home > data and maps > derived data > tassel-eared squirrel

ForestERA Data Layer Overview: Predicted Tassel-eared Squirrel Density and Recruitment

Description

The Tassel-eared Squirrel (Sciurus aberti) is a ponderosa pine obligate species, endemic to the southwestern United States (Keith 1965). It is a key species in these systems, where it facilitates essential symbiotic interactions of mycorrhizal fungi with ponderosa pine through consumption of fruiting bodies and dispersal of spores (States and Gaud 1997, States and Wettstein 1998). It also serves as an important prey for the southwestern subspecies of the northern goshawk (Accipiter gentilis; Reynolds et al. 1992, Beier and Drennan 1997), which is federally listed as threatened. Previous research has suggested that squirrel population parameters are highly dependent on forest structure, particularly canopy cover and ponderosa pine basal area (Ratcliff et al. 1975, Patton 1984, Patton et al. 1985, Dodd et al. 1998, unpublished; Dodd 2003). For these reasons, the tassel-eared squirrel is considered an important management indicator species in southwestern forests. The layers presented here represent predicted Tassel-eared Squirrel density (squirrels / ha) and recruitment (juveniles / ha) at a resolution of 90m (0.8 ha or 2 acres).

Purpose

These data layers were created as part of the ForestERA project to support landscape-scale forest restoration planning efforts by a broad group of stakeholders including federal and state agencies, academic institutions, and non-governmental entities. These data are intended for regional analyses over spatial extents on the order of tens to hundreds of thousands of acres, and were not developed for use at finer spatial scales, although they may be useful for some applications at finer scales.

Development

These layers were developed from field data collected by Norris Dodd of the Arizona Game and Fish Department on nine 280 ha study sites spread across the Western Mogollon Rim between 2000 and 2003. The sites were oriented along a gradient of habitat quality so that differences in forest structure could be assessed in relation to squirrel population parameters (Dodd 2003, Dodd et al. unpublished). On each site, squirrel density was estimated using a feeding sign index techniques and recruitment were estimated using a trapping grid (Dodd et al. 1998).

In our modeling approach, we used Norris Dodd’s estimates of squirrel density and recruitment from 2 sampling plots, each 24 ha in size, from each of his study areas (n = 18 sampling plots). Using linear regression, we linked these to layers representing forest structural attributes developed from remotely sensed imagery by the ForestERA project (Hampton et al. 2003, Xu et al. 2005). We developed a-priori hypotheses about the influence of forest structure on squirrel density and recruitment and used the relationships between the remotely-sensed data layers and squirrel population parameters to develop a set of candidate models predicting squirrel density and recruitment. We used Akaike’s Information Criterion (AIC, Akaike 1973) and an information-theoretic approach to assess the strength of these models and choose the most strongly supported model from the set of candidate models (Burnham and Anderson 2002, Anderson et al. 2000).
Our methodology resulted in a final set of 17 candidate models to be assessed for each population parameter. From our candidate set of models, the best predictor of squirrel density was the model that included local basal area (m2 / ha) as the only variable. Unlike squirrel density, there was no single model that predicted squirrel recruitment with overwhelming support. A total of six models had ?AICc values within approximately 2 of the best model. All of these models included basal area (m2 / ha) and a variation on percent canopy cover over a larger spatial extent. For more information on model development and selection please see Prather et al. (2005). The layers available for download on the ForestERA web site were built using the following formulas;

Squirrel Density = -0.1815 + (0.0206 * basal area)

Squirrel Recruitment = -0.1710 + (0.0076 * basal area) + (0.0007 * canopy cover variant) + (0.0001 * interaction effect).

In the recruitment model, canopy cover variant was created by determining the number of cells with cover greater than 50% over a 160 ha extent. In ArcGIS this involved converting the canopy cover layer to a binary layer (cells with cover > 50% reclassified as 1, other cells reclassified as 0). Then, a neighborhood focalsum operation was used on this layer (focalsum of a circle with a radius of 8 pixels).

Accuracy Assessment

We used Akaike’s Information Criterion adjusted for small sample size (AICc; Akaike 1973, Burnham and Anderson 2002) to determine the amount of support for each model from the candidate set. The models presented here were the most strongly supported from among their representative candidate sets (see Prather et al. 2005). Adjusted r² values indicated that the models were doing a good job of explaining variance in squirrel density (r² = 0.84) and recruitment (r² = 0.72).

To do a rigorous accuracy assessment for these layers would require the collection of a great deal of ground data on squirrel density and recruitment. However, we were able to do a limited assessment using data collected by Norris Dodd on seven independent study plots in the mid 1990’s (Dodd et al. 1998). We obtained the boundaries for the plots and estimated squirrel density and recruitment on each plot using our models. Using linear regression, we compared these estimates with estimates of squirrel density and recruitment obtained during the field study. This analysis indicated that there were statistically significant relationships between the two estimates for both squirrel density (F = 7.6, P = 0.04, r² = 0.60, m = 0.84, d.f. = 6) and recruitment (F = 9.3, P = 0.03, r² = 0.65, m = 0.47, d.f. = 6).

Sources of errors

We believe this layer does a relatively good job at predicting relative Tassel-eared Squirrel density and recruitment. However, we note that the field data were collected over a relatively short (4 year) time period. Squirrel populations can vary dramatically from year to year depending on weather conditions and food resources. We do not suggest that these layers can accurately predict actual values for squirrel density or recruitment in a given year. However, the relative patterns of high and low density and recruitment across the landscape should be reasonably robust.

We also note that the predicted values for density and recruitment were based on total basal area. In areas where other types of trees are mixed with ponderosa pine, actual values may be lower than expected. Likewise in areas with high basal area and canopy cover, but few mature trees (e.g., doghair thickets) the model is likely to overpredict density and recruitment.

Recommendations

We recommend that this layer be used at a minimum resolution of 90m (0.8 ha or 2 acres) for purposes of analysis and display. However, ForestERA data layers were not designed for analyses at the level of individual pixels, and uncertainty in the data will generally decline over greater spatial extents. Therefore, we recommend using larger analysis units, with groupings of at least 50 cells (40 ha or 100 acres). Finally, we reiterate that ForestERA data layers were developed for the purpose of regional landscape-level planning, and we suggest that the analyses be applied over spatial extents of tens to hundreds of thousands of acres. We recognize, however, that this layer may be useful for analyses over smaller spatial extents depending on the type and purpose of those analyses.

Literature Cited

Akaike, H. 1973. Information theory as an extension of the maximum likelihood principle. Pp. 267-281 in B. N. Petrov and F. Csaksi, editors. 2nd International Symposium on Information Theory. Akademiai Kiado, Budapest, Hungary.

Anderson, D. R., K. P. Burnham, and W. L. Thompson. 2000. Null hypothesis testing: problems, prevalence, and an alternative. Journal of Wildlife Management 64: 912-923.

Burnham, K. P. and D. R. Anderson. 2002. Model Selection and Multi-model Inference: a Practical Information-Theoretic Approach (2nd Edition). Springer-Verlag, New York, USA.

Beier, P., and J. E. Drennan. 1997. Forest structure and prey abundance in foraging areas of northern goshawks. Ecological Applications 7: 564-571.

Dodd, N. L. 2003. Landscape-scale habitat relationships to tassel-eared squirrel population dynamics in north-central Arizona. Technical Guidance Bulletin 6, Arizona Game and Fish Department, Phoenix, Arizona, USA.

Dodd, N. L., R. E. Schwiensburg, and S. Boe. Unpublished. Landscape-scale forest habitat relationships to tassel-eared squirrel populations: forest restoration implications. Arizona Game and Fish Department.

Dodd, N. L., S. S. Rosenstock, C. R. Miller, and R. E. Schweinsburg. 1998. Tassel-eared squirrel population dynamics in Arizona: index techniques and relationships to habitat condition. Arizona Game and Fish Department Technical Report 27, Phoenix, Arizona, USA.

Hampton, H. M., Y. Xu, J. W. Prather, E. A. Aumack, B. G. Dickson, M. M. Howe, and T. D. Sisk. 2003. Spatial tools for guiding forest restoration and fuel reduction efforts. Proceedings of the 23rd Annual Environmental Systems Research Institute (ESRI) International User Conference. Available at http://gis.esri.com/library/userconf/proc03/p0679.pdf

Keith, J. O. 1965. The Abert squirrel and its dependence on ponderosa pine. Ecology 46: 150-163.

Patton, D. R. 1984. A model to evaluate Abert’s squirrel habitat in uneven-aged ponderosa pine. Wildlife Society Bulletin 12: 408-413.

Patton, D. R. , R. L. Wadleigh, and H. G. Hudak. 1985. The effects of timber harvest on the Kaibab squirrel. Journal of Wildlife Management 49: 14-19.

Prather, J. W., N. L. Dodd, B. G. Dickson, H. M. Hampton, Y. Xu, E. N. Aumack, and T. D. Sisk. (2005). Developing spatially-explicit models of Tassel-eared Squirrel density and recruitment: an information-theoretic approach. Journal of Wildlife Management.

Ratcliff, T. D., D. R. Patton, and P. F. Ffolliott. 1975. Ponderosa pine basal area and the Kaibab squirrel. Journal of Forestry 73: 284-286.

Reynolds, R.T., R.T. Graham, M.H. Reiser, R.L. Bassett, P.L. Kennedy, D.A. Boyce, Jr., G. Goodwin, R. Smith, and E.L. Fisher. 1992. Management recommendations for the northern goshawk in the southwestern United States. U.S. Forest Service General Technical Report RM-GTR-217, Fort Collins, Colorado, USA.

States, J. S., and W. S. Gaud. 1997. Ecology of hypogeous fungi associated with ponderosa pine. I. Patterns of distribution and scorocarp production in some Arizona forests. Mycologia 89: 712-721.

States, J. S., and P. J. Wettstein. 1998. Food habitats and evolutionary relationships of the tassel-eared squirrel (Sciurus aberti). Pages 185-194 in M. A. Steele, J. F. Merritt, and D. A. Zegers, editors. Ecology and evolutionary biology of tree squirrels. Virginia Museum of Natural History Special Publication No. 6. Martinsville, Virginia, USA.

Xu, Y., J. W. Prather, H. M. Hampton, E. N. Aumack, B. G. Dickson, and T. D. Sisk. 2005. Advanced exploratory data analysis for mapping regional canopy cover. Photogrammetric Engineering and Remote Sensing XX: xx-xx.

Last updated February 23, 2005