Austral Ecology (2011) 36, 35–45
Mammal responses to matrix development intensity
aec_2110
35..45
MEGAN J. BRADY,1,2,3* CLIVE A. MCALPINE,3,4 CRAIG J. MILLER,2 HUGH P. POSSINGHAM3 AND GREG S. BAXTER1,3 1 The University of Queensland, School of Integrative Systems, Gatton, Qld. 4343 (Email:
[email protected]), 2CSIRO Sustainable Ecosystems, 3The University of Queensland, The Ecology Centre, School of Biological Sciences, and 4The University of Queensland, Landscape Ecology and Conservation Group, School of Geography, Planning and Environmental Management, St. Lucia, Queensland 4072, Australia
Abstract The landscape matrix is increasingly being recognized as important to biodiversity conservation. The nature of the matrix impacts the persistence of species in human-modified landscapes through its pervasive influence on adjacent habitat and through the habitat value of the matrix itself. However, previous studies have not isolated the effects of the matrix from the effects of other aspects of landscape modification, such as habitat loss and fragmentation, and much remains to be understood about the independent impact of the matrix on wildlife. We investigated the effects of the matrix on mammal abundance and landscape use in south-east Queensland, Australia. Mammals were surveyed in patch ‘core’, patch ‘edge’ and ‘matrix’ landscape elements along a rural–suburban gradient of matrix development intensity quantified by a weighted road-length metric, which was significantly correlated with housing density, while controlling for potentially confounding patch and landscape attributes. Response to increasing matrix development intensity was highly species-specific. Several native species declined in abundance; however, others were more resilient to moderate levels of matrix intensity, one species increased in abundance, and at least one species appeared unaffected by matrix intensity. Native species richness peaked at moderate levels of matrix development intensity. Exotic species richness and feral predators increased with matrix intensity and were negatively correlated with native species. Species response to matrix intensity appeared related to their use of edge or matrix habitat. An ability to use the matrix per se, however, may not translate into an ability to persist in a landscape where development substantially reduces the habitat or movement value of the matrix. Key words: edge, mammal, matrix, peri-urban landscape, urban–rural gradient.
INTRODUCTION The matrix is now globally recognized as an important influence on biodiversity in human-modified landscapes (Franklin & Lindenmayer 2009). In this study we define the matrix as the human-modified area of the landscape that is not ‘traditional’ habitat for native species but potentially once was, as opposed to areas that are ‘non-habitat’ for a species, for example, a different vegetation type. Across Australia and worldwide, this modified matrix is becoming increasingly ubiquitous. In many regions remnant vegetation is still rapidly disappearing with rampant road expansion (Laurance 2009), while rural landscapes on city fringes are coming under increasing pressure from housing development (Queensland Government 2005), being transformed into mosaics of patches of remnant vegetation surrounded by a matrix of varying densities of roads, buildings, people and modified vegetation (Brady et al. 2009). Fragmentation research *Corresponding author. Accepted for publication November 2009.
© 2010 The Authors Journal compilation © 2010 Ecological Society of Australia
has historically focused on patch attributes as the key factors influencing vertebrate distribution and abundance, with limited conceptual understanding and analysis of the importance of the matrix. However, the matrix is important for at least two separate reasons; for (i) its impact on remnant habitat, whether through altering landscape habitat patterns (McKinney 2002), or on disturbance of a habitat patch (Brady et al. 2009) and for (ii) the habitat value of the matrix itself (Barlow et al. 2007), including its impact on functional landscape connectivity (Ricketts 2001). New ways of conceptualizing landscapes are emerging, for example, as mosaics of resource availability, which recognize the fact that species need to move through and utilize the entire landscape mosaic (Wiens 1997; Pope et al. 2000).This landscape mosaic perspective seems especially pertinent in highly modified landscapes with low levels of suitable habitat remaining as isolated fragments, where species may be forced to change their traditional patterns of landscape use. Diffendorfer et al. (1995) found that as fragmentation increased, small mammal species moved further, but less often, within their daily foraging doi:10.1111/j.1442-9993.2010.02110.x
36
M . J. B R A DY ET AL.
range. Species may increase or decrease their exposure to edge or matrix habitats in response to, for example, changing habitat quality or anthropogenic disturbance. An ability to tolerate and/or use modified areas of the landscape, such as edge or matrix habitats, can increase resources available to an individual (Withey & Marzluff 2009) and/or allow movement to where other resources occur, including seasonal food sources or mates. Matrix tolerance, or abundance in the matrix, has been proposed as a predictor of species success in modified landscapes (Laurance 1991; Gascon et al. 1999; Jules & Shahani 2003). Importance of matrix tolerance may depend, however, on the matrix itself. The matrix is not uniform in character or effect and both anthropogenic and habitat attributes of the matrix can shape its impact. Use of the matrix may depend on matrix type and similarity to the primary habitat of a species, or an area. For example, Perault and Lomolino (2000) found forest mammals used matrix of secondary forest much more often than clear-cut matrix. Matrix surrounding a remnant forest that is similar in habitat structure to the forest may even mediate the effects of habitat loss and fragmentation (Debinski 2006).Tolerance of the matrix may be more important in landscapes where the matrix is highly dissimilar structurally and floristically to the native habitat of the area, such as may be the case in highly urbanized landscapes. As a landscape is developed, not only do matrix attributes change, but remnant habitat attributes can covary, and habitat loss and fragmentation usually increase (Moffatt et al. 2004). This can confound the effects of the matrix. Attributes, such as patch size and the amount of habitat in the landscape, are known to determine wildlife species assemblages (Cox et al. 2004; Vieira et al. 2009), as well as influence matrix use (Tubelis et al. 2004) and could be disguising the role of the matrix in species’ distribution patterns. Other studies have suggested that observed ‘matrix effects’ could often be due to a confounding with patch quality (Haynes & Cronin 2004). It is vital we understand what the independent effects of the matrix are on species distribution and abundance to help curb the impact of continued human landscape modification. We studied the mammal community throughout modified landscapes along a gradient of matrix intensity in south-east Queensland, Australia. Our aim was to investigate the impact of matrix development intensity on the mammal community, independent of patch and landscape attributes, including patch size and the amount of habitat in the landscape. We quantified individual mammal abundance and distribution across modified landscape elements. We hypothesized that matrix development intensity will impact species by affecting their landscape use and that species landscape abundance along the gradient of matrix intensity doi:10.1111/j.1442-9993.2010.02110.x
will be related to their use of patch edge and/or matrix habitat. We placed a high priority on controlling for potentially confounding patch and landscape attributes in the landscape selection process, while systematically varying the intensity of matrix development based on a weighted road-length (WRL) metric, which was also correlated with housing density and human disturbance throughout landscapes (hereafter termed a gradient of matrix development intensity). METHODS Study area The study was conducted in the Toowoomba Regional Shire, south-east Queensland, Australia (supplementary material Fig. S1). Across the study region, remnant eucalyptus forest mainly exists as small isolated patches surrounded by a cleared matrix, historically used for cattle grazing. Grazing is still the largest single land use; however, the rural matrix is being converted to residential use of varying housing densities, making it an ideal region to study the effects of matrix development intensity. A matrix development intensity gradient was quantified by a weighted road-length (WRL) metric (described below) and was characterized by increased anthropogenic disturbance, such as increased housing density, closer proximity of sample sites to houses and higher human disturbance across the entire landscape mosaic, including inside remnant forest patches. Mean patch size was 5.9 ha (⫾1.4 ha) representing an average of 14.7% (⫾3.3% or 2.6 ha) remnant vegetation remaining in landscapes (Brady et al. 2009). Landscapes with this level of habitat loss and fragmentation are common across the region (Manning et al. 2003). Study remnant patches consisted of tall open forest and overall, remnant patch floristics and habitat structure did not change significantly with matrix development intensity (Brady et al. 2009). Common canopy species included tallowwood (Eucalyptus microcorys), small-fruited grey gum (E. propinqua) and pink bloodwood (Corymbia intermedia). Exotic species lantana (Lantana camara) and broad-leaf privet (Ligustrum lucidum) were common in the midstorey of patches along the entire matrix gradient. Introduced Kikuyu grass (Pennisetum clandestinum) dominated the ground layer of the matrix at the urban end of the gradient, while blady grass (Imperata cylindrica) commonly occurred throughout landscapes along the gradient.
Landscape selection To be as ecologically relevant as possible to the wide range of mammal species likely to be encountered, a study landscape was defined as the area within a 500-m radius (78.5 ha) of a remnant forest patch edge (Fig. 1). Home range estimates and between-remnant movements can vary widely, however, for the species of interest, and are probably flexible depending on patch and landscape characteristics (Pope et al. 2004). In urban landscapes in south-east Queensland, similar to many in this study, Fitzgibbon et al. (2007) observed north-
© 2010 The Authors Journal compilation © 2010 Ecological Society of Australia
M A M M A L R E S P O N S E S TO M AT R I X I N T E N S I T Y
37
fore provide a more ecologically appropriate classification of roads within our study landscapes, we developed a WRL metric, based on road surface, width and verge vegetation management, and also informed by a road kill study carried out in the region just prior to fieldwork in this study (Johnston 2008). Length of road within each landscape was measured in ArcView 9.2 (Environmental Systems Research Institute, Redlands, California) and then multiplied by its assigned weighting. For further details of data and methods used to derive these weightings see Brady et al. (2009).
Landscape elements (sites)
Fig. 1. Landscape survey design showing landscape elements (study sites) sampled. C, core; E, edge; M, matrix.
ern brown bandicoots (Isoodon macrourus) to rarely move more than 500 m. Between-remnant movements of yellowfooted antechinus (Antechinus flavipes) averaged 352 m in highly fragmented landscapes in South Australia (Marchesan & Carthew 2008). Medium and large macropods also respond to habitat within a 500-m radius of their refuge and only rarely disperse distances greater than this (Arnold et al. 1993; Laws & Goldizen 2003; Murray et al. 2008). Landscapes were selected with the aim of controlling for potentially confounding patch and landscape attributes. Selection was based on compliance with criteria related to matrix land use type and intensity, focal patch characteristics, such as size, shape, remnant vegetation type and position in the landscape, landscape composition and temporal patterns of land use within the surrounding landscape (supplementary material Table S1). Following ground-truthing to check compliance and validity of Geographic Information System datasets, 19 landscapes were selected that spanned the widest and most evenly spread range possible of matrix development intensity, based on a WRL metric, described below.
Weighted road-length (WRL) metric We characterized the change in landscape from rural to suburban by using a metric based on the length of roads in each landscape, the logic being that road density and increased intensity of land use are interdependent (Forman & Collinge 1997; Watson et al. 2005). However, we also recognized that all roads are not equal in their impact on wildlife. Multiple aspects of roads and road corridors, such as width, traffic volume, surface type and properties of vegetation in the verge, can influence their level of impact (Forman & Alexander 1998; Bellamy et al. 2000; Rico et al. 2007). To there-
© 2010 The Authors Journal compilation © 2010 Ecological Society of Australia
In each landscape three different landscape elements were sampled for mammals and habitat attributes: patch ‘core’, patch ‘edge’ and ‘matrix’, giving a total of 57 sites. The landscape study design ensured that the three elements in each landscape had similar landscape context, while mammal data of each element can also be aggregated for comparisons of ‘whole’ landscapes. Edge sites were located on the very edge of the focal patch, at the centre of the landscape unit, equidistant (50 m) between the core and matrix sites (Fig. 1). Edges were defined by the limit of the continuous canopy (Harper et al. 2005). Core sites were located at least 70 m from any patch edge. Matrix sites were located 50 m into the matrix from the patch edge.
Mammal surveys Mammals were surveyed by a combination of Elliott traps, wire cages, hair funnels, scats and direct sightings. All 57 sites across the 19 landscapes were surveyed twice, once in spring and once in summer of 2007–2008, with at least 4 months between first and second surveys. The core, edge and matrix sites of each landscape were sampled simultaneously. Traps were set for three consecutive nights on the two occasions, resulting in 11 970 trap-nights. A pilot study conducted in 18 sites in spring and summer of 2006–2007 resulted in no new species being captured in a landscape after the 3rd consecutive trap-night (M. Brady, unpubl. data, 2007). We also considered the use of sand plots and pitfall traps during the pilot study. However, the high human visitation rate to many field sites and presence of livestock made both these methods inappropriate for this study. We recognize that some species may have therefore gone undetected; however, this limitation would be equal for all sites. Trapping grids were 110 ¥ 20 m in size, with 10 m between traps (supplementary material Fig. S2). Traps were opened and baited late in the afternoon and closed during the day. Hair funnels were baited and left in situ for the entire survey period but were checked each afternoon and bait was replaced if it had been removed (or mostly removed) by ants. The bait used in traps and hair funnels was a mixture of 1 part rolled oats, 0.5 parts peanut butter, 0.5 parts honey and 0.3 parts sardines in oil. In addition, a piece of fresh apple and half a slice of white bread spread with plum jam were used in wire traps. Scat surveys were conducted during trapping and predator scats were also analysed for content. Upon capture animals were identified, weighed, sexed, measured (head–body length and tail length), marked (unique number or marking to site/species on the inside of
doi:10.1111/j.1442-9993.2010.02110.x
38
M . J. B R A DY ET AL.
Table 1. Total detections, number of landscapes and proportion of detections in landscape elements of mammals along gradient of matrix development intensity Total detections
Species
Number of landscapes
Pale Field Rat (Rattus tunneyi)
3
1
Bush Rat (R. fuscipes)
8
1
18
9
Brown antechinus (A. stuartii)
Yellow-footed antechinus (Antechinus flavipes)
2
2
Rufous bettong (Aepyprymnus rufescens)
8
3
Northern brown bandicoot (Isoodon macrourus)
29
11
Long-nosed bandicoot (Perameles nasuta)
13
7
4
3
31
12
128
19
Echidna (Tachyglossus aculeatus) Macropods (Macropus rufogriseus, M. giganteus, Wallabia bicolor) Common brushtail possum (Trichosurus vulpecula) †
House mouse (Mus musculus)
39
12
Dog (dingo, wild or domestic) (Canis familiaris)†‡
62
18
Cat (wild or domestic) (Felis catus)†
25
12
2
2
1
1
Black ship rat (R. rattus)† Norway rat (R. norvegicus) †
†
% Core
% Edge
% Matrix
Introduced species. ‡Canis familiaris hair detections could not distinguish between dingoes or feral dogs.
the ear with a permanent marker pen), and their reproductive and general condition assessed. Pilot study results showed this method of marking to be effective for three consecutive nights while being quick and easy to apply, causing minimal stress to the animal. Marked individuals were used to identify recaptures over the three consecutive trap-nights. Hair funnel detections were only included in site abundance calculations if the species was not detected in the site by any other method. Multiple hair funnel detections per site of the same species were only counted as single detections, as it cannot be guaranteed that the same animal did not visit multiple hair funnels. Species abundance was estimated by the minimum number known alive method (Krebs 1999). Data from the first and second samplings were combined and species abundance in the core, edge and matrix sites of each landscape, excluding recaptures across landscape elements, were combined to give abundance for that landscape.
Habitat attributes Focal patches were selected for their vegetation similarity (see Table 1 for remnant ecosystems) in order to remove the influence of local habitat attributes on species occurrences. However, habitat structure and vegetation composition were measured at every core, edge and matrix site in each landscape during summer of 2007–2008. Patch and landscape-scale variables were also measured using ArcView 9.2 and ground-truthed.
doi:10.1111/j.1442-9993.2010.02110.x
Data analysis Many species were captured in low numbers and a high proportion of zeros often limited statistical analysis. Spearman’s rank correlations were performed between the species’ abundance data, across landscape elements and landscapes of varying matrix development intensities. Detrended correspondence analysis and simple linear regressions were applied to establish responses of mammal species and species richness across modified landscape elements to matrix development intensity. Ordinations were performed in canoco 4.5 (Ter Braak & Šmilauer 2002) and all other analyses were performed in spss 16.0 (SPSS 16.0 for Windows 2008).
RESULTS Across the 57 sites, 373 individual mammals were detected within 12 small mammal species, two feral predator species and several medium-large macropodid species (pooled as ‘macropod’). The 12 small mammal species included nine native species with an adult body weight under 3 kg and three exotic rodent species all under 500 g (Table 1). Echidnas (Tachyglossus aculeatus) were also caught at three sites but were not included in native species richness calculations because they were not targeted by fauna surveys and may therefore have been missed at other sites. Only the © 2010 The Authors Journal compilation © 2010 Ecological Society of Australia
M A M M A L R E S P O N S E S TO M AT R I X I N T E N S I T Y
three exotic rodent species (but not feral predators) were included in exotic species richness calculations. No new species were detected from scats or scat content alone, so scats were only used to identify possible predation. Even in the most rural landscape, dog scats contained possum and macropod hair. Other mammals observed in the study region, most often as road kill, not included in the analyses were koalas (Phascolarctus cinereus), ringtail possums (Pseudocheirus peregrinus), and the introduced cape hare (Lepus capensis) and rabbit (Oryctolagus cuniculus).
39
occurred in relatively high abundance in every landscape. Isoodon macrourus appeared unaffected by matrix development intensity (R2 = 0.001; Fig. 2), while P. nasuta peaked in abundance at mid-intensity matrix levels of 14 420 m. Trichosurus vulpecula occurred in its highest abundance in the most urban landscape in the study (number 19; Fig. 3), which is located near the centre of the ordination triplot. Antechinus flavipes also occurred in the patch core in this landscape.
Species correlations Species and community response to matrix development intensity Individual species differed in distribution and abundance along the gradient of matrix development intensity (Fig. 2). The most abundant species captured was the common brushtail possum (Trichosurus vulpecula); present in all 19 landscapes. This was the only native species that was higher in abundance in landscapes with higher matrix development intensity (R2 = 0.341, P = 0.009; Fig. 2). The least abundant native species was the brown antechinus (A. stuartii), captured in only two landscapes (Table 1). Highest native species richness occurred at intermediate levels of matrix development intensity (11 400 m and 15 400 m). Exotic species richness in landscapes increased significantly with increasing matrix development intensity (R2 = 0.493, P < 0.001; Fig. 2). Four general responses of species to matrix development intensity were apparent; (i) species abundance decreased significantly (P < 0.02) as matrix development intensity increased (Aepyprymnus rufescens, macropods); (ii) species abundance increased significantly (P < 0.02) as matrix development intensity increased (T. vulpecula, exotic species, predators); (iii) species abundance remained stable along the gradient (I. macrourus); and (iv) species peaked in abundance at intermediate levels of matrix development intensity (Perameles nasuta). Several species’ responses, such as A. stuartii and native rat species Rattus tunneyi and R. fuscipes, were hard to gauge because of the low capture rate (Fig. 2). The detrended correspondence analysis ordered landscapes based on their species assemblage. Mus musculus and R. rattus increased in landscape abundance along the gradient and clustered at the right end of the ordination, associated with moderate-high levels of matrix development intensity (Fig. 3). Antechinus flavipes and Ae. rufescens decreased in abundance and occurred at the left end of the ordination, strongly associated with the lowest levels of matrix intensity in the most rural landscapes. Although T. vulpecula increased in abundance along the gradient, it was positioned in the middle of the ordination triplot because it © 2010 The Authors Journal compilation © 2010 Ecological Society of Australia
Native species richness in the core was significantly positively correlated with native species richness in the edge (Table 2). Specifically, T. vulpecula and A. flavipes core abundances were significantly correlated with their edge abundances. Canis familiaris abundance in the edge was negatively correlated with most native mammal species abundance, and significantly so with I. macrourus abundance in the core. Felis catus abundance in the edge was significantly positively correlated with house mouse abundance in the core. Across whole landscapes, there was a significant positive correlation between macropods and native species richness and a significant negative correlation between macropods and exotic species richness (Table 3). Macropods were also significantly negatively correlated with C. familiaris. Mus musculus was significantly positively correlated with C. familiaris. However, there was no significant correlation between C. familiaris and F. catus.
Patterns of landscape utilization Use of core, edge and matrix landscape elements varied both between and within species. Of the nine native small mammal species detected, seven species were detected using the edge and five species were detected using the matrix (Table 1). The common brushtail possum, northern brown bandicoot, long-nosed bandicoot and house mouse were detected in all landscape elements.The rufous bettong was captured in core and matrix sites, but more often in the core. Both Antechinus species were caught most often in core sites and also in edge sites, but not in the matrix. Trichosurus vulpecula was detected in similar numbers in the core and edge, consistently along the gradient, whereas A.flavipes used the core more often and declined in the use of edge sites along the gradient, with no edge use recorded in the three most intense landscapes in which it occurred (Fig. 2). Neither native rat species was detected in the matrix. The pale field rat (R. tunneyi) only occurred in core sites, while the bush rat (R. fuscipes) occurred in core and edge sites (Table 1). Age structuring of some populations across modified landscape elements may doi:10.1111/j.1442-9993.2010.02110.x
40
M . J. B R A DY ET AL.
(a)
Trichosurus vulpecula
(b)
Antechinus flavipes
20
4 R2 = 0.341
15
3 10
2
5
R2 = 0.3663
R2 = 0.1314
1 0
0 0
5000
(c)
10000
15000
20000
25000
30000
0
5000
(d)
Isoodon macrourus
10000
15000
20000
25000
30000
25000
30000
Antechinus stuartii
5 4 1
3 2
R2 = 0.0002
1 0
0
0
5000
(e)
10000
15000
20000
25000
30000
0
5000
10000
15000
20000
(f) Rattus fuscipes & Rattus tunneyi
Perameles nasuta
8
4
R.fuscipes
7 6
3
R.fuscipes
5 4
2
3
R.tunneyi
2
1
R.fuscipes
1 0
0 0
5000
(g)
10000
15000
20000
25000
0
30000
5000
(h)
Aepyprymnus rufescens
10000
15000
20000
25000
30000
25000
30000
Large macropods
5
4
4
3 3
2
2
1
R2 = 0.525
1 0
0 0
5000
10000
15000
20000
25000
Matrix development intensity (WRL)
30000
0
5000
10000
15000
20000
Matrix development intensity (WRL)
Fig. 2. Species abundance (a–l) and species richness (m–n) along gradient of matrix intensity. (✕) landscape, (䉬) core, ( ) edge, (䉭) matrix. Solid linear regression lines are for landscape abundance or richness, dashed regression lines are for matrix abundance or richness. See methods for description of matrix development intensity WRL metric.
doi:10.1111/j.1442-9993.2010.02110.x
© 2010 The Authors Journal compilation © 2010 Ecological Society of Australia
M A M M A L R E S P O N S E S TO M AT R I X I N T E N S I T Y
(i)
(j)
Rattus rattus & Rattus norvegicus
41
Mus musculus
9 8 R.rattus R.rattus
R.norvegicus
1
7 6 5 4 3 2 1 0
0 0
5000
(k)
10000
15000
20000
25000
0
30000
5000
10000
9
20000
25000
30000
20000
25000
30000
Felis catus
(l)
Canis familiaris
15000
5
8 7
4
6
3
5 4
2
3 2
1
1 0
0 0
5000
(m)
10000
15000
20000
25000
30000
0
5000
(n)
Native small species richness
10000
15000
Exotic species richness
5
R2 = 0.4933
2
4 3 R2 = 0.0644
1
2 2
R = 0.0262
1 0
0
0
5000
10000
15000
20000
25000
30000
Matrix development intensity (WRL) Fig. 2.
0
5000
10000
15000
20000
25000
30000
Matrix development intensity (WRL)
Continued
be occurring but low capture rates made it difficult to assess. The R. fuscipes individuals that were caught on the edge (adjoining a busy sealed road), were young females and visible signs suggested both had recently been in a fight, whereas five out of six individuals caught in the core were adults, including two males, the sixth being a very young female. Of the four A. flavipes individuals caught in the edge, two individuals were juvenile and two were breeding females, one of which had again visibly been in a recent fight. Exotic rats were never captured in the core; R. rattus was only captured in the edge and R. norvegicus was only captured in the matrix. © 2010 The Authors Journal compilation © 2010 Ecological Society of Australia
DISCUSSION This study showed that the intensity of matrix development in a landscape affected individual mammal species abundance within remnant habitat patches and across landscape mosaics. Response to increased matrix development intensity was similar to the general response to fragmentation in that it was highly speciesspecific (Nupp & Swihart 2000; Lampila et al. 2005). Unfortunately several species were captured in low numbers that made interpretation of response difficult. While increased trap-nights may have resulted in greater capture rates, low numbers of several species doi:10.1111/j.1442-9993.2010.02110.x
0.05 -0.17 -0.15 -0.471* -0.14 -0.10 -0.33 -0.37 0.31 0.475* 0.33 -0.09 -0.37 1.00 -0.41 0.10 0.00 0.12 -0.23 -0.15 0.43 0.33 -0.20 -0.20 0.02 0.38 1.00 -0.23 0.057 0.00 -0.033 0.11 -0.15 -0.03 -0.18 -0.20 -0.20 -0.28 1.00 -0.32 -0.37 -0.475* 0.21 -0.07 -0.12 0.16 0.22 0.07 0.580** 1.00 -0.21 -0.37 -0.37 -0.02 0.22 -0.26 -0.35 -0.30 0.36 1.00 0.14 0.14 -0.02 -0.15 -0.08 -0.26 -0.18 -0.06 1.00 -0.10 0.22 0.02 0.38 -0.11 -0.22 0.539* 1.00 0.09 0.35 0.09 -0.06 -0.44 -0.15 1.00 -0.08 -0.14 0.11 0.44 1.00
*P < 0.05, **P < 0.001.
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.
Trichosurus vulpecula core T. vulpecula edge T. vulpecula matrix Isoodon macrourus core I. macrourus edge I. macrourus matrix Antechinus flavipes core A. flavipes edge Mus musculus core M. musculus edge M. musculus matrix Aepyprymnus rufescens core A. rufescens matrix Canis familiaris edge Felis catus edge
0.726** 1.00
0.44 0.570* 1.00
-0.15 0.05 0.16 1.00
0.35 0.11 0.17 0.28 -0.02 1.00
14. 13. 12. 11. 10. 5. 2.
3.
4.
6.
7.
8.
9.
doi:10.1111/j.1442-9993.2010.02110.x
Spearman’s rank correlations of species abundance across landscape elements
may also be due to the high level of habitat loss and fragmentation in our study landscapes (Lindenmayer et al. 2000; Harrington et al. 2001). It would be a valuable, although difficult exercise to repeat this study in landscapes that had a large proportion of large patches. Species response to matrix development intensity appeared to have a complex relationship with both matrix use and matrix tolerance. For example, the rufous bettong (A. rufescens) displayed the most acute response to increased matrix development intensity, rapidly declining in abundance and disappearing from landscapes at very low levels of matrix development intensity. The rufous bettong traditionally uses open habitat areas for foraging but constructs nests in forested areas (Wallis & Green 1992), and despite often being termed an ecological specialist, can make use of cleared anthropogenic matrix with high food-plant diversity (Southwell 1987). In landscapes in which the bettong was detected in this study, it was regularly caught in the matrix. However, its ability to use the matrix per se did not translate into an ability to tolerate matrix land use intensification. Rather, the increase in matrix intensity may have had a greater impact on this species and possibly also other ‘open-area’ species,
Table 2.
Fig. 3. Detrended correspondence analysis triplot. Arrow shows direction and strength of matrix intensity on landscape order. Proximity of a species (䉭) to a landscape (䊉) indicates likelihood of the species occurring in that landscape. Species abbreviations are first letter of genus and first three letters of species as per Table 2. Landscapes are numbered in order along matrix intensity gradient; 1 being most rural, 19 being most suburban.
0.14 0.11 0.03 -0.07 -0.05 -0.22 -0.05 0.02 0.891** 0.40 0.12 -0.18 -0.18 0.25 1.00
M . J. B R A DY ET AL.
15.
42
© 2010 The Authors Journal compilation © 2010 Ecological Society of Australia
M A M M A L R E S P O N S E S TO M AT R I X I N T E N S I T Y
Table 3.
1. 2. 3. 4. 5. 6.
43
Spearman’s rank correlations of species richness, macropods and predators across whole landscapes
Macropods Canis familiaris Felis catus Total species richness Native species richness Exotic species richness
1.
2.
3.
4.
5.
6.
1.00
-0.604** 1.00
-0.235 0.45 1.00
0.2 -0.047 0.23 1.00
0.488* -0.407 0.04 0.698** 1.00
-0.476* 0.481* 0.21 0.407 -0.3 1.00
*P < 0.05, **P < 0.001.
such as macropods, than forest species, by translating into a habitat loss. Plant species richness in the matrix decreased significantly with increased matrix development intensity (Brady et al. 2009) potentially reducing the foraging value of the matrix, even if they were able to tolerate the increased human and predator presence. Matrix use in combination with intolerance of increased matrix disturbance may have directly contributed to this species decline. Other matrix users, however, such as the northern brown bandicoot (I. macrourus), were unaffected by development intensity of the matrix. This is highlighted by the fact that their use of the matrix did not decline along the gradient of development intensity. The combined effect of matrix use and tolerance of increased matrix intensity appears to have worked in this species’ favour. The yellow-footed antechinus (A. flavipes), often regarded as a forest dependent species, declined in abundance along the gradient of matrix intensity. No matrix use was observed by A. flavipes in this study; however, others (Marchesan & Carthew 2008) have observed A. flavipes using a pasture matrix, which may suggest greater tolerance of a less intense matrix, similar to the rufous bettong. Laurance (1994) found A. flavipes to strongly favour edges and disturbed forest. Once again, this could be indicative of the lowintensity matrix adjoining edges in Laurance’s (1994) study, compared with the gradient of matrix intensity in this study. In our study, abundance of A. flavipes in core sites was correlated with abundance in edge sites, edge use declined along the gradient of matrix development and no edge use was recorded in the three most intense landscapes in which it occurred. Therefore, although the species may be tolerant of some disturbance (Laurance 1994), higher rates of disturbance in the matrix and also across the edge into the patch with increased matrix intensity (Brady et al. 2009) could be causing this species to retreat from edges further into the core of patches, contributing to the overall decline of A. flavipes with increasing matrix development intensity. Antechinus flavipes are believed to exist as metapopulations in fragmented landscapes with females able to persist in very small fragments, and males moving © 2010 The Authors Journal compilation © 2010 Ecological Society of Australia
large distances across the matrix to disperse and breed with resident females (Marchesan & Carthew 2008). This marked male dispersal strategy also exhibited by other Antechinus species may aid persistence in fragmented agricultural landscapes with low matrix development intensity (Lindenmayer et al. 1999; Marchesan & Carthew 2004). However, as matrix development intensity increases, this strategy might be increasingly risky. Increased road density, predator density and acute changes in habitat structure in the matrix and across edges (Brady et al. 2009) would likely result in lower landscape functional connectivity, with fewer individuals able or willing to move between isolated habitat patches to maintain the species (Stokes et al. 2004; Hodgson et al. 2007). Common brushtail possums were the only native species to increase in landscapes with higher matrix intensity. This is consistent with existing literature (Kerle 1984) and attributed to the possum’s adaptability in both behaviour and diet, including nesting in human structures, also observed in this study. Similar to A. flavipes, possum abundance in the core was correlated with abundance in the edge, and further, highest possum abundance in the core occurred when they were also present in the matrix, in highly urbanized landscapes. Harper et al. (2008) found that brushtail possums increased in abundance in remnants with the availability of resources in the adjacent urban matrix. Therefore, the possum’s ability to not only tolerate but also make use of the urban matrix (supplementing habitat, Dunning et al. 1992), may allow the species to increase in abundance in modified landscapes. Laurance (1991) showed matrix abundance to be the best ecological correlate of extinction proneness for mammals in fragmented landscapes, above species traits, such as body size, natural rarity or dietary specialization. Bentley et al. (2000) suggested, however, that floristic habitat specialization was a better predictor of species vulnerability to fragmentation than matrix abundance, although the two are correlated (Laurance 1991). We suggest that utility of both these traits, and possibly others, are dependent on the nature of the landscape matrix. For example, matrix abundance or tolerance, might be more impordoi:10.1111/j.1442-9993.2010.02110.x
44
M . J. B R A DY ET AL.
tant for certain matrix types or at certain levels of matrix development intensity. At a lower level of intensification, many species can safely use the matrix (e.g. the rufous bettong; and voles, Microtus ochrogaster, Cook et al. 2004); hence, matrix use in those landscapes may be less deterministic for the species survival than, for example, changes in habitat structure or floristics, important to habitat specialists. At higher intensities of matrix development, higher levels of human disturbance (Brady et al. 2009) and/or an increasing resistance of the matrix to species movement may all make the matrix and consequently a species tolerance of it, more important. In conclusion, this study confirms that patches in modified landscapes cannot be viewed in isolation from the attributes of the surrounding matrix, but that the nature of the entire landscape mosaic is important to the mammal community. The intensity of development in the matrix affects distribution and abundance of most, but not all, mammal species persisting in human-modified landscapes in this region, independent of patch size or the amount of habitat remaining in the landscape, although the response is highly species-specific. Matrix development intensity may impact species in multiple and complex ways, including through affecting the habitat and dispersal values of the matrix itself. An ability to use the matrix per se, however, may not translate into an ability to persist in a landscape with increased modification if that modification reduces the habitat or movement value of the matrix. ACKNOWLEDGEMENTS Mammal surveys were conducted under UQ AEC permit NRSM/266/06/NRSM/CSIRO and a Queensland Government EPA scientific purposes permit WISP03682706. B. Triggs analysed hair wafers and scats. M. Mayhew and G. Johnston provided invaluable field assistance. Funding was provided by a UQPRS and CSIRO Postgraduate Top-Up Scholarship. REFERENCES Arnold G. W., Steven D. E., Weeldenburg J. R. & Smith E. A. (1993) Influences of remnant size, spacing pattern and connectivity on population boundaries and demography in Euros Macropus robustus living in a fragmented landscape. Biol. Conserv. 64, 219–30. Barlow J., Gardner T. A., Araujo I. S. et al. (2007) Quantifying the biodiversity value of tropical primary, secondary, and plantation forests. Proc. Natl. Acad. Sci. U.S.A. 104, 18555– 60. Bellamy P. E., Shore R. F., Ardeshir D., Treweek J. R. & Sparks T. H. (2000) Road verges as habitat for small mammals in Britain. Mamm. Rev. 30, 131–9.
doi:10.1111/j.1442-9993.2010.02110.x
Bentley J. M., Catterall C. P. & Smith G. C. (2000) Effects of fragmentation of araucarian vine forest on small mammal communities. Austral Ecol. 14, 1075–87. Brady M. J., McAlpine C. A., Miller C. J., Possingham H. P. & Baxter G. S. (2009) Habitat attributes of landscape mosaics along a gradient of matrix development intensity: matrix management matters. Landscape Ecol. 24, 879–91. Cook W. M., Anderson R. M. & Schweiger E. W. (2004) Is the matrix really inhospitable? Vole runway distribution in an experimentally fragmented landscape. Oikos 104, 5–14. Cox M. P., Dickman C. R. & Hunter J. (2004) Effects of rainforest fragmentation on non-flying mammals of the Eastern Dorrigo Plateau, Australia. Biol. Conserv. 115, 175–89. Debinski D. M. (2006) Forest fragmentation and matrix effects: the matrix does matter. J. Biogeogr. 33, 1791–2. Diffendorfer J. E., Gaines M. S. & Holt R. D. (1995) Habitat fragmentation and movements of three small mammals (Sigmodon, Microtus, and Peromyscus). Ecology 76, 827. Dunning J. B., Danielson B. J. & Pulliam H. R. (1992) Ecological processes that affect populations in complex landscapes. Oikos 65, 169–75. FitzGibbon S. I., Putland D. A. & Goldizen A. W. (2007) The importance of functional connectivity in the conservation of a ground-dwelling mammal in an urban Australian landscape. Landscape Ecol. 22, 1513–25. Forman R. T. T. & Collinge S. K. (1997) Nature conserved in changing landscapes with and without spatial planning. Landsc. Urban Plan. 37, 129–35. Forman R. T. T. & Alexander L. E. (1998) Roads and their major ecological effects. Annu. Rev. Ecol. Syst. 29, 207–31. Franklin J. F. & Lindenmayer D. B. (2009) Importance of matrix habitats in maintaining biological diversity. Proc. Natl. Acad. Sci. U.S.A. 106, 349–50. Gascon C., Lovejoy T. E., Bierregaard R. O. et al. (1999) Matrix habitat and species richness in tropical forest remnants. Biol. Conserv. 91, 223–9. Harper K. A., MacDonald S. E., Burton P. J. et al. (2005) Edge influence on forest structure and composition in fragmented landscapes. Conserv. Biol. 19, 768–82. Harper M. J., McCarthy M. A. & Van Der Ree R. (2008) Resources at the landscape scale influence possum abundance. Austral Ecol. 33, 243–52. Harrington G. N., Freeman A. N. D. & Crome F. H. J. (2001) The effects of fragmentation of an Australian tropical rain forest on populations and assemblages of small mammals. J. Trop. Ecol. 17, 225–40. Haynes K. J. & Cronin J.T. (2004) Confounding of patch quality and matrix effects in herbivore movement studies. Landscape Ecol. 19, 119–24. Hodgson P., French K. & Major R. E. (2007) Avian movement across abrupt ecological edges: differential responses to housing density in an urban matrix. Landsc. Urban Plan. 79, 266–72. Johnston G. F. (2008) Wildlife vehicle collision GIS mapping and modelling. Dissertation Faculty of Engineering and Surveying 102 pp. University of Southern Queensland, Toowoomba. Jules E. S. & Shahani P. (2003) A broader ecological context to habitat fragmentation: why matrix habitat is more important than we thought. J.Veg. Sci. 14, 459–64. Kerle J. (1984) Variation in the ecology of Trichosurus: its adaptive significance. In: Possums and Gliders (eds A. Smith & I. Hume) pp. 115–128. Surrey Beatty and Sons, Sydney. Krebs C. J. (1999) Ecological Methodology. Addison-Welsley Educational Publishers Inc., Menlo Park.
© 2010 The Authors Journal compilation © 2010 Ecological Society of Australia
M A M M A L R E S P O N S E S TO M AT R I X I N T E N S I T Y
Lampila P., Monkkonen M. & Desrochers A. (2005) Demographic responses by birds to forest fragmentation. Conserv. Biol. 19, 1537–46. Laurance W. F. (1991) Ecological correlates of extinction proneness in Australian Tropical Rain Forest Mammals. Conserv. Biol. 5, 79–88. Laurance W. F. (1994) Rain-forest fragmentation and the structure of small mammal communities in tropical Queensland. Biol. Conserv. 69, 23–32. Laurance W. F. (2009) Roads to ruin. New Sci. 24–5. Laws R. J. & Goldizen A. W. (2003) Nocturnal home ranges and social interactions of the Brush-tailed Rock-wallaby Petrogale penicillata at Hurdle Creek, Queensland. Aust. Mamm. 25, 169–76. Lindenmayer D. B., Cunningham R. B. & Pope M. L. (1999) A large-scale ‘experiment’ to examine the effects of landscape context and habitat fragmentation on mammals. Biol. Conserv. 88, 387–403. Lindenmayer D. B., McCarthy M. A., Parris K. M. & Pope M. L. (2000) Habitat fragmentation, landscape context, and mammalian assemblages in southeastern Australia. J. Mamm. 81, 787–97. McKinney M. L. (2002) Urbanization, biodiversity and conservation. BioScience 52, 883–90. Manning D., Galbraith R. & Clewett J. (2003) Crows Nest Land Management Manual. Crows Nest Shire Council, Crows Nest. Marchesan D. & Carthew S. M. (2004) Autecology of the yellowfooted antechinus (Antechinus flavipes) in a fragmented landscape in southern Australia. Wildl. Res. 31, 273–82. Marchesan D. & Carthew S. M. (2008) Use of space by the yellow-footed antechinus, Antechinus flavipes, in a fragmented landscape in South Australia. Landscape Ecol. 23, 741–52. Moffatt S. F., McLachlan S. M. & Kenkel N. C. (2004) Impacts of land use on riparian forest along an urban – rural gradient in southern Manitoba. Plant Ecol. 174, 119–35. Murray J.V., Low Choy S., McAlpine C. A., Possingham H. P. & Goldizen A. W. (2008) The importance of ecological scale for wildlife conservation in naturally fragmented environments: a case study of the brush-tailed rock-wallaby (Petrogale penicillata). Biol. Conserv. 141, 7–22. Nupp T. E. & Swihart R. K. (2000) Landscape-level correlates of small-mammal assemblages in forest fragments of farmland. J. Mamm. 81, 512–26. Perault D. R. & Lomolino M. V. (2000) Corridors and mammal community structure across a fragmented, old growth forest landscape. Ecol. Monogr. 70, 401–22. Pope M. L., Lindenmayer D. B. & Cunningham R. B. (2004) Patch use by the greater glider (Petauroides volans) in a fragmented forest ecosystem. I. Home range size and movements. Wildl. Res. 31, 559–68. Pope S. E., Fahrig L. & Merriam N. G. (2000) Landscape complementation and metapopulation effects on leopard frog populations. Ecology 81, 2498–508. Queensland Government (2005) South East Queensland Regional Plan 2005–2026. Office of Urban Management, Department of Local Government, Planning, Sport and Recreation, Queensland Government, Brisbane. Ricketts T. H. (2001) The matrix matters: effective isolation in fragmented landscapes. Am. Nat. 158, 87–99.
© 2010 The Authors Journal compilation © 2010 Ecological Society of Australia
45
Rico A., Kindlmann P. & Sedlacek F. (2007) Road crossing in bank voles and yellow-necked mice. Acta Theriol. 52, 85–94. Sattler P. & Williams R. (1999) The Conservation Status of Queensland’s Bioregional Ecosystems. Environmental Protection Agency, Brisbane. Southwell C. (1987) Macropod studies at Wallaby Reek II. Density and distribution of Macropod species in relation to environmental variables. Aust.Wildl. Res. 14, 15–33. SPSS 16.0 for Windows (2008) SPSS 16.0 for Windows. SPSS Inc., Chicago. Stokes V. L., Pech R. P., Banks P. B. & Arthur A. D. (2004) Foraging behaviour and habitat use by Antechinus flavipes and Sminthopsis murina (Marsupialia: Dasyuridae) in response to predation risk in eucalypt woodland. Biol. Conserv. 117, 331–42. Ter Braak C. J. F. & Šmilauer P. (2002) CANOCO Reference Manual and CanoDraw forWindows User’s Guide: Software for Canonical Community Ordination (Version 4.5). Microcomputer Power, Ithaca. Tubelis D. P., Lindenmayer D. B. & Cowling A. (2004) Novel patch-matrix interactions: patch width influences matrix use by birds. Oikos 107, 634–44. Vieira M. V., Olifiers N., Delciellos A. C. et al. (2009) Land use vs. fragment size and isolation as determinants of small mammal composition and richness in Atlantic Forest remnants. Biol. Conserv. 142, 1191–200 (in press). doi: 10.1016/j.biocon.2009.02.006 Wallis I. R. & Green B. (1992) Seasonal field energetics of the Rufous Rat-kangaroo (Aepyprymnus rufescens). Aust. J. Zool. 40, 279–90. Watson J. E. M., Whittaker R. J. & Freudenberger D. (2005) Bird community responses to habitat fragmentation: how consistent are they across landscapes? J. Biogeogr. 32, 1353–70. Wiens J. A. (1997) The emerging role of patchiness in conservation biology. In: The Ecological Basis of Conservation: Heterogeneity, Ecosystems and Biodiversity (eds S. Pickett, R. Ostfeld, M. Schachak & G. Likens) pp. 93–107. Chapman and Hall, New York. Withey J. & Marzluff J. (2009) Multi-scale use of lands providing anthropogenic resources by American Crows in an urbanizing landscape. Landscape Ecol. 24, 281–93.
SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Fig. S1. Study area showing landscapes (circles) numbered in order of their position on the matrix development intensity gradient; 1 being most rural, 19 being most suburban. Fig. S2. Mammal sampling design. This trap configuration was chosen in order to have the greatest spread of traps through a site, within the constraints of small patch size, but also considering matrix heterogeneity along the gradient. Table S1. Landscape selection criteria.
doi:10.1111/j.1442-9993.2010.02110.x
