The field data were acquired during February–May and September–December 1999, and May–September 2000. Wet-season data, months in which precipitation exceeds 100 mm, include data from February, April, May, September–November 1999, and March–June, September 2000. Dry-season data include data from March and December 1999, and July-August 2000. We chose sites within existing macrohabitats (primary / secondary forest) to establish a total of ten rectangular transects (six in primary forest, four in secondary forest). Each had a north–south extension of 200 m and an east–west extension of 100 m. For data acquisition the complete transect length was subdivided into 25-m subunits (SU; 24 SUs / transect). Transect paths were kept open so that walking at a constant speed was possible at all times. We avoided extensive cutting and thus manipulation of important habitat features. Transects were arranged in pairs, thus ensuring that all habitat types of a certain area within the inhomogeneous forest were covered. The rectangular transect design is a combination of two widely used standard techniques (see Heyer et al. 1994, Rödel and Ernst 2004). For a discussion of the advantages of rectangular sample units see Krebs (1989) and McCune and Grace (2002).

We characterized all 25-m SUs using several variables that were recorded at two defined points (beginning and midpoint of each SU). These variables included vegetation density in four strata (canopy: > 20 m, lower tree stratum: 3–10 m, bush and shrub stratum: 0.5–1.5 m, understory: < 0.5 m) divided into seven categories (Table A1). Soil was assigned to one of seven categories (compare Lieberoth 1982 and Table A1). Leaf-litter coverage was estimated according to Braun-Blanquet (1964). The vegetation of all 25-m SUs within a distance of about 100 cm left and right of the transect was recorded by counting the number of plants belonging to a certain category (plant stem diameter at breast height in centimeters, dbh1: 0–5 cm; dbh2: 6–10 cm; dbh3: 11–20 cm; dbh4: 21–50 cm; dbh5: > 50 cm). Definitions of habitat variables are summarized in Table A1. In order to quantify the availability of potential aquatic breeding sites, every aquatic habitat (lentic and lotic) located at a maximum distance of 25 m from either side of the transect was recorded with respect to type, surface, and depth. Substrate moisture was determined in four categories (from dry to saturated wet) during every transect walk. The habitat variables considered in the analyses were substrate type, substrate moisture, leaf-litter coverage, vegetation density in four strata, number of plants in the five respective dbh-categories, lotic habitats, and lentic habitats, thus summing up to a total of 14 variables.

Sampling was performed independent of prevailing weather conditions. Repeated controls of identical transects on consecutive days were avoided to ensure independence of samples. Transects were intensively patrolled at a constant speed (0.30–0.35 m/s), visually recording all amphibians within a distance of 100 cm from either side of the path. To avoid duplicate records, captured frogs were marked by toe clipping (Donnelly et al. 1994, Henle et al. 1997). Recaptures were excluded from the analyses. Individuals below nine mm SVL were not marked due to their small size. Probability of detection was not tested but seemed to be independent of frogs’ sizes. Capture probability varied according to frog size and climatic conditions, but generally was higher than 90%. Frogs were more likely to escape after rain and larger frogs were more likely to escape than smaller ones. In a comparative analysis of amphibian monitoring programs using transects in East Africa, West Africa (including the data presented herein), Madagascar and Borneo, we recently have shown that ≥ 20 independent transect walks seem to be necessary to achieve a species saturation (Veith et al. 2004). During this study, every Taï transect was walked independently at least 41 times. It is thus justified to assume that the local species assemblages have been almost completely recorded. Relative abundances of all leaf-litter frogs recorded during transect walks are summarized in Table A2.

TABLE A1. Three habitat variables measured on transects, and the categories to which their measurements were assigned.

		Vegetation†		Substrate types		Leaf cover (%)
Category		Absent		forest soil		0–20
		Transition		arenaceous forest soil		21–40
		Gaps predominating		loamy soil		41–60
		Transition				61–80
		Closed areas predominating		sabulose soil		81–100
		Transition		muddy soil
		Swampy soil

† Vegetation density was measured in four strata (see second paragraph, above), and assigned to one of seven categories.

 

TABLE A2. Abundance matrix of species recorded in primary and secondary forest habitats during the study in Taï National Park, Ivory Coast. Relative abundance, as considered in analyses, is given as individuals per transect hour (i/th).

Species	i/th primary	i/th secondary	i/th total	Total no. specimens
Bufonidae
Bufo maculatus	-	0.0108	0.0026	1
Bufo taiensis	0.0035	-	0.0026	1
Bufo togoensis	0.2280	0.0968	0.1961	75
Ranidae
Amnirana albolabris	0.1485	0.1075	0.1386	53
Amnirana occidentalis	0.0173	0.0108	0.0157	6
Aubria occidentalis	0.0207	-	0.0157	6
Ptychadena aequiplicata	0.3040	0.0215	0.2353	90
Ptychadena longirostris	-	0.0108	0.0026	1
Petropedetidae
Phrynobatrachus accraensis	-	0.0108	0.0026	1
Phrynobatrachus alleni	5.0639	0.4624	3.9451	1,509
Phrynobatrachus annulatus	0.0069	-	0.0052	2
Phrynobatrachus fraterculus	0.0104	0.0108	0.0105	4
Phrynobatrachus guineensis	0.0484	0.0323	0.0444	17
Phrynobatrachus gutturosus	0.0967	-	0.0732	28
Phrynobatrachus liberiensis	1.0363	1.5591	1.1634	445
Phrynobatrachus phyllophilus	0.7081	0.0430	0.5464	209
Phrynobatrachus plicatus	1.0155	0.0538	0.7817	299
Phrynobatrachus tokba	0.0242	1.1613	0.3007	115
Phrynobatrachus villiersi	1.7098	0.2151	1.3464	515
Astylosternidae
Astylosternus occidentalis	0.0035	0.0215	0.0078	3
Arthroleptidae
Arthroleptis comb.	1.0328	1.3871	1.1190	428
Cardioglossa leucomystax	0.0242	0.0108	0.0209	8
Hyperoliidae
Kassina lamottei	0.0933	-	0.0706	27
i/th total	11.5959	5.2258	10.0471	3,843

   Notes: Arthroleptis species (two species known to occur, distinction possible by means of genetic and acoustic characters) could not be separated morphologically and thus were treated in the analysis as a single artificial taxon Arthroleptis comb.; i/th primary = no. of frogs per species per transect hour in primary forest transects; i/th secondary =  no. of frogs per species per transect hour in secondary forest transects; i/th total = total number of individuals per species per transect hour in primary and secondary forest combined (rows) and total number of individuals per transect hour in primary forest, secondary forest, and both types of forest combined (columns); last column = absolute number of individuals recorded per species, column sum = absolute number of individuals recorded; - = absent; total number of transect walks = 765 (one transect walk lasted app. 30 min); nomenclature of amphibian species is according to Frost (2002). A dash in a species cell indicates the species could potentially be found (because it lives in the forest, but it was not recorded).