How to count kangaroos

8 Pages·1999·1.3 MB·English

by  H. I. Mccallum

How to count kangaroos
H. I. McCdum
Department of Zoology, University of Queensland, Queensland 4072
ABSTRACT
The fundamental reason for counting kangaroos is to ensure that harvests are
demonstrably sustainable. Given this goal, it is essential that the survey method should be
able to detect broad scale trends in population size. Aerial counts will detect trends only if
the constant ("correction factor") that relates the number of sightings to actual population size
either remains constant, or the way in which it changes depending on environmental conditions
is known. Actual counts, and not an index, are essential if harvesting policy is to be proactive,
rather than simply reacting to previous changes in population size. It is evident from other
papers in this volume that there is a particular problem in estimating appropriate correction
factors for grey kangaroos. Several contributors suggested a research programme to produce
better correction factors, which would then be used in the future to correct raw sighting data.
A better solution may be to use 'double sampling", which is a well-established sampling
method. Each time an aerial survey is conducted, a subsample of the transects surveyed would
be re-surveyed using helicopter counts, which are more accurate, but more expensive. The
ratio of the helicopter to fixed-wing counts is then used to produce a survey-specific correction
factor. Double sampling is more expensive than fixed-wing surveys alone, so it would be
necessary to decrease the intensity of the fixed-wing surveys to allow for the extra cost of
the helicopter surveys. The experiments necessary to see whether double sampling is
worthwhile are verv similar to those that are necessarv to develoo better correction factors. I
suggest that any eiperlrnenls deslgneo to Improve cor;ectlon factors snould also be analySe0
to nvest gate tne feaston ty of Jslng oo~bles ampl ng to estlmate kangaroo popu atlon slze
WHY COUNT KANGAROOS? harvests sustainably, particularly when social
and political issues are important. There is,
There is never much point in collecting
however, one very significant difference
data without thinking about what those data
between managing fish populations and
are to be used for. The underlying reason
managing populations of large terrestrial
that all Australian mainland States (except
vertebrates, such as kangaroos. Most of the
Victoria and the Northern Territory) spend
data used to monitor fish populations are
considerable resources in counting kangaroos
catch data. They come from the commercial
is to ensure that the kangaroo harvest is
exploitation itself. Many of the problems
managed appropriately. So, any discussion of
that have occurred in fisheries management
the adequacy of current methods of counting
are direct results of the tendency for catch
kangaroos must start with a discussion of the
per unit effort to remain high as abundance
objectives of the kangaroo management, and
decreases through overexploitation (Hilborn
the way in which data on kangaroo popula-
and Walters 1992). In contrast, the Australian
tions relate to these management objectives.
kangaroo harvest is monitored primarily
The general consensus of the workshop was by surveys, independent of the harvest.
that the fundamental objective of kangaroo Independent survey data do not have most
management is to maintain viable populations of the inherent problems of catch per unit
of all exploited species over their current effort data.
range, while allowing for both a sustainable
I do not propose to discuss sustainability in
and viable kangaroo industry and for reductions
detail, other than to note that a sustainable
in populations where they contribute to over-
harvest is one that does not lead to a
grazing.
continuing decline in the population being
Sustainability of harvests of wild animals is exploited, and that sustainable harvests may
discussed in great detail in the fisheries be at, below, or above, the point of maximum
literature (Hilborn and Walters 1992). However, sustainable yield. To my knowledge, no-one
few of the major exploited fish stocks have has ever suggested that an objective of
been managed sustainably (Beddington and kangaroo harvesting should be to extract the
Basson 1994; Hutchings and Myers 1994). maximum sustainable yield from the popula-
The Grand Banks cod fishery was arguably the tion. A viable harvest is not necessarily the
most important in the world, from a historical same thing as a sustainable one, although
and social perspective (Kurlansky 1997). It it is obvious that an unsustainable harvest
has now collapsed entirely. We should not be cannot be via6le in the long term. Viability
overconfident about our ability to manage means that the harvest must be capable of
June 1999 Australian Zoologist 31(1) 309
economically supporting participants in it. For Pollock 1995), the use of an index relies on
the specific purposes of this discussion, it is an implicit assumption that
essential to recognize that a viable harvest
depends crucially on the existence of
continuing markets for kangaroo products. where N is the actual population size, I is
Those markets will be threatened if the the index, and a is a constant of pmportion-
harvest cannot he demonstrated to be sustain- ality connecting the index to the actual
able, in the face of substantial hostility from population size.
certain organizations. The monitoring system
There are two implicit assumptions in eqn
must not only ensure that the harvest is
(1). First, there is the assumption of direct
sustainable, it must ensure that the harvest is
proportionality. If there is a curvilinear
seen to be sustainable.
relationship between the index and population
There is no general consensus about size, or if there is a non-zero intercept, the
the extent to which kangaroos actually do index cannot be used to compare relative
contribute to overgrazing, with the con- density through space or time. However,
sequences to either Land degradation or if the nonlinear relationship between the
competition with livestock. This is an index and actual population size can be
important debate, but is beyond the scope of described, the index can he still be converted
this article, and was not considered in detail into an estimate of population size. Cairns
at the workshop. Most primary producers (1999) provided evidence that the problem
believe that kangaroo populations do need of nonlinearity is not a major one if aerial
to he reduced, at least some of the time, sightings are used as an index of kangaroo
to prevent overgrazing. They are supported population.
in this belief by most state departments
responsible for primary industries, and by The second, and more serious, implicit
some biologists outside such departments assumption is that the constant a is indeed
(W~lsone l al. 1984; Freudenberger el al. 1997). constant, both through time and across
However, other biologists, particularly those space. This assumption is impossible to test
who have undertaken controlled experiments, if only the index I is available. If it can be
doubt that overgrazing by kangaroos is often tested and verified against the true population
a significant land management issue. For size, there is no need to use the index anyway,
example, Edwards el al. (1996) suggested that as eqn (1) could then be used to obtain an
competition between sheep and kangaroos estimate of actual population size. In some
rarely occurred. Whatever the actual reality, cases, this second assumption can be circum-
the political reality is that kangaroo shooting vented by estimating a series of constants.
for "pest reduction" purposes will continue Each might be appropriate for a particular
for the foreseeable future in most Australian habitat or season, or some combination
States. An objective of kangaroo monitoring is of both. To estimate these constants, or
therefore to ensure that such "pest reduction" "correction factors", it would be necessary
is consistent with the overall objective of to have a series of index values and actual
maintaining kangaroo populations as viable counts in various habitats, seasons etc.
across their range. To determine whether
Whenever an index is used alone, instead
kangaroo populations are at a level such that
of being converted to an actual estimate, what
they do cause land degradation is not a major
is being asserted is that, while the constant
objective of current monitoring schemes,
is unknown, it is known to be constant. This
although South Australia is moving in this
is a difficult assertion to maintain. In the
direction (Alexander 1997). It cannot become
context of kangaroo counting, the index is
a central objective until the relationship
typically an uncorrected aerial count, and
between kangaroo grazing and land degrada-
the constant is the "correction factor". For
tion is better established.
red kangaroos, there is good evidence that
correction factors do not change markedly
AN INDEX OR AN ESTIMATE
through time (Pople 1999). The 20 years
OF ABUNDANCE?
of red kangaroo survey data from South
An important issue, discussed at some length Australia (Grigg et al. 1999) are therefore a
in the workshop, is whether it is essential to regular and consistent record of the dynamics
produce an estimate of the actual number of of a large, wild herbivore over a spatial scale
kangaroos, or whether an index of abundance and length of time that is probably unique.
is adequate. It is often assumed that an index The only comparable data are for caribou
is an easier quantity to obtain and work with in the Arctic (Caughley and Gunn 1993), or
than an actual population estimate. As has possibly for wildebeest in Africa (Campbell
been pointed out a number of times (e.g., and Borner 1995), although neither of these
310 Australian Zoologist 31(1)
covers the spatial scale and has the regularity methods (see Seher 1982 for a review).
of the red kangaroo data. Although mark-recapture or resight methods
have been successful in estimating the local
Assuming that a is constant through time
ahundance of smaller macropods, there is
and space (something I will discuss later), the
little doubt that only direct visual counts are
question that then can be asked is whether
able to be used to estimate the abundance of
it is necessary to know its value. If we have
the larger macropods over the spatial scales
convinced ourselves (and hopefully others)
necessary to set harvest policies. It is also
that aerial counts are directly proportional
generally accepted that, of all current visual
to kangaroo ahundance, do we really need
census methods, line transects carried out on
to know what the constant of proportionality
foot produce estimates that are closest to
is? Given that a sustainable harvest is one
actual kangaroo abundance (Southwell 1994).
that does not lead to a continuing decline in
While they are valuable to calibrate other
kangaroo numbers, an index that does not
methods, foot transects are not possible at
decline through time demonstrates that the
the large scales of state-wide surveys. States
previous harvests have been sustainable. One
therefore base their population estimates on
could go further and suggest that if current
aerial surveys.
harvests are set on the same policy as
before, then an index that maintains its Queensland currently uses annual helicopter
value through time gives confidence that surveys using line transect methods (Lundie-
current harvests are sustainable too. The Jenkins et al. 1999). A line transect (Buckland
difficulty is that kangaroo populations et al. 1993) uses the frequency distribution
vary greatly through time, largely due to of distances of sighted individuals from
the influence of rainfall, hut the periods of the transect line to estimate ahundance. All
increase and decrease generated by rainfall individuals on the line are assumed to be
variation may he many years long (Caughley seen, but the method assumes that some
1987a; Grigg et al. 1999). This means that animals away from the line will be missed. A
evidence that a harvest level (whether defined strip transect, in contrast, simply counts all
as effort, numbers or proportion) is not individuals seen within a certain distance of
sustainable may not be clear for a number of the line). Evidence from Queensland research
years. Any setting of harvest levels only with (Clancy et al. 1997) suggests that helicopter
respect to an index therefore operates with a surveys record similar population densities
substantial lag. to walked transects, for red and eastern grey
kangaroos, but still undercount wallaroos
Currently, kangaroo harvest levels are
(euros). Helicopter surveys are expensive,
approved by the Commonwealth as quotas:
although it may be that ultralight aircraft
actual numbers that may he taken in a given
can be used to produce similar results for
year. Quotas are currently set and justified
less cost (Grigg et al. 1997). The expense
as proportions of the total population. This
and short range of helicopters means that
is because determining the impact that a
Queensland monitors populations on several
given quota will have on a population is a
fixed monitor blocks only (Lundie-Jenkins
matter of determining the per capita level of
et al. 1999), rather than extensively across the
mortality that the quota is equivalent to. This
entire state.
requires an estimate of the actual population
size, not simply an index, connected to the The remaining states count kangaroos using
population size by some unknown constant. strip transects surveyed from fixed-wing
Without an absolute population estimate, aircraft. New South Wales and South Australia
any harvest policy can only be reactive, rather use extensive annual surveys with two (New
than pro-active. It may be possible to measure South Wales) or four (South Australia) transect
the impact that a harvest with a given quota lines being flown per degree square through-
has had on a population, but not to predict out the main harvest area (Grigg et al. 1999;
what effect it will have on a population, except Gilroy 1999), covering a large part of the
to conjecture that it may be similar to kangaroo range within each state. Western
the effect of a previous quota of similar size Australia uses similar extensive surveys, but on
in the past. It will also be difficult to defend a less frequent basis. All three states adjust the
publicly a quota based purely on an index. actual numbers seen to produce an estimated
actual population, using correction factors
originally derived by Graeme Caughley for
HOW KANGAROOS ARE COUNTED
red kangaroos (Caughley et al. 1976). There
CURRENTLY
is consensus that the correction factors for red
Population size of animals can be estimated kangaroos return reasonable estimates, but
in a number of ways, including mark-recapture that grey kangaroo and wallaroo populations
methods, visual counts and a variety of indirect are grossly underestimated.
Australian Zoologist 31(1) 311
SIGHTABILITY OF KANGAROOS survey, and experiments to ensure that 100 m
FROM THE AIR is not closer to this optimum than 200 m are
warranted. Stuart Cairns (University of New
All kangaroos present on the ground will
England) and Josh Gilroy (NSW National
not be seen from the air, particularly in a
Parks and Wildlife Service) will undertake
fixed-wing survey. To translate an aerial count
a large-scale experimental programme to
into an estimated population size, it is
address this issue, among others.
necessary to estimate the sightability, which
is the probability of seeing an animal, given A related suggestion to increase the sight-
that it is present within the area being ability of kangaroos on the transect being
counted. As Pople (1999) pointed out, visibility surveyed, at the cost of surveying less area
is a function of species, behaviour, habitat for the same flight time, is to count from one
and observer. The likelihood is that it also side of the aircraft only. It is clear that "away
depends on interactions between these factors, from the sun" counting is much easier than
rather than simply being able to be estimated "towards the sun" counting. Using only one
as a multiple of individual probabilities for observer to count away from the sun would
each of the above factors. reduce personnel requirements considerably,
and possibly increase the reliability of the
In ground counts and helicopter surveys
counts, but it would also reduce the total area
using line transect methods, the distance to
surveyed.
the sighted individual or group is recorded.
Provided all animals actually on the transect Observer training is a crucial issue in both
line (i.e., zero distance) are seen, the use increasing the mean sightability and reducing
of appropriate software (Laake et al. 1993) its coefficient of variation. Beard (1999)
enables these counts to be corrected for produced data showing that the number of
sightability. Strip transect methods, however, kangaroos seen by a trainee observer, relative
require sightability to be estimated from to that seen by a trained observer, does not
information additional to that provided by the reach an asymptote until after at least 10
survey itself. flights. Similarly, the number of kangaroos
seen per transect by a trainee is much more
To improve the accuracy of fixed-wing aerial
variable than that counted by an experienced
counts, three things can be done with sight-
observer. Some observers, despite experience
ability. It can be improved, to increase the
with kangaroos, do not seem to approach
mean value of the raw count. It can be
the sighting levels of experienced observers
standardized, to reduce the variation in
at all. Clearly, there is a major problem
counts between replicate aerial transects.
in setting fixed "correction factors" unless
Finally, if it can be estimated, it can be
all observers have been calibrated against
used to correct the raw sighting index, and
a single "experienced observer" standard.
translate it into an actual estimate. Improving
Such a standard probably does not exist, as
mean sightability and reducing its variation
even experienced observers differ in the
(or coefficient of variation) are issues that
mean number of kangaroos that they will see
cannot be discussed separately. Some measures
on the same transect.
may improve both, whereas others may
trade one off against the other. Few ways
of improving mean sightability have no STRATIFICATION
consequences for its coefficient of variation.
One way to produce major improvements in
First, there is the possibility of reducing the precision of overall estimates of kangaroo
the strip width over which kangaroos are density is to use stratified sampling methods.
counted from the current 200 m to 100 m. The idea is extremely simple. If some
This is likely to increase the probability of mappable characteristic of the environment is
an animal on the strip being sighted, as thought to influence kangaroo density, then
the search pattern of the observers can be the total area for which the estimate is
concentrated, and there will no longer be required is divided into a small number of
attempts to count animals away from the separate subareas (or strata) on the basis of
track of the aircraft. On the other hand, this characteristic. The population density,
fewer animals will be on a strip of half and its standard error, is then estimated
the width, and it is likely that fewer will separately for each stratum, and the results am
be counted for the same expense and effort. combined to give an overall estimate. Strata
All other things being equal, the larger the should be selected so that variation between
total number of individual animals counted them in density is maximized, with the
in a survey, the lower will be the resulting consequence that variation within them is
standard error in the final estimated density. minimized. The obvious mappable character-
There is an optimum strip width for any visual istic to use in this particular situation would
312 Australian Zwlogist 31(1)
be vegetation type. Precision is maximized in others. Second, it is well known that
if sampling effort is allocated between kangaroos have different habitat preferences in
strata proportional to the overall kangaroo different environmental conditions (Caughley
population in each stratum, but substantial 198713; Hill et al. 1987; Pople 1989; Southwell
gains in precision are possible if sampling is 1989). If such habitat shifts occur at a large
random or systematic across the entire study scale, then habitat-dependent correction
area, or even if stratification is done after factors may allow for consistent estimation of
sampling is completed. population size in different environmental
conditions. However, correction factors would
necessarily be applied at the transect scale,
THE PROBLEM WITH GREY
of several kilometres. If habitat selection
KANGAROOS
occurs at the fine scale of patches within
The biggest problems with aerial strip transects, then habitat-specific correction
transects occur when counting grey kangaroos factors would not assist. Third, drought is
(both eastern and western). There are similar likely to change sightability within particular
diff~cultiesw ith wallaroos, but as the com- habitat types. Fourth, a fixed set of correction
mercial harvest of this species is relatively factors does not allow for differences between
small, the implications are not as serious. observers. Finally, the problem with a single
There was total consensus among workshop set of correction factors is that they are
participants that the correction factors likely to become fixed in usage, and not
originally derived by Graeme Caughley grossly reviewed until their faults are gross. The
undercount greys, and that they need to be experience with the original correction factors
at least doubled. Some participants suggested derived by Graeme Caughley epitomizes
quadrupling them was more appropriate. this last point.
Whatever the most appropriate correction
Is there an alternative to fixed correction
factor is, such ad hoc doubling and re-doubling
factors?
is better suited to a poker game than to the
management of a major wildlife harvest.
IS DOUBLE SAMPLING A SOLUTION?
Sightability of grey kangaroos is much lower
than that for reds primarily because they A method known as double sampling offers
prefer less open habitats than do reds, and an alternative to the use of fixed correction
are therefore more likely to be hidden from factors. The basic idea would be to survey
an aerial observer. Several participants in a large number of transects using fixed-
the workshop suggested that the solution wing aerial survey methods, and then to
to the correction factor problem for grey use a more intensive and accurate sampling
kangaroos is to undertake a substantial method (such as helicopter surveys) to obtain
research programme to establish improved a more accurate estimate on a subsample
correction factors. The idea would be to of those transects. The ratio of helicopter
survey grey kangaroos in a variety of habitats, counts to fixed-wing counts is estimated
using both aerial strip counts and helicopter from the subsample, and used to correct the
counts. Assuming that the helicopter counts counts on the full sample. Details of the
are close to being complete counts, an method are in standard textbooks on sampling
improved set of correction factors would be strategies (e.g., Thompson 1992) and it is
developed. This would represent a consider- widely used in wildlife surveys in North
able improvement over the current situation. America (Prenszlow and Lovvorn 1996).
I have several misgivings about this as a There are two major advantages in using
complete solution to the problem. First, "open" double sampling compared with fixed
or "closed" habitats may be quite different correction factors. First, because the correction
things in terms of vegetation structure or is recalculated in each survey, changes in
composition across the range of grey average sightability that occur through
kangaroos. Even if objective definitions in time are allowed for. This is the case whether
terms of percentage tree cover were developed, they occur because of changes in actual
the grain or patchiness in the habitat will visibility in a given habitat, habitat shift,
vary considerably, as will the tree species or inter-observer variability. Second, the
making up the canopy. Such factors may well estimated error in the final estimate incor-
influence sightability considerably. It would porates a component due to the error in
certainly be possible to test whether this the estimation of the correction factor. This
problem is substantial, but the costs would means that the precision of the final estimate
be considerable. Realistically, the likelihood is better described than it is by applying a
is that factors established in one part of fixed correction factor. The error in an
the country would be applied uncritically estimate based on a fixed correction factor is
Australian Zoologist 31(1) 313
underestimated, as well as the estimate itself surveys, while placing the whole procedure in
being biased in an unknown fashion. a more secure theoretical framework. I would
see the broad design as operating as follows:
There is an obvious disadvantage in double
sampling: cost. To add an expensive, but 1. Identify a small number of major zones
accurate estimation method on top of the within the state that are geographically
current intensity of fixed-wing aerial surveys contiguous, and across which kangaroo
will considerably increase the total cost of populations can be expected to vary, in
kangaroo surveys. The suggestion should not species composition or density. These are
necessarily be that the resources dedicated the basic strata.
to kangaroo surveys should be increased to 2. strata, identify (preferably at random)
allow for double sampling. It may be that the a small number (5-10) of blocks, or strips,
existing resources would be better allocated by within which kangaroo populations are to
reducing the intensity of fixed-wing aerial be monitored. Fixed-wing transects would
survey to allow for some double sampling. then be flown across these areas. These are
then cluster samples.
EXTENSIVE VERSUS INTENSIVE
3. A subsample of the fixed-wing transects
SURVEYS
within each block is then selected for
The current fixed-wing aerial surveys are helicopter or ground sampling. This is the
expensive. Despite this cost, all participants double sampling step.
in the workshop agreed that the estimates
For this approach to be feasible financially,
of grey kangaroo numbers produced may
it is clear that the total number of strip
be out by a factor of at least two. This is not
transects would need to be fewer than the
a satisfactory state of affairs. It is almost
current number. It should be possible to
certainly the case that the surveys may detect
model the alternative strategies, including
changes in grey kangaroo numbers with far
costs and reasonable assumptions about
greater accuracy than they estimate actual
kangaroo distributions, to see if the alternative
numbers, hut this relies on the assumption
approach is feasible.
that the uncorrected counts are an adequate
index of grey kangaroo population size. As A decision that must be taken is whether the
I discuss above, if only changes in numbers blocks and transects should be kept fixed,
are estimated, harvest management becomes or re-randomized each year. A general rule
reactive, and cannot be proactive. is that fixed sampling units are better able
to detect changes through time, but re-
It is certainly worth exploring the possibility
randomized sampling units will better estimate
of changing the basis of the surveys. The
population size at any given time.
current design in all states except Queensland
attempts a complete geographical coverage A problem with the above proposal is that
of the main kangaroo populations, but uses the optimal sample allocations to monitor
a method that is known to be inaccurate for grey and red kangaroos will be different.
grey kangaroos. The alternative is to use a Some sort of compromise would need to be
more accurate estimation process, carried out established, but there would still be major
over an appropriately selected sample from gains in precision compared with a random
the total geographic area. This is essentially allocation.
the approach Queensland currently takes,
using helicopter surveys over selected monitor WHERE TO FROM HERE?
blocks (Lundie-Jenkins et al. 1999). Whether
To immediately replace the current monitor-
the 10 Queensland monitor blocks can really
ing system with double sampling would be
be considered an appropriate sample from
premature. It is necessary to first demonstrate
the total area of kangaroo distribution is
that, for the same cost, double sampling
doubtful. They bear some resemblance to a
will lead to better estimates of kangaroo
stratified sample, as there is some replication
population size than provided by the current
within each of the major biogeographic zones
fixed-wing surveys. The previous discussion
of the state within which kangaroos are
merely suggests that it may. Much of the
common. These biogeographic zones could be
information necessary to make an informed
considered as strata. Clancy (1999) discussed
decision is already available. The records of
the advantages and disadvantages of the
aerial surveys across extensive areas of
Queensland approach relative to the approach
New South Wales and South Australia for a
of the other states.
number of years form a database that can
A double sampling approach offers the easily be subsampled to simulate stratified
prospect of combining some of the best surveys from only a sample of areas. Sub-
features of both helicopter and fixed-wing sampling would result in some loss of precision
314 Australian Zoologist 31(1)
in the estimate of the uncorrected mean REFERENCES
number of sightings per unit area, as fewer
Alexander, P., 1997. Kangaroo culling. harvesting
individual transects would be used in the and farming in South Australia - an ecological
estimate. Simulation could explore the extent approach. Aurt. Biol. 10: 23-29.
that the standard error in the mean
Beard, L. A,, 1999. Training observers. Awl. Zaol. 31:
uncorrected index increased, and more
287-91.
importantly, whether the same trends that
Beddington, J. R. and Basson, M., 1994. The limits to
are observed in the extensive counts can
exploitation on land and sea. Phil. Tranr. R. Soc.
be consistently picked up in subsamples. It
Lond. B 343: 87-92.
would not be difficult to cost subsamples
Buckland. S. T., Anderson, D. R., Burnham, K. P. and
in various locations across a state, including
Laake, J. L., 1993. Distance sampling: estimating
personnel, flight survey time and ferry time. abundance of biological populations. Chapman and
Hall: London.
The element of double sampling that cannot
easily be investigated using currently available Campbell, K. and Borner, M., 1995. Population trends
data is the strength of correlation between and distribution of Serengeti herbivores: implica-
tians for management. Pp. 117-45 in Snngcli XI:
fixed-wing and helicopter counts in different
dynamics, management and carcrvolion of an ccoryrtem
locations and at different times. If this ed by A. R. E. Sinclair and P. Arcese. University of
correlation is very strong and consistent, then Chicago Press: Chicago.
double sampling is unnecessary. All that is
Caughley, G. and Gunn, A,, 1993. Dynamics of large
needed is a single set of correction factors herbivores in deserts: kangaroos and caribou. Oikor
calculated once, and precision would be 67: 47-55.
maximized in the future by directing all
Caughley, G. C., 1987h. The effect of draught on
available resources to fixed-wing surveys. On kangaroo populations: a response. J. Wildl. Manage.
the other hand, if the correlation were entirely 51: 603-04.
absent, then fixed-wing surveys would be
Cairns, S. C., 1999. Accuracy and consistency in the
useless, and all resources should be dedicated aerial survey of kangaroos. Aust. Zool. 31: 275-79.
to helicopter surveys. Based on evidence from
Caughley, G., 1987a. Ecological relationships. Pp. 159-87
Queensland (Clancy et al. 1997; Pople et al.
in Kollgaroor: their Ecology and Management in the
1998a,h), the real answer lies somewhere Sheep Rangelands of Awtralia ed by G. Caughley,
in between: there is a good, hut imperfect N. Shepherd and J. Short. Cambridge University
correlation between fixed-wing and helicopter Press: Cambridge.
counts, and both the strength of the relation- Caughley, G., Sinclair, R,a nd Scott-Kemmis, D., 1976.
ship and its regression coefficient vary some- Experiments in aerial survey. J. Wildl. Manage. 40:
what through space and time. If this is 290-300.
the case, then the best sampling strategy is Clancy, T. F., 1999. Choice of survey platforms and
double sampling, with the optimal allocation technique for braad-scale monitoring of kangaroo
of effort between fixed-wing and helicopter populations. Aurt. Zool. 31: 267-74.
surveys depending both on the strength of Clancy, T. F., Pople, A. R. and Gibson, L. A,, 1997.
the correlation between the two and their Comparison of helicopter line transects with walked
relative costs. line transects for estimating densities of kangaroos.
Wildl. Res. 24: 397-409.
Fortunately, the experiments necessary to
Edwards, G. P., Croft, D. B. and Dawson, T. J., 1996.
determine the optimal allocation of effort
Competition between red kangamos (Marropur rufus)
for double sampling are very similar to and sheep (Ovir arirs) in the arid rangelands of
those needed to produce improved correction Australia. Aurt. j. Ecol. 21: 165-72.
factors. Essentially, what is needed for either
Freudenberger, D., Noble, J. and Hodgkinson, K., 1997.
purpose is a series of counts on the same Management for production and conservation goals
transects, repeated both using fixed-wing and in rangelands. Pp. 93-106 in Landscape Ecology,
helicopter surveys. For either purpose, these Fundion and Management: Prinincipler from Australia's
surveys would need to he carried out in a Rangelondr ed by J. Ludwig, D. Tongway, D.
Freudenberger, J. Noble and K. Hodgkinson. CSlRO
variety of habitats and geographical locations,
Publishing: Collingwood.
and for a variety of kangaroo densities. My
proposal, therefore, is that any experiments Gilroy, J., 1999. Kangaroo monitoring in relation to
kangaroo management in New South Wales. Aurt.
designed to produce improved correction
Zool. 31: 30608.
factors be so designed that they also act as a .
pilot double sampling survey. Gri"e-e.. G. C.. Poole.. A. R. and Beard. L. A,. 1997.
Apl~ltrariono l an ultraltgh~a nrcraft ru acrtdl ,urrev
of kangaruo, un grazing prol,rrric:i l4ilrll Rrr 24:
ACKNOWLEDGEMENTS 359-72.
Grigg, G. C., Beard, L. A,, Alexander, P., Pople, A. R.
I would like to thank Gordon Grigg and
and Cairns, S. C., 1999. Aerial survey of kangaroos
Tony Pople for comments on an earlier draft in South Australia 1978-1998: a brief report
of this manuscript. focusing on methodology. Aut. Zoal. 31: 292-300.
June 1999 Australian Zoologisl 31(1) 315
Hilborn, R. and Walterr, C. J., 1992. Quantitative Pople, A. R., Cairns, S. C., Clancy, T F., Grigg. G. C.,
fisheries stock assessment: choice dynamics and Beard, L. A. and Southwell, C. J., 1998~.A n
uncertainty. Chapman and Hall: New York. assessment of the accuracy of kangaroo surveys using
fixed-wing aircraft. Wildl. Ra. 25: 315-26.
Hill, G. J. E., Barnes, A. and Wilson, G. R., 1987.
The effect of drought on kangaroo populations: a Pople, A. R., Cairns, S. C., Clancy, T. F., Grigg,
comment. J. Wildl. Manage. 51: 602-03. G. C., Beard, L. A. and Southwell, C. J., 1998b.
Comparison of suzveys of kangaroos in Queensland
Hutchings, J. A. and Myers, R. A,, 1994. What Can
using helicopters and fixed-wing aircraft. Rnngcl.
Be Learned from the Collapse of a Renewable
J. 20: 92-103.
Resource? Atlantic Cod, Gndus marhuo, of
Newfoundland and Labrador Can. J. Fish. Aqwt. SG~. Prensdow, D. M. and Loworn, J. R., 1996. Evaluation
51: 2126-346. of visibility correction factors for waterfowl surveys
in Wyoming. J. Wildl. Manngc. 60: 286-97.
Kurlansky, M., 1997. Cod. A biography of the fish that
changed the world. Johnathon Cape: London. Seber, G. A. F., 1982. The estimation of animal
abundance and related parameters, 2nd ed. Griffin:
Laake, J. L., Buckland, S. T., Anderson, D. R, and
London.
Burnham, K. P., 1993. DISTANCE User's Guide
V2.0. Colorado Cooperative Fish and Wildlife Southwell, C.. 1989. Techniques for monitoring the
Research Unit, Colorado State University: Fort abundance of kangaroo and wallaby populations.
Collins, CO. (http:iiwww.mwpa.st-and.ac.uWdistance) Pp. 659-93 in Kangaroos, Wallnbics and Ral-hangarnos,
Vol. 2 ed by G. Grigg, P. Jarman and I. Hume.
Lundie-Jenkins, G., Hoolihan, D. W. and Maag, G. W.,
Surrey Beatty & Sons: Chipping Norton, Sydney.
1999. An overview of the Queensland macropod
monitoring programme. Ausl. Zool. 31: 801-05. Southwell, C., 1994. Evaluation of walked line transect
counts for estimating macropod density. J. Wildl.
Pollock, K. H., 1995. The challenges of measuring
Mamgc. 58: 348-56.
change in wildlife populations: a biometrician's
perspective. Pp. 117-21 in Consctvalion Thmugh tht Thompson, S. K., 1992. Sampling. Wiley-Interscience:
Sustainable Use qf Wildlife ed by C. C. Grigg, New York.
P. T Hale and D. Lunney. Centre for Conservation
Biology, University of Queensland: Brisbane. Wilson, A. D., Harrington, G. N. and Beale. I. F., 1984.
Grazing management. Pp. 129-39 in Manngrminl
Pople, A., 1989. Habitat associations of Australian o f Aurtmlia's Rangelands ed by G. N. Harrington,
Mampodoidea. Pp. 755-66 in Kangomor, Wollnbicr A. D. Wilson and M. D. Young. CSlRO Publications:
and Rat-hangaroor, Vol. 2 ed by G. Grigg, P. Jarman East Melbourne.
and I. Hume. Surrey Beatty & Sons: Chipping
Norton. Sydney.
Pople, A. R., 1999. Repeatability of aerial surveys. Aurt.
Zool. 31: 280-86.
316 Australian Zoologist 31(1)