To print this page, select File-> Print from your browser menu.

Why a Picture is Worth a Thousand Data Points

Diana Stralberg

Habitat for the marsh-dwelling Black Rail was delineated in a PRBO spatial model. Photo by Peter LaTourrette.
Many of the important decisions made by land managers and conservation planners involve questions of space: Which areas to protect or restore? Where to plan trails? Where are the critical habitats for a species? Bird survey data collected by PRBO, documenting the abundance and diversity of birds found at particular places, are key for informing such decisions. But field biologists can't be everywhere, and inevitably there are spatial gaps in the available data.

Spatial predictive models help fill these gaps. Among the products of spatial models are maps, and these particular maps let birds and other organisms tell us, through data collected by field biologists, which areas are most suitable for them. In this way, large volumes of data about birds and their habitats can be distilled into simple maps that give managers, planners, and biologists a bird's-eye view of the landscape.

As with any model, the quality of the end product depends on the quality of the information that was used to develop the model. If the pieces used to build the model are of high quality (bird and vegetaion data that are current and accurate, for example), so will be the model. Fortunately, PRBO's long history of data collection has provided spatial modelers with large volumes of reliable data to choose from, especially through the recently established California Avian Data Center online resource (

With the development of geographic information systems (GIS) and improvements in computer processing capabilities, spatial predictive models have become easier and faster to implement. Increased availability of aerial photos, satellite imagery of vegetation, and other environmental data has also improved the quality of the predictions that are possible.

Despite the wide array of modeling techniques and data sets, the basic spatial modeling premise is generally as follows. Location data for the subject of interest (bird abundance, perhaps) are related to environmental data (such as vegetation and topography) for those same locations and used to develop a mathematical equation describing the relationship. Those relationships can then be expressed as maps, which are, in essence, spatial predictions.

Spatial models can range from very large-scale (such as North America) to very small (such as a county park). Their resolution (see Glossary) can also vary, much like that of a digital photograph.
Figure 1. For six tidal marsh study sites in San Francisco Bay, PRBO used data from locations we had surveyed (open circles) to map the density of birds such as the Black Rail (above). Where the model shows areas of high density (deeper colors), Black Rail habitat is most suitable. Click on red type for larger view of Figure 1.

For local management questions, fine-scale models are usually most appropriate (but are labor-intensive!). As part of the San Francisco Bay Integrated Regional Wetlands Monitoring project (, we developed models of bird density at a one-meter resolution for six tidal marsh study sites that we had previously surveyed (Figure 1). Our goal was to compare bird use of newly restored and reference tidal marshes across a range of different salinities and elevations and with respect to specific marsh plants and distances to levees and tidal channels.

Among the bird species included was the California Black Rail, a state-listed threatened species. With the majority of its population concentrated in northern San Francisco Bay tidal marshes, the Black Rail is an important indicator of restoration success. Spatial models can be used to predict habitat suitability for such species in marsh areas that have not yet been surveyed. The final maps from this modeling effort were used to compare variability within and across sites and to identify areas of greatest suitability for tidal marsh birds.

Larger scale and resolution

For other management questions, fine-scale detail is neither cost-effective nor particularly helpful. At the scale of a national park or forest, planning and management activities are usually conducted with respect to more general vegetation types or wildlife habitats.
Figure 2. A map of habitat suitability for management-sensitive birds in the Golden Gate National Recreation Area and Point Reyes National Seashore. The darker the shading, the more such birds an area is likely to hold. Click on red type for larger view of Figure 2.

At the request of the Golden Gate National Recreation Area, PRBO developed a set of spatial distribution models for bird species of management interest. The goal was to help the park use information about birds to prioritize habitats, plan trails, and address other issues relevant to their general management plan.
One species sensitive to park management decisions and included in the GGNRA model is the Western Meadowlark, vulnerable because it nests on the ground. Photo by Peter LaTourrette.

We took advantage of PRBO's long-term comprehensive bird survey data and the park's detailed vegetation map to develop spatial predictive models at a 30-meter spatial resolution (Figure 2). We focused on species considered particularly sensitive to the park's primary management activities—birds that nest low to the ground, like the Western Meadowlark; those sensitive to disturbance, like the Swainson's Thrush; and others.

The long-term partnership between PRBO and the National Park Service fostered an interactive process and heightened this project's success.

Other areas of study—for example the potential effects of climate change on species' distributions—require still broader perspectives. PRBO has developed current and future distribution models for a number of species, the results of which will be highlighted in a future Observer issue (see note below).

Above all, spatial models help people visualize the landscape in new ways, from the perspective of a bird or any other plant or animal of interest, beyond what data alone can provide. Ultimately, that is why a picture is often worth a thousand words (or data points).

PRBO is grateful to photographer Peter LaTourrette for generously sharing his beautiful images of birds (go to with the Observer by special arrangement.