Measuring stream network connectivity: how close is close enough?

Kevin Collier
Ude Shankar
Peter Smith

The success of stream restoration projects may depend on understanding the links between landscape patchiness and aquatic insect dispersal.

New Zealand’s landscape has become highly fragmented over the last century with the clearance of large tracts of forest and the conversion of land to pasture. The remaining patches of forest are often isolated from each other, and populations of some sensitive terrestrial animals have suffered as a result. For example, native forest birds must now cross an inhospitable matrix of developed land to move between remaining forest patches. In other words, the connectivity of their habitat has been lost.

Although not widely recognised, loss of connectivity also affects streams and rivers. Sensitive adult aquatic insects may not be able to fly overland between catchments. Within streams, barriers such as culverts or degraded water may limit dispersal by invertebrate larvae and fish.

Understanding the implications of this loss of connectivity between different parts of a stream network and adjacent catchments is important in planning stream restoration in which the aim is to re-establish sensitive aquatic species. This is because the success of natural recolonisation depends partly on whether patches of suitable habitat are large enough and spaced close enough together between the area being restored and the colonist populations, which are often holding out in headwater streams surrounded by remnant native forest.

GIS study

Novel methods have been developed to analyse spatial patterns of landscapes fragmented by landuse change. The same approach could be useful for measuring connectivity of stream riparian habitats. We are conducting a study to analyse the spatial structure of forested patches near stream boundaries. We are using vegetation data from the New Zealand Land Cover Database, and stream networks from the NZ TOPO datasets available from Land Information New Zealand. This analysis was carried out using a Geographic Information System (GIS) linked to a software package for landscape patch analysis. We wanted to see whether measures of connectivity, when applied to riparian areas, were related to the genetic differences found in a stream caddisfly from a previous study (see “Dispersal of a North Island caddisfly” in an earlier issue of Water & Atmosphere).

We measure genetic differences as genetic distance, which is estimated from gene frequencies using DNA or allozymes. These distances indicate how related different populations are, and provide some clues to how often populations mix between sites, at least over long timescales. In the earlier study, we collected caddisflies for genetic analysis from a range of stream sites. Using GIS, we mapped all areas of forest and scrub within 200 m of a stream edge at these sites. Using the resulting network of forest/scrub riparian strips (see example) we calculated several measures – or metrics – of connectivity among sites. We made comparisons within regions because our previous genetic work demonstrated that historical landmass changes still influence current genetic patterns for the same species occurring in different regions. Several landscape metrics showed interesting relationships with genetic distance. The graph shows how low genetic distance is associated with higher levels of connectedness. In other words, genetically similar populations tend to occur at pairs of sites that have higher connectivity. These initial results support the hypothesis that the degree of riparian connectivity may influence the permeability of the landscape to adult aquatic insects, and therefore the interchange between populations along and between stream networks.

Integrated approach needed

This work is in its preliminary stage. In the future we plan to develop and apply measures suitable for analysing connectivity within streams. We will also look in more detail at the quality of riparian vegetation connecting stream segments. In addition, the spatial resolution of the data will be improved by using better aerial photographs. Spatial studies such as these are useful for providing a basis for planning the arrangement of stream restoration works at the landscape scale. However, simply getting to a site is only half the recolonisation battle. Freshwater organisms must be able to complete their life-cycles at a site in order to establish self-sustaining populations. This means that habitat at restored sites must be suitable for colonists. So, understanding the likelihood of restoration sites being recolonised by sensitive species depends on:

• developing ecologically appropriate measures of habitat connectivity;
• integrating these measures with knowledge of species' habitat requirements for successful life-cycle completion.

Kevin Collier is based at NIWA in Hamilton, Ude Shankar is at NIWA in Christchurch, and Peter Smith is with NIWA in Wellington.