Study area

We conducted this study from January 2009 to July 2011 on a 1360 ha private pine plantation (31°31′57.2″N, 94°42′91.2″W), and a 5000 ha commercial timber property (31°21′28.1″N, 94°24′54.4″W) in Nacogdoches and Angelina counties in east Texas. The vegetation resembles the southeastern mixed and southeastern coniferous forests. Longleaf pine Pinus palustris forests have largely been replaced by even-aged loblolly pine P. taeda plantations. The natural vegetation has been transformed by the planting of pine stands and the exclusion of fire (Omernick et al. 2008). The topography is undulating hills with swampy low-lying areas. Historically these pine forests were successional to hardwood forests. The mean annual rainfall in the Pineywoods is 1192 mm, with monthly means from 55 mm in July to 116 mm in May (NOAA 2012). The temperatures vary from > 38°C in summer to < –5°C in winter.

Data analysis

We calculated the linear distance between consecutive nocturnal locations, for each animal, using the Pythagorean formula:

We used linear displacement as a proxy for distance moved each hour (hereafter ‘movement'; Palomares and Delibes 1991, Rockhill et al. 2013). Each movement was associated with Sex, Astronomical season, Biological season, Lunar phase, Moonlight intensity, Night period, Crepuscularity and Temperature. Prior to analysis, we removed movement intervals that had missing variables from the data sets.

From collar data, we calculated 11 940 (x̄ = 1194) discrete movement intervals for bobcats (3530 for females (x̄ = 1177) and 8410 for males (x̄ = 1201)) and 12 701 (x = 1270) for coyotes (5695 for females (x̄ = 1424), and 7006 for males (x̄ = 1167)). Coyotes moved farther each hour than bobcats. Bobcat movements varied from 0.6 to 2059 m h–¹ (x̄ = 230 m h–¹, n = 11 935, SE = 2.39), while coyote movements varied from 0.2 to 4636 m h–1 (x̄ = 336 m h–¹, n = 12 686, SE = 4.18).

We used generalized linear mixed effects models (GLMMs) (Bolker et al. 2009, Harrison et al. 2018) to model hourly movements relative to nocturnal variables. We used the distance moved (mh–¹) as the dependent variable, the variables Astronomical season, Biological season, Moonphase, Moonlight intensity, Night period, Crepuscular period and Temperature as fixed effects and the individual animals as random effects (Harrison et al. 2018). This obviated any concerns of autocorrelation between consecutive data points (Harrison et al. 2018). We used program R (< www.r-project.org>) and the packages ‘lme4' (Bates et al. 2015), ‘lmerTest' (Kuznetsova et al. 2017) and ‘emmeans' (Lenth 2019) to model the effects of variables associated with bobcat and coyote movements. Visual assessment of the data suggested that the residuals were not normally distributed, consequently we identified the most appropriate data transformation using the ‘bestNormalize’ package (Peterson and Cavanaugh 2019) in program R. This identified the ordered quantile normalization transformation (Peterson and Cavanaugh 2019) as most appropriate for all three data sets. We applied this transformation to ‘distance moved’, for each of our data sets (bobcat movements, coyote movements and the combined bobcat and coyote movements). After transformation the residuals were normally distributed, and we performed our analyses on the transformed data.

We suspected that some variables might be collinear, particularly Astronomical season with Biological season and Moon phase with Moonlight intensity. To avoid inferential errors associated with collinearity, we the calculated variable inflation factors (VIF) for each independent variable. Where variables had VIF ≥ 3 (Zuur et al. 2010, Thompson et al. 2017), we dropped one of the collinear variables from our candidate set. We used the bobcat data to refine the variable set and then checked the VIF values for the coyote and the combined data sets.

Having defined our candidate variables, we used model selection to identify the combination of variables best supported by our data using the corrected Akaike's information criterion (AICc) (Hurvich and Tsai 1989). We calculated the AICc values of each independent variable and used these to establish a hierarchy of influence within each candidate set of variables. We started with the full model (all variables included) and then sequentially removed the variable with the highest AICc score from the model. The sequence in which we removed variables from bobcat models was: 1) Moonlight intensity, 2) Sex, 3) Crepuscularity, 4) Biological season, 5) Night period and 6) Temperature. The sequence of variable removal from coyote models was: 1) Sex, 2) Night period, 3) Moonlight intensity, 4) Biological season, 5) Crepuscularity and 6) Temperature. The sequence of variable removal from the combined model was: 1) Species, 2) Sex, 3) Moonlight intensity, 4) Biological season, 5) Night period, 6) Crepuscularity and 7) Temperature. In addition, we evaluated each combination of two-way variable interactions. Where models contained interactions, we started with a model containing all two-way interactions and then applied the same hierarchical removal of variables from the models. For models that contained both species, only interactions of the variable ‘Species’ with other variables was considered. In all cases, delta AICc scores greater than six were considered to have substantially less support than the best model (Harrison et al. 2018). We calculated marginal and conditional R² values for each model (Nakagawa and Schielzeth 2013). The marginal R² values represent the variance attributable to the fixed effects and the conditional R² values relate to the variance explained by the entire models (Nakagawa and Schielzeth 2013).

Bobcats

Female bobcats consistently moved shorter distances (x̄ = 160 m h–¹, n = 3, SE = 62) than males (x̄ = 236 m h–¹, n = 7, SE = 43). Bobcat movements tended to increase from the breeding season through the kitten rearing season (t = –4.40, df = 11 910, p < 0.001) and winter (t = –4.06, df = 1910, p < 0.001) (Fig. 1) (Supplementary material Appendix 1 Table A1). This trend was consistent between sexes, variation in seasonal movement was significant for females between the breeding and kitten rearing seasons (t = –6.03, df = 3528.94, p < 0.001), however male movement differed between breeding season and the kitten rearing season (t = –1.96, df = 8383.18, p = 0.05) and between the breeding season and winter (t = –3.95, df = 8383.03, p < 0.001) (Supplementary material Appendix 1 Table A1). Bobcat movements were not greatly influenced by Moonlight intensity (t = –0.52, df = 11 900, p = 0.61) (Fig. 2) (Supplementary material Appendix 1 Table A1). Night period influenced bobcat movements (Fig. 3) decreasing from the period prior to 22:00 through to 02:00 h (t = –10.89, df = 11 900, p < 0.001), but then increasing between 02:01 and 07:00 h (t = –7.48, df = 11 900, p < 0.001). Generally, bobcats moved marginally farther in the crepuscular period than during the night (Fig. 4) (Supplementary material Appendix 1 Table A1). This pattern was not consistent for males and females. Females moved farther in the night than in the crepuscular period (although not significantly so) whereas males moved farther in the crepuscular period (t = 3.25, df=8378, p=0.001) (Supplementary material Appendix 1 Table A1). Bobcats movements declined as temperature increased (t=–26.84, df=11 910, p<0.001) (Fig. 5).

Figure 1.

Rates of movement (± 1 SE) by bobcats Lynx rufus and coyotes Canis latrans relative to biological seasons in east Texas from January 2009 to July 2011. A, B, C = Statistically significant influence of biological seasons within species (Supplementary material Appendix 2 Output A1).

Temperature and its interaction with Sex, Biological season, Night portion and Crepuscularity was the model best supported by our data (Model 1, Table 1). Although not included in the top model, the penalty incurred by including Moonlight intensity does not preclude it from influencing the movements of bobcats.

Coyotes

There was little difference in movements of female (x̄ = 326 m h–¹, n = 4, SE = 29) and male (x̄ = 337 m h–¹, n = 6, SE = 48) coyotes. Coyote movements varied seasonally (Fig. 1) (Supplementary material Appendix 1 Table A2), however only the comparison between spring and summer was significant (t = 6.45, df = 12 600, p < 0.001). Male movements increased more than females from the spring to summer but declined less in winter (Supplementary material Appendix 1 Table A2). Coyotes' movements increased with Moonlight intensity (t = 3.40, df = 12 600, p < 0.001) (Fig. 2). Coyotes moved farther in the middle of the night than from 17:00 to 22:00 h and from 02:01 to 07:00 h (Fig. 3) (Supplementary material Appendix 1 Table A2), however, these relationships were non-significant. Coyotes moved farther in the night time than in the crepuscular period (t = –10.62, df = 12 680, p < 0.001) (Fig. 4) (Supplementary material Appendix 1 Table A2), and their movements declined as temperatures increased (t = –45.66, df = 12 680, p < 0.001) (Fig. 5).

Biological season and its interaction with Night period, Sex, Crepuscularity, Moonlight intensity and Temperature was the model best supported by our data (Model 1, Table 2). Even the removal of the interaction of Biological season with Sex from the model substantially weakened its predictive value (Table 2).

Bobcats versus coyotes

Whereas bobcats' movements seemed to decline from breeding season through summer and then winter, coyotes' movements increased from spring to summer and then declined in winter (Fig. 1) (Supplementary material Appendix 1 Table A1, A2). Bobcats did not display a response to increased Moonlight intensity, whereas coyotes moved farther as Moonlight intensity increased (Fig. 2) (Supplementary material Appendix 1 Table A1, A2).

Bobcats seemed to have bimodal movement patterns declining in the middle if the night, whereas coyote movements peaked in the middle of the night (Fig. 3) (Supplementary material Appendix 1 Table A1, A2). The crepuscular movements of bobcats did not differ from their nighttime movements. However, coyotes moved farther at night than during the crepuscular period (Fig. 4) (Supplementary material Appendix 1 Table A1, A2). Both bobcats and coyotes reduced their movements as temperatures increased (Fig. 5).

The model that best explains the differences between coyotes and bobcats excludes Sex (Model 1, Table 3). However, the inclusion of Sex does not substantially increase the model's AICc score (Table 3). The removal of Moonlight intensity from the model substantially reduces the predictive ability of the model (Table 3).

Figure 2.

Response of bobcats Lynx rufus and coyotes Canis latrans to increasing Moonlight intensity from January 2009 to July 2011. *Statistically significant influence of moonlight (Supplementary material Appendix 2 Output A1). Note: Highlighted ribbons represent 95% confidence intervals.

Bobcats

Longer movement distances of male than of female bobcats is in line with previous studies for bobcats (Bailey 1974, Chamberlain et al. 2003) and other lynx species (e.g. Eurasian lynx, Lynx lynx: Sunde et al. 2000, Jedrzejewski et al. 2002) and can probably be explained by female movements being governed by foraging concerns while male movements optimize mating opportunities (Sandell 1989). In the southern United States, bobcats breed between February and May, and rear kittens from June to September (Chamberlain et al. 2003). As in Mississippi (Chamberlain et al. 1998), female bobcats in east Texas moved the shortest distances during spring and summer, which may be associated with kitten rearing. Furthermore, this seasonal trend aligns with bobcats' reduced movements at high temperatures (Fig. 5).

Figure 4.

Rates of movement (± 1 SE) by bobcats Lynx rufus and coyotes Canis latrans relative to crepuscular period in east Texas from January 2009 to July 2011. *Statistically significant difference between night time and crepuscular period (Supplementary material Appendix 2 Output A1).

Figure 5.

Response of bobcats Lynx rufus and coyotes Canis latrans to temperature in east Texas from January 2009 to July 2011. *Statistically significant influence of temperature (Supplementary material Appendix 2 Output A1). Note: highlighted ribbons represent 95% confidence intervals.

Although bobcats are dependent on their sight during hunting (Sunquist and Sunquist 2002, Rockhill et al. 2013), our analysis did not show an increase in bobcat movements as Moonlight intensity increased. This is contrary to Rockhill et al. (2013), who found that bobcats increased their movements at lunar light intensities between 10 and 49%. The lack of response to increasing illumination is surprising, especially as bobcat movements are thought to be inhibited by low levels of illumination (Kavanau 1971, Rockhill et al. 2013). It seems that despite bobcats having poor night vision (Rockhill et al. 2013), bobcat movements in east Texas are not influenced by lunar light.

Like observations elsewhere in the USA (Chamberlain et al. 1998, Tigas et al. 2002, Elizalde-Arellano et al. 2012), bobcat nocturnal movements were bimodal with a clear reduction in movement during the middle of the night. Although generally we found little difference between nocturnal and crepuscular movements for bobcats, this differs from other studies that show bobcats being more active during the crepuscular period than at night (Rockhill et al. 2013, Symmank et al. 2014). It seems that in east Texas, female and male bobcats have different crepuscular activity patterns. Such differential crepuscular activity may have evolved to dampen intraspecific intersexual competition (Rolley and Warde 1985). However, the intrasexual differences in movements between crepuscular and night periods are small and may be of little biological importance.

As in other studies (Zezulak and Schwab 1980, Elizalde-Arellano et al. 2012, Rockhill et al. 2013), our data suggests that temperature is the primary factor influencing bobcat movements in east Texas. It appears that bobcat activity is curtailed at extreme high temperatures (Zezulak and Schwab 1980, Elizalde-Arellano et al. 2012). During winter, when the ambient temperatures are lowest, bobcats move the longest distances. It may be that at low temperatures bobcats need to move farther to satisfy their increased metabolic requirements (Sandell 1989) or to compensate for seasonally lower prey availability.

Coyotes

As has been demonstrated elsewhere (Shivik et al. 1997, McClennen et al. 2001, Elfelt 2014) our data confirm that coyote movements are not influenced by Sex. Our analysis highlights the importance of differential movements between Biological seasons for coyotes in east Texas. In east Texas, as in Mississippi (Chamberlain et al 2000), increased summer movements may be necessitated by provisioning food for pups. Our analysis showed that male and female coyotes in east Texas, like those in the Sierra Nevada (Shivik et al. 1997), increased their nocturnal movements during the summer. This differs from other studies that found that only females travel farther during pup rearing (Holzman et al. 1992, Chamberlain et al. 2000). Their increased summer movement is counterintuitive as high ambient temperature inhibits coyote movements. The relatively short distances travelled by coyotes in spring and winter coincided with pulses of increased resource availability and the pup rearing period. In spring, coyotes use a high proportion of seasonally available fruits (Dumond et al. 2001, Schrecengost et al. 2008, Melville et al. 2015a), this likely reduces the distances that they need to travel to fulfill their metabolic requirements. Winter coincides with the onset of white-tailed deer hunting season (Kilgo et al. 1998). Both research sites were intensively hunted by recreational hunters. Resultant carcass remains, deposited on ‘gut piles’, provided easily accessible, predictable, localized, seasonal resource nodes during deer hunting season (Crimmins et al. 2012, Gomo et al. 2017).

The increase in movement by coyotes as Moonlight intensity increased is logical as sight is their primary sense for hunting (Kavanau and Ramos 1975). In east Texas, as had been demonstrated for captive individuals (Kavanau and Ramos 1975), coyotes increased their movements as nocturnal light increased. Like coyotes in Arizona (Grinder and Krausman 2001), coyotes in east Texas moved farthest in the middle of the night. This is contrary to the behavior of coyotes in northern Mexico (Arias-Del Razo et al. 2011) where their activity peaked in the mornings and evenings. As in California (Neale and Sacks 2001) and Massachusetts (Way et al. 2004), coyotes in east Texas moved farther during the nocturnal than the crepuscular period. Increased movement after the crepuscular period may be a behavioral response to avoid humans (Chamberlain and Leopold 1999, Kitchen et al. 2000, McClennen et al. 2001, Riley et al. 2003, Wang et al. 2015). It is possible that harvesting of coyotes by humans induces a more nocturnal habit in coyotes in east Texas.

Table 1.

Model selection for variables thought to influence bobcat Lynx rufus movements in east Texas from January 2009 to July 2011.

Table 2.

Model selection for variables thought to influence coyote Canis latrans movements in east Texas from January 2009 to July 2011.

Table 3.

Model selection for variables thought to influence both bobcat Lynx rufus coyote Canis latrans movements in east Texas from January 2009 to July 2011.

Coyote movements consistently decreased as collar temperature increased. Coyote body temperatures are independent of ambient temperature below ca 25°C but rise steadily above this level. At ambient temperatures >34°C desert coyotes pant, and as temperatures approach 40°C coyotes become stressed (Golightly and Ohmart 1983). Coyotes depend on evaporative cooling to maintain body temperature at high ambient temperatures (Golightly and Ohmart 1983). Unlike the desert environment, where low humidity promotes evaporative cooling, the Pineywoods are hot and humid, especially during summer. These conditions reduce locomotor performance and may cause hyperthermia (Zub et al. 2013). High humidity inhibits evaporative cooling and coyotes may reduce activity to avoid overheating. This may explain coyotes' reduced movements at high perceived temperatures, and why coyote movement peaks in the middle of the night (Fig. 3).

Bobcats versus coyotes

Both bobcat and coyote populations on our study sites were harvested continuously, and consequently, the populations remained at relatively low densities (Davis 2010) throughout the study. Therefore, neither populations are likely to have been resource constrained relative to prey availability (Melville et al. 2015a). Despite this, bobcats tended to move farthest in winter when bobcat diets were least diverse (Melville et al. 2015a), suggesting an increased energy expenditure while hunting. Coyotes, on the other hand, moved farthest in summer when they were provisioning pups and when resource supplements such as fruits and animal remains were not available.

Bobcat night time movements in east Texas were bimodal with reduced movements during the middle of the night, like bobcats elsewhere in the USA (Chamberlain et al. 1998, Tigas et al. 2002, Elizalde-Arellano et al. 2012). Coyotes in east Texas, as in Arizona (Grinder and Krausman 2001), moved farthest in the middle of the night. This could be due to differential nocturnal movement patterns between bobcats and coyotes (Fig. 3), with bobcats reducing their movements when coyotes are most active.

Although activity patterns are governed primarily by nutrition and reproduction (Arias-Del Razo et al. 2011), bobcat and coyote movements are clearly influenced by several abiotic variables. In addition, the responses to these variables cannot be viewed in isolation. The complex interactions of variables that influence bobcat and coyote movements makes it clear that their movements are influenced by mechanisms that are more complex than mere resource availability. For future research, biologists should be cognizant of the complex system of constraints under which bobcats and coyotes operate and ensure that these are incorporated into the development of hypotheses to explain their movements.