7. “Best” Abundance Estimates and Best Management
97
Background
Abundance Estimation
To understand the effect of human-caused mortality on a population, we need to
know the size of the population. No population of marine mammals can be
counted in its entirety (a true census). Instead, the population is sampled, and
mathematical techniques are used to estimate the absolute abundance. The precision of abundance estimates depends not only on the effort made to make the
estimate but also on properties inherent to the populations themselves. To understand the concepts of precision and bias, consider the analogy of archery (Fig. 7.1;
White et al. 1982). For the small remaining population of Hawaiian Monk Seals,
nearly every individual is identified. Thus, the estimate should be both precise and
unbiased (Fig. 7.1a). Seals and sea lions are photographed and counted during
maximum abundance on land (breeding or molting). Seasonal counts from animals on land are accurate (coefficients of variation in abundance (CV) often
<10%). Because some unknown proportion of the population is at sea, the estimate would be precise but biased (Fig. 7.1b). Estimates of the proportion at sea
could be made to correct for the bias. Most whale and dolphin populations and
some seal populations must be estimated with distance sampling techniques.
Obtaining precise estimates is frequently difficult. An uncommon but highly
visible species, such as the Killer Whale, would have an imprecise estimate (Fig.
7.1c). Both imprecision and bias (Fig. 7.1d) would be expected for an animal such
as a Sperm Whale, which is both uncommon and easily missed even when close
because it typically dives for 40 minutes. A few examples will illustrate the
difficulties in estimating abundance.
The most common technique for abundance estimation is line-transect (Buckland et al. 1993). Observers on ships or planes traveling along survey lines record
number of animals seen, species identification, and perpendicular distance (Fig.
7.2). Not all animals are seen, and observers have a better chance of seeing
animals that are close than those that are more distant. Data are used to estimate
the total number of animals. For a small population, few sightings will be made. If
the survey were replicated, the resulting abundance estimate would be different
(possibly substantially) due to many random factors. If you could repeat the
survey many times, the distribution of resulting estimates would be relatively
wide for rare species and would be narrow for common species. For Vaquita, an
endangered porpoise, Taylor and Gerrodette (1993) showed that the precision of
the abundance estimate drops sharply with decreasing population size. Thus, one
reason for poor precision is small population size.
A second reason for poor precision is that the species may be difficult to see.
Consider again populations A and B. Assume A is Blue Whales and B is Beaked
Whales. Blue Whales are conspicuous. Not only are they large but blows can be
seen for great distances. Thus, the probability of sighting does not decrease with
distance until distance becomes large. The smaller Beaked Whales surface
quickly, often erratically, and have no conspicuous blow. Sighting probability
97
Background
Abundance Estimation
To understand the effect of human-caused mortality on a population, we need to
know the size of the population. No population of marine mammals can be
counted in its entirety (a true census). Instead, the population is sampled, and
mathematical techniques are used to estimate the absolute abundance. The precision of abundance estimates depends not only on the effort made to make the
estimate but also on properties inherent to the populations themselves. To understand the concepts of precision and bias, consider the analogy of archery (Fig. 7.1;
White et al. 1982). For the small remaining population of Hawaiian Monk Seals,
nearly every individual is identified. Thus, the estimate should be both precise and
unbiased (Fig. 7.1a). Seals and sea lions are photographed and counted during
maximum abundance on land (breeding or molting). Seasonal counts from animals on land are accurate (coefficients of variation in abundance (CV) often
<10%). Because some unknown proportion of the population is at sea, the estimate would be precise but biased (Fig. 7.1b). Estimates of the proportion at sea
could be made to correct for the bias. Most whale and dolphin populations and
some seal populations must be estimated with distance sampling techniques.
Obtaining precise estimates is frequently difficult. An uncommon but highly
visible species, such as the Killer Whale, would have an imprecise estimate (Fig.
7.1c). Both imprecision and bias (Fig. 7.1d) would be expected for an animal such
as a Sperm Whale, which is both uncommon and easily missed even when close
because it typically dives for 40 minutes. A few examples will illustrate the
difficulties in estimating abundance.
The most common technique for abundance estimation is line-transect (Buckland et al. 1993). Observers on ships or planes traveling along survey lines record
number of animals seen, species identification, and perpendicular distance (Fig.
7.2). Not all animals are seen, and observers have a better chance of seeing
animals that are close than those that are more distant. Data are used to estimate
the total number of animals. For a small population, few sightings will be made. If
the survey were replicated, the resulting abundance estimate would be different
(possibly substantially) due to many random factors. If you could repeat the
survey many times, the distribution of resulting estimates would be relatively
wide for rare species and would be narrow for common species. For Vaquita, an
endangered porpoise, Taylor and Gerrodette (1993) showed that the precision of
the abundance estimate drops sharply with decreasing population size. Thus, one
reason for poor precision is small population size.
A second reason for poor precision is that the species may be difficult to see.
Consider again populations A and B. Assume A is Blue Whales and B is Beaked
Whales. Blue Whales are conspicuous. Not only are they large but blows can be
seen for great distances. Thus, the probability of sighting does not decrease with
distance until distance becomes large. The smaller Beaked Whales surface
quickly, often erratically, and have no conspicuous blow. Sighting probability