Excerpt
Table of Contents
1. Introduction
2. The Grey Parrot - Psittacus erithacus
2.1 Ecology
2.2 Conservation Status
3. Study Area & Data sampling
4. Data Analysis
4.1 Evaluating Compliance to Model Assumptions
4.2 Coping with evasive movements: Model Selection & Grouping
4.3 Coping with evasive movements: Activity segregation
4.4 Left Truncation
4.5 Amelioration Measures in the Field
4.5.1 Transect Sampling
4.5.2 Point Counts
5. Results
6. Conclusion
References
Abstract
We examined field data from surveys exercised in three consecutive years (1999-2000) for four different sample sites, situated in the Korup Project Area in western Cameroon, in order to estimate population densities of the Congo African Grey Parrot Psittacus erithacus erithacus. Data was acquired by line transect sampling, and was analysed using the program DISTANCE 6.2 (Laake et al . 2009). Exploratory analyses were performed as suggested by Buckland et al. (2001), looking for evidence of evasive movement before detection, ‘rounding’ and ‘heaping’ of data. Outlier data were truncated when it improved the fit of the model and model selection was based on Akaike’s information criterion (AIC) values, as well as Chi-square goodness of fit tests (χ2). Evasive Movements and heaping were detected in most of the surveys, and different approaches to ex post compensation measures were compared in their suitability for this study. The data suggest, that acoustic detection is very likely to cause heaping at distances further away from the transect line, as observers tend to round estimates and increment in fives or tens. This proofed true when detection histograms where compared including or respectively excluded acoustic data. While heaping events could be considerably alleviated by omitting acoustic data, a complete obliteration was not given, and further accounting to reduce bias was needed. In conclusion, results had the least prospect to be biased, when all visual detections of birds in any activity (including aerial birds) at the moment of sighting were accounted for when estimating parameters. Evasive movements and heaping could be leveled to a great extent by lavish grouping for intervals closest to the transect line, in which way detections where evenly redistributed to restore random distributions in approximation, and was, in this case study, found to be the best possible approach to account for responsive movements in line transect data sets.
1. Introduction
The Grey Parrot is one of the best-known flagship species of equatorial Africa. Its fame for being an entertaining and clever pet resulted in the species being the third most commonly traded wild bird species, worldwide (Chupezi et al. 2006). Besides habitat destruction, capturing for the domestic and international pet trade is therefore the single most imminent threat to the species. Formerly widespread it is now listed as near threatened on the IUCN redlist, with predicted further declining population trends. Cameroon is home to a large proportion of the population of one of two subspecies, namely the Congo African Grey Parrot Psittacus erithacus erithacus. Internationally, concerns of declining populations lead to a proposal of more severe conservation measures, such as trade restriction enforcements (BirdLife International 2009, CITES 2004). Furthermore, monitoring of wild population trends was declared to be a missing stepping stone towards effective conservation measures, since too little exact data on the species is available until today (CITES 2004). With this case study for the Korup Project Area we wish to contribute to the latter proposal.
While population estimates derived from line transects are known to produce results relatively robust to a variety of error sources, some commonly encountered issues when it comes to bird sampling go beyond the scope of tolerance, e.g. when individuals purposefully peer away from the transect line. This responsive movement to the observer prior to detection is problematic, because animals are assumed to be located randomly in relation to the line transect (Thomas et al. 2009), and thus clearly violates one of the eight assumptions hold by Distance (see 4.1; Evaluating Compliance to Model Assumptions). In literature different approaches can be found for dealing with bird distance sampling related problems (Thompson et al. 2002), such as incomplete detections near the transect for instance due to concealment behavior of certain species (Turnock et al. 1991, Borchers et al. 1998) and for responsive movements to the approaching observer (Mardsen et al. 1999, Bårdsen et al. 2006 , Palka and Hammond 2001). However, most of the proposed solution statements are made in relation to aerial or marine surveys, need special data collection in the field, and are difficult to apply to bird counts. While ex ante measures that can minimize the magnitude of adverse effects are rather plentiful, addressing the problem ex post is restricted to limited options, of which the most common will be discussed and assessed for this case study.
2. The Grey Parrot - Psittacus erithacus
The Grey Parrot, Psittacus erithacus also referred to as Jacko or Jacquot (CITES 2004) is one of the most renowned species of the African representatives of the Psittacidae Family, and in Africa the only member of the genus Psittacus (Snyder et al. 2000). Its fame is largely due to its long history as an entertaining pet around the world. Scientific attention was draught to the species by ‘Alex’, an African Grey that was object to comparative studies of cognitive communicative abilities of Grey parrots (Pepperberg 2006) and the fundamentals of language and communication, at Harvard University.
P. erithacus is a medium sized parrot of approximately 33cm height, which nevertheless makes it the largest African Parrot. The lower mantle and back and belly are of a mottled grey, while the upper- and undertail- coverts are scarlet red. The lores and area around the eyes are naked and of white color. Assumptions about the total number of subspecies vary, while officially is restricted to the Congo African Grey Parrot Psittacus erithacus erithacus and the Timneh African Grey Parrot Psittacus erithacus timneh. The latter is generally darker and slightly smaller than the nominate, has a reddish upper mandible, and a distinctive call. As the naming suggests the subspecies are of allopatric distribution, with the Bandama River in Côte d'Ivoire. P. e. erithacus can be encountered anywhere east of the river up to Kenya including the Gulf of Guinea Islands, while P. e. timneh is restricted to the west of the river including southern Guinea, Liberia, Sierra Leone and extending to Guinea-Bissau. It has been estimated that the total range comprises an area of about 3 million sq km (CITES 2004) and total population estimates aggregate to 680,000 – 13 million individuals (BirdLife International 2009).
Abbildung in dieser Leseprobe nicht enthalten
Fig.1: Geographic range of the two acknowledges subspecies, Psittacus erithacus timeh to the west, and Psittacus erithacus erithacus to the east of Bandama River in Côte d'Ivoire. The Yellow rhomb in Cameroon indicates the study area (IUCN 2009, modified).
2.1 Ecology
The species proofs to have a rather wide range of tolerance, concerning their habitat. Primarily dwelling in dense primary and secondary forest, they can also be encountered at forest edges, clearings, gallery forest, mangroves, wooded savannah, cultivated areas, and even gardens. However, it is likely that at least some of the populations found in these habitats depict sink populations (BirdLife International 2009, Fry et al. 1988). Depending on the habitat, population densities vary significantly from 0.15 birds to two breeding pairs per km² (CITES 2004). The species roosts in large flocks preferentially in primary forests during the rainy season and migrates into villages and plantations in the dry season (Fry et al. 1988). In the choice of breeding sites the hole-nesting P. erithacus shows a rather wide flexibility, as nests have been sighted in cocoa plantations, rocky crevices and even in hollows in shrub coverage (Juste 1996). Typical for a species with a long life-expectancy, population growth rates are relatively low. Approximately 15-30% of the population breeds per year, with an average productivity of 0.4 nestlings per nest (CITES 2004). It is a monogamous species that establishes colonies in some areas, such as the Niger delta or on Principe Island, while elsewhere breeding more solitary (Fry et al. 1988).
2.2 Conservation Status
Population declines have been noted in large parts of the distribution ranges of the species (CITES 2004). This resulted in a reclassification of the conservation status from least concern to near threatened by the IUCN, in 2007. Major threats to the species can be summarized to the following (BirdLife International 2009, Chupezi et al. 2006):
- Agriculture and aquaculture expansions (àhabitat loss and fragmentation etc.)
- Small-holder farming practices (slash and burn, harvest of non-timber crops etc.)
- Increasing commercial timber exploitation
- Small Scale selective logging & wood harvesting and related destruction of feeding trees and nesting sites as a result of selective felling of the larger trees.
- Corruption and illegality in the pet trade
- Weak legislation and poor enforcement of laws where they do exist
- Unmanaged parrot trapping and unsustainable subsistence hunting
- Insecurity and weaker legislation in neighboring countries
Of these often interwoven causes for population declines habitat destruction throughout West and East Africa is thought to make up for a significant part. The only factor outranging this in its impact is legal and especially illegal pet trade. Due to its longevity, cleverness and ability to mimic human speech Psittacus erithacus is one of the most popular avian pets in Europe, the United States of America the Middle East and China. Unfortunately, a conceivable proportion of the demand is met with illegally imported birds, even though the species is easily bred in captivity and wild-caughts birds are likely to develop behavioral and post traumatic stress disorders. Recent analysis suggests that up to 21% of the global population may be harvested annually (BirdLife International 2009). While there is some demand on inner-African markets, the majority of wild caught birds are exported, and trade has flourished despite international protection (CITES Annex II) (Juste et a. 1995). While so far the main market was in western countries, demand for wild birds is also increasing in China, and is likely to be further spurred by increasing business interactions with central Africa (particularly for mining, oil and logging) (BirdLife International 2009). In-between 1994 and 2003 CITES reported a total of 360.000 officially traded specimen. The estimated number of unreported cases however, exceeds this figure by the manifolds. Implementation of trade controls is often subject to institutional constraints and corruption, and fails to ensure that exports are in accordance with national legislation and/or CITES (2004).
Cameroon’s export quota has been estimated to 12 thousand individuals per year, with a value of approximately US$ 12 million in European markets, while a very small proportion of the revenues stay within the source country (Chupezi et al. 2006). While Lobeke National Park is known to be the highest parrot-trapping zone in Cameroon with 80% of parrots from Cameroon caught there, the other 20% are trapped from other tropical rainforest ecosystems such as those found in the Korup National Park and Lomie regions (Chupezi et al. 2006, ref; Tieguhong and Ndoye, 2004). Literature reviews however indicate, that monitoring of these processes are negligible for most areas, and that changes in inner country capturing activity shifts would be likely to go undetected.
In 2004, a trade review was issued by CITES strongly recommending a two-year export ban from January 2007 on from four West African countries (Cote d'Ivoire, Liberia, Sierra Leone and Guinea), as well as for Cameroon. It was also suggested that quotas in Congo and DRC should be halved to 4,000 and 5,000 individuals exported per year. The proposed conservation measures stress trade restriction enforcement and monitoring of wild population trends (BirdLife International 2009, CITES 2004). With this case study of population estimates in the Korup Project area we wish to contribute to the latter proposal.
Abbildung in dieser Leseprobe nicht enthalten
Fig.2: Survey sites within the Korup project area in south-west Cameroon. (Waltert et al. 2002).
3. Study Area & Data sampling
Abbildung in dieser Leseprobe nicht enthalten
The Survey was conducted in the Korup Project Area (Fig.2), comprising the Korup National Park and surrounding areas, situated along the western border of Cameroon. This area is known to harbor one of Africa's rainforests, richest in biodiversity (Chuyong et al. 2004). Data was collected in the framework of a community-based survey in three consecutive years, from 1999 to 2001. The paramount intent was to provide data on abundance, population trends and impact of logging on several target species (Waltert et al. 2002), of which the African Grey parrot will here be discussed. The four sample sites differ in their suitability for this species; Umbrella trees of major importance for this frugivore species can be found to a larger extent in the logged areas III and IV. Of the two other areas which remained unlogged, area III has a lesser umbrella tree density than sample site I. For further information on the Project area, as well as detailed description of survey design and data collection please see Waltert et al. (2002).
4. Data Analysis
4.1 Evaluating Compliance to Model Assumptions
Distance modeling for data collected on transect line surveys incorporates a number of underlying assumptions that have to be granted for unbiased results (Bibby et al. 1993). Ranked from the most to least critical these assumptions are;
1) All Individuals close to the transect line are detected with certainty.
2) Birds do not move prior to detection. Density estimations assume that birds are randomly distributed in relation to the transect line.
3) Distances are measured accurately. Range finders are off great help to get exact data, however for birds, fast movements not only prior to but also at the very detection moment, are common and aggravate correct measures.
4) Individual birds are only counted once. This is especially problematic when observers face large flocks during lively activity, and can merely be addressed by experience and training.
5) Birds are detected independently. Detection probabilities can vary with densities, e.g. when higher densities lead to higher frequency of vocal interactions.
6) Bias from seasons, weather and observers is understood. Transect lines are specifically subject to observer skills. All ecologically highly relevant influences such as day time and weather conditions have to be accounted for in the survey design.
Histogram plots depicting the number of observations in relation to their perpendicular distances, as well as Chi-square goodness of fit tests (χ2) with special emphasis on the fit near zero distances, were used to descry violations of these assumptions. The datasets of the Grey Parrot in the different study areas and regions all suggest to a minor or major degree, that assumption two was not met. Most birds were rather detected at some distance than directly at the transect line, indicating an avoidance movement away from approaching humans and thus the 100% detection area. Responsive movements will result in biased estimators of abundance, for instance will density most likely be underestimated (E(Ḋ)<D) (Buckland et al. 1993, Turnock et al. 1991). In a study focusing on parrots and hornbills in Indonesia, Mardsen (1999) found that in all but one parrot’s detection curves no birds were recorded at a 10 meter distance from the observer, while highest densities were detected at the subsequent interval stretching up to 30 meters. While his casual approach, that birds could remain cryptic in the above canopy can most likely be dismissed for an apparent species like P. erithacus, his second conclusion suggesting that the birds flush in response to the recorder's presence seems to depict an axiomatic problem, which also applies for this case study.
Abbildung in dieser Leseprobe nicht enthalten
Fig.2: Examples for the probability density functions when responsive movements occur (Turnock et al. 1991), in comparison to case study of Cameroon study (Bajo 1999-2000). (X-Axis: perpendicular distance, Y-axis: Detection Probability).
4.2 Coping with evasive movements: Model Selection & Grouping
The selection of models was carried out, after an appropriate right truncation had been defined, since the AIC a reconciling approach between model fit and parameter complexity, cannot be used to choose between histograms of different truncation distances (Jathanna et al. 2003, Buckland et al. 1993). Data was right-truncated at perpendicular distances of 80-200 meters. The appropriate model for each analysis was therefore chosen on the ground of the lowest AIC value and data was grouped manually in unequal intervals to reach good χ2 fitting values where necessary. A disadvantage of the goodness-of-fit tests test data is that data has to be arranged into intervals before the test can be performed, and the selection of cutpoints can have a strong influence on the outcome of the test (Laake et al. 2009). However, an examination of the output histograms showed strong evidence for evasive movement and heaping, so that ex post measures had to be taken to alleviate those. It is widely acknowledged that sensitive grouping can improve the fit, while resulting in little change of density estimates Ḋ (Bårdsen et al. 2006, Jathanna et al. 2003). Furthermore, if named effects are present, grouping of data into appropriate distance intervals can even improve precision and provide more realistic Ḋ values (Buckland et al. 1993, Bårdsen et al. 2006, Jathanna et al. 2003). In order to apply this outsmoothing of evasive movements the two following assumptions where made; First it is very unlikely that more individuals close to the transect remained cryptic than in further distances, and that thus the commonly observed low detection numbers at close distances are a direct result of fleeing individuals. Second that heaping events in larger distances where primarily caused by the alarmed birds still in flight. This could be fostered by a comparison of histograms which accounted for, respectively omitting the observations of aerial birds, since heaping and evasive movement indicators were significantly reduced for the latter (Fig.3). Thus, the evasive gap near zero should in approximation equal the heaping event in total numbers, since heaping events where actually made up by those individuals, which fled from the transect line. In order to provide for the made assumption, that heaping had little different causes only visual detections where accounted for (Fig.3). Acoustic estimations are very likely to be made in certain distance increments which will result in artificial overestimation of numbers in rounded distances.
[...]