Although various methods for monitoring parrot populations are available, the utility of these methods is not uniform among species because of species differences in behaviour and ecology. Among the methods that have been used with various parrots are roost counts, nest enumerations, river transects, mark-resighting studies, and fixed lookout counts. All have weaknesses of one sort or another, and only direct field experience is likely to reveal the most useful and practical techniques for a particular species. All methods have key assumptions that need to be met for applications to be reliable (see Casagrande and Beissinger 1997).

Roost counts have been used with good success in achieving population counts for some species [e.g., the Bahama parrot (Amazona leucocephala bahamensis) (Gnam and Burchsted 1991), and the Puerto Rican parrot (Amazona vittata) (Snyder et al. 1987)], but proper use of such counts necessitates finding all significant roosts for the population in question and determining which time of year the birds tend to clump most consistently in roosts. Preferably, all roosts should be monitored simultaneously, although this requirement can be relaxed for species whose roost-use tends to be stable over long periods.

Unfortunately, some species do not clump together in obvious roosts, while others approach and leave roosts in the subcanopy, making them difficult to enumerate accurately. Others change roost locations so frequently that it can be difficult to keep current on roost locations. Thus, while roost counts can be a very good method with species that do not present the above problems, it is not a method that can be used effectively with others.

Nest enumeration is currently being employed in status work on the maroon-fronted parrot (Rhynchopsitta terrisi), a species that nests colonially in cliffs (Enkerlin in litt. 1997). As with roost counts, success in using this method as a population monitoring technique depends on locating all significant colonies of the species and determining which time of year is best for counting. The maroonfronted parrot also roosts communally, but frequent changes in roost locations make monitoring of roosts difficult. Moreover, the habitat of this species is sufficiently difficult to access that getting close enough to count some roosts poses severe logistic problems. Nevertheless, results of roost counts to date show clearly that a large fraction (perhaps on the order of 80%) of the population does not show up in nest enumerations. While long-term monitoring of the size of the nesting population appears to be a relatively practical goal and may prove to be an important component of efforts to follow the overall health of the population it appears unlikely to track total population numbers closely on a year to year basis, because of large fluctuations in food supplies with this species. Nest enumerations may well give more useful population trend information on the basis of longer time spans. The optimal monitoring strategy with this species appears to be efforts to utilise both roost counts and nest enumerations, despite the practical difficulties in roost counts.

For species that nest in dispersed fashion, nest enumeration often has little potential for overall monitoring of populations because it is often extraordinarily labourintensive to locate nests for such species. Nevertheless, nest enumeration has proved useful in tracking the population health of the dispersed-nesting goldenshouldered parrot (Psephotus chrysopterygius), which utilises termite mounds for breeding (which can be located with some efficiency). The density and distribution of nests of this species are monitored annually over 250 square kilometres of terrain as a measure of effectiveness of conservation actions (Garnet and Crowley 1995).

Counts of birds assembling at clay licks or waterholes can also be useful, particularly when they can be converted into density figures or total population counts. Such conversions, however, require knowledge of the areas serviced by such features and the frequency of visitation by individuals. Waterhole counts tend to be most useful for species in relatively arid habitats, especially during the dry season when the number of water sources is minimal, forcing the birds to concentrate on relatively few sites. However, since individuals may visit water sources more than once a day or move between water sources, to extrapolate such counts to population counts some individual birds must be marked (for example with radiotags) to ascertain frequency of visits. As with roost counts and nest enumerations, efforts have to be made to locate all water holes in use and to monitor them simultaneously. Counts at clay licks, coupled with individual identifications of birds achieved through photography, have been used to generate population density figures for certain macaws (Munn 1992).

Moving transects (line transects), such as counts from boats along rivers, can give useful indices of abundance of some species. They are often very difficult to convert into accurate population estimates, however, as the areas serviced and the detection efficiencies can be difficult to specify. Moreover, behavioural characteristics of some species may strongly bias their detectability by such methods. Nevertheless, such counts can be used to compare species abundances in different areas of similar habitat, to gain trend information on specific populations, and to document seasonal changes in habitat use (Munn 1992, Renton 1994, Robinet et al. 1996).

Under some circumstances, counts from stationary locations can give useful monitoring data, particularly if stations are established along important flight lines. The problems here are that parrots are often highly patchy in distribution, and it is often difficult to establish how representative the observation points are and what areas are effectively covered in the counts. Such counts are often most useful as indexes of abundance if carried out over long periods of time, but they are difficult to convert into absolute abundances accurately. Flight lines of species can change, seasonally or more permanently, relative to changes in distributions of food supplies, so counts in fixed locations can give spurious trend information if not coupled with other indices of abundance. Point surveys can often be expected to be more biased than line transect counts, but there are circumstances where they are a preferable technique (see Casagrande and Beissinger 1997).

Mark-resighting techniques are potentially useful with some species (Casagrande and Beissinger 1997), but such methods are highly labour-intensive compared to other methods and often are impractical because of difficulties in capturing birds for marking. In addition, there can also be an increased risk of predation for marked animals in some species (see Saunders 1988).

The aforementioned methods are not the only methods that might be applied to parrots. For example, variable circular plot methodology has been used in many studies in Asia and Africa, and offers a number of advantages in some contexts. However this method, like some others, often yields such wide variability for population sizes that it is sometimes of limited value in monitoring population trends.

Although it is highly desirable and valuable to develop techniques that may give accurate total population counts for any endangered species, this may simply not be feasible for some parrots. In such cases, it may alternatively be possible to devise ways of indexing abundance that can give reliable trend information over the long term. This is usually the most important information for conservation purposes. Additionally, relative differences in density between areas may be important, and even order of magnitude estimates for poorly known species may be better than no estimates at all. Mail surveys have been used successfully to monitor declines in populations of species that were once common and widespread in Western Australia. This method is cheap, quick, and well suited to species that are readily recognised and familiar to amateurs (see Mawson and Long 1996). In addition, information on trade volume can sometimes be used to infer population trends, provided certain assumptions about harvest intensity and reporting uniformity are met.

The literature on bird censusing is large, and the reader should consult general reviews on bird censusing methods, such as Ralph and Scott (1981), Davis (1982), Verner (1985), Taylor et al. (1985), Seber (1986), and Bibby et al. (1992) for a critical discussion of other methods that may have value with some species. Accurate censusing of wild bird populations remains one of the more difficult tasks confronting researchers and conservationists. There is no one universal method for estimating bird abundances and densities, and appropriate methods vary according to species, time, and location. The desire to find a single technique that might work well for all parrot species will surely remain unfulfilled.

Reproduced with permission of the International Union for Conservation of Nature and Natural Resources and the World Parrot Trust.