Lecture Outline: Estimation of dispersal

1
Lecture Outline Estimation of dispersal

• Mark-recapture estimates of dispersal distances
• Biases including the missing tail
• Partial corrections
• Multistate M-R models

• Radio telemetry and harmonic radar

• Genetic approaches

• Dispersal distances inferred from patch
  colonization

• Cross-species predictions

2
Three stages of dispersal
Immigration Colonization
Transfer
Emigration
Social pressure Habitat quality Density
dependence Inbreeding avoidance Mate
competition Landscape context
Habitat selection/imprinting Conspecific
attraction Social fence
Matrix/Mosaic effects Search behavior Mortality
risk Perceptual range
(modified from Ims and Yoccoz 1997)
3
Mark-Recapture and Dispersal Distances

• Systematic bias occurs in the dispersal
  distributions obtained from studies using
  plot-based recapture or resighting data.

• the vast majority of intensive field studies
  conducted thus far on birds and mammals provide
  data on dispersal that are highly biased and
  virtually useless in determining either the true
  distribution of dispersal distances in the
  population or the extent of gene flow. (Koenig
  et al. 1996. TREE 11514-517)

• Degree of bias depends on dispersal distance
  relative to size of study area.

4
Mark-Recapture and Dispersal Distances
5
(Koenig et al. 1996)
6
Mark-Recapture Partial correction for bias

• Partial solution for underestimated dispersal
  distances is to correct observed distribution of
  movements for distance-specific detection
  probabilities.

• Detection probabilities reflect likelihood that
  individuals end up within plot boundaries and
  thus are available for detection.

• Allows for correction of dispersal distances
  within the maximum detection distance of plot.
  Longer movements are still not detected.

• Detection probabilities estimated via simulation
  in which many individuals (10,000) move a given
  distance in random direction from random starting
  points from within plot.

7
50 m

• Maximum detectable movement distance was 127 m.

• Probability of detecting a movement of 12 m was
  0.97, whereas probability of detecting a movement
  of 40 m was 0.27.

• Divide number of observed moves of a given
  distance by detection probability.

8
Dispersal distances corrected for detection
probabilities
9
Mark-Recapture Multistate/Mulstistrata Design

• Allows animals to move between geographic areas
  that differ in survival and recapture
  probabilities.

• Estimate transition probabilities between pairs
  of sites.

• One approach involves extension of
  Cormack-Jolly-Seber model.

• Data hungry design with many potential
  parameters to estimate.

10
Multistate Design
11
Radio Telemetry

• Greatly reduces biases typical of mark-recapture
  estimates of dispersal distance distributions
  such as truncation of the tail.

• However, often have some fixed detection radius
  around study area.
• Movements beyond search zone still go undetected
  unless you can use aircraft to extend search for
  lost animals.

• Restricted search issue is avoided by use of
  satellite or GPS based transmitters, but these
  methods had reduced accuracy until recently.

• Key negative for radio telemetry is cost
• Small mammal ear tag few cents
• PIT tag 4-10
• Radio collar 175-350
• GPS collar 2,500

12
Dispersal estimates Mark-recapture vs. telemetry
Females
Males
13
Estimating emigration with telemetry
14
Movement routes of green sea turtles determined
by satellite telemetry
15
Harmonic radar

• New opportunities to study movements of small
  animals, especially invertebrates.

• Exampleforaging paths of bumble bees.

• Range of 700 m in ideal conditions.

(Osborne JL et al. 1999. J. Appl. Ecol. 36519-533
16
Genetic approaches

• Sometimes called indirect estimates of
  dispersal and gene flow.

• Often easier to obtain spatially referenced
  genetic samples than to track actual dispersal
  movements.

• Methods based on allele frequencies of multiple
  populations often provide different picture of
  dispersal when compared to direct measures based
  on tracking individuals.
• Traditional genetic approaches include
  equilibrium assumptions and estimate historical
  dispersal over long time scales (dozens or
  hundreds of generations) compared to direct
  estimates.

• Newer genetic approaches such as assignment tests
  do not assume genetic
• equilibrium and provide more direct estimates of
  current gene flow (see Mills).

17
Medium to high levels of gene flow for
lynx (equilibrium approach)

• Low Fst values indicate high gene flow.

18
Dispersal and genetic structure of bannertail
kangaroo rats

• High degree of natal philopatry based on
  extensive mark-recapture studies.

• Effects on spatial genetic structure?

Waser and Elliott. 1991. Evolution 45935-943.
19
No evidence for spatial clustering of alleles

• Suggested that discrepancy between natal
  dispersal patterns and genetic structure might be
  due to gamete dispersal due to movements by
  males away from their residences during the
  breeding season.

20
Inferring dispersal distances from patch
recolonizations

• Approach applicable to spatially structured
  populations and metapopulations.

• Examine distances between newly colonized patches
  and closest potential source patches.

• Provides a minimum estimate in that individuals
  could have come from patch that was farther away
  than nearest source.

• Not ideal, but can provide a general idea of
  probable spatial scale of dispersal for rare
  species lacking dispersal data.

21
(No Transcript)
22
Wetland colonization by round-tailed muskrats

• Recolonized wetlands typically were within 2 km
  of closest potential source of immigrants.
• Median distances were 450-470 m.

• Only 1 of 50 wetlands gt2 km from known source was
  colonized between years.

23
Cross-species predictions of dispersal distances

• If dispersal distance was correlated with a
  variable that was easier to measure than
  dispersal, managers could use surrogate variable
  to obtain ballpark estimate of dispersal.

• Could be useful for identifying species that are
  likely to be vulnerable to landscape-level
  habitat changes such as fragmentation.

• What might be a potentially useful surrogate?

BODY MASS
24
Dispersal distance in relation to body mass
Sutherland GD et al. 2000. Conservation Ecology
4(2) Online.
25
Dispersal distance in relation to body mass
26

• Bowman et al. hypothesized for mammals that some
  species have an inherent capacity for movement
  that is independent of body size.

• They suggested that this mobility was reflected
  by typical home range sizes.

• Predicted that species with relatively large home
  ranges (for their body mass) would have
  relatively long dispersal distances (for their
  body mass).

Bowman J. et al. 2002. Ecology 832049-2055.
27
Bowman J. et al. 2002. Ecology 832049-2055.
28
What important factor is ignored by these
cross-species predictive equations?
WITHIN-SPECIES VARIATION IN DISPERSAL RELATED TO
LANDSCAPE STRUCTURE
29
Landscape structure dispersal distances