Measuring local biodiversity with Diversity Measures at Distance
Marcon E. and F. Puech

• Working paper on request.

• Poster (old version!).

• Abstract: This article introduces the diversity measures at distance (DMD), a new class of indicators, that are able to evaluate local diversity. The difference between those measures and all existing ones is that DMD provide, at any scale of observation, an interpretive value of the diversity according to different conceptual approaches. This is possible because this new class of measures focuses on the entities that are analyzed and, based on continuous geographically data, they are powerful to measure very precisely all the diversity levels around the observed geolocalized entities. Diversity results are intelligible thanks to the Hill numbers (Hill, 1973). Visualizing the results by plotting them provide a very detailed and intelligible map of the local diversity everywhere on the territory analyzed at the scale chosen. After introducing our methodology, a first application as a proof of concept is proposed. The entities in our empirical study are trees located in the Parc des Buttes Chaumont, a Parisian park. Our application illustrates the potential of this new approach to gauge local biodiversity by focusing on trees as taxocenes. Results and maps of local biodiversity levels are provided.

   Photo: Parc des Buttes Chaumont (Nov. 2024), © ATP.

On the computation of large spatial datasets with M
Marcon E. and F. Puech

• Working paper (pdf) and its appendix (R code).

• Abstract: Increasing access to large individual and spatial datasets, coupled with the development of computing power, has encouraged the search for suitable statistical tools to best analyse such data. In a recent article, Tidu et al. (2024) highlight the qualities of the M function (Marcon et Puech, 2010), a measure of spatial concentration in continuous space. They also express reservations about the computation times required. Our methodological work seeks to specify the processing of large spatialized data sets with M using R software. Two avenues are being explored to determine the computational performance of M. Firstly, a precise evaluation of the computational time and memory requirements for geo-localised data is carried out using the dbmss package in R by means of performance tests. Then, as suggested by Tidu et al. (2024), we also consider the possibility of approximating the geographical positions of the entities analysed. The extent of the deterioration in the estimate of M that this approach creates can thus be estimated, as can the gains in computation time made possible by the spatial approximation of locations. The complete R code is given for the reproducibility of the results. 

Diffusion of short-time work

Nevoux S., Marcon E. and F. Puech

• Working paper on request.

• Abstract: In France, short-time work was not a well-known policy before the Covid-19 pandemic. The sanitary crisis shed the light on the importance of sharing information on this public program. In this paper, we assess the local diffusion of short-time work before the Covid-19 pandemic. We argue that the geographical proximity of establishments having already used short-time work in the past constitutes an information channel regarding this scheme. Relying on distance-based methods, our stylized facts highlight the spatial and dynamic concentration of short-time work use in France between 2002 and 2014. Our econometric analysis reveals that (i) the local information about short-time work constitutes a determinant of its use, (ii) it attenuates rapidly in the first kilometers and (iii) this information is both transmitted within and between sectors. We interpret this significant spatial concentration of short-time work use, after controlling for the determinants of short-time work at the establishment level, as an evidence of information spillovers.