Mosaic cycles in agricultural landscapes of Northwest Europe

Abstract

Mosaic cycles were originally understood as cyclical regeneration phases in forests. In this review, we shall examine how far the concept can be extended towards cyclical mosaics of habitat quality in patterned landscapes as a special case of ‘dynamic landscapes’. We will concentrate on habitats and plants in European temperate agricultural landscapes and grasslands in particular. Mosaic cycles of habitat quality are characterised by spatiotemporal shifts between disturbance and secondary succession. We found evidence for mosaic cycles in traditional agricultural systems, modern crop farming, and in recent conservation management. The relevant disturbance parameters to describe land-use drivers of mosaic cycles are spatial extent, frequency, and magnitude (biomass loss). Land-use-related drivers are usually regular and deterministic in space and time, with the exception of year-round grazing by free-ranging large herbivores. Fluctuating soil resources such as water and nutrients in interaction with climate variability add a stochastic component to these (land-use-related) drivers. The proportion of deterministic and stochastic components and their autocorrelation in time and space divides purely deterministic mosaic cycles from purely stochastic dynamic landscapes. In a second part, we briefly review plant life-history traits that may facilitate survival of plants in mosaic cycles of habitat quality. Theoretical studies emphasise (i) dispersal functions for extinction and recolonisation processes of metapopulations, (ii) storage effects as a component of buffered population growth in response to temporal fluctuations of habitat quality, and (iii) competitive ability in metacommunities. We propose a simple scheme relating these functions to the temporal and spatial correlation of patterned landscapes. There are only a very limited number of field studies available that give some support for the proposed scheme. We provide perspectives for further research in this field.

Zusammenfassung

Mosaikzyklen sind ursprünglich als zyklische Regenerationsphasen von naturnahen Wäldern aufgefasst worden. In dieser Übersicht dehnen wir das Konzept auf zyklisch wiederkehrende Habitatqualitäten in Landschaften aus und betrachten es als Spezialfall von ‘dynamischen Landschaften’. Wir konzentrieren uns auf Agrarlandschaften der europäischen gemäßigten Zone, speziell auf Grünland, aus dem Blickwinkel von Naturschutz und Landschaftspflege. Mosaikzyklen sind durch räumliche und zeitliche Wechsel zwischen Störungen und Sekundärsukzessionen gekennzeichnet. Wir fanden Belege für Mosaikzyklen in traditionellen agrarischen Anbausystemen und in aktuellen Richtungen der Landschaftspflege. Räumliche Ausdehnung, Frequenz und Biomasseentzug sind wesentliche Störungsparameter für die Charakterisierung des Einflusses der Landnutzung auf Mosaikzyklen.

Landnutzungen als Treiber von Mosaikzyklen sind üblicherweise regelmäßig und deterministisch, mit der Ausnahme von ganzjähriger Beweidung durch freilaufende Großherbivoren. Jährlich schwankende Wasser- und Nährstoffangebote fügen dem eine stochastische Komponente hinzu. Die Anteile deterministischer und stochastischer Komponenten und ihre Autokorrelationsstruktur in Zeit und Raum trennen rein deterministische Mosaikzyklen von rein stochastischen dynamischen Landschaften.

Im zweiten Teil geben wir einen kurzen Überblick über biologische Merkmale (Life-History Traits) von Pflanzen, welche das Überleben in Mosaikzyklen ermöglichen. Theoretische Studien betonen die Bedeutung von (i) Ausbreitungsmechanismen für Rekolonisationsprozesse im Rahmen von Metapopulationen, (ii) Speichereffekten reproduktiver Kapazität als Puffermechanismus von Populationen bei zeitlicher Änderung der Habitatqualität, sowie (iii) Konkurrenzkraft im Rahmen von Meta-Gemeinschaften. Wir schlagen ein einfaches Schema vor, das diese Funktionen in Abhängigkeit von der räumlichen und zeitlichen Korrelation des Landschaftsmusters darstellt. Bisher gibt es nur eine sehr begrenzte Zahl von empirischen Studien zu diesem Thema. Diese unterstützen das vorgeschlagene Schema zum Teil. Wir geben Hinweise für weitere Forschungen in diesem Feld.

Introduction

In the past decades, classical patch-centred ecological theory has been broadened by introducing the effects of spatial habitat configuration and temporal habitat availability on the survival of species (Fahrig, 1992; Hanski, 1994; Levin & Paine, 1974). This has resulted in the concepts of hierarchical patch dynamics and dynamic landscapes (White & Pickett, 1985; Wu & Loucks, 1995) where species are facing habitats which are randomly changing in quality, shape, and spatial location (May, 1974). Correspondingly species can only persist within these dynamic landscapes when the population dynamics are in pace with landscape dynamics.

Dynamic landscape theory has important implications when trying to understand the survival of species in traditional agricultural systems as well as in contemporary conservation management. Many traditional land uses were characterised by shifts between periods of cultivation and secondary succession. Similar shifts are now often encountered in conservation management especially when financial restrictions call for extended fallow periods between disturbances such as grazing, mowing, or burning. Consequently, as usually a different part of a reserve is managed each year, these disturbances also vary in space (e.g. Radeloff, Mladenoff, & Boyce, 2000; Wegener, 1998). Extended fallow periods can promote changes within habitat quality by means of succession. Moreover, the successional stages may differ in habitat quality thus rendering habitat quality a function of successional change (McCook, 1994). The species will experience those changes as spatiotemporal cycles in habitat quality. If conservation management generates such cycles rather than promote constant habitats, the question will arise on how this would influence biodiversity and which life-history traits would enable survival.

However, the successional pattern which is generated by agricultural or conservation land use may significantly differ from the random patterns of dynamic landscapes. Firstly, the habitat quality is mainly decreased by stronger competitors appearing in the course of succession and not (alone) by drastic exterior events, such as frost, drought, or flooding. Secondly, if human disturbances repeatedly set back this secondary succession without altering the abiotic environment, an orderly return of successional stages could be expected. Hence the resulting cycles of habitat quality represent a special case of dynamic landscapes. Moreover, they point to the mosaic cycle concept as yet another strand of theory dealing with dynamic habitats.

The classical mosaic cycle concept (coined by Remmert, 1991; building on Aubreville, 1936; Watt, 1947) and the concept of shifting mosaic steady states as proposed by Bormann and Likens (1979) describe cyclical fluctuations in age-structured forest stands. Here cycles are regarded as a sequence of community states which depend on the building, maturation, and degeneration phases of a dominant plant (Herben, During, & Law, 2000). The degeneration of dominant plant stands consequently results in patches that are open for recolonisation. As local cycles tend to be out of phase, maturation and degeneration will lead to a mosaic of community states that shift across the habitat.

Some well-known examples also describe these shifting mosaics in various ecosystems, for instance grasslands (Watt, 1947), heathlands (Gimingham, 1988), wetlands (e.g. Bornette & Amoros, 1996), tall-grass prairie (e.g. Fuhlendorf & Engle, 2004), arid steppes (e.g. Soriano, Sala, & Perelman, 1994), and river ecosystems (e.g. Kollmann, Vieli, Edwards, Tockner, & Ward, 1999; Stanford, Lorang, & Hauer, 2005). In some of these examples the shifting mosaics are driven internally, i.e. by biotic interactions (Watt, 1947) whereas others are driven by external disturbances, such as flooding (Kollmann et al., 1999).

In this review, we will apply the mosaic cycle concept to agricultural landscapes in temperate Europe. We will focus on mosaic cycles of habitat quality of plants and restrict our review to the conservation context. The classical mosaic cycle concept was originally developed to describe the spatiotemporal patterns and regeneration processes within a community (Herben et al., 2000). Our approach differs from the classical concept in so far as (1) several communities in a successional sequence are involved, (2) larger spatial scales than mosaic cycles in forests are involved, and (3) the dynamics are mainly driven by human disturbances and variations in resource supply. We will present parameters that allow to describe the frequency, spatial distribution and randomness of disturbances that will strongly influence the spatiotemporal pattern of mosaic cycles.

Species can only persist in mosaic cycles if they are able to track the shifts in habitat quality and they will become extinct if they fail to keep up with this (Thomas, 1994). On assuming that prolonged phases of succession are accepted in conservation management, the question arises which plant species or life histories will be at risk in the process. In our second part, we will therefore review the life-history traits that can facilitate plant survival in mosaic cycles. Our main assumption will be that different spatiotemporal arrangements of mosaic cycles will filter different trait expressions, i.e. there is no single life history that can survive in every kind of mosaic cycle. We will end our review with promising further research directions.