abundance, movements and habitat use of coastal … · abundance, movements and habitat use of...

Abundance, movements and habitat use of coastal dolphins in the Darwin region

Abundance, movements and

habitat use of coastal dolphins in

the Darwin region

Analysis of the first four primary samples (October 2011 to April 2013)

STATPLAN CONSULTING PTY LTD

November 4, 2013

Lyndon Brooks & Kenneth Pollock (2013). Abundance, movements and habitat use of

coastal dolphins in the Darwin region: Analysis of the first four primary samples (October

2011 to April 2013). Draft report for the Northern Territory Government Department of

Land Resource Management.

INPEX Doc. Number L384-AH-REP-10009_0


Page 1

Contents Contents .......................................................................................................................................................... 1

Executive Summary ........................................................................................................................................ 2

Context ........................................................................................................................................................ 2

Methods ...................................................................................................................................................... 2

Results......................................................................................................................................................... 2

Abundance .............................................................................................................................................. 3

Movements.............................................................................................................................................. 3

Apparent Survival ................................................................................................................................... 4

Habitat Use ............................................................................................................................................. 4

Conclusion .................................................................................................................................................. 4

Introduction ..................................................................................................................................................... 5

Methods .......................................................................................................................................................... 6

Sampling ..................................................................................................................................................... 6

Data analysis ............................................................................................................................................... 9

Abundance, apparent survival and movements ....................................................................................... 9

Habitat use ............................................................................................................................................ 12

Results........................................................................................................................................................... 15

Abundance, Apparent Survival and Movements ...................................................................................... 15

Humpback dolphin ................................................................................................................................ 15

Bottlenose dolphin ................................................................................................................................ 16

Snubfin dolphin ..................................................................................................................................... 18

Summary of abundance estimates ......................................................................................................... 19

Habitat use ................................................................................................................................................ 21

Summary of habitat use ........................................................................................................................ 25

Conclusion .................................................................................................................................................... 26

REFERENCES ............................................................................................................................................. 27

Appendix ......................................................................................................................................................... 0

Table A1 Dates of secondary samples by site for each primary sample ..................................................... 1

Table A2 Summary of captures by species, site and primary sample ......................................................... 2

Table A3 Model comparison results for models on each species ............................................................... 4

Table A4 Parameter estimates from selected models for each species ....................................................... 7



Page 2

Executive Summary

Context

The Darwin Harbour Coastal Dolphin Monitoring Program was initiated as part of the environmental

approvals under the Commonwealth Environment Protection and Biodiversity Conservation Act 1999 and

the Northern Territory Environmental Assessment Act 1982 for the INPEX Ichthys Gas Field Development

project. The stated aim of the program is:

“To detect change, beyond natural spatial and temporal variation, in coastal dolphin abundance and

distribution during near shore Project construction activities in Darwin Harbour, including pre- and post-

construction phase monitoring.”

Methods

The demographic parameters, abundance and apparent survival, were estimated using an extension of the

Robust Design model known as the Multistate Robust Design model. This model provides an integrated

analysis of the data over the three sites sampled as part of the monitoring program (Bynoe Harbour,

Darwin Harbour and Shoal Bay) including estimates of rates of movement between the three sites between

primary samples.

Sufficient data were available to build a Multistate Robust Design model for all three sites for humpback

dolphins (Sousa sp.). Conversely, for bottlenose dolphins (Tursiops sp.) there were too few data from some

sites to allow analysis, and therefore, the data for Bynoe Harbour and Shoal Bay was pooled to yield

estimates for Darwin Harbour and elsewhere in the local region. Similarly, the data for the Australian

snubfin dolphin (Orcaella heinsohni) were pooled across all three sites to yield a single set of estimates for

the whole local region.

A Binary Logistic Mixed Effects model was employed to estimate the relative rates of use of locations

(sub-sites) in the local region over the four primary samples. The model was fitted to the data for all

species combined as there were too few data on bottlenose and snubfin dolphins to support separate

models for these species.

Results

The ability to model more complex effects or to obtain estimates of demographic parameters for all sites

was limited for snubfin and bottlenose dolphins due to the small sizes of these populations. This limitation

was more acute for snubfin than bottlenose dolphins due to the irregularity of their visitation to the area

and greater difficulty of capturing and recapturing them. The population size and recapture rate for

humpback dolphins were sufficient however to allow models to generate estimates with good precision at

all three sites.



Page 3

Abundance

The abundance estimates of the three species have remained relatively stable over the four primary

samples. Humpback dolphins were the most abundant of the three species in the region and at all three

sites. Moreover, their numbers have remained stable with 20-29 using Bynoe Harbour, 37-41 using Darwin

Harbour and 13-21 using Shoal Bay. Overall, around 80 humpback dolphins use the region in a primary

sample (i.e., a three week survey period).

Although the population of bottlenose dolphins in the region is small, capture and recapture rates for this

species were relatively high and consistent over time allowing estimates with good precision to be

obtained. Bottlenose dolphins rarely use Bynoe Harbour and the few data that were available for that site

were pooled with those from Shoal Bay to obtain estimates for Bynoe Harbour/Shoal Bay and Darwin

Harbour. Five to ten bottlenose dolphins were found to use Bynoe Harbour/Shoal Bay and 15-25 were

found to use Darwin Harbour over the primary samples. Overall, 22-31 bottlenose dolphins use the region

during a primary sample.

The Australian snubfin may be least abundant or use the region in about the same numbers as the

bottlenose dolphin. Visitation of the region by this species is irregular, and their capture and recapture rates

are highly variable over time. The snubfin data were the most difficult to model and it was necessary to

pool the data over all three sites to obtain estimates for the region. Around 17-29 snubfin dolphins use the

region in a primary sample.

Movements

There were sometimes substantial movements of humpback dolphins between Bynoe Harbour and Darwin

Harbour, varying between six and 38 percent in one direction or another between successive primary

samples. No movements were observed between either of Darwin Harbour or Bynoe Harbour and Shoal

Bay. We found no evidence of temporary emigration between primary samples.

Movements of bottlenose dolphins between sites were observed in only two intervals between primary

samples but when they occurred, they were quite substantial at 39 and 11 percent. Around six percent of

the Darwin Harbour population were estimated to have temporarily emigrated from the local region

between successive primary samples.

While movements of snubfin between sites were observed, it was not possible to model them. Temporary

emigration of snubfin from the local region between successive primary samples was high at around 63

percent, indicating that they move around a large area and occur in numbers substantially larger than the

number that use the region in a primary sample (perhaps as large as 70 or more). That is, the area sampled

(Bynoe Harbour, Darwin Harbour and Shoal Bay) appears to be smaller than total area used by the snubfin

population that uses the local area, of which only about 37 percent were estimated to be present in the

sample area during a primary sample.



Page 4

Apparent Survival

The annual apparent survival rate of humpback dolphins varied markedly between Darwin Harbour at

around 0.93 and Bynoe Harbour and Shoal Bay at around 0.49 each, suggesting substantial emigration

from the latter two sites. Abundance at these sites has remained relatively constant however, suggesting

complementary recruitment to the sites from outside the local region.

The annual apparent survival rate (alive and on site) for bottlenose dolphins was around 0.81, a rate which

is likely to be lower than the true biological survival rate (alive) suggesting some, relatively small rate of

emigration from the local region.

The annual apparent survival rate for snubfin dolphins was around 0.94 which is quite high and may be

close to the true biological survival rate for the species. While the confidence interval was very wide for

this estimate (0.0:99.9), taken at face value, this suggests that, although there is substantial temporary

emigration from the local region between primary samples, most snubfin that use the area at some time are

likely to return at some stage.

Habitat Use

We found no evidence of change in the relative rates of use of ten locations (sub-areas) defined within the

local region between primary samples. It’s notable that the outer Harbour areas (outer Bynoe Harbour,

outer Darwin Harbour and Gunn Point) are relatively well used and that this is consistent with and

complementary to evidence from the capture-recapture models of movement into and out of the Darwin

region: i.e., there is evidently a flow of dolphins not only across the outer Harbour areas in the Darwin

region but also to and from sites west of Bynoe Harbour and east of Shoal Bay.

Conclusion

While the abundances of the three species are quite small, they have remained stable over the duration of

the study so far, and the estimated rates of habitat use show no significant variation between locations over

time.



Page 5

Introduction

As part of the environmental approvals under the Commonwealth Environment Protection and

Biodiversity Conservation Act 1999 and the Northern Territory Environmental Assessment Act 1982 for the

INPEX Ichthys Gas Field Development project, a monitoring program for coastal dolphins was required

for Darwin Harbour. The stated aim of the program is: “To detect change, beyond natural spatial and

temporal variation, in coastal dolphin abundance and distribution during near shore Project construction

activities in Darwin Harbour, including pre- and post- construction phase monitoring.”

Three species of coastal dolphin inhabit the Darwin Harbour region (extending from Gunn Point to Bynoe

Harbour): Indo-Pacific humpback (Sousa sp.; hereafter referred to as humpback, see Mendez et al. 2013

for recent taxonomic results), Australian snubfin (Orcaella heinsohni, hereafter referred to as snubfin) and

bottlenose (Tursiops sp., hereafter referred to as bottlenose) dolphins. All three species are listed as marine

and migratory species (and hence are matters of National Environmental Significance) under the

Environment Protection and Biodiversity Conservation Act 1999 (EPBC Act).

Monitoring of the local populations of the three species of dolphin in Darwin Harbour for the Ichthys

project commenced in October 2011. The population sizes of each species in Darwin Harbour have

previously been estimated as small, with approximately 40, 18 and eight individual humpback, bottlenose

and snubfin dolphins, respectively (Brooks and Pollock, 2012). Population sizes in Bynoe Harbour and

Shoal Bay were smaller with approximately 30 humpback, four bottlenose and 10 snubfin in Bynoe

Harbour, and 15 humpback, 11 bottlenose and a small but unknown number of snubfin in Shoal Bay.

The first two primary sample surveys may be viewed as constituting a pre-construction phase baseline as

no construction activity occurred (although shipping activity increased) during this period. Dredging first

began in October 2012 and continued through the 2012/2013 wet season which includes primary samples

three and four.

The objectives for this report are, for the available data, to:

1. Assess changes in population dynamic parameters of the three coastal dolphin species at the three

sites, including population size, losses (mortality + emigration), gains (births + immigration) and

temporary emigration (assumes sufficient population sizes of a species at a site to yield adequate

samples for analysis) before, during and after the construction of the Ichthys LNG facility in

Darwin Harbour.

2. Assess changes in the spatio-temporal distribution (pattern of habitat use) of the three coastal

dolphin species at each of the three sites, including movements between the sites (assumes

adequate data on movements) and site fidelity before, during and after the construction of the

Ichthys LNG facility in Darwin Harbour.



Page 6

Methods

Sampling

The sampling design for the Darwin Harbour Dolphin Monitoring Program is based on a Robust Design

sampling structure (Brooks and Pollock 2011, Pollock et al. 1990, Williams et al. 2002) of two primary

samples per year (wet and dry season samples), each consisting of nine secondary samples at each of three

sites (Darwin Harbour, Bynoe Harbour, Shoal Bay). Each secondary sample was defined as a complete set

of transects through a site (Brooks and Pollock 2011, Griffiths and Palmer 2011). It was planned that,

weather permitting, each secondary sample would be taken in one day using four boats in Darwin Harbour,

three in Bynoe Harbour and one in Shoal Bay. The four boats alternate each three days between all four in

Darwin Harbour, and three in Bynoe harbour and one in Shoal Bay.

The principal data collected on survey are the locations and species of dolphin groups, and photographs of

the dorsal fins of individual dolphins. The data on the locations of sighted dolphin groups were used to

model their spatial distribution, while the photographs were used to identify individuals from the nicks and

scars on their dorsal fins to yield capture-recapture data to model abundance, apparent survival and

movements.

Capture-recapture methods have been widely used to estimate demographic parameters for a number of

dolphin species including snubfin, humpback and bottlenose dolphins (Würsig and Jefferson 1990; Parra et

al. 2006; Nicholson et al. 2012). Many cetaceans bear nicks and marks that allow identification of

individuals from photographs, and such identifiers provide a mechanism for population estimation based

on capture-recapture methods, where re-sightings of individuals with distinctive natural marks constitute

re-captures (Hammond and Thompson 1990). A general overview of capture-recapture models is found in

Amstrup et al. (2005) while more detailed coverage is found in Williams et al. (2002). The Robust Design

model is described and its parameters specified in Pollock et al. (1990), Kendall and Nichols (1995), and

Kendall, et al. (1995, 1997).

Figure 1 shows a research crew on survey on a calm morning, Figure 2 shows a humpback dolphin

observing a research crew, Figure 3 shows a crew photographing a dolphin with a distinctively-marked

dorsal fin, and Figure 4 maps the transect lines followed nine times in each primary sample at the three

sites.



Page 7

Figure 1 Research crew on survey

Figure 2 A humpback dolphin observing a research crew



Page 8

Figure 3 A crew photographing a dolphin with a distinctively-marked dorsal fin

Figure 4 Map of locations of transect lines in Bynoe Harbour, Darwin Harbour and Shoal Bay



Page 9

Sampling was completed for the first four primary samples at all sites between:

20th

October and 18th

November 2011 (primary sample one pre-construction),

26th

March and15th

April 2012 (primary sample two – pre-construction),

8th

October 2012 and 27th

October 2012 (primary sample three – dredging season one), and

13th

March 2013 and 2nd

April 2013 (primary sample four – dredging season one).

The dates of the primary samples and their secondary samples at each site are reported in the Appendix,

Table A1.

Data analysis

The number of secondary samples with non-zero captures, the number of individuals captured, the

maximum number of captures per individual and the total number of captures of all individuals are

reported for each species at each site in each primary sample in the Appendix, Table A2.

Abundance, apparent survival and movements

The Multistate Closed Robust Design Model (MSCRD, Nichols & Coffman 1999, Kendall & Nichols

2002, Kendall 2013) was employed for analysis of the capture-recapture data to estimate abundance,

apparent survival, and movements between sites and temporary emigration between primary samples. This

is at once an extension of the Closed Robust Design model (CRD, Pollock 1982, Kendall and Nichols

1995, Kendall et al. 1997) and the multistate model for recapture data (Arnason 1972, 1973; Brownie et al.

1993; Schwarz et al. 1993).

The CRD model was employed in the previous report (Brooks & Pollock 2013) for analysis of abundance

and apparent survival, and a multistate model based on data collapsed to primary samples (i.e., not robust

design) was employed for analysis of movements between the three sites. Here these two kinds of models

are combined in the MSCRD model.

The MSCRD model provides estimates of:

1. Apparent survival between primary samples (probabilities, S parameters)

2. Movements between sites and temporary emigration between primary samples (probabilities, psi

parameters). Whereas temporary emigration is modelled in terms of gamma” and gamma’

parameters in the CRD, temporary emigration is included among the movements (psi parameters)

in the MSCRD by defining an ‘unobservable’ state for dolphins that are temporarily absent in a

primary sample.

3. Abundance at each primary sample (N parameters).



Page 10

Whereas the CRD model deals with only one site (here, Bynoe Harbour, Darwin Harbour, Shoal Bay or all

considered as one regional site) at a time, the MSCRD model can simultaneously provide these estimates

for multiple states (here multiple sites, Bynoe Harbour, Darwin Harbour and Shoal Bay).

With three sites (Bynoe Harbour, Darwin Harbour and Shoal Bay) four states were defined: three

observable states (the three sites) and one unobservable state for temporary absence from all three sites.

Dolphins may move between all four states between pairs of primary samples, with such movements being

modelled as transition probabilities. With four primary samples, the complete set of possible between-state

movements (transition probabilities) for the intervals between primary samples one and two, two and three

and three and four is:

1. Movements between Bynoe Harbour and Darwin Harbour.

2. Movements between Bynoe Harbour and Shoal Bay.

3. Movements between Darwin Harbour and Bynoe Harbour.

4. Movements between Darwin Harbour and Shoal Bay.

5. Movements between Shoal Bay and Bynoe Harbour.

6. Movements between Shoal Bay and Darwin Harbour.

7. Movements between Bynoe Harbour and the unobservable state (absent from Bynoe Harbour,

Darwin Harbour and Shoal Bay).

8. Movements between Darwin Harbour and the unobservable state (absent from Bynoe Harbour,

Darwin Harbour and Shoal Bay).

9. Movements between Shoal Bay and the unobservable state (absent from Bynoe Harbour, Darwin

Harbour and Shoal Bay).

10. Movements from the unobservable state to Bynoe Harbour.

11. Movements from the unobservable state to Darwin Harbour.

12. Movements from the unobservable state to Shoal Bay.

Movements between an observable and the unobservable state (temporary emigration, 7 to 12 above) may

be modelled as:

1. Random, where, for each interval, the probability of staying away after an absence from the local

region is equal to the probability of leaving (i.e., is independent of previous state).

2. Markovian, where, for each interval, the probability of staying away after an absence from the

local region is not equal to the probability of leaving (i.e., depends on previous state).

3. Even flow, where, for each interval, the probability of returning to a site after an absence from the

local region is equal to the probability of leaving the site to the unobservable state.

Movements to and from the unobservable state in the MSCRD are equivalent to the temporary emigration

(gamma’’ and gamma’ parameters, for temporary absence given presence or absence respectively in the

last primary sample) in the CRD. Consequently, for Markovian models, in the MSCRD as in the CRD, if



Page 11

the probability of apparent survival varies by primary sample (apparent survival varies over intervals

between primary samples), the last transition probability must be constrained to equal a transition

probability from an earlier interval. For such models we’ve set the last transition probability (primary

sample three to primary sample four) equal to the second last transition probability (primary sample two to

primary sample three).

Similarly, in the MSCRD as in the CRD, the probability of apparent survival for temporarily absent

(unobservable) dolphins must be constrained to be equal to the probability of apparent survival for

dolphins in an observable state. In the present, multisite case, this may be any one of the sites: if there is no

temporary emigration, models with apparent survival constrained to any one of the three sites are

equivalent but may differ when temporary emigration is modelled.

The MSCRD model estimates transitions between but not within primary samples, and requires that a

single site be identified for each individual in each primary sample. We nominated the site in which each

dolphin was last observed in a primary sample as its state for that primary sample.

The probabilities of movement between a pair of sites can only be estimated for intervals between primary

samples for which there was at least one observed movement. We’ve set the transition probabilities

between pairs of sites for intervals for which no movements were observed to zero.

The probability of capture (p) may be specified to vary by any or all of site, primary sample and secondary

sample. After elimination of poorly fitting models, we found no need to model the variation over

secondary samples within primary samples for humpback or bottlenose dolphins but this was necessary for

snubfin dolphins due to their very uneven capture rates over time. Models are reported with p varying by

site and by site by primary sample [p(site) and p(site*time] for humpbacks where there were three

observable states, by site or constant over sites [p(site) and p(constant)] for bottlenose where there were

two observable states, and by primary sample by secondary sample [p(primary*secondary)] for snubfin

where there was only one observable state.

The probability of capture (p) may be distinguished from the probability of recapture (c) following first

capture to indicate a behavioural response to first capture to persistent avoidance of recapture (‘trap-

shyness’) or to persistent seeking of recapture (‘trap-happiness’). While this sort of effect has been

observed in classic trapping studies with small mammals for example, it is considered unlikely in the

present situation where the dolphins are not actually trapped or otherwise interfered with and largely

habituated to the presence of vessels prior to their first capture. We consider only models in which the

probabilities of capture and recapture are equal on all occasions.

The modelling process involves fitting a set of models with alternative parameter structures and comparing

them for fit to data and parsimony. Models were compared with the Akaike Information Criterion

corrected for small sample sizes (AICc, Burnham and Anderson 2002), with smaller values of AICc



Page 12

indicating better fitting models, and with AICc weights, which measure the relative likelihoods of the

models in the set. When one model in the set had a clearly lower AICc than all others and attracted the

major proportion of the AICc weight, the parameter estimates from this ‘best’ model are reported; when

several models had similar AICc values and shared the AICc weight model-averaging may be applied

(Buckland et al. 1997) whereby a weighted average of the parameter estimates from several models are

reported.

Program MARK (V6.1; White and Burnham 1999) and the Multistate Closed Robust Design model were

employed for the analysis. The models, their AICc values, AICc weights, likelihoods and numbers of

parameters are reported in the Appendix, Table A3. The parameter estimates from the best fitting model or

model averaged estimates for each species at each site in each primary sample are reported in the

Appendix, Table A4.

Habitat use

A dolphin population may change the relative frequency with which different parts of its habitat are used.

We used a Binary Logistic Mixed Effects model to examine changes in spatial habitat use of dolphins in

the three sites. A 3.2 km x 3.2 km grid was placed over a map of the three sites (Bynoe Harbour, Darwin

Harbour and Shoal Bay) and data on sightings of dolphin groups and sampling intensity in each grid cell in

each secondary sample were modelled to estimate the relative probability of sighting at least one dolphin

in a transect pass of a given length in each grid cell.

The grid cells were grouped into 10 coherent locations in the three sites, with three, five and two locations

in Bynoe Harbour, Darwin Harbour and Shoal Bay respectively: in Bynoe Harbour, 18 cells were grouped

as outer, 15 as middle and 13 as upper; in Darwin Harbour, nine cells were grouped as outer, 10 as East

Arm, 12 as Middle Arm, seven as central and nine as West Arm; and in Shoal Bay, seven cells were

grouped as Gunn Point and 10 as Hope Inlet/Buffalo Creek.

The transects through the region moved slightly between primary samples and the interaction of variable

transect locations and fixed grid cells resulted in 110 cells being sampled in at least one of the four primary

samples: 105 cells were sampled in all four primary samples, one was sampled in three and four were

sampled in only one. A map of the grid cells overlaid on the transect lines in primary sample four is shown

in Figure 5.



Page 13

Figure 5 Map of grid cells overlaid on transect lines (primary sample four)



Page 14

While the data modelled were the binary response ‘dolphin sighted’ / ‘no dolphin sighted’ of each cell in

each secondary sample, the results from the model are estimated binomial means (proportions or

probabilities of sighting a dolphin) based on the binary data accumulated over secondary samples and

locations. The principal factors assessed were the primary sample and the location. Changes in habitat use

between the primary samples are modelled by the primary sample by location interaction effect which

assesses whether the relative rates of usage of the locations varies significantly between primary samples.

A non-significant interaction effect would indicate a lack of evidence of change in the relative rates of use

of the locations over the primary samples. A significant interaction effect indicate a correlation with

potential inference to impact only if the interaction involved locations closer and further way from

construction activities between primary samples.

The grid cells were sampled with different intensities, measured here in terms of the total length of transect

through a cell in each secondary sample. A single transect through and aligned with the cell would be 3.2

km long but most transects passed through cells at angles and were not aligned with them, and there was

more than one transect segment through some cells.

Primary sample, location and their interaction were initially fitted together with a linear function of

transect length through the cell to adjust for sampling intensity. The model was systematically reduced by

eliminating non-significant effects (p≥0.05) one at a time in order of their p-value sizes.

We report a model for the sightings of any species of dolphin (i.e., all three species pooled). It may be

possible in future to fit a model for humpbacks but there are too few sightings of bottlenose and snubfin to

model these species separately.

The model was fitted to the sightings data with random factors for the cell and the repeated measures on

the cell with a first order autoregressive structure (AR1) fitted to the repeated measures residuals. The

‘Genlinmixed’ procedure in SPSS V21 was employed for the analysis. Genlinmixed uses a pseudo-

likelihood method to obtain estimates with fixed effects tested by F statistics.



Page 15

Results

Abundance, Apparent Survival and Movements

Humpback dolphin

Model selection

Model comparison results are reported for a set of sixteen of the best-fitting (smallest AICc) models in

Table A3. The best fitting model had apparent survival varying by site [S(site)], transitions between Bynoe

Harbour and Darwin Harbour and between Darwin Harbour and Bynoe Harbour varying by primary

sample, and no temporary emigration [psi(a_b*t, b_a*t, No TE] and capture probability varying by both

site and primary sample [p(site*time)]. This model attracted 90% of the AICc weight. The second best-

fitting model attracted a further 8% of the AICc weight and differed from the best-fitting model only in

having variation on capture probability by site but not by primary sample. The parameter estimates from

the best-fitting model are reported in Table A4.

Apparent survival

There was no support for variation in apparent survival (the annual probability of both remaining alive and

on site) of humpback dolphins between successive pairs of the four primary samples, but strong support for

differences among the three sites (Table A3). Apparent survival was greater in Darwin Harbour than either

of Bynoe Harbour or Shoal Bay, with estimated apparent survival in Darwin Harbour = 0.93 (95%CI =

0.74:0.98) per annum, in Bynoe Harbour = 0.49 (95%CI = 0.31:0.67) per annum, and in Shoal Bay = 0.49

(95%CI = 0.24:0.75) per annum.

With no reason to expect differences in biological survival between sites, the lower rates of apparent

survival in Bynoe Harbour and Shoal Bay than in Darwin Harbour indicate emigration from those sites,

and with no observed movements between Bynoe Harbour and Shoal Bay, it seems likely that the

emigration from Bynoe Harbour is to the west and from Shoal Bay to the east.

Movements

No movements were observed between Bynoe Harbour or Darwin Harbour and Shoal Bay in either

direction. Movements were observed between Bynoe Harbour and Darwin Harbour in both directions

between each successive pair of primary samples. A total of fifteen such movements were observed with

most occurring between Bynoe Harbour and Darwin Harbour between primary samples two and three

(five), and between Darwin Harbour and Bynoe Harbour between primary samples three and four (five).

There was significant variation in humpback dolphin movement rates between the four primary samples.

The estimated rates of movement from Bynoe Harbour to Darwin Harbour between primary samples one



Page 16

and two, two and three, and three and four were 6% (95%CI = 1:33%), 38% (95%CI = 17:64%) and 13%

(95%CI = 2:53%) respectively. The estimated rates of movement in the opposite direction, from Darwin

Harbour to Bynoe Harbour, between successive primary samples were 10% (95%CI = 2:32%), 4% (95%CI

= 1:21%) and 20% (95%CI = 9:38%) respectively.

These rates indicate movement of occasionally substantial proportions of the population between Darwin

Harbour and Bynoe Harbour in either direction, while no movements were observed between Shoal Bay

and either of the other two sites. Although the population was relatively small in Shoal Bay (see below),

limiting the probability of observing movements to or from this site, these results suggest that the Shoal

Bay population may be relatively independent of the more closely related Bynoe and Darwin populations.

Abundance

The estimated size of the Bynoe Harbour humpback population in primary samples one, two, three and

four was 29 (95%CI=26:41), 29 (95%CI=25:42), 20 (95%CI=14:41), and 29 (95%CI=24-44) respectively.

In Darwin Harbour there were 39 (95%CI=35:51), 37 (95%CI=33:48), 41 (95%CI=36:53), and 37

(95%CI=35:44) respectively, while in Shoal Bay they were 14 (95%CI=12:28), 13 (95%CI=12:22), 21

(95%CI=17:34), and 17 (95%CI=10:42) respectively.

These estimates are quite stable over primary samples, with the total for all three sites varying only slightly

with 82, 79, 82 and 83 over primary samples. The reduction in the Bynoe Harbour estimate between

primary samples two and three corresponds with the larger rate of movement from Bynoe Harbour and

Darwin Harbour than from Darwin Harbour and Bynoe Harbour in the interval between primary samples

two and three. That the Darwin Harbour population increased slightly from primary samples two and three

is consistent with this.

While the Shoal Bay population also increased between primary samples two and three, there were no

observed movements from either Bynoe Harbour or Darwin Harbour to account for it. While some such

movement may have occurred, the increase may also have been due to recruitment into Shoal Bay.

Bottlenose dolphin

Very few bottlenose dolphins were ever observed in Bynoe Harbour, with four observed there in single

secondary samples in each of primary samples one and three (Table A2). There were too few data to derive

estimates for Bynoe Harbour and barely sufficient to derive estimates for Shoal Bay. The data from these

two sites were pooled and Bynoe Harbour and Shoal Bay modelled as a single site (BH&SB).

No movements were observed between Bynoe Harbour or Shoal Bay and Darwin Harbour within any

primary sample.



Page 17

Model selection

Model comparison results from a set of eight better fitting (lower AICc; except for models without

temporary emigration which were included for comparison) models are reported in Table A3. All models

have capture probability varying by both site and primary sample [p(site*time)], and all estimate only the

transitions between either Bynoe Harbour or Shoal Bay and Darwin Harbour between primary samples two

and three, and between Darwin Harbour and either Bynoe Harbour or Shoal Bay between primary samples

three and four [psi(ac_b2, b_ac3,…)]. The models vary in whether they estimate temporary emigration as

time-varying and random [psi(…, random)], time-varying and Markovian [psi(…, Markov)], constant and

random [psi(…, constant random)], or as having no temporary emigration [psi(…, No TE)]. The models

also vary in whether they estimate apparent survival as constant over both sites and primary samples

[S(constant)] or as varying by site [S(site)].

The best fitting model had a very simple structure with apparent survival constant over both sites and

primary samples and constant random temporary emigration. This model attracted 97% of the AICc weight

in the set. Models with no temporary emigration fitted very poorly. The parameter estimates from the best-

fitting model are reported in Table A4.

Apparent survival

There was no evidence of apparent survival of bottlenose dolphins differing between each of the four

primary samples or between the two sites (Bynoe and Shoal Bay combined and Darwin Harbour) (Table

A4). The estimated apparent survival was 0.81 (95%CI=0.63:0.91) per annum in both sites over all

intervals between primary samples. This is likely to be lower than the rate of biological survival and

indicates some degree of emigration from the local region.

Movements

Three individuals were observed to have moved between either Bynoe Harbour or Shoal Bay and Darwin

Harbour between primary samples two and three, and one individual was observed to have moved from

Darwin Harbour and either Bynoe Harbour or Shoal Bay between primary samples three and four.

There was significant variation in bottlenose dolphin movement rates in the intervals between the four

primary samples. Thirty nine percent (95%CI=13%:73%) of the population present in either Bynoe

Harbour or Shoal Bay in primary sample two were estimated to have moved to Darwin Harbour before

primary sample three, and eleven percent (95%CI=1%:52%) of the population in Darwin Harbour in

primary sample three were estimated to have moved to either Bynoe Harbour or Shoal Bay before primary

sample four.

Temporary emigration from either Bynoe Harbour or Shoal Bay was estimated at zero, while temporary

emigration from Darwin Harbour was estimated at 6% (95%CI=1%:25%). While the estimated rate of



Page 18

temporary emigration was relatively small and the confidence interval was wide, there was strong evidence

that it was non-zero from the very poor fit of models in which it was assumed to be zero.

Abundance

The estimated size of the Bynoe Harbour/Shoal Bay population in primary samples one, two, three and

four was 5 (95%CI=5:5), 10 (95%CI=9:19), 6 (95%CI=5:16), and 7 (95%CI=5:21) respectively, while in

Darwin Harbour there were 19 (95%CI=19:25), 15 (95%CI=14:22), 25 (95%CI=20:40), and 15

(95%CI=15:15).

These are small populations but the estimates are reasonably stable over primary samples. There were

slightly more in Bynoe Harbour and Shoal Bay in primary sample two and in Darwin Harbour in primary

sample three. The total for all three sites varied only slightly with 24, 25, 31 and 22 over primary samples.

It was possible to derive these estimates only because of the relatively high capture probability for this

species.

Snubfin dolphin

There were too few data to obtain separate estimates for Darwin Harbour or Shoal Bay, and while there

were more captures in Bynoe Harbour, the rates of capture and recapture there were highly irregular. Only

four snubfin were observed at more than one site over primary samples one, two, three and four. The data

from all three sites were combined and models fitted for the whole local region.

Model selection

Model comparison results from a set of eight better fitting (lower AICc) models are reported in Table A3.

All models have capture probability varying by both primary sample and secondary sample

[p(primary*secondary)].

The best fitting (lowest AICc) model had apparent survival constant over primary samples and constant

random temporary emigration (Table A3). This model attracted 73% of the AICc weight. A model with

constant Markovian temporary emigration attracted a further 15% of the AICc weight. There is good

evidence of temporary emigration with the best-fitting model with no temporary emigration having an

AICc of 5.9 greater than the best-fitting model. We report estimates from the best-fitting random temporary

emigration model and comment on the difference in the temporary emigration estimates between this and

the Markovian temporary emigration model. The estimates from the best fitting model are reported in

Table A4.



Page 19

Apparent survival

There was no evidence of variation in apparent survival of snubfin dolphins over the intervals between

successive primary samples. The estimated apparent survival was 0.94 (95%CI=0.00:1.00) per annum with

a very wide confidence interval indicating that this estimate was based on very little information.

Movements

The only movements in this model are movements to and from the unobservable state, or temporary

emigration estimates. The temporary emigration estimate from the best-fitting, random temporary

emigration model was 0.63 (95%CI=0.44:0.79). This model assumes that, for each interval between

successive primary samples, the probability of staying away after an absence is the same as the probability

of leaving. The temporary emigration estimate from the Markovian temporary emigration model indicates

that the probability of staying away after an absence was very slightly lower than the probability of

leaving. There is clearly substantial temporary emigration from the local region for this species, suggesting

that the sample area may be only a small part of the home range of the population that uses it. Only about

37 percent of that population are present in the sample area during a primary sample.

Abundance

The estimated size of the local regional population in primary samples one, two, three and four was 164

(95%CI=65:555), 18 (95%CI=16:28), 29 (95%CI=24:44), and 17 (95%CI=17:17) respectively. The very

large estimate for primary sample one appears to be anomalous with only one of thirty individuals

recaptured. Overall, and discounting the estimate for primary sample one, it appears that while the

numbers vary between primary samples, 30 (the number captured in Bynoe Harbour in primary sample

one) or more snubfin dolphins may use the local region during a primary sample.

Summary of abundance estimates

The abundance estimates at the three sites (Bynoe Harbour, Darwin Harbour, Shoal Bay) for humpback

dolphins, at two sites (Bynoe Harbour and Shoal Bay, Darwin Harbour) for bottlenose dolphins and in the

local region (Bynoe Harbour, Darwin Harbour and Shoal Bay) for snubfin dolphins by primary sample are

summarised in Table 1.



Page 20

Table 1 Summary of abundance estimates for the three species over four primary samples

Species Site Primary sample Estimate SE L95%CI U95%CI

Humpback Bynoe Harbour 1 29 3.16 26.27 40.56

2 29 3.78 25.03 41.67

3 20 6.09 13.52 41.02

4 29 4.53 24.37 44.34

Darwin Harbour 1 39 3.73 35.16 51.45

2 37 3.34 33.03 47.56

3 41 3.99 36.25 53.19

4 37 1.72 35.32 43.85

Shoal Bay 1 14 3.31 11.65 27.92

2 13 1.82 12.19 21.86

3 21 3.73 17.33 34.44

4 17 6.76 10.30 41.45

Bottlenose

Tursiops Bynoe Harbour 1 5 0.00 5.00 5.00

and Shoal Bay 2 10 1.77 9.10 19.26

3 6 1.90 5.10 16.10

4 7 2.94 5.29 21.06

Darwin Harbour 1 19 0.92 19.00 25.12

2 15 1.33 14.05 22.14

3 25 4.55 20.42 40.47

4 15 0.00 15.00 15.00

Snubfin Bynoe Harbour, 1 1641 105.58 64.66 554.55

Darwin Harbour 2 18 2.14 16.23 27.55

and Shoal Bay 3 29 4.49 23.92 44.11

4 17 0.00 17.00 17.00

Notes: 1 Estimate appears anomalous and is based on a single recapture



Page 21

Habitat use

A map of the sighting positions of dolphin groups over the four primary samples is shown in Figure 6.

Figure 6 Map of sighting positions of dolphin groups over the four primary samples



Page 22

Results of the tests of fixed effects for a series of 3 Binary Logistic models are reported in Table 2.

Table 2 Tests of fixed effects in 3 models

Model Effect F df1 df2 p

1 transect length 104.294 1 3802 0.000

primary 1.656 3 3802 0.174

location 2.862 9 3802 0.002

primary * location 1.362 27 3802 0.101

2 transect length 107.259 1 3829 0.000

primary 1.615 3 3829 0.184

location 2.489 9 3829 0.008

3 transect length 107.041 1 3832 0.000

location 2.481 9 3832 0.008

Model 1. The primary sample by location interaction effect was non-significant and was removed to build

model 2.

Model 2. The primary sample effect was non-significant removed to build model 3.

Model 3. Model 3 fitted only the main effects for transect length and location. Both the transect length

(sample intensity) covariate and location effects were clearly significant.

Although the primary by location effect was not significant (p=0.101), the estimated mean probability of

sighting a dolphin group from a secondary sample with the mean transect length of 3.78 km is plotted by

location by primary sample in Figure 7 for descriptive purposes. The estimated mean probability of

sighting a dolphin group from a secondary sample with the mean transect length of 3.78 km is plotted with

95% confidence interval by location in Figure 8.

A map of the region with grid cells colour-coded to show relative rates of habitat usage in the locations is

shown in Figure 9.



Page 23

Figure 7 Estimated relative probability of sighting a dolphin group in a secondary sample of transect length

= 3.78 km with 95%CI by location by primary sample.

Figure 8 Estimated relative probability of sighting a dolphin group in a secondary sample of transect length

= 3.78 km by location with 95% confidence intervals.

.000

.020

.040

.060

.080

.100

.120

.140

.160

.180

.200

BH outer BH

middle

BH

upper

DH

outer

DH East

Arm

DH

Middle

Arm

DH

central

DH

West

Arm

SB Gunn

Point

SB Hope

Inlet /

Buffalo

Creek

P1

P2

P3

P4

.000

.050

.100

.150

.200

.250

BH outer BH

middle

BH upper DH outer DH East

Arm

DH

Middle

Arm

DH

central

DH West

Arm

SB Gunn

Point

SB Hope

Inlet /

Buffalo

Creek



Page 24

Figure 9 Map of Darwin region showing relative rates of habitat use over locations



Page 25

With 45 pairwise tests between locations, there was no significant difference between any pair of locations

when sequential Bonferroni adjustment was applied. This is a very conservative criterion with many tests

and the closest pairwise difference was a greater probability of sighting in Gunn Point than Darwin central

(p = 0.232 Bonferroni adjusted).

It may be noted however, that probabilities of sighting were relatively high overall in Shoal Bay (Gunn

Point and Hope Inlet/Buffalo Creek) and in the outer Harbours (Bynoe outer, Darwin outer and Gunn

Point), and within Darwin Harbour, in West Arm.

Summary of habitat use

The estimated mean probabilities of sighting in the locations had reasonable precision with coefficients of

variation (CV = standard error/estimate, a measure of precision) ranging from 14% (Bynoe outer) to 20%

(Darwin central) (mean CV = 17% ±1.9%).

We have provided estimates of the relative rates of use of the locations and information about the natural

variation in these estimates over primary samples. It’s notable that the outer harbour areas are relatively

well used and that this is consistent with and complementary to evidence from the capture-recapture

models of movement into and out of the Darwin region: i.e., there is evidently a flow of dolphins not only

across the outer Harbour areas in the Darwin region but also to and from sites west of Bynoe Harbour and

east of Shoal Bay.



Page 26

Conclusion The major constraint on obtaining estimates of abundance, movement and apparent survival was the small

sizes of the populations of bottlenose and snubfin dolphins, with the snubfin appearing to have irregular

patterns of visitation to the sampling areas.

Where too few data were available to model for each site separately, estimates were obtained by pooling

the data over sites. Adequate data were available to obtain estimates of abundance for humpback dolphins

in all three sites, to estimate their rates of movement between Bynoe Harbour and Darwin Harbour,

temporary emigration from the local region, and their apparent survival at all three sites.

The data for Bynoe Harbour and Shoal Bay were pooled for bottlenose dolphins and estimates were

obtained of abundance, apparent survival for Bynoe Harbour/Shoal Bay and Darwin Harbour, temporary

emigration from the local region, and rates of movement between the two sites.

The data from all three sites were pooled for snubfin dolphins and estimates were obtained of abundance,

temporary emigration and apparent survival for the whole local region.

While the habitat use model was fitted only for sighting of dolphin groups independently of the species

(i.e., all three species combined), it should be possible in future to fit the model to the data on sightings of

the humpback dolphin. We found no significant change in the relative rates of use of the defined locations

between primary samples and note that, over all primary samples, the outer Harbour areas (outer Bynoe

Harbour, outer Darwin Harbour and Shoal Bay, especially Gunn Point) are relatively well used and that

this is consistent with and complementary to evidence from the capture-recapture models of movement

into and out of the Darwin region: i.e., there is evidently a flow of dolphins not only across the outer

Harbour areas in the Darwin region but also to and from sites west of Bynoe Harbour and east of Shoal

Bay.

Overall, although it was necessary to pool data over some sites for bottlenose and snubfin dolphins, the

sampling design has generated data of a quantity and structure that allows the models to generate

reasonably precise estimates of abundance, movements and other demographic parameters, and of the

relative frequency of use of parts of the sample area.

While the abundances of the three species are small, they have remained stable over the duration of the

study so far, and the relative rates of use of different parts of the habitat have not changed.



Page 27

REFERENCES

Amstrup, S.C., McDonald, T.L., & Manly, B.F.J. (Eds). (2005). Handbook of Capture-Recapture Analysis.

Princeton University Press, Princeton, NJ.

Arnason, A.N. (1972). Parameter estimates for mark-recapture experiments on two populations subject to

migrations and death. Researches in Population Ecology 13: 99-113.

Arnason, A.N. (1973). The estimation of population size, migration rates and survival in a stratified

population. Researches in Population Ecology 15: 1-8.

Brooks, L. and Pollock, K.H. (2011). A Sampling Design for Monitoring Dolphin Populations in Darwin

Harbour. Unpublished report held by NTDLRM.

Brooks, L. and Pollock, K.H. (2012). Assessment of Coastal Monitoring Program for Darwin Harbour: An

evaluation of the sampling design after two primary samples. Unpublished report held by NTDLRM.

Brooks, L. & Pollock, K. (2013). Abundance, movements and habitat use of coastal dolphins in the Darwin

region: Analysis of the first three primary samples. Unpublished report held by NTDLRM.

Brownie, C., Hines, J.E., Nichols, J.D., Pollock, K.H., & Hestbeck, J.B. (1993). Capture-recapture studies

for multiple strata including non-Markovian transition probabilities. Biometrics 49: 1173-1187.

Buckland, S. T., Burnham, K. P. and Augustin, N. H. (1997). Model selection: an integral part of

inference. Biometrics 53:603-618.

Burnham, K. P., & Anderson, D. R. (2002). Model selection and multi-model inference: a practical

information-theoretic approach. Springer.

Griffiths, A.D. and Palmer, C.P. 2011 Monitoring Plan for Coastal Dolphin Species in Darwin Harbour.

Department of Natural Resources, Environment, the Arts and Sport, Palmerston.

Hammond, P. S. and Thompson, P. M. (1990). Minimum estimate of the number of bottlenose dolphins

(Tursiops truncatus) in the Moray Firth. Biological Conservation 56, 79-88.

Kendall, W. L., and J.D. Nichols. 1995. On the use of secondary capture-recapture samples to estimate

temporary emigration and breeding proportions. Journal of Applied Statistics, 22, 751-762.

Kendall; W.L., Pollock; K.H., & Brownie, C. (1995). A Likelihood-Based Approach to Capture-Recapture

Estimation of Demographic Parameters under the Robust Design. Biometrics, 51, No. 1: 293-308.

Kendall, W.L., Nichols, J.D. & Hines, J.E. (1997). Estimating Temporary Emigration Using Capture-

Recapture Data with Pollock's Robust Design. Ecology 78, No. 2: 563-578.

Kendall,W. L., and J.D. Nichols. 2002. Estimating state-transition probabilities for unobservable states

using capture-recapture/resighting data. Ecology, 83, 3276-3284.



Page 28

Kendall, W.L. 2013. The robust design. In Cooch, E.G. & White, G.C. “Program Mark – a gentle

introduction (Edition 13), Chapter 15. http://www.phidot.org/software/mark/docs/book/.

Mendez, M., Jefferson, T. A., Kolokotronis, S.-O., Krützen, M., Parra, G. J., Collins, T., Minton, G.,

Baldwin, R., Berggren, P., Särnblad, A., Amir, O. A., Peddemors, V. M., Karczmarski, L., Guissamulo, A.,

Smith, B., Sutaria, D., Amato, G. and Rosenbaum, H. C. (2013), Integrating multiple lines of evidence to

better understand the evolutionary divergence of humpback dolphins along their entire distribution range: a

new dolphin species in Australian waters?. Molecular Ecology. doi: 10.1111/mec.12535

Nichols, James D., and Cynthia J. Coffman. "Demographic parameter estimation for experimental

landscape studies on small mammal populations." In Landscape Ecology of Small Mammals, pp. 287-309.

Springer New York, 1999.

Nicholson, K., Bejder, L., Allen, S., Krützen, K. and Pollock, K. (2012). Abundance, survival and

temporary emigration of bottlenose dolphins (Tursiops sp.) off Useless Loop in the western gulf of Shark

Bay, Western Australia. Marine and Freshwater Research, 63: 1059-68.

Parra, G. J., Corkeron, P. J. and Marsh, H. (2006). Population sizes, site fidelity and residence patterns of

Australian snubfin and Indo-Pacific humpback dolphins: Implications for conservation. Biological

Conservation, 129,167-180.

Pollock, K. H. (1982). A capture-recapture design robust to unequal probability of capture. Journal of

Wildlife Management, 46: 752-757.

Pollock, K. H., Nichols, J. D., Brownie, C., and Hines, J. E. (1990). Statistical Inference for Capture-

Recapture Experiments, Wildlife Society Monographs (no. 107).

Program MARK V6.1. 2012. http://warnercnr.colostate.edu/~gwhite/MARK/MARK.htm .

SPSS V21. IBM Corp. Released 2012. IBM SPSS Statistics for Windows, Version 21.0. Armonk, NY:

IBM Corp.

Schwarz, C.J., Schweigert, J.F. and Arnason, A. N. (1993). Estimating migration rates using tag recovery

data. Biometrics, 49: 177-193.

White, G.C. and K.P. Burnham. (1999). Program MARK: survival estimation from populations of marked

animals. Bird Study, Vol. 46, No. S., pp. 120-139.

Williams, B. K., Nichols, J. D. and Conroy, M. J. (2002). Analysis and Management of Animal

Populations. Academic Press, San Diego, California.

Würsig, B. and Jefferson, T. A. (1990). Methods of photo-identification for small cetaceans. In Individual

Recognition of Cetaceans: Use of Photo-Identification and Other Techniques to estimate population



Page 29

parameters, eds. P. S. Hammond, S. A. Mizroch, G. P. Donovan, pp43-52, Special Issue 12. International

Whaling Commission, Cambridge.



Appendix

Table A1 Dates of secondary samples by site for each primary sample Page 1

Table A2 Summary of captures by species, site and primary sample Pages 2-3

Table A3 Model comparison results for models on each species Pages 4-6

Table A4 Parameter estimates from selected models for each species Pages 7-12



Page 1

Table A1 Dates of secondary samples by site for each primary sample

Primary Secondary sample

sample Site 1 2 3 4 5 6 7 8 9

1 Darwin 02.11.2011 03.11.2011 04.11.2011 09.11.2011 10.11.2011 11.11.2011 16.11.2011 17.11.2011 18.11.2011

1 Bynoe 29.10.2011 30.10.2011 31.10.2011 05.11.2011 06.11.2011 07.11.2011 12.11.2011 13.11.2011 14.11.2011

1 Bynoe

15.11.2011

1 Shoal Bay 20.10.2011 21.10.2011 22.10.2011 05.11.2011 06.11.2011 07.11.2011 12.11.2011 13.11.2011 14.11.2011

1 Shoal Bay

08.11.2011

2 Darwin 26.03.2012 27.03.2012 28.03.2012 02.04.2012 03.04.2012 04.04.2012 09.04.2012 10.04.2012 11.04.2012

2 Darwin

12.04.2012

2 Bynoe 29.03.2012 30.03.2012 31.03.2012 05.04.2012 06.04.2012 07.04.2012 13.04.2012 14.04.2012 15.04.2012

2 Shoal Bay 29.03.2012 30.03.2012 31.03.2012 05.04.2012 06.04.2012 07.04.2013 12.04.2012 13.04.2012 14.04.2012

3 Darwin 08.10.2012 09.10.2012 10.10.2012 15.10.2012 16.10.2012 17.10.2012 22.10.2012 23.10.2012 24.10.2012

3 Bynoe 11.10.2012 12.10.2012 13.10.2012 18.10.2012 19.10.2012 20.10.2012 25.10.2012 26.10.2012 27.10.2012

3 Shoal Bay 11.10.2012 12.10.2012 13.10.2012 18.10.2012 19.10.2012 20.10.2013 25.10.2012 26.10.2012 27.10.2012

3 Shoal Bay

19.10.2012

4 Darwin 13.03.2013 14.03.2013 15.03.2013 20.03.2013 21.03.2013 22.03.2013 27.03.2013 28.03.2013 29.03.2013

4 Bynoe 16.03.2013 17.03.2013 18.03.2013 23.03.2013 24.03.2013 25.03.2013 31.03.2013 1.04.2013 2.04.2013

4 Shoal Bay 16.03.2013 17.03.2013 18.03.2013 23.03.2013 24.03.2013 25.03.2013 31.03.2013 1.04.2013 2.04.2013



Page 2

Table A2 Summary of captures by species, site and primary sample

Species Site Primary

sample

Secondary

samples1 Individuals Max.Caps.

2 Captures

Snubfin Bynoe 1 5 25 2 26




Snubfin Darwin 1 1 5 1 5




Snubfin Shoal Bay 1 0 0 0 0




Snubfin All sites 1 5 30 2 31




Humpback Bynoe 1 9 26 4 49




Humpback Darwin 1 9 33 4 62





Page 3


Humpback Shoal Bay 1 5 11 3 18




Humpback All sites 1 9 69 4 129




Bottlenose Bynoe 1 1 4 1 4




Bottlenose Darwin 1 5 18 4 45




Bottlenose Shoal Bay 1 3 5 3 10




Bottlenose All sites 1 6 23 4 59



Bottlenose All sites 4 6 20 4 43 1

Number of secondary samples with non-zero captures

2 Maximum number of captures per individual



Page 4

Table A3 Model comparison results for models on each species Notes:

1. S = probability of apparent survival, psi = probability of transition including temporary emigration, p = probability of capture, c= probability of recapture, *t

indicates variation over primary samples.

2. a indicates Bynoe Harbour (BH), b indicates Darwin Harbour (DH), c indicates Shoal Bay (SB), ab indicates Bynoe Harbour & Shoal Bay, u indicates

unobservable (temporarily absent from region), a_b indicates transition from a to b.

3. S(…, u=b) indicates apparent survival for temporarily absent dolphins set equal to Darwin Harbour.

AICc Model Number of

Species Sites Model AICc ∆AICc Weight Likelihood Parameters

Humpback BH,DH,SB {S(site, u=b), 1625.516 0.000 0.902 1.000 33

psi(a_b*t, b_a*t, No TA),

p(site*time)}



p(site)}

Humpback BH,DH,SB {S(site*time, u=b), 1634.189 8.673 0.012 0.013 39


p(site*time)}

Humpback BH,DH,SB {S(site*time, u=b), 1638.279 12.763 0.002 0.002 30


p(site)}


psi(a_b*t, b_a*t, a_u*t=u_a*t, b_u*t=u_b*t, c_u*t=u_c*t, a_c=0, b_c=0 - even flow),

p(site*time)}

Humpback BH,DH,SB {S(site, u=c), 1640.509 14.993 0.001 0.001 42



Page 5


p(site*time)}

Humpback BH,DH,SB {S(site, u=a), 1642.350 16.834 0.000 0.000 33


p(site)}


psi(a_b*t, b_a*t ,a_u*t=u_a*t, b_u*t=u_b*t, c_u*t=u_c*t, a_c=0, b_c=0 -even flow),

p(site)}


psi(a_b*t, b_a*t, a_u*t=u_a*t ,b_u*t=u_b*t, c_u*t=u_c*t, a_c=0, b_c=0 - even flow),

p(site)}



p(site*time)}


psi(a_b*t, b_a*t, a_u*t=(1-u_a*t), b_u*t=(1-u_b*t), c_u*t=(1-u_c*t), a_c=0, b_c=0 - random),

p(site*time)}



p(site)}



p(site*time)}



p(site)}





Page 6

p(site)}



p(site*time)}

Bottlenose BH&SB,DH {S(constant), psi(ac_b2, b_ac3, constant random), p(site*time)} 1922.371 0.000 0.970 1.000 21

Bottlenose BH&SB,DH {S(constant), psi(ac_b2, b_ac3, random), p(site*time)} 1931.471 9.100 0.010 0.011 25

Bottlenose BH&SB,DH {S(site, u=b), psi(ac_b2, b_ac3, random), p(site*time)} 1931.793 9.422 0.009 0.009 26

Bottlenose BH&SB,DH {S(site, u=ac), psi(ac_b2, b_ac3, random), p(site*time)} 1931.940 9.569 0.008 0.008 26

Bottlenose BH&SB,DH {S(site, u=b), psi(ac_b2, b_ac3, Markov), p(site*time)} 1934.387 12.016 0.002 0.003 27

Bottlenose BH&SB,DH {S(site, u=ac), psi(ac_b2, b_ac3, Markov), p(site*time)} 1937.154 14.783 0.001 0.001 28

Bottlenose BH&SB,DH {S(site, u=b), psi(ac_b2, b_ac3, No TE), p(site*time)} 3305.358 1382.987 0.000 0.000 20

Bottlenose BH&SB,DH {S(site, u=ac), psi(ac_b2, b_ac3, No TE), p(site*time)} 3305.358 1382.987 0.000 0.000 20

Snubfin BH&DH&SB {S(constant), psi(constant random), p(primary*secondary)} 142.562 0.000 0.728 1.000 24

Snubfin BH&DH&SB {S(constant), psi(constant Markov), p(primary*secondary)} 145.717 3.155 0.150 0.207 25

Snubfin BH&DH&SB {S(primary), psi(No TE), p(primary*secondary)} 148.463 5.901 0.038 0.052 23

Snubfin BH&DH&SB {S(constant), psi(primary random), p(primary*secondary)} 148.631 6.069 0.035 0.048 26

Snubfin BH&DH&SB {S(constant), psi(primary Markov), p(primary*secondary)} 148.915 6.353 0.030 0.042 26

Snubfin BH&DH&SB {S(primary), psi(primary Markov), p(primary*secondary)} 150.456 7.894 0.014 0.019 28

Snubfin BH&DH&SB {S(primary), psi(primary random), p(primary*secondary)} 152.741 10.179 0.004 0.006 28

Snubfin BH&DH&SB {S(constant), psi(No TE), p(primary*secondary)} 169.214 26.652 0.000 0.000 23



Page 7

Table A4 Parameter estimates from selected models for each species

Notes: Notation for the model is described above Table A3

Species Parameter Site Primary sample Secondary Estimate SE LCI UCI

Humpback Model: {S(site, u=b), psi(a_b*t, b_a*t, No TA), p(site*time)}

Apparent survival BH 1 to 2 = 2 to 3 = 3 to 4 0.49 0.09 0.31 0.67

Apparent survival DH 2 to 2 = 2 to 3 = 3 to 4 0.93 0.05 0.74 0.98

Apparent survival SB 3 to 2 = 2 to 3 = 3 to 4 0.49 0.14 0.24 0.75

Transition BH to DH 1 to 2 0.06 0.06 0.01 0.33



Transition DH to BH 1 to 2 0.10 0.07 0.02 0.32



Capture BH 1 0.18 0.03 0.13 0.25

Capture BH 2 0.15 0.03 0.11 0.21

Capture BH 3 0.08 0.03 0.04 0.15

Capture BH 4 0.14 0.03 0.09 0.20

Capture DH 1 0.18 0.03 0.13 0.24

Capture DH 2 0.18 0.02 0.14 0.23

Capture DH 3 0.16 0.02 0.13 0.20

Capture DH 4 0.27 0.03 0.22 0.33

Capture SB 1 0.14 0.04 0.07 0.25

Capture SB 2 0.20 0.04 0.13 0.29

Capture SB 3 0.14 0.03 0.09 0.22

Capture SB 4 0.07 0.03 0.03 0.14

Abundance BH 1 29 3.16 26.27 40.56

Abundance BH 2 29 3.78 25.03 41.67

Abundance BH 3 20 6.09 13.52 41.02

Abundance BH 4 29 4.53 24.37 44.34



Page 8

Abundance DH 1 39 3.73 35.16 51.45

Abundance DH 2 37 3.34 33.03 47.56

Abundance DH 3 41 3.99 36.25 53.19

Abundance DH 4 37 1.72 35.32 43.85

Abundance SB 1 14 3.31 11.65 27.92

Abundance SB 2 13 1.82 12.19 21.86

Abundance SB 3 21 3.73 17.33 34.44

Abundance SB 4 17 6.76 10.30 41.45

Bottlenose Model: {S(constant), psi(ac_b2, b_ac3, constant random), p(site*time)}

Apparent survival BHSB=DH All 0.81 0.07 0.63 0.91

Transition BHSB to DH 2 to 3 0.39 0.18 0.13 0.73

Transition (Random TE) BHSB to Absent 1 to 2 = 2 to 3 = 3 to 4 0.00 0.00 0.00 0.00

Transition DH to BHSB 3 to 4 0.11 0.11 0.01 0.52

Transition (Random TE) DH to Absent 1 to 2 = 2 to 3 = 3 to 4 0.06 0.05 0.01 0.25

Capture BHSB 1 0.29 0.08 0.16 0.45

Capture BHSB 2 0.32 0.08 0.18 0.50

Capture BHSB 3 0.21 0.07 0.10 0.39

Capture BHSB 4 0.16 0.07 0.06 0.35

Capture DH 1 0.39 0.05 0.30 0.48

Capture DH 2 0.40 0.06 0.28 0.53

Capture DH 3 0.18 0.04 0.12 0.26

Capture DH 4 0.39 0.05 0.30 0.50

Abundance BHSB 1 5 0.00 5.00 5.00

Abundance BHSB 2 10 1.77 9.10 19.26

Abundance BHSB 3 6 1.90 5.10 16.10

Abundance BHSB 4 7 2.94 5.29 21.06

Abundance DH 1 19 0.92 19.00 25.12

Abundance DH 2 15 1.33 14.05 22.14

Abundance DH 3 25 4.55 20.42 40.47

Abundance DH 4 15 0.00 15.00 15.00



Page 9

Snubfin Model: {S(constant), psi(constant random TE), p(primary*secondary)}

Apparent survival BH&DH&SB 1 to 2 = 2 to 3 = 3 to 4 0.94 0.22 0.00 1.00

Transition (Random TE) 1 to 2 = 2 to 3 = 3 to 4 0.63 0.09 0.44 0.79

Capture 1 2 0.09 0.06 0.02 0.31

Capture 1 3 0.02 0.02 0.00 0.11

Capture 1 5 0.01 0.01 0.00 0.06

Capture 1 7 0.01 0.01 0.00 0.06

Capture 1 9 0.07 0.05 0.02 0.26

Capture 2 2 0.40 0.13 0.19 0.65

Capture 2 4 0.34 0.12 0.15 0.59

Capture 2 6 0.06 0.06 0.01 0.31

Capture 2 8 0.68 0.14 0.39 0.88

Capture 3 3 0.04 0.04 0.00 0.22

Capture 3 5 0.04 0.04 0.00 0.22

Capture 3 7 0.39 0.11 0.21 0.61

Capture 3 8 0.25 0.09 0.12 0.46

Capture 3 9 0.43 0.11 0.24 0.64

Capture 4 1 0.29 0.11 0.13 0.54

Capture 4 2 0.94 0.06 0.67 0.99

Capture 4 5 0.06 0.06 0.01 0.32

Capture 4 8 0.41 0.12 0.21 0.65

Abundance 1 164 105.58 64.66 554.55

Abundance 2 18 2.14 16.23 27.55

Abundance 3 29 4.49 23.92 44.11

Abundance 4 17 0.00 17.00 17.00


abundance, movements and habitat use of coastal … · abundance, movements and habitat use of...

Documents