Trailblazers in the Forest: Response of Endangered Mt. Graham Red Squirrels to Severe Insect Infestation

Item Type text; Electronic Thesis

Authors Zugmeyer, Claire Ann

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 23/09/2021 14:33:31

Link to Item http://hdl.handle.net/10150/193336 TRAILBLAZERS IN THE FOREST: RESPONSE OF ENDANGERED MT. GRAHAM RED SQUIRRELS TO SEVERE INSECT INFESTATION

by

Claire Ann Zugmeyer

______

A Thesis Submitted to the faculty of the

SCHOOL OF NATURAL RESOURCES

In Partial fulfillment of the Requirements For the Degree of

MASTERS OF SCIENCE WITH A MAJOR IN WILDLIFE AND FISHERIES SCIENCES

In the Graduate College

THE UNIVERSITY OF ARIZONA

2007 2

STATEMENT BY AUTHOR

This thesis has been submitted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library.

Brief quotations from this thesis are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author.

SIGNED: ______Claire Zugmeyer______

APPROVAL BY THESIS DIRECTOR

This thesis has been approved on the date shown below:

______April 27, 2007_____ John L. Koprowski Date Professor of Wildlife and Fisheries Science 3

ACKNOWLEDGMENTS

I thank my family for their love and support throughout my thesis research. I am indebted to my advisor John Koprowski, whose guidance, encouragement, enthusiasm, and friendship were critical components in the completion of this research. I also thank my committee members, Courtney Conway and R. William Mannan, for their helpful advice and direction. I would like to thank Sadie Bertelsen, Debbie Buecher, Seafha Blount, Kathi Borgmann, Nichole Cudworth, Sandy Doumas, Andrew Edelman, Aaron Flesch, Erika Geiger, Vicki Garcia, Vicki Greer, Nate Gwinn, Sky Jacobs, Julia Jolley, Sarah King, Kate Leonard, Andrea Litt, Melissa Merrick, Karen Munroe, Danielle O’Dell, Mark Ogonowski, Geoff Palmer, Bret Pasch, Kitty Pokorny, Nicolas Ramos- Lara, Mags Rheude, Bob Steidl, Eric Wallace, Marissa Watson, Bruce Weise, Ryan Wilson, David Wood and past University of Arizona Red Squirrel Monitoring Program employees for valuable assistance and support throughout the project. Field efforts were conducted under permits from the USDA Forest Service, Arizona Game and Fish Department, US Fish and Wildlife Service, and the University of Arizona’s Institutional Animal Care and Use Committee. Special thanks to Bill Van Pelt, Arizona Game and Fish Department, for his efforts to procure funds. Funding and in-kind support for this research were provided by Arizona Game and Fish Department, U. S. Department of Agriculture, U. S. Forest Service, University of Arizona’s Office of the Vice President for Research, Steward Observatory, Arizona Agricultural Experiment Station, and T&E Incorporated. 4

TABLE OF CONTENTS

LIST OF TABLES………………………………………………………………………...7

LIST OF FIGURES……………………………………………………………………….8

ABSTRACT………………………………………………………………………………9

CHAPTER 1: INTRODUCTION………………………………………………………..10

CHAPTER 2: PRESENT STUDY………………………………………………………12

REFERENCES…………………………………………………………………………..14

APPENDIX A: UNALTERED HABITAT SELECTION AFTER SEVERE INSECT INFESTATION: CONCERNS FOR FOREST-DEPENDENT SPECIES ……………..16

Abstract…………………………………………………………………….……16

Introduction……………………………………………………………………...17

Methods………………………………………………………………………….21

Study area……………………………………………………………….21

Midden occupancy………………………………………………………21

Midden characteristics…………………………………………………..22

Level of insect damage…………………………………………………..23

Statistical analyses………………………………………………………23

Results…………………………………………………………………………...25

Midden occupancy………………………………………………………25 5

TABLE OF CONTENTS – Continued

Midden characteristics…………………………………………………...25

Midden microclimates……………………………………………………27

Level of insect damage…………………………………………………..28

Discussion………………………………………………………………………..29

References………………………………………………………………………..33

APPENDIX B: SEVERELY INSECT-DAMAGED FOREST MAY FUNCTION AS AN ECOLOGICAL TRAP FOR RED SQUIRRELS………………………….…………….42

Abstract…………………………………………………………………………..42

Introduction…………………………………………………………….………..43

Methods………………………………………………………………….….…..45

Model organism and system…………………………………………….45

Study area and trapping…………………………………………………46

Habitat preference………………………………………………………47

Home range and space use………………………………………………47

Body mass and demography……………………………………………..49

Statistical analyses……………………………………………………….50

Results……………………………………………………………………………51

Habitat is available in healthy and insect-damaged forest ……………..51

Habitat selection in areas with less damage……………………………..52

Home ranges enlarge…………………………………………………….54

Body mass did not differ…………………………………………………55 6

TABLE OF CONTENTS – Continued

Demographic differences exist…………………………………………...55

Discussion………………………………………………………………………..56

References………………………………………………………………………..61 7

LIST OF TABLES

TABLE 1. Mean (± SE) physical and vegetation measurements for random locations (n = 26), and Mt. Graham red squirrel (Tamiasciurus hudsonicus grahamensis) unoccupied middens (n = 28), and occupied middens (n = 18) in the insect-damaged forest………..39

TABLE 2. Results of step-wise logistic regression models to assess odds of a) a site being a Mt. Graham red squirrel (Tamiasciurus hudsonicus grahamensis) midden (n = 46) or random location (n = 26), or b) a midden being occupied (n = 18) or unoccupied (n = 28) in the insect-damaged forest……………………………………………………….40

TABLE 3. Seasonal fluctuations in mean 95% range and 50% core (± SE) for Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis) in insect-damaged and healthy forest………………………………………………………………………….….68

TABLE 4. Mean 95% range and 50% core (± SE) for male and female Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis) in insect-damaged and healthy forest in the Pinaleño Mountains…………………………………………………………….…….69 8

LIST OF FIGURES

FIGURE 1. Mean (± SE) a) maximum, b) minimum, and c) maximum daily range of ground surface temperature from September 2004 to August 2005 for random locations and unoccupied and occupied Mt. Graham red squirrel (Tamiasciurus hudsonicus grahamensis) middens in insect-damaged forest……...………………………………....41

FIGURE 2. Seasonal fluctuations in mean percent pixels (± SE) classified as a) dead trees and b) open areas, within ranges, cores, and random areas for Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis) in insect-damaged forest in the Pinaleño Mountains……………………………………………………………..……….70

FIGURE 3. Seasonal fluctuations in mean number of pixels (± SE) classified as a) live and b) dead trees, within ranges, cores, and random areas for Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis) in insect-damaged forest in the Pinaleño Mountains………………………………………………………………………..……....71

FIGURE 4. Mean body mass (± SE) for adult male and female Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis) in insect-damaged and healthy forest in the Pinaleño Mountains……………………………………………………………………...72

FIGURE 5. Proportion of Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis) surviving since first captured in insect-damaged and healthy forests in the Pinaleño Mountains…...…………………………………………………………………73 9

ABSTRACT

I examined habitat selection of middens within insect-damaged forest and compared home range and survival for Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis) in insect-damaged and healthy forest. Squirrels used habitat in areas with <69% tree mortality. Basal area, canopy cover, and log volume were greater at middens than random locations. Within midden sites, only greater basal area of live trees distinguished occupied sites from unoccupied sites. Surface temperature at occupied middens tended to be cooler than unoccupied middens. Squirrels living in insect-damaged forest had larger home ranges than in healthy forest. Squirrel body mass and reproductive condition did not differ between forest types, suggesting that insect- damaged forest provided adequate resources. However, squirrels inhabiting insect- damaged forest experienced lower survivorship and 50% fewer potential reproductive events than squirrels in healthy forest, implicating presence of an ecological trap.

Preservation of remaining healthy forest is a priority for management of this endangered species. 10

CHAPTER 1: INTRODUCTION

Disturbance creates temporal and spatial heterogeneity of resources that can cause short- and long-term changes to habitat and disrupt interactions between individuals and resources (Sousa 1984). Severe disturbance may dramatically alter habitat structure, causing reduced reproductive success or site abandonment (Jones et al. 2001, Penteriani et al. 2002). Rates and impacts of natural disturbances are predicted to increase with current trends in climate change (Dale et al. 2000), thus highlighting the need to understand species’ response to disturbance.

Responses of small mammals to forest disturbance from insect infestation are largely unknown but some studies suggest decreased abundance of individuals after insect damage, particularly for seed-eating forest obligates (Matsuoka et al. 2001,

Koprowski et al. 2005, McDonough and Rexstad 2005). Tree squirrels can be considered indicators of forest health (Carey 2000) due to their dependence on mature trees for food, nests, and cover (Gurnell 1987, Steele and Koprowski 2001), thus red squirrels

(Tamiasciurus hudsonicus) can be used as a model organism to determine the effects of insect infestation. In the western portion of their range, red squirrels are territorial and defend conspicuous larderhoards or middens (Steele 1998). Therefore, occupancy of an area by red squirrels is easily determined. Furthermore, if insect infestation severely damages a forest, red squirrels will likely demonstrate a demographic and behavioral response due to their dependence on conifer trees as a primary food source (Gurnell

1987). The Mt. Graham red squirrel (T. h. grahamensis), a federally listed subspecies of red squirrel (USFWS 1987, Sullivan and Yates 1995) occupies high elevation spruce-fir 11

and mixed-conifer forests of the Pinaleño Mountains in southeastern Arizona

(Hoffmeister 1986). Recent recolonization of insect-damaged forest provides an opportunity to examine response of red squirrels to insect infestation. I examined habitat selection, home range, body mass, and demography of Mt. Graham red squirrels inhabiting insect-damaged forest. 12

CHAPTER 2: PRESENT STUDY

The methods, results, and conclusions of this study are presented in detail in the papers appended to this thesis. The following is a summary of the major conclusions in these papers.

I examined habitat selection within insect-damaged forests by comparing physical and vegetative characteristics at 1) occupied middens, 2) a random subset of unoccupied historical middens, and 3) random locations. Mt. Graham red squirrels established middens in areas with <64% tree mortality. Greater basal area, canopy cover, and log volume best distinguished middens from random locations. Only greater basal area of live trees significantly distinguished occupied middens from unoccupied middens.

Midden surface temperature depended on surrounding vegetation structure and tended to be cooler at occupied middens than unoccupied middens. These habitat components are congruent with features found at middens prior to the insect infestation as well as those found in other parts of the red squirrel distribution, indicating that red squirrels do not alter habitat selection after insect disturbance.

I used radiotelemetric techniques and live-trapping to examine home range, body mass, and demographics of Mt. Graham red squirrels inhabiting insect-damaged and healthy forest. Squirrels living within insect-damaged forest had larger home ranges than squirrels within healthy forest. Females had larger home ranges than males in healthy forest, but I detected no difference between sexes in insect-damaged forest. Squirrel body mass, reproductive condition, and litter size did not differ between forest types, suggesting insect-damaged forest provided adequate resources. However, squirrels 13

inhabiting insect-damaged forest experienced lower survivorship and 50% fewer potential reproductive events than squirrels in healthy forest. Although Mt. Graham red squirrels demonstrate equal-preference for habitat within insect-damaged forests and reproduce as well as individuals in healthy forest, poor survivorship and reduced potential to reproduce suggest insect-damaged forest may function as an ecological trap and has potential to threaten the persistence of this isolated population. Therefore, preservation of remaining healthy forest is a priority for management of this endangered species. 14

REFERENCES

Carey, A. B. 2000. Effects of new forest management strategies on squirrel populations. Ecological Applications 10:248-257.

Dale, V. H., L. A. Joyce, S. McNulty, and R. P. Neilson. 2000. The interplay between climate change, forests, and disturbances. The Science of the Total Environment 262:201-204.

Gurnell, J. 1987. The natural history of squirrels. Facts on File Inc., New York, New York, USA.

Hoffmeister, D. F. 1986. Mammals of Arizona. University of Arizona Press, Tucson, Arizona, USA.

Jones, J., R. D. DeBruyn, J. J. Barg, and R. J. Robertson. 2001. Assessing the effects of natural disturbance on a neotropical migrant songbird. Ecology 82:2628-2635.

Koprowski, J. L., M. I. Alanen, and A. M. Lynch. 2005. Nowhere to run and nowhere to hide: Response of endemic Mt. Graham red squirrels to catastrophic forest damage. Biological Conservation 126:491-498.

Matsuoka, S. M., C. M. Handel, and D. R. Ruthrauff. 2001. Densities of breeding birds and changes in vegetation in an Alaskan boreal forest following a massive disturbance by spruce beetles. Canadian Journal of Zoology 79:1678-1690.

McDonough, T. J., and E. Rexstad. 2005. Short-term demographic response of the red- backed vole to spruce beetle infestations in Alaska. Journal of Wildlife Management 69:246-254.

Penteriani, V., M. Mathiaut, and G. Boisson. 2002. Immediate species responses to catastrophic natural disturbances: windthrow effects on density, productivity, nesting stand choice, and fidelity in northern goshawks (Accipiter gentilis). The Auk 119:1132-1137.

Sousa, W. P. 1984. The role of disturbance in natural communities. Annual Review of Ecology and Systematics 15:353-391.

Steele, M. A. 1998. Tamiasciurus hudsonicus. Mammalian Species 586:1-9.

Steele, M. A., and J. L. Koprowski. 2001. North American tree squirrels. Smithsonian Institution Press, Washington D. C., USA. 15

Sullivan, R. M., and T. L. Yates. 1995. Population genetics and conservation biology of relict populations of red squirrels. Pages 193-208 in C. A. Istock, and R .S. Hoffmann, editor. Storm over a mountain island: Conservation biology and the Mt. Graham affair. University of Arizona Press, Tucson, Arizona, USA.

U. S. Fish and Wildlife Service (USFWS). 1987. Endangered and threatened wildlife and plants; determination of endangered status for the Mount Graham Red Squirrel. Federal Register 52:20994-20999. 16

APPENDIX A: UNALTERED HABITAT SELECTION AFTER SEVERE INSECT INFESTATION: CONCERNS FOR FOREST-DEPENDENT SPECIES

Abstract:

Disturbance creates spatial and temporal heterogeneity of resources. Severe

disturbance may alter habitat structure and change habitat selection. Habitat selection is

critical for completion of life history stages and is likely influenced by different

environmental cues. Habitat selection for sites to cache and preserve food is important

for survival of larderhoarding species. The recent recolonization of an insect-damaged

forest by the endangered Mt. Graham red squirrel (Tamiasciurus hudsonicus

grahamensis) provided an opportunity to examine habitat selection for midden (cache) sites following disturbance. From September 2003 to December 2005, we examined surface temperature and physical and vegetative characteristics associated with random locations and midden sites (unoccupied and occupied) in insect-damaged forests. Red squirrels do not appear to be changing habitat selection in disturbed areas, as results are consistent with other studies of red squirrel midden sites. However, differences between occupied and historical sites that are now unoccupied suggest the severity of insect infestation tolerated by red squirrels. Occupied middens were found in areas with <64% tree mortality, characterized by high basal area of live trees, and had cooler surface temperatures during snow-free months. Forest areas receiving higher damage would likely not represent habitat, threatening the persistence of an isolated population with little opportunity to shift to undamaged forest. Although a few individuals may find habitat, insect-damaged forests may function as ecological traps if fitness declines. High 17

rates of habitat loss, combined with predictions of increased rate and intensity of disturbance due to climate change, highlight the importance of documenting the flexibility of habitat selection and response to disturbance.

Introduction:

Disturbance, both natural and anthropogenic, creates temporal and spatial heterogeneity of resources that can cause short- and long-term changes to habitat and disrupt interactions between individuals and resources (Sousa 1984, Bengtsson et al.

2000). Severe disturbance may dramatically alter habitat structure, causing reduced reproductive success or site abandonment (Jones et al. 2001, Penteriani et al. 2002). In addition, rapid disturbance can function as an ‘ecological trap’ if cues used for habitat selection no longer correlate with habitat quality (Schlaepfer et al. 2002, Robertson and

Hutto 2006). Rates and impacts of natural disturbances are predicted to increase with current trends in climate change (Dale et al. 2000), thus highlighting the need to understand species’ response to disturbance and document potential changes in habitat selection in response to the disturbed environment.

Habitat is any area that provides resources and conditions that promote survival and reproduction of an individual (Hall et al. 1997). Therefore, habitat selection is critical for completion of life history stages. Habitat selection can be influenced by environmental cues such as structural complexity (McNett and Rypstra 2000), chemical properties (Lecchini et al. 2005), or the presence or absence of other species (Williams and Batzli 1979, Werner et al. 1983). Although understanding the actual process of 18

habitat selection may not be feasible, examining areas used by individuals may suggest environmental variables used in the selection process. Furthermore, identification of important microhabitat features is necessary for effective conservation efforts, particularly if habitat is limited (Frenandez and Palomares 2000, Oppel et al. 2004).

Habitat selection of sites to cache and hoard food for later consumption has evolved in many birds, mammals, and arthropods (Smith and Reichman 1984, Vander

Wall 1990). Loss of cached items due to decomposition or pilferage is detrimental to hoarding animals. Consequently, food-hoarding animals evolved many food-handling behaviors, including selection of the cache site (Vander Wall 1990). Cache-site selection helps to preserve and protect stored food and can vary with species needs. For example, cache-sites can be dry (burrows of steppe rodents; Formozov 1966), inaccessible (deep crevices sealed by red-headed woodpeckers; MacRoberts 1975), or wet (scavenged carcasses placed in water by hyenas; Kruuk 1972). Disturbance may damage or alter features about cache sites and can reduce quality of a site to preserve food or cause changes in habitat selection.

Red squirrels (Tamiasciurus hudsonicus) defend territories surrounding conspicuous, centrally located larderhoards (Smith, C. C. 1968) and therefore provide an excellent model system to examine questions relating to habitat selection of cache locations following disturbance. Red squirrels cache conifer cones in large cone-scale piles or middens resulting from concentrated feeding (Finley 1969). Middens create a cool, moist environment that prevents desiccation of stored cones (Shaw 1936, Brown

1984, USFWS 1993), thereby providing an important food source for winter and times of 19

cone failure (Smith, M. C. 1968, Steele and Koprowski 2001). Midden locations are associated with forest vegetation characteristics, such as high basal area and canopy cover, which help to maintain microclimates necessary to preserve cones (Finley 1969,

Vahle and Patton 1983, Uphoff 1990, Smith and Mannan 1994, Mattson and Reinhart

1997). Vegetation structure surrounding middens may be damaged or altered after disturbance such as severe insect infestation. Frequency and impacts of insect infestations are projected to increase with current trends in global climate change (Ayres and Lombardero 2000); however, the response of squirrels to structural and vegetation disturbance is poorly known. While red squirrel numbers decrease in insect disturbed areas (Matsuoka et al. 2001, Koprowski et al. 2005), little is known about response to changes in forest structure and impacts on habitat selection. Red squirrel establishment of territories and middens has been investigated by documenting displacement (Smith, C.

C. 1968), conducting removal experiments (Larsen and Boutin 1995), and determining settlement of dispersing individuals (Larsen and Boutin 1994, Haughland and Larsen

2004a, b). Recolonization of an unoccupied area is an ideal opportunity to understand territory establishment, as individuals are able to select habitat from a suite of environments with little or no influence of conspecific presence.

The Mt. Graham red squirrel (T. h. grahamensis), a federally listed subspecies of

North American red squirrel (USFWS 1987, Sullivan and Yates 1995), occupies high elevation coniferous forests (spruce-fir and mixed-conifer) of the Pinaleño Mountains,

Arizona (Hoffmeister 1986). Recent insect infestations of western balsam bark beetle

(Drycoetes confusus), spruce beetle (Dendroctonus rufipennis), and the introduced spruce 20

aphid (Elatobium abietinum) beginning in the late 1990’s severely damaged spruce-fir forest and reduced basal area of live trees and seed crop (USDA 2000, Koprowski et al.

2005). Occupied middens in an area with severe insect damage decreased 96% from 23 in 1997 to 1 in 2001 (Koprowski et al. 2005). Subsequently, red squirrels began to recolonize high elevation forests in 2003 that now consist of a mosaic of forest conditions with varying levels of insect damage. Prior to infestations, Mt. Graham red squirrel middens were characterized by indicators of high forest density such as high basal area, log volume, and canopy cover (Smith and Mannan 1994). Midden sites may be limited in the Pinaleño Mountains because of location at the southern terminus of red squirrel distribution and exposure to hotter and drier conditions than habitat in higher latitudes

(Smith and Mannan 1994). Furthermore, isolation in upper elevations of a single mountain range (Hoffmeister 1986) reduces ability to shift to undamaged forest areas in response to disturbance. Recent recolonization of insect-damaged forests by Mt. Graham red squirrels provides an opportunity to examine habitat selection of midden sites and identify possible changes in habitat selection due to disturbance. Our objectives were to:

1) determine which traits distinguish midden locations from random forest locations, 2) determine if any traits distinguish occupied midden locations from historical middens

(those occupied prior to insect damage that now remain unoccupied), and 3) examine relationships between level of insect damage and habitat selection and occupancy. 21

Methods:

Study area

We collected data from September 2003 to December 2005 within the insect-

damaged spruce-fir forest (529 ha, >3048 m) in the Pinaleño Mountains of southeastern

Arizona. The forest is dominated by Engelmann spruce (Picea engelmannii) and cork-

bark fir (Abies lasiocarpa), but includes Douglas fir (Pseudotsuga menziesii), white fir

(Abies concolor), southwestern white pine (Pinus strobiformis), and aspen (Populus

tremuloides).

Midden occupancy

Historical middens, those found during initial surveys by the Arizona Game and

Fish Department in the late 1980s, are permanently marked and location coordinates

documented. We searched for new middens and sign of squirrel activity during ground

surveys each fall (September to November, 2003 - 2005) and while tracking radio-

collared squirrels (see Appendix B). We conducted quarterly surveys of middens (n =

377) to monitor and evaluate midden occupancy. Middens were recorded as occupied or

unoccupied based on sign of red squirrel activity and/or observations of squirrels

displaying resident behavior (Mattson and Reinhart 1997). Sign of red squirrel activity

included remnants of feeding (fresh cone scales and cores), cone caching, and clipped

cones (Finley 1969). Resident behaviors were defined as territorial vocalizations,

chasing of other squirrels, or caching cones within a midden (Gurnell 1987). 22

Midden characteristics

To determine whether features at midden sites within insect-damaged forests were

similar to features found prior to insect infestation (Smith and Mannan 1994), we

compared characteristics among 1) occupied middens (n = 18), 2) a random subset of

unoccupied historical middens (n = 28), and 3) random forest locations (n = 26)

generated in ArcView 3.3 (Environmental Systems Research Institute 2002). To

determine if squirrels were selecting specific historical middens to occupy, we compared

characteristics of occupied and unoccupied middens.

At each occupied midden, unoccupied historical midden, and random location, we

measured 3 physical and 14 vegetation characteristics within circular plots (radius = 10m,

area = 0.3ha) centered on a tree. We measured percent slope, slope aspect, and elevation

from the center of the plots. We used a GPS (GeoExplorer II; Trimble Navigation,

Sunnyvale, CA, USA) to obtain elevation. Direction was determined from slope aspect

as: N, if 315º < aspect ≤ 45º; E, if 45º < aspect ≤ 135º; S, if 135º < aspect ≤ 225º; and W,

if 225º < aspect ≤ 315º. We recorded species, diameter at breast height (DBH), and snag

class (dead or alive) for all trees > 3cm DBH, and noted number of logs and decayed logs

( 20cm DBH at the large end and  2m in length) within each circular plot. We used densiometer readings at 0, 5, and 10m in the four cardinal directions to determine canopy cover. Finally, we placed temperature data loggers (Thermochron iButtons, Model

DS1921G-F5, Dallas Semiconductor, Dallas, TX) at each site to measure surface temperature from 1 September 2004 to 31 August 2005. Data loggers were housed in open-ended plastic containers covered in hardware cloth for protection and placed open 23

side down, 1m north of the center of the plot, and elevated approximately 1cm to allow air circulation. We programmed data loggers to record temperature every 91min from

May – November and 121min from December – April. While downloading data loggers on 11 May 2005, we visually estimated percent snow cover remaining on each plot.

Level of insect damage

To assess level of insect damage throughout the study area, we used high- resolution satellite imagery (0.6m Quickbird) classified by pixel into 9 ground cover classes: 1) deep shadow; 2) shadow; 3) healthy large conifers; 4) small healthy trees or shrubs; 5) dying or dead trees; 6) cienega, grass, or aspen; 7) tan soil; 8) bright rock; and

9) asphalt road, dark rock, dark soil (Wood et al. 2007); shadows were unlit sides of live tree canopies. We used the classified image to generate a map of insect damage where each pixel represented the percent of dead trees (total number of pixel 5 / sum of pixels 1-

6) found within a 10m circular buffer. Using this map, we examined relationships between level of insect damage and midden location and occupancy.

Statistical analyses

We examined physical and vegetative characteristics of all three sites (random, unoccupied midden, and occupied midden) separately with linear contrasts and simultaneously with stepwise logistic regression. For each variable, we used linear contrasts in two comparisons: middens (occupied and unoccupied) versus random 24

locations, and occupied versus unoccupied middens. When examining differences in vegetation structure, we included all physical characteristics as covariates.

We used logistic regression to determine variables that were most important in predicting whether a site was a midden or random location, and occupied or unoccupied midden. Since we had no a priori models to explain differences between middens and random locations, we used stepwise selection (P < 0.25 for entry, P > 0.1 for removal) of variables to include in a logistic regression model. We only included variables for selection that had low correlation (correlation < 0.75) with each other to reduce impacts of multicolinearity.

To examine monthly variation in surface temperature, we compared 3 measures of surface temperature between random locations, unoccupied middens, and occupied middens. Maximum, minimum, and maximum daily range temperatures were the highest, lowest, and highest range (daily high – daily low) of temperatures respectively recorded during the first seven days of every month. For each measure of temperature in each of the 12 months, we used linear contrasts to make the same comparisons as previously described. Elevation, slope, and direction were included as covariates. Since vegetation measures were obtained during summer, we examined July temperatures with stepwise (P < 0.25 for entry, P > 0.1 for removal) multiple regression to determine the variables that best explained variation in midden temperature.

Insect infestation has primarily affected the upper elevations. Proportion of dead trees increases with elevation (F1,64 = 7.39, P = 0.008) and was included as a covariate when examining variation in level of insect damage, measured as percent dead trees from 25

the damage map. We used ANOVA to examine relationships between level of insect damage and midden type (occupied vs unoccupied) and linear regression to examine relationship between insect damage and midden occupancy (percent of quarterly surveys that a midden is occupied).

Results:

Midden occupancy

Forty-five middens were occupied during ≥1 of 10 surveys from September 2003 to December 2005; 31% were either new middens (n = 8) or shifts of previously designated midden sites (n = 6, mean distance from original scale pile = 16m, range = 7.5 to 26m). New middens had minimal or no scale pile, with cached cones (>30) sometimes found on the ground rather than in a scale pile.

Variation in percent occupancy for middens was not associated with percent dead trees (n = 24, F1,21 = 0.12, P = 0.73). Historical middens that were reoccupied (n = 17, mean time since last occupied = 29 months, 95% CI 22 to 36 months) during our study had been occupied more recently than middens that remained unoccupied (n = 100, mean time since last occupied = 42 months, 95% CI 39 to 45 months; t115 = 3.21, P = 0.002).

Midden characteristics

Physical characteristics of midden and random sites differed only in slope and elevation (Table 1). All midden sites (occupied and unoccupied) were located in areas 26

with lower slope than random locations. Occupied middens were on lesser slopes and located at lower elevations than unoccupied middens (Table 1).

Middens and random locations did not differ in number of total trees, live trees, dead trees, decayed logs, and logs, or in average canopy cover at the plot center (Table

1). However, midden sites had higher canopy cover at 5 and 10m, total canopy cover, number of large live trees (>40cm DBH), basal area of total, live, and dead trees, and volume of logs. Basal area of live trees was the only vegetation measurement that was higher at occupied middens than at unoccupied middens (Table 1).

Total basal area, log volume, number of decayed logs, and number of live trees were important variables for distinguishing middens from random locations (Table 2). A location increased in likelihood of being a midden by 3% (95% CI 1 to 5%) with a 1m2 increase in total basal area, by 206% (95% CI 43 to 558%) with a 1m3 increase in log volume, by 40% (95% CI 2 to 98%) with an increase of one decayed log, and decreased by 4% (95% CI 2 to 6%) with a decrease of 1 live tree. Slope, basal area of live trees, and elevation were important variables for distinguishing occupied middens from unoccupied middens (Table 2). A location decreased in likelihood of being an occupied midden by 13% (95% CI 3 to 33%) with a 1° decrease in slope, increased in likelihood by 2% (95% CI 100 to 105%) with a 1m2 increase in basal area of live trees, and decreased in likelihood by 2% (95% CI 1 to 3%) with a 1m decrease in elevation. 27

Midden microclimates

Surface temperatures varied monthly (Figure 1). May temperatures varied

greatly; percent snow cover in May differed between random and midden locations (F2,63

= 9.10, P < 0.001; site mean % cover ± SD: random = 22 ± 32%, unoccupied = 45 ±

41%, occupied = 70 ± 32%). With exception of July, when maximum temperatures at

midden sites tended to be higher (t57 = 1.95, P = 0.057), maximum surface temperatures

tended to be lower at midden sites than at random locations (Figure 1a; October, t58 = -

1.71, P = 0.093; November, t62 = -3.44, P = 0.001; December, t52 = -2.69, P = 0.01;

January, t52 = -2.16, P = 0.036; all other months, |t| < 1.8, P > 0.31). Minimum

temperatures at midden sites tended to be warmer than random locations during fall and

early winter months (Figure 1b; September, t58 = 1.79, P = 0.079; October, t58 = 1.95, P =

0.056; January, t52 = 2.31, P = 0.025; all other months, |t| < 1.4, P > 0.16). Differences in

maximum and minimum surface temperatures were reflected in the lower maximum daily

ranges of midden sites (Figure 1c; October, t58 = -2.09, P = 0.041; November, t62 = -2.97,

P = 0.004; December, t52 = -2.01, P = 0.051; January, t52 = -2.51, P = 0.016; all other

months, |t| < 1.5, P > 0.14). Patterns were similar when comparing occupied and unoccupied middens. Occupied middens had lower maximum temperatures and higher minimum temperatures than unoccupied middens during snow-free months (Figure 1a, b; maximum: September, t58 = -1.87, P = 0.067; November, t62 = -1.98, P = 0.052; May, t52

= -2.21, P = 0.032; July, t57 = -2.04, P = 0.046; August, t56 = -1.87, P = 0.067; all other

months, |t| < 1.5, P > 0.13; minimum: May, t52 = 2.59, P = 0.013; June, t57 = 2.13, P =

0.038; all other months, |t| < 1.6, P > 0.11). Differences were reflected in the lower 28

maximum daily range of temperatures at occupied middens during snow-free months

(Figure 1c; October, t58 = -1.95, P = 0.057; November, t62 = -1.85, P = 0.069; May, t52 = -

2.30, P = 0.026; July, t57 = -2.09, P = 0.042; August, t56 = -1.84, P = 0.072; all other months, |t| < 1.5, P > 0.15).

Maximum July temperature was cooler at random locations and occupied middens than at unoccupied middens (F1,54 = 7.03, P = 0.011), was cooler with higher average canopy cover at plot center (F1,54 = 12.23, P < 0.001) and was warmer with greater number of logs (F1,54 = 3.89, P < 0.053). Minimum July temperature was cooler at middens facing south and east than at middens facing north and west (F1,55 = 43.30, P <

0.001) and was warmer with higher basal area of live trees (F1,55 = 25.22, P < 0.001) and higher average canopy cover at plot center (F1,55 = 10.53, P = 0.002). Maximum July daily range was smaller at random locations and occupied middens than at unoccupied middens (F1,55 = 7.35, P = 0.009), was smaller with higher average canopy cover at plot center (F1,55 = 17.92, P < 0.001), and larger with higher numbers of logs (F1,55 = 4.56, P =

0.037).

Level of insect damage

Insect damage did not differ among random (n = 26, mean = 52.1%, range = 16.2 to 78.4%), unoccupied (n = 26, mean = 53.4%, range = 38.2 to 74.2%), and occupied sites (n = 16, mean = 45.3%, range = 24.5 to 63.2%) where vegetation was measured

(F2,64 = 2.28, P = 0.110). Percent dead trees determined from imagery was similar to those measured in the field (paired t-test: t71 = 0.487, P = 0.63). Percent dead trees at 29

unoccupied middens (n = 237, mean = 59.1%, 95% CI 57.6 to 60.7%) was 7.07% higher than occupied middens (n = 39, mean = 50.8, 95% CI 46.4 to 55.3%; F1,273 = 12.24, P <

0.001).

Discussion:

Despite dramatic changes to forest structure, midden site selection by Mt. Graham red squirrels within insect-damaged forest was characterized by indicators of high forest density, such as high basal area, log volume, and canopy cover. Results were congruent with features at middens prior to insect infestation (Smith and Mannan 1994) as well as across red squirrel distribution (Finley 1969, Vahle and Patton 1983, Uphoff 1990,

Mattson and Reinhart 1997). However, differences between occupied middens and historical middens, which are presently unoccupied, indicate severity of insect infestation that red squirrels will tolerate. Midden sites occupied during our study were found in areas with <64% dead trees. Large trees produce the most cones (USDA and USFS

1965) and proximity to a food source is economical for hoarding due to reduced time and energy spent traveling (Charnov 1976). Although number of large trees did not differ, occupied middens had greater potential for cone production with greater basal area of live trees than historic midden sites. Establishment of middens near a food source may be important in disturbed forests, such as our study area, where availability of food is reduced and basal area of live trees around middens has decreased 63.3% following onset of insect infestation (Koprowski et al. 2005). 30

Occupied middens experienced cooler surface temperatures than unoccupied middens during snow-free months when sun and heat have strong desiccating effects on middens. As cones age, scales open and seeds disperse by gravity and wind (USDA and

USFS 1965), at which point squirrels no longer benefit from cone harvesting (Shaw

1936). Cone opening is prevented by storage in a midden’s cool moist environment

(Shaw 1936, Brown 1984), suggesting that locations with cool surface temperatures will better preserve cones. Squirrels are not likely selecting midden locations solely on surface temperature, as random locations occasionally had temperature regimes similar to occupied middens. Cool surface temperatures were associated with high basal area, log volume, and canopy cover, likely cues used for selection since squirrels demonstrate preference and ability to discern differences in these microhabitat characteristics (Bakker

2006). The ability of vegetation structure to create a cool moist microclimate was further demonstrated in May when percent snow cover remaining on occupied middens was 3 times greater than cover at random locations. Canopy cover at the center of the plot was associated with maximum, minimum, and maximum daily range of surface temperature and may be more important in insect-damaged forest where canopy cover at midden sites decreased 68% from 1990 to 2002 (Koprowski, unpublished data). Sites with greater canopy cover may also provide protection from raptors, the most commonly observed predator (Schauffert et al. 2002). Furthermore, areas of reduced canopy cover equate to areas with increased dead trees and fuel loads that can result in increased frequency and intensity of fire that further decreases survival (Koprowski et al. 2006). Thus reduced 31

canopy cover from severe insect infestation has potential to alter midden microclimate and increase rates of predation and fire, important influences on red squirrel survival.

Recolonization of insect-damaged forest provided increased understanding of what constitutes midden habitat, as individuals were able to select a midden location within a mosaic of forest conditions and levels of insect-damage without residents already in the forest to influence site selection. Many habitat selection models assume that individuals are selecting the highest quality habitat available (Fretwell and Lucas

1969, Pulliam and Danielson 1991). Although individuals presumably did not sample the entire 529ha study area, Mt. Graham red squirrels have large home ranges (mean ± SD, year: 3.5 ± 6.2ha, summer: 8.2 ± 8.0ha; Koprowski et al. in review) and explore large areas of forest while foraging. Squirrels presumably selected the best unoccupied midden location within areas sampled (Pulliam and Danielson 1991). Social dominance and availability of vacant territories influences habitat selection and territory establishment in many taxa (Petit and Petit 1996, Jacquot and Solomon 2004, Stiver et al. 2006). Red squirrels are excluded from areas by territorial defense of conspecifics (Smith, C. C.

1968) and midden acquisition is often influenced by vacant territories (Gurnell 1984,

Price et al. 1986) or bequeathal by females (Price and Boutin 1992, Boutin and Price

1993). Such effects are minimized in our study due to recolonization of unoccupied forest.

The importance of middens for red squirrel survival, particularly during times of low food availability, has long been acknowledged (Smith, M. C. 1968, Steele and

Koprowski 2001). However, characteristics that define a midden location may not be 32

consistent throughout the range or under varying forest conditions. Aspects of midden

structure associated with surface temperature may be more important in the southern

portion of red squirrel distribution where temperatures are warmer and drier. In contrast,

in parts of the Rocky Mountains, risk of midden excavation by grizzly bears, Ursus

arctos horribilis, (Mattson and Reinhart 1997) may have a greater influence on midden site selection than features influencing surface temperature. Since the midden is critical for survival, selection of midden location will influence selection of habitat at a larger scale. If numbers of live trees in a forest are severely reduced (>64% tree mortality in this study), squirrels may not find adequate midden locations and as a result may not select or use any forest in the area. Studies of forest thinning practices suggest similar results with a decline in squirrel density after removal of >50% stem density (Koprowski

2005b). Red squirrels do not appear to alter habitat selection following severe insect infestation; forests receiving greater damage will likely no longer represent habitat. In addition, although red squirrels equally prefered insect-damaged to healthy forest, survival was significantly lower compared to healthy forest (Appendix B). Thus, equal- preference of insect-damaged forests combined with low survival suggests response of squirrels to insect disturbance may result in an ecological trap (Schlaepfer et al. 2002,

Robertson and Hutto 2006). Decreased food storage abilities, increased risk of predation, and greater likelihood of catastrophic fire, further reduce the quality of insect-damaged habitat, which provides additional support for the possibility of an ecological trap.

Although ecological traps do not necessitate population declines (Robertson and Hutto 33

2006), they remain a concern for the persistence of populations in isolated or fragmented

forests with no opportunity to move to undamaged habitat.

Impacts of forest insect infestations and other natural disturbances are projected to

increase in the future with the current trends in global climate change (Ayres and

Lombardero 2000, Dale et al. 2000) and therefore have the potential to affect many

forest-dwelling species. Ecological and evolutionary traps (Schlaepfer et al. 2002) may

become more common as climate change creates asynchronous relationships between

species and their environments (eg. Yellow-bellied marmots, Marmota flaviventris, emerging early from hibernation causing asynchrony with phenology of food plants;

Parmesan 2006). Disturbance and climate change commonly result in range shifts

(Parmesan 2006), but high rates of habitat loss may eliminate that option for many species. Therefore, documenting response and ability, or lack there of, to adapt to impacts of disturbance is a priority for future research.

References:

Ayres, M. P., and M. J. Lombardero. 2000. Assessing the consequences of global change for forest disturbance from herbivores and pathogens. The Science of the Total Environment 262:263-286.

Bakker, V. J. 2006. Microhabitat features influence the movements of red squirrels (Tamiasciurus hudsonicus) on unfamiliar ground. Journal of Mammalogy 87:124- 130.

Bengtsson, J., S. G. Nilsson, A. Franc, and P. Menozzi. 2000. Biodiversity, disturbances, ecosystem function and management of European forests. Forest Ecology and Management 132:39-50. 34

Boutin, S., and K. Price. 1993. Post-breeding dispersal by female red squirrels (Tamiasciurus hudsonicus): the effect of local vacancies. Behavioral Ecology 4:151-155.

Brown, D. E. 1984. Arizona's tree squirrels. Arizona Game and Fish Department, Phoenix, Arizona.

Charnov, E. L. 1976. Optimal foraging, the Marginal Value Theorem. Theoretcial Population Biology 9:129-136.

Dale, V. H., L. A. Joyce, S. McNulty, and R. P. Neilson. 2000. The interplay between climate change, forests, and disturbances. The Science of the Total Environment 262:201-204.

Environmental Systems Research Institute. 2002. ArcView GIS: Release 3.3 [software]. in. Environmental Systems Research Institute, Redlands, California.

Finley, R. B. J. 1969. Cone caches and middens of tamiasciurus in the Rocky Mountain region. University Kansas Museum Natural History Miscellaneous Publication 51:233-273.

Formozov, A. N. 1966. Adaptive modifications of behavior in mammals of the Eurasian steppes. Journal of Mammalogy 47:208-223.

Frenandez, N., and F. Palomares. 2000. The selection of breeding dens by the endangered Iberian lynx (Lynx pardinus): implications for its conservation. Biological Conservation 94:51-61.

Fretwell, S. D., and H. L. Lucas, Jr. 1969. On territorial behavior and other factors influencing habitat distribution in birds, I. theoretical development. Acta- Biotheoretica 19:16-36.

Gurnell, J. 1984. Home range, territoriality, caching behaviour and food supply of the Red Squirrel (Tamiasciurus hudsonicus fremonti) in a subalpine lodgepole pine forest. Animal Behaviour 32:1119-1131.

Gurnell, J. 1987. The natural history of squirrels. Facts on File Inc., New York, New York.

Hall, L. S., P. R. Krausman, and M. L. Morrison. 1997. The habitat concept and a plea for standard terminology. Wildlife Society Bulletin 25:173-182. 35

Haughland, D. L., and K. W. Larsen. 2004a. Ecology of North American red squirrels across contrasting habitats: relating natal dispersal to habitat. Journal of Mammalogy 85:225 - 236.

Haughland, D. L., and K. W. Larsen. 2004b. Exploration correlates with settlement: red squirrel dispersal in contrasting habitats. Journal of Animal Ecology 73:1024- 1034.

Hoffmeister, D. F. 1986. Tamiasciurus hudsonicus. Mammals of Arizona:213-217.

Jacquot, J. J., and N. G. Solomon. 2004. Experimental manipulation of territory occupancy: effects on immigration of female prairie voles. Journal of Mammalogy 85:1009-1014.

Jones, J., R. D. DeBruyn, J. J. Barg, and R. J. Robertson. 2001. Assessing the effects of natural disturbance on a neotropical migrant songbird. Ecology 82:2628-2635.

Koprowski, J. L. 2005. Management and conservation of tree squirrels: the importance of edemism, species richness, and forest condition. Pages 245-250 in G. J. Gottfried, B. S. Gebow, L. G. Eskew, and C. B. Edminster, editors. Connecting mountain islands and desert seas: biodiversity and management of the Madrean archipelago II. U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Tucson, Arizona.

Koprowski, J. L., M. I. Alanen, and A. M. Lynch. 2005. Nowhere to run and nowhere to hide: Response of endemic Mt. Graham red squirrels to catastrophic forest damage. Biological Conservation 126:491-498.

Koprowski, J. L., S. R. B. King, and M. Merrick. in review. Home ranges are expanded in a peripheral population: space use of endangered Mt. Graham red squirrels. Canadian Journal of Zoology.

Koprowski, J. L., K. M. Leonard, C. A. Zugmeyer, and J. L. Jolley. 2006. Direct effects of fire on endangered Mt. Graham red squirrels. Southwestern Naturalist 51:59- 63.

Kruuk, H. 1972. The spotted hyena: A study of predation and social behavior. University of Chicago Press, Chicago, Illinois.

Larsen, K. W., and S. Boutin. 1994. Movements, survival, and settlement of red squirrel (Tamiasciurus hudsonicus) offspring. Ecology 75:214-223. 36

Larsen, K. W., and S. Boutin. 1995. Exploring territory quality in the North American red squirrel through removal experiments. Canadian Journal of Zoology 73:1115- 1122.

Lecchini, D., S. Planes, and R. Galzin. 2005. Experimental assessment of sensory modalities of coral-reef fish larvae in the recognition of their settlement habitat. Behavioral Ecology and Sociobiology 58:18-26.

MacRoberts, M. H. 1975. Food storage and winter territory in red-headed woodpeckers in northwestern Louisiana. Auk 92:382-385.

Matsuoka, S. M., C. M. Handel, and D. R. Ruthrauff. 2001. Densities of breeding birds and changes in vegetation in an Alaskan boreal forest following a massive disturbance by spruce beetles. Canadian Journal of Zoology 79:1678-1690.

Mattson, D. J., and D. P. Reinhart. 1997. Excavation of red squirrel middens by grizzly bears in the whitebark pine zone. Journal of Applied Ecology 34:926-940.

McNett, B. J., and A. L. Rypstra. 2000. Habitat selection in a large orb-weaving spider: vegetational complexity determines site selection and distribution. Ecological Entomology 25:423-432.

Oppel, S., H. M. Schaefer, V. Schmidt, and B. Schroder. 2004. Habitat selection by the pale-headed brush-finch (Atlapetes pallidiceps) in southern Ecuador: implications for conservation. Biological Conservation 118:33-40.

Parmesan, C. 2006. Ecological and evolutionary responses to recent climate change. Annual Review of Ecology, Evolution, and Systematics 37:637-669.

Penteriani, V., M. Mathiaut, and G. Boisson. 2002. Immediate species responses to catastrophic natural disturbances: windthrow effects on density, productivity, nesting stand choice, and fidelity in northern goshawks (Accipiter gentilis). The Auk 119:1132-1137.

Petit, L. J., and D. R. Petit. 1996. Factors governing habitat selection by prothonotary warblers: field tests of the Fretwell-Lucas model. Ecological Monographs 66:367- 387.

Price, K., and S. Boutin. 1992. Territorial bequeathal by red squirrel mothers. Behavioral Ecology 4:144-150.

Price, K., K. Broughton, S. Boutin, and A. R. E. Sinclair. 1986. Territory size and ownership in red squirrels: response to removals. Canadian Journal of Zoology 64:1144-1147. 37

Pulliam, H. R., and B. J. Danielson. 1991. Sources, sinks, and habitat selection: A landscape perspective on population dynamics. The American Naturalist 137:S50- S66.

Robertson, B. A., and R. L. Hutto. 2006. A framework for understanding ecological traps and an evaluation of existing evidence. Ecology 87:1075-1085.

Schauffert, C. A., J. L. Koprowski, V. L. Greer, M. I. Alanen, K. A. Hutton, and P. J. Young. 2002. Interactions between predators and Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis). The Southwestern Naturalist 47:498- 501.

Schlaepfer, M. A., M. C. Runge, and P. W. Sherman. 2002. Ecological and evolutionary traps. Trends in Ecology and Evolution 17:474-480.

Shaw, W. T. 1936. Moisture and its relation to the cone-storing habit of the western pine squirrel. Journal of Mammalogy 17:337-349.

Smith, A. A., and R. W. Mannan. 1994. Distinguishing characteristics of Mount Graham red squirrel midden sites. Journal of Wildlife Management 58:437-445.

Smith, C. C. 1968. The adaptive nature of social organization in the genus of three squirrels Tamiasciurus. Ecological Monographs 38:31-63.

Smith, C. C., and O. J. Reichman. 1984. The evolution of food caching by birds and mammals. Annual Review of Ecology and Systematics 15:329-351.

Smith, M. C. 1968. Red squirrel responses to spruce cone failure in interior Alaska. Journal of Wildlife Management 32:305-317.

Sousa, W. P. 1984. The role of disturbance in natural communities. Annual Review of Ecology and Systematics 15:353-391.

Steele, M. A., and J. L. Koprowski. 2001. North American tree squirrels. Smithsonian Institution Press, Washington D. C.

Stiver, K. A., J. Fitzpatrick, J. K. Desjardins, and S. Balshine. 2006. Sex differences in rates of territory joining and inheritance in a cooperatively breeding cichlid fish. Animal Behaviour 71:449-456.

Sullivan, R. M., and T. L. Yates. 1995. Population genetics and conservation biology of relict populations of red squirrels. Pages 193-208 in C. A. Istock, and R .S. 38

Hoffmann, editor. Storm over a mountain island: conservation biology and the Mt. Graham affair. University of Arizona Press, Tucson, Arizona.

U. S. Department of Agriculture, and U. S. Forest Service (USDA and USFS). 1965. Silvics of forest trees of the United States. Agriculture Handbook No. 271, Washington D. C.

U. S. Department of Agriculture (USDA). 2000. Forest insect and disease conditions in the Southwestern region. Flagstaff, Arizona.

U. S. Fish and Wildlife Service (USFWS). 1987. Endangered and threatened wildlife and plants; determination of endangered status for the Mount Graham Red Squirrel. Federal Register 52:20994-20999.

U. S. Fish and Wildlife Service (USFWS). 1993. Mount Graham red squirrel recovery plan. Phoenix, Arizona, USA.

Uphoff, K. C. 1990. Habitat use and reproductive ecology of red squirrels (Tamiasciurus hudsonicus) in central Arizona. MS thesis. Arizona State University, Tempe.

Vahle, J. R., and D. R. Patton. 1983. Red squirrel cover requirements in Arizona mixed conifer forests. Journal of Forestry 81:14-15.

Vander Wall, S. B. 1990. Food hoarding in animals. University of Chicago Press, Chicago, Illinois.

Werner, E. E., J. F. Gilliam, D. J. Hall, and G. G. Mittelbach. 1983. An experimental test of the effects of predation risk on habitat use in fish. Ecology 64:1540-1548.

Williams, J. B., and G. O. Batzli. 1979. Competition among bark-foraging birds in central Illinois: experimental evidence. Condor 81:122-132.

Wood, D. J. A., S. Drake, S. P. Rushton, D. Rautenkranz, P. W. W. Lurz, and J. L. Koprowksi. 2007. Fine scale analysis of Mount Graham red squirrel habitat following disturbance. Jounal of Wildlife Management. In press. 39 P* 0.949 0.587 0.752 0.660 0.957 0.942 0.106 0.018 0.896 0.953 0.855 0.001 0.582 0.950 0.469 0.155 < 0.001 Unocc vs Occ P* 0.017 0.007 0.589 0.512 0.097 0.022 0.008 0.368 0.019 0.012 0.603 0.410 0.022 0.413 0.169 <0.001 <0.001 Rand vs Midd 78 ± 15 52 ± 10 Midden 133 ± 24 239 ± 16 78.6 ± 3.4 74.3 ± 3.8 944 ± 113 Occupied 82.9 ± 4.1 78.5 ± 4.1 10.2 ± 1.7 16.1 ± 2.4 3,118 ± 10 1,115 ± 144 2,059 ± 175 3,314 ± 347 8,070 ± 476 4,756 ± 455 60 ± 8 89 ± 11 Midden 988 ± 85 155 ± 18 225 ± 12 3,146 ± 7 73.6 ± 2.8 839 ± 108 73.1 ± 2.6 73.7 ± 3.1 71.8 ± 3.1 11.8 ± 1.3 25.6 ± 1.9 1,827 ± 132 1,602 ± 261 6,856 ± 358 5,253 ± 343 Unoccupied ) unoccupied middens (n = 28) and occupied middens (n = 18) in the insect-damaged (n = 18) in the insect-damaged (n = 28) and occupied middens ) unoccupied middens 27 ± 8 72 ± 12 82 ± 19 217 ± 12 685 ± 87 1.4 ± 1.3 Random Location 3,115 ± 7 29.0 ± 2.0 60.9 ± 2.9 894 ± 111 64.7 ± 2.6 72.5 ± 3.1 60.5 ± 3.1 1,578 ± 135 1,583 ± 267 5,485 ± 367 3,903 ± 351 /ha) 2 /ha) 3 (>40cm DBH) Tamiasciurus hudsonicus grahamensis Tamiasciurus hudsonicus ( forest above 3000m in the Pinaleño Mountains, Arizona, Fall 2003 to Summer 2005. in the Pinaleño Mountains, Arizona, forest above 3000m 10m from center Total trees (no./ha) Live trees Live trees Dead trees Total basal area (m Live tree basal area Total logs (no./ha) Volume logs (m Total decayed logs (no./ha) Dead tree basal area Aspect (º) Total canopy cover (%) Center 5m from center Physical/Vegetation Characteristic Slope (%) Elevation (m) *Results of linear contrasts*Results of linear occupied middens vs. middens and 2) unoccupied middens, locations vs. comparing 1) random TABLE 1. Mean (± SE) physical and vegetation measurements for random locations (n = 26) and Mt. Graham red squirrel Graham red squirrel Mt. = 26) and locations (n for random measurements physical and vegetation 1. Mean (± SE) TABLE 40

TABLE 2. Results of step-wise logistic regression models to assess odds of a) a site being a Mt. Graham red squirrel (Tamiasciurus hudsonicus grahamensis) midden (n = 46) vs random location (n = 26), or b) a midden being occupied (n = 18) vs unoccupied (n = 28) in the insect-damaged forest above 3000m in the Pinaleño Mountains, Arizona, Fall 2003 to Summer 2005.

Source Estimate Standard P Odds Ratio Error a) Intercept -7.46 2.32 0.001 - Total basal area (m2) 0.035 0.011 0.002 1.03 Log volume (m3) 1.12 0.39 0.004 3.06 Number decayed logs 0.35 0.17 0.036 1.4 Number live trees -0.039 0.021 0.058 0.96 b) Intercept 73.60 38.78 0.058 - Slope (º) -0.14 0.058 0.013 0.869 Live basal area (m2) 0.028 0.014 0.043 1.02 Elevation (m) -0.023 0.012 0.057 0.977 41

1* 2 a) 50

C 40 2*

º 2* 1* p 30 m e

T 1 2 2 20 x a

M 10 1 1 0

10 b) C

º 5 1* 2

p 2 m 1 e 0 1 T

n i

M -5

-10 2 c) 40 C º

1 2* e

g 30 2* n 1 2* a 2 R 20 y l i

a 1

D 10 1 x a

M 0

Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug

FIGURE 1. Mean (± SE) a) maximum, b) minimum, and c) maximum daily range of ground surface temperature from September 2004 to August 2005 for random locations ( ), and unoccupied ( ) and occupied ( ) Mt. Graham red squirrel (Tamiasciurus hudsonicus grahamensis) middens the insect-damaged forest above 3000m in the Pinaleño Mountains, Arizona. A “1” indicates P <0.05 for linear contrast of middens and random locations, “2” indicates P <0.05 for linear contrast of unoccupied and occupied middens. Asterisk (*) indicates 0.1> P >0.05. Sample sizes are (random, unoccupied, occupied): Sept – Oct (26, 23, 12); Nov (26, 23, 16); Dec – May (24, 20, 11); Jun – Jul (22, 23, 15); Aug (22, 23,14). 42

APPENDIX B: SEVERELY INSECT-DAMAGED FOREST MAY FUNCTION AS AN ECOLOGICAL TRAP FOR RED SQUIRRELS

Abstract:

Current climate trends predict rapid alterations of the frequency, intensity, and

extent of natural forest disturbances. These rapid changes in forest structure may create

ecological traps. Isolated populations face increased risk of extinction because of

susceptibility to stochastic events, thus highlighting the importance of understanding

response to disturbance and ability of disturbance to create ecological traps. Recent

insect infestations in the spruce-fir forest in the Pinaleño Mountains of southeastern

Arizona provided an opportunity to document response to severe forest disturbance and

existence of an ecological trap in an endemic mountain isolate, the endangered Mt.

Graham red squirrel (Tamiasciurus hudsonicus grahamensis). From September 2003 to

December 2005 we assessed habitat selection and monitored potential correlates of fitness (body mass, reproduction, and survivorship) in red squirrels living in insect- damaged forest, while drawing comparisons to squirrels inhabiting healthy or undamaged mixed-conifer forest. Although Mt. Graham red squirrels demonstrate equal-preference for habitat within insect-damaged forests and reproduce as well as individuals in healthy forest, poor survivorship and reduced potential to reproduce suggest insect-damaged forest may function as an ecological trap. In addition, areas selected within insect- damaged forest had <69% dead trees, suggesting an upper limit to extent of tree mortality tolerated by red squirrels. Habitat selection and sensitivity to disturbance will influence use of insect-damaged areas by forest-dwelling species. Although insect-damaged forest 43

may remain habitat for a few individuals, low survivorship may generate an ecological trap.

Introduction:

Natural and anthropogenic disturbances create temporal and spatial heterogeneity of resources and can cause short- and long-term changes to habitat (Sousa 1984,

Bengtsson et al. 2000). Severe disturbance may dramatically alter habitat structure, causing reduced reproductive success and site abandonment (Jones et al. 2001, Penteriani et al. 2002). Disturbance often results in habitat loss and degradation that fragment and isolate populations (Ceballos and Ehrlich 2002, Ewers and Didham 2006). Current trends in climate change are predicted to increase frequency, intensity, and extent of natural forest disturbance (Dale et al. 2001), thus increasing possible loss and degradation of forests. Geographic isolation often increases risk of population extinction by increasing susceptibility to stochastic events (Lande 1998, Alvarez 2001), thus, isolation combined with climate altered disturbance regimes highlight the important future role of disturbance.

Disturbance due to insect infestation has ecological and economical impacts that will increase with elevated global temperatures (Ayres and Lombardero 2000). Insect infestation alters forest structure by reducing live basal area and increasing understory vegetation, which changes animal guilds (Matsuoka et al. 2001, Suring et al. 2006).

Responses of small mammals to habitat change from insect infestation are largely unknown but studies suggest decreased abundance after insect damage, particularly for 44

seed-eating forest obligates (Matsuoka et al. 2001, Koprowski et al. 2005, McDonough and Rexstad 2005). Rapid changes to forest composition and structure may function as an ‘ecological trap’ (Flaspohler et al. 2001). Ecological traps occur when animals make maladaptive habitat selection using cues that formerly correlated with habitat quality

(Schlaepfer et al. 2002). Equal-preference traps, when animals equally prefer high and low quality habitat, and severe traps, when animals prefer lower quality habitat, are known (Robertson and Hutto 2006). A key requirement of ecological traps is documentation of preference, which is often influenced by competition and aggression from conspecifics (Robertson and Hutto 2006). Therefore, recolonization of an unoccupied area is an ideal opportunity to understand individual habitat preference and, when coupled with measures of individual fitness, determine the existence of an ecological trap. In addition, with climate change increasing the intensity and extent of insect infestation, ecological traps may be increasingly frequent in forest environments.

Recent insect infestations of western balsam bark beetle (Drycoetes confusus), spruce beetle (Dendroctonus rufipennis), and introduced spruce aphid (Elatobium abietinum) beginning in the late 1990’s severely damaged spruce-fir forest throughout the southwestern United States, reducing seed crop and basal area of live trees (USDA 2000,

Koprowski et al. 2005). Subsequently, recovering forests serve as habitat for forest- dwelling species and recolonization of areas offers an excellent opportunity to examine response to insect infestation and assess if insect-damage can create an ecological trap.

Therefore, our objectives were to: 1) assess habitat selection, and 2) examine potential correlates of fitness (body mass, reproduction, survivorship) in Mt. Graham red squirrels 45

(Tamiasciurus hudsonicus grahamensis), a seed-eating forest obligate that recently

recolonized insect-damaged forest. We draw comparisons to squirrels inhabiting lower

elevation mixed-conifer forest that is relatively undamaged and identify a possible

ecological trap that results from insect-damage.

Methods:

Model organism and system

Tree squirrels are an indicator of forest health (Carey 2000) due to their

dependence on mature trees for food, nests, and cover (Gurnell 1987, Steele and

Koprowski 2001). Red squirrels (T. hudsonicus) are an excellent model species to

determine effects of insect infestation. In the western part of their range, red squirrels are

territorial and defend conspicuous larderhoards or middens (Steele 1998); therefore,

occupancy of an area by red squirrels is easily determined. Furthermore, if insect

infestation severely damages a forest, red squirrels will likely demonstrate a demographic

and behavioral response due to their dependence on conifer trees as a primary food

source (Gurnell 1987). The Mt. Graham red squirrel (T. h. grahamensis), a federally listed subspecies of North American red squirrel (USFWS 1987, Sullivan and Yates

1995), occupies high elevation spruce-fir and mixed-conifer forests of the Pinaleño

Mountains in southeastern Arizona (Hoffmeister 1986). Recent recolonization of insect- damaged forest provides an opportunity to examine response to insect infestation and possible existence of an ecological trap. 46

Study area and trapping

We studied Mt. Graham red squirrels from September 2003 to December 2005

within the insect-damaged, spruce-fir forest (529ha, 3048 - 3267m) in the Pinaleño

Mountains of southeastern Arizona dominated by Engelmann spruce (Picea engelmannii)

and cork-bark fir (Abies lasiocarpa) in the upper elevations (>3048m), but includes

Douglas fir (Pseudotsuga menziesii), white fir (Abies concolor), southwestern white pine

(Pinus strobiformis), ponderosa pine (Pinus ponderosa), and aspen (Populus tremuloides) as elevation decreases. All of the spruce-fir forest (hereafter termed ‘insect-damaged’) in the Pinaleño Mountains was impacted by insect infestation and occurs within the area studied (Wood et al. 2007). Therefore, we simultaneously collected data in a portion of lower elevation, undamaged mixed-conifer forest (hereafter termed ‘healthy’, 130ha,

2850 – 2979m) dominated by cork-bark fir, but includes previously listed species. These two study areas contain 34% of all known Mt. Graham red squirrel middens in the

Pinaleño Mountains (Koprowski et al. 2005) and are demographically linked by dispersal in both directions (Zugmeyer, unpublished data).

We used live traps (Model 202, 48.3x15.2x15.2cm, Tomahawk Live Trap,

Tomahawk, WI) to capture squirrels at middens, transferred individuals to a handling bag

(Koprowski 2002), and assessed body mass, sex, age class, and reproductive status. We fitted adult squirrels (>200g) with radio collars (Model SOM 2190, Wildlife Materials

International Inc., Carbondale, IL) and marked all squirrels for permanent identification with uniquely numbered, metal ear tags (Model 1005-1, National Band and Tag Co., 47

Newport, KY) and color-coded, plastic washers (Model 1841 3/8”, National Band and

Tag Co., Newport, KY).

Habitat preference

To investigate preference of insect-damaged over healthy forest, we quantified available habitat within each forest type by evaluating midden occupancy each

December. Available middens had been occupied at least once in the previous three years, but were unoccupied at time of survey. Middens were recorded as occupied or unoccupied based on sign of red squirrel activity and/or observations of squirrels displaying resident behavior (Finley 1969, Gurnell 1987, Mattson and Reinhart 1997).

We used records of midden occupany from 1990-1997 to examine historical occupancy of both forest types prior to insect infestation.

Home range and space use

To investigate home range and space use, we obtained radio-telemetric locations by homing and biangulation (White and Garrott 1990) using a receiver (Model TRX-

2000S, Wildlife Materials International Inc., Carbondale, IL) and Yagi antenna (Model

F164-165-3FB, Wildlife Materials International Inc., Carbondale, IL). We tracked squirrels in daylight throughout each month. Consecutive locations within the same day were separated by a minimum of 1 hour to reduce autocorrelation.

We calculated seasonal home ranges from January 2004 to August 2005. Seasons were defined as winter (December to February), spring (March to May), summer (June to 48

August), and fall (September to November). Thirty locations are standard for home

range calculation (Kenward 1987). Incremental area plots reached an asymptote for most

squirrels at 20 locations, setting our minimum number of locations. Average number of

locations per calculated home range was 47 (± 12 SD), with only 12% (n = 25) with <30

locations. We calculated fixed-kernel home ranges (Millspaugh and Marzluff 2001) and

include minimum convex polygons (MCP) for comparison with other studies. We used

Animal Movements (Hooge and Eichenlaub 2000), with smoothing parameters calculated

by least squares cross-validation (LSCV), to determine 95% (hereafter termed ‘range’)

and 50% (hereafter termed ‘core’) fixed-kernels, and Ranges 6 (Anatrack Ltd 2003) to

calculate both 95% and 50% MCP.

We used two simultaneous directional bearings, with interbearing angles between

70-110° (White and Garrott 1990), to obtain biangulations and calculated locations with

LOAS (Ecological Software Solutions 2003). We used a beacon to estimate location

error (Millspaugh and Marzluff 2001). Within insect-damaged forests, steep drainages

made biangulation impossible or less accurate (mean ± SD, n = 32, bearing error, 8.6º ±

6.2º; linear error 42.9m ± 28.7m). We increased homing in insect-damaged forests (mean

± SD, % homing 2004 = 35.9% ± 30.9%, % homing 2005 = 87.6% ± 17.3%); range and

core size did not differ between years (95% range, t44 = 1.32, P = 0.19; 50% core area, t44

= 1.07, P = 0.29). Error within healthy forests was low (mean ± SD, n = 64, bearing

error, 5.6º ± 4.5º; linear error, 12m ± 6m; Koprowski et al. in review); biangulation was used most frequently (mean homing = 28% ± 23% SD). 49

To assess areas selected by squirrels within insect-damaged forests, we used high- resolution satellite imagery (0.6m Quickbird) classified by pixel into 9 ground cover classes: 1) deep shadow; 2) shadow; 3) healthy large conifers; 4) small healthy trees or shrubs; 5) dying or dead trees; 6) cienega, grass, or aspen; 7) tan soil; 8) bright rock; and

9) asphalt road, dark rock, dark soil (Wood et al. 2007). We used Hawth’s tools (Beyer

2006) to clip seasonal home range polygons and determine cover composition. We compared cover composition in random home ranges created by centering each squirrel’s home range about a random location in the study area. Shadows were unlit sides of live tree canopies (Wood et al. 2007); for each actual and random range and core we calculated percent dead tree pixels (total number of pixel 5 / sum of pixels 1-6), percent open pixels (sum of pixels 7, 8, and 9 / total pixels), and number of live (sum of pixels 1-

4, and 6) and dead tree pixels (total number of pixel 5).

Body mass and demography

We determined seasonal body mass and reproductive condition (Koprowski

2005a) for adult males and females in both study areas. Litter size was determined by homing to nests of all lactating females during 2004 and 2005. We calculated the number of potential reproductive events attained since first capture for all squirrels that died prior to the end of the study. We used peak frequency of scrotal condition and presence of lactation in Mt. Graham red squirrels (Koprowski 2005) to set limits for potential reproduction to March 31 for males and July 31 for females (ie. death prior to these dates

= no reproduction in a given year). We examined the ratio of adults to subadults and males to females for newly captured squirrels. 50

We used radio-telemetry, trapping, and observation at middens to monitor survival of squirrels. If a squirrel could not be found or the radio-signal remained in one location, we tried to capture missing squirrels at middens to determine mortality or collar failure. Squirrels were considered a mortality if remains and radio collar were found; radio collar was found with no new sign and/or a new squirrel captured at the midden; and if the missing individual was not captured and a new squirrel was observed at the missing individual’s midden. Red squirrels are territorial, and if alive, tolerance of other squirrels at their midden is unlikely (Smith, C. C. 1968).

Statistical analyses

We conducted statistical analysis in JMP 5.1 (SAS Institute Inc 2003). We report means (± SE) as untransformed values. To determine squirrel preference for each forest type, we used ANOVA and Fisher’s exact to compare historical and current occupancy of available middens. To determine severity of insect-damage tolerated by squirrels, we compared percent dead trees, percent open, and number of live and dead trees between actual and random ranges and cores. We used the natural logarithm transformation on numbers of live and dead trees. We used multiple-regression to examine effect of type

(random vs actual), while accounting for size and season, and ANOVA to examine effect of type in each season, while accounting for size.

We used the natural logarithm transformation on home range and core size.

Range and core size did not vary between years in either forest type (insect-damaged: range, t44 = 1.32, P = 0.20; core, t44 = 1.07, P = 0.29; healthy: range, t168 = -0.10, P = 51

0.92; core, t168 = -0.73, P = 0.47) and were pooled. We used ANOVA, multiple

regression, and Tukey-Kramer post hoc analyses, to examine effect of season, home

range estimator, sex, and forest type on range and core size.

To determine dispersal patterns, we used Fisher’s exact test to compare ratio of

adults to subadults and males to females among newly captured individuals. We used t-

tests and ANOVA to compare litter size, number of potential reproductive events, and

body mass of squirrels between forest types. Mt. Graham red squirrels can demonstrate

sex and seasonal differences in body mass (Koprowski 2005a) so we compared body

mass separately for each sex, while including season as a covariate. Body mass did not

vary with year (insect-damaged: male F3,42 = 0.09, P = 0.96; female, F3,37 = 1.8, P = 0.16; healthy: male F3,148 = 0.90, P = 0.44, female F3,122 = 1.45, P = 0.23) so data were pooled.

We used Kaplan-Meier analysis (Kaplan and Meier 1958) to examine differences in

survival between forest type and the sexes; we determined fate, number of days survived

since first capture, of each squirrel as of December 2005. We used linear regression to

examine the relationship between survival within insect-damaged forests and home range

size and percent of home range classified as dead trees or open; we used the last home

range calculated prior to death.

Results:

Habitat is available in healthy and insect-damaged forest

Ample opportunities appear to exist for settlement in each of forest types. Prior to

insect infestation (1990-1997), percent of available middens occupied within mixed- 52

conifer (‘healthy forest’, 59% ± 5%, range = 42% to 85%; occupied middens n = 38 ± 5;

available middens n = 65 ± 6) did not differ from spruce-fir (‘insect-damaged forest’,

53% ± 5%, range = 35% to 81%; occupied middens n = 82 ± 17; available middens n =

153 ± 22; F1,14 = 0.69, P = 0.42).

During the recolonization period (this study), the percent of available middens

occupied within healthy forest (20% ± 3%, range = 15% to 25%; occupied middens, n =

13 ± 1; available middens, n = 65 ± 4) did not differ from insect-damaged forest (14% ±

7%, range 3% to 28%; occupied middens, n = 8 ± 5; available middens, n = 46 ± 11;

Fisher’s exact, P = 0.67). As recent as December 2000, percent of available middens

occupied within healthy forest was 50% (occupied middens, n = 42; available middens, n

= 84).

Habitat selection in areas with less damage

Random and actual ranges and cores differed in percent dead trees (full model:

range, F5,86 = 3.35, P = 0.008; core, F5,86 = 3.54, P = 0.006). Percent dead trees in actual

ranges (47.9% ± 0.8%) and cores (47.3% ± 1.3%) were 4.1% and 6.7% lower than

respective random ranges (52.1% ± 0.8%) and cores (54% ± 1.3%) (range, F1,86 = 12.25,

P < 0.001; core, F1,86 = 13.97, P < 0.001). Except for one male with a winter core of 84% dead trees, actual ranges and cores had less than 69% dead trees (core, n = 45, Range =

12.0 to 64.7%; range, n = 46, Range = 30.9 to 68.5%). Actual ranges and cores had less percent dead trees in all seasons (Figure 2A, ANOVA, all F > 6.6, all P < 0.02), except

summer ranges and winter, which did not differ (ANOVA, all F < 2.3, all P > 0.14). 53

Random and actual ranges and cores differed in percent open area (full model:

range, F5,86 = 8.6, P < 0.001; core, F5,86 = 6.1, P < 0.001). Percent open area in actual

ranges (16.8% ± 0.9%) and cores (13.8% ± 1.2%) were 6.2% and 8.2% lower than

respective random ranges (23.0% ± 0.9%) and cores (22.0% ± 1.2%; range, F1,86 = 23.6,

P < 0.001; core, F1,86 = 23.5, P < 0.001). Actual ranges and cores were less open in all

seasons (Figure 2B, ANOVA, all F > 4.5, all P ≤ 0.07), except in fall (range, F1,15 = 2.5,

P = 0.13).

Random and actual ranges and cores differed in number of live trees (full model:

range, F5,86 = 865.1, P < 0.001; core, F5,86 = 380.3, P < 0.001). Median number of live

trees in actual ranges and cores was 8.3% (95%CI 3.4 to 14.0) and 12.1% (95%CI 4.5 to

20.2) higher than respective random ranges and cores (range: F1,86 = 11.1, P = 0.001;

core: F1,86 = 10.4, P = 0.002). Actual ranges and cores had more live trees in all seasons

(Figure 3a, ANOVA, all F > 5.4, all P < 0.04), except for winter (ANOVA, range, F1,17 =

0.03, P = 0.87; core, F1,17 = 0.12, P = 0.74).

Random and actual ranges and cores did not differ in number of dead trees (full model: range, F5,86 = 1768.4, P < 0.001; core, F5,86 = 739.1, P < 0.001). Median number

of dead trees within actual ranges and cores were 0.3% (95%CI -3 to 3.7) and 2.5%

(95%CI 2.3 to 7.1) lower than respective random ranges and cores (range, F1,86 = 0.03, P

= 0.86; core, F1,86 = 1.1, P = 0.31). Actual ranges and cores had similar numbers of dead

trees in all seasons (Figure 3b, ANOVA, all F < 2.3, all P > 0.14), except spring with few

dead trees (ANOVA, range, F1,25 = 3.8, P = 0.06). 54

Home ranges enlarge

Ranges varied by season but not by type of estimator used, fixed-kernel or MCP

(Table 3, ANOVA: season, F3,427 = 19.08, P < 0.0001; range estimator, F1,427 = 1.38, P =

0.24). Cores varied by season and type of estimator used, with fixed-kernel cores larger than MCP cores (Table 3, kernel core = 0.45 ha, 95% CI 0.35 to 0.55 ha; MCP core =

0.36 ha, 95% CI 0.29 to 0.43 ha; ANOVA, season, F3,427 = 11.53, P < 0.001; range

estimator, F1,427 = 6.79, P = 0.010). With little to no variation in home range estimator,

from this point forward, we report only fixed-kernel ranges and cores.

Ranges varied with season (F3,210 = 9.41, P < 0.001), forest type (F1,210 = 12.48, P

< 0.001), and sex (F1,210 = 4.56, P = 0.034), as did cores (season, F3,210 = 11.03, P <

0.001; forest type, F1,210 = 11.85, P < 0.001; sex, F1,210 = 3.86, P = 0.051). Squirrels

living in healthy forest demonstrated great seasonal variation in ranges and cores;

summer ranges were larger than fall and winter, spring ranges larger than winter, and

summer and spring cores larger than winter and fall (Table 3, ANOVA, range, F3,165 =

6.81, P < 0.001; core, F3,165 = 8.51, P < 0.001; Tukey-Kramer post hoc, all P < 0.05).

Squirrels living in insect-damaged forest demonstrated weaker seasonal variation in range

and core, only summer range was larger than winter range (Table 3, ANOVA, range, F3,41

= 2.68, P = 0.059; core, F3, 41 = 2.58, P = 0.066; Tukey-Kramer post hoc, P < 0.05). In

healthy forest, females had larger ranges and cores than males (Table 4, ANOVA, range,

F1,165 = 6.30, P = 0.013; core, F1,165 = 4.42, P = 0.037), whereas sex did not influence

range or core size in insect-damaged forest (Table 4, ANOVA, range, F1,41 = 0.12, P =

0.73; core, F1,41 = 0.16, P = 0.69). 55

Squirrels living in insect-damaged forest ranged more widely than squirrels living in healthy forest during summer and spring (Table 3, ANOVA, summer range, F1,68 =

13.30, P < 0.001; summer core, F1,68 = 8.50, P = 0.005; spring range F1,56 = 2.97, P =

0.090; other seasons all F < 2.1, all P > 0.15).

Body mass did not differ

Body condition, as indicated by body mass, varied by season in both sexes

(Figure 4; males, F3,193 = 2.9, P = 0.037; females, F3,162 = 7.1, P < 0.001). Females were heavier in insect-damaged forest than in healthy forest (insect-damaged, 232.4g ± 2.1g; healthy, 226.2g ± 1.2g; F1,162 = 5.6, P = 0.019); male body mass did not vary with forest type (insect-damaged, 233.8g ± 1.9g; healthy, 231.6g ± 1.1g; F1,193 = 0.8, P = 0.37).

Demographic differences exist

Neither the ratio of adults:subadults (% adult:% subadult, insect-damaged: 85:15, n = 82; healthy: 85:15, n = 269; Fisher’s exact: P = 1.00), nor males:females (% male:% female, insect-damaged: 55:45, n = 82; healthy: 52:48, n = 269; Fisher’s exact: P =

0.707) differed between forest type.

All females alive during the breeding season within insect-damaged forest (n =

15) and healthy forest (n = 23) lactated at least once. Litter size was similar between females in insect-damaged (n = 7, mean = 2.7 young, 95% CI 1.9 to 3.4 young) and healthy forest (n = 17, mean = 3.1 young, 95% CI 2.6 to 3.6 young; t = -0.91, P = 0.37). 56

All males alive during the breeding season within insect-damaged (n = 20) and healthy

forest (n = 32) were reproductive at least once.

Survivorship was lower in insect-damaged relative to healthy forest (Figure 5,

median days alive since first capture: insect-damaged, 247d, 95%CI 141 to 345d; healthy,

483d, 95% CI 286 to 710d; Log-Rank χ2 = 11.57, df = 1, P < 0.001). Median survival

since first capture was similar between males and females living in insect-damaged forest

(males, n = 20, 211d, 95% CI 115 to 364d; females, n = 15, 247d, 95% CI 99 to 557d;

Log-Rank χ2 = 0.25, df = 1, P = 0.62), and also in healthy forest (males, n = 32, 458d,

95% CI 236 to 747d; females, n = 25, 560d, 95% CI 240 to 1317d; Log-Rank χ2 = 0.58, df = 1, P = 0.45). Reduced survivorship also results in the reduction of the number of

potential reproductive events since first capture; squirrels in the healthy forest (1.93 ±

0.23, n = 13) had nearly double the opportunities to reproduce as squirrels living in

insect-damaged forest (1.25 ± 0.21, n = 16; t27 = -2.09, P = 0.045).

Within insect-damaged forest, days alive since first capture and range or core size

were not related (linear regression: range, F1,12 = 0.026, P = 0.81; core area, F1,12 = 0.057,

P = 0.82); nor were days survived and percent of range and core classified as dead trees

or open area (linear regression, range: % dead, F1,12 = 0.19, P = 0.67, % open F1,12 =

0.043, P = 0.84; core: % dead, F1,12 = 1.41, P = 0.26, % open, F1,12 = 0.18, P = 0.68).

Discussion:

Although Mt. Graham red squirrels select habitat within insect-damaged forests

and reproduce as well as individuals in healthy forest, poor survivorship and reduced 57

potential to reproduce suggest that the insect-damaged forest represents an ecological trap. Insect-damaged forests were essentially unoccupied as of 2002 (Koprowski et al.

2005), and colonizers dispersed from contiguous undamaged forest at low elevations.

Dispersal patterns in small mammals are often sex or age biased, the latter towards subadults (Gaines and McClenaghan 1980, Berteuax and Boutin 2000). Therefore, similar age (adult:subadult) and sex ratios among newly caught individuals in both forests suggest that natal and breeding dispersal are occurring in both forest types. In addition, current occupancy of middens in healthy forest is lower than recent occupancy, suggesting habitat is available in healthy forest during the recolonization. Therefore, available habitat and similar age and sex ratios indicate individuals were not relegated to insect-damaged forest. Although spruce-fir forest was thought the preferred forest type prior to infestation (USFWS 1993), similar midden occupancy, both historically and during recolonization, suggests squirrels equally prefer both forest types, with individuals currently inhabiting insect-damaged forest possibly succumbing to an equal-preference ecological trap (Robertson and Hutto 2006).

While preference may be equal, habitat selection within insect-damaged forest indicates the severity of insect infestation that individuals will tolerate. Squirrels inhabiting insect-damaged forest demonstrated little seasonal variation in habitat selection with nearly all individuals selecting home ranges with <69% dead trees. Food resources likely influence home range selection, as number of live trees was greater than found at random. Additionally, live trees create more canopy cover that affords protection from predators (Hughes and Ward 1993). Selection of greater cover suggests 58

influence of predation risk; red squirrels perceive greater predation risk in and avoid crossing forest clearcuts (Bakker and Van Vuren 2004). However, food resources can also influence selection, as dry open areas offer fewer mushrooms, a secondary food resource for squirrels (Steele 1998). Territory or midden establishment may influence habitat selection at the larger scale of the home range, as squirrels living in insect- damaged forest occupied middens in areas with <64% dead trees (Appendix A). Studies of forest thinning practices suggest similar results with a decline in squirrel density after removal of >50% stem density (Koprowski 2005b); if damage were more severe, squirrels may not find an area habitable. Sites selected within insect-damaged forests have similar vegetation structure to those in healthy forests on Mt. Graham (Smith and

Mannan 1994, Merrick et al. in press) and elsewhere (Finley 1969, Vahle and Patton

1983, Uphoff 1990, Mattson and Reinhart 1997), suggesting that cues for habitat selection remain unchanged.

Squirrels living within insect-damaged forest had larger home ranges than individuals living in healthy forest. Home ranges must satisfy an individual’s food and resource needs (Gurnell 1987); thus home range size may increase when food is less abundant (Lurz et al. 2000, Dussault et al. 2005). Because insect infestation reduced seed crop and basal area of live conifer trees (Koprowski et al. 2005), food abundance is much lower, requiring individuals to cover larger areas to meet energetic demands. Although squirrels in healthy forest demonstrated clear seasonal variation in home range size, individuals in insect-damaged forest had large home ranges year round, with home ranges only reduced in winter when individuals often exhibit subnivean activity restricted about 59

the midden and nest tree (Steele 1998). Fluctuations in red squirrel home ranges are associated with harsh winter weather and seasonal behaviors related to resource use

(Zirul and Fuller 1970, Vlasman and Fryxell 2002, Koprowski et al. 2007); a mild winter may result in little seasonal variation in home range size within insect-damaged forest.

While sexual dimorphism in home range occurs in Mt. Graham red squirrels primarily in mixed-conifer (Koprowski et al. 2007), we found no evidence of sexual dimorphism within insect-damaged forest, as all home ranges were large.

Food resources and high body mass are important for reproductive success

(Wauters and Dhondt 1989, Becker et al. 1998, Boutin et al. 2006), thus decreased seed and epigeous fungi crops in insect-damaged forest (Koprowski et al. 2005) and large home ranges would suggest decreased reproduction within insect-damaged forests.

However, squirrels appear to obtain adequate resources as body masses did not differ between forest types and all females and males were in reproductive condition.

Furthermore, litter size did not differ between forest types. Juvenile survival and dispersal may differ between forest type, but remain unknown. Population density can impact reproductive success (Wauters and Lens 1995); however, density likely had minimal or no impact on reproduction with few squirrels living in insect-damaged forest and the notably low density of this endangered species (USFWS 1993).

While most demographic characteristics did not differ between forest types, squirrels living in insect-damaged forest had low survivorship relative to individuals in healthy forest. Differences in survival can be linked to low food availability leading to decreased body size and condition (Wauters and Dhondt 1989); however, body mass of 60

squirrels was similar in the forest types. Large home ranges may decrease survival by increasing energy expenditure (Smith, C. C. 1968, Covell et al. 1996) and exposure to predators because of greater movement (Martel and Dill 1995); however, we found no relationship between home range size and survival, perhaps due to considerable size of home ranges in the Pinaleños. Raptors are the main predators of Mt. Graham red squirrels (Schauffert et al. 2002) and reduced canopy cover may increase exposure to avian predators, particularly northern goshawks (Accipiter gentilis) that inhabit similar old growth forests and kill in areas with high numbers of dead trees (Greenwald et al.

2005). Although we found no relationship between survival and percent dead trees or percent open area within home ranges, factors influencing predation may occur on a finer scale in both forest types. Regardless of the cause, reduced survivorship reduces the number of potential reproductive events attained by an individual by 50%, thus dramatically reducing individual fitness for squirrels living in insect-damaged forest.

Therefore, equal-preference for both forest types combined with low survival suggest that insect-damaged forest currently represents an ecological trap (Schlaepfer et al. 2002,

Robertson and Hutto 2006) for red squirrels in this habitat isolate.

Habitat selection at small and large scales influences use of insect-damaged areas by forest dwelling species. Abundance of individuals dependent on resources previously provided by infected trees will likely decrease (Matsuoka et al. 2001). Forests with severe insect infestation (>69% tree mortality as suggested by this study) may no longer represent habitat. Although insect-damaged forest may provide cues for habitat selection, low survivorship suggests an ecological trap (Schlaepfer et al. 2002, Robertson and Hutto 61

2006) and potentially a population sink if overall net fitness is negative (Pulliam and

Danielson 1991). While their existence does not necessitate population decline

(Robertson and Hutto 2006), ecological traps are a concern for the persistence of isolated

and/or endangered populations inhabiting insect-damaged forests. In addition, fuel build-

up from fire suppression (Grissino-Mayer and Fritts 1995) combined with large quantities

of dead trees increase risk of wildfire for individuals inhabiting remnant forest patches.

In June and July of 2004 a stand-replacing wildfire severely burned a large portion of the

insect-damaged forest in the Pinaleño Mountains (Koprowski et al. 2006), destroying

habitat not only for red squirrels, but many forest-dwelling species. Climate change is

altering natural disturbance regimes (Dale et al. 2000) and generating asynchronous

relationships between species and their environments (eg. Yellow-bellied marmots,

Marmota flaviventris, emerging early from hibernation causing asynchrony with phenology of food plants; Parmesan 2006). While species may adapt to changing climate, the response of many species is a shift in range distribution (Parmesan 2006).

Populations occupying isolated habitats, those on the tops of mountains and in areas fragmented by habitat loss and degradation, will not have that option and thus face a greater risk of extinction that can be exacerbated by ecological traps created by an array of disturbances.

References:

Alvarez, L. H. R. 2001. Does increased stochasticity speed up extinction? Journal of Mathematical Biology 43:534-544.

Anatrack Ltd. 2003. Ranges 6. Version 1.208. Anatrack Ltd., Wareham, England, UK. 62

Ayres, M. P., and M. J. Lombardero. 2000. Assessing the consequences of global change for forest disturbance from herbivores and pathogens. The Science of the Total Environment 262:263-286.

Bakker, V. J., and D. H. V. Vuren. 2004. Gap-crossing decisions by the red squirrel, a forest-dependent small mammal. Conservation Biology 18:689-697.

Becker, C. D., S. Boutin, and K. W. Larsen. 1998. Constraints on first reproduction in North American red squirrels. Oikos 81:81-92.

Bengtsson, J., S. G. Nilsson, A. Franc, and P. Menozzi. 2000. Biodiversity, disturbances, ecosystem function and management of European forests. Forest Ecology and Management 132:39-50.

Berteuax, D., and S. Boutin. 2000. Breeding dispersal in female North American red squirrels. Ecology 81:1311-1326.

Beyer, H. L. 2006. Hawth's analysis tools for ArcGIS ver. 3.26. Available at http://www.spatialecology.com/htools.

Boutin, S., L. A. Wauters, A. G. McAdam, M. M. Humphries, G. Tosi, and A. A. Dhondt. 2006. Anticipatory reproduction and population growth in seed predators. Science 314:1928-1930.

Carey, A. B. 2000. Effects of new forest management strategies on squirrel populations. Ecological Applications 10:248-257.

Ceballos, G., and P. R. Ehrlich. 2002. Mammal population losses and the extinction crisis. Science 296:904-907.

Covell, D. F., D. S. Miller, and W. H. Karasov. 1996. Cost of locomotion and daily energy expenditure by free living swift foxes (Vulpes velox): A seasonal comparison. Canadian Journal of Zoology 74:283-290.

Dale, V. H., L. A. Joyce, S. McNulty, and R. P. Neilson. 2000. The interplay between climate change, forests, and disturbances. The Science of the Total Environment 262:201-204.

Dale, V. H., L. A. Joyce, S. McNulty, R. P. Neilson, M. P. Ayres, M. D. Flannigan, P. J. Hanson, L. C. Irland, A. E. Lugo, C. J. Peterson, D. Simberloff, F. J. Swanson, B. J. Stocks, and B. M. Wotton. 2001. Climate change and forest disturbances. Bioscience 51:723-734. 63

Dussault, C., R. Courtois, J.-P. Ouellet, and I. Girard. 2005. Space use of moose in relation to food availability. Canadian Journal of Zoology 83:1431-1437.

Ecological Software Solutions. 2003. LOAS version 2.10.1. Ecological Software Solutions, Sacramento, California, USA.

Ewers, R. M., and R. K. Didham. 2006. Confounding factors in the detection of species responses to habitat fragmentation. Biological Reviews 81:117-142.

Finley, R. B. J. 1969. Cone caches and middens of tamiasciurus in the Rocky Mountain region. University Kansas Museum Natural History Miscellaneous Publication 51:233-273.

Flaspohler, D, J., S. A. Temple, and R. N. Rosenfield. 2001. Species-specific edge effects on nest success and breeding bird density in a forested landscape. Ecological Applications 11:32-46.

Gaines, M. S., and L. R. McClenaghan. 1980. Dispersal in small mammals. Annual Review of Ecology and Systematics 11:163-196.

Greenwald, D. N., D. C. Crocker-Bedford, L. Broberg, K. F. Suckling, and T. Tibbitts. 2005. A review of northern goshawk habitat selection in the home range and implications for forest management in the western United States. Wildlife Society Bulletin 33:120-129.

Grissino-Mayer, H. D., and H. C. Fritts. 1995. Dendroclimatology and dendroecology in the Pinaleño Mountains. Pages 100-120 in C. A. Istock and R. S. Hoffmann, editors. Storm over a mountain island: Conservation biology and the Mt. Graham affair. University of Arizona Press, Tucson, Arizona, USA.

Gurnell, J. 1987. The natural history of squirrels. Facts on File Inc., New York, New York, USA.

Hoffmeister, D. F. 1986. Mammals of Arizona. University of Arizona Press, Tucson, Arizona, USA.

Hooge, P. N., and B. Eichenlaub. 2000. Animal movement extension to Arcview, ver. 2.0. Alaska Science Center - Biological Science Office, U.S. Geological Survey, Anchorage, AK, USA.

Hughes, J. J., and D. Ward. 1993. Predation risk and distance to cover affect foraging behavior in Namib Desert gerbils. Animal Behavior 46:1243-1245. 64

Jones, J., R. D. DeBruyn, J. J. Barg, and R. J. Robertson. 2001. Assessing the effects of natural disturbance on a neotropical migrant songbird. Ecology 82:2628-2635.

Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. American Statistical Association Journal 53:457-481.

Kenward, R. E. 1987. Wildlife radio tagging: Equipment, field techniques and data analysis. Academic Press, London, England.

Koprowski, J. L. 2002. Handling tree squirrels with a safe and efficient restraint. Wildlife Society Bulletin 30:101-103.

Koprowski, J. L. 2005a. Annual cycles in body mass and reproduction of endangered Mt. Graham red squirrels. Journal of Mammalogy 86:309-313.

Koprowski, J. L. 2005b. Management and conservation of tree squirrels: the importance of edemism, species richness, and forest condition. Pages 245-250 in G. J. Gottfried, B. S. Gebow, L. G. Eskew, and C. B. Edminster, editors. Connecting mountain islands and desert seas: biodiversity and management of the Madrean archipelago II. U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Tucson, Arizona.

Koprowski, J. L., M. I. Alanen, and A. M. Lynch. 2005. Nowhere to run and nowhere to hide: Response of endemic Mt. Graham red squirrels to catastrophic forest damage. Biological Conservation 126:491-498.

Koprowski, J. L., S. R. B. King, and M. Merrick. 2007. Home ranges are expanded in a peripheral population: space use of endangered Mt. Graham red squirrels. Endangered Species Research. In press.

Koprowski, J. L., K. M. Leonard, C. A. Zugmeyer, and J. L. Jolley. 2006. Direct effects of fire on endangered Mt. Graham red squirrels. Southwestern Naturalist 51:59- 63.

Lande, R. 1998. Anthropogenic, ecological, and genetic factors in extinction and conservation. Researches on Population Ecology 40:259-269.

Lurz, P. W. W., P. J. Garson, and L. A. Wauters. 2000. Effects of temporal and spatial variations in food supply on the space and habitat use of red squirrels (Sciurus vulgaris L.). Journal of Zoology, London 251:167-178.

Martel, G., and L. M. Dill. 1995. Influence of movement by coho salmon (Oncorhynchus kisutch) parr on their detection by common mergansers (Mergus merganser). Ethology 99:139-149. 65

Matsuoka, S. M., C. M. Handel, and D. R. Ruthrauff. 2001. Densities of breeding birds and changes in vegetation in an Alaskan boreal forest following a massive disturbance by spruce beetles. Canadian Journal of Zoology 79:1678-1690.

Mattson, D. J., and D. P. Reinhart. 1997. Excavation of red squirrel middens by grizzly bears in the whitebark pine zone. Journal of Applied Ecology 34:926-940.

McDonough, T. J., and E. Rexstad. 2005. Short-term demographic response of the red- backed vole to spruce beetle infestations in Alaska. Journal of Wildlife Management 69:246-254.

Merrick, M. J., S. R. Bertelsen, and J. L. Koprowski. in press. Characteristics of Mount Graham red squirrel nest sites in a mixed conifer forest. Journal of Wildlife Management.

Millspaugh, J. J., and J. M. Marzluff. 2001. Radio tracking and animal populations. Academic Press, San Diego, California, USA.

Parmesan, C. 2006. Ecological and evolutionary responses to recent climate change. Annual Review of Ecology, Evolution, and Systematics 37:637-669.

Penteriani, V., M. Mathiaut, and G. Boisson. 2002. Immediate species responses to catastrophic natural disturbances: windthrow effects on density, productivity, nesting stand choice, and fidelity in northern goshawks (Accipiter gentilis). The Auk 119:1132-1137.

Pulliam, H. R., and B. J. Danielson. 1991. Sources, sinks, and habitat selection: A landscape perspective on population dynamics. The American Naturalist 137:S50- S66.

Robertson, B. A., and R. L. Hutto. 2006. A framework for understanding ecological traps and an evaluation of existing evidence. Ecology 87:1075-1085.

SAS Institute Inc. 2003. JMP, version 5.1. Cary, North Carolina, USA.

Schauffert, C. A., J. L. Koprowski, V. L. Greer, M. I. Alanen, K. A. Hutton, and P. J. Young. 2002. Interactions between predators and Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis). The Southwestern Naturalist 47:498- 501.

Schlaepfer, M. A., M. C. Runge, and P. W. Sherman. 2002. Ecological and evolutionary traps. Trends in Ecology and Evolution 17:474-480. 66

Smith, A. A., and R. W. Mannan. 1994. Distinguishing characteristics of Mount Graham red squirrel midden sites. Journal of Wildlife Management 58:437-445.

Smith, C. C. 1968. The adaptive nature of social organization in the genus of three squirrels Tamiasciurus. Ecological Monographs 38:31-63.

Sousa, W. P. 1984. The role of disturbance in natural communities. Annual Review of Ecology and Systematics 15:353-391.

Steele, M. A. 1998. Tamiasciurus hudsonicus. Mammalian Species 586:1-9.

Steele, M. A., and J. L. Koprowski. 2001. North American tree squirrels. Smithsonian Institution Press, Washington D. C., USA.

Sullivan, R. M., and T. L. Yates. 1995. Population genetics and conservation biology of relict populations of red squirrels. Pages 193-208 in C. A. Istock, and R .S. Hoffmann, editor. Storm over a mountain island: Conservation biology and the Mt. Graham affair. University of Arizona Press, Tucson, Arizona, USA.

Suring, L. H., M. I. Goldstein, S. Howell, and C. S. Nations. 2006. Effects of spruce beetle infestations on berry productivity on the Kenai Peninsula, Alaska. Forest Ecology and Management 227:247-256.

U. S. Department of Agriculture (USDA). 2000. Forest insect and disease conditions in the southwestern region. Flagstaff, Arizona, USA.

U. S. Fish and Wildlife Service (USFWS). 1987. Endangered and threatened wildlife and plants; determination of endangered status for the Mount Graham Red Squirrel. Federal Register 52:20994-20999.

U. S. Fish and Wildlife Service (USFWS). 1993. Mount Graham red squirrel recovery plan. Phoenix, Arizona, USA.

Uphoff, K. C. 1990. Habitat use and reproductive ecology of red squirrels (Tamiasciurus hudsonicus) in central Arizona. MS thesis. Arizona State University, Tempe.

Vahle, J. R., and D. R. Patton. 1983. Red squirrel cover requirements in Arizona mixed conifer forests. Journal of Forestry 81:14-15.

Vlasman, K. L., and J. M. Fryxell. 2002. Seasonal changes in territory use by red squirrels, Tamiasciurus hudsonicus, and responses to food augmentation. Canadian Journal of Zoology 80:1957-1965. 67

Wauters, L. A., and A. A. Dhondt. 1989. Body weight, longevity, and reproductive success in red squirrels (Sciurus vulgaris). Journal of Animal Ecology 58:637- 651.

Wauters, L. A., and L. Lens. 1995. Effects of food availability and density on red squirrel (Sciurus vulgaris) reproduction. Ecology 76:2460-2469.

White, G. C., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data. Academic Press, Inc., San Diego, California, USA.

Wood, D. J. A., S. Drake, S. P. Rushton, D. Rautenkranz, P. W. W. Lurz, and J. L. Koprowksi. 2007. Fine scale analysis of Mount Graham red squirrel habitat following disturbance. Journal of Wildlife Management. In press.

Zirul, D. L., and W. A. Fuller. 1970. Winter fluctuation in size of home range of the red squirrel (Tamiasciurus hudsonicus). Transaction of the 35th North American Wildlife and Natural Reservation Conference 70:115-127. 68 MCP 0.2 ± 0.03 0.38 ± 0.17 0.38 ± 0.09 0.25 ± 0.04 1.11 ± 0.35 0.45 ± 0.08 0.36 ± 0.13 0.21 ± 0.04 Tamiasciurus Tamiasciurus 50% Core (ha) Kernel 0.50 ± 0.16 0.16 ± 0.03 0.62 ± 0.16 0.38 ± 0.07 1.25 ± 0.45 0.60 ± 0.13 0.33 ± 0.10 0.21 ± 0.06 b MCP 3.57 ± 0.91 0.92 ± 0.12 4.39 ± 1.29 1.74 ± 0.19 7.74 ± 1.75 2.76 ± 0.43 3.25 ± 0.97 1.30 ± 0.29 a 95% Range (ha) Kernel 3.33 ± 0.89 1.24 ± 0.17 4.29 ± 0.98 2.22 ± 0.31 8.77 ± 2.06 3.66 ± 0.62 3.54 ± 1.07 1.54 ± 0.33 ) in insect-damaged and healthy forest in the Pinaleño Mountains, Arizona, winter 2004 to summer Mountains, Arizona, winter and healthy forest in the Pinaleño ) in insect-damaged Insect Non-insect Insect Non-insect Insect Non-insect Insect Non-insect Forest Type Minimum convex polygon b Fixed-kernel, Fixed-kernel, Fall Spring Summer Winter Season a Table 3: Seasonal fluctuations in mean 95% range and 50% core (± SE) for Mt. Graham red squirrels ( Mt. Graham core (± SE) for range and 50% in mean 95% Seasonal fluctuations Table 3: 58; fall, n = 9, n = 14, 45; summer, n = 13, winter, n = 10, 42; spring, (insect damage, non-insect damage): 2005. Sample sizes 25. hudsonicus grahamensis 69

Table 4: Mean 95% range and 50% core (± SE) for male and female Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis) in insect-damaged and healthy forest in the Pinaleño Mountains, Arizona, winter 2004 to summer 2005. Sample sizes (male, female): insect-damaged, n = 26, 20; healthy, n = 98, 72.

95% Range (ha) 50% Core (ha)

Forest Type Male Female Male Female

Insect 4.41 ± 0.70 6.23 ± 1.51 0.59 ± 0.11 0.88 ± 0.30

Non-insect 1.76 ± 0.23 3.20 ± 0.47 0.29 ± 0.06 0.49 ± 0.09 70

Actual range a) Random range 60 Actual core ** Random core ** 55 ** s

e ** ** e r T

d 50 a e D

% 45

40 b) 30 ** ** * 25 ** * ** **

n 20 e p O

15 %

10

5 Winter Spring Summer Fall Season

FIGURE 2. Seasonal fluctuations in mean percent pixels (± SE) classified as a) dead trees and b) open areas, within ranges, cores, and random areas for Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis) in insect-damaged forest in the Pinaleño Mountains, Arizona, winter 2004 to summer 2005. Asterisk (*) denotes 0.05 < P < 0.1, 2 asterisks (**) denote P < 0.05 comparing actual to random. Sample sizes: winter, n = 10; spring, n = 14; summer, n = 13; fall, n = 9. 71

a) 12 Actual range ** Random range Actual core ) s 11 Random core e e

r ** t

e **

v 10 i l

r

e **

b 9 m

u ** n (

n 8

l **

7

12 b) )

s 11 e e r

t *

d

a 10 e d

r e

b 9 m u n ( 8 n l

7 Winter Spring Summer Fall Season

FIGURE 3. Seasonal fluctuations in mean number of pixels (± SE) classified as a) live and b) dead trees, within ranges, cores, and random areas for Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis) in insect-damaged forest in the Pinaleño Mountains, Arizona, winter 2004 to summer 2005. Asterisk (*) denotes 0.05 < P < 0.1, 2 asterisks (**) denote P < 0.05 comparing actual to random. Sample sizes: winter, n = 10; spring, n = 14; summer, n = 13; fall, n = 9. 72

Insect males Non-insect males Insect females 250 Non-insect females

240 ) g (

s 230 s a m

y 220 d o B 210

200 Winter Spring Summer Fall FIGURE 4. Mean body mass (± SE) for adult male and female Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis) in insect-damaged and healthy forest in the Pinaleño Mountains, Arizona, spring 2002 to fall 2005. Sample sizes (insect-damaged, healthy): male: winter = 7, 34; spring = 12, 33; summer = 17, 44; fall = 10, 41; female: winter = 7, 22; spring = 10, 35; summer = 14, 38; fall = 10, 31. 73

1.0

0.9 Non-insect damaged

g 0.8 Insect damaged n i

v 0.7 i v r 0.6 u s

n 0.5 o i

t 0.4 r o

p 0.3 o r

P 0.2 0.1 0.0 0 200 400 600 800 1000 1200 1400 Days since first capture

FIGURE 5. Proportion of Mt. Graham red squirrels (Tamiasciurus hudsonicus grahamensis) surviving since first captured in insect-damaged and healthy forest in the Pinaleño Mountains, Arizona, summer 2003 to summer 2005. Dotted line indicates median number days alive since first capture. Sample sizes: insect-damaged, n = 35; healthy, n = 57.