Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Habitat selection during ungulate dispersal and exploratory movement at broad and fine scale with implications for conservation management

Habitat selection during ungulate dispersal and exploratory movement at broad and fine scale with... Background: Dispersal has a critical influence on demography and gene flow and as such maintaining connectivity between populations is an essential element of modern conservation. Advances in satellite radiotelemetry are providing new opportunities to document dispersal, which previously has been difficult to study. This type of data also can be used as an empirical basis for defining landscapes in terms of resistance surfaces, enabling habitat corridors to be identified. However, despite the scale-dependent nature of habitat selection few studies have investigated selection specifically during dispersal. Here we investigate habitat selection during and around dispersal periods as well as the influence of age and sex on dispersal for a large ungulate. Results: Of 158 elk (Cervus elaphus) tracked using GPS radiotelemetry almost all dispersers were males, with individuals dispersing up to 98 km. The dispersal period was distinct, with higher movement rates than before or after dispersal. At fine scale elk avoided the most rugged terrain in all time periods, but to a greater extent during and after dispersal, which we showed using step selection functions. In contrast, habitat selection by resident elk was less affected by ruggedness and more by an attraction to areas of higher forage availability. At the broad scale, however, movement corridors of dispersers were characterized by higher forage availability and slightly lower ruggedness then expected using correlated random walks. Conclusions: In one of the first examples of its kind we document complete long-distance dispersal events by an ungulate in detail. We find dispersal to be distinct in terms of movement rate and also find evidence that habitat selection during dispersal may differ from habitat selection in the home-range, with potential implications for the use of resistance surfaces to define conservation corridors. Keywords: Alberta, Cervus elaphus, Dispersal, Elk, Habitat selection, Migration, Step selection functions Background developed, empirical studies are generally lacking because Dispersal is a fundamental process in ecology and evolu- of the difficulties associated with observing and quanti- tion, affecting individual fitness as well as demography, fying the dispersal process [6]. However, new advances in genetic structure, and species distributions [1-4]. Disper- satellite radiotelemetry now permit opportunities to docu- sal is likewise of central importance for managing animal ment dispersal that were previously unattainable [7]. populations; maintaining connectivity among populations Landscape connectivity describes how the movement is considered to be an essential part of modern conser- of animals is linked to landscape structure [8]. The way vation [5]. Although theoretical work on dispersal is well in which movement among populations is affected by environmental conditions is important for predicting the effects of landscape modification and habitat fragmenta- * Correspondence: [email protected] 1 tion, and in prioritising which habitats to protect. The Department of Biological Sciences, University of Alberta, Edmonton, AB T6G 2E9, Canada most widely used method for maintaining connectivity is Marine Evolution and Conservation Group, Centre of Evolutionary and the conservation corridor, a protected area of landscape Ecological Studies, University of Groningen, PO Box 11103 CC, Groningen, that facilitates the movement of organisms between po- The Netherlands Full list of author information is available at the end of the article pulations [5]. One approach has been to map resistance © 2014 Killeen et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Killeen et al. Movement Ecology 2014, 2:15 Page 2 of 13 http://www.movementecologyjournal.com/content/2/1/15 surfaces to characterize how environmental variables af- During 2007-2011 we monitored elk (Cervus elaphus) fect animal movement, and to use these surfaces to model in the Rocky Mountains of North America using GPS connectivity and thus to inform management decisions telemetry to address these questions. Elk herds in this relating to corridors [9]. As the use of Geographic Infor- area may be migratory, in which animals move to higher mation Systems (GIS) has become more widespread, land- altitudes in spring and summer to gain access to high- scapes are increasingly defined using resistance surfaces, quality forage, or partially migratory, in which some ani- allowing the use of methods such as least-cost path ana- mals remain resident in winter ranges throughout the year lysis to help determine the most effective placement of [22]. Dispersal occasionally takes place, when an animal corridors [10]. However, the effectiveness of such an ap- leaves its home range and does not return. We expected proach is highly dependent on the accurate assignment of that young male elk would be more likely to disperse than resistance values to landscape units, yet this has mostly females [23]. Many male cervids are known to disperse been a subjective process [9,11]. Using empirical data gen- from their natal home range, including elk [24-26]. We erally leads to more robust and more readily justifiable also predicted that dispersers would select habitats based conclusions [12]. more on ease of movement through the landscape than Where empirical data have been used, it frequently has forage quality and we expected this pattern to be clear been assumed that habitat selection parameters derived during dispersal, while before and after dispersal we ex- from movement within home ranges can be extrapolated pected forage quality to be more important in habitat to movement during dispersal [13,14]. Corridors aim to selection. The dispersal period also was expected to be promote connectivity between populations and therefore distinct in terms of movement parameters, with faster and need to facilitate dispersal movement, making this as- more directional movement. Furthermore we expected sumption a potential source of error [9]. There is currently there to be distinct movement behaviours at the fine- little information to suggest whether this assumption is scale; shorter foraging movements and longer directional reasonable or not, although Soulsbury et al. [15] found movements [27]. We predicted that longer steps would be that red foxes (Vulpes vulpes) have unique patterns of selected based more on ease of movement while shorter habitat selection during dispersal, while Newby [16] found steps would be selected based more on forage quality. We that cougars (Puma concolor) do not. The general lack of used resident individuals for further comparison, because data on this subject probably reflects the difficulty of of their consistent home-ranging behaviour, and we ex- measuring and tracking dispersal [6]. Predicting which an- pected habitat selection to be consistently less affected by imals are likely to disperse can be difficult, so tracking ease of movement and more affected by forage quality for those specific individuals is often not possible. Even with a these individuals. large sample size the number of dispersing individuals We tested these predictions by first differentiating dis- within a population can be small [13,17]. Given that badly persal from other types of movement, such as migration designed corridors run the risk of acting as population and residency, on the basis of the measurement Net- sinks or simply wasting financial resources and eroding Squared Displacement (NSD) [28] and extracting tele- the support of stakeholders [11,18], this is an important metry points within the dispersal period. We quantified issue to address. Greater understanding of habitat selec- the distance and duration of dispersal using these data. tion by animals during dispersal should lead to better in- We then used step selection functions (SSF) and step- formed conservation management decisions. length analysis to compare fine-scale habitat selection Here we investigate how age and sex influence the between time periods for dispersing individuals and for likelihood of dispersal for an ungulate. Sex differences matched resident individuals [29]. We also investigated in juvenile dispersal are common and for mammals it is the dispersal period itself using segmented regression to most often young males that disperse [19]. In a polygyn- split movement behaviours into different scales of move- ment, before analysing each separately [30]. Finally, we ous mating system males are expected to have more variation in reproductive success because a male must used correlated random walks to analyse selection at a secure a territory or become dominant to increase re- broad scale. While step selection functions can be used to analyse animal behavioural decisions during move- productive success [20]. Therefore it is important for ju- venile males to avoid sexual competition with older and ment at a scale measured in hours, correlated random more powerful males, and consequently dispersal is fa- walks can be used to analyse movement at the scale of a complete dispersal pathway. voured [21]. We then investigate whether habitat se- lection during dispersal differs from habitat selection during home-ranging and other behaviours. We hypo- Results thesise that habitat selection during dispersal will have a Dispersal and movement parameters unique pattern based on ease of movement through the Of the 132 elk (94 female, 38 male) with complete data, landscape. 97 were identified as migratory and 17 as resident. There Killeen et al. Movement Ecology 2014, 2:15 Page 3 of 13 http://www.movementecologyjournal.com/content/2/1/15 were a total of 16 dispersal events, all but one under- although movement rate in all was lower compared to taken by males, 9 of which completed dispersal and 7 of that recorded during the rest of the year (reference ca- which were exploratory movements. Of the males tracked tegory, β =0). st st for the required time period (at least 1 April to 31 Movement data during the dispersal period were fur- October), this translates to 39% of males dispersing (24% ther divided into two parts using segmented regression, completing dispersal, 16% showing an exploratory move- representing two distinct movement behaviours [30]. ment) (Table 1). The breakpoint was identified as a movement rate of -1 Dispersal events ranged in length from 29.16 km to 7.12 m min (SE = 1.04) which corresponds to a step 98.01 km (straight line distance from first to last loca- length of 854 m in 2 hours (Figure 3). tion, actual distance travelled greater) and lasted from This resulted in two subsets of data – one of steps 12 to 47 days (M = 25.9 days), taking place between 18- under 854 m, representing shorter steps during foraging May and 04-August, with the majority of movement or resting (n = 1,292 steps), and the other of steps over occurring in June and July. Dispersal routes traversed a 854 m, representing longer movement steps (n = 355 range of landscapes, with some animals travelling from steps). Turning-angle distribution was distinct between Alberta into British Columbia or Montana (Figure 1). the groups, with longer steps having smaller turning an- Exploratory movements were similar to dispersal move- gles and therefore being more directional (Figure 4). ments in terms of timing and distance travelled but the Turning angles associated with short steps also are bur- animals returned to their starting locality. These move- dened with greater sampling error [31]. ments occurred on a timescale of weeks, inconsistent with migration, and we consider them as explorations Fine-scale habitat selection (Additional file 1: Figure S1). All males were captured at For dispersers the relationship between step length and approximately age 1.5 and therefore were dispersing terrain ruggedness was strongly non-linear with selection when just over 2 years old. for intermediate values of ruggedness during all time pe- For dispersers, according to the predictions of linear riods (Table 2). They avoided more rugged terrain in all -1 mixed effects analysis, movement rate (m hr ) recorded time periods, but this avoidance was weaker before disper- during different time periods (Figure 2) had the follow- sal (Figure 5a). Dispersers also had a weak attraction to ing pattern: during dispersal (β = 173.9; 95% CI 162.4, areas with high Normalized Difference Vegetation Index 185.3) > > before dispersal (β = 57.5; 95% CI 46.5, 68.4) > > (NDVI) values and to some extent avoided areas with high after dispersal (β = 4.0, 95% CI -7.28, 15.3). Movement rate canopy cover, strongly before dispersal and weakly during after dispersal did not differ from that recorded throughout and after dispersal, with no evidence for a non-linear ef- the rest of the year (reference category, β =0). Resi- fect. Distance to roads had no or a weak effect in all pe- dents meanwhile did not show significant differences riods examined. in movement rate between the periods during (β = -10.5, During the dispersal period, the shorter steps, as iden- 95% CI -20.7, -0.48; Figure 2), before (β = -14.3; 95% tified by segmented regression, were still characterised CI -24.5, -4.1) and after (β = -20.6; 95% CI -31.5, -9.7), by strong avoidance of more rugged terrain and selec- tion for intermediate values. Longer steps showed a lin- ear avoidance of ruggedness (Figure 5b), but this model was not highly predictive. The short-step group of steps Table 1 Movement classifications using net squared were more likely to end in resource units with higher displacement NDVI values while the long-step group had no effect of Male Female Total Percentage (%) NDVI (Table 2). (n = 38) (n = 94) For residents there was also a non-linear relationship Migratory 26 71 97 73 between step length and terrain ruggedness but this ef- Resident 2 15 17 13 fect was only strong in one time period. There was in Other/not classified 10 8 18 14 general a weak avoidance of more rugged terrain, but in Dispersal event 15 (6) 1 (1) 16 contrast to dispersers there was a strong attraction to A total of 132 animals were observed between 2007 and 2011, with data from areas with high NDVI values in all time periods. There st st at least April 1 to October 31 , using GPS radiotelemetry and were classified was also a non-linear relationship with canopy cover in by movement type; migrant or resident. Some animals had an atypical movement pattern and were classified as other/not classified. The number of all time periods with residents selecting intermediate animals which had a dispersal period (number of those which were values while strongly avoiding areas with high canopy exploratory movements in brackets) was counted. Depending on the cover (Table 2). availability of data in the year following dispersal some dispersing animals could also be classified as migrants or residents. Those without data in the All models, except for the long-steps model, were following year fall under the ‘other’ category. Classifications were made by found to be useful predictors of habitat selection, with classifying and inspecting Net Squared Displacement graphs sensu Mysterud et al. 2011 [44] (see Additional file 1: Figure S1). the mean observed r of the models greater than 0.65, S Killeen et al. Movement Ecology 2014, 2:15 Page 4 of 13 http://www.movementecologyjournal.com/content/2/1/15 Figure 1 The study area and telemetry data overview. Map of the study area, located in Alberta and British Columbia, Canada and Montana, USA. Telemetry data was collected between 2007 and 2011. Locations from dispersing elk, during their dispersal period, are shown as points (each colour corresponds to an individual elk) on a digital elevation model of the area (lighter areas representing higher elevation). Minimum convex polygons for the full year of data for each matched resident are also shown in black. and the distribution of r higher than expected by chance found no significant pattern with canopy cover nor dis- alone (Additional file 2: Table S1). tance to roads (Wilcoxon signed rank test, mean β =-0.03, P = 0.85; mean β =0.25 P = 0.28 respectively). Broad-scale habitat selection Using correlated random walks with turning angle and Discussion step-lengths from the distribution of observed during- Dispersal and movement parameters dispersal long steps, we compared each dispersal path- We observed a number of substantial long-distance dis- way to random paths. For 7 of the 10 animals there was persal events, all but one of which was made by males. an avoidance of rugged terrain at this broad scale but Of the 38 males tracked for the required time period overall this was not significantly different from zero 39% had a dispersal period (24% completed) which is (mean β = -0.66, Wilcoxon signed rank test, P = 0.28). comparable to previous estimates of 27% for herds in However, for NDVI there was a more consistent attraction Colorado [25] and 40% in a Montana herd [32]. This to higher values (mean β = 0.99, Wilcoxon signed rank pattern of male-biased dispersal is expected for a po- test, P = 0.03) with 8 of 10 animals showing selection. We lygynous mammal species, because males have highly Killeen et al. Movement Ecology 2014, 2:15 Page 5 of 13 http://www.movementecologyjournal.com/content/2/1/15 -1 Figure 2 Movement rate of dispersers and residents. Movement rates (m hr ) before, during and after the dispersal event and throughout the rest of the year for a) dispersers and b) residents The mean movement rate in each period is shown with error bars representing ±1 SE. For dispersers, movement rate in the during-dispersal time period was significantly greater than in all other time periods and the before-dispersal period was also greater than the after-dispersal period, which was not different from the rest of the year. For residents, there was no significant difference in mean movement rate before, during and after the dispersal period, and in all these periods they moved slower than during the rest of the year. variable reproductive success [19]. Our estimate is, if may affect movement rate of animals and this was not ex- anything, likely to be an underestimate because many plicitly examined, although resident and dispersing ani- males (28% of those tracked) were killed before we were mals were paired in time to account for seasonal effects. able to gather enough data to confirm if they were dis- Resident animals had similar movement rates in all time persers or migrants. Given the lack of male residents periods, while dispersers did not. and female dispersers, we cannot assess the effect of sex on our models. However, we had a large sample size and Fine-scale habitat selection age range of females and it is clear that dispersal is al- Steps taken by dispersing elk showed a non-linear re- most totally male-dominated and therefore the charac- lationship with terrain ruggedness in all time periods teristics of dispersal behaviour appear inextricably linked examined, selecting intermediate values of ruggedness, to males. while avoiding the most rugged terrain. Avoidance of Movement rate for dispersers during the dispersal pe- rugged terrain has been observed for other mammals riods was greater than movements before or after. This [13,34] and presumably reduces energy expenditure and shows that the extracted dispersal period is indeed dis- facilitates ease of movement [35]. Unexpectedly, the re- tinct in terms of movement rate. Dispersing animals sponse by dispersing elk to terrain ruggedness was simi- move faster and further and maintain a persistent direc- lar both during and after dispersal. This may be because tion. There was a significant difference between the be- these individuals were experiencing novel environments fore dispersal movement rate in comparison to the rest and minimising energy expenditure while exploring and of the year, possibly representing a period of restlessness forming a new home-range after dispersal. We also split in the pre-dispersal phase. In the period after dispersal the during-dispersal period into two movement be- movement rate reduced, during which time there was haviours. A variety of methods have been used to infer likely to be a slow formation of a new home-range. Within different movement behaviours from step-length data, their home ranges animals are likely to switch between providing insight into movement decisions [27,30,36]. long directional movements between patches and periods We found that when only looking at long movement of short non-directional movements within a patch [27]. steps, in comparison to short steps during foraging or This is probably based on spatial memory of resources resting, the avoidance of ruggedness was linear for longer within their home-range [33]. However, seasonality also steps. More avoidance of rugged terrain during longer Killeen et al. Movement Ecology 2014, 2:15 Page 6 of 13 http://www.movementecologyjournal.com/content/2/1/15 Figure 3 Segmented regression of movement rate. The log frequency distribution of movement rates for all dispersers combined, during their dispersal period. Movement rates were binned and the log frequency in each bin was plotted, sensu Johnson et al. 2002 [30]. A segmented 2 -1 linear regression model was fit (red lines, r = 0.93) for which a breakpoint of 7.118 m min was identified using the R package segmented. A null model linear regression model is also shown for comparison (black, dashed line, r = 0.82). steps might be expected given that only when there is con- weak, and it may be that dispersers foraged wherever pos- siderable movement is the energy saved by avoiding rug- sible, rather than being highly selective, or that the pattern ged terrain likely to be substantial, although steps may be of foraging might be obscured with steps during resting shorter in rugged terrain due to the difficulty of move- behaviour, all of which fall within the same group. Dif- ment. However, the long-steps model was not validated, ferentiating resting and foraging steps is difficult given showing that habitat selection for these long movement the imprecision inherent in GPS data [36]. Residents also steps may be somewhat unpredictable. showed avoidance of areas with high canopy cover and se- In contrast, resident elk had inconsistent responses to lected movement toward localities with lower canopy ruggedness, although did still avoid more rugged terrain. cover, consistent with elk use of forest cover for protection However unlike dispersers, they showed strong and highly while using open areas for foraging [39]. consistent selection for areas with higher NDVI values. High values of NDVI indicate green herbaceous phyto- mass or forage quantity [37] suggesting that residents are Broad-scale habitat selection selecting areas with high forage availability the majority of While habitat selection at fine scale examines behav- the time. This is consistent with previous studies showing ioural decisions over the course of hours, we also used that foraging dominates summer elk activity [36,38]. Dur- broad-scale analyses to examine characteristics of the ing dispersal we expected that short steps would show a dispersal pathway as a whole. At this larger scale we strong attraction to high NDVI values in comparison to found that there was no clear overall avoidance of rug- long steps, because they should be associated with for- ged terrain, which was surprising given the strength of aging when moving less rapidly. In fact this pattern was the relationship at fine scale. This suggests that although Killeen et al. Movement Ecology 2014, 2:15 Page 7 of 13 http://www.movementecologyjournal.com/content/2/1/15 Figure 4 Turning angle distributions for short and long steps during dispersal. Turning angle distributions combined for the 10 dispersers during their dispersal period are shown. The upper graph shows the distribution of turning angles for steps longer than 854 m (n = 355 steps) and the lower graph shows the distribution of turning angles for steps shorter than 854 m (n = 1,292 steps). This boundary was derived from the segmented regression breakpoint in Figure 3. dispersing animals appear to select habitats for ease of but perhaps also because they select the direction of dis- movement at fine scale during movement, at a larger persal based on a broad assessment of habitat and for- scale they do not necessarily select their overall route age quality and along that route select habitats for ease in this way. This could be because of imperfect know- of movement at the fine scale. This is supported by the ledge of their environment – dispersers are travelling overall positive relationship of movement toward re- through areas that they have never encountered before, source units with high NDVI values. Table 2 SSF model parameter estimates (fine-scale) Ruggedness Ruggedness ^2 NDVI Distance roads Canopy cover Canopy cover ^2 Beta (se) Beta (se) Beta (se) Beta (se) Beta (se) Beta (se) Dispersers before 0.434 (0.100) - 0.202 (0.052) 0.623 (0.416) 0.457 (0.242) - 0.238 (0.108) - 0.185 (0.172) Dispersers during 0.155 (0.097) - 0.135 (0.044) 0.288 (0.175) 0.113 (0.158) - 0.113 (0.082) - 0.053 (0.068) Dispersers after 0.144 (0.106) - 0.156 (0.035) 0.860 (0.443) 0.354 (0.203) - 0.186 (0.139) - 0.002 (0.129) Residents before - 0.112 (0.125) - 0.281 (0.081) 1.031 (0.345) 0.202 (0.153) 0.132 (0.092) - 0.291 (0.078) Residents during - 0.087 (0.146) - 0.101 (0.085) 0.996 (0.309) 0.177 (0.169) 0.238 (0.072) - 0.216 (0.065) Residents after 0.111 (0.075) - 0.103 (0.085) 0.598 (0.209) 0.047 (0.188) 0.229 (0.095) - 0.263 (0.086) Disp. during long-steps - 0.236 (0.151) - 0.054 (0.054) 0.066 (0.278) - 0.114 (0.298) - 0.121 (0.108) - 0.029 (0.119) Disp. during short-steps 0.276 (0.117) - 0.171 (0.057) 0.385 (0.227) 0.211 (0.136) - 0.073 (0.102) - 0.099 (0.085) Beta values with standard errors in brackets are shown, from each of the SSF models. Beta values greater than 2 SE from 0 are shown in bold. Killeen et al. Movement Ecology 2014, 2:15 Page 8 of 13 http://www.movementecologyjournal.com/content/2/1/15 Figure 5 The response of dispersers to ruggedness. The relationship between the selection coefficients (beta values) for ruggedness. In a) dispersers are shown for the before, during and after periods of dispersal. Avoidance of ruggedness is stronger during and after dispersal compared to before dispersal. In b) the long step and short step groups, as derived from segmented regression (Figure 3), are shown for the dispersers during their dispersal period. Avoidance of ruggedness appears stronger for long steps than for short steps. Conclusions about habitat selection throughout a landscape. For ungu- In one of the first examples of its kind we were able to lates, it would be most useful to track young males, espe- track dispersing elk using high fix-rate GPS satellite ra- cially during spring and early summer, because these are diotelemetry throughout the entirety of their dispersal the animals most likely to disperse. periods, providing detailed information on the distance Overall we believe that the processes of dispersal amongst and direction travelled and the likelihood of dispersal. large mammals warrant further investigation and modern By analysing habitat selection with these data we find GPS radiotelemetry provides an excellent tool in this en- evidence to suggest that the most-supported step selec- deavour. We also find step selection functions to be a tion functions during dispersal are not the same as those powerful statistical tool for analysing habitat selection dur- selected for home-ranging behaviour. Where possible it ing dispersal. would be beneficial to estimate and account for the dif- ferences when using cost-distance modelling. If managers Methods are to implement initiatives to promote habitat connectiv- Study area ity, such as corridors, it would be ideal to prioritise data The study took place in southwest Alberta and extended collection from individuals that are most likely to disperse. into northwest Montana and southeast British Columbia This serves a double purpose. One, it ensures that the data (Figure 1). The majority of the area within Alberta is obtained are as representative as possible for characteris- provincial forest reserve and on the eastern boundary of ing dispersal movement, the primary purpose for habitat our study area there is mixed livestock ranching and corridors. Second, dispersers cover much larger distances cropland. This boundary is a transition zone from grass- during their travel than other animals, making them a land into the Rocky Mountains and several different elk cost-effective way to get the maximum amount of data populations are present. Natural predators of elk in the Killeen et al. Movement Ecology 2014, 2:15 Page 9 of 13 http://www.movementecologyjournal.com/content/2/1/15 area are wolf (Canis lupus), cougar (Puma concolor), and these animals as migrant or resident, as well as their dis- grizzly bear (Ursus arctos) [40]. There is considerable persal event. Those without long enough periods of data human presence in the study area, including industrial following dispersal were classified as other/not classified, activities such as forestry and natural gas extraction, as along with a number of animals with atypical movement well as recreational activities, especially during summer patterns. Several males had NSD graphs suggestive of and the autumn hunting season [41]. dispersal, but were killed by hunters during the autumn hunting season (September to end of November for rifle Elk data hunting [45]), making it difficult to confirm if they were A total of 158 elk (105 female, 53 male) were captured truly dispersers, or migratory. A number of individuals on winter ranges between 2007 and 2011 using helicopter also showed an ‘exploratory movement’ (Additional file 1: net-gunning. Each was fitted with a GPS-radiotelemetry Figure S1). These movements were similar to dispersal collar (either ARGOS GPS for males or Lotek GPS 4400 movements, occurring within the same timeframe before for females, Lotek Wireless Inc., Ontario, Canada) and all migration, rapidly travelling a long distance, but then units were programmed to obtain locations every 2 hrs. returning to or close to their original range within a short Satellite transmitted data from GPS collars fitted with period of time. These appear to be exploration events, and Argos communication devices were received weekly via therefore were included as dispersers in our analyses, al- email, while other data were downloaded remotely in the though we note that our results were not notably altered field. Radiocollars from elk that died were located and by inclusion of these individuals. Our categorisation re- re-fitted to new animals. All males were approximately sulted in a group of elk-years with clearly distinguishable 1.5 yrs old at time of capture and females varied be- movement patterns (Additional file 1: Figures S1, S2). We tween 1.5 yrs and 19 yrs old. A vestigial canine was re- calculated NSD and plotted the associated graphs using moved during capture and used to determine age by the adehabitat package version 1.8 [46] in version 2.15 cementum analysis (Matson’s Laboratory, MT, USA). of R [47]. Locations were screened following the method of Lewis et al. 2007 [42] but a few large measurement errors Data selection and movement parameters remained. These outliers were easily identified as loca- To investigate habitat selection by dispersers we selected tions which were an unreasonable distance from the individuals that had a dispersal period (n =10, 7 of which previous and next location (10s of km round trips in completed dispersal, 3 of which had exploratory move- 4 hrs) and were removed. Data sets also were trimmed ments) and for which there were resident individuals with at beginning and end to remove data where an elk’sbe- data spanning the same timeframe (n =10) available for haviour might have been influenced by capture or where comparison (n = 20 individuals, 57,637 telemetry reloca- the elk had died or where the collar failed to function. tions in total, locations shown in Figure 1). Residents were Fix rate for the 20 animals included in the analysis was used for comparison because they have a simple move- 81.7% for Lotek 4400 (females) and 66.2% for ARGOS ment pattern in which they stay within the same home GPS (males). range throughout the year. We did not investigate mi- grants because migration in this population is typically Distinguishing movement types characterised by frequent stop-overs while following Only data from individuals with relocations from at least spring green-up towards summer ranges, rather than st st 1 April to 31 October were included and of the 158 a single directed movement period [48]. This makes radio-collared animals, 26 (15 males, 11 females) were it difficult to compare the movement period of a migra- excluded. This was to ensure that the entirety of the mi- tion directly with that of a dispersal period. We selected gration or dispersal period was covered. To distinguish the telemetry points that occurred during the dispersal between different movement behaviours we used the mea- movement event itself, to the nearest day. This was done surement Net-Squared Displacement (NSD) [28]. This is a using NSD graphs, with the start of dispersal identified by time-dependent statistic that measures the straight-line a steep increase in the value of NSD and the end of dis- distance between a starting location and subsequent loca- persal identified by NSD plateauing as the animal settles tions in a movement path of a given individual. We used into a new home range (Additional file 1: Figure S1). If the method of Bunnefeld et al. [43] to classify each animal there was a short movement followed by a period of a as migrant or resident and to identify dispersal events. month or more of stationary behaviour, before the main Graphs of NSD were then inspected visually to identify dispersal event, this early movement was not considered the type of movement, similar to Mysterud et al. 2011 to be part of the dispersal event itself. [44]. After a successful dispersal animals will settle into a Also, we selected telemetry locations in the period of new home range and become either migrants or residents 26 days (corresponding to the average duration of a disper- and in some cases there were available data to classify sal event) immediately preceding and after the dispersal Killeen et al. Movement Ecology 2014, 2:15 Page 10 of 13 http://www.movementecologyjournal.com/content/2/1/15 event. This gave us a group of data points from dispersers A measure of distance to roads was used, derived from before they had undertaken their dispersal and after they road maps of the study area including both paved and had completed it. These same time periods also were ana- gravel roads (Governments of AB & BC: National Topo- lysed for matched residents even though there was no dis- graphic Database 1:50,000; U.S. Census Bureau Tiger/Line persal among these individuals. files, 2000). Finally we used a percent canopy cover surface -1 Movement rate (m hr ) was calculated by dividing the at 30 m resolution (U.S. National Land Cover Database, step length between two successfully obtained locations Governments of Alberta/British Columbia). Again, we in- by the time elapsed between those locations. We used a cluded both a linear and a squared term given an expected mixed effects model with individual elk as a random fac- non-linear relationship to canopy cover. tor to assess the effect of the time period (before, during, after dispersal, and the rest of the year) on movement Modelling fine-scale habitat selection using step selection rate. Movement rate throughout the rest of the year was functions set as reference category in the mixed model. A separate The straight-line segments linking successive animal lo- model was used for dispersers and for residents, using cations, taken at regular intervals, are defined as steps the R package lme4 [49]. In both models, statistically sig- [53]. We calculated step length in metres and turning nificant differences (α = 0.05) between movement rates angle (angle between previous and next location) using recorded during different time periods were assessed by GME with ARCMAP v. 10.1 (ESRI Inc., Redlands, CA). checking the overlap of 95% Confidence Intervals (CIs), In analyses involving step length, we used only steps assuming movement rates to be statistically different where the time interval was within 10 minutes of the (P < 0.05) if related 95% CIs didn’toverlap. typical two-hour interval. Steps with intervals much shorter or longer were removed to ensure fair compari- Landscape variables son of step length. We identified a number of environmental covariates that Step selection functions (SSF) are used for incorpo- are known to influence elk movement behaviour [45,50]. rating movement into habitat selection analysis, provid- Each covariate was imported into a GIS system and values ing a more fine-scale and mechanistic movement model for each were then attached to the points in the telemetry than the original Resource Selection Function (RSF) [54]. dataset using ESRI ArcGIS 10.1 (ESRI Inc., Redlands, CA) In an SSF each observed step is compared with a number and GME (Geospatial Modelling Environment, http:// of random steps that have the same starting point. Ran- www.spatialecology.com/gme/). dom step lengths and turning angles are drawn from dis- We used a 30-m resolution Digital Elevation Model tributions taken from observed data, allowing comparison (DEM) and the Spatial Analyst extension in ESRI ArcGIS of the observed step to a sample of those that could have 10.1 to produce a terrain ruggedness variable (30-m reso- been taken in the local environment. Here we analyse the lution), which quantifies topographic heterogeneity [51]. endpoints of observed and random steps, similar to a con- This is a measurement of the average elevation difference ditional RSF but with controls drawn from a domain most between a point on a digital elevation model grid and likely to represent those truly available given observed the surrounding cells. We included both a linear and movement patterns [29]. a squared term given an expected non-linear relation- Step lengths and turn angles for the random steps were ship to ruggedness. drawn independently from distributions of the observed As a proxy for forage quality we used monthly Nor- values, which was reasonable because of low observed cor- malized Difference Vegetation Index (NDVI) measure- relation between step length and turning angle [29]. Ob- ments at 250-m resolution (1 layer per month, MODIS servations of turn angle and step length were placed into Science Team, http://modis.gsfc.nasa.gov/). Chlorophyll 10° bins and 50 m bins respectively. A unique distribution strongly absorbs visible light but strongly reflects near- for each individual was created by calculating a probability infrared light allowing the use of remote sensing to distribution using values from all other animals exclu- measure and compare the intensity of light emitted in ding the individual itself, thereby avoiding problems these wavelengths, thereby enabling photosynthetic cap- of circularity [54]. These distributions were then used acity of vegetation to be quantified [37]. This measure- to create 10 random steps per real step using GME with ment is widely used in ecological studies, being highly ARCMAP v. 10.1. correlated with green herbaceous phytomass [52]. Al- though high NDVI values from conifer forested areas may not necessarily represent the ground-level forage SSF analysis availability, ground estimates of herbaceous forage bio- We developed separate models for dispersers in the mass for elk correlate with satellite derived NDVI values time periods before, during and after dispersal as well in tree-covered vegetation types [22]. as models for matched residents during these same Killeen et al. Movement Ecology 2014, 2:15 Page 11 of 13 http://www.movementecologyjournal.com/content/2/1/15 time periods. We split the during-dispersal subset of data Habitat selection at the broad-scale using correlated using a segmented regression or ‘broken-stick’ model to random walk analysis of dispersers define two types of movement behaviour (shorter steps We tested for the landscape characteristics defining real during resting and foraging and less-frequent long move- dispersal routes at a broad-scale by comparing each route ment steps) [30]. This was carried out using the R package to 10 random alternatives created using the simple.crw segmented [55]. Two further models were then fitted to function in GME. Step length and turning angle distribu- each of these subsets. For ease of comparison between tions were specified from the observed distributions from time periods we used the same explanatory variable struc- each individual, using only the long movement steps, as ture for all models, sensu Muhly et al. 2010 [50]. To allow defined by the segmented regression model. This ensured comparison between variables we scaled each variable by that random alternatives were of a comparable length to subtracting the mean and dividing by the standard devi- the actual dispersal route. We used the same landscape ation of the input variable. variables as in the previous analysis and compared the For all models we used a two-stage modelling approach, observed points during dispersal to the random points fitting models first to individuals and then averaging pa- derived from the correlated random walks. We used a rameters across individuals to estimate population-level generalised linear model per individual with a binomial selection parameters [56]. Before including variables we link function and averaged the resulting coefficients for evaluated them in terms of collinearity and multicollinear- a population estimate. We then performed one-sample ity using Pearson correlation coefficients (r) and variance Wilcoxon signed rank tests for each variable to determine inflation factors (all r < 0.3 and all VIF < 2) as well as for if the beta averages were significantly different from zero. biological meaningfulness. We used conditional logistic regression to estimate SSFs assuming an exponential se- Additional files lection function of the form: Additional file 1: Net Squared Displacement graphs for individuals included in SSF analyses. Figure S1. Net Squared Displacement wðÞ x ¼ exp β x þ β x þ β x þ … þ β x 1 2 3 p (NSD) calculated for n = 10 dispersers (D), all of which are male. Of the 1 2 3 p dispersers, E007, E010 and E095 undergo exploratory movements in which they return to, or close to, previous ranges. Beside each individual is the NSD graph for the extracted dispersal period. Figure S2. Net where β to β are coefficients estimated by conditional 1 p Squared Displacement (NSD) calculated for n = 10 residents (R), all of logistic regression, which are associated with a vector x, which are female. of environmental variables x to x respectively. The 1 p Additional file 2: Table S1. k-fold cross-validation results. Robustness of higher the value of wðÞ x the more likely that step will be models was evaluated by k-fold cross validation for case-control design, following the method of Fortin et al. 2009 [60]. The SSFs were built using chosen by an animal [57]. a random selection of 80% of the data strata and then used to predict The analysis was carried out using version 1.2 of the R the SSF scores for the remaining 20% of strata. The observed location package TwoStepCLogit [58]. To correctly analyse data was then ranked against the random locations, for each stratum, and ranks were tallied. Spearman rank correlations (r ) were carried out with in which there are multiple controls per case it is neces- S the bin’s ranking and associated frequency. The procedure was repeated sary to use conditional logistic regression. However, it is 100 times and the mean and range of r are shown. The mean and difficult to take into account cluster-level variation (i.e., expected range of r are also shown, calculated by ranking one random location against the other random locations, per strata, tallying ranks and random variation among elk) using this method, and this using Spearman rank correlations (r ). Again this was repeated 100 times. is important to address given that resource selection can vary considerably between individuals within a popula- Competing interests tion and large differences in individual behaviour have The authors declare that they have no competing interests. been observed amongst elk [45,59]. To account for vari- ation among individuals the TwoStepCLogit package was Authors’ contributions used to first fit fixed-effects regression models to each All co-authors participated in writing the manuscript. SC, DP, MM and MB collected and processed data and JK and HT conducted analyses. All authors individual elk and then combine those estimates using read and approved the final manuscript. restricted maximum likelihood (REML) estimation [58]. This method gives stable and consistent estimations of Acknowledgements the parameters in mixed conditional logistic regression We thank the Natural Sciences and Engineering Research Council of Canada models when the number of strata is large, as it was in (NSERC-CRD), Shell Canada Limited, Alberta Conservation Association (Grant Eligible Conservation Fund), Alberta Environment and Sustainable Resource this dataset. By using this procedure we account for the Development, Safari Club International, Alberta Parks, and Parks Canada for large inter-individual variation observed amongst the elk. funding and support. We also thank the Carl Tryggers foundation for The two-stage approach also helps to account for cor- scientific research (Swedish Post-doc grant) and the European Union (Erasmus Mundus Masters scholarship). The funders had no role in study relation within individuals, common in habitat-selection design, data collection and analysis, decision to publish, or preparation of studies (autocorrelation) [56]. Models were validated using the manuscript. We thank Daniel Fortin and an anonymous reviewer for their k-fold cross validation [60]. comments which greatly improved the manuscript. Killeen et al. Movement Ecology 2014, 2:15 Page 12 of 13 http://www.movementecologyjournal.com/content/2/1/15 Animal care 20. Ciuti S, Apollonio M: Do antlers honestly advertise the phenotypic quality Our data collection complied with all relevant federal laws of Canada and of fallow buck (Dama dama) in a lekking population? Ethology 2011, provincial laws of Alberta. Procedures adopted in this study were reviewed 117:133–144. and approved by the University of Alberta Animal Care and Use Committee 21. Dobson FS: Competition for mates and predominant juvenile male ACUC – Biosciences (Animal care protocol # 536-1003 AR University of dispersal in mammals. Anim Behav 1982, 30:1183–1192. Alberta, Edmonton, Canada), by all jurisdictions of the Alberta Government 22. Hebblewhite M, Merrill E, McDermid G: A multi-scale test of the forage (Permit Numbers: BI-2008-19, RC-06SW-001 and 23181CN), and by Parks maturation hypothesis in a partially migratory ungulate population. Canada (Permit Numbers: WL-2010-7292, WL-2010-5755). Ecol Monogr 2008, 78:141–166. 23. Boyce MS: The Jackson Elk Herd: Intensive Wildlife Management in North Author details America. Cambridge, U.K.: Cambridge University Press; 1989. Department of Biological Sciences, University of Alberta, Edmonton, AB T6G 24. Kalb DM, Bowman JL, Eyler TB: Dispersal and home-range dynamics of 2E9, Canada. Marine Evolution and Conservation Group, Centre of exotic, male sika deer in Maryland. Wildl Res 2013, 40:328–335. Evolutionary and Ecological Studies, University of Groningen, PO Box 11103 25. Petersburg ML, Alldredge AW, Vergie WJ: Emigration and survival of CC, Groningen, The Netherlands. Department of Biometry and 2-year-old male elk in northwestern Colorado. Wildl Soc Bull 2000, Environmental System Analysis, University of Freiburg, Freiburg 79106, 28:708–716. Germany. Faculty of Environmental Design, University of Calgary, Calgary, AB 26. Long ES, Diefenbach DR, Rosenberry CS, Wallingford BD: Multiple T2N 1 N4, Canada. proximate and ultimate causes of natal dispersal in white-tailed deer. Behav Ecol 2008, 19:1235–1242. Received: 4 March 2014 Accepted: 12 June 2014 27. Fryxell JM, Hazell M, Borger L, Dalziel BD, Haydon DT, Morales JM, Published: 1 July 2014 McIntosh T, Rosatte RC: Multiple movement modes by large herbivores at multiple spatiotemporal scales. Proc Natl Acad Sci U S A 2008, 105:19114–19119. 28. Borger L, Fryxell JM: Quantifying individual differences in dispersal using References net squared displacement. In Dispersal and Spatial Evolutionary Ecology. 1. Bohonak AJ: Dispersal, gene flow, and population structure. Q Rev Biol Edited by Clobert MB J, Benton T, Bullock J. Oxford, UK: Oxford University 1999, 74:21–45. Press; 2012. 2. Clobert J, Danchin E, Dhondt AA, Nichols D: Editors: Dispersal. Oxford: Oxford University Press; 2001:311–340. 29. Thurfjell H, Ciuti S, Boyce M: Applications of step-selection functions in 3. Dunning JB, Stewart DJ, Danielson BJ, Noon BR, Root TL, Lamberson RH, ecology and conservation. Movement Ecol 2014, 2:4. Stevens EE: Spatially explicit population models - current forms and 30. Johnson CJ, Parker KL, Heard DC, Gillingham MP: Movement parameters of future uses. Ecol Appl 1995, 5:3–11. ungulates and scale-specific responses to the environment. J Anim Ecol 4. Bowler DE, Benton TG: Causes and consequences of animal dispersal 2002, 71:225–235. strategies: relating individual behaviour to spatial dynamics. Biol Rev 31. Jerde CL, Visscher DR: GPS measurement error influences on movement 2005, 80:205–225. model parameterization. Ecol Appl 2005, 15:806–810. 32. Edge WD, Marcum CL, Olson SL, Lehmkuhl JF: Nonmigratory cow elk herd 5. Chetkiewicz CLB, Clair CCS, Boyce MS: Corridors for conservation: ranges as management units. J Wildl Manage 1986, 50:660–663. integrating pattern and process. Ann Rev Ecol Evol Syst 2006, 37:317–342. 6. Tesson S, Edelaar P: Dispersal in a changing world: opportunities, insights 33. Van Moorter B, Visscher D, Benhamou S, Börger L, Boyce MS, Gaillard J-M: and challenges. Movement Ecol 2013, 1:10. Memory keeps you at home: a mechanistic model for home range 7. Cagnacci F, Boitani L, Powell RA, Boyce MS: Animal ecology meets GPS- emergence. Oikos 2009, 118:641–652. based radiotelemetry: a perfect storm of opportunities and challenges. 34. Bruggeman JE, Garrott RA, White PJ, Watson FGR, Wallen R: Covariates Philos Trans R Soc B-Biol Sci 2010, 365:2157–2162. affecting spatial variability in bison travel behavior in Yellowstone 8. Taylor PD, Fahrig L, Henein K, Merriam G: Connectivity is a vital element of National Park. Ecol Appl 2007, 17:1411–1423. landscape structure. Oikos 1993, 68:571–573. 35. Parker KL, Robbins CT, Hanley TA: Energy expenditures for locomotion by 9. Zeller KA, McGarigal K, Whiteley AR: Estimating landscape resistance to mule deer and elk. J Wildl Manage 1984, 48:474–488. movement: a review. Landsc Ecol 2012, 27:777–797. 36. Frair JL, Merrill EH, Visscher DR, Fortin D, Beyer HL, Morales JM: Scales of movement by elk (Cervus elaphus) in response to heterogeneity 10. Adriaensen F, Chardon JP, De Blust G, Swinnen E, Villalba S, Gulinck H, in forage resources and predation risk. Landsc Ecol 2005, 20:273–287. Matthysen E: The application of 'least-cost' modelling as a functional landscape model. Landsc Urban Plan 2003, 64:233–247. 37. Pettorelli N, Vik JO, Mysterud A, Gaillard JM, Tucker CJ, Stenseth NC: Using 11. Beier P, Majka DR, Spencer WD: Forks in the road: choices in procedures the satellite-derived NDVI to assess ecological responses to environmental for designing wildland linkages. Conserv Biol 2008, 22:836–851. change. Trends Ecol Evol 2005, 20:503–510. 12. Pullinger MG, Johnson CJ: Maintaining or restoring connectivity of 38. Green RA, Bear GD: Seasonal cycles and daily activity patterns of Rocky modified landscapes: evaluating the least-cost path model with multiple mountain elk. J Wildl Manage 1990, 54:272–279. sources of ecological information. Landsc Ecol 2010, 25:1547–1560. 39. Skovlin JM, Zager P, Johnson BK: Elk habitat selection and evaluation. In 13. Squires JR, DeCesare NJ, Olson LE, Kolbe JA, Hebblewhite M, Parks SA: Elk of North America: Ecology and Management. Edited by Toweill DE, Combining resource selection and movement behavior to predict Thomas JW. Washington, DC: Smithsonian Institution Press; 2002:531–555. corridors for Canada lynx at their southern range periphery. Biol Conserv 40. Muhly TB, Semeniuk C, Massolo A, Hickman L, Musiani M: Human activity 2013, 157:187–195. helps prey win the predator-prey space race. Plos One 2011, 6:8. 14. Thatcher CA, van Manen FT, Clark JD: A habitat assessment for Florida 41. Ciuti S, Northrup JM, Muhly TB, Simi S, Musiani M, Pitt JA, Boyce MS: Effects panther population expansion into Central Florida. J Mammal 2009, of humans on behaviour of wildlife exceed those of natural predators in 90:918–925. a landscape of fear. Plos One 2012, 7:13. 15. Soulsbury CD, Iossa G, Baker PJ, White PCL, Harris S: Behavioral and spatial 42. Lewis JS, Rachlow JL, Garton EO, Vierling LA: Effects of habitat on GPS analysis of extraterritorial movements in red foxes (Vulpes vulpes). collar performance: using data screening to reduce location error. J Appl J Mammal 2011, 92:190–199. Ecol 2007, 44:663–671. 16. Newby J: Puma dispersal ecology in the Central Rocky Mountains. In 43. Bunnefeld N, Borger L, van Moorter B, Rolandsen CM, Dettki H, Solberg EJ, Master's thesis. Missoula, MT: University of Montana; 2011. Ericsson G: A model-driven approach to quantify migration patterns: 17. Singh NJ, Borger L, Dettki H, Bunnefeld N, Ericsson G: From migration to individual, regional and yearly differences. J Anim Ecol 2011, 80:466–476. nomadism: movement variability in a northern ungulate across its 44. Mysterud A, Loe LE, Zimmermann B, Bischof R, Veiberg V, Meisingset E: latitudinal range. Ecol Appl 2012, 22:2007–2020. Partial migration in expanding red deer populations at northern 18. Simberloff D, Farr JA, Cox J, Mehlman DW: Movement corridors - latitudes - a role for density dependence? Oikos 2011, 120:1817–1825. conservation bargains or poor investments. Conserv Biol 1992, 6:493–504. 45. Ciuti S, Muhly TB, Paton DG, McDevitt AD, Musiani M, Boyce MS: Human 19. Greenwood PJ: Mating systems, philopatry and dispersal in birds and selection of elk behavioural traits in a landscape of fear. Proc R Soc B-Biol mammals. Anim Behav 1980, 28:1140–1162. Sci 2012, 279:4407–4416. Killeen et al. Movement Ecology 2014, 2:15 Page 13 of 13 http://www.movementecologyjournal.com/content/2/1/15 46. Calenge C: The package "adehabitat" for the R software: A tool for the analysis of space and habitat use by animals. Ecol Model 2006, 197:516–519. 47. R Core Team: R: A Language and Environment for Statistical Computing. In R Foundation for Statistical Computing. 2013. 48. Paton DG: Connectivity of elk migration in southwestern Alberta. In Master's thesis. Calgary, Alberta: University of Calgary; 2012. 49. Bates D, Maechler M, Bolker B, Walker S: lme4: Linear mixed-effects models using Eigen and S4. R package version 1.0-5. 2013. http://cran.R-project. org/package=lme4. 50. Muhly TB, Alexander M, Boyce MS, Creasey R, Hebblewhite M, Paton D, Pitt JA, Musiani M: Differential risk effects of wolves on wild versus domestic prey have consequences for conservation. Oikos 2010, 119:1243–1254. 51. Riley SJ, DeGloria SD, Elliot R: A terrain ruggedness index that quantifies topographic heterogeneity. Intermountain J Sci 1999, 5:23–27. 52. Pettorelli N, Ryan S, Mueller T, Bunnefeld N, Jedrzejewska B, Lima M, Kausrud K: The Normalized Difference Vegetation Index (NDVI): unforeseen successes in animal ecology. Clim Res 2011, 46:15–27. 53. Turchin P: Quantitative Analysis of Movement: Measuring and Modeling Population Redistribution in Plants and Animals. Sunderland, MA: Sinauer Associates; 1998. 54. Fortin D, Beyer HL, Boyce MS, Smith DW, Duchesne T, Mao JS: Wolves influence elk movements: behavior shapes a trophic cascade in Yellowstone National Park. Ecology 2005, 86:1320–1330. 55. Muggeo VMR: segmented: an R package to fit regression models with broken-line relationships. R News 2008, 8:20–25. 56. Fieberg J, Matthiopoulos J, Hebblewhite M, Boyce MS, Frair JL: Correlation and studies of habitat selection: problem, red herring or opportunity? Philos Trans R Soc B-Biol Sci 2010, 365:2233–2244. 57. Lele SR, Merrill EH, Keim J, Boyce MS: Selection, use, choice and occupancy: clarifying concepts in resource selection studies. J Anim Ecol 2013, 82:1183–1191. 58. Craiu RV, Duchesne T, Fortin D, Baillargeon S: Conditional logistic regression with longitudinal follow-up and individual-level random coefficients: A stable and efficient two-step estimation method. J Comput Graph Stat 2011, 20:767–784. 59. Bolnick DI, Svanback R, Fordyce JA, Yang LH, Davis JM, Hulsey CD, Forister ML: The ecology of individuals: incidence and implications of individual specialization. Am Nat 2003, 161:1–28. 60. Fortin D, Fortin ME, Beyer HL, Duchesne T, Courant S, Dancose K: Group-size-mediated habitat selection and group fusion-fission dynamics of bison under predation risk. Ecology 2009, 90:2480–2490. doi:10.1186/s40462-014-0015-4 Cite this article as: Killeen et al.: Habitat selection during ungulate dispersal and exploratory movement at broad and fine scale with implications for conservation management. Movement Ecology 2014 2:15. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Movement Ecology Springer Journals

Habitat selection during ungulate dispersal and exploratory movement at broad and fine scale with implications for conservation management

Download PDF
13 pages

Loading next page...
 
/lp/springer-journals/habitat-selection-during-ungulate-dispersal-and-exploratory-movement-R0UxRywZFf

References (69)

Publisher
Springer Journals
Copyright
Copyright © 2014 by Killeen et al.; licensee BioMed Central
Subject
Life Sciences; Animal Ecology; Conservation Biology/Ecology; Terrestial Ecology
eISSN
2051-3933
DOI
10.1186/s40462-014-0015-4
pmid
27148450
Publisher site
See Article on Publisher Site

Abstract

Background: Dispersal has a critical influence on demography and gene flow and as such maintaining connectivity between populations is an essential element of modern conservation. Advances in satellite radiotelemetry are providing new opportunities to document dispersal, which previously has been difficult to study. This type of data also can be used as an empirical basis for defining landscapes in terms of resistance surfaces, enabling habitat corridors to be identified. However, despite the scale-dependent nature of habitat selection few studies have investigated selection specifically during dispersal. Here we investigate habitat selection during and around dispersal periods as well as the influence of age and sex on dispersal for a large ungulate. Results: Of 158 elk (Cervus elaphus) tracked using GPS radiotelemetry almost all dispersers were males, with individuals dispersing up to 98 km. The dispersal period was distinct, with higher movement rates than before or after dispersal. At fine scale elk avoided the most rugged terrain in all time periods, but to a greater extent during and after dispersal, which we showed using step selection functions. In contrast, habitat selection by resident elk was less affected by ruggedness and more by an attraction to areas of higher forage availability. At the broad scale, however, movement corridors of dispersers were characterized by higher forage availability and slightly lower ruggedness then expected using correlated random walks. Conclusions: In one of the first examples of its kind we document complete long-distance dispersal events by an ungulate in detail. We find dispersal to be distinct in terms of movement rate and also find evidence that habitat selection during dispersal may differ from habitat selection in the home-range, with potential implications for the use of resistance surfaces to define conservation corridors. Keywords: Alberta, Cervus elaphus, Dispersal, Elk, Habitat selection, Migration, Step selection functions Background developed, empirical studies are generally lacking because Dispersal is a fundamental process in ecology and evolu- of the difficulties associated with observing and quanti- tion, affecting individual fitness as well as demography, fying the dispersal process [6]. However, new advances in genetic structure, and species distributions [1-4]. Disper- satellite radiotelemetry now permit opportunities to docu- sal is likewise of central importance for managing animal ment dispersal that were previously unattainable [7]. populations; maintaining connectivity among populations Landscape connectivity describes how the movement is considered to be an essential part of modern conser- of animals is linked to landscape structure [8]. The way vation [5]. Although theoretical work on dispersal is well in which movement among populations is affected by environmental conditions is important for predicting the effects of landscape modification and habitat fragmenta- * Correspondence: [email protected] 1 tion, and in prioritising which habitats to protect. The Department of Biological Sciences, University of Alberta, Edmonton, AB T6G 2E9, Canada most widely used method for maintaining connectivity is Marine Evolution and Conservation Group, Centre of Evolutionary and the conservation corridor, a protected area of landscape Ecological Studies, University of Groningen, PO Box 11103 CC, Groningen, that facilitates the movement of organisms between po- The Netherlands Full list of author information is available at the end of the article pulations [5]. One approach has been to map resistance © 2014 Killeen et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Killeen et al. Movement Ecology 2014, 2:15 Page 2 of 13 http://www.movementecologyjournal.com/content/2/1/15 surfaces to characterize how environmental variables af- During 2007-2011 we monitored elk (Cervus elaphus) fect animal movement, and to use these surfaces to model in the Rocky Mountains of North America using GPS connectivity and thus to inform management decisions telemetry to address these questions. Elk herds in this relating to corridors [9]. As the use of Geographic Infor- area may be migratory, in which animals move to higher mation Systems (GIS) has become more widespread, land- altitudes in spring and summer to gain access to high- scapes are increasingly defined using resistance surfaces, quality forage, or partially migratory, in which some ani- allowing the use of methods such as least-cost path ana- mals remain resident in winter ranges throughout the year lysis to help determine the most effective placement of [22]. Dispersal occasionally takes place, when an animal corridors [10]. However, the effectiveness of such an ap- leaves its home range and does not return. We expected proach is highly dependent on the accurate assignment of that young male elk would be more likely to disperse than resistance values to landscape units, yet this has mostly females [23]. Many male cervids are known to disperse been a subjective process [9,11]. Using empirical data gen- from their natal home range, including elk [24-26]. We erally leads to more robust and more readily justifiable also predicted that dispersers would select habitats based conclusions [12]. more on ease of movement through the landscape than Where empirical data have been used, it frequently has forage quality and we expected this pattern to be clear been assumed that habitat selection parameters derived during dispersal, while before and after dispersal we ex- from movement within home ranges can be extrapolated pected forage quality to be more important in habitat to movement during dispersal [13,14]. Corridors aim to selection. The dispersal period also was expected to be promote connectivity between populations and therefore distinct in terms of movement parameters, with faster and need to facilitate dispersal movement, making this as- more directional movement. Furthermore we expected sumption a potential source of error [9]. There is currently there to be distinct movement behaviours at the fine- little information to suggest whether this assumption is scale; shorter foraging movements and longer directional reasonable or not, although Soulsbury et al. [15] found movements [27]. We predicted that longer steps would be that red foxes (Vulpes vulpes) have unique patterns of selected based more on ease of movement while shorter habitat selection during dispersal, while Newby [16] found steps would be selected based more on forage quality. We that cougars (Puma concolor) do not. The general lack of used resident individuals for further comparison, because data on this subject probably reflects the difficulty of of their consistent home-ranging behaviour, and we ex- measuring and tracking dispersal [6]. Predicting which an- pected habitat selection to be consistently less affected by imals are likely to disperse can be difficult, so tracking ease of movement and more affected by forage quality for those specific individuals is often not possible. Even with a these individuals. large sample size the number of dispersing individuals We tested these predictions by first differentiating dis- within a population can be small [13,17]. Given that badly persal from other types of movement, such as migration designed corridors run the risk of acting as population and residency, on the basis of the measurement Net- sinks or simply wasting financial resources and eroding Squared Displacement (NSD) [28] and extracting tele- the support of stakeholders [11,18], this is an important metry points within the dispersal period. We quantified issue to address. Greater understanding of habitat selec- the distance and duration of dispersal using these data. tion by animals during dispersal should lead to better in- We then used step selection functions (SSF) and step- formed conservation management decisions. length analysis to compare fine-scale habitat selection Here we investigate how age and sex influence the between time periods for dispersing individuals and for likelihood of dispersal for an ungulate. Sex differences matched resident individuals [29]. We also investigated in juvenile dispersal are common and for mammals it is the dispersal period itself using segmented regression to most often young males that disperse [19]. In a polygyn- split movement behaviours into different scales of move- ment, before analysing each separately [30]. Finally, we ous mating system males are expected to have more variation in reproductive success because a male must used correlated random walks to analyse selection at a secure a territory or become dominant to increase re- broad scale. While step selection functions can be used to analyse animal behavioural decisions during move- productive success [20]. Therefore it is important for ju- venile males to avoid sexual competition with older and ment at a scale measured in hours, correlated random more powerful males, and consequently dispersal is fa- walks can be used to analyse movement at the scale of a complete dispersal pathway. voured [21]. We then investigate whether habitat se- lection during dispersal differs from habitat selection during home-ranging and other behaviours. We hypo- Results thesise that habitat selection during dispersal will have a Dispersal and movement parameters unique pattern based on ease of movement through the Of the 132 elk (94 female, 38 male) with complete data, landscape. 97 were identified as migratory and 17 as resident. There Killeen et al. Movement Ecology 2014, 2:15 Page 3 of 13 http://www.movementecologyjournal.com/content/2/1/15 were a total of 16 dispersal events, all but one under- although movement rate in all was lower compared to taken by males, 9 of which completed dispersal and 7 of that recorded during the rest of the year (reference ca- which were exploratory movements. Of the males tracked tegory, β =0). st st for the required time period (at least 1 April to 31 Movement data during the dispersal period were fur- October), this translates to 39% of males dispersing (24% ther divided into two parts using segmented regression, completing dispersal, 16% showing an exploratory move- representing two distinct movement behaviours [30]. ment) (Table 1). The breakpoint was identified as a movement rate of -1 Dispersal events ranged in length from 29.16 km to 7.12 m min (SE = 1.04) which corresponds to a step 98.01 km (straight line distance from first to last loca- length of 854 m in 2 hours (Figure 3). tion, actual distance travelled greater) and lasted from This resulted in two subsets of data – one of steps 12 to 47 days (M = 25.9 days), taking place between 18- under 854 m, representing shorter steps during foraging May and 04-August, with the majority of movement or resting (n = 1,292 steps), and the other of steps over occurring in June and July. Dispersal routes traversed a 854 m, representing longer movement steps (n = 355 range of landscapes, with some animals travelling from steps). Turning-angle distribution was distinct between Alberta into British Columbia or Montana (Figure 1). the groups, with longer steps having smaller turning an- Exploratory movements were similar to dispersal move- gles and therefore being more directional (Figure 4). ments in terms of timing and distance travelled but the Turning angles associated with short steps also are bur- animals returned to their starting locality. These move- dened with greater sampling error [31]. ments occurred on a timescale of weeks, inconsistent with migration, and we consider them as explorations Fine-scale habitat selection (Additional file 1: Figure S1). All males were captured at For dispersers the relationship between step length and approximately age 1.5 and therefore were dispersing terrain ruggedness was strongly non-linear with selection when just over 2 years old. for intermediate values of ruggedness during all time pe- For dispersers, according to the predictions of linear riods (Table 2). They avoided more rugged terrain in all -1 mixed effects analysis, movement rate (m hr ) recorded time periods, but this avoidance was weaker before disper- during different time periods (Figure 2) had the follow- sal (Figure 5a). Dispersers also had a weak attraction to ing pattern: during dispersal (β = 173.9; 95% CI 162.4, areas with high Normalized Difference Vegetation Index 185.3) > > before dispersal (β = 57.5; 95% CI 46.5, 68.4) > > (NDVI) values and to some extent avoided areas with high after dispersal (β = 4.0, 95% CI -7.28, 15.3). Movement rate canopy cover, strongly before dispersal and weakly during after dispersal did not differ from that recorded throughout and after dispersal, with no evidence for a non-linear ef- the rest of the year (reference category, β =0). Resi- fect. Distance to roads had no or a weak effect in all pe- dents meanwhile did not show significant differences riods examined. in movement rate between the periods during (β = -10.5, During the dispersal period, the shorter steps, as iden- 95% CI -20.7, -0.48; Figure 2), before (β = -14.3; 95% tified by segmented regression, were still characterised CI -24.5, -4.1) and after (β = -20.6; 95% CI -31.5, -9.7), by strong avoidance of more rugged terrain and selec- tion for intermediate values. Longer steps showed a lin- ear avoidance of ruggedness (Figure 5b), but this model was not highly predictive. The short-step group of steps Table 1 Movement classifications using net squared were more likely to end in resource units with higher displacement NDVI values while the long-step group had no effect of Male Female Total Percentage (%) NDVI (Table 2). (n = 38) (n = 94) For residents there was also a non-linear relationship Migratory 26 71 97 73 between step length and terrain ruggedness but this ef- Resident 2 15 17 13 fect was only strong in one time period. There was in Other/not classified 10 8 18 14 general a weak avoidance of more rugged terrain, but in Dispersal event 15 (6) 1 (1) 16 contrast to dispersers there was a strong attraction to A total of 132 animals were observed between 2007 and 2011, with data from areas with high NDVI values in all time periods. There st st at least April 1 to October 31 , using GPS radiotelemetry and were classified was also a non-linear relationship with canopy cover in by movement type; migrant or resident. Some animals had an atypical movement pattern and were classified as other/not classified. The number of all time periods with residents selecting intermediate animals which had a dispersal period (number of those which were values while strongly avoiding areas with high canopy exploratory movements in brackets) was counted. Depending on the cover (Table 2). availability of data in the year following dispersal some dispersing animals could also be classified as migrants or residents. Those without data in the All models, except for the long-steps model, were following year fall under the ‘other’ category. Classifications were made by found to be useful predictors of habitat selection, with classifying and inspecting Net Squared Displacement graphs sensu Mysterud et al. 2011 [44] (see Additional file 1: Figure S1). the mean observed r of the models greater than 0.65, S Killeen et al. Movement Ecology 2014, 2:15 Page 4 of 13 http://www.movementecologyjournal.com/content/2/1/15 Figure 1 The study area and telemetry data overview. Map of the study area, located in Alberta and British Columbia, Canada and Montana, USA. Telemetry data was collected between 2007 and 2011. Locations from dispersing elk, during their dispersal period, are shown as points (each colour corresponds to an individual elk) on a digital elevation model of the area (lighter areas representing higher elevation). Minimum convex polygons for the full year of data for each matched resident are also shown in black. and the distribution of r higher than expected by chance found no significant pattern with canopy cover nor dis- alone (Additional file 2: Table S1). tance to roads (Wilcoxon signed rank test, mean β =-0.03, P = 0.85; mean β =0.25 P = 0.28 respectively). Broad-scale habitat selection Using correlated random walks with turning angle and Discussion step-lengths from the distribution of observed during- Dispersal and movement parameters dispersal long steps, we compared each dispersal path- We observed a number of substantial long-distance dis- way to random paths. For 7 of the 10 animals there was persal events, all but one of which was made by males. an avoidance of rugged terrain at this broad scale but Of the 38 males tracked for the required time period overall this was not significantly different from zero 39% had a dispersal period (24% completed) which is (mean β = -0.66, Wilcoxon signed rank test, P = 0.28). comparable to previous estimates of 27% for herds in However, for NDVI there was a more consistent attraction Colorado [25] and 40% in a Montana herd [32]. This to higher values (mean β = 0.99, Wilcoxon signed rank pattern of male-biased dispersal is expected for a po- test, P = 0.03) with 8 of 10 animals showing selection. We lygynous mammal species, because males have highly Killeen et al. Movement Ecology 2014, 2:15 Page 5 of 13 http://www.movementecologyjournal.com/content/2/1/15 -1 Figure 2 Movement rate of dispersers and residents. Movement rates (m hr ) before, during and after the dispersal event and throughout the rest of the year for a) dispersers and b) residents The mean movement rate in each period is shown with error bars representing ±1 SE. For dispersers, movement rate in the during-dispersal time period was significantly greater than in all other time periods and the before-dispersal period was also greater than the after-dispersal period, which was not different from the rest of the year. For residents, there was no significant difference in mean movement rate before, during and after the dispersal period, and in all these periods they moved slower than during the rest of the year. variable reproductive success [19]. Our estimate is, if may affect movement rate of animals and this was not ex- anything, likely to be an underestimate because many plicitly examined, although resident and dispersing ani- males (28% of those tracked) were killed before we were mals were paired in time to account for seasonal effects. able to gather enough data to confirm if they were dis- Resident animals had similar movement rates in all time persers or migrants. Given the lack of male residents periods, while dispersers did not. and female dispersers, we cannot assess the effect of sex on our models. However, we had a large sample size and Fine-scale habitat selection age range of females and it is clear that dispersal is al- Steps taken by dispersing elk showed a non-linear re- most totally male-dominated and therefore the charac- lationship with terrain ruggedness in all time periods teristics of dispersal behaviour appear inextricably linked examined, selecting intermediate values of ruggedness, to males. while avoiding the most rugged terrain. Avoidance of Movement rate for dispersers during the dispersal pe- rugged terrain has been observed for other mammals riods was greater than movements before or after. This [13,34] and presumably reduces energy expenditure and shows that the extracted dispersal period is indeed dis- facilitates ease of movement [35]. Unexpectedly, the re- tinct in terms of movement rate. Dispersing animals sponse by dispersing elk to terrain ruggedness was simi- move faster and further and maintain a persistent direc- lar both during and after dispersal. This may be because tion. There was a significant difference between the be- these individuals were experiencing novel environments fore dispersal movement rate in comparison to the rest and minimising energy expenditure while exploring and of the year, possibly representing a period of restlessness forming a new home-range after dispersal. We also split in the pre-dispersal phase. In the period after dispersal the during-dispersal period into two movement be- movement rate reduced, during which time there was haviours. A variety of methods have been used to infer likely to be a slow formation of a new home-range. Within different movement behaviours from step-length data, their home ranges animals are likely to switch between providing insight into movement decisions [27,30,36]. long directional movements between patches and periods We found that when only looking at long movement of short non-directional movements within a patch [27]. steps, in comparison to short steps during foraging or This is probably based on spatial memory of resources resting, the avoidance of ruggedness was linear for longer within their home-range [33]. However, seasonality also steps. More avoidance of rugged terrain during longer Killeen et al. Movement Ecology 2014, 2:15 Page 6 of 13 http://www.movementecologyjournal.com/content/2/1/15 Figure 3 Segmented regression of movement rate. The log frequency distribution of movement rates for all dispersers combined, during their dispersal period. Movement rates were binned and the log frequency in each bin was plotted, sensu Johnson et al. 2002 [30]. A segmented 2 -1 linear regression model was fit (red lines, r = 0.93) for which a breakpoint of 7.118 m min was identified using the R package segmented. A null model linear regression model is also shown for comparison (black, dashed line, r = 0.82). steps might be expected given that only when there is con- weak, and it may be that dispersers foraged wherever pos- siderable movement is the energy saved by avoiding rug- sible, rather than being highly selective, or that the pattern ged terrain likely to be substantial, although steps may be of foraging might be obscured with steps during resting shorter in rugged terrain due to the difficulty of move- behaviour, all of which fall within the same group. Dif- ment. However, the long-steps model was not validated, ferentiating resting and foraging steps is difficult given showing that habitat selection for these long movement the imprecision inherent in GPS data [36]. Residents also steps may be somewhat unpredictable. showed avoidance of areas with high canopy cover and se- In contrast, resident elk had inconsistent responses to lected movement toward localities with lower canopy ruggedness, although did still avoid more rugged terrain. cover, consistent with elk use of forest cover for protection However unlike dispersers, they showed strong and highly while using open areas for foraging [39]. consistent selection for areas with higher NDVI values. High values of NDVI indicate green herbaceous phyto- mass or forage quantity [37] suggesting that residents are Broad-scale habitat selection selecting areas with high forage availability the majority of While habitat selection at fine scale examines behav- the time. This is consistent with previous studies showing ioural decisions over the course of hours, we also used that foraging dominates summer elk activity [36,38]. Dur- broad-scale analyses to examine characteristics of the ing dispersal we expected that short steps would show a dispersal pathway as a whole. At this larger scale we strong attraction to high NDVI values in comparison to found that there was no clear overall avoidance of rug- long steps, because they should be associated with for- ged terrain, which was surprising given the strength of aging when moving less rapidly. In fact this pattern was the relationship at fine scale. This suggests that although Killeen et al. Movement Ecology 2014, 2:15 Page 7 of 13 http://www.movementecologyjournal.com/content/2/1/15 Figure 4 Turning angle distributions for short and long steps during dispersal. Turning angle distributions combined for the 10 dispersers during their dispersal period are shown. The upper graph shows the distribution of turning angles for steps longer than 854 m (n = 355 steps) and the lower graph shows the distribution of turning angles for steps shorter than 854 m (n = 1,292 steps). This boundary was derived from the segmented regression breakpoint in Figure 3. dispersing animals appear to select habitats for ease of but perhaps also because they select the direction of dis- movement at fine scale during movement, at a larger persal based on a broad assessment of habitat and for- scale they do not necessarily select their overall route age quality and along that route select habitats for ease in this way. This could be because of imperfect know- of movement at the fine scale. This is supported by the ledge of their environment – dispersers are travelling overall positive relationship of movement toward re- through areas that they have never encountered before, source units with high NDVI values. Table 2 SSF model parameter estimates (fine-scale) Ruggedness Ruggedness ^2 NDVI Distance roads Canopy cover Canopy cover ^2 Beta (se) Beta (se) Beta (se) Beta (se) Beta (se) Beta (se) Dispersers before 0.434 (0.100) - 0.202 (0.052) 0.623 (0.416) 0.457 (0.242) - 0.238 (0.108) - 0.185 (0.172) Dispersers during 0.155 (0.097) - 0.135 (0.044) 0.288 (0.175) 0.113 (0.158) - 0.113 (0.082) - 0.053 (0.068) Dispersers after 0.144 (0.106) - 0.156 (0.035) 0.860 (0.443) 0.354 (0.203) - 0.186 (0.139) - 0.002 (0.129) Residents before - 0.112 (0.125) - 0.281 (0.081) 1.031 (0.345) 0.202 (0.153) 0.132 (0.092) - 0.291 (0.078) Residents during - 0.087 (0.146) - 0.101 (0.085) 0.996 (0.309) 0.177 (0.169) 0.238 (0.072) - 0.216 (0.065) Residents after 0.111 (0.075) - 0.103 (0.085) 0.598 (0.209) 0.047 (0.188) 0.229 (0.095) - 0.263 (0.086) Disp. during long-steps - 0.236 (0.151) - 0.054 (0.054) 0.066 (0.278) - 0.114 (0.298) - 0.121 (0.108) - 0.029 (0.119) Disp. during short-steps 0.276 (0.117) - 0.171 (0.057) 0.385 (0.227) 0.211 (0.136) - 0.073 (0.102) - 0.099 (0.085) Beta values with standard errors in brackets are shown, from each of the SSF models. Beta values greater than 2 SE from 0 are shown in bold. Killeen et al. Movement Ecology 2014, 2:15 Page 8 of 13 http://www.movementecologyjournal.com/content/2/1/15 Figure 5 The response of dispersers to ruggedness. The relationship between the selection coefficients (beta values) for ruggedness. In a) dispersers are shown for the before, during and after periods of dispersal. Avoidance of ruggedness is stronger during and after dispersal compared to before dispersal. In b) the long step and short step groups, as derived from segmented regression (Figure 3), are shown for the dispersers during their dispersal period. Avoidance of ruggedness appears stronger for long steps than for short steps. Conclusions about habitat selection throughout a landscape. For ungu- In one of the first examples of its kind we were able to lates, it would be most useful to track young males, espe- track dispersing elk using high fix-rate GPS satellite ra- cially during spring and early summer, because these are diotelemetry throughout the entirety of their dispersal the animals most likely to disperse. periods, providing detailed information on the distance Overall we believe that the processes of dispersal amongst and direction travelled and the likelihood of dispersal. large mammals warrant further investigation and modern By analysing habitat selection with these data we find GPS radiotelemetry provides an excellent tool in this en- evidence to suggest that the most-supported step selec- deavour. We also find step selection functions to be a tion functions during dispersal are not the same as those powerful statistical tool for analysing habitat selection dur- selected for home-ranging behaviour. Where possible it ing dispersal. would be beneficial to estimate and account for the dif- ferences when using cost-distance modelling. If managers Methods are to implement initiatives to promote habitat connectiv- Study area ity, such as corridors, it would be ideal to prioritise data The study took place in southwest Alberta and extended collection from individuals that are most likely to disperse. into northwest Montana and southeast British Columbia This serves a double purpose. One, it ensures that the data (Figure 1). The majority of the area within Alberta is obtained are as representative as possible for characteris- provincial forest reserve and on the eastern boundary of ing dispersal movement, the primary purpose for habitat our study area there is mixed livestock ranching and corridors. Second, dispersers cover much larger distances cropland. This boundary is a transition zone from grass- during their travel than other animals, making them a land into the Rocky Mountains and several different elk cost-effective way to get the maximum amount of data populations are present. Natural predators of elk in the Killeen et al. Movement Ecology 2014, 2:15 Page 9 of 13 http://www.movementecologyjournal.com/content/2/1/15 area are wolf (Canis lupus), cougar (Puma concolor), and these animals as migrant or resident, as well as their dis- grizzly bear (Ursus arctos) [40]. There is considerable persal event. Those without long enough periods of data human presence in the study area, including industrial following dispersal were classified as other/not classified, activities such as forestry and natural gas extraction, as along with a number of animals with atypical movement well as recreational activities, especially during summer patterns. Several males had NSD graphs suggestive of and the autumn hunting season [41]. dispersal, but were killed by hunters during the autumn hunting season (September to end of November for rifle Elk data hunting [45]), making it difficult to confirm if they were A total of 158 elk (105 female, 53 male) were captured truly dispersers, or migratory. A number of individuals on winter ranges between 2007 and 2011 using helicopter also showed an ‘exploratory movement’ (Additional file 1: net-gunning. Each was fitted with a GPS-radiotelemetry Figure S1). These movements were similar to dispersal collar (either ARGOS GPS for males or Lotek GPS 4400 movements, occurring within the same timeframe before for females, Lotek Wireless Inc., Ontario, Canada) and all migration, rapidly travelling a long distance, but then units were programmed to obtain locations every 2 hrs. returning to or close to their original range within a short Satellite transmitted data from GPS collars fitted with period of time. These appear to be exploration events, and Argos communication devices were received weekly via therefore were included as dispersers in our analyses, al- email, while other data were downloaded remotely in the though we note that our results were not notably altered field. Radiocollars from elk that died were located and by inclusion of these individuals. Our categorisation re- re-fitted to new animals. All males were approximately sulted in a group of elk-years with clearly distinguishable 1.5 yrs old at time of capture and females varied be- movement patterns (Additional file 1: Figures S1, S2). We tween 1.5 yrs and 19 yrs old. A vestigial canine was re- calculated NSD and plotted the associated graphs using moved during capture and used to determine age by the adehabitat package version 1.8 [46] in version 2.15 cementum analysis (Matson’s Laboratory, MT, USA). of R [47]. Locations were screened following the method of Lewis et al. 2007 [42] but a few large measurement errors Data selection and movement parameters remained. These outliers were easily identified as loca- To investigate habitat selection by dispersers we selected tions which were an unreasonable distance from the individuals that had a dispersal period (n =10, 7 of which previous and next location (10s of km round trips in completed dispersal, 3 of which had exploratory move- 4 hrs) and were removed. Data sets also were trimmed ments) and for which there were resident individuals with at beginning and end to remove data where an elk’sbe- data spanning the same timeframe (n =10) available for haviour might have been influenced by capture or where comparison (n = 20 individuals, 57,637 telemetry reloca- the elk had died or where the collar failed to function. tions in total, locations shown in Figure 1). Residents were Fix rate for the 20 animals included in the analysis was used for comparison because they have a simple move- 81.7% for Lotek 4400 (females) and 66.2% for ARGOS ment pattern in which they stay within the same home GPS (males). range throughout the year. We did not investigate mi- grants because migration in this population is typically Distinguishing movement types characterised by frequent stop-overs while following Only data from individuals with relocations from at least spring green-up towards summer ranges, rather than st st 1 April to 31 October were included and of the 158 a single directed movement period [48]. This makes radio-collared animals, 26 (15 males, 11 females) were it difficult to compare the movement period of a migra- excluded. This was to ensure that the entirety of the mi- tion directly with that of a dispersal period. We selected gration or dispersal period was covered. To distinguish the telemetry points that occurred during the dispersal between different movement behaviours we used the mea- movement event itself, to the nearest day. This was done surement Net-Squared Displacement (NSD) [28]. This is a using NSD graphs, with the start of dispersal identified by time-dependent statistic that measures the straight-line a steep increase in the value of NSD and the end of dis- distance between a starting location and subsequent loca- persal identified by NSD plateauing as the animal settles tions in a movement path of a given individual. We used into a new home range (Additional file 1: Figure S1). If the method of Bunnefeld et al. [43] to classify each animal there was a short movement followed by a period of a as migrant or resident and to identify dispersal events. month or more of stationary behaviour, before the main Graphs of NSD were then inspected visually to identify dispersal event, this early movement was not considered the type of movement, similar to Mysterud et al. 2011 to be part of the dispersal event itself. [44]. After a successful dispersal animals will settle into a Also, we selected telemetry locations in the period of new home range and become either migrants or residents 26 days (corresponding to the average duration of a disper- and in some cases there were available data to classify sal event) immediately preceding and after the dispersal Killeen et al. Movement Ecology 2014, 2:15 Page 10 of 13 http://www.movementecologyjournal.com/content/2/1/15 event. This gave us a group of data points from dispersers A measure of distance to roads was used, derived from before they had undertaken their dispersal and after they road maps of the study area including both paved and had completed it. These same time periods also were ana- gravel roads (Governments of AB & BC: National Topo- lysed for matched residents even though there was no dis- graphic Database 1:50,000; U.S. Census Bureau Tiger/Line persal among these individuals. files, 2000). Finally we used a percent canopy cover surface -1 Movement rate (m hr ) was calculated by dividing the at 30 m resolution (U.S. National Land Cover Database, step length between two successfully obtained locations Governments of Alberta/British Columbia). Again, we in- by the time elapsed between those locations. We used a cluded both a linear and a squared term given an expected mixed effects model with individual elk as a random fac- non-linear relationship to canopy cover. tor to assess the effect of the time period (before, during, after dispersal, and the rest of the year) on movement Modelling fine-scale habitat selection using step selection rate. Movement rate throughout the rest of the year was functions set as reference category in the mixed model. A separate The straight-line segments linking successive animal lo- model was used for dispersers and for residents, using cations, taken at regular intervals, are defined as steps the R package lme4 [49]. In both models, statistically sig- [53]. We calculated step length in metres and turning nificant differences (α = 0.05) between movement rates angle (angle between previous and next location) using recorded during different time periods were assessed by GME with ARCMAP v. 10.1 (ESRI Inc., Redlands, CA). checking the overlap of 95% Confidence Intervals (CIs), In analyses involving step length, we used only steps assuming movement rates to be statistically different where the time interval was within 10 minutes of the (P < 0.05) if related 95% CIs didn’toverlap. typical two-hour interval. Steps with intervals much shorter or longer were removed to ensure fair compari- Landscape variables son of step length. We identified a number of environmental covariates that Step selection functions (SSF) are used for incorpo- are known to influence elk movement behaviour [45,50]. rating movement into habitat selection analysis, provid- Each covariate was imported into a GIS system and values ing a more fine-scale and mechanistic movement model for each were then attached to the points in the telemetry than the original Resource Selection Function (RSF) [54]. dataset using ESRI ArcGIS 10.1 (ESRI Inc., Redlands, CA) In an SSF each observed step is compared with a number and GME (Geospatial Modelling Environment, http:// of random steps that have the same starting point. Ran- www.spatialecology.com/gme/). dom step lengths and turning angles are drawn from dis- We used a 30-m resolution Digital Elevation Model tributions taken from observed data, allowing comparison (DEM) and the Spatial Analyst extension in ESRI ArcGIS of the observed step to a sample of those that could have 10.1 to produce a terrain ruggedness variable (30-m reso- been taken in the local environment. Here we analyse the lution), which quantifies topographic heterogeneity [51]. endpoints of observed and random steps, similar to a con- This is a measurement of the average elevation difference ditional RSF but with controls drawn from a domain most between a point on a digital elevation model grid and likely to represent those truly available given observed the surrounding cells. We included both a linear and movement patterns [29]. a squared term given an expected non-linear relation- Step lengths and turn angles for the random steps were ship to ruggedness. drawn independently from distributions of the observed As a proxy for forage quality we used monthly Nor- values, which was reasonable because of low observed cor- malized Difference Vegetation Index (NDVI) measure- relation between step length and turning angle [29]. Ob- ments at 250-m resolution (1 layer per month, MODIS servations of turn angle and step length were placed into Science Team, http://modis.gsfc.nasa.gov/). Chlorophyll 10° bins and 50 m bins respectively. A unique distribution strongly absorbs visible light but strongly reflects near- for each individual was created by calculating a probability infrared light allowing the use of remote sensing to distribution using values from all other animals exclu- measure and compare the intensity of light emitted in ding the individual itself, thereby avoiding problems these wavelengths, thereby enabling photosynthetic cap- of circularity [54]. These distributions were then used acity of vegetation to be quantified [37]. This measure- to create 10 random steps per real step using GME with ment is widely used in ecological studies, being highly ARCMAP v. 10.1. correlated with green herbaceous phytomass [52]. Al- though high NDVI values from conifer forested areas may not necessarily represent the ground-level forage SSF analysis availability, ground estimates of herbaceous forage bio- We developed separate models for dispersers in the mass for elk correlate with satellite derived NDVI values time periods before, during and after dispersal as well in tree-covered vegetation types [22]. as models for matched residents during these same Killeen et al. Movement Ecology 2014, 2:15 Page 11 of 13 http://www.movementecologyjournal.com/content/2/1/15 time periods. We split the during-dispersal subset of data Habitat selection at the broad-scale using correlated using a segmented regression or ‘broken-stick’ model to random walk analysis of dispersers define two types of movement behaviour (shorter steps We tested for the landscape characteristics defining real during resting and foraging and less-frequent long move- dispersal routes at a broad-scale by comparing each route ment steps) [30]. This was carried out using the R package to 10 random alternatives created using the simple.crw segmented [55]. Two further models were then fitted to function in GME. Step length and turning angle distribu- each of these subsets. For ease of comparison between tions were specified from the observed distributions from time periods we used the same explanatory variable struc- each individual, using only the long movement steps, as ture for all models, sensu Muhly et al. 2010 [50]. To allow defined by the segmented regression model. This ensured comparison between variables we scaled each variable by that random alternatives were of a comparable length to subtracting the mean and dividing by the standard devi- the actual dispersal route. We used the same landscape ation of the input variable. variables as in the previous analysis and compared the For all models we used a two-stage modelling approach, observed points during dispersal to the random points fitting models first to individuals and then averaging pa- derived from the correlated random walks. We used a rameters across individuals to estimate population-level generalised linear model per individual with a binomial selection parameters [56]. Before including variables we link function and averaged the resulting coefficients for evaluated them in terms of collinearity and multicollinear- a population estimate. We then performed one-sample ity using Pearson correlation coefficients (r) and variance Wilcoxon signed rank tests for each variable to determine inflation factors (all r < 0.3 and all VIF < 2) as well as for if the beta averages were significantly different from zero. biological meaningfulness. We used conditional logistic regression to estimate SSFs assuming an exponential se- Additional files lection function of the form: Additional file 1: Net Squared Displacement graphs for individuals included in SSF analyses. Figure S1. Net Squared Displacement wðÞ x ¼ exp β x þ β x þ β x þ … þ β x 1 2 3 p (NSD) calculated for n = 10 dispersers (D), all of which are male. Of the 1 2 3 p dispersers, E007, E010 and E095 undergo exploratory movements in which they return to, or close to, previous ranges. Beside each individual is the NSD graph for the extracted dispersal period. Figure S2. Net where β to β are coefficients estimated by conditional 1 p Squared Displacement (NSD) calculated for n = 10 residents (R), all of logistic regression, which are associated with a vector x, which are female. of environmental variables x to x respectively. The 1 p Additional file 2: Table S1. k-fold cross-validation results. Robustness of higher the value of wðÞ x the more likely that step will be models was evaluated by k-fold cross validation for case-control design, following the method of Fortin et al. 2009 [60]. The SSFs were built using chosen by an animal [57]. a random selection of 80% of the data strata and then used to predict The analysis was carried out using version 1.2 of the R the SSF scores for the remaining 20% of strata. The observed location package TwoStepCLogit [58]. To correctly analyse data was then ranked against the random locations, for each stratum, and ranks were tallied. Spearman rank correlations (r ) were carried out with in which there are multiple controls per case it is neces- S the bin’s ranking and associated frequency. The procedure was repeated sary to use conditional logistic regression. However, it is 100 times and the mean and range of r are shown. The mean and difficult to take into account cluster-level variation (i.e., expected range of r are also shown, calculated by ranking one random location against the other random locations, per strata, tallying ranks and random variation among elk) using this method, and this using Spearman rank correlations (r ). Again this was repeated 100 times. is important to address given that resource selection can vary considerably between individuals within a popula- Competing interests tion and large differences in individual behaviour have The authors declare that they have no competing interests. been observed amongst elk [45,59]. To account for vari- ation among individuals the TwoStepCLogit package was Authors’ contributions used to first fit fixed-effects regression models to each All co-authors participated in writing the manuscript. SC, DP, MM and MB collected and processed data and JK and HT conducted analyses. All authors individual elk and then combine those estimates using read and approved the final manuscript. restricted maximum likelihood (REML) estimation [58]. This method gives stable and consistent estimations of Acknowledgements the parameters in mixed conditional logistic regression We thank the Natural Sciences and Engineering Research Council of Canada models when the number of strata is large, as it was in (NSERC-CRD), Shell Canada Limited, Alberta Conservation Association (Grant Eligible Conservation Fund), Alberta Environment and Sustainable Resource this dataset. By using this procedure we account for the Development, Safari Club International, Alberta Parks, and Parks Canada for large inter-individual variation observed amongst the elk. funding and support. We also thank the Carl Tryggers foundation for The two-stage approach also helps to account for cor- scientific research (Swedish Post-doc grant) and the European Union (Erasmus Mundus Masters scholarship). The funders had no role in study relation within individuals, common in habitat-selection design, data collection and analysis, decision to publish, or preparation of studies (autocorrelation) [56]. Models were validated using the manuscript. We thank Daniel Fortin and an anonymous reviewer for their k-fold cross validation [60]. comments which greatly improved the manuscript. Killeen et al. Movement Ecology 2014, 2:15 Page 12 of 13 http://www.movementecologyjournal.com/content/2/1/15 Animal care 20. Ciuti S, Apollonio M: Do antlers honestly advertise the phenotypic quality Our data collection complied with all relevant federal laws of Canada and of fallow buck (Dama dama) in a lekking population? Ethology 2011, provincial laws of Alberta. Procedures adopted in this study were reviewed 117:133–144. and approved by the University of Alberta Animal Care and Use Committee 21. Dobson FS: Competition for mates and predominant juvenile male ACUC – Biosciences (Animal care protocol # 536-1003 AR University of dispersal in mammals. Anim Behav 1982, 30:1183–1192. Alberta, Edmonton, Canada), by all jurisdictions of the Alberta Government 22. Hebblewhite M, Merrill E, McDermid G: A multi-scale test of the forage (Permit Numbers: BI-2008-19, RC-06SW-001 and 23181CN), and by Parks maturation hypothesis in a partially migratory ungulate population. Canada (Permit Numbers: WL-2010-7292, WL-2010-5755). Ecol Monogr 2008, 78:141–166. 23. Boyce MS: The Jackson Elk Herd: Intensive Wildlife Management in North Author details America. Cambridge, U.K.: Cambridge University Press; 1989. Department of Biological Sciences, University of Alberta, Edmonton, AB T6G 24. Kalb DM, Bowman JL, Eyler TB: Dispersal and home-range dynamics of 2E9, Canada. Marine Evolution and Conservation Group, Centre of exotic, male sika deer in Maryland. Wildl Res 2013, 40:328–335. Evolutionary and Ecological Studies, University of Groningen, PO Box 11103 25. Petersburg ML, Alldredge AW, Vergie WJ: Emigration and survival of CC, Groningen, The Netherlands. Department of Biometry and 2-year-old male elk in northwestern Colorado. Wildl Soc Bull 2000, Environmental System Analysis, University of Freiburg, Freiburg 79106, 28:708–716. Germany. Faculty of Environmental Design, University of Calgary, Calgary, AB 26. Long ES, Diefenbach DR, Rosenberry CS, Wallingford BD: Multiple T2N 1 N4, Canada. proximate and ultimate causes of natal dispersal in white-tailed deer. Behav Ecol 2008, 19:1235–1242. Received: 4 March 2014 Accepted: 12 June 2014 27. Fryxell JM, Hazell M, Borger L, Dalziel BD, Haydon DT, Morales JM, Published: 1 July 2014 McIntosh T, Rosatte RC: Multiple movement modes by large herbivores at multiple spatiotemporal scales. Proc Natl Acad Sci U S A 2008, 105:19114–19119. 28. Borger L, Fryxell JM: Quantifying individual differences in dispersal using References net squared displacement. In Dispersal and Spatial Evolutionary Ecology. 1. Bohonak AJ: Dispersal, gene flow, and population structure. Q Rev Biol Edited by Clobert MB J, Benton T, Bullock J. Oxford, UK: Oxford University 1999, 74:21–45. Press; 2012. 2. Clobert J, Danchin E, Dhondt AA, Nichols D: Editors: Dispersal. Oxford: Oxford University Press; 2001:311–340. 29. Thurfjell H, Ciuti S, Boyce M: Applications of step-selection functions in 3. Dunning JB, Stewart DJ, Danielson BJ, Noon BR, Root TL, Lamberson RH, ecology and conservation. Movement Ecol 2014, 2:4. Stevens EE: Spatially explicit population models - current forms and 30. Johnson CJ, Parker KL, Heard DC, Gillingham MP: Movement parameters of future uses. Ecol Appl 1995, 5:3–11. ungulates and scale-specific responses to the environment. J Anim Ecol 4. Bowler DE, Benton TG: Causes and consequences of animal dispersal 2002, 71:225–235. strategies: relating individual behaviour to spatial dynamics. Biol Rev 31. Jerde CL, Visscher DR: GPS measurement error influences on movement 2005, 80:205–225. model parameterization. Ecol Appl 2005, 15:806–810. 32. Edge WD, Marcum CL, Olson SL, Lehmkuhl JF: Nonmigratory cow elk herd 5. Chetkiewicz CLB, Clair CCS, Boyce MS: Corridors for conservation: ranges as management units. J Wildl Manage 1986, 50:660–663. integrating pattern and process. Ann Rev Ecol Evol Syst 2006, 37:317–342. 6. Tesson S, Edelaar P: Dispersal in a changing world: opportunities, insights 33. Van Moorter B, Visscher D, Benhamou S, Börger L, Boyce MS, Gaillard J-M: and challenges. Movement Ecol 2013, 1:10. Memory keeps you at home: a mechanistic model for home range 7. Cagnacci F, Boitani L, Powell RA, Boyce MS: Animal ecology meets GPS- emergence. Oikos 2009, 118:641–652. based radiotelemetry: a perfect storm of opportunities and challenges. 34. Bruggeman JE, Garrott RA, White PJ, Watson FGR, Wallen R: Covariates Philos Trans R Soc B-Biol Sci 2010, 365:2157–2162. affecting spatial variability in bison travel behavior in Yellowstone 8. Taylor PD, Fahrig L, Henein K, Merriam G: Connectivity is a vital element of National Park. Ecol Appl 2007, 17:1411–1423. landscape structure. Oikos 1993, 68:571–573. 35. Parker KL, Robbins CT, Hanley TA: Energy expenditures for locomotion by 9. Zeller KA, McGarigal K, Whiteley AR: Estimating landscape resistance to mule deer and elk. J Wildl Manage 1984, 48:474–488. movement: a review. Landsc Ecol 2012, 27:777–797. 36. Frair JL, Merrill EH, Visscher DR, Fortin D, Beyer HL, Morales JM: Scales of movement by elk (Cervus elaphus) in response to heterogeneity 10. Adriaensen F, Chardon JP, De Blust G, Swinnen E, Villalba S, Gulinck H, in forage resources and predation risk. Landsc Ecol 2005, 20:273–287. Matthysen E: The application of 'least-cost' modelling as a functional landscape model. Landsc Urban Plan 2003, 64:233–247. 37. Pettorelli N, Vik JO, Mysterud A, Gaillard JM, Tucker CJ, Stenseth NC: Using 11. Beier P, Majka DR, Spencer WD: Forks in the road: choices in procedures the satellite-derived NDVI to assess ecological responses to environmental for designing wildland linkages. Conserv Biol 2008, 22:836–851. change. Trends Ecol Evol 2005, 20:503–510. 12. Pullinger MG, Johnson CJ: Maintaining or restoring connectivity of 38. Green RA, Bear GD: Seasonal cycles and daily activity patterns of Rocky modified landscapes: evaluating the least-cost path model with multiple mountain elk. J Wildl Manage 1990, 54:272–279. sources of ecological information. Landsc Ecol 2010, 25:1547–1560. 39. Skovlin JM, Zager P, Johnson BK: Elk habitat selection and evaluation. In 13. Squires JR, DeCesare NJ, Olson LE, Kolbe JA, Hebblewhite M, Parks SA: Elk of North America: Ecology and Management. Edited by Toweill DE, Combining resource selection and movement behavior to predict Thomas JW. Washington, DC: Smithsonian Institution Press; 2002:531–555. corridors for Canada lynx at their southern range periphery. Biol Conserv 40. Muhly TB, Semeniuk C, Massolo A, Hickman L, Musiani M: Human activity 2013, 157:187–195. helps prey win the predator-prey space race. Plos One 2011, 6:8. 14. Thatcher CA, van Manen FT, Clark JD: A habitat assessment for Florida 41. Ciuti S, Northrup JM, Muhly TB, Simi S, Musiani M, Pitt JA, Boyce MS: Effects panther population expansion into Central Florida. J Mammal 2009, of humans on behaviour of wildlife exceed those of natural predators in 90:918–925. a landscape of fear. Plos One 2012, 7:13. 15. Soulsbury CD, Iossa G, Baker PJ, White PCL, Harris S: Behavioral and spatial 42. Lewis JS, Rachlow JL, Garton EO, Vierling LA: Effects of habitat on GPS analysis of extraterritorial movements in red foxes (Vulpes vulpes). collar performance: using data screening to reduce location error. J Appl J Mammal 2011, 92:190–199. Ecol 2007, 44:663–671. 16. Newby J: Puma dispersal ecology in the Central Rocky Mountains. In 43. Bunnefeld N, Borger L, van Moorter B, Rolandsen CM, Dettki H, Solberg EJ, Master's thesis. Missoula, MT: University of Montana; 2011. Ericsson G: A model-driven approach to quantify migration patterns: 17. Singh NJ, Borger L, Dettki H, Bunnefeld N, Ericsson G: From migration to individual, regional and yearly differences. J Anim Ecol 2011, 80:466–476. nomadism: movement variability in a northern ungulate across its 44. Mysterud A, Loe LE, Zimmermann B, Bischof R, Veiberg V, Meisingset E: latitudinal range. Ecol Appl 2012, 22:2007–2020. Partial migration in expanding red deer populations at northern 18. Simberloff D, Farr JA, Cox J, Mehlman DW: Movement corridors - latitudes - a role for density dependence? Oikos 2011, 120:1817–1825. conservation bargains or poor investments. Conserv Biol 1992, 6:493–504. 45. Ciuti S, Muhly TB, Paton DG, McDevitt AD, Musiani M, Boyce MS: Human 19. Greenwood PJ: Mating systems, philopatry and dispersal in birds and selection of elk behavioural traits in a landscape of fear. Proc R Soc B-Biol mammals. Anim Behav 1980, 28:1140–1162. Sci 2012, 279:4407–4416. Killeen et al. Movement Ecology 2014, 2:15 Page 13 of 13 http://www.movementecologyjournal.com/content/2/1/15 46. Calenge C: The package "adehabitat" for the R software: A tool for the analysis of space and habitat use by animals. Ecol Model 2006, 197:516–519. 47. R Core Team: R: A Language and Environment for Statistical Computing. In R Foundation for Statistical Computing. 2013. 48. Paton DG: Connectivity of elk migration in southwestern Alberta. In Master's thesis. Calgary, Alberta: University of Calgary; 2012. 49. Bates D, Maechler M, Bolker B, Walker S: lme4: Linear mixed-effects models using Eigen and S4. R package version 1.0-5. 2013. http://cran.R-project. org/package=lme4. 50. Muhly TB, Alexander M, Boyce MS, Creasey R, Hebblewhite M, Paton D, Pitt JA, Musiani M: Differential risk effects of wolves on wild versus domestic prey have consequences for conservation. Oikos 2010, 119:1243–1254. 51. Riley SJ, DeGloria SD, Elliot R: A terrain ruggedness index that quantifies topographic heterogeneity. Intermountain J Sci 1999, 5:23–27. 52. Pettorelli N, Ryan S, Mueller T, Bunnefeld N, Jedrzejewska B, Lima M, Kausrud K: The Normalized Difference Vegetation Index (NDVI): unforeseen successes in animal ecology. Clim Res 2011, 46:15–27. 53. Turchin P: Quantitative Analysis of Movement: Measuring and Modeling Population Redistribution in Plants and Animals. Sunderland, MA: Sinauer Associates; 1998. 54. Fortin D, Beyer HL, Boyce MS, Smith DW, Duchesne T, Mao JS: Wolves influence elk movements: behavior shapes a trophic cascade in Yellowstone National Park. Ecology 2005, 86:1320–1330. 55. Muggeo VMR: segmented: an R package to fit regression models with broken-line relationships. R News 2008, 8:20–25. 56. Fieberg J, Matthiopoulos J, Hebblewhite M, Boyce MS, Frair JL: Correlation and studies of habitat selection: problem, red herring or opportunity? Philos Trans R Soc B-Biol Sci 2010, 365:2233–2244. 57. Lele SR, Merrill EH, Keim J, Boyce MS: Selection, use, choice and occupancy: clarifying concepts in resource selection studies. J Anim Ecol 2013, 82:1183–1191. 58. Craiu RV, Duchesne T, Fortin D, Baillargeon S: Conditional logistic regression with longitudinal follow-up and individual-level random coefficients: A stable and efficient two-step estimation method. J Comput Graph Stat 2011, 20:767–784. 59. Bolnick DI, Svanback R, Fordyce JA, Yang LH, Davis JM, Hulsey CD, Forister ML: The ecology of individuals: incidence and implications of individual specialization. Am Nat 2003, 161:1–28. 60. Fortin D, Fortin ME, Beyer HL, Duchesne T, Courant S, Dancose K: Group-size-mediated habitat selection and group fusion-fission dynamics of bison under predation risk. Ecology 2009, 90:2480–2490. doi:10.1186/s40462-014-0015-4 Cite this article as: Killeen et al.: Habitat selection during ungulate dispersal and exploratory movement at broad and fine scale with implications for conservation management. Movement Ecology 2014 2:15. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit

Journal

Movement Ecology – Springer Journals

Published: Jul 26, 2014

There are no references for this article.