Skip to content

Advertisement

Open Access

Adult bobcat (Lynx rufus) habitat selection in a longleaf pine savanna

Ecological Processes20187:20

https://doi.org/10.1186/s13717-018-0129-5

Received: 15 September 2017

Accepted: 16 April 2018

Published: 7 May 2018

Abstract

Background

Pine savannas are primarily managed with frequent prescribed fire (≤ 3 years) to promote diversity of flora and fauna, and to maintain open, park-like conditions needed by species such as the endangered red-cockaded woodpecker (Picoides borealis). However, a knowledge gap exists in our understanding of bobcat (Lynx rufus) habitat selection in longleaf pine savannas and research is warranted to direct our future management decisions.

Methods

We examined bobcat habitat selection in a pine savanna managed with frequent fires at two spatial scales (i.e., study area boundary [hereafter, landscape scale]) and annual area of use [95% kernel density; local scale]), and assessed effects of prescribed fire on bobcat habitat selection. Specifically, we monitored 45 bobcats (16 males and 29 females) during 2001–2007.

Results

We found differential habitat selection by sex. At the landscape scale, female bobcats were closer to mixed pine-hardwoods, young pine, and secondary roads, but farther from mature pine and hardwoods stands relative to males. We found no difference in selection of agriculture, shrub-scrub, and primary roads between sexes. At the annual area of use scale, female bobcats were closer to secondary roads, but farther from agriculture and shrub-scrub relative to males. We found no difference in selection of mature pine, mixed pine-hardwoods, hardwoods, young pine, and primary roads between sexes. Bobcats primarily selected for stands burned ≤ 1.1 years post-fire.

Conclusions

Our results show that bobcats exploit a broad range of habitat types in pine landscapes managed with frequent fire and commonly use recently burned stands (≤ 1.1 year post-fire), suggesting prey in many areas of this system are at risk of bobcat predation. Additionally, we suggest land managers consider scale of selection by bobcats when developing habitat management strategies.

Keywords

GeorgiaHabitat selection Lynx rufus Pinus palustris Prescribed fire

Background

Longleaf pine (Pinus palustris) savannas historically occupied over 30 million hectares in the southeastern USA (Brockway et al. 2005, Van Lear et al. 2005). This unique ecosystem is considered one of the most biologically diverse in North America because it supports hundreds of species of flora and fauna (Alavalapati et al. 2002). Longleaf pine savannas were shaped over millennia by frequent, low-intensity fires ignited by man and lightning (Brockway et al. 2005, Van Lear et al. 2005). Today, longleaf pine savannas are primarily managed by frequent prescribed fires (≤ 3 years return interval) to reduce undesirable vegetation while stimulating growth and development of a diverse understory (Waldrop et al. 1992, Cain et al. 1998, Barnett 1999, Steen et al. 2013). Frequent fire enables pine savannas to support an abundant diversity of flora and fauna (Alavalapati et al. 2002, Van Lear et al. 2005). For example, approximately 40% of the 1600+ plant species (including many rare species) found in the Atlantic and Gulf coastal plains are found in longleaf pine savannas (Walker 1998).

Fire is the primary management tool in longleaf pine savannas and can increase understory plant species richness, diversity, and evenness (Brockway and Lewis 1997). Without fire disturbance, hardwood encroachment reduces plant diversity (Landers et al. 1995, Glitzenstein et al. 2012). Fire also benefits fauna found in this system including common game birds (e.g., eastern wild turkey [Meleagris gallopavo silvestris], northern bobwhite quail [Colinus virginianus]) and species of conservation concern (e.g., red-cockaded woodpeckers [Picoides borealis], gopher tortoise [Gopherus Polyphemus]). Frequent fire can also reduce foraging habitat for common avian nest predators such as raccoons (Procyon lotor; Jones et al. 2004), and thus, conservation practices that reduce use by predators may result in increased nest success.

Bobcats are the most widely distributed wild felid in North America, and population abundance is increasing in many regions (Roberts and Crimmins 2010). Additionally, bobcats are a top carnivore and a vital component of most terrestrial ecosystems in the USA. The role bobcats play as a predator of game animals is still poorly understood in many systems. For example, what role do bobcats play in ecosystem function including direct and indirect effects (i.e., predation risk) on game and non-game populations? Additionally, their value as a furbearer requires management attention, especially in the absence of other larger carnivores such as mountain lions (Felis concolor) and wolves (Canis lupus). Bobcats are also closely tied to their prey and have previously been recommended to serve as indicators of ecosystem health (Conner and Leopold 1996). However, further research is needed to understand bobcat ecology in pine savannas.

Bobcat space use is influenced by prey abundance, season, breeding behaviors, and intraspecific relationships (Chamberlain et al. 2003). Bobcats commonly select mature pine, young pine, hardwood, and agricultural habitat types albeit at different spatial scales (Conner and Leopold 1996, Chamberlain et al. 2003, Godbois et al. 2003a), and prey abundance has been shown to influence bobcat habitat selection (Miller and Speake 1978, Maehr and Brady 1986, Conner and Leopold 1996, Chamberlain et al. 2003, Godbois et al. 2003a). For example, bobcat prey items (e.g., small mammals; hispid cotton rat [Sigmodon hispidus]) were most abundant in early successional habitats containing woody debris and significant herbaceous growth, and least abundant in pine stands with high basal area and little ground-level debris (Chamberlain 1999). Coincidently, bobcats often select habitat types that support abundant small mammal communities (Conner and Leopold 1996, Chamberlain 1999). Additionally, hardwood stands are commonly used by bobcats for refugia (i.e., den sites, cover, protection from summer heat; [Hall and Newsome 1976, Godbois et al. 2003a]) or travel corridors between foraging patches (Godbois et al. 2003a), and are particularly important in pine-dominated landscapes (Conner et al. 1992, Conner and Leopold 1996, Chamberlain et al. 2003). Roads and other linear features are also considered important to bobcats as they also serve as travel corridors (Lovallo and Anderson 1996).

Multi-scale studies are important for understanding how patterns and processes interact and operate on the landscape and influence animal behaviors. Previous research has indicated that bobcat habitat selection occurs at multiple spatial scales (Conner and Leopold 1996, Chamberlain et al. 2003, Godbois et al. 2003a). For example, Godbois et al. (2003a) found bobcats selected mixed pine-hardwood habitat at the landscape scale, while selecting for agriculture and hardwoods at the home range scale. Fire can also influence wildlife habitat selection by altering vegetation structure (e.g., Harper et al. 2016); however, little is known about how time-since-fire influences bobcat habitat selection. For example, small mammals (e.g., cotton rats [Sigmodon hispidus]) are likely exposed to predation from bobcats following a fire event due to reductions in herbaceous cover (Conner et al. 2011, Morris et al. 2011). However, in a pine savanna ecosystem managed by frequent fire (≤ 3 years) and at small scales (≤ 100 ha), fire may have minimal influence on bobcat selection due to the diversity of cover created by frequent fire. Improving our knowledge of how these fires influence bobcat habitat selection can enable land managers to make informed management decisions, especially in cases where biodiversity management is a primary objective. Therefore, our objective was to examine bobcat habitat selection in a longleaf pine savanna managed by frequent-fire.

Methods

Study area

We conducted our research on the 11,735 ha privately owned Joseph W. Jones Ecological Research Center at Ichauway (hereafter, Jones Center; Fig. 1) located in Baker County, Georgia. The Jones Center contains a diversity of habitat types including a variety of forested cover such as longleaf pine, loblolly pine (Pinus taeda), slash pine (P. elliottii), mixed-pine and hardwood forests, oak barrens, lowland hardwood hammocks, and cypress-gum (Taxodium ascendens-Nyssa biflora) limesink ponds (Boring 2001). The Ichawaynochaway Creek bisects the property and the Flint River borders the property to the southeast. The Jones Center was historically managed as a northern bobwhite quail hunting plantation. Today, the Jones Center is an ecological research site while also serving as a quail hunting plantation. The Jones Center was comprised of approximately 31.2% mixed-pine hardwood, 31.1% mature pine (> 20 years old), 11.2% agriculture/food plot, 9.8% young pine (< 20 years old), 9.8% hardwoods, 2.6% open water, 1.8% wetlands, 1.5% shrub-scrub, and 0.9% urban/barren (Fig. 2). Wiregrass and old-field grasses (e.g., Andropogon spp.) were the dominant understory habitat in pine and pine/hardwood stands (Goebel et al. 1997), but > 1000 vascular plant species occur on the site (Drew et al. 1998). Road density was 5.48 km/km2. Total annual rainfall during the study period ranged from 95.9 to 154.8 cm (Jones Center; Georgia Automated Environmental Monitoring Network, http://georgiaweather.net).
Figure 1
Fig. 1

Study area boundary. The dashed line shows the spatial extent of the Joseph W. Jones Ecological Research Center at Ichauway located in southwestern Georgia, USA, and the solid line shows the spatial extent of available habitat used in the study area scale habitat analysis

Figure 2
Fig. 2

Vegetation communities located at the Joseph W. Jones Ecological Research Center at Ichauway located in southwestern Georgia, USA

We classified stands as pine if they consisted of > 90% loblolly, longleaf, slash, and/or shortleaf (P. echinata) pine. We further classified stands as mature pine if trees were > 20 years old and were in large pole (12.6–25.4 cm) or saw timber (>25.4 cm) size classes. We classified stands as young pine if trees were ≤ 20 years old and were considered seedling-sapling or small pole timber (< 12.6 cm).

Mixed pine-hardwood stands contained a variety of species (e.g., loblolly, longleaf, slash pine, southern red oak [Quercus falcata], turkey oak [Q. laevis], live oak [Q. virginiana], laurel oak [Q. laurifolia], and sweetgum [Liquidambar styraciflua]). We classified stands as mixed pine-hardwoods if they were 50–89% hardwoods or pine. Within these stands, tree sizes ranged from seedling-sapling through saw timber; however, most (> 90%) were in the pole to saw timber classes.

We classified stands as hardwoods if they consisted of > 90% hardwood species such as southern red, turkey, live, and laurel oaks, and sweetgum) and trees ranged in size from seedling-sapling to saw timber. The majority of these stands were in pole to saw timber classes (> 90%). Agricultural stands consisted of cropland, pastureland, wildlife food plots, and horticultural crops (e.g., pecan orchard). Shrub-scrub stands consisted of abandoned agricultural fields and pastures, clear-cuts, grassland, and shrubby areas.

To successfully restore and maintain pine savannas on our study sites, land managers used prescribed fire and mechanical hardwood removal. Fire was applied to mature pine, young pine, mixed pine/hardwood, and shrub-scrub habitats. Prescribed fire was conducted throughout the year with > 50% of burns conducted during the growing season (15 April–15 August). Prescribed fire application occurred in a mosaic fashion, which promoted landscape diversity. Average patch size burned at the Jones Center during the study was 35.60 ha (range = 0.01–373.68 ha). Fire-return interval typically ranged from 1 to 3 years with an average fire return interval ≤ 2 years. Land managers often used mechanical removal to remove large off-site hardwoods (e.g., water oak [Q. nigra]) from within mature pine stands.

Bobcat capture and monitoring

We captured adult bobcats using Victor No. 1.75 offset and No. 3 Soft Catch traps (Woodstream Corp., Litiz, Pennsylvania) during December 2000–May 2007. We fitted adults with a 180-g VHF radiocollar (Advanced Telemetry Systems, Isanti, MN), and each individual bobcat received a uniquely numbered ear tattoo. Additional information regarding the capture and tagging of bobcats can be found in Godbois et al. (2003a). The Institutional Animal Care and Use Committee at the University of Georgia approved trapping and handling procedures (Animal Use Proposal no. A990159). Georgia Department of Natural Resources–Wildlife Resources Division issued the scientific collection permit (no. 29-WJH-16-190).

We monitored bobcats from 6 times/day to 4–6 times per week depending on the season using a 3-element Yagi antenna (Sirtrack, New Zealand) and hand-held receiver (Wildlife Materials, Carbondale, IL). Our telemetry data set was compiled from five separate research projects on the same study area, which led to the differences in telemetry sampling. Due to potential observer biases, we conducted an accuracy test for all field personnel and the standard deviation associated with bearing accuracy was 7° (Kirby et al. 2010). To reduce additional error with VHF locations, we limited time between consecutive bearings to < 15 min, though most consecutive bearings were taken within 7 min (White and Garrott 1990).

Habitat selection

To investigate bobcat habitat selection, we examined selection at two spatial scales: landscape (i.e., study area boundary) and annual area of use scale (hereafter, AAU). Bobcats frequently moved off the study area during the study period. Therefore, to assess landscape scale habitat selection, we calculated the median linear distance of all bobcat locations occurring outside of the Jones Center boundary to the property boundary line, which resulted in a median distance from the study area boundary of 237 m. We then buffered the Jones Center boundary by 237 m (i.e., available habitat; see Fig. 1) and excluded all bobcat locations occurring outside of this boundary. This resulted in the loss of 5.3% of locations from our master data set, which we felt was an acceptable loss as some of the locations were likely erroneous fixes due to triangulation errors. To assess AAU habitat selection, we calculated 95% fixed kernel utilization distributions (KDE) in Geospatial Modeling Environment (Ver. 0.7.4; Beyer 2012). Due to a lack of defined breeding and kitten rearing seasons on our study area (L. M. Conner, Joseph W. Jones Ecological Research Center, unpublished data), we chose to focus on annual habitat selection. We required each bobcat to have ≥ 40 locations during each year and ≥ 6 months of telemetry locations to calculate AAU. We clipped the 95% AAU to the available study area boundary because we were only interested in bobcat habitat selection in the fire-maintained pine savanna, and we lacked fire data off-site.

To quantify habitat selection, we used a geographic information system (ArcGIS® 10.3.1, Environmental Systems Research Institute Inc., Redlands, CA) to create a vector layer of habitat types available on our study area. We converted the vector layer to a 30-m raster layer and used the Euclidean Distance tool in ArcGIS® 10.3.1 to calculate distance (m) from every 30-m pixel to the nearest patch of each habitat type. To evaluate the influence of roads as travel corridors, we converted a digitized road layer consisting of county (paved), primary (graded and dirt), secondary (harrowed), and tertiary (mowed and firebreaks) roads to a 30-m raster grid. We combined road classes into two categories based on traffic-levels (L.M. Conner, Joseph W. Jones Ecological Research Center, personal observation): (1) primary roads (county and primary; paved or sand/dirt roads); and (2) secondary roads (firebreaks). Primary roads were defined as roads that received vehicle travel daily, whereas, secondary roads (firebreaks) rarely received vehicle travel and served as an edge between vegetation communities. We used a distance-based approach because distance-based metrics are not restricted to linear or point habitat features, require no explicit error handling, and permit extraction of more information than classification-based analyses (Conner et al. 2003). At the landscape scale, we evaluated non-random habitat selection with a ratio of 1 bobcat location to 5 available (random) locations to characterize available habitat at a larger extent (Northrup et al. 2013). At the AAU scale, we characterized available habitat using a ratio of 1 bobcat location to 3 available (random) locations due to the smaller spatial extent. Finally, we extracted values from the raster habitat variables to both used and available locations.

Model development and analysis

We modeled habitat selection at each scale using a generalized linear mixed-effects model (GLMM) implemented in program R (V. 3.3.3, R Core Team 2017) and a use vs. availability habitat selection approach to evaluate non-random habitat selection (Manly et al. 2002). We compared used (bobcat locations) to available (random) locations in a logistic regression framework where used and random locations were represented as a binary response (1 = bobcat location; 0 = random location). We included sex (male and female) as an interaction term with each stand type in the model to evaluate whether habitat selection differed between sexes. Males were the reference group for the interaction. We included bobcat identification and year as random effects to account for variability among individual bobcats and year-to-year variation in habitat selection, respectively (Gillies et al. 2006).

Prior to data analysis, we scaled all distance values for used and available locations by dividing the linear distance by 200 m to reduce model convergence issues. We evaluated pairwise correlations between explanatory variables at each scale using Pearson correlation. We considered any variables that were highly correlated (|r| > 0.7) and retained the variable that provided the simplest biological interpretation (Dormann et al. 2013). We then evaluated variance inflation factors of all variables to assess the extent of any remaining collinearity. All variables contained a variance inflation factor ≤ 1.12, which suggested that collinearity was not likely an issue (Zuur et al. 2009). We then constructed a full model for each scale. We made inference to only those variables that were statistically significant (α = 0.05). For easier interpretation, we calculated scaled odds ratios (OR) and associated 95% confidence intervals for parameter estimates. We only considered parameter estimates with 95% confidence intervals that excluded 0 to be informative (Miller and Conner 2007). Lastly, we ranked the absolute value of each standardized beta coefficient estimate from largest to smallest to assess relative importance of each landscape variable (Hamilton 1992).

Habitat selection in relation to prescribed fire

To evaluate whether prescribed fire influenced bobcat habitat selection, we used a subset of our radio-telemetry data that included only those bobcat locations occurring on our study area during the growing season (15 April–15 August) and in stands that received frequent fire (i.e., mature pine, young pine, and mixed pine-hardwoods). We chose to focus the analysis during this period because it corresponded to the growing season when most prescribed fire was applied and prescribed fire may be a useful management tool to reduce incidental encounters between bobcats and prey species (e.g., Jones et al. 2004). Additionally, we used only locations within our study area boundary because fire history data were not available outside of this area. We used the union tool in ArcGIS® 10.2 to create a days-since-fire (no. of days) map for the study area. We determined the radius of the buffer by calculating the median 100% daily area of use minimum convex polygon (MCP) for bobcats from 15 April to 15 August using Geospatial Modeling Environment. We required a minimum of six telemetry locations/bobcat/day from 15 April to 15 August to calculate a daily area of use MCP (n = 9 males; n = 14 females). These 23 individuals were a subset of the larger data set because they met the minimum required number of locations/bobcat/day. We calculated an MCP rather than a KDE because the average number of locations per bobcat/day was 7.6 (range 6–9), which would likely lead to over-smoothing of the kernel density estimate (Seaman and Powell 1996). We placed a buffer (170 m radius) based on the median 100% daily area of use minimum convex polygon (MCP) around each used telemetry location to delineate available habitat. Within each buffer, we generated five random points that intersected with the days-since-fire map. We calculated a days-since fire value for each used and random location by subtracting the day we obtained the location from the day the fire occurred for that patch.

We modeled effect of days-since-fire on habitat selection by using a generalized linear mixed-effects model (GLMM) in program R. We included bobcat identification and year as random effects to account variability among individual bobcats and year-to-year variation in habitat selection, respectively (Gillies et al. 2006).

Results

We monitored 64 adult bobcats (27 males and 37 females) during 2001–2007. We removed 19 bobcats (12 males and 7 females) prior to analysis because they contained less than 40 locations during each year and < 6 months of telemetry locations. Our final dataset contained 45 bobcats (16 males and 29 females; Table 1) and generated 144 AAU. We found no highly correlated variables at landscape and AAU scales; therefore, we retained all variables in our modeling efforts. At the landscape scale, female bobcats were closer to mixed pine-hardwoods (β = − 0.172, P < 0.001), young pine (β = − 0.083, P < 0.001), and secondary roads (β = − 0.165, P < 0.001), whereas females were farther from mature pine (β = 0.085, P < 0.001) and hardwoods (β = 0.119, P < 0.001) relative to males. We found no difference in selection of agriculture (β = 0.011, P = 0.723), shrub-scrub (β = 0.001 P = 0.964), and primary roads (β = 0.018, P = 0.284) relative to males.
Table 1

Mean number (and SE) of radio-telemetry locations/bobcat/year/sex

Sex

Year

n a

\( \overline{x} \) number of locations/bobcat

SE

Male

2001

5

104.8

4.4

2002

7

201.9

23.8

2003

8

136.9

11.1

2004

6

116.0

21.6

2005

5

125.8

6.1

2006

7

62.6

2.4

2007

6

60.3

4.1

Female

2001

7

104.7

9.6

2002

15

222.1

21.0

2003

16

147.1

10.4

2004

20

131.1

8.1

2005

18

118.3

6.3

2006

14

66.7

1.3

2007

10

65.1

1.5

an, number of radio-marked bobcats

This research was conducted at the Joseph W. Jones Ecological Research Center, southwestern Georgia, USA, 2001–2007

Because of differences in habitat selection by sex at the landscape scale, we developed two separate sex-specific models and provided sex-specific parameter estimates (Table 2; Fig. 3a–h). Based on standardized coefficient estimate rankings at the landscape scale, distance to primary road was the most important landscape variable to male bobcats followed by young pine, agriculture, mature pine, mixed pine-hardwood, and shrub-scrub; distance to nearest secondary road was least important (Table 2). Specifically, male bobcats were closer to mature pine (β = − 0.289, SE = 0.026, OR = 0.749), mixed pine-hardwoods (β = − 0.262, SE = 0.035, OR = 0.769), hardwoods (β = − 0.268, SE = 0.024, OR = 0.765), agriculture (β = − 0.349, SE = 0.026, OR = 0.705), shrub-scrub (β = − 0.097, SE = 0.013, OR = 0.907), and primary roads (β = − 0.241, SE = 0.014, OR = 0.786), and farther from young pine (β = 0.139, SE = 0.010, OR = 1.150). Male bobcat selection of secondary roads did not occur (P > 0.05; Table 2). At the landscape scale, distance to nearest primary road was the most important landscape variable to female bobcats followed by agriculture, mixed pine-hardwood, mature pine, shrub-scrub, hardwoods, and young pine; distance to nearest secondary road was least important (Table 2). Specifically, female bobcats were closer to mature pine (β = − 0.204, SE = 0.026, OR = 0.816), mixed pine-hardwoods (β = − 0.434, SE = 0.024, OR = 0.648), hardwoods (β = − 0.149, SE = 0.014, OR = 0.862), agriculture (β = − 0.338, SE = 0.017, OR = 0.713), shrub-scrub (β = − 0.097, SE = 0.008, OR = 0.908), primary roads (β = − 0.223, SE = 0.009, OR = 0.800), and secondary roads (β = − 0.139, SE = 0.023, OR = 0.870).
Table 2

Parameter estimates for bobcat habitat selection at the landscape spatial scale

Variablea

Sex

β

SE

Z b

P

Scaled odds ratioc

Scaled lower 95%

Scaled upper 95%

Mature pine

Male

− 0.289

0.026

− 11.104

< 0.001

0.749

0.712

0.788

Female

− 0.204

0.016

− 13.106

< 0.001

0.816

0.791

0.841

Mixed pine/hardwoods

Male

− 0.262

0.035

− 7.463

< 0.001

0.769

0.718

0.824

Female

− 0.434

0.024

− 18.440

< 0.001

0.648

0.619

0.678

Hardwoods

Male

− 0.268

0.024

− 11.384

< 0.001

0.765

0.731

0.801

Female

− 0.149

0.014

− 10.334

< 0.001

0.862

0.838

0.886

Young pine

Male

0.139

0.010

14.524

< 0.001

1.150

1.128

1.171

Female

0.056

0.007

8.590

< 0.001

1.057

1.044

1.071

Agriculture

Male

− 0.349

0.026

− 13.365

< 0.001

0.705

0.670

0.743

Female

− 0.338

0.017

− 20.251

< 0.001

0.713

0.690

0.737

Shrub/scrub

Male

− 0.097

0.013

− 7.261

< 0.001

0.907

0.884

0.931

Female

− 0.097

0.008

− 11.433

< 0.001

0.908

0.893

0.923

Primary roads

Male

− 0.241

0.014

− 16.795

< 0.001

0.786

0.764

0.808

Female

− 0.223

0.009

− 24.755

< 0.001

0.800

0.786

0.815

Secondary roads

Male

0.026

0.036

0.713

0.476

1.026

0.956

1.101

Female

− 0.139

0.023

− 5.989

< 0.001

0.870

0.831

0.911

aDistance to nearest habitat patches (m)

bStandardized coefficient estimates

cScalar, 200 m

This research was conducted at the Joseph W. Jones Ecological Research Center, southwestern Georgia, USA, 2001–2007

Figure 3
Fig. 3

Relative probability of selection at the landscape and annual area of use spatial scales This research was conducted at the Joseph W. Jones Ecological Research Center, southwestern Georgia, USA, 2001–2007. Landscape scale predictive probability figures: mature pine (a), mixed pine-hardwoods (b), hardwoods (c), young pine (d), agriculture (e), shrub-scrub (f), primary roads (g), and secondary roads (h). Annual area of use predictive probability figures: mature pine (i), mixed pine-hardwoods (j), hardwoods (k), young pine (l), agriculture (m), shrub-scrub (n), primary roads (o), and secondary roads (p)

Table 3

Parameter estimates for bobcat habitat selection at the annual area of use spatial scale

Variablea

Sex

β

SE

Z b

P

Scaled odds ratioc

Scaled lower 95%

Scaled upper 95%

Mature pine

Male

− 0.079

0.029

− 2.741

0.006

0.924

0.873

0.978

Female

− 0.078

0.017

− 4.469

< 0.001

0.925

0.894

0.957

Mixed pine/hardwoods

Male

− 0.026

0.040

− 0.646

0.518

0.975

0.902

1.053

Female

− 0.081

0.027

− 3.019

0.003

0.923

0.876

0.972

Hardwoods

Male

− 0.140

0.025

− 5.671

< 0.001

0.869

0.828

0.912

Female

− 0.111

0.015

− 7.257

< 0.001

0.895

0.868

0.922

Young pine

Male

0.040

0.011

3.649

< 0.001

1.041

1.019

1.063

Female

0.023

0.007

3.319

0.001

1.023

1.009

1.037

Agriculture

Male

− 0.280

0.027

− 10.562

< 0.001

0.756

0.717

0.796

Female

− 0.169

0.017

−9.764

< 0.001

0.845

0.816

0.874

Shrub/scrub

Male

− 0.070

0.015

− 4.783

< 0.001

0.932

0.906

0.960

Female

− 0.009

0.009

− 1.025

0.305

0.991

0.973

1.009

Primary roads

Male

0.005

0.016

0.290

0.772

1.005

0.973

1.038

Female

0.004

0.010

0.412

0.680

1.004

0.984

1.025

Secondary roads

Male

0.049

0.037

1.333

0.183

1.051

0.977

1.130

Female

− 0.072

0.025

− 2.919

0.004

0.931

0.887

0.977

aDistance to nearest habitat patches (m)

bStandardized coefficient estimates

cScalar, 200 m

This research was conducted at the Joseph W. Jones Ecological Research Center, southwestern Georgia, USA, 2001–2007

At the AAU scale, female bobcats were closer to secondary roads (β = − 0.122, P = 0.006), but farther from agriculture (β = 0.092, P = 0.003) and shrub-scrub (β = 0.053, P = 0.002) relative to males. We found no difference in selection of mature pine (β = − 0.005, P = 0.873), mixed pine-hardwoods (β = − 0.049 P = 0.309), hardwoods (β = 0.030, P = 0.310), young pine (β = − 0.008, P = 0.526), and primary roads (β = − 0.002, P = 0.930).

Because of differences in habitat selection by sex at the AAU scale, we developed two separate sex-specific models and provided sex-specific parameter estimates (Table 3; Fig. 3i–p). At this scale, distance to agriculture was the most important landscape variable to male bobcats followed by hardwoods, shrub-scrub, young pine, mature pine, secondary roads, and mixed pine-hardwoods; distance to nearest primary road was least important (Table 3). Specifically, male bobcats were closer to mature pine (β = − 0.079, SE = 0.029, OR = 0.924), hardwoods (β = − 0.140, SE = 0.025, OR = 0.869), agriculture (β = − 0.280, SE = 0.027, OR = 0.756), shrub-scrub (β = − 0.070, SE = 0.015, OR = 0.932), and farther from young pine (β = 0.040, SE = 0.011, OR = 1.041). Selection of mixed pine-hardwoods, primary roads, and secondary roads did not differ (P > 0.05; Table 3) from expectation. Distance to agriculture was the most important landscape variable to female bobcats followed by hardwoods, mature pine, young pine, mixed pine-hardwoods, secondary roads, and shrub-scrub; distance to nearest primary road was least important (Table 3). Specifically, female bobcats were closer to mature pine (β = − 0.078, SE = 0.017, OR = 0.925), mixed pine-hardwoods (β = − 0.081, SE = 0.027, OR = 0.923), hardwoods (β = − 0.111, SE = 0.015, OR = 0.895), agriculture (β = − 0.169, SE = 0.017, OR = 0.845), secondary roads (β = − 0.072, SE = 0.025, OR = 0.931), and farther from young pine (β = 0.023, SE = 0.007, OR = 1.023). Selection of shrub-scrub and primary roads did not differ (P > 0.05; Table 3) from expectation.

Days-since-fire did not influence male and female bobcat use of pine stands that received frequent fire (β = 1.303 e-05, SE = 4.372 e-05, P = 0.766). Median use of stands was 404.0 ± 8.6 days after fire, whereas expected use was 402.0 ± 3.8 days.

Discussion

Male and female bobcats selected vegetation communities differently based on scale. Female bobcats presumably establish their home ranges to ensure sufficient resources for individual survival and survival of their young (Anderson 1987, Sandell 1989) compared to males that likely establish their home range based on breeding opportunities (Sandell 1989). Males and females generally selected mature pine and mixed pine-hardwoods, which is contrary to previous studies in pine-dominated systems (Conner and Leopold 1996, Chamberlain et al. 2003). However, these studies were conducted in systems that were not managed by frequent fire. For example, Conner et al. (1992) reported that mature pine and mixed pine-hardwood stands were not managed by frequent fire, resulting in prominent mid-stories, little herbaceous understory, and few bobcat prey. In a system managed by frequent fire return intervals (≤ 3 years), herbaceous plant communities do not shift to dense hardwood understory communities (Glitzenstein et al. 2012). Therefore, these systems contain diverse understory plant communities (Brockway and Lewis 1997), which can produce abundant small mammal populations (Golley et al. 1965, Schnell 1968, McMurry et al. 1994, Masters et al. 1998). We suspect bobcats selected mature pine and mixed pine-hardwoods due to the availability of prey (see also Sasmal et al. 2017); however, we lack data on prey availability among the vegetation types on our study area.

We found that male and female bobcats were farther from young pines, which is also contrary to previous studies (Conner and Leopold 1996, Chamberlain et al. 2003). We expected bobcats to select for areas with lower basal areas managed by frequent fire that contained an abundance of understory vegetation. Selection of young pines in the Mississippi studies was primarily due to relative abundance of small mammals in young pines (Conner 1991, Conner et al. 1992). Many young pines in our study area contained little herbaceous vegetation and few small mammals (Golley et al. 1965, Schnell 1968, McMurry et al. 1994, Masters et al. 1998), which may explain why bobcats were farther from young pines.

Bobcats were found to be closer to agriculture fields and shrub-scrub vegetation communities at both spatial scales. Our findings were consistent with previous studies that found agricultural areas were important to bobcats (Conner et al. 1992, Conner and Leopold 1993, Godbois et al. 2003a). Hardwoods were also important to bobcats; perhaps because they serve as travel corridors between forage patches (Godbois et al. 2003a). Likewise, hardwood stands may also serve as refugia (i.e., den sites, cover, and protection from summer heat; [Hall and Newsome 1976, Godbois et al. 2003a]) within otherwise pine-dominated systems.

Our interpretation of results differed based on sex and the scale examined, suggesting that scale acts as a hierarchy distinguishing broad-level population questions from more fine-scale activity patterns for bobcats. Our findings suggest that primary roads are important to bobcats at the landscape scale but their use of primary roads was random within the AAU scale. Bobcats commonly use roads as travel corridors (Lovallo and Anderson 1996, Conner and Leopold 1998), and roads provide valuable edge habitats (Frey and Conover 2006, Barding and Nelson 2008). Use of primary roads may provide others benefits to bobcats such as reduction in energetic costs related to space use. The lack of importance of primary roads at the AAU scale may be due to increased risk of mortality (e.g., Benson et al. 2015) or availability of other important habitat types or edge. Secondary roads were of minimal importance at both spatial scales, which is likely due to their structure as these roads were primarily disked roads separating pine stands and serving as firebreaks. These roads received rarely received vehicle traffic and maintenance; therefore, they may not have provided enough edge habitat to benefit bobcats.

Days-since-fire did not appear to influence bobcat habitat selection. However, our results may be partially due to a lack of stands with longer burn rotations (> 3 years). With frequent fire-return intervals, herbaceous plant communities do not shift to dense hardwood understory communities (Glitzenstein et al. 2012). In addition, frequent use of prescribed fire in longleaf pine forests increases understory plant species richness, diversity, and evenness (Brockway and Lewis 1997), which may influence where bobcats search for prey items. Small mammals (e.g., cotton rats [Sigmodon hispidus], Conner et al. 2011, Morris et al. 2011) are commonly exposed to predation from bobcats and other predators following a fire event due to reductions in herbaceous cover. Our results may also be due to the scale of fire relative to bobcat home ranges. Bobcat home range size varies from 2.76 to 36.5 km2 (Conner et al. 1992, Cochrane 2003, Doughty 2004). With the average patch size burned on our study area < 40 ha in size, our findings suggest that small scale (< 40 ha) frequent fires (≤ 3 years) may have minimal influence on bobcat habitat selection. Additionally, Sasmal et al. (2017) found no difference in abundance of Peromyscus spp. among five vegetation types and three burn regimes (1, 2, and 3 years post-fire) in pine savannas, suggesting bobcats likely do not have to move far to locate available prey in a system managed by frequent fire. Lastly, bobcats were monitored from 6 times a day to 4–6 times per week depending on the season. The infrequency of locations may have led to biased results in cases where bobcats rapidly exploited a burned area and moved away before the subsequent telemetry location(s). We suggest future research evaluate bobcat foraging strategies in pine savannas to potentially reduce incidental encounters between bobcats and ground-dwelling prey species. For example, Kaufmann et al. (2007) found landscape heterogeneity shaped predation with specific landscape features and vegetation structure influencing the probability of detection between gray wolves (Canis lupus) and elk (Cervus elaphus). We also suggest future research collect finer-temporal data using GPS technology to improve our understanding of how rapidly bobcats locate and exploit recently burned areas in fire-maintained pine savannas and examine the influence of larger scales of fire (e.g., > 40 ha) on bobcat habitat selection.

Conclusions

Our findings demonstrate the importance of vegetation diversity (e.g., mature pine, mixed pine/hardwoods, hardwoods, agriculture/food plots, shrub-scrub) for bobcats in a forested landscape managed by frequent (≤3 years) and small scale (< 40 ha) fires. We found no apparent fire-associated selection at the smaller spatial scale, which is likely due to frequency of telemetry locations, and scale and frequency of fire return intervals. Interestingly, at larger spatial scales, frequent burns led to selection patterns contrary to our current knowledge of bobcat habitat selection, which are based on studies where frequent fire was absent. For example, bobcats generally selected mature pine and mixed pine-hardwoods, which were managed by frequent fire in our system. However, prior studies found bobcats avoided these vegetation communities primarily due to the absence of herbaceous understory vegetation beneficial to prey species (Conner et al. 1992). Small mammals are the primary prey of bobcats on our study area (Godbois et al. 2003b). Bobcats selected mature pine and mature pine-hardwood stands managed by frequent fire, but also selected other habitat types such as agricultural fields and shrub-scrub. Additionally, fire events facilitate predation of small mammal communities by temporarily reducing cover and exposing prey to predators (Conner et al. 2011, Morris et al. 2011). Therefore, there is a chance that bobcats—and other predators—may be temporarily attracted to recently burned areas, suggesting a tradeoff between managing for prey that may be particularly sensitive to predators following fire (e.g., wild turkey nests) and those species that rely on frequent fire (e.g., gopher tortoise [Gopherus Polyphemus and red-cockaded woodpecker (Picoides borealis; Alavalapati et al. 2002).

Abbreviations

AAU: 

Annual area of use

GLMM: 

Generalized linear mixed-effects models

MCP: 

Minimum convex polygon

VHF: 

Very high frequency

Declarations

Acknowledgements

We are grateful for funding provided by the Joseph W. Jones Ecological Research Center at Ichauway, Robert W. Woodruff Foundation, and the Warnell School of Forestry and Natural Resources at the University of Georgia. The Georgia Department of Natural Resources–Wildlife Resources Division provided in-kind support. We thank the staff at the aforementioned organizations for their technical assistance, and J. Rutledge, J. Doughty, I. Godbois, J. Keenan, A. Reid, B. Howze, and G. Lynch for their assistance with data collection.

Authors’ contributions

ARL designed the study, managed and analyzed the data, wrote the statistical code, and drafted the manuscript. LMC designed the study, managed data collection, provided statistical assistance, and reviewed the manuscript drafts. MJC designed the study and reviewed the manuscript drafts. NPN designed the study, provided statistical assistance, and reviewed the manuscript drafts. RJW designed the study, managed data collection, and reviewed the manuscript drafts. All authors read and approved the final manuscript.

Competing interests

The author(s) declare that they have no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Authors’ Affiliations

(1)
Warnell School of Forestry and Natural Resources, University of Georgia, Athens, USA
(2)
Joseph W. Jones Ecological Research Center at Ichauway, Newton, USA

References

  1. Alavalapati JRR, Stainback GA, Carter DR (2002) Restoration of the longleaf pine ecosystem on private lands in the US south: an ecological economic analysis. Ecol Econ 40:411–419View ArticleGoogle Scholar
  2. Anderson E (1987) A critical review and annotated bibliography of literature on bobcat. Colorado Division Wildlife, Boulder, CO R-S- 62-87. 61 p.Google Scholar
  3. Barding EE, Nelson TA (2008) Raccoons use habitat edges in northern Illinois. Am Midl Nat 159:394–402View ArticleGoogle Scholar
  4. Barnett JP (1999) Longleaf pine ecosystem restoration: the role of fire. J Sust For 9:89–96View ArticleGoogle Scholar
  5. Benson JF, Mahoney PJ, Patterson BR (2015) Spatiotemporal variation in selection of roads influences mortality risk for canids in an unprotected landscape. Oikos 124:1664–1673View ArticleGoogle Scholar
  6. Beyer HL (2012) Geospatial Modelling Environment (Vers. 0.7.4). Available via: http://www.spatialecology.com/gme
  7. Boring LR (2001) The Joseph W. Jones Research Center: Co-directed applied and basic research in the private sector. In: Barrett GW, Barret TL (eds) Holistic science: the evolution of the Georgia Institute of Ecology (1940-2000). Taylor and Francis, New York, New York., pp. 233–258.Google Scholar
  8. Brockway DG, Lewis CE (1997) Long-term effects of dormant-season prescribed fire on plant community diversity, structure and productivity in a longleaf pine wiregrass ecosystem. Ecol Manage 96:167–183View ArticleGoogle Scholar
  9. Brockway DG, Outcalt KW, Tomczak DJ, Johnson EE (2005) Restoring longleaf pine forest ecosystems in the southern U.S. In: Stanturf JA, Madsen P (eds) Restoration of temperate and boreal forests. CRC Press, Boca RatonGoogle Scholar
  10. Cain MD, Wigley TB, Reed DJ (1998) Prescribed fire effects on structure in uneven-aged stands of loblolly and shortleaf pines. Wildl Soc Bull 26:209–218Google Scholar
  11. Chamberlain MJ (1999) Ecological relationships among bobcats, coyotes, gray fox, and raccoons, and their interactions with wild turkey hens. Dissertation, Mississippi State University, Mississippi State, MississippiGoogle Scholar
  12. Chamberlain MJ, Leopold BD, Conner LM (2003) Space use, movements and habitat selection of adult bobcats (Lynx rufus) in Central Mississippi. Am Midl Nat 149:395–405View ArticleGoogle Scholar
  13. Cochrane JC (2003) Ecology of bobcats in a longleaf pine forest in southwestern Georgia,. M.S. Thesis, University of Georgia, Athens, GeorgiaGoogle Scholar
  14. Conner LM (1991) Bobcat home range, habitat use, and core use areas in east-central Mississippi. M.S. Thesis, Mississippi State University, Mississippi State, MississippiGoogle Scholar
  15. Conner LM, Castleberry SB, Derrick AM (2011) Effects of mesopredators and prescribed fire on hispid cotton rat survival and cause-specific mortality. J Wildl Manag 75:938–944View ArticleGoogle Scholar
  16. Conner LM, Leopold BD (1993) Habitat characteristics of bobcat core use areas in Mississippi. Proc Annu Conf Southeast Assoc Fish Wildl Agencies 47:53–61Google Scholar
  17. Conner LM, Leopold BD (1996) Bobcat habitat use at multiple spatial scales. Proc Annu Conf Southeast Assoc Fish Wildl Agencies 50:622–631Google Scholar
  18. Conner LM, Leopold BD (1998) A multivariate habitat model for female bobcats: a GIS approach. Proc Annu Conf Southeast Assoc Fish Wildl Agencies 52:232–243Google Scholar
  19. Conner LM, Leopold BD, Sullivan K (1992) Bobcat home range, density, and habitat use in east-Central Mississippi. Proc Annu Conf Southeast Assoc Fish Wildl Agencies 46:147–158Google Scholar
  20. Conner LM, Smith MD, Burger LW (2003) A comparison of distance-based and classification-based analyses of habitat use. Ecology 84:526–531View ArticleGoogle Scholar
  21. Dormann CF, Elith J, Bacher S, Buchmann C, Carl G, Carré G, García Marquéz JR, Gruber B, Lafourcade B, Leitão PJ, Münkemüller T, McClean C, Osborne PE, Reineking B, Schröder B, Skidmore AK, Zurell D, Lautenbach S (2013) Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography 36:027–046View ArticleGoogle Scholar
  22. Doughty J (2004) Bobcat (Lynx rufus) ecology in a longleaf pine ecosystem in southwestern Georgia. M.S. Thesis, University of Georgia, Athens, GeorgiaGoogle Scholar
  23. Drew MB, Kirkman LK, Gholson AK Jr (1998) The vascular flora of Ichauway, Baker County, Georgia: a remnant longleaf pine/wiregrass ecosystem. Castanea 63:1–24Google Scholar
  24. Frey SN, Conover MR (2006) Habitat use by meso-predators in a corridor environment. J Wildl Manag 70:1111–1118View ArticleGoogle Scholar
  25. Gillies CS, Hebblewhite M, Nielsen SE, Krawchuk MA, Aldridge CL, Frair JL, Saher DJ, Stevens CE, Jerde CL (2006) Application of random effects to the study of resource selection by animals. J Anim Ecol 75:887–898View ArticleGoogle Scholar
  26. Glitzenstein JS, Streng DR, Masters RE, Robertson KM, Hermann SM (2012) Fire frequency effects on vegetation in North Florida pinelands: another look at the long-term Stoddard fire research plots at tall timbers Research Station. For Ecol Manage 264:197–209View ArticleGoogle Scholar
  27. Godbois IA, Conner LM, Warren RJ (2003a) Habitat use of bobcats at two spatial scales in southwestern Georgia. Proc Annu Conf Southeast Assoc Fish Wildl Agencies 57:228–234Google Scholar
  28. Godbois IA, Conner LM, Warren RJ (2003b) Bobcat diet on an area managed for northern bobwhite. Proc Annu Conf Southeast Assoc Fish Wildl Agencies 57:222–227Google Scholar
  29. Goebel PC, Palik BJ, Kirkman LK (1997) Landscape ecosystem types of Ichauway. Technical report 97–1. Joseph W. Jones Ecological Research Center, NewtonGoogle Scholar
  30. Golley FB, Gentry JB, Caldwell LD, Davenport LB Jr (1965) Number and variety of small mammals on the AEC Savannah River plant. J Mammal 46:1–18View ArticleGoogle Scholar
  31. Hall HT, Newsome JD (1976) Summer home ranges and movements of bobcats in bottomland hardwoods of southern Louisiana. Proc Annu Conf Southeast Assoc Fish Wildl Agencies 30:427–436Google Scholar
  32. Hamilton LC (1992) Regression with graphics: a second course in applied statistics. Brooks and Cole, Pacific GroveGoogle Scholar
  33. Harper CA, Ford WM, Lashley MA, Moorman CE, Stambaugh MC (2016) Fire effects on wildlife in the central hardwoods and Appalachian regions, USA. Fire Ecol 12:127–159View ArticleGoogle Scholar
  34. Jones DD, Conner LM, Storey TH, Warren RJ (2004) Prescribed fire and raccoon use of longleaf pine forests: implications for managing nest predation. Wildl Soc Bull 32:1255–1259View ArticleGoogle Scholar
  35. Kauffman MJ, Varley N, Smith DW, Stahler DR, MacNulty DR, Boyce MS (2007) Landscape heterogeneity shapes predation in a newly restored predator-prey system. Ecol Lett 10:690–700View ArticleGoogle Scholar
  36. Kirby JD, Rutledge JC, Jones IG, Conner LM, Warren RJ (2010) Effects of time of day and activity status on bobcat (Lynx rufus) cover-type selection in southwestern Georgia. Southeast Nat 9:317–326View ArticleGoogle Scholar
  37. Landers JL, Van Lear DH, Boyer WD (1995) The longleaf pine forest of the southeast: requiem or renaissance? J For 93:39–43Google Scholar
  38. Lovallo MJ, Anderson EM (1996) Bobcat movements and home ranges relative to roads in Wisconsin. Wildl Soc Bull 24:71–76Google Scholar
  39. Maehr DS, Brady JR (1986) Food habits of bobcats in Florida. J Mammal 67:133–138View ArticleGoogle Scholar
  40. Manly BF, McDonald LL, Thomas DL, McDonald TL, Erickson WP (2002) Resource selection by animals: statistical analysis and design for field studies, Second edn. Kluwer Academic, BostonGoogle Scholar
  41. Masters RE, Lochmiller RL, McMurry ST, Bukenhofer GA (1998) Small mammal response to pine-grassland restoration for red-cockaded woodpeckers. Wildl Soc Bull 26:148–158Google Scholar
  42. McMurry ST, Lochmiller RL, Boggs JF, Leslie DM, Engle DM (1994) Demographic profiles of populations of cotton rats in a continuum of habitat types. J Mammal 75:50–59View ArticleGoogle Scholar
  43. Miller DA, Conner LM (2007) Habitat selection within seasonal use areas of female eastern wild turkeys in a managed pine landscape in Mississippi. J Wildl Manag 71:744–751View ArticleGoogle Scholar
  44. Miller SD, Speake DW (1978) Prey utilization by bobcats on quail plantations in southern Alabama. Proc Annu Conf Southeast Assoc Fish Wildl Agencies 32:100–111Google Scholar
  45. Morris G, Hostetler JA, Conner LM, Oli MK (2011) Effects of prescribed fire, supplemental feeding, and mammalian predator exclusion on hispid cotton rat populations. Oecologia 167:1005–1016View ArticleGoogle Scholar
  46. Northrup JM, Hooten MB, Anderson CR Jr, Wittemyer G (2013) Practical guidance on characterizing availability in resource selection functions under a use–availability design. Ecology 94:1456–1463View ArticleGoogle Scholar
  47. R Core Team (2017) A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna. Available via: https://www.r-project.org/
  48. Roberts NM, Crimmins SM (2010) Bobcat population status and management in North America: evidence of large-scale population increase. J Fish Wildl Manage 1:169–174View ArticleGoogle Scholar
  49. Sandell M (1989) The mating tactics and spacing patterns of solitary carnivores. In: Gittleman JL (ed) Carnivore behavior, ecology, and evolution. Cornell University Press, New YorkGoogle Scholar
  50. Sasmal I, DePerno CS, Swingen MB, Moorman CE (2017) Influence of vegetation type and prescribed fire on Peromyscus abundance in a longleaf pine ecosystem. Wildl Soc Bull 41:49–54View ArticleGoogle Scholar
  51. Schnell JH (1968) The limiting effects of natural predation on experimental cotton rat populations. J Wildl Manag 32:698–711View ArticleGoogle Scholar
  52. Seaman DE, Powell RA (1996) An evaluation of the accuracy of kernel density estimators for home range analysis. Ecol 77:2075–2085View ArticleGoogle Scholar
  53. Steen DA, Conner LM, Smith LL, Provencher L, Hiers JK, Pokswinski S, Helms BS, Guyer C (2013) Bird assemblage response to restoration of fire-suppressed longleaf pine sandhills. Ecol Appl 23:134–147View ArticleGoogle Scholar
  54. Van Lear DH, Carroll WD, Kapeluck PR, Johnson R (2005) History and restoration of the longleaf pine-grassland ecosystem: implication for species at risk. For Ecol Manage 211:150–165View ArticleGoogle Scholar
  55. Waldrop TA, White DL, Jones SM (1992) Fire regimes of pine-grassland communities in the southeastern United States. Forest Ecol Manage 47:195–210View ArticleGoogle Scholar
  56. Walker J (1998) Ground layer vegetation in longleaf pine landscapes: an overview for restoration management. In: Proceedings of the longleaf pine ecosystem restoration symposium. Longleaf Alliance Report no. 3, pp 2–13Google Scholar
  57. White GC, Garrott RA (1990) Analysis of wildlife radio-tracking data. Academic Press, New YorkGoogle Scholar
  58. Zuur A, Ieno EN, Walker N, Saveliev AA, Smith GM (2009) Mixed effects models and extensions in ecology with R. Springer, New YorkView ArticleGoogle Scholar

Copyright

© The Author(s). 2018

Advertisement