Introduction

Public lands are important for supporting a variety of multiple uses, two of which include cattle grazing and timber production. Balancing these uses is fundamental to maintaining the ecological integrity and economic sustainability of forested rangelands. Lack of cooperation between forest and cattle industries as they pursue independent goals on a common land base can create conflict, thereby complicating the process of integrated resource management on public land (Mitchell et al. 1982).

Grazing by cattle of coniferous cut blocks has been associated with poor tree regeneration, growth and survival (Zimmerman and Neuenschwander 1984). Tree seedlings are most likely to be detrimentally affected by grazing when livestock stocking rates are high (Newman et al. 1994; Pitt et al. 1998). Although cattle normally do not browse conifer trees due to their low palatability (Lewis 1980a), damage to conifer seedlings can still occur from trampling (McLean and Clark 1980; Newman et al. 1994), leading to increased seedling mortality (Eissenstat et al. 1982). Physical damage to the tree’s cambium layer not only hinders tree physiological processes and associated growth (Lewis 1980b), but increases tree susceptibility to environmental stress, including attack by pathogens and insects (Frey et al. 2003).

The primary factors influencing habitat selection by cattle are forage quantity and quality, as animals generally strive to optimize nutrient intake at minimal energetic cost (Senft et al. 1987). Roath and Krueger (1982) found that cattle distribution and forage use across forested ranges was related to vegetation type and feed quality. Factors affecting forage quantity and quality include plant community composition, vegetation morphology, the presence of secondary compounds and corresponding palatability, as well as growing conditions and associated phenological stage of growth (Senft et al. 1987; Senft 1989; Chapman et al. 2007). Locations that provide the most favourable forage are likely to experience greater cattle use compared to areas with less desirable vegetation.

Cattle distribution and associated use of vegetation is further affected by accessibility to preferred habitats, temperature, shelter from the elements, and protection from biting insects (Owens et al. 1991; Beaver and Olson 1997; Launchbaugh and Howery 2005). Distance to free standing water is also known to influence cattle locations and grazing preference, with cattle use typically peaking near sources of drinking water (Roath and Krueger 1982; Kie and Boroski 1996). Additionally, livestock avoid steep slopes within mountainous forested landscapes, as well as areas with sparse herbaceous vegetation (Roath and Krueger 1982; Kie and Boroski 1996).

Following forest harvest, large quantities of slash (i.e., wood residue) are often left on and above the soil surface. Slash can act as physical and visual barriers to cattle, thereby reducing opportunities for grazing. In some cases, slash has been deliberately used to reduce cattle impact on regenerating trees (Kota and Bartos 2010). Accumulated logging slash, however, presents a challenge for livestock by physically preventing access to forage, or risking injury when attempting to walk over or around it (Jones 1983; Morgan 1991). Over time animals can associate areas of abundant slash with high energy costs and low forage availability, leading to avoidance of these areas (Parker et al. 1984; Forester et al. 2007).

Livestock producers can alter cattle foraging behavior, including the extent of use on young trees, by managing the timing, intensity and duration of grazing (McLean and Clark 1980; Eissenstat et al. 1982; Newman et al. 1994; Pitt et al. 1998). Range use by cattle is also influenced by season of grazing as a result of changes in forage quantity and quality. Forage depletion resulting from high stocking rates can motivate cattle to access areas that are relatively unproductive, nutrient poor or less accessible (Senft et al. 1983; Hart et al. 1991). In this situation stocking rates need to be reduced and livestock distribution adjusted to limit the effects of grazing on tree regeneration (Newman et al. 1994; Pitt et al. 1998; Krzic et al. 2003, 2004). In general, damage to conifer seedlings can be minimized when stocking rates are matched to available forage resources (McLean and Clark 1980; Cutter et al. 1999).

In most jurisdictions, including Alberta, Canada, timber companies are responsible for the reforestation of cut blocks, the success of which is evaluated through regeneration surveys normally completed 4–8 years after harvest (Alberta Environment and Sustainable Resource Development (AESRD) 2012). Inadequate tree stocking necessitates further reforestation efforts. Government mandated regeneration assessments require that at least 80 % of plots examined during these surveys contain an undamaged conifer seedling of adequate size (Alberta Sustainable Resource Development (ASRD) 2008). Any factor causing substandard regeneration will increase reforestation costs, reduce profit, and jeopardize future timber harvesting opportunities. To mitigate cattle-induced damage to regenerating trees and minimize conflict between the forest and ranching industries, a greater understanding is required of the extent to which cattle utilize cut blocks following conifer forest harvest under typical open range conditions in this region.

A previous investigation reported on the comparative selection patterns of cattle among diverse habitats under open range conditions within a Montane landscape in SW Alberta (Kaufmann et al. 2013). While that study revealed cattle had greater relative preferences for open grassland habitats over intact conifer forests and recent clear cut areas, it did not quantify detailed levels or patterns of cattle use within cut blocks. The objectives of the current study were to; (1) quantify cattle presence within a representative sample of cut blocks in several large pastures exposed to open-range grazing; (2) explore and describe environmental factors associated with observed patterns of cattle presence within cut blocks, including topography, distance to water, forage characteristics and slash properties; and (3) document cattle impacts on conifer seedlings.

Methodology

Study area

This investigation was conducted within two public grazing allotments (Sharples Creek and Skyline) situated in the Porcupine Hills (49°53′47″N; 114°01′41″W), approximately 30 km north of Pincher Creek in southwest Alberta, from June 2008 through October 2009. This area is part of the Rocky Mountain Forest Reserve, and is located within the Montane Natural Subregion. The two allotments include three large pastures (Damon-836 ha, North Sharples-364 ha, and Prong Coulee-1035 ha) administered by AESRD.

The Porcupine Hills are characterized by strongly rolling topography containing large-scale relief, with slopes from 6 to 30 %, and elevations between 1409 and 1689 m ASL. Different plant communities, including conifer, mixedwood and aspen forests among interspersed grasslands, are present across the landscape as a result of marked variation in topographic relief, aspect and moisture availability. Dense lodgepole pine (Pinus contorta Loudon) stands commonly occupy uplands, particularly on north-facing slopes, with succession to Douglas-fir [Pseudotsuga menziesii (Mirb.) Franco] and white spruce [Picea glauca (Moench)] (Willoughby et al. 2005). Within harvested cut blocks, common grasses include pinegrass (Calamagrostis rubescens Buckley), hairy wild rye [Leymus innovatus (Beal) Pilg.], and bluegrasses (Poa spp.). Native forbs include heart leaved arnica (Arnica cordifolia Hook.), showy aster [Eurybia conspicua (Lindl.) Nesom], and bunchberry (Cornus canadensis L.), whereas Canada thistle [Cirsium arvense (L.) Scop.] is invasive in more heavily disturbed sites. Widespread shrub species in cut blocks include wild rose (Rosa woodsii Lindl.), raspberry (Rubus idaeus L.), thimbleberry (Rubus parviflorus Nutt.), and white meadowsweet (Spiraea betulifolia Pallas). Soils associated with conifer forests are largely Gray Luvisols and Eutric Brunisols (Willoughby et al. 2005). The primary source of open water is flowing creeks within valley bottoms.

This area of the Montane has a bimodal precipitation pattern, with peaks in May–June and August–September (Willoughby et al. 2005). The nearby Connelly Creek weather station received 379 mm in 2008, only 65 % of normal (Environment Canada, unpublished data, 1971–2000). Precipitation was 521.7 mm in 2009, 90 % of normal. Average annual temperatures during 2008 and 2009 were 5.1 and 4.1 °C, respectively, close to the long-term norm of 4.8 °C. Summer maximum temperatures can exceed 35 °C.

Conifer cut blocks represented an average of 5 % of the total area of pastures, with a total of five cut blocks, ranging from 2.1 to 18.6 ha in size, assessed in this investigation. Timber harvest was conducted by Spray Lakes Sawmills between December 2005 and March 2006. Trees were felled, stump side processed (delimbed and topped), and skidded to the side of temporary unimproved in-block roads for loading and removal. After harvest, the temporary in-block roads were ripped to de-compact the soil, and where necessary, re-contoured. Coarse woody debris was left in place throughout the cut blocks. Where debris was considered excessive a grapple skidder was used to redistribute slash, including onto reclaimed in-block roads. Shortly after logging, cut blocks were directly planted to achieve approximately a 2.1 m spacing resulting in a planting density of about 2200 trees ha−1. Lodgepole pine were the predominant seedlings planted (75 %), although white spruce seedlings were planted into more mesic sites. Planting occurred in 2007, 1–2 years after harvest. Seedlings were 1–2 years of age with a minimum height of 8 cm and a minimum root crown diameter of 2.3 mm.

A single herd of 632 animals, comprised of 311 black angus cross cow-calf pairs and 10 bulls, grazed the Damon, North Sharples, and Prong Coulee pastures sequentially during July, August and September of both 2008 and 2009, although the sequence of grazing varied from year to year (Table 1). Resultant stocking rates ranged from 0.35 to 0.59 AUM ha−1 (Table 1) and were consistent with those used on other publicly grazed Montane rangelands in the region. While a variety of native ungulates are found in the area, including moose, elk, mule deer and whitetail deer, their densities are relatively low, and therefore contribute little to overall herbivory.

Table 1 Grazing periods and stocking rates within each of the three distribution units in the Sharples Creek and Skyline grazing allotments
Full size table

Data collection

Field sampling occurred immediately following each grazing period. In each pasture, one large cut block (Prong Coulee unit) or two smaller cut blocks (North Sharples and Damon units) were examined to assess voluntary use by cattle of these areas under open range (i.e., free-choice) conditions, cattle damage to tree seedlings, levels of slash loading, and environmental characteristics.

A systematic sampling grid was established across each cut block consisting of plots situated 50 or 60 m apart: a narrower grid spacing was used in the four smaller cut blocks to maintain adequate plot numbers for sampling these areas. Final plot numbers ranged from 6 to 62 among cut blocks. Plot locations were recorded with a geographic information system (GIS) to facilitate sampling in consecutive years. Regeneration assessments followed the methodology described in the Alberta Regeneration Survey Manual (ASRD 2008). Each individual plot was circular and 10 m2 in area, and relocated in 2009 using a GPS. Within each pasture, two range cages, 1.5 × 1.5 m in size, were paired with two randomly selected sample plots for the assessment of non-grazed biomass. Cages were moved to different plots 1 year later. To gather information on the factors responsible for cattle selection across cut blocks and the effect of grazing on tree regeneration, the following data were collected in each sample plot:

  1. 1.

    Evidence of cattle visitation in the form of recent cattle tracks, grazing, or dung-pats (presence/absence per plot);

  2. 2.

    Biomass clippings within a 0.5 m2 quadrat from within those plots where range cages were present, with samples taken inside and outside cages, and separated into grass, forb and shrub (i.e., current annual growth of shrub leaves and twigs) components;

  3. 3.

    Ocular estimates of standing biomass by growth form (kg ha−1) in all plots, which were subsequently validated by correlating estimated biomass against actual biomass (e.g., r = 0.55 and 0.88 in 2008 and 2009, respectively) for the subset of plots within which biomass was harvested (see Kaufmann et al. 2013; Girard et al. 2013);

  4. 4.

    Ocular assessment of forage utilization (%). Where range cages were present, actual utilization was calculated by comparing dry matter biomass (kg ha−1) inside and outside cages using Eq. (1);

    $$ {\text{Forage utilization }}\left( \% \right) \, = \, {{\left( {{\text{biomass inside cage }}{-}{\text{ biomass outside cage}}} \right)} \mathord{\left/ {\vphantom {{\left( {{\text{biomass inside cage }}{-}{\text{ biomass outside cage}}} \right)} {{\text{inside cage biomass }} \times \, 100}}} \right. \kern-0pt} {{\text{inside cage biomass }} \times \, 100}} \, $$
    (1)
  5. 5.

    Topographic attributes, including slope gradient (%), aspect (°) and elevation (m);

  6. 6.

    Quantification of slash, including ocular assessment of the proportion (%) of ground covered by slash, maximum slash height (cm), and distance to the nearest grazed patch from the point of greatest slash height within the plot;

  7. 7.

    Number, height, and basal diameter of seedlings for each of three conifer tree species present;

  8. 8.

    Damage to trees including browsing, leader breakage, basal scarring, and vertical displacement (≥15 degrees from vertical), as defined by Irving (2001); and

  9. 9.

    Distance to the nearest water source from each plot, as calculated using ArcGIS 9.2 software.

Plant biomass samples were dried at 45 °C for 48 h to constant mass, and weighed. Standing biomass values were combined with utilization estimates to derive forage biomass available prior to grazing (see Eq. 2).

$$ {\text{Forage biomass }}\left( {{\text{kg ha}}^{ - 1} } \right) \, = \, {{\text{post grazing biomass estimate}} \mathord{\left/ {\vphantom {{\text{post grazing biomass estimate}} \quad {\left( {1 - {\text{estimate of proportion forage removed}}} \right)}}} \right. \kern-0pt} {\left( {1 - {\text{estimate of proportion forage removed}}} \right)}} \, $$
(2)

Samples of grass, forb and shrub were ground to 1 mm using a Wiley Mill, and assessed for nitrogen (N) concentration using the Dumas method (Dumas 1831), with samples combusted using a LECO TruSpec FP-428 auto-analyzer (St. Joseph, MI, USA). Nitrogen values were subsequently converted to crude protein (CP) concentrations by multiplying N values by 6.25, based on standardized conversions. Crude protein concentration for all forage components at the plot level was obtained using the mean contribution of CP from each of the three growth forms, weighted by their estimated biomass contribution (see Eq. 3).

$$ {\text{Total plot CP concentration }}\left( \% \right) \, = \, {{\left( {{\text{biomass}}_{\text{grass}} \times {\text{ CP}}_{\text{grass}} + {\text{ biomass}}_{\text{forb}} \times {\text{ CP}}_{\text{forb}} + {\text{ biomass}}_{\text{shrub}} \times {\text{ CP}}_{\text{shrub}} } \right)} \mathord{\left/ {\vphantom {{\left( {{\text{biomass}}_{\text{grass}} \times {\text{ CP}}_{\text{grass}} + {\text{ biomass}}_{\text{forb}} \times {\text{ CP}}_{\text{forb}} + {\text{ biomass}}_{\text{shrub}} \times {\text{ CP}}_{\text{shrub}} } \right)} {\left( {{\text{biomass}}_{\text{grass}} + {\text{ biomass}}_{\text{forb}} + {\text{ biomass}}_{\text{shrub}} } \right)}}} \right. \kern-0pt} {\left( {{\text{biomass}}_{\text{grass}} + {\text{ biomass}}_{\text{forb}} + {\text{ biomass}}_{\text{shrub}} } \right)}} \, $$
(3)

Analysis

An information-theoretic approach (Burnham and Anderson 1998; Anderson 2008) was used to evaluate models describing the probability of cattle presence within cut block plots. Cattle presence was characterized using binary values of 1 (presence) and 0 (absence). Cattle presence was then related to four main independent variables (and associated sub-variables) as follows: (1) forage [biomass (kg ha1) and crude protein concentration (%)], (2) topography [elevation (m), slope (%) and aspect (degrees) cosine transformed (Platts et al. 2008)], (3) logging slash abundance [maximum height (cm) and cover (%)], and (4) distance to the nearest water source (m) (Table 2). Correlations were conducted to identify collinearity and eliminate redundant independent variables (|r| > 0.6).

Table 2 Units, means, 5th and 95th percentile values for independent variables analyzed for their effect on habitat occupancy
Full size table

A multi-step process was then used to identify the best models. At each step, general linear mixed model analysis (PROC GLIMMIX with LAPLACE estimation; SAS Institute Inc. 2009) was used to model the logarithm of the likelihood of cattle presence (i.e., ln [p/(1−p)], where p is the probability of cattle presence). Block, year and their interaction were considered random factors in the analysis. Two criteria were used to evaluate model appropriateness. The first criterion was ω i , defined as “the weight of evidence in favor of model i being the actual best model for the situation at hand, given that one of the models must be the best model in that set” (Burnham and Anderson 1998, p. 75). The second was McFadden’s pseudo R2 (Windmeijer 1995), calculated as:

$$ 1 - ({\text{log likelihood candidate model}}/{\text{log likelihood null model}}) $$
(4)

where the null model did not include any of the environmental variables.

In the first step, the relative strength of each of the four main variables was evaluated. This process revealed that topography was the most important independent variable whereas the other three variables had little explanatory power.

In step two, nine candidate models were compared to determine if water, forage or slash provided any additional explanatory power when compared to topography alone. These models were:

  1. 1.

    Null model with no environmental variables

  2. 2.

    Topography (= elevation + slope + aspect)

  3. 3.

    Topography + forage (= biomass + protein)

  4. 4.

    Topography + distance to water

  5. 5.

    Topography + slash accumulation (= slash maximum height + ground covered by coarse woody debris)

  6. 6.

    Topography + distance to water + forage

  7. 7.

    Topography + distance to water + slash accumulation

  8. 8.

    Topography + slash accumulation + forage

  9. 9.

    Topography + distance to water + slash accumulation + forage

In step three, once the best primary model was identified, sub-models were developed to assess the strength and relative contribution of each sub-variable within the topography and forage categories in accounting for additional cattle use. Topography sub-models included elevation + distance to water + forage, slope + distance to water + forage, and aspect + distance to water + forage. Forage sub-models included topography + distance to water + forage biomass, and topography + distance to water + forage CP concentration. Topography sub-models were only compared against one another, as were the forage sub-models. In step 4, the relative explanatory power of the best six models were compared using values of ω i .

Finally, to confirm the absence of a slash impact on cattle use, two analyses were performed, separately by pasture, to determine the effects of slash. In the first, PROC GLIMMIX (SAS Institute Inc. 2009) was used to analyze the relationship between cattle presence and percent slash cover. To help visualize these relationships, six bins with approximately equal numbers of observations were created, average percent cover and average presence were calculated, and plots constructed. In the second, PROC REG (SAS Institute Inc. 2009) was used to calculate the relationship between the distance from each plot to the nearest grazed patch and maximum slash height per plot; residuals were examined for normality and equality of variances. For both analyses one-tailed tests were used to test the hypothesis that cattle use would be deterred by slash.

Results

Factors affecting cattle presence in cut blocks

Of the 154 plots examined across all clear cuts, the average number of plots with evidence of cattle presence in 2008 and 2009 was 50 and 43 %, respectively, for an overall annual average of 46 %. A total of 65 % of plots received some use over the 2 year period. Among those plots with cattle presence, the most common evidence of cattle was grazing of herbage (40 %), followed by tracks (33 %) and the presence of dung-pats (6 %), with up to 36 % of plots having multiple signs of cattle presence. Comparison of the four main parameters (topography, forage, water distance, and slash accumulation) hypothesized to influence cattle presence within cut blocks indicated topography played the primary role, with a ω i value of 1.000 as opposed to values of 0.000 for the other three variables (Table 3).

Table 3 Summary of results in the four step process comparing AIC models predicting the presence of cattle within cut block plots. All analyses are based on 306 plot-years of data
Full size table

Subsequent comparison of topography sub-models indicated elevation was by far the most important landscape factor affecting cattle presence across cut blocks with a ω i value of 1.000 as opposed to values of 0.000 for slope and aspect (Table 3). Plots of predicted cattle presence versus elevation illustrate the marked role of elevation in influencing cattle presence (Fig. 1). Closer inspection of animal locations indicated that when cattle did enter cut blocks, they tended to concentrate at low elevations within these areas.

Fig. 1
figure 1

Modelled probability of cattle presence as a function of elevation and distance to water. To illustrate the influence of elevation and distance to water on cattle utilization, the top ranked model was used to predict cattle presence with slope, aspect and crude protein held constant at their mean values. Elevation values range from the 5th to the 95th percentiles (1480 and 1620 m, respectively). Distance to water values represent the 5th and 95th percentiles, as well as the mean distance from water (37, 364 and 181 m, respectively)

Full size image

Comparison of forage sub-models indicated CP concentration had a greater influence than biomass in regulating cattle presence across cut blocks, as shown by ω i values of 0.68 and 0.32, respectively (Table 3). As expected, biomass was associated with a positive regression coefficient in all models. In contrast, CP was associated with a negative coefficient in all models, indicating cattle presence was lower in areas with greater CP concentration. Further examination of the unexpected negative association between CP content and cattle presence revealed that forbs sampled in cut blocks were greater (p < 0.001) in CP concentration than grasses (10 vs. 6 %), and plots with evidence of cattle had, on average, 162 kg ha−1 more grass than forb biomass (598 vs. 436 kg ha−1). In contrast, plots without cattle had similar grass and forb biomass.

All of the best models explaining cattle presence contained distance to water as an explanatory variable (Table 3). Using the top-ranked model and a baseline elevation of 1540 m, predicted cattle presence decreased from 65 % at the 5th percentile of distance from water (37 m) to 22 % at the 95th percentile of distance from water (364 m) (Fig. 1). Information-theoretic analysis suggested the accumulation of logging slash had comparatively little influence on cattle presence across cut blocks, as evidenced by the drop in ω i from 26 to 7 % when the slash variable was added to the model containing topography and distance from water (Table 3). Maximum slash height of plots across cut blocks averaged 0.85 m. Separate analysis of slash influences by pasture however, revealed a significant positive relationship between distance to the nearest grazed patch and maximum slash height for the Damon pasture only (Fig. 2) and no relationship between cattle presence and percent slash cover for any of the three pastures (Fig. 3).

Fig. 2
figure 2

Relationship between distance to the nearest grazed patch and maximum slash height, evaluated separately by pasture. p-values for the Damon, Prong Coulee and North Sharples pastures were 0.03, 0.23 and 0.06, respectively, and R 2 values were 0.33, 0.02, and 0.19, respectively

Full size image
Fig. 3
figure 3

Relationship between cattle presence and percent slash cover, evaluated separately by pasture. To aid visualization, average cattle presence and slash cover were determined for six bins per pasture, and linear regression performed using generalized linear models. For the Damon, Prong Coulee and North Sharples pastures, R 2 values were 0.17, 0.21 and 0.32, respectively. p-values for the Damon, Prong Coulee and North Sharples pastures were 0.32, 0.22 and 0.06, respectively

Full size image

Damage to regenerating trees

Damage to all coniferous seedlings inspected (N = 278 total trees examined across all plots in 2009) due to cattle activity was very low, being limited to the vertical displacement of two naturally regenerating Douglas-fir seedlings within the Damon pasture grazed during August of 2009. Total coniferous regeneration densities also increased in each pasture from 2008 through 2009 (Table 4), with these increases reflective of lodgepole pine and Douglas-fir seedling responses. Douglas-fir seedlings were the result of natural regeneration. In contrast, white spruce decreased in density during this period. Dead and browned spruce seedlings were frequently observed throughout cut blocks despite no apparent damage.

Table 4 Mean regeneration densities, heights and basal diameters (±SD) of coniferous tree seedlings observed within cut blocks of each of the three distribution units (Damon, North Sharples and Prong Coulee), for the years 2008 and 2009
Full size table

Tree seedling height and diameter increased over the study period concurrently with exposure to cattle grazing (Table 4). Average measurements of seedlings indicated these trees were likely to meet the minimum 30 cm height requirement for conifer regeneration standards in the Montane region of Alberta during subsequent re-establishment surveys (AESRD 2012). Additionally, a minimum 80 % stocking is required within regenerating cut blocks at the time of survey. At the end of our study in 2009, overall tree stocking rates were 77, 69 and 68 % in cut blocks assessed within the Prong Coulee, North Sharples and Damon pastures, respectively.

Discussion

Cattle selection and associated explanatory factors

While cattle were previously found to avoid cut blocks at the larger landscape scale, instead preferring adjacent open grasslands in valley bottoms near water (Kaufmann et al. 2013), we nevertheless found evidence of cattle presence in about two-thirds of all plots examined in this study. Cattle did not exhibit widespread presence across all cut blocks, however, a result that can be attributed to a combination of low stocking rates of cattle within pastures, combined with the fact that cut blocks represented only one habitat among a complex mix of habitats within the greater landscape. These findings are important in that they indicate that cattle still occupy cut block areas despite the latter not being a preferred habitat type under typical open-range grazing.

Habitat use by cattle involves complex processes including experiential learning, genetic predispositions, interaction with surrounding environmental characteristics, and the social dynamics of a herd (Launchbaugh and Howery 2005). Herbivores employ a range of physical senses (e.g., sight, smell and taste) to judge the desirability of habitats, and have the ability to associate environmental features with positive or negative consequences (Bailey et al. 1996; Launchbaugh and Howery 2005). Moreover, the relative importance of factors regulating cattle use of cut blocks will vary based on spatial variation in soils, elevation, topography, precipitation, temperature, logging method, post-harvest site treatment, distance to water, accessibility and herd management (McLean and Clark 1980; Eissenstat et al. 1982; Allen and Bartolome 1989; Pitt et al. 1998).

The relatively low R 2 values (max of 21 %) observed in this study between cattle presence across cut blocks and habitat characteristics, in spite of the large number of plots sampled and independent variables tested, could arise from several sources. These include the presence of non-linear relationships between cattle use and habitat attributes, complex interactions among independent variables not measured here (and reviewed above), the difficulty of describing a dynamic process based on measurements taken at a single point in time, or perhaps most likely, the possibility that when travelling through cut blocks cattle were simply using these areas at random in response to the measured environmental variables.

Nevertheless, when cattle did travel into cut blocks, habitat attributes of individual plots still accounted for a significant proportion of cattle presence across these areas (21 %). The best fitting models had topography (and in particular, elevation), distance to water, and to a lesser extent, forage characteristics, as the most important explanatory variables. Increased cattle presence in low elevation areas of cut blocks is attributed to closer proximity to, and improved access from, adjacent preferred habitats within each pasture, specifically low-lying grasslands near streams (Kaufmann et al. 2013). Despite spending much of their time in lowland grasslands near water, cattle may have travelled upslope from these areas into nearby cut blocks to investigate this novel habitat, which in turn, would concentrate their presence in the lowest elevation portions of cut blocks. This observation is supported by Kie and Boroski (1996), who hypothesized that cattle use of logged areas in California was related to their proximity to riparian areas. Vertical distance above water is a key factor affecting cattle use in several other investigations (Holechek et al. 1982; Ganskopp and Bohnert 2009). In the current study, livestock attractants (such as salt) were available but never placed within cut blocks to be consistent with management practices intended to dissuade cattle use of these areas.

Although cattle are known to concentrate use on slopes of 10 % or less (Pinchak et al. 1991; Bailey et al. 1996), no evidence was found in this study that slope gradient altered cattle use within cut blocks. This finding contrasts Gillen et al. (1984) who concluded slope was the only physical factor consistently associated with cattle distribution in Oregon. Similarly, the limited role of aspect in altering cattle presence within the present study may be explained by the restricted range in this variable within and between cut blocks, as cut blocks (and the occurrence of conifer forests in general) were typically constrained to north and north-west facing slopes (Table 2). Small increases in observed cattle presence on north-west aspects could arise as these areas receive greater solar radiation, and as a result, may offer more productive forage, particularly if prior to logging these sites had low density forests that are commonly found under warmer, drier conditions of the Montane (Willoughby et al. 2005).

Forage characteristics were also associated with cattle presence within cut blocks, although not entirely as expected. Cattle preferred areas with greater biomass, consistent with the notion that cattle optimize intake by selecting more productive habitats (Senft et al. 1987). However, contrary to our assumption that cattle would actively choose sites with greater CP, this did not occur, and may reflect the trade-off observed between grass and forb biomass within plots. As greater CP levels were associated with forb rather than grass biomass, but grasses provided the majority of herbage biomass (58 % of total), cattle selection for biomass may be antagonistic to selection for CP. This is particularly likely given that cattle diets consist mostly of graminoids (Holechek et al. 1982), which in turn, would account for the selection of plots with greater grass biomass. A final explanation for the avoidance of plots high in forb biomass and CP is the potential for these plots to be colonized by disturbance-adapted weed species unpalatable to cattle. For example, a number of plots had significant Canada thistle, which although high in protein content, is known to deter herbivory by cattle, particularly under low stocking densities (De Bruijn and Bork 2006). Overall, forage characteristics alone offered limited explanatory ability for cattle presence, suggesting hunger and satiation may not have been major factors motivating cattle to select different areas of cut blocks.

Contrary to predictions that cattle would avoid sites with expansive and vertically elevated logging debris, slash accumulation did not exhibit a strong association with cattle presence across cut blocks, at least relative to the other environmental factors investigated. As such, it appears that altering levels of slash would not result in marked changes to cattle presence at the spatial scale examined in this study. Nevertheless, our empirical observations of reduced cattle presence with increasing slash height in at least one pasture is consistent with previous studies demonstrating woody debris can obstruct herbivore movement or prevent entry into potential foraging sites (Ripple and Larsen 2001; Casabon and Pothier 2007; de Chantal and Granström 2007), and warrants further investigation. Wildlife studies have shown that the energy cost of travel through woody debris increases in relation to obstacle density and depth. Slash depths over 0.5 m (Parker et al. 1984) and 0.8 m (Ripple and Larsen 2001) have been associated with reduced wild ungulate use, with depth thresholds for cattle probably much less than this given their reduced mobility. Mean maximum slash height (0.85 m) among the plots sampled here may therefore explain the low overall presence of cattle within cut blocks across our study area relative to other vegetation types (Kaufmann et al. 2013). As slash decays over time any detrimental effects on cattle movement should diminish (Casabon and Pothier 2007).

Forest regeneration

Grazing by cattle in this montane landscape was compatible with forest management in this area given observations of only two instances of cattle-induced damage to conifer seedlings, in addition to ongoing increases in tree regeneration and growth (height and diameter). Conifer seedling densities in 2009 were above those of Zimmerman and Neuenschwander (1984), who reported 1330 stems ha−1 of Douglas-fir and ponderosa pine in Idaho, but below those of McLean and Clark (1980) in British Columbia (1700–6100 stems ha−1). The favorable increase in naturally regenerating Douglas-fir is consistent with studies suggesting this species is more tolerant of cattle presence than some pine species (Zimmerman and Neuenschwander 1984). Although the cut blocks examined here remained below provincial stocking guidelines in 2009 (68–77, vs. 80 % requirement), no evidence was found that cattle grazing was responsible for this shortfall. Furthermore, continued natural regeneration in conjunction with the planted stock may lead to tree stocking levels sufficient to meet regeneration requirements in the near future.

Damage and mortality to tree seedlings can be expected to increase as livestock stocking rates become greater (Lewis 1980a; Newman et al. 1994; Pitt et al. 1998). The limited occurrence of damage in this study suggests overall cattle stocking rates are aligned with available forage resources at the landscape-scale, leading to low associated use of cut blocks by cattle. Substantial seedling death from non-cattle related causes has previously been attributed to weather conditions during planting, poor quality planting stock, and even rodent damage (McLean and Clark 1980; Allen and Bartolome 1989; Newman et al. 1994). The poor condition of many spruce seedlings in our study cut blocks, despite limited livestock presence or damage, could be related to drought conditions during 2008 when rainfall was 35 % below normal.

Cases of successful integration of timber and cattle industries have been documented in British Columbia (McLean and Clark 1980; Newman et al. 1994), California (Allen and Bartolome 1989), Louisiana (Cutter et al. 1999), Idaho (Eissenstat et al. 1982; Zimmerman and Neuenschwander 1984), and now Alberta. These studies consistently attribute negligible seedling damage and sufficient tree performance to well managed grazing practices that take into account an appropriate timing, duration, and intensity of forage use. Our results indicate livestock producers, timber companies, and public land managers should work together to lay out future cut blocks in these Montane landscapes, implement appropriate grazing systems, and monitor cattle activity and impact to ensure damage to regenerating seedlings remains within acceptable tolerances (Mitchell et al. 1982; Allen and Bartolome 1989).

Based on the present study, cattle presence and associated foraging in cut blocks may be less likely if timber harvests are planned for higher elevations within the local landscape. This process may reduce the likelihood that cattle leave preferred habitats and move upslope into adjacent cut blocks. Similarly, keeping cut blocks further from available water, typically streams within valley bottoms, may reduce cattle presence in cut blocks. Additionally, the continued use of conservative cattle stocking rates on adjacent grasslands, where forage removal was only 43 % (Kaufmann et al. 2013), is likely to limit cattle presence of nearby cut blocks, in turn minimizing undesirable impact on tree seedlings (Newman et al. 1994). Further studies may be warranted examining how increases in cattle stocking rates (i.e., moderate to high levels) or changes in the relative availability of cut block habitat within pastures, alters patterns and impacts of cattle use. Increased forage use within cut blocks may be possible once seedlings are established and resistant to damage, but before growing trees become large enough to provide a physical barrier to cattle (Lewis 1980c; Eissenstat et al. 1982; Mitchell et al. 1982; Newman et al. 1994). Overall, the results of this study support the notion that cattle grazing, coniferous forest harvest, and subsequent tree regeneration lead to little detrimental impact on one another, and are therefore compatible uses on this publicly managed land base.