1
This is the accepted version of the article. For the final version visit 1
http://www.cell.com/trends/ecology-evolution/ 2
3
4
Conceptual domain of the matrix in fragmented landscapes 5
6
Don A. Driscoll, Sam C. Banks, Philip S. Barton, David B. Lindenmayer, Annabel L. Smith 7
8
ARC Centre of Excellence for Environmental Decisions, the National Environmental Research 9
Program Environmental Decisions Hub, Fenner School of Environment and Society, The 10
Australian National University, Canberra ACT 0200, Australia. 11
12
In extensively modified landscapes, how the matrix is managed determines many 13
conservation outcomes. Recent publications revise popular conceptions of a homogeneous 14
and static matrix, yet we still lack an adequate conceptual model of the matrix. Here, we 15
identify three core effects that influence patch-dependent species, through impacts 16
associated with movement and dispersal, resource availability and the abiotic environment. 17
These core effects are modified by five 'dimensions': (i) spatial and (ii) temporal variation 18
in matrix quality, (iii) spatial scale, (iv) temporal scale of matrix variation, and (v) 19
adaptation. The conceptual domain of the matrix, defined as three core effects and their 20
interaction with the five dimensions, provides a much-needed framework to underpin 21
management of fragmented landscapes and highlights new research priorities. 22
23
2
A matrix focus is now both important and possible 24
Biodiversity conservation often focusses on patches of native vegetation in a surrounding matrix 25
that is highly modified by agriculture or urbanisation [18, 19]. The patch-matrix model of 26
landscapes [20] includes patches that are useful for conservation and the matrix in which the 27
patches are embedded [21] (see Glossary). Assumptions underpinning the patch-matrix model 28
are reasonable in many situations, particularly in fragmented and relictual landscapes where 29
there are patch-dependent species [22-24]. However, the matrix surrounding remnant vegetation 30
can have a strong influence on species occurrence and spatial dynamics [25, 26] and can be more 31
important than the size and spatial arrangement of remnant patches [2, 27, 28]. The growth in 32
knowledge about the matrix means it is now possible to develop a detailed synthesis of the 33
mechanisms by which the matrix directly, or indirectly drives the distribution of patch-dependent 34
species in space and time. 35
36
Not only is such a synthesis possible, it is also urgent. The nature of the matrix has profound 37
implications for conserving biodiversity [28, 29]. Management of the matrix can limit or 38
exacerbate the impacts of habitat loss and fragmentation [30]. Habitat loss and fragmentation are 39
the biggest threat to biodiversity globally [31]. In highly modified landscapes, further loss of 40
remnant vegetation is limited because most of it is already gone, or because what remains is 41
legally protected [32, 33]. Where this is the case, modifying the matrix will be the major form of 42
landscape change in the future, and will therefore likely be the main process influencing 43
biodiversity conservation. There is now a pressing need for a comprehensive theoretical 44
framework of the matrix to guide the way scientists and land managers think about matrix 45
ecology. 46
3
47
While there has been much conceptual development in the habitat fragmentation literature [22, 48
26, 34], the concepts related to how the matrix influences patch-dependent species have not been 49
thoroughly synthesised. In this review, we build on progress made within ecological sub-50
disciplines [25, 35, 36], and on research into edge-effects [37] and habitat fragmentation [26, 51
34], to describe the conceptual domain of the matrix in fragmented landscapes. 52
53
Our approach to understanding the conceptual domain of the matrix is to synthesise ideas from 54
the empirical literature. However, instead of providing a list of matrix effects [e.g. 25, 35, 36, 55
38, 39], we illustrate relationships among mechanisms in a conceptual model. We demonstrate 56
through the conceptual model that what previously were considered primary effects of the matrix 57
are actually secondary outcomes of three 'core effects' (see Boxes 1 and 2). In the second part of 58
our review we identify five influential 'dimensions' and show how these modify the way that core 59
effects play out. The resulting conceptual model of the matrix can help to improve 60
communication of matrix ideas, and guide future research, including research that addresses new 61
questions about interactions between core effects and dimensions associated with time, space and 62
adaptation. 63
64
Core effects of the matrix 65
After considering the range of effects that the matrix can have on patch-dependent species [using 66
empirical literature, also canvased in numerous reviews: 19, 25, 34-36], we identified three 67
fundamental ways that the matrix influences the spatial dynamics of populations and species 68
occurrence in fragmented landscapes. The matrix can influence population persistence in 69
4
fragmented systems through effects associated with (i) movement and dispersal; (ii) resource 70
availability, and; (iii) the abiotic environment (Figure 1). 71
72
Movement and Dispersal. Matrix quality influences the outcome of movement into the matrix 73
Recent reviews report that movement between patches is enhanced as the matrix becomes 74
structurally more similar to the remnant patches [40, 41]. For example, when pastures are 75
replaced by tree plantations, colonisation of forest patches by forest specialists can increase [4]. 76
However, the matrix can influence immigration and emigration in other ways. Sharp ecotonal 77
boundaries between a patch and the matrix can cause individuals to cluster inside remnants 78
('fence effects') [1]. If a species does venture into the matrix, rapid movement through 79
unfavourable habitat could enhance connectivity between separated habitat patches [42]. On the 80
other hand, dispersal or movement between disjunct habitat patches might decline due to altered 81
behaviour, or increased mortality [2, 5, 26, 43]. The influence of the matrix as a demographic 82
sink has received little research attention, although in theory, density-independent emigration can 83
increase the risk of local extinctions [44]. 84
85
Resource availability. Matrix resources could aid patch-dependent species or support matrix 86
specialists. 87
The role of the matrix as a resource base for species that invade remnant patches has long been 88
understood [19] (Box 3). For example, red squirrel Tamiasciurus hudsonicus populations thrived 89
on pine-seeds in Canadian pine plantations. Squirrels subsequently invaded remnant broad-leaf 90
forest and ate Brown Creeper Certhia americana eggs, increasing the rate of nest failure of this 91
patch-dependent bird [16]. On the other hand, if the right resources are provided, the matrix can 92
5
be converted to habitat and desirable native species can live throughout the landscape [e.g. 45]. 93
However, if species remain patch-dependent, they might nevertheless use resources within the 94
matrix as a food subsidy [34]. With the possible exception of bees that can forage outside of the 95
nesting patch [e.g. 14], evidence that patch-dependent species gather resources outside of the 96
patch to support higher population densities inside the patch is limited [e.g. 46]. 97
98
Abiotic environment. The matrix influences microclimate and disturbance regimes of patches. 99
The physical structure of the matrix is often different from habitat patches and can alter the 100
environmental conditions within patches [19, 37], particularly when treed landscapes are cleared 101
[25]. Microclimatic changes associated with increased light and wind penetration can have far-102
reaching effects on patch-dependent species, increasing the risk of local extinction [7, 47]. In 103
addition, species that prosper under the altered microclimate can colonise remnant vegetation 104
and drive edge-sensitive species into the remnant core [37, 48]. 105
106
Changes to disturbance regimes in the matrix can also affect patch-dependent species. Larger 107
and more frequent fires can occur if there are more ignitions in the matrix [11], or when the fuel 108
structure in the matrix is changed by forest logging [11, 49] or by invasive grasses [17]. 109
Conversely, active fire suppression in matrix environments can reduce rates of natural 110
disturbance in patches [3]. Altered microclimate and disturbance regimes can advantage some 111
species, often invasive exotic species [6, 17], but disadvantage others, often species that depend 112
on remnant vegetation [8]. Increased disturbance associated with urban or mining landscapes can 113
also drive local extinctions in patches [9, 10]. 114
115