Etude coopérative EWRS: la détermination du potentiel semencier: II—estimation de la précision relative sur la moyenne à partir de composites

Des simulations d'échantillonnage ont été realisées sur des données provenant d'une expérimentation commune du groupe de travail EWRS ‘Potentiels semenciers des sols agricoles’. Cinq parcelles agricoles ont été choisies et leurs stocks semenciers analyses. Une étude par simulation de la relation liant la précision relative et le nombre de prélèvements regroupés en composites a montre (1) que l'hypothfèse de normalité des distributions des moyennes spécifiques n'est réalisée que pour environ 30% des séries lorsque le nombre de prélèvements est égal a 30 et pour 75% des séries lorsque ce nombre est de 150; (2) que la précision relative est fonction d'une part de la taille du composite (nombre de prelevements) et d'autre part du nombre moyen de semences par prélévement; (3) I'existence d'une relation de type exponentiel liant la précision relative et le nombre moyen de semences par prélevement sur les différentes parcelles expérimentales. EWRS collaborative study of seed bank estimation, II. Estimation of the relative accuracy of the mean using the compound sample procedures Sampling simulations were made using the data from a joint experiment of the EWRS working group ‘Seed banks of agricultural land’. Five agricultural plots were selected and the seed banks analysed. A simulation study of the relationship between the relative accuracy and the size of the compound sample (number of soil cores) showed (1) that the hypothesis of normality of the distribution of the species means was true for only about 30% of the sets when the compound sample size was 30 and for about 75% of the sets when the size was 150, (2) that the relative accuracy was dependent partly on the size of the compound sample and partly on the mean number of seeds per soil core, and (3) that there was an exponential relationship between relative accuracy and the mean number of seeds per soil core for the various experimental plots. EWRS-Ringsversuch zur Bestimmung der Samenbank im Boden. II. Schatzung der re-lativen Treffsicherheit des Mittelwerts von Sammelproben Mit Datenmaterial aus einem Ringversuch der EWRS-Arbeitsgruppe ‘Samenbank land-wirtschaftlich genutzter Boden’ wurden Pro-bennahme-Simulationen vorgenommen. Dazu wurden 5 Felder ausgewahlt und deren Samenvorrat analysiert. Eine Simulations-Studie über die Beziehung zwischen der relativen Treffsicherheit und der Sammel-probengröße (Anzahl der Bodenproben) ergab zum einen, daß die Hypothese der Normalverteilung der Mittelwerte von Arten nur in etwa 30% der Datensätze bei einer Sammelproben-größe von 30 und in etwa 75% bei einer Sammelprobengröße von 150 zutraf. Zum anderen hing die relative Treffsicherheit teils von der Größe der Sammelproben, teils von der mittleren Samenzahl pro Bodenprobe ab. Drittens bestand für die einzelnen Versuchsflächen eine exponentielle Beziehung zwischen der relativen Treffsicherheit und der mittleren Samenzahl pro Bodenprobe.     

Precision of soil seedbank sampling: how many soil cores?

A prerequisite for a precise estimate of the abundance of weed seeds in sou is prior knowledge of sampling variability, Based on an analysis of EWRS working group trials, it is shown that the sampling variance (s²) of seed counts increases with the sampling mean (S²) and can be predicted with the logarithmic form of Taylor's power law: log₁₀(s^z)=a+b log₁₀(x?). This relationship is constant over time for means greater than 0.1 seeds per core within each of the five different sites studied (temporal variability) but differs slightly among sites (geographical variability). An attempt is made to use a general s² : x? relationship to predict the number of samples as a function of precision and density.     

La répartition spatiale du stock semencier: comparaison de techniques statistiques

The spatial distribution of the seedbank; comparison of statistical procedures The spatial distribution of a seedbank was analysed with simulated values and data from an experimental plot. Three statistical approaches were compared: (1) adjustment of the observed frequency distributions to a theoretical distribution (Poisson and negative binomial); (2) calculation of indices based on the mean (m) and the variance (V) of samples (V/m, Lloyd's ‘mean crowding’ and aggregation index); (3) calculation of an autocorrelation coefficient which takes account of the relative positions of samples. Results using the simulated values showed that only the autocorrelation coefficient fitted the spatial distribution. It allowed the spatial distribution of the principal species in the experimental plot to be defined.     

Etude de la dynamique d'une communauté adventice

Studies of the dynamics of a weed community: II. Long-term influence of cultivation techniques on the seedbank The development of the seedbank was followed for eight years on four experimental plots with deep or shallow cultivation and with or without herbicide treatment (growing a sequence of soft winter wheat, spring barley and spring peas). The results show that the vertical distribution of seeds in the soil was not affected by cultivation in the herbicide treated plots although differences appeared in the fifth year in the unweeded plots. The total seedbank increased in the unweeded plots although it was relatively stable in the others. The increase was greater in the shallow cultivated plots. Nevertheless, a more detailed analysis of the changes in numbers at three levels of burial allowed three types of behaviour of seeds in the topsoil to be defined. The species present in the seedbank remained stable but the structure of the community was modified by increases or reductions in the numbers of some species depending on whether or not herbicides were used.     

Etude de la dynamique d'une communaute adventice

Studies of the dynamics of a weed community: III. Long-term influence of cultivation techniques on the species composition of the seedbank The species composition of the seedbank was followed for eight years on four experimental plots differing in cultivation treatment (deep or shallow) and either treated or untreated with herbicides. The results show that the species composition was not affected by herbicides but there were quantitative changes: some species, such as Euphorbia exigua and Anagallis arvensis became more dominant without increasing in number. Conversely, in unweeded plots there were qualitative and quantitative changes in the species composition with the arrival of ‘opportunistic’ species such as Thlaspi arvense and Sinapis arvensis, which considerably increased in number although they were practically absent at the start. The type of cultivation modified these changes a little by mixing, in the cultivated layer, the annual additions of new seeds with the existing seedbank.     

Influence of weed management strategies on soil seedbank diversity

The role of three weed management strategies on weed populations and community dynamics was investigated from 1989 to 1994. These were chemical weed control (CH), integrated weed control (IN) and mechanical weed control (ME). Weed populations and communities were analysed by univariate (species richness, rank:frequency diagrams) and multivariate (correspondence analysis) methods. Species richness of the soil seedbank differed with weed management strategies and over time, with an observed annual number of species that ranged from 18 to 32. The seedbank was dominated by the same pool of species in all cases, but these species responded differently among years and management strategies; their rank (relative abundance) and densities changed between strategy and within a strategy among the years. In the last 2 years of the study, the soil seedbank was characterized mainly by #Capsella bursa-pastoris     

Etude coopérative EWRS: la détermination du potentiel semencier

EWRS collaborative study of seed bank estimation: I. Studies of the relation between the mean and the variance with sampling procedure The weed seedbank was analysed in cropped plots situated in five countries. Analysis of single core and composite samples showed that there was a relation between mean and variance which was the same for all 5 sites. It followed the Taylor equation: log(variance) = a+b log (mean). However, the regression of log (variance) on log (mean) given by the different species varied with the site and within the same plot differed according to the number of species considered. The same type of study using simulations led to the same conclusions.