Copyright © 1997 Province of British Columbia
Published by the Resources Inventory Committee
[Back to TOC] [Previous] [Next]
3.2.1 Standards for accuracy and precision
Accuracy refers to how close the parameter estimate is to the true population parameter, and it can be improved by accounting for biases such as sightability. Precision is the closeness of repeated measurements to the mean population estimate. Precision is quantified by the sampling variance, and can be improved by replicating surveys, increasing the number of sample units, stratifying samples into groups where variation is expected to be lower, and by optimal allocation of sampling effort.
Accuracy and precision are both important for good survey estimates (Gill et al. 1983). Without those measures it is difficult, if not impossible, to compare studies over time or between areas. Bias can be classified into two types: 1) small sample bias; and 2) model bias. Model bias is the most serious of the two, since increasing sample size usually does not reduce the magnitude of the bias. All sample-based estimates are based on statistical models which depend on one or more assumptions. If all of the assumptions of the statistical model are not met, model bias results.
Precision is commonly indicated by associating confidence intervals with the estimate. A confidence interval gives the known probability (1 - a) that the actual value of a parameter will be included within the interval. It is recommended that a = 0.10 so that confidence intervals will provide a 90% probability that the actual value of a parameter will be included within an interval.
The reliability levels (accuracy and precision) required for population surveys depend on at least three factors: (1) the decision risk, i.e., the "cost" and likelihood of being wrong in the projected effect; (2) the natural variation in the parameters or characteristics to be measured; and (3) the technology and resources (i.e., people, time and money) available to measure the key parameters (Salwasser et al. 1983). Three reliability levels are recommended for aerial ungulate inventories, based on intended uses (Table 4).
| Level | Confidence Interval | Allowable Error | Intended Use |
| 1 | 90% | + 15% | Inventory Development Population Research Inventory of Red/Blue Listed Species |
| 2 | 90% | + 25% | Intensive Population Management Inventory of Yellow Listed Species |
| 3 | 90% | + 50% | Less Intensive Management |
In addition to establishing Type 1 errors (a levels), Type 2 errors ( b levels) should also be specified when using statistical tests. Gasaway et al. (1986:61) and Gerrodette (1987) provide a good discussion of Type 1 and 2 errors as they relate to population estimation. Briefly, the test of a null hypothesis results in accepting or rejecting the hypothesis, based on some estimated risk of being wrong. The probability of rejecting the hypothesis when it is true is referred to as a Type 1 error. The largest recommended acceptable risk of committing a Type 1 error is a = 0.10. A Type 2 error ( b) is the probability of concluding that the null hypothesis is true when in fact it is false. The largest recommended acceptable risk of committing a Type 2 error is b = 0.20. This will provide for statistical power (probability of rejecting the null hypothesis when null hypothesis is false) of 0.80.
3.2.2 Standards for sex and age classification
Ungulate surveys are often used to provide information on population sex and age structure, in addition to data on population size or distribution. The usual procedure has been to conduct intensive flights over selected areas of wildlife concentration, and to classify animals according to specified sex and age criteria. However, classification surveys at wildlife concentrations are often biased, sometimes severely so. A representative portion of the population should be surveyed to account for differential habitat preferences between sex and age classes. Replicating classification surveys will not overcome bias, if differences in habitat preferences between sex/age classes are not considered.
Jones (1984) proposed a classification level system for ungulates, which is used here, with some slight modifications (Table 5, 6, 7 and 8). The simplest levels of classification require distinguishing between adults and juveniles (Level 1), or between adult males, adult females, and juveniles (Level 2). Level 3 distinguishes between adult and yearling males. Level 4 includes 3 to 4 classes of mature males, which are based on horn curl, and antler size and shape.
Juvenile animals or young (less than 1 year old), can be distinguished for all species based on body size. For deer, elk, and moose, males can be distinguished from females when they have their antlers. For cervids, large antlered males drop their antlers earlier than others. Antler drop varies regionally, but usually begins in late November for caribou, in late December for moose, in January for deer, and late February for elk. For sheep, goats, bison and caribou, distinguishing between young males and females can be difficult, since both carry small horns or antlers. Correct classification requires close observation from a helicopter, or from the ground using spotting scopes. Adult moose and caribou can be classified as male or female by using the white (moose) or black (caribou) vulval patch to identify females. Close viewing of the antler scars on males is sometimes possible, but usually difficult. Urination posture may also be used, but requires longer term observations. Using standard aerial inventory methods, young males may be unavoidably included in the adult female class for mountain sheep, mountain goat, and bison. Animals which cannot be identified with certainty should be counted as unclassified. These include animals which slip into dense cover before they can be viewed more closely, or can not be identified in the midst of large groups. There are two categories for them; unclassified adults (unsexed but known not to be juveniles) and unknown (neither sex nor age is determined). All levels of classification should use those categories for clarity. There is little benefit to guessing at classifications.
When surveys are focused on obtaining population estimates, and many animals are enumerated quickly from the air, only unambiguous classification criteria should be used. The choice will depend on survey objectives, type of aircraft and seasonal timing (note: Level 4 classifications may require ground stations using spotting scopes). Although some experienced personnel may be able to classify animals to a higher level (e.g., Level 4), for consistency and accuracy we recommend using straight forward criteria requiring minimal judgment and time. The Level 2 or 3 classifications will normally be sufficient. Level 4 classifications are generally only required where detailed information is male age structure is required (e.g. mountain sheep). Consult Simpson et al. (1993) for diagrams of Level 4 classification of ungulates. The four levels of ungulate classification, and their intended uses are summarized in Table 5.
It is important to maintain consistent classification standards between surveys. Where data are being compared between areas, managers should insure that the same standards are used throughout. Consistent classification is required to calculate standard population ratios and allow comparison between populations. Numerous factors can affect the results of classification surveys, particularly where some sex/age classes in a population are more dispersed or less visible than others. In stratified random sampling, animals must be classified in all strata. Although focusing on high strata may allow classification of the most animals, the sex/age structure of populations often differs between strata. This is most notable for elk where many mature bulls are found in the low strata. Similar relationships have been noted for other species. A representative count can be assured if all parts of the area occupied by a population are surveyed adequately. Common biases reported for each species are identified in Section 4.0.
Table 5. Classification criteria for aerial ungulate surveys.
| Level | Classification Criteria | Population Ratio(4) |
| 1 | juvenile ( 1 yr) | · natality (births/100 adults) · recruitment (short yearlings/100 adults) |
| 2 | male/female/juvenile | · sex ratios (males/100 females) · recruitment (juveniles/100 females) |
| 3 | adult male/yrlg male / female/juvenile | · male recruitment (yearling males/100 · adult males or yearling males/100 adult females) |
| 4 | see individual species | · mature male age structure (classI/classII/class III/class IV) |
| Species | Sex/Age | Composition | Description |
| Black-tailed Deer | juvenile | fawn | · spotted pelage in summer · smaller body size and shorter nose in winter |
| female | doe | · medium size and no antlers · adult does may be accompanied by fawns | |
| male | yearling buck | · spike or antlers with 1 and/or 2 points | |
| Class I | · small 2 point or 2/3 points per antler | ||
| Class II | · medium 2 or 3 point or small 3 point, light antlers | ||
| Class III | · large 3 or 4 points/antler, or 5 points, heavy antlers | ||
| Mule Deer | juvenile | fawn | · spotted pelage in summer · smaller body size and shorter nose in winter |
| female | doe | · medium size and no antlers · adult does may be accompanied by fawns | |
| male | yearling buck | · spike or 2-points on one or both antlers | |
| Class I | · large 2 point or small 3 point antlers | ||
| Class II | · medium size antlers with 3 points/antler | ||
| Class III | · medium size with 3 or 4 points/antler · moderate to large bodied | ||
| Class IV | · large antlers with 4 or 5 points/antler | ||
| White-tailed Deer | juvenile | fawn | · spotted pelage in summer · smaller body size and shorter nose in winter |
| female | doe | · medium size and no antlers · adult does may be accompanied by fawns | |
| male | yearling buck | · spike or 2-points on one or both antlers | |
| Class I | · large 2 point or small 3 point antlers | ||
| Class II | · medium size antlers with 3 points/antler | ||
| Class III | · medium size with 3 or 4 points/antler · moderate to large bodied | ||
| Class IV | · large antlers with 4 or 5 points/antler |
| Species | Sex/Age | Composition | Description |
| Elk(5) | juvenile | calf | · small body size without antlers |
| female | cow | · medium size without antlers | |
| male | yearling bull | · spike antlers or with light 1 to 2 point antlers | |
| Class I | · small antlers with 3 or 4 points (raghorn) | ||
| Class II | · large 4 pt antler, small 5 pt antler, spindly (raghorn) | ||
| Class III | · large 5 pt antler, small 6 pt antler, heavy antlers | ||
| Class IV | · large antlers with 6 or 7 pts/antler, massive | ||
| Moose | juvenile | calf | · small body size without antlers |
| female | cow | · no antlers and short bell, medium size · distinguished by white vulval patch · usually has a light brown face colour · sometimes accompanied by calf | |
| male | yearling bull | · antler, if palmated, does not extend beyond eartip · antler pole type, usually a spike or fork | |
| Class I | · antler palmated, extends beyond tip of ear · browtine a spike or fork | ||
| Class II | · antler palmated, extends beyond tip of ear · brow tine palmated with usually 2 or more points · innermost points of brow palm close over face | ||
| Class III | · antlers palmated, but smaller than Class II · brown tine usually a spike or fork, like Class I | ||
| Caribou | juvenile | calf | · antlers (if any) are short (spikes) with velvet · darker body and smaller than adults |
| female | cow | · small antlers 2-3 times the ear length · black vulval patch | |
| male | yearling/ Class I | · small antlers which are 2-3x the ear length · similar to females, but no vulval patch | |
| Class II | · antlers larger than females · antlers are lighter and smaller than Class III bulls · antlers without shovels | ||
| Class III | · large, heavy-beamed antlered males · antlers with many points and a palmated brow tine · may have shovel with few points, but heavy beams |
| Species | Sex/Age | Composition | Description |
| Bison | juvenile | calf | · small, light brown |
| female | cow | · medium size, difficult to distinguish from yearling males | |
| male | bull | · largest in size, horns also largest | |
| Mountain Goat(6) | juvenile | kid | · small body size, short horns, usually with female |
| yearling | · larger than kid but smaller than adult · juvenile face, shorter muzzle · horns shorter than ears in early-mid summer · usually in groups with females and kids | ||
| female | nanny (2+ yrs) | · shaggy coat in July · squatting urination posture · black vulval patch when tail raised · horns have a fairly sharp kink at the tip · horns thinner and more wide-spread than males · horns V-shaped from the front · head and shoulders less massive than males | |
| male | billy (2+ yrs) | · smooth coat in July · scrotum visible in summer coat · stiff-legged “rocking horse” gait · horns thicker and closer together than in female · stretching urination posture · head and neck massive compared to female | |
| Mountain Sheep | juvenile | lamb | · small with small horns |
| female | yearling | · larger than lambs in horn and body size but not as large as adult females or yearling males | |
| female | adult | · < 1/2 curl horns, often difficult to distinguish from yearlings | |
| male | yearling | · identified in bighorn sheep, but often difficult to distinguish from adult ewes in thinhorn sheep · horn size greater and bases are slightly wider and more divergent than ewes | |
| Class I | · horns larger than females or yearling rams · often difficult to separate from yearling rams · 2 year old ram (Peace-Liard) | ||
| Class II | · horns over 1/2 curl, but less than 3/4 curl · body size smaller than Class III ram · 3 year old ram (Peace-Liard) | ||
| Class III | · horns over 3/4 curl but less than full curl · 4 year to full curl ram (Peace-Liard) | ||
| Class IV | · full curl horns (to bridge of nose) or greater |
3.2.3 Habitat Data Standards
A minimum amount of habitat data must be collected for each survey type. The type and amount of data collected will depend on the scale of the survey, the nature of the focal species, and the objectives of the inventory. As most, provincially-funded wildlife inventory projects deal with terrestrially-based wildlife, the terrestrial Ecosystem Field Form developed jointly by MOF and MELP (1995) will be used. However, under certain circumstances, this may be inappropriate and other RIC-approved standards for ecosystem description may be used. For a generic but useful description of approaches to habitat data collection in association with wildlife inventory, consult the manual, “Introduction to RIC Wildlife Inventory”.
[Back to TOC] [Previous] [Next]