Analysis

Traffic

Calculating average traffic at each lodge, for the five year period concerned, was no easy task. Firstly, the monthly totals provided for CAPH, EI, SACHA and TRC were compared to the official MITINCI and INRENA registers. In percentage terms there was not much difference between these sets of data (+/- 7%), although the data provided by the lodges was marginally greater on the whole in terms of numbers of tourists. For this reason lodge data formed the basis for calculating traffic at these 4 lodges. Secondly, the number of guides that accompanied the tourists was calculated by dividing the monthly tourist visitor totals by the average group size during excursions, a figure obtained after questioning guiding staff and analysing trail-use registers. This figure was subsequently added to the number of tourists to give yearly totals of potential traffic, i.e. the maximum number of tourists that were available to ventured along the trails.

The information gained from the trail registers showed that, for any given period, more tourists were present at a lodge than actually used the tourist transects under investigation. This is due to the fact that the probability of a tourist venturing out along a T transect is a variable of the tour program itinerary, guide preference, tourist preference, and weather conditions. The potential traffic levels per year were therefore adjusted accordingly. Finally, the yearly traffic totals for the period 1994-98, inclusive, were averaged. This figure was that used to analyse impact, and was measured in terms of the average number of people per year who ventured out along the T transects.

In the case of ECO which did not provide detailed data on monthly tourist receipts, the data collected from the trail registers was used to extrapolate the data provided by MITINCI for the years 1996-98 in an upwards direction. For the period 1994-95, when MITINCI was provided with inadequate data for this lodge, we had to estimate tourist arrivals using the trends  for this period observed at the other lodges.

Habitat and Fruit Resource

T-tests were used to compare the abundance between T and C at all lodges for each of the 5 most common plant genera under investigation i.e. those that made up between 70-90% of the sample in most cases (Iriartea, Scheelea, Astrocaryum, Pseudolmedia, and Ficus). To determine if there were any differences in abundance of each of these genera between lodges as a whole, all the available data for each lodge was first combined and subsequently analysed using t-tests. The program SYSTAT was also used to undertake a series of multivariate hierarchical cluster analyses, based on simple Euclidean distances, using data from all 13 genera simultaneously, in order to gauge the level of similarity between lodges.

Hunting Pressure

To determine if the hunting index of a lodge had any effect on mammal abundance a simple Spearman rank correlation coefficient was calculated using combined abundance data for several species known to be commonly sought by hunters in Madre de Dios (collared peccary, brazilian tapir, red howler, brown capuchin, red brocket deer. The relative abundance data collected on the Spix’s Guan (P. jacquacu) was also regressed against hunting pressure and proved a useful indicator.

Transect Sampling Effort

Transect sampling effort was recorded as the distance (km) censused along T and C transects at each lodge within the permitted time period and weather conditions. The time it took to complete all censuses was also calculated and converted into an average velocity (km/hr).

Mammal Abundance

Encounters with each mammal species during the transects were converted into sighting frequencies, or relative indices of abundance. Initially these were measured in terms of the number of groups sighted per km walked (grps/km). Data collected on average group size for each species was then analysed using t-tests to determine if there were any significant differences in group size between T and C and between lodges as a whole. [Some data was also provided for this test from opportunistic encounters along the T and C transects although outside the census period, particularly where data was scarce.] Subsequently the data for grps/km was converted into individuals per km (ind/km). This index was used in the analysis of impact. For species where sufficient data were available, absolute density estimates (individuals/km²) were computed using the program Distance developed by Laake et al. (1991), although these results were not used to analyse impacts and are included in this report for reference purposes only.

Forage Scrapings

Paired t-tests (1-tailed) were utilised to determine if there was a marked difference in forage scraping density between T and C from one lodge to another. Furthermore, a Pearson correlation coefficient was calculated to investigate whether there was a link between scraping density and the magnitude of traffic along T.

Tracks

Paired t-tests (1-tailed) were used to determine if there was a significant difference in the abundance of tracks between T and C. Pearson correlation coefficients were calculated to compare T track abundance and traffic, to determine whether any species was significantly affected by varying traffic intensity.

Visibility Levels

To determine if visibility levels were significantly different between T and C and between lodges as a whole the average straight-line observer-animal distances (AD) for the most abundant terrestrial species, the brown agouti, and the most abundant whole arboreal species, the saddleback tamarin, were compared using t-tests. By using these two species we could gauge the visibility through the shrub layer as well as the lower canopy.

Traffic Effects

For each of 13 species, paired t-tests (1-tailed) were used to test for overall significant differences in abundance between T and C. This test was undertaken after the data for these species had first been standardised using percentage abundance, in order to reduce the effects of major differences in abundance between lodges brought about by variables other than traffic. The Pearson correlation coefficient was calculated to investigate the relationship between the abundance of a species and the intensity of traffic along T. Hierarchical cluster analyses were used to determine if there were any similarities between T and C across all lodges for the species that showed the greatest indication of being affected by traffic. The same analysis were used to determine similarities between lodges as a whole based on average abundance per lodge of the 13 species.

Community Effects

The structure of four taxonomic groups: primates (6 species), ungulates (4 species), carnivores (4 species) and rodents (4 species), was determined using relative indices of biomass, i.e. kg observed per km walked, as insufficient data was available to determine absolute density with any degree of accuracy for all the species concerned. Weights for each species were obtained from the literature (Clutton-Brock et al. 1977, Emmons 1984, Robinson et al. 1986, Ayres et al. 1991, Mittermeier 1991, Peres 1993) and multiplied against the relative abundance of each species. Each community was compared between T and C and between lodges as a whole using paired t-tests (1-tailed) after having first converted biomass indices into percentage biomass indices. Finally, hierarchical cluster analyses were used to investigate the similarities between lodges in terms of each community.

Results