1) Effective estimation of wildlife population abundance is an important component of population monitoring, and ultimately essential for the development of conservation actions. Diurnal line transect surveys are one of the most applied methods for abundance estimations. Local ecological knowledge (LEK) is empirically acquired through the observation of ecological processes by local people. LEK-based methods have only been recognized as valid scientific methods for surveying fauna abundance in the last three decades. However, the agreement between both methods has not been extensively analysed.

2) We compared concomitant abundance data for 91 wild species (mammals, birds and tortoises) from diurnal line transects (9,221 km of trails) and a LEK-based method (291 structured interviews) at 18 sites in Central and Western Amazonia. We used biological and socioecological factors to assess the agreements and divergences between abundance indices obtained from both methods.

3) We found a significant agreement of population abundance indices for diurnal and game species. This relationship was also positive regardless of species sociality (solitary or social), body size and locomotion mode (terrestrial and arboreal); and of sampled forest type (upland and flooded forests). Conversely, we did not find significant abundance covariances for nocturnal and non-game species. Despite the general agreement between methods, line transects were not effective at surveying many species occurring in the area, with 40.2% and 39.8% of all species being rarely and never detected in at least one of the survey sites. On the other hand, these species were widely reported by local informants to occur at intermediate to high abundances.

4) Although LEK-based methods have been long neglected by ecologists, our comparative study demonstrated their effectiveness for estimating vertebrate abundance of a wide diversity of taxa and forest environments. This can be used simultaneously with line transects surveys to calibrate abundance estimates and record species that are rarely sighted during surveys on foot, but that are often observed by local people during their daily extractive activities. Thus, the combination of local and scientific knowledge is a potential tool to improve our knowledge of tropical forest species and foster the development of effective strategies to meet biodiversity conservation goals. --

Data collection

Between 2011 and 2017, we surveyed the abundance of a set of species through line transects, and through interviews with local people from 17 different indigenous and non-indigenous riverine villages. All villages were settled in or near the sites where transects were surveyed, and local people use these sites for different purposes (hunting, harvesting of forest products, etc.); therefore, each village offered information about at least one correspondent site. Sixteen villages informed LEK about a single correspondent site, and 1 village informed LEK for 2 correspondent sites. Interviews and line transects within each study area were conducted in a mean lapse time of 8.6 months, ranging from 0 to 24 months.

Line transects

We estimated the population abundance of each species from direct diurnal sightings conducted on 31 line transects, with a total surveyed distance of 9,221 km (ranging from 42 km to 2,687 km surveyed per site; mean= 512 Km, SD= 707) (Supplementary material, Table 1). Each transect was randomly positioned in all study areas and transects were opened prior to the surveys. Two observers (at least one of them was a local monitor) walked the trails between 6:00 h and 15:00 h at an average speed of 1.5 km/h. When a group of animals was encountered, the number of individuals and species was recorded. From the collected data, we calculated the sighting rates (individuals/km, calculated as the total number of individuals observed divided by the total effort in km travelled on any given transect during all seasons), which were used as our abundance index since higher abundance increases species detections (e.g. Paim et al., 2019). The value of zero was assigned to species whose occurrence is confirmed in the area but that were not detected on any transect sampled near that village.

Local Ecological Knowledge

We interviewed 291 local people from the sampled villages (average interviewees per village =16.16, SD=6.62) using a snowball sampling technique (Bailey, 1994) through the indication by each interviewee of another local expert on fauna. The interviewees' ages ranged between 16 to 75 years old (average = 37.75; SD=13.29). We conducted interviews individually to collect the interviewee perception of the abundance through LEK-based methods of each species that occur in the area in which the interviewee lives. Interviews did not require local translators as both the interviewers and the interviewee, including those from indigenous territories, were fluent in Portuguese or Spanish. All researchers conducting the interviews were already working in each site and had built relationships of trust in the communities. 

Data were collected through structured interviews with an illustrated checklist, which provided colour plates of species expected in each study area. During each interview we asked the local vernacular name for each species illustration, often corresponding to the species common nomenclature in Portuguese or Spanish. For each species, we asked the interviewee to estimate their abundance on a Likert scale; 0 (when the species was “absent”), 1 (low abundance), 2 (medium abundance) and 3 (high abundance) (Van Holt et al., 2010; Van Holt et al., 2016). The value assigned by each interviewee for each species was considered as our abundance index for the LEK-based method. The value of zero was assigned only to species whose occurrence is expected for the area by previous studies but was mentioned as absent by a specific interviewee. We validated the consistency of the responses through a cultural consensus analysis (Borgatti and Halgin 2011), which consists in a multivariate test based on the degree of similarity between respondents’ answers. In this case, respondents showed a personal consensus higher than 0.6 (indicating a high consensus) regarding the abundance indices of each species population.

Article title associated with this data: Congruence of local ecological knowledge (LEK)-based methods and line-transect surveys in estimating wildlife abundance in Amazonian forests. In Methods in Ecology and Evolution journal.

The bioecological characteristics of the species can be found in the supplementary material of the article.
The abundance information database presented here is complete.