Alaska-wide populations. Using the same principle, the mean mass for species categories may be adjusted for smaller geographic scales. Use of means weighted by population size is relevant for categories that include species of very different sizes. However, to simplify the application of CF and facilitate comparison among studies, whenever appropriate, CF should refer to relatively large geographic areas and encompass all seasons of the year.
Regardless of the level of analytical complexity researchers can implement when using CF, we offer four recommendations for this method. First, surveyors must be prepared to assist respondents in accurately reporting the number of eggs harvested, instead of volume. Second, considerations and assumptions used in CF must be clearly explained. Third, recovery rates and mass of birds and eggs used to generate CF must be reported (with citation of the source) so that users can assess which of these two variables accounts for potential differences from CF in other studies. Fourth, if using CF from previous studies, citations must refer to original sources, avoiding second-hand citations. Attention to these points will improve the accuracy of food production estimates and our ability to compare them across time and geographic areas.