You are here:Empirical Support

Empirical Support

Several studies have evaluated and demonstrated the ability of the new habitat availability (reachability) metrics implemented in Conefor (IIC, PC) to explain or predict ecological processes related to landscape connectivity, including species distributions, colonization events, seed dispersal or genetic diversity patterns at the landscape scale.

Most of these studies have been performed by other research groups and institutions different from the one that developed Conefor and the metrics there implemented, which provides an independent assessment and empirical support to these quantitative developments and metrics.

Some of these studies have evaluated the ability of IIC or PC to explain ecological processes and have compared it with the performance of other existing connectivity metrics. When this has been done, IIC or PC have been shown to outperform the other analyzed connectivity metrics by presenting a stronger relationship with the analyzed ecological processes and empirical data.

The main results on this empirical support and validation can be found in the following peer-reviewed scientific articles:

  • Pereira, M., Segurado, P, Neves, N. 2011. Using spatial network structure in landscape management and planning: a case study with pond turtles. Landscape and Urban Planning 100: 67-76.

Pereira et al. (2011) analyzed a pond system used by an endangered species, the European pond turtle (Emys orbicularis), at a coastal region under strong agriculture intensification in southwestern Portugal. Their study took into account different types of species-specific behavioural and ecological information: a surface of resistance to movement through the landscape matrix, the maximum traveled distance, observed pond presence/absence data, and an empirical habitat suitability model based on field sampling.

The authors evaluated a set of graph metrics in their ability to relate to the actual species pond occupancy patterns in the study area. They found that, among all the analyzed metrics, PC was the one that allowed a better discrimination of the occupied and unoccupied sites by adult and juvenile turtles (see the results for the three metrics in figure 6 in that article, plus the lack of relationship with turtle presence that is reported for the betweenness centrality in page 74). PC also outperformed the results obtained when considering only the information provided by the habitat suitability model. In words of the authors, “the difference in dPC values between occupied and unoccupied ponds was slightly more pronounced than habitat suitability values alone. This was especially true for the presence-absence of hatchlings and juvenile individuals”. They concluded that “among the different estimated parameters, this [PC] is possibly the most relevant for management purposes”.

  • Awade, M., Boscolo, D., Metzger, J.P. 2012. Using binary and probabilistic habitat availability indices derived from graph theory to model bird occurrence in fragmented forests. Landscape Ecology 27: 185-198.

Awade et al. (2012) used playback techniques to empirically determine the inter-patch movements and occurrence patterns of a rainforest insectivorous bird (Pyriglena leucoptera) in three fragmented Atlantic forest landscapes in Brazil. They considered several models as candidates for predicting the observed presence/absence patterns of the species in the study area, including variables at the patch and landscape levels and different ways to characterize the links between patches.

Awade et al. (2012) found that the Equivalent Connected Area (ECA) for the PC index (ECA(PC)) was the best predictor of the analysed bird occurrence patterns, and that the probabilistic connection model in ECA(PC) outperformed in this case the results provided by the binary connection model in the ECA(IIC) index. In their words, “the single landscape-level model including ECA(PC) was found to be the best”, “all three best-supported models included ECA(PC)”, and “we advise the use of probabilistic metrics, such as ECA(PC), when inferring the effects of habitat availability on the occurrence of a certain species”. They stated that “models should include at least one landscape-level habitat availability predictor and that a probabilistic measure of connectivity is mandatory to obtain good predictions”. They concluded that “habitat availability is an essential factor determining species occurrence in fragmented landscapes” and that “evaluation of the abovementioned habitat availability aspects is strongly recommended to properly guide management decisions”.

  • Rodríguez-Pérez, J., García, D., Martínez, D. 2014. Spatial networks of fleshy-fruited trees drive the flow of avian seed dispersal through a landscape. Functional Ecology 28: 990-998.

Rodríguez-Pérez et al. (2014) analyzed avian seed dispersal of fleshy-fruited trees in a secondary forest of the northern Iberian Peninsula. In this forest they set up an 18 hectare plot where they recorded the standing fruit crop of all trees and the abundance of seeds deposited by birds below the canopy of each tree.

They aimed to infer the effects of connectivity among trees on seed deposition patterns in each tree resulting from seed dispersal by frugivorous birds. They calculated the contribution of each tree to network connectivity for frugivorous birds through different connectivity metrics, which included one only accounting for interpatch connectivity (number of links, NL) but also the habitat availability metrics IIC and PC that account both for intrapatch and interpatch connectivity.

Rodríguez-Pérez et al. (2014) found that “from the univariate models of fruit crop, NL, dIIC and dPC, we found that tree contribution to dPC best predicted (lowest AIC) seed abundance (Table 1)”. They concluded that “trees with higher dPC had higher seed abundance below them (Fig. 3), and this connectivity metric (PC) was the most important factor explaining seed abundance (see Table S5, Supporting information)”. PC was therefore able to capture factors driving seed dispersal and the abundance of seed deposition by frugivorous birds. Trees with the highest dPC values accumulated larger seed clumps under their canopies, demonstrating agreement between the network connectivity as characterized by PC and the actual process of seed dispersal flow by birds.

  • Neel, M.C. 2008. Patch connectivity and genetic diversity conservation in the federally endangered and narrowly endemic plant species Astragalus albens (Fabaceae). Biological Conservation 141: 938-955.

Neel (2008) analysed the patterns of genetic diversity of the endangered plant species Astragalus albens, a geographically and edaphically narrowly distributed herb that is found in San Bernardino Mountains (California, USA). Thirty occurrences (patches) of Astragalus albens were sampled for genetic diversity using allozymes, and the non-parametric Gamma statistic was used to correlate the genetic diversity of the species in each patch with a set of other patch characteristics and connectivity metrics.

Among other results, Neel (2008) found that the IIC values calculated for each patch (dIIC) at different threshold distances presented significant correlations with four of the seven analysed genetic diversity statistics. The dIIC values were those presenting the highest correlations with the genetic diversity patterns among the metrics considered (this was the case for six of the seven genetic diversity statistics). In this case IIC performed better than PC, which only correlated significantly with one of the genetic diversity statistics.

  • Ribeiro, R., Carretero, M.A, Sillero, N., Alarcos, G., Ortiz-Santaliestra, M.,Lizana, M., Llorente, G.A. 2011. The pond network: can structural connectivity reflect on (amphibian) biodiversity patterns? Landscape Ecology 26: 673-682.

Ribeiro et al. (2011) sampled the amphibian species present in 51 ponds in two areas within the province of Salamanca (Central Western Spain). They used the IIC index to calculate, for different threshold distances, the importance of each pond for habitat availability and connectivity in the landscape. They related the IIC pond importance values with the amphibian species richness in each pond. To test if the connectivity effects (as quantified by IIC) would remain after considering the individual pond traits relevant for amphibians, they applied logistic regression for each species presence/absence with the IIC pond importance and four pond characteristics (area, depth, hydroperiod, and aquatic vegetation) as independent variables.

They found that the IIC values were significantly related to the amphibian species richness in the ponds, and that the network characteristics evaluated through IIC were an important factor for the presence of some species by determining the patterns of amphibian colonization. In their words, “the influence of a [IIC] connectivity factor determining the presence of amphibian species in the ponds is striking with a total of nine final models (five in Puerto Seguro and four in Vilvestre) including at least one [distance] threshold of pond importance”. They concluded that “the relation of the landscape connectivity with species richness is unambiguous in both areas indicating that the accumulation of amphibian species in the ponds is, even if partially, related to the spatial position of the pond in relation to the other ponds nearby”.

  • Ishiyama, N., Akasaka, T., Nakamura, F. 2014. Mobility-dependent response of aquatic animal species richness to a wetland network in an agricultural landscape. Aquatic Sciences 76:437-449.

Ishiyama et al. (2014) evaluated the relationships between species richness of aquatic animals and wetland connectivity in the Tokachi plain in central Hokkaido (northern Japan). Their study area consisted of a set of 50 wetlands, the half of which was surveyed for the presence of fish and aquatic insects. Five variables related to habitat quality (water quality and habitat structure) were assessed in each of the wetlands. IIC was used to quantify the connectivity of each wetland considering both species movement through watercourses (for fish and insects) and overland (for flying insects).

They found that the species richness of highly-mobile fishes and insects was significantly and positively related to the IIC values (dIIC). In particular, the species richness of high-swimming ability fish was only significantly correlated with dIIC; none of the five variables related to local water quality or habitat structure was significant. The species richness of high-flying ability insects was also significantly and positively correlated with dIIC. One of the five local habitat quality variables was also significant for this latter type of species, although with a lower correlation coefficient (in absolute value) than dIIC.

  • Zozaya, E., Brotons, L., Saura, S. 2012. Recent fire history and connectivity patterns determine bird species distribution dynamics in landscapes dominated by land abandonment. Landscape Ecology 27: 171-184.

Zozaya et al. (2012) evaluated the role of forest fires occurred in recent decades in Catalonia (NE Spain) as drivers of the pattern of expansion of early-successional, open-habitat bird species. 44 large forest fires occurring between 2000 and 2005 (with sizes ranging from 50 to more than 6000 hectares) were analysed and surveyed in the field for the presence of six bird species with preference for open habitats in Mediterranean landscapes. Several scenarios of potential colonizer sources were assessed: open habitats created by previous recent fires, shrublands and farmlands. They used the flux fraction of the PC index to estimate potential species colonization dynamics on the selected fires, and differentiated the nodes comprising the sources of colonisers and the nodes corresponding to the new suitable habitat patches originated by the wildfires. They evaluated the capacity of that PC flux metric to explain the empirically observed species occurrences in the studied sites by using generalized linear mixed models.

They found that the occurrence and colonization patterns of the focal species on the newly burnt sites were significantly related to the potential flux as estimated by PC. This occurred for five of the six surveyed open-habitat bird species and for all the scenarios of potential colonizer sources (although with a stronger signal for the scenario that considered previous wildfires). The authors concluded that “fires occurring in the last decades are acting as sources of immigrants to the new suitable habitats appearing in the landscape. Overall, the probability of colonisation of a recently burnt area was greatest in those sites well connected by dispersal to other previously burnt areas”.

An additional study of interest in this respect is the following:

  • Visconti, P., Elkin, C. 2009. Using connectivity metrics in conservation planning: when does habitat quality matter? Diversity and Distributions 15: 602-612.

Visconti and Elkin (2009) assessed five connectivity metrics for their ability to predict the contribution of each patch to metapopulation viability. The probability of connectivity (PC) was one of the five evaluated metrics. The study was based in simulating the dynamics and viability of species occupying the landscapes using a metapopulation model linked to continuous time logistic population growth models. The authors compared the results from that model with the patch importance rankings derived from each connectivity metric.

They concluded that from all the five evaluated metrics “only the metapopulation capacity and the PC [probability of connectivity] index were reasonably successful in predicting patch value in over-dispersed, heterogeneous landscapes”. They also concluded that “to assess the persistence of a species in a landscape, [interpatch] connectivity is a necessary but by itself insufficient factor to consider. It is also necessary to consider the other components of metapopulation dynamics, which are taken into account in metapopulation capacity and PC [probability of connectivity] index”.

This is not an empirical validation study since it is based in landscape simulations and modelling, but provides another kind of support to the PC metric by comparing it with a much more complex, biologically detailed and data hungry spatially explicit population model.