The concept of ecological niche as a hypothetical multidimensional space  has boosted the exploration of niche properties [2–4] and has enabled generalist and specialist species to be distinguished according to their respective niche breadth. Theoretical studies have tried to evaluate the origins and/or the consequences of niche-breadth differences among species. The "jack-of-all-trades is master of none" hypothesis states that the existence of generalist versus specialist species is the result of an evolutionary trade-off between the ability of species to use an extended range of resources and their capacity to exploit each one with a level of performance above those of competing species [5–8]. This trade-off has been associated with several life-history traits: generalist species are supposed to maintain higher dispersal abilities [9, 10] and to cope more easily with environmental stochasticity  while, conversely, specialists would be strongly shaped by intra-specific competition . The generalist-specialist concept could thus be appropriate to find community assembly rules , in particular in habitats where communities are subjected to varying levels of disturbance. The distinction between generalist and specialist species can pinpoint general mechanisms of species filtering, similarly to approaches using species traits rather than the species themselves. Indeed, several authors have recently focused on the processes leading to the replacement of many specialist species by a few generalist species that take place in diverse phylogenetic groups such as fish , bird  or plant assemblages . In parallel to species extinction, this so called 'biotic homogenization' process would characterize the next biodiversity crisis .
The weed communities of arable land provide an interesting model for exploring the generalist/specialist concept because the arable field habitat is characterized by an intense disturbance regime and by varying ecological conditions, both within a year (because of management practices) and across years (because of crop rotation). Compared to other vascular plants occurring in more stable habitats, one would expect arable weeds to be generalist species. Indeed, it has been hypothesized that "under a constant environment or slow environmental changes, inter-specific competition involves local processes that favour specialist species at the expense of generalist species, while under moderate to high rates of environmental change, local population dynamics increasingly favour high immigration rates of the generalist over the local competitive ability of the specialist" . However, arable fields can also be viewed as a particular habitat harbouring a set of specialized species (i.e. arable weeds) adapted to frequent but also specific disturbances. Within a year, agricultural practices can be perceived as highly specialized with the single aim of favouring a particular species, i.e. the crop, so that all species in the seed bank sharing the same requirements might well find regularly optimal conditions for completing their life cycle. Large differences between habitat breadth of weed species are observed with some species confined to arable fields in Western Europe (e.g., weeds of winter cereal fields: Agrostemma githago, Bupleurum rotundifolium, etc.) while others are able to grow both in crops and in other less disturbed habitats (e.g., Galium aparine, Lapsana communis). Even within arable fields, differences can be observed in niche position and breadth between weeds that are specialized to a particular crop type or to particular soil conditions, and weeds that are present almost everywhere [19, 20]. Finally, even if arable fields are characterized by stochastic conditions, a specialized species could persist temporally, for example, with dormant diaspores that would wait for optimal environmental conditions e.g., high relative summer air humidity for arable bryophyte species  or each time a favourable crop occurs in the rotation . Therefore, even though most agroecosystems experience high levels of disturbance, it is not clear whether the current arable weed flora is dominated by specialist or by generalist species and what would be the resulting variation in ecosystem function . The present paper addresses this question and classifies the French arable weed flora along a specialist-generalist gradient, using vegetation records from a national monitoring network for applying and comparing six different specialisation indices available in the ecological literature. It should be noted here that what is meant by generalist species are species able to exploit many or all the niches within the "arable field" habitat regardless of their ability to occupy other habitats.
A second set of questions relates to possible relationships between the degree of specialisation of weed communities and disturbance regime that are related to the choice of contrasted agricultural management options. In this paper, we focus on two sets of situations that result in different levels and regime of disturbance.
The first situation compares the weed flora in maize cultivated as a monoculture and in a crop rotation. Monoculture means here that the same crop species is cultivated for several consecutive years. Crop rotation means that each year a different crop species is cultivated. Crop rotation induces disturbances that vary with the crop grown each year (planting or maturation dates, growth habit, competitive ability, associated cultural practices, fertiliser requirements and more or less specific herbicides) while the disturbance regime is constant in monocultures. While many recent studies focused on the impact of crop rotation on weed diversity [24–26], the present study aimed to quantify the functional shift in weed composition. There could be two alternative responses of weed communities to crop rotation (i) crop rotation could either favour generalist species and monoculture specialist species, or, (ii) crop rotation could favour specialist species of each crop (i.e., species that are associated with conditions of a particular crop), so that weed communities would mainly be composed of specialist species alternating each year, persisting within the seed bank during the unfavourable years.
The second situation analyses the shift in weed communities that has taken place in the same arable fields between the 1970 s and the 2000 s. It is here assumed that the level of disturbance has significantly increased with agriculture intensification (i.e., increasing number of herbicide treatments, increasing depth and frequency of tillage, see ) between the two surveys. Increased N-fertilization and systematic liming or drainage have homogenized soil conditions across the sampled fields. In addition, recurrent changes in cultivation techniques since the 1970 s (tillage or no-tillage systems, new herbicides) are likely to have translated into continuous environmental changes for the arable flora. These changes may have hampered specialist species and/or favoured generalist weed species.