Habitat area, quality and connectivity: striking the balance for efficient conservation


Correspondence author. E-mail: jenny.hodgson@york.ac.uk


1. Population viability can depend on habitat area, habitat quality, the spatial arrangement of habitats (aggregations and connections) and the properties of the intervening non-breeding (matrix) land. Hodgson et al. [Journal of Applied Ecology46 (2009) 964] and Doerr, Barrett & Doerr (Journal of Applied Ecology, 2011) disagree on the relative importance of these landscape attributes in enabling species to persist and change their distributions in response to climate change.

2. A brief review of published evidence suggests that variations in habitat area and quality have bigger effects than variations in spatial arrangement of habitats or properties of the intervening land. Even if structural features in the matrix have a measurable effect on dispersal rates, this does not necessarily lead to significant increases in population viability.

3. Large and high-quality habitats provide source populations and locations for colonisation, so they are the main determinants of the capacity of species to shift their distributions in response to climate change because populations must be established successively in each new region.

4.Synthesis and applications. Retaining as much high quality natural and semi-natural habitat as possible should remain the key focus for conservation, especially during a period of climate change.


Hodgson et al. (2009) noted that habitat area, quality, and aggregation were key components of landscape-scale conservation, and that prioritising habitat area and quality would be a robust way to facilitate connectivity and the persistence of biodiversity in the face of climate change. Doerr, Barrett & Doerr (2011) suggest that structural features to facilitate dispersal in non-breeding habitat are valuable targets in themselves, and can be measured with certainty equal to that of area, quality, and aggregation. They further suggest that we misunderstood the term ‘connectivity conservation’. In this reply to the Forum by Doerr, Barrett & Doerr in this issue, we begin by summarising where we are in agreement. We then present our case that Doerr, Barrett & Doerr, in focussing narrowly on non-habitat structural features, have ignored the bigger picture of relative priorities for conservation under climate change.

Hodgson et al. (2009) and Doerr, Barrett & Doerr (2011) present different ways of approaching the same overall goal; limiting extinctions and maintaining functioning ecosystems. Because of the complexities of ecology and human impacts in different regions, no single prescription for conservation will work everywhere. So, we agree with Doerr, Barrett & Doerr (2011) that conservation strategies need to be tailored to the landscape. We also fundamentally agree about which factors affect the persistence of populations and metapopulations (Doerr, Barrett & Doerr’s Fig. 1 essentially expands our boxes for ‘dispersal mechanism’ and ‘potential for barriers or conduits in non habitat’). However, we disagree about the relative importance of the various factors: habitat area, habitat quality, the spatial arrangement of habitats (aggregations and connections) and the properties of the intervening non-breeding (matrix) land. Hence, we provide here evidence in support of our assessment of the relative importance of these landscape attributes.

Relative importance of landscape attributes

Habitat area

Successful reproduction is confined to natural and semi-natural habitats for the majority of terrestrial species. In the context of climate change, multi-generational range shifts are facilitated initially by large habitat areas that support large source populations, by substantial intervening habitat areas to support breeding and generate propagules en route, and by high habitat availability within target regions to ensure eventual persistence (Table 1). Maximising area is also likely to increase diversity via increased habitat heterogeneity. Contrary to Doerr, Barrett & Doerr (2011), we contend that there is generally a large amount of natural or semi-natural habitat with insufficient protection. Tropical forest, for example, is converted to other land uses at around 0·5% annually (FAO 2005). It is usually much more effective and cheaper to retain what is still present than to attempt to recreate it.

Table 1.   Brief overview of evidence for the importance of different landscape-scale factors in conservation. Priority has been given to review papers and meta-analyses, papers that compare more than one factor, and papers that relate factors specifically to range expansion
Landscape attributeSummary of evidenceSources
Habitat areaEastern US trees’ range expansions dependent on source populationsIverson, Schwartz & Prasad (2004)
Patch area has consistently bigger effect (up to 100 times greater F ratio in ANOVA) than connectivity measures on species richness and abundance of plants and butterflies in fragmented grasslands in southern GermanyBruckmann, Krauss & Steffan-Dewenter (2010)
Plant and butterfly species richness of calcareous grasslands in northern Germany is affected by patch area, and not significantly by isolation or surrounding landscape heterogeneityKrauss, Steffan-Dewenter & Tscharntke (2003);Krauss et al. (2004)
Simulated and observed expansion rates of butterfly Pararge aegeria >40% slower in landscape containing 24% less woodland coverHill et al. (2001)
Hesperia comma butterfly range expansion rates in UK strongly related to habitat quantityWilson, Davies & Thomas (2010)
Habitat quality*Positive effects of quality on occupancy for plants, butterflies, moths, other insects, amphibians, birds, small mammals, primates and carnivoresReviewed by Mortelliti, Amori & Boitani (2010)
Correct vegetation management can increase butterfly densities 10- to 100-foldThomas, Simcox & Hovestadt (2010)
Macaw (Ara and Orthopsittaca spp) density in Amazon varies >10-fold with respect to indicators of human disturbanceKarubian et al. (2005)
Sea otters (Enhydra lutris nereis) range expansion dependent on population growth rateTinker, Doak & Estes (2008)
Quality the dominant factor affecting butterfly and moth abundance and diversity in a Finnish landscapePöyry et al. (2009)
Spatial pattern 1: aggregation (aka spatial autocorrelation)Isolation negatively affects colonisation rate and occupancyReviewed in Hanski (1999: Chapter 9)
Negative effect of isolation on patch occupancy in three English butterflies (effect of habitat quality is 2–3 times larger).Thomas et al. (2001)
Positive effects of aggregation on flower visitation and seed set in pan-European study of 10 plant species (although patch area effect is bigger)Dauber et al. (2010)
Aggregation measures have positive effects on species richness and abundance of butterflies (and to lesser extent plants) in fragmented grasslands, though not as large as the effects of areaBruckmann, Krauss & Steffan-Dewenter (2010)
Spatial pattern 2: habitat connections (corridors and stepping stones)Average 1·6-fold increases in exchange rates between patches, based on systematic review in which the corridor is the same type of habitat as the connected patchesEycott et al. (2009)
Modest positive effect of corridors on dispersal rate: standardised effect size <0·5 if study controlled for distance between patches. Smaller effect for created corridors than natural corridors.Gilbert-Norton et al. (2010)
Substantial evidence that hedgerows are used by animals as corridors between woods, but very little on the quantitative effect this has on population viabilityDavies & Pullin (2007)
Corridors of habitat relatively unimportant for spread of an invasive plantAndrew & Ustin (2010)
Matrix (non-breeding habitat) qualityOn average, 1·3-fold increases in exchange rates between patches, based on systematic review of movements between patches separated by more versus less benign matrixEycott et al. (2009)
Meta-analysis found an effect of the type of matrix surrounding habitat patches in most studies, but such effects were smaller than patch size or isolation effects, and strongly species-specific (whenever species were compared).Prevedello & Vieira (2010)
Ranking of alternative landscape configurations was the same with a simple habitat-patch model as one with complex matrix-dependent movement behaviourJepsen et al. (2005)

Habitat quality

Quality improves persistence by increasing population growth, resulting in larger propagule numbers, increased likelihood of colonisation, and higher population growth rates following colonisation. Many aspects of habitat quality are, as Doerr, Barrett & Doerr (2011) suggest, species-specific and difficult to control. However, some aspects of quality affect many species similarly, such as nutrient pollution, the spread of invasive species or major habitat disturbances. Much of the world’s biomes are now partly or substantially altered by, for example, selective logging, partial fertilisation of grasslands, drainage of wetlands, and elimination of fire or large mammals. Preventing further degradation and increasing the quality of already-degraded areas can generate extremely large differences in population densities (e.g. Table 1) and, therefore, colonisation and population growth, including range expansion.

Spatial arrangement of habitats

The locations of remaining habitat fragments in landscapes are known to be important to long-term population persistence (Hanski & Ovaskainen 2000), but the size of this effect is smaller than the effects of quantity and quality (Table 1), principally because the production of new individuals takes place within habitats, regardless of their location (Ovaskainen 2002). Habitat aggregation generally increases the chance of a propagule landing in suitable habitat, and therefore of a patch of habitat being colonised/occupied, but it can only compensate a little for deficiencies in quantity and quality (Table 1). Under climate change, the benefit of aggregation may be less certain because aggregating remaining habitat within a few regions may leave dispersal barriers that will eventually need to be bridged. When suitable habitat is a very low proportion of the landscape, there may be a trade-off between maximising aggregation and reducing the largest dispersal barriers. There is evidence that corridors may increase dispersal rates between patches to some extent (Table 1), but an increase in dispersal per se is not direct evidence of an increase in population viability.

Non-breeding habitat (or the matrix)

Interventions in the matrix could contribute to population viability by overcoming behavioural barriers to crossing certain boundaries or land-cover types, or by reducing dispersal mortality. Managing the matrix can result in modest increases in dispersal between nearby habitat patches (by ca. 25%, Table 1). However, we could find no robust evidence that matrix condition alters long-distance, multi-generational range changes. The likelihood of leaving an individual’s natal habitat patch may increase if the intervening matrix is favourable, but movements typically become faster and straighter when an individual is in a hostile matrix environment, and this leads to much longer realised dispersal distances (Ovaskainen et al. 2008; Zheng, Pennanen & Ovaskainen 2009). The longest dispersal distances are the most important for maintaining genetic diversity and for range expansions under climate change (Neubert & Caswell 2000; Trakhtenbrot et al. 2005). We also note that a whole different set of considerations apply to wind- and water-dispersed species, and to the four kingdoms other than animals. The suggestion of Doerr, Barrett & Doerr (2011) that dispersal behaviour could be universal is centred on an example that involved a small number of well-studied and related species with high cognitive abilities.

Uncertainty and robustness

Doerr, Barrett & Doerr (2011) argue that structural connectivity can be measured with reasonable certainty. The issue to us is not the development of a repeatable metric, but whether variation in such a metric is a good predictor of multi-generational range expansions for a wide range of different taxa. The effect of the matrix on long-distance dispersal and colonisation is virtually unknown; and we already know that short-distance matrix effects are species specific (Eycott et al. 2009; Prevedello & Vieira 2010). Doerr, Barrett & Doerr (2011) reasonably counter that measurement of habitat quality is an issue, and also species-specific, but it is much easier to investigate, for example by relating the population densities of individual species or broader measures of diversity to environmental and biotic variables. We strongly disagree with the Doerr, Barrett & Doerr (2011) view of area, quality and connectivity that ‘all provide the same degree of conservation certainty because their benefits depend on the interactions between them’. It is clear to us (e.g. Table 1) that different variables that interact in a given model do not all have the same effect sizes or associated uncertainties. Uncertainty about functional connectivity is higher than any of the above-mentioned factors because it adds uncertainty about dispersal distances and behaviour on top of uncertainty about what constitutes habitat quality. Crucially, though, this uncertainty does not negate the large effects of habitat area and quality over large numbers of species and landscapes, because only metapopulations close to their extinction threshold are substantially limited by connectivity (Hodgson et al. 2009: Fig. 1).

This leads to the issue of robustness. Understanding and modelling the detailed behavioural responses of individuals to multiple landscape elements is an interesting area of research. However, if such details are to be incorporated within multi-generational distribution models, we would argue that other details (e.g. habitat changes, local adaptation, species interactions, evolution of dispersal) are equally relevant, and that they all have associated uncertainties. The robustness of such complicated modelling will almost necessarily be low. The intention of the original Hodgson et al. (2009) argument was to identify robust, easy to understand and easy to implement conservation strategies.

What role for connectivity conservation?

It is possible that everybody understands ‘connectivity conservation’ to mean a holistic, landscape-to-continental scale conservation method that accounts for all important factors (including habitat quality, quantity, spatial habitat arrangements, the character of the intervening landscape), and the consequences of these for persistence and resilience of multiple species, in a location-specific manner (Doerr, Barrett & Doerr 2011). If so, our fears about an over-emphasis on connectivity may be unfounded. But if ‘connectivity conservation’ is all this, then the term is actually an unneeded alias for ‘biodiversity conservation’. Such a re-branding may be useful to stimulate investment in conservation, but we should still be aware of the potential for confusion it creates. Our original concern was not that there were problems with every activity labelled ‘connectivity conservation’, but that the review of Heller & Zavaleta (2009) found that ‘increasing connectivity’ was the predominant proposed solution for conservation under climate change.

We think that considerations of connectivity have an important place in conservation planning (e.g. Moilanen et al. 2005; Moilanen & Wintle 2007; Moilanen, Wilson & Possingham 2009; Carroll, Dunk & Moilanen 2010). Recent awareness of connectivity has undoubtedly been helpful to conservation: it has freed conservationists from focusing too narrowly on individual protected areas, and has brought landscape-scale and spatial considerations into conservation. But our concern was how connectivity is often seen as ‘the solution’ when the fundamental problem is an inadequate quantity of high-quality habitats. It is fortunate that many of the existing connectivity conservation programmes such as the Great Eastern Ranges (GER) initiative are heavily focused on increasing and consolidating natural habitat area (the first three of five GER goals; Mackey, Watson & Worboys 2010). This serves to reinforce the impression that connectivity conservation may be a new name for the business of conservation that has long recognised the role of habitat quantity, quality, and connectivity (e.g. Diamond 1975).

Based on the weight of current evidence (Table 1) and our original arguments (Hodgson et al. 2009), we remain of the opinion that maintaining (and where feasible restoring) large areas of environmentally diverse, high-quality (low human impact) breeding habitats should be the primary focus of conservation when resources limit the range of conservation actions that can take place. Compared to this, structural features of the matrix which are not breeding habitat for many or any species are a minor consideration.


We thank Amy Eycott, Jeremy Thomas and Kevin Watts for providing material/permission to cite work in press.