L’analyse des peuplements d’oiseaux, éléments d’un diagnostic écologique I. la méthode des échantillonnages fréquentiels progressifs (E.F.P.)

This paper describes a new sampling method for bird populations making use of the frequency of occurrence of each species per unit-time of 20 min (Echantillonnage Fréquentiel Progressif, E. F. P.). This technique is derived from the I.P.A. method (Blondel et al., 1970) but is not intended to replace the latter ; the two methods are designed to solve different theoretical and practical problems in ecology. 1. In an introductory chapter, it is argued that the « bird- model » allows one to put forward an ecological diagnosis of ecosystems and that bird communities can be used to test the level of complexity of ecosystems. Furthermore, from a practical point of view, we may make use of bird communities for environmental planning and monitoring. The frequency-sampling method is interesting because it permits one to gather rapidly quantitative and reliable data. 2. The material consists of 279 I.P.A. censuses or 558 frequency units, since 1 I.P.A. = 2 E.F.P., collected in 23 study areas of French Mediterranean Biome. These data were used to deter mine the composition and structure of avian communities by means of densities (I.P.A.) and frequencies (E.F.P.). Comparisons of the two sets of data allows one to determine for each species the relationship between its density and its frequency. The composition of the community involves two parameters : A) richness, that is to say the number of species. A distinction is made between total richness S or absolute number of species noted at least once in a set of N censuses, and mean richness s or mean number of species per count ; B) abundance of individuals in the communities. This parameter can be expressed either as an index of specific abundance (I.P.A. 8) which can be converted to density dt if one knows the coefficient by which to multiply I.P.A. in order to obtain (dt = I.P.A.t X c,), or by mean of fre quencies FC. The specific frequency is shown to be correlated to the logarithm of the density of the species (r between logdt and FCi = 0,84 ; P < 0,001) ; the lower the frequency of a given species, the better the correlation. Thus, frequencies of occurrence can be used as an objective measure of the number of individuals included in the community. The structure of a community accounts for its organization, that is the distribution of individuals among the different species. We recognize three ways of expressing community structure : A) the law of distribution ; the species-abundance and the species- frequency distributions are shown to fit the Galton log-normal model at a high level of statistical significance (P < 0,001). If bird communities are actually organized according to this model, censusing methods give a reliable sampling of communities. Furthermore, this adjustement allows us to check the level of homogeneity of a community ; B) the diversity of a community ; we used here Shannon’s measure of diversity : H’ = — 2 log2 p». i = 1 In all cases frequencies-diversity H’F is higher than abundance- diversity H\, which means a better equitability of individuals in the frequential method of censusing. But the two indices are highly correlated with r = 0,99. Of the two components of diver sity — richness and equitability or evenness — richness is the most important in the determination of diversity as is shown by the high correlation beween S and H’ (r = 0,93) ; C) equitability J’ ; this index shows how far the maximal potential diversity H’max is actually realized by a community of S species. Thus J’ = H’/H’max = log2 S. In most samples, J’ is very high, which suggests that breeding bird communities are relatively constant with respect to the evenness component of diversity except in uncertain environments. 3. In the third part of the paper, the practical problems of bird censuses with the frequential method are discussed. The sampling of plots must be achieved by probabilistic methods in order to have statistical interpretations of the results. Several examples are given. Advantages and disavantages of various ways of sampling are discussed, chiefly in respect with stratified sampling which results in a classification of patches of environ ment according to the ecological criteria, the importance of which one wishes to measure in bird communities. As far as bird censuses are concerned, some practical indi cations are given, concerning especially the differences between the frequential and the abundance methods (E.F.P. versus I.P.A.). The E.F.P. method is highly standardized, which is very useful for a rigorous statistical interpretation of data. This standardi zation together with the simplicity of the actual recording of data («yes» or « no » for presence or absence of each species per 20 min.) allows for an extension of the diagnosis of communities in space and time. This means that we have the possibility to undertake a semi-quantitative areography of birds at the species and at the community levels. 4. In the last chapter three concrete examples are given ; they emphasize the many ways to work out original data. A) the list of 84 breeding bird species from Mont Ventoux was obtained from 310 censuses, the sampling method is explained. Some distributional maps of birds were drawn on a geographical grid. Relative frequencies of the species were indicated in each square of the grid (5 classes of frequencies, cf. Fig. 7) ; B) ecological profiles and niche breadth were calculated for each species. Niche breadth was worked out for altitude (8 clas ses of 200 m each). The formula used is eH’x (see text). Selected examples from the case of Mont Ventoux (Fig. 8) show that species with narrow niche breadth occur at the foothill of the mountain (Sylvia melanocephala) or at its top (Anthus spinoletta) and that other species with large niche breadth (Fringilla coelebs or Sylvia atricapilla ) inhabit almost all altitude classes (excellent equita bility in each of them). Niche breadths of the six species of the genus Sylvia are shown and discussed. C) the structure of bird communities was analyzed according to the structure of vegetation. A stratified sampling led to the diagnosis of communities in ten natural and reforested areas of Mont Ventoux. For each community the following parameters are given (Fig. 9) : total richness S, mean richness s, theoretical- information diversity H’, equitability and the level of fit to Gallon’s log-normal model. The influence of reforestation on bird communities is discussed. In many instances, conifer plan tations, especially Cedar, seem successful in the rebuilding of communities. In conclusion, the standardized and time saving method discussed in this paper seems very well adapted to solve problems of theoretical and applied ecology at the community level, and can be used fruitfully for environmental monitoring