Brief review of previous lecture
•Observational vs. experimental evaluations
•Alternatives to enhance habitat connectivity
Lecture Outline: Population Viability Analysis & Conservation Genetics
•Extinctions: rates, causes, species characteristics
•Genetic variation: heterozygosity & allelic diversity
Recent extinctions of animal species
Are current extinction rates elevated relative to background rates?
•Most ecologists agree that current rates are relatively high and
mainly due to human effects
•However, disagreement exists regarding the magnitude of the
current extinction period—some equate current period with historical
mass extinctions; others think predictions are overestimates.
Problems with evaluating current and future extinction rates
•We don’t know how many species there are on earth
(perhaps 6-20 million but only 1.5 million described)
•Extinctions of certain groups are well documented (mammals, birds)
but others are not (insects, plants).
•Background levels are estimated from fossil record and have great
deal of uncertainty associated with them.
•Future predictions might overestimate species loss because the
species most susceptible to human impacts might be lost first.
Causes of Endangerment for Vertebrates
Species traits that might increase extinction risk
•Habitat overlap with humans
•Sensitive to disturbance
•Limited dispersal ability
•Rare—low population density, restricted geographic range
•Low growth rate capacity (life-history constraints)
•Large space requirements (e.g., large carnivores)
The demise of the passenger pigeon
•Probably the most abundant bird in North America.
•Continental population might have been 6 billion and represented 25-40% of
all of the birds in North America.
•In 1866, a cloud of birds passed into southern Ontario that was a mile wide,
300 miles long, and took 14 hours to pass a single point.
•The last passenger pigeon, Martha, died
in the Cincinnati Zoo in 1914.
•Species extinction was due to overharvesting,
loss of forest habitat, and a low reproductive
Population Viability Analysis (PVA)
•A quantitative assessment of a population’s risk of extinction, quasi-
extinction, or projected growth rate given current conditions or those
expected due to proposed management.
•We already have conducted various sorts of PVA in lab with Ramas
•Effect of TEDs on loggerhead sea turtles
•Habitat improvement for a spotted owl metapopulation
•Environmental stochasticity and helmeted honeyeater demography
•Perhaps first use of PVA was by Mark Shaffer
(1978) to evaluate the viability of the Yellowstone
grizzly bear population. PVA has since become a
cornerstone of the field of conservation biology.
Potential Uses of PVA
1. Assessment of extinction risk
•Assessing risk of single population
•Comparing relative risks of two or more populations
•Analyzing and synthesizing monitoring data
2. Guiding management
•Identifying key life stages or demographic processes
•Determining required size for a reserve
•Determining number of individuals to release in introduction
•Setting harvest limits
•Deciding number of populations needed for regional persistence
from Morris and Doak. 2002. Quantitative Conservation Biology.
•Initial focus of PVA was to determine Minimum Viable Population
For instance, what is the population size of grizzly bears
needed to be 95% certain that the population should remain
extant for 100 years?
•Emphasis shifted to using
extinction (or quasi-extinction)
curves to conduct sensitivity
analysis and to compare relative
merits of proposed conservation
Main Types of PVA
1. Deterministic Single Population Models
e.g., age-structured matrix model using only mean vital values
2. Stochastic Single Population Models
•e.g., age-structured model with environmental & demographic stochasticity
•Yields probabilistic results
•Most common type of PVA
3. Metapopulation Models
•Include spatial structure via dispersal among local populations
•Two forms: Patch-Occupancy Models, Detailed Multi-site models
4. Spatially-explicit Individual Based Model (IBM)
•Location, movement, breeding, mortality of each individual is tracked
•Data hungry; difficult to test with field data
Increasing (Detailed Multi-site)
Criticisms of PVA: Are Models Reliable?
1. Poor data quality
•Data requirements for even deterministic model are not trivial
•Good estimates of means (and variances) of vital rates are
difficult to obtain for endangered species
•Dispersal is especially tough to estimate with certainty
2. Form of density dependence unknown
3. Patterns of environmental stochasticity might not hold in future
•Catastrophic events could alter extinction risk dramatically
4. “Canned” programs
•Allow novice users to conduct PVA without enough understanding
•Different programs can produce dissimilar results with same data
5. Models not validated with field data
Test of PVA1
•Extensive evaluation of PVA using data from 21 long-term studies
•Used first half of data to develop and parameterize models; used
latter half to test the accuracy of PVA predictions.
•Accuracy assessed by comparing predicted quasi-extinction risk and
population size projections with reality.
•Surprisingly close match between model predictions and real outcomes.
•Actual population sizes fell within bounds predicted by stochastic
“PVA is the best tool we have for estimating extinction risk, and
the alternatives are subjective, less rigorous, and likely to
provide poorer predictions”
1Brooks,BW et al. 2000. Predictive accuracy of population viability
analysis in conservation biology. Nature 404:385-387.
Critique of Brook et al.1
•Argue that Brook et al.’s conclusions were worded too strongly and
a result of bias in studies included in the evaluation.
•Only used long-term studies with high-quality data and these conditions
are the exception for populations of endangered species.
•Suggested that PVA will only be accurate for predicting extinction
probability if data are extensive and reliable and if estimated vital
rates are likely to apply into the future.
“PVAs could be useful for comparing the consequences of
different management or conservation strategies…However, we
doubt the general claim that they can be accurate in their
ability to predict the future status of wild populations”
Coulson et al. 2001. The use and abuse of population viability analysis. TREE 16:219-221.
General Recommendations for using PVA
1. PVA should be treated as a model. Validity of models should be
tested with independent field data and PVA adjusted accordingly.
2. Evaluate relative rather than absolute rates of extinction or growth.
3. Do not focus on single value such as MVP; models are not accurate
enough to make such precise predictions.
4. Include uncertainty analysis in the broadest sense (vital rate
estimates, model structure and assumptions).
5. Compare short-term and long-term projections.
•Integration of genetic and demographic approaches in wildlife ecology
has become more common.
•Many insights are possible from genetic tools:
•Dispersal and connectivity
•Estimating abundances with mark-recapture approaches
•Levels of genetic variation and historical population sizes
•Taxonomic relationships and hybridization
Genetic variation (nuclear)
•Homozygous: two alleles (different form of gene) at a locus are the same.
•Heterozygous: two alleles at a locus are different.
Hardy-Weinberg Principle and Heterozygosity
If two alleles (A1 and A2) at a locus have frequency of p and q, then after
one generation of random mating, the genotype frequencies are:
A1 A1 = p2 A1 A2 = 2pq A2 A2 = q2 and p2 + 2pq + q2 = 1
If we extend concept to multiple alleles in which homozygote frequency for any
allele i with frequency p is pi2, the expected Hardy-Weinberg frequency
of heterozygotes, given k alleles at a locus, is:
1 - ∑pi2
Expected heterozygosity for single locus
•Example for endangered Hawaiian Laysan finch.
•One locus with three alleles:
p1 = 0.364 p2 = 0.352 p3 = 0.284
1 - ∑pi2 = 1 – (0.3642 + 0.3522 + 0.2842) = 0.663
Expected heterozygosity based on Hardy-Weinberg Principle
•Allele and genotype frequencies will remain constant over time if they are at
•Equilibrium means that they are not affected by evolutionary forces.
Natural selection Mutation
Genetic drift Gene flow
•Hardy-Weinberg equilibrium is unlikely for real populations, but expected
heterozygosity can serve as benchmark. That is, we can ask why population
deviates from expected.
Number of alleles at a locus
•Polymorphic: >1 allele detected at a locus across all individuals (vs. monomorphic)
•Allelic diversity or richness: average number of alleles per locus
Allelic diversity is lost more quickly during a severe population bottleneck
than is heterozygosity, because heterozygosity not affected as much by
changes in frequencies of rare alleles.
= mean heterozygosity
Population bottlenecks and loss of genetic diversity
•Two separate studies compared genetic diversity before and after a severe
population reduction for northern elephant seals and Guadalupe fur seals.
•Both species hunted to near extinction during 18th and 19th centuries.
• For instance, northern elephant seals might have been reduced to 10-30
individuals. The population has recovered to 100,000 seals.
•Compared variation from bones of pre-bottleneck seals to extant populations.
•Both species showed loss of genetic diversity.
•For Guadalupe fur seal, pre-bottleneck sample included 25 mtDNA genotypes,
whereas only 7 genotypes found for extant seals.
Determinants of genetic variation in populations
Natural selection Mutation
Genetic drift Gene flow
•Genetic drift is only one of BIG FOUR that always acts to decrease
variation in populations over the long haul.
•With drift, allelic frequencies change randomly across generations.
•Drift leads to loss of alleles as they are FIXED (frequency = 1.0) and to
a decrease in heterozygosity.
•Genetic drift most important for small populations (very much like
Habitat fragmentation and genetic variation
•Small, isolated populations can be
dominated by genetic drift leading
to reduced heterozygosity and
•Genetic diversity can still be
maintained across populations.
Why might fragmentation not lead to loss of genetic variation?
1. Fragmentation might not lead to genetic isolation. Only a small level of
gene flow is needed to maintain variation.
2. Fragmentation event might have been too recent for genetic effects to
3. Populations might be historically fragmented due to natural dispersal
barriers, which can add noise and make it difficult to detect
Inbreeding: loss of heterozygosity due to mating of individuals of related
“That any evil directly follows from the closest inbreeding has been denied by
many persons; but rarely any practical breeder; and never, as far as I know, by
one who has largely bred animals which propagate their kind quickly.”
Charles Darwin (from Mills)
Total Inbreeding Coefficient (Fit) includes components due to non-random
mating and due to genetic drift.
Inbreeding depression: decrease in demographic vital rates due to inbreeding.
Inbreeding depression in red-cockaded woodpeckers
•All vital rates are not
•Is inbreeding affecting vital
rate with strong influence on
population growth rate?
Some genetic markers used in wildlife biology1
•Restriction fragment length polymorhphism (RFLP)
Modern techniques greatly aided by polymerase chain reaction (PCR)
process for amplifying small samples of DNA.
1See Mills’ Text (Chapter 3) for descriptions