Interactions of management and white pine blister rust on Pinus strobiformis regeneration abundance in southwestern USA

Interactions of management and white pine blister rust on Pinus strobiformis regeneration... Abstract Modelling natural regeneration is complex, and both natural and anthropogenic disturbances can alter forest trajectories. Pinus strobiformis (southwestern white pine, SWWP) is an important component of mixed conifer forests in the Southwest and management recommendations related to natural and planted regenerations are needed to guide conservation of SWWP in the face of an invasive disease (white pine blister rust, WPBR). Regeneration was surveyed across six mountain ranges, three silviculture treatments and two levels of disease severity in the Southwest US. Key findings were: (1) SWWP regeneration in stands with no recent management (<20 years) and high disease severity had unsustainable WPBR infection, (2) SWWP regeneration was less abundant but less likely to be infected in stands with recent management, (3) stands with high disease severity had fewer SWWP seedlings than stands with no or low disease severity and (4) SWWP regeneration densities were best predicted by other understory species abundance. We recommend silviculture treatments that reduce basal area to 9–10 m2 ha−1 and leave large canopy openings to enhance natural SWWP regeneration. Without creating conditions for disease-free regeneration to reach reproductive maturity, some stands may lose SWWP as an overstory component. Results may help refine SWWP management guidelines and expand conservation efforts in forests threatened by WPBR. Introduction Natural regeneration is a complex, stochastic process and understanding the dynamics associated with tree regeneration is critical to forest sustainability (Nyland, 1996; Puhlick et al., 2012; Bataineh et al., 2013). Regeneration abundance can be difficult to predict because it may be directly or indirectly influenced by abiotic factors, biotic factors or interactions between both (Puhlick et al., 2012; Fisichelli et al., 2013). Abiotic factors such as geology, soils, climate and fire influence the potential for germination, establishment and survival of regeneration, and also influence the current biotic conditions of overstory and understory species composition and structure (Paluch, 2005; Puhlick et al., 2012; Fisichelli et al., 2013). Biotic disturbances, such as disease or insect outbreaks, or anthropogenic disturbances including previous silviculture treatments, can alter the trajectory of stand dynamics by creating establishment sites and increasing resource availability to regeneration (Gray et al., 2005; Waring and O’Hara, 2005; Bataineh et al., 2013). However, disease or insect outbreaks may also cause infection and mortality to regeneration, so the affected species may not be able to take advantage of increased resources. Currently, forest structure in many dry mixed conifer forests of the Western US is considered dense and overcrowded with novel species mixtures due to disruption of historical fire regimes and intense logging practices (Gray et al., 2005; Korb et al., 2012; Rodman et al., 2016). Silviculture treatments (reducing basal area (BA) through regeneration treatments or intermediate treatments, such as thinning) may be effective tools for successful regeneration of desired species (e.g. Stoddard et al., 2015) but disease and insect disturbances can alter desired outcomes of such treatments and should be considered in order to create the most resilient future forest structures (Waring and O’Hara, 2005; Zald et al., 2008). Anticipating changes, realignment of disturbed systems and promoting resilience are all important tools in adaptive management in the face of invasive pests and a changing climate (Waring and O’Hara, 2005; Millar et al., 2007; O’Hara and Ramage, 2013; DeRose and Long, 2014). More effective silviculture treatments may be developed when primary abiotic and biotic factors, and regulatory patterns that govern natural regeneration success have been identified. Understanding the baseline ecology of natural regeneration is essential when species sustainability is threatened by an invasive pest (Schoettle and Sniezko, 2007; Loo, 2009; Goodrich and Waring, 2017). Southwestern white pine (Pinus strobiformis Engelm., SWWP) is a five-needle pine species that occurs across mixed conifer forests throughout Mexico and more disjunctly in the Southwest US (Arizona, New Mexico, southwestern Colorado and western Texas) (Little, 1971; Steinhoff and Andresen, 1971; Looney and Waring, 2013). Like all North American five-needle pine species, SWWP is susceptible to infection by Cronartium ribicola (J.C. Fisch.), the invasive fungal pathogen that causes white pine blister rust (WPBR). WPBR is a canker-causing disease that can lead to branch dieback, topkill and mortality of all sizes of SWWP (Conklin et al., 2009; Looney et al., 2015). Although SWWP is generally a minor species component in southwestern US, this species contributes to biodiversity and ecosystem resiliency as a food and habitat resource for wildlife and through the ability of mature trees to survive fire (Grissino-Mayer et al., 1995; Samano and Tomback, 2003; Mattson and Arundel, 2013). High rates of WPBR mortality, delayed forest recovery and loss of ecosystem services in other North America five-needle pine systems threatened by WPBR preview what may occur in some Southwest ecosystems with no proactive conservation and management strategies in place (Schoettle, 2004; Schoettle and Sniezko, 2007; Tomback and Achuff, 2010; Keane and Schoettle, 2011). Conservation strategies for other five-needle pines include mechanical cutting and prescribed burning to reduce competition and enhance natural regeneration and promoting rust resistance through planting genetically resistant seedlings (Schoettle and Sniezko, 2007; Keane and Schoettle, 2011; Schoettle et al., 2012; Keane et al., 2017). Genetic resistance to C. ribicola has been found in SWWP and current work is in progress to quantify the frequency of resistance across the SWWP range (Conklin et al., 2009; Sniezko et al., 2011; Waring et al., 2017; Goodrich et al., in press). The suitable distribution range for SWWP is expected to contract across southwestern US by 2090 in most future climate predictions (Crookston and Rehfeldt, 2008; Shirk et al., 2018), furthering the need to enhance resiliency in SWWP forests. Silviculture treatments best suited to stimulate and encourage SWWP regeneration are not well-defined, in part because we lack knowledge of the species’ ability to grow and compete under different light regimes, amongst different species and across various stand conditions. Retaining overstory SWWP in silviculture prescriptions to increase SWWP regeneration is recommended, as is removing competing vegetation in intermediate treatments (Conklin et al., 2009; Looney and Waring, 2012). Recent work illustrates that SWWP seedlings grow most quickly in open conditions created by regeneration treatments (Goodrich and Waring, 2017). There is a lack of published SWWP regeneration density data across both managed and non-managed southwestern US forests (Jones, 1971; Looney and Waring, 2013, but see Ffolliott and Gottfried, 1991; Looney and Waring, 2012; Goodrich and Waring, 2017). A gradient of WPBR incidence exists across southwestern mixed conifer forests (Conklin, 2004; Fairweather and Geils, 2011; Looney et al., 2015), creating opportunities to study SWWP regeneration dynamics with both the presence and absence of the disease. Our overarching goal was to survey for regeneration within non-managed stands and recently (<20 years) managed stands with a range of previous silviculture treatments and WPBR severities in order to provide better management recommendations to sustain SWWP through enhanced regeneration success. Results may be helpful in guiding conservation and management strategies of SWWP in the face of an invasive pathogen by providing direction on site selection and prescriptions for a variety of objectives (e.g. intermediate treatments, regeneration treatments and seedling outplanting efforts) when restoration is warranted (Schoettle and Sniezko, 2007; Keane and Schoettle, 2011). Our research questions were the following: What are average overstory and regeneration densities and proportions of SWWP compared with co-occurring species across six mountain ranges in southwestern US? What are average regeneration densities of SWWP across past silviculture treatments and two levels of WPBR severity? What between- and within-stand variables are associated with SWWP regeneration abundance? Do differences in SWWP size class distributions occur across silviculture treatments or WPBR severity levels? Methods Study area We sampled mixed conifer forests with SWWP as a substantial overstory component across Arizona and New Mexico National Forest and Native American tribal land in 2012–13 (Figure 1; Goodrich and Waring, 2017). Our surveys included National Forests (NF) land, the Fort Apache Indian Reservation (owned by the White Mountain Apache tribe) in Arizona and the Mescalero Apache Indian Reservation (owned by the Mescalero Apache tribe) in New Mexico. We grouped stands by the mountain ranges (hereafter referred to as sites) in which they occurred: San Francisco Peaks (n = 3 stands), Mogollon Rim (n = 6), White Mountains (n = 11), Mt Graham (n = 4), Signal Mountain (n = 3) and Sacramento Mountains (n = 28) (Figure 1). Our sampling represented a range of topographic relative moisture index (TRMI) values, a scalar index ranging from 0 (xeric) to 60 (mesic) and intended to represent site moisture characteristics by combining aspect, slope and topographic variables (Parker, 1982). Figure 1 View largeDownload slide Fifty-five stands sampled for regeneration across six mountain ranges (sites) in Arizona and New Mexico. White pine distribution layer credit of United States Geological Survey (2006) based on Little (1971). WPBR = white pine blister rust. Figure 1 View largeDownload slide Fifty-five stands sampled for regeneration across six mountain ranges (sites) in Arizona and New Mexico. White pine distribution layer credit of United States Geological Survey (2006) based on Little (1971). WPBR = white pine blister rust. Silviculture treatments and WPBR severity We sampled stands across two current WPBR severity levels and three silviculture treatment categories within ranges of TRMI values. Within categories of low (10–<25), medium (>25–<40) and high (>40−55) TRMI values we attempted to sample two levels of WPBR severity (low and high, defined below), and include at least three stands within each of the following levels of silviculture treatments (representing a gradient of residual BA from high to low): (1) stands with no recent silviculture treatments (no tree removal or prescribed fires) in the past 20 years, (2) recent (<20 years prior) uneven-aged (UEA) regeneration treatments or timber stand improvement treatments and (3) recent (<20 years prior) even-aged (EA) regeneration treatments. UEA structures were generally created through single-tree selection regeneration methods, thinning or a fuels treatment (Supplementary data Table S1). EA structures were created through establishment phase of shelterwood with reserves or clearcut with reserves regeneration methods, leaving two-aged residual structures (Supplementary data Table S1). EA regeneration methods were only found on the Mescalero Apache Indian Reservation in the Sacramento Mountains. We sampled a minimum of three stands per silviculture treatment in every TRMI category with the exception of high TRMI (>40−55), where only two recently-managed UEA stands and one recently-managed two-aged stand were sampled. Stands with high WPBR severity were defined as those with >10% stand incidence of WPBR (percent of overstory trees infected with WPBR, range: 0–88%) and at least 5% mean stand severity (crown dieback due to WPBR cankers, range: 0–25.8%) of infected trees. Stands with no WPBR and low severity WPBR were grouped together because early effects of the disease should not affect overstory or understory composition. Five percent mean severity was selected as the cutoff point where changes in overstory crown conditions could influence understory conditions and has been used as a cutoff between healthy and declining health at the individual tree level in previous research (Kearns and Jacobi, 2007). There were no pre-treatment SWWP regeneration density data or WPBR severity data available for any surveyed stands. Stand selection and plot sampling Stands were defined by stand delineation boundaries on NF lands or treatment boundaries on tribal lands. We followed stand selection methods defined in Looney and Waring (2012) and described in Goodrich and Waring (2017). In brief, we selected stands to scout for suitability by limiting our queries to NF stand exam data that contained ≥6.9 m2 ha−1 BA of SWWP. Permanent plot data were used as scouting information for stands on the Fort Apache Reservation (Looney and Waring, 2012) and local knowledge was used to identify potential reconnaissance stands on the Mescalero Apache Indian Reservation (William Hornsby, pers. comm., Bureau of Indian Affairs Branch of Forestry, 2012–13) where stand exam data were unavailable. Within all potential stands, a random point was generated within the stand boundaries and located in the field using Global Positioning System (GPS). Plots (50 m × 20 m, 0.1 ha) were established if a minimum SWWP overstory density occurred and randomly relocated within the stand if adequate densities did not occur (see Goodrich and Waring, 2017 for details). Fifty-five stands were surveyed using a combination of 0.1 ha rectangular plots and transects with three 0.0016 ha plots per transect using the same methods as described in Goodrich and Waring (2017). Each 0.1 ha rectangular plot was oriented parallel to slope contour and size corrected for slope. For all overstory trees, we recorded species, diameter at breast height (DBH) and condition (live or dead). Tree height, height to base of live crown, crown widths and damage agents were measured on all overstory SWWP stems. All SWWP were visually inspected using binoculars for signs or symptoms (e.g. aecia, branch flagging, branch swelling, squirrel chewing or roughened bark) of WPBR (Looney et al., 2015). Crown dieback due to WPBR was recorded by tracing tree crowns and dieback (branch flagging and recently killed branches) on transparencies and later estimated using Adobe Photoshop CS5.1 (Millers et al., 1991). If a tree had ≤2 branch cankers, an estimate of dieback was visually made, bypassing the transparency sketch (17/83 trees with visual crown dieback estimated). Additional site characteristics measured at each stand included GPS coordinates, percent slope, aspect and topographic and slope positions. Northness (−1 to 1 where −1 represents south and 1 represents north) and eastness (−1 to 1, west to east, respectively) were calculated from aspect (Roberts, 1986). Derived normalized 30-year mean (1961–90) climate variables were estimated from stand coordinates and elevations using thin plate spline surfaces of Rehfeldt (2006). Three 0.01 ha (10 m × 10 m) subplots were nested within the 0.1 ha plot located at distances of 0–10, 20–30 and 40–50 m above or below the centre line of the rectangular plot. All seedlings (<140 cm height) and saplings (≥140 cm height and ≤12.75 cm DBH) were measured for species, height, crown width and condition in each subplot. Within each subplot a line intercept technique (Husch, 2002) was used by placing a ground cover transect across the diagonal of the plot and ground cover type (i.e. grass, forb, shrub, litter, rock, bare, other) was recorded every 0.3 m. Canopy closure was averaged over four cardinal directions at each subplot centre with a spherical densiometer held at 1.2 m above the ground. A random azimuth and distance from the plot corner were used to start an additional 50 m transect with three additional 4 m × 4 m (0.0016 ha) regeneration subplots at 0, 25 and 50 m. A variable radius plot using an angle gauge with a BA factor of 2.3 m2 ha−1 was used to record overstory species BA in each subplot. All overstory and understory measurements were similar to those made in the 0.1-ha rectangular plot in the first transect, but overstory SWWP stems were not tagged and severity was not estimated with transparencies. In each subplot ground cover was estimated using a 1m × 1 m square and crown closure was estimated using a spherical densitometer at 1.2 m above the ground at subplot centre. Statistical analyses All statistical analyses were completed in R statistical programing environment (v. 3.1.1, R Core Team, 2014). We used Kruskal–Wallis (K–W) one-way analysis of variance by ranks tests, which are non-parametric tests to assess whether independent samples originate from the same distribution, to compare regeneration densities and probabilities across sites (n = 6 sites), silviculture treatment and WPBR severity levels (n = 3 levels of silviculture treatment and two levels of WPBR severity). Kendall’s τ correlation analyses were used to compare SWWP seedling and sapling densities to between- and within-stand variables. Subplot-level densities were compared with subplot-level factors (e.g. other species densities, BA and percent canopy closure) and mean stand densities relationship with between-stand variables (e.g. climate variables, percent slope and aspect) were assessed. Because of the wide geographic area that was sampled, common species varied between stands. Regeneration (with seedlings and saplings separated as they may occupy different niche spaces) and overstory species were grouped into (1) common shade-tolerant species: white fir (Abies concolor (Gord. & Glend.) Lindl. Ex Hildebr), corkbark fir (Abies lasiocarpa (Hook.) Nutt. var. arizonica (Merriam) Lemmon), Engelmann spruce (Picea engelmannii Parry ex Engelm.) and blue spruce (Picea pungens Engelm.) and (2) common shade-intolerant species: trembling aspen (Populus tremuloides Michx.), Gambel or other oak species (Quercus gambelii Nutt. or various species of Quercus), alligator juniper (Juniperus deppeana Steud.), Utah juniper (Juniperus osteosperma), Rocky Mountain juniper (Juniperus scopulorum Sarg.) and New Mexico locust (Robinia neomexicana A. Gray). SWWP were neither grouped nor were Douglas-fir (DF; Pseudotsuga menziesii (Mirb.) Franco) and ponderosa pine (PP; Pinus ponderosa Dougl. ex Laws) due to their common occurrence across our study areas. Hierarchical, mixed-effects models were used to compare SWWP regeneration densities across silviculture and WPBR severity levels and build predictive models. Normally, discrete count data are modelled using Poisson regression, so we modelled our seedling and sapling data as stem counts per subplot. Counts within the three smaller subplots were scaled to 0.01 ha counts and the three nested 0.01 ha subplots were averaged (four total subplots per stand). Regeneration count data often exhibit overdispersion (i.e. the variance is greater than the mean) and in such cases a negative binomial (NB) model is more appropriate where the mean and variance can be independently estimated (Ridout et al., 2001; Li et al., 2011; Crotteau et al., 2014). We confirmed SWWP seedling and sapling data were overdispersed by using the ‘dispersiontest’ function in package ‘AER’ (Supplementary data Table S2). Regeneration data also tend to contain more zeros than would be expected under a normal Poisson or NB distribution (zero-inflation) (Crotteau et al., 2014). We confirmed SWWP seedling and sapling data were zero-inflated by comparing the observed and expected proportions of zeros in each data set under a Poisson distribution (Supplementary data Table S2). Zero-inflation is typically accounted for by jointly modelling the zero/non-zero component (using logistic regression) and count component (which can include zeros under a Poisson or NB distribution) (Li et al., 2011). The zero-inflated NB (ZINB) distribution provides flexibility in the variance assumptions and accounts for overdispersion, so we modelled the data with a ZINB probability mass function (Supplementary data equations (S1) and (S2)). Because our data were hierarchical, with subplots nested within stands and stands clustered within sites, we used a mixed-effects modelling approach, which allows for the inclusion of both fixed and random effects (Supplementary data equation (S3)). Models included nested random effects of ‘site’ and ‘stand within-site’ and were fitted using the ‘glmmADMB’ package in R (Skaug et al., 2011). To answer our research questions, we developed two ZINB models for both SWWP seedling and sapling counts that included the following explanatory variables: Silviculture treatment, WPBR severity and an interaction term included as fixed effects (research question 2). Silviculture treatment, WPBR severity, an interaction term and between- and within-stand variables included as fixed effects and reduced to retain best models (research question 3). We chose other potential between- and within-stand variables if they were significant (P ≤ 0.05) in the Kendall’s τ correlation analyses. We then took all potential covariates and reduced them further by viewing all possible ZINB univariate models and only keeping variables that were significantly related to SWWP seedling or sapling counts at P ≤ 0.10. We then built predictive ZINB models by modelling all combinations of the significant variables and compared the models with Akaike’s information criteria (AIC) (Akaike, 1974), the change in AIC value for each model relative to the best model (∆AIC) and Akaike’s weights (AICw) (‘qpcR’ package, akaike.weights function, Spiess, 2014) (Burnham and Anderson, 2002). We chose the best candidate model as the model with the lowest AIC and in which all covariates were significant at P ≤ 0.05. We assessed the predictive ability of the top candidate models by examining plots of observed and predicted seedling and sapling frequencies (Li et al., 2011; Auty et al., 2014) and calculated the root mean square error (RMSE) from the predicted and observed values for each candidate model. Kolmogorov–Smirnov (K–S) bootstrap tests (1000 replicates) were used to compare SWWP diameter distributions (in 10-cm DBH classes) between WPBR severities within each silviculture treatment level (‘Matching’ package, ks.boot function, Sekhon, 2011). We also compared distributions of WPBR-infected SWWP between silviculture treatments in high WPBR severity stands with K–S tests. We used diameter distributions to calculate ratios of each diameter size class to the next largest size class. The average ratio across size classes in none/low WPBR and non-managed stands within each site was used to calculate the number of seedlings needed to regenerate a stand if one tree per hectare (TPH) of SWWP was desired in the largest size class (≥62.7 cm DBH). The number of seedlings needed in high WPBR stands was calculated by taking this value and dividing it by the average incidence of WPBR-infected TPH to account for WPBR-induced mortality. Results SWWP seedlings occurred in 86% and SWWP saplings occurred in 84% of the 55 stands sampled, and densities ranged from 0 to 3 931 seedlings ha−1 and from 0 to 1 304 saplings ha−1 per stand. There were differences in mean ranks for SWWP seedling densities but not SWWP sapling densities across sites (Table 1 and Figure 2a,c). The San Francisco Peaks had the highest SWWP seedling and sapling densities (Figure 2a,c). The proportion of SWWP seedlings to total seedling densities varied across sites (Table 1) and ranged from 2.7% in the Sacramento Mountains to 42.5% on the San Francisco Peaks (Figure 2a). The ratio of SWWP saplings to total sapling densities also varied widely, from 7.1% on Mount Graham to 60.6% on the San Francisco Peaks (Figure 2c), although mean ranks were not significantly different (Table 1). Mean ranks did not differ between sites for DF seedlings or saplings, but there were site-to-site differences in PP seedlings, shade-intolerant seedlings/suckers and shade-tolerant seedlings and saplings (Table 1 and Figure 2a,c). Mean overstory BA ranks differed between sites for live SWWP, dead SWWP, live DF, live PP, live shade-tolerant and live shade-intolerant species (Table 1 and Figure 3a). In general, the San Francisco Peaks had the highest SWWP live BA and Mount Graham had the highest total BA (Table 1 and Figure 3a). The proportion of SWWP BA significantly varied across sites (Table 1) and ranged from 16.3% on Signal Peak to 50.8% on the San Francisco Peaks (Figure 3a). Mean rank differences of some between-stand variables occurred among sites, including elevation, climate variables (mean annual temperature, mean annual precipitation and the proportion of mean annual precipitation that occurs in July, August and September) and percent canopy closure (Supplementary data Table S3). Table 1 Kruskal–Wallis’ (K–W) Chi-squared H statistics, degrees of freedom (df) and P-values for common species regeneration densities (trees per hectare, TPH), proportions and overstory basal areas (BA, m2 ha−1) amongst six sites and six silviculture treatment/WPBR disease severity levels. Kruskal–Wallis’ H statistics significant at P ≤ 0.05 (bolded P-values) indicate differences in the distribution of ranks. Kruskal–Wallis mean ranks by site (n = 6) Kruskal–Wallis mean ranks in silviculture treatment/disease severity levels (n = 6) K–W Chi-squared P-value (df = 5) K–W Chi-squared P-value (df = 5) SWWPa seedlings 18.79 0.002 9.440 0.093 SWWP saplings 5.69 0.337 10.420 0.064 SWWP seedlings/total seedlings 23.95 <0.0001 9.370 0.100 SWWP saplings/total saplings 8.03 0.155 10.970 0.052 DFb seedlings 6.38 0.271 16.540 0.005 DF saplings 2.18 0.824 3.500 0.623 PPc seedlings 34.31 <0.0001 20.770 0.001 PP saplings 7.82 0.166 2.680 0.750 Shade-intolerant seedlings 63.66 <0.001 24.280 0.0002 Shade-intolerant saplings 10.51 0.062 9.180 0.102 Shade-tolerant seedlings 28.02 <0.001 8.440 0.134 Shade-tolerant saplings 23.63 0.0002 2.350 0.799 Live SWWP BA 22.91 0.0003 15.240 0.009 Dead SWWP BA 29.87 <0.0001 11.830 0.037 Live SWWP BA/Live total BA 13.37 0.020 11.84 0.037 Live DF BA 16.77 0.005 57.230 0.0001 Live PP BA 76.36 <0.0001 24.730 0.0002 Live shade-intolerant BA 84.75 <0.0001 25.370 0.0001 Live shade-tolerant BA 41.71 <0.0001 6.530 0.258 Kruskal–Wallis mean ranks by site (n = 6) Kruskal–Wallis mean ranks in silviculture treatment/disease severity levels (n = 6) K–W Chi-squared P-value (df = 5) K–W Chi-squared P-value (df = 5) SWWPa seedlings 18.79 0.002 9.440 0.093 SWWP saplings 5.69 0.337 10.420 0.064 SWWP seedlings/total seedlings 23.95 <0.0001 9.370 0.100 SWWP saplings/total saplings 8.03 0.155 10.970 0.052 DFb seedlings 6.38 0.271 16.540 0.005 DF saplings 2.18 0.824 3.500 0.623 PPc seedlings 34.31 <0.0001 20.770 0.001 PP saplings 7.82 0.166 2.680 0.750 Shade-intolerant seedlings 63.66 <0.001 24.280 0.0002 Shade-intolerant saplings 10.51 0.062 9.180 0.102 Shade-tolerant seedlings 28.02 <0.001 8.440 0.134 Shade-tolerant saplings 23.63 0.0002 2.350 0.799 Live SWWP BA 22.91 0.0003 15.240 0.009 Dead SWWP BA 29.87 <0.0001 11.830 0.037 Live SWWP BA/Live total BA 13.37 0.020 11.84 0.037 Live DF BA 16.77 0.005 57.230 0.0001 Live PP BA 76.36 <0.0001 24.730 0.0002 Live shade-intolerant BA 84.75 <0.0001 25.370 0.0001 Live shade-tolerant BA 41.71 <0.0001 6.530 0.258 aSWWP = southwestern white pine. bDF = Douglas-fir. cPP = ponderosa pine. Table 1 Kruskal–Wallis’ (K–W) Chi-squared H statistics, degrees of freedom (df) and P-values for common species regeneration densities (trees per hectare, TPH), proportions and overstory basal areas (BA, m2 ha−1) amongst six sites and six silviculture treatment/WPBR disease severity levels. Kruskal–Wallis’ H statistics significant at P ≤ 0.05 (bolded P-values) indicate differences in the distribution of ranks. Kruskal–Wallis mean ranks by site (n = 6) Kruskal–Wallis mean ranks in silviculture treatment/disease severity levels (n = 6) K–W Chi-squared P-value (df = 5) K–W Chi-squared P-value (df = 5) SWWPa seedlings 18.79 0.002 9.440 0.093 SWWP saplings 5.69 0.337 10.420 0.064 SWWP seedlings/total seedlings 23.95 <0.0001 9.370 0.100 SWWP saplings/total saplings 8.03 0.155 10.970 0.052 DFb seedlings 6.38 0.271 16.540 0.005 DF saplings 2.18 0.824 3.500 0.623 PPc seedlings 34.31 <0.0001 20.770 0.001 PP saplings 7.82 0.166 2.680 0.750 Shade-intolerant seedlings 63.66 <0.001 24.280 0.0002 Shade-intolerant saplings 10.51 0.062 9.180 0.102 Shade-tolerant seedlings 28.02 <0.001 8.440 0.134 Shade-tolerant saplings 23.63 0.0002 2.350 0.799 Live SWWP BA 22.91 0.0003 15.240 0.009 Dead SWWP BA 29.87 <0.0001 11.830 0.037 Live SWWP BA/Live total BA 13.37 0.020 11.84 0.037 Live DF BA 16.77 0.005 57.230 0.0001 Live PP BA 76.36 <0.0001 24.730 0.0002 Live shade-intolerant BA 84.75 <0.0001 25.370 0.0001 Live shade-tolerant BA 41.71 <0.0001 6.530 0.258 Kruskal–Wallis mean ranks by site (n = 6) Kruskal–Wallis mean ranks in silviculture treatment/disease severity levels (n = 6) K–W Chi-squared P-value (df = 5) K–W Chi-squared P-value (df = 5) SWWPa seedlings 18.79 0.002 9.440 0.093 SWWP saplings 5.69 0.337 10.420 0.064 SWWP seedlings/total seedlings 23.95 <0.0001 9.370 0.100 SWWP saplings/total saplings 8.03 0.155 10.970 0.052 DFb seedlings 6.38 0.271 16.540 0.005 DF saplings 2.18 0.824 3.500 0.623 PPc seedlings 34.31 <0.0001 20.770 0.001 PP saplings 7.82 0.166 2.680 0.750 Shade-intolerant seedlings 63.66 <0.001 24.280 0.0002 Shade-intolerant saplings 10.51 0.062 9.180 0.102 Shade-tolerant seedlings 28.02 <0.001 8.440 0.134 Shade-tolerant saplings 23.63 0.0002 2.350 0.799 Live SWWP BA 22.91 0.0003 15.240 0.009 Dead SWWP BA 29.87 <0.0001 11.830 0.037 Live SWWP BA/Live total BA 13.37 0.020 11.84 0.037 Live DF BA 16.77 0.005 57.230 0.0001 Live PP BA 76.36 <0.0001 24.730 0.0002 Live shade-intolerant BA 84.75 <0.0001 25.370 0.0001 Live shade-tolerant BA 41.71 <0.0001 6.530 0.258 aSWWP = southwestern white pine. bDF = Douglas-fir. cPP = ponderosa pine. Figure 2 View largeDownload slide (a–d) Mean southwestern white pine (SWWP), ponderosa pine (PP), Douglas-fir (DF), shade-tolerant and shade-intolerant regeneration densities across six sites in Arizona and New Mexico (a. seedlings and c. saplings) and across six silviculture treatment/white pine blister rust (WPBR) severity levels (b. seedlings and d. saplings). Percentages indicate proportion of SWWP as total regeneration TPH at the subplot level (Note: subplots without regeneration were calculated as ‘NA’ and not zero, so subplot-level percentages are slightly different than strata-level means in the graph). Silviculture treatments: no recent mgmt (20 years), recent (20 years) uneven-aged (rUEA) treatments and recent (20 years) two-aged (rTWO) treatments. Sites ordered by longitude (west–east). Figure 2 View largeDownload slide (a–d) Mean southwestern white pine (SWWP), ponderosa pine (PP), Douglas-fir (DF), shade-tolerant and shade-intolerant regeneration densities across six sites in Arizona and New Mexico (a. seedlings and c. saplings) and across six silviculture treatment/white pine blister rust (WPBR) severity levels (b. seedlings and d. saplings). Percentages indicate proportion of SWWP as total regeneration TPH at the subplot level (Note: subplots without regeneration were calculated as ‘NA’ and not zero, so subplot-level percentages are slightly different than strata-level means in the graph). Silviculture treatments: no recent mgmt (20 years), recent (20 years) uneven-aged (rUEA) treatments and recent (20 years) two-aged (rTWO) treatments. Sites ordered by longitude (west–east). Figure 3 View largeDownload slide (a and b) Southwestern white pine (SWWP), ponderosa pine (PP), Douglas-fir (DF), shade-tolerant and shade-intolerant overstory basal areas (BA, m2 ha−1) across six sites in Arizona and New Mexico (a) and six silviculture treatment/white pine blister rust (WPBR) severity levels. Percentages indicate proportion of SWWP as total BA at the subplot level (Note: subplots without overstory trees were calculated as ‘NA’ and not zero, so subplot-level percentages are slightly different than strata-level means in the graph). Silviculture treatments: no recent mgmt, recent uneven-aged (rUEA) treatments and recent two-aged (rTWO) treatments. Sites ordered by longitude (west–east). Figure 3 View largeDownload slide (a and b) Southwestern white pine (SWWP), ponderosa pine (PP), Douglas-fir (DF), shade-tolerant and shade-intolerant overstory basal areas (BA, m2 ha−1) across six sites in Arizona and New Mexico (a) and six silviculture treatment/white pine blister rust (WPBR) severity levels. Percentages indicate proportion of SWWP as total BA at the subplot level (Note: subplots without overstory trees were calculated as ‘NA’ and not zero, so subplot-level percentages are slightly different than strata-level means in the graph). Silviculture treatments: no recent mgmt, recent uneven-aged (rUEA) treatments and recent two-aged (rTWO) treatments. Sites ordered by longitude (west–east). There was considerable variation in densities of several species of regeneration across the six levels of silviculture treatments and WPBR severity but there were no differences in mean ranks in SWWP seedling or SWWP sapling densities (P = 0.06 and 0.09, respectively) or the proportions of SWWP seedlings and saplings to total seedling and sapling densities (P = 0.10 and 0.052, respectively) across the six levels in the non-parametric analyses (Table 1 and Figure 2b,d). However, we found significant effects of both silviculture treatment and WPBR severity on SWWP regeneration abundance with the ZINB models that supported trends in non-parametric analyses. A significant effect of silviculture treatment occurred on SWWP seedling counts (P = 0.041) where SWWP seedling counts were greater in non-managed and recently-managed two-aged stands compared with recently-managed UEA stands (Figure 4). There was also a significant effect of WPBR on seedling counts; they were lower in high WPBR severity stands compared with none/low WPBR severity stands across all silviculture treatments (P = 0.032) (Figure 4). SWWP sapling counts were higher in non-managed stands compared with recently-managed two-aged and UEA stands in low/none WPBR severity areas (P = 0.034, Figure 4). There was no effect of WPBR severity alone on sapling counts (P = 0.99) but the interaction between management and WPBR severity was significant (P = 0.027). There were higher SWWP sapling counts in high WPBR severity stands in recently-managed two-aged residual structures compared with none/low severity stands with a two-aged structure (Figure 4, P = 0.03). Figure 4 View largeDownload slide Seedling and sapling zero-inflated negative binomial mixed effect model estimations (seedling and sapling counts per subplot) for the interaction of silviculture treatment (no mgmt, recently treated uneven-aged residual structure (rUEA) and recently treated two-aged residual stand (rTWO) and WPBR severity (none/light and heavy) levels. df = degrees of freedom; Chisq = Chi-square; Pr(Chisq) = P-value. Error bars represent ±1 SE. Figure 4 View largeDownload slide Seedling and sapling zero-inflated negative binomial mixed effect model estimations (seedling and sapling counts per subplot) for the interaction of silviculture treatment (no mgmt, recently treated uneven-aged residual structure (rUEA) and recently treated two-aged residual stand (rTWO) and WPBR severity (none/light and heavy) levels. df = degrees of freedom; Chisq = Chi-square; Pr(Chisq) = P-value. Error bars represent ±1 SE. Overstory species composition and mean BA varied significantly across silviculture treatment and WPBR severity levels (Table 1 and Figure 3b). Live SWWP, dead SWWP, live DF, live PP and live shade-intolerant species BA all varied across levels; non-managed stands had the highest total BA (>25 m2 ha−1), recently-managed two-aged structures had the lowest total BA (<10 m2 ha−1) and recently-managed UEA structures had total BA between these values (~20 m2 ha−1) (Figure 3b). The proportion of SWWP BA to total BA varied across levels and was generally highest in recently-managed two-aged stands (Table 1 and Figure 3b). Mean ranks of between-stand attributes were commonly variable; factors that varied between management and disease levels were eastness, percent slope, percent canopy closure and proportion of mean annual precipitation that occurs July–September (Supplementary data Table S3). Stands with high WPBR severity were more east-facing than the stands with none/low WPBR severity (Supplementary data Table S3). Recently managed stands were less steep than non-managed stands and two-aged stands had lower percent canopy closure than non-managed or recently-managed UEA stands (Supplementary data Table S3). Non-managed stands had the largest proportion of WPBR-infected SWWP regeneration stems in high severity stands (30% of regeneration infected), 4.0% were infected in recently-managed UEA stands and two-aged structures had no regeneration infected in high severity areas (Table 2). Incidence of WPBR in overstory trees and mean crown dieback of infected trees were highest in non-managed, high severity stands but high across all silviculture treatments in stands with high severity WPBR (Table 2). There were no significant differences in mean ranks of Ribes species densities (the most common alternate host in the WPBR disease cycle in the Southwest) across silviculture treatment/WPBR severity levels, although the highest densities occurred in UEA stands and lowest in recently-managed two-aged structures (Table 2). Table 2 Disease related variable means and standard deviation (SD) and Kruskal–Wallis’ one-way analysis of variance by rank test for differences in ranks among disease and silviculture treatment levels. Kruskal–Wallis’ Chi-squared (H statistic) significant at P ≤ 0.05 (bolded values) indicate differences in the distribution of ranks between silviculture treatments and disease levels (n = number of stands in each level). Silviculture treatmenta Disease levelsb n SWWP regeneration WPBR incidence (%) SWWP overstory WPBR incidence (%) SWWP crown dieback from WPBR (%) Ribes density (SPHc) Mean SD Mean SD Mean SD Mean SD No mgmt None/low severity 26 0.02 0.13 0.06 0.38 0.37 0.98 64.8 144.7 rUEA 9 0.00 0.00 0.01 0.46 0.25 0.67 81.0 153.8 rTWO 6 0.00 0.00 0.06 0.16 0.38 0.80 28.6 75.6 No mgmt High severity 8 0.30 0.38 0.50 0.20 13.41 6.64 58.3 101.9 rUEA 3 0.04 0.09 0.49 0.04 9.76 3.02 244.4 234.1 rTWO 3 0.00 0.00 0.15 0.25 11.44 8.94 0.0 0.0 Kruskal–Wallis Chi-squared, P-value (dfd= 5 for all) H = 55.8, P< 0.0001 H = 72.16, P < 0.0001 H = 76.34, P< 0.0001 H = 6.18, P = 0.289 Silviculture treatmenta Disease levelsb n SWWP regeneration WPBR incidence (%) SWWP overstory WPBR incidence (%) SWWP crown dieback from WPBR (%) Ribes density (SPHc) Mean SD Mean SD Mean SD Mean SD No mgmt None/low severity 26 0.02 0.13 0.06 0.38 0.37 0.98 64.8 144.7 rUEA 9 0.00 0.00 0.01 0.46 0.25 0.67 81.0 153.8 rTWO 6 0.00 0.00 0.06 0.16 0.38 0.80 28.6 75.6 No mgmt High severity 8 0.30 0.38 0.50 0.20 13.41 6.64 58.3 101.9 rUEA 3 0.04 0.09 0.49 0.04 9.76 3.02 244.4 234.1 rTWO 3 0.00 0.00 0.15 0.25 11.44 8.94 0.0 0.0 Kruskal–Wallis Chi-squared, P-value (dfd= 5 for all) H = 55.8, P< 0.0001 H = 72.16, P < 0.0001 H = 76.34, P< 0.0001 H = 6.18, P = 0.289 aSilviculture treatments: rUEA = recently-managed, uneven-aged; rTWO = recently-managed, two-aged, disease levels. bWPBR severity level (none/low severity and high severity). cStems per hectare. dDegrees of freedom. Table 2 Disease related variable means and standard deviation (SD) and Kruskal–Wallis’ one-way analysis of variance by rank test for differences in ranks among disease and silviculture treatment levels. Kruskal–Wallis’ Chi-squared (H statistic) significant at P ≤ 0.05 (bolded values) indicate differences in the distribution of ranks between silviculture treatments and disease levels (n = number of stands in each level). Silviculture treatmenta Disease levelsb n SWWP regeneration WPBR incidence (%) SWWP overstory WPBR incidence (%) SWWP crown dieback from WPBR (%) Ribes density (SPHc) Mean SD Mean SD Mean SD Mean SD No mgmt None/low severity 26 0.02 0.13 0.06 0.38 0.37 0.98 64.8 144.7 rUEA 9 0.00 0.00 0.01 0.46 0.25 0.67 81.0 153.8 rTWO 6 0.00 0.00 0.06 0.16 0.38 0.80 28.6 75.6 No mgmt High severity 8 0.30 0.38 0.50 0.20 13.41 6.64 58.3 101.9 rUEA 3 0.04 0.09 0.49 0.04 9.76 3.02 244.4 234.1 rTWO 3 0.00 0.00 0.15 0.25 11.44 8.94 0.0 0.0 Kruskal–Wallis Chi-squared, P-value (dfd= 5 for all) H = 55.8, P< 0.0001 H = 72.16, P < 0.0001 H = 76.34, P< 0.0001 H = 6.18, P = 0.289 Silviculture treatmenta Disease levelsb n SWWP regeneration WPBR incidence (%) SWWP overstory WPBR incidence (%) SWWP crown dieback from WPBR (%) Ribes density (SPHc) Mean SD Mean SD Mean SD Mean SD No mgmt None/low severity 26 0.02 0.13 0.06 0.38 0.37 0.98 64.8 144.7 rUEA 9 0.00 0.00 0.01 0.46 0.25 0.67 81.0 153.8 rTWO 6 0.00 0.00 0.06 0.16 0.38 0.80 28.6 75.6 No mgmt High severity 8 0.30 0.38 0.50 0.20 13.41 6.64 58.3 101.9 rUEA 3 0.04 0.09 0.49 0.04 9.76 3.02 244.4 234.1 rTWO 3 0.00 0.00 0.15 0.25 11.44 8.94 0.0 0.0 Kruskal–Wallis Chi-squared, P-value (dfd= 5 for all) H = 55.8, P< 0.0001 H = 72.16, P < 0.0001 H = 76.34, P< 0.0001 H = 6.18, P = 0.289 aSilviculture treatments: rUEA = recently-managed, uneven-aged; rTWO = recently-managed, two-aged, disease levels. bWPBR severity level (none/low severity and high severity). cStems per hectare. dDegrees of freedom. Most factors significantly associated with SWWP seedling and sapling densities varied within stands, and several of these variables were other species’ regeneration abundance. An increase in both DF and PP seedling densities was significantly correlated with an increase in SWWP seedlings (τ = 0.292 and 0.289, respectively; Supplementary data Table S4). Increases in shade-tolerant sapling densities, shade-tolerant BA and downed-woody debris ground cover were associated with decreases in SWWP seedlings (τ = −0.304, −0.269 and −0.19, respectively; Supplementary data Table S4). The only climate variable that correlated significantly with SWWP seedling densities was the proportion of mean annual precipitation that occurs in July through September (τ = −0.221; Supplementary data Table S4). There were no significant correlations between SWWP sapling densities and any climate or between-stand variables. An increase in DF sapling densities was significantly correlated with increasing SWWP sapling densities (τ = 0.394; Supplementary data Table S4). SWWP sapling densities were also significantly correlated with an increase in canopy closure (τ = 0.35), live SWWP BA (τ = 0.242) and live DF BA (τ = 0.185) (Supplementary data Table S4). The top candidate ZINB model predicted a decrease in SWWP seedling counts in high WPBR severity stands, with higher percent ground cover of downed-woody debris and in recently-managed UEA stands (Tables 3 and 4). A significant interaction between WPBR severity and DF seedling count indicated that the relationship between SWWP seedling counts and DF seedling count varied between disease levels (Tables 3 and 4). This model had a ∆AIC = 9.0 compared with the saturated model, indicating a better fit (Table 3). The RMSE of the top model was 2.55 seedlings (the range of SWWP seedlings was 0–69 seedlings per subplot). The top candidate SWWP sapling count ZINB model predicted higher SWWP sapling counts in recently-managed, two-aged stands and with increasing canopy closure (Tables 3 and 4). A significant interaction between two-aged stands and canopy closure indicates the relationship between SWWP sapling count and canopy closure varied between management levels. The top model had an AIC value 8.4 lower than the saturated model and a RMSE of 1.21 saplings (range of SWWP saplings from 0 to 31 saplings per subplot; Table 3). The best ZINB models predicted zero counts extremely well but were less accurate in predicting regeneration counts above six stems per subplot (Supplementary data Figure S1). Table 3 Zero-inflated negative binomial (ZINB) models predicting southwestern white pine (SWWP) seedling and sapling counts (per subplot). Shaded value indicates the top candidate model based on AIC and all covariates significant at P < 0.05. Dependent variable Model name Parameters AICa ∆AICb AICwc RMSEd SWWP seedling count ZINBseed1 WPBR severity* DF seedling count + Management + DWDe 718 0.0 0.67 2.55 ZINBseed2 WPBR severity* DF seedling count + Management 721.3 3.3 0.13 2.14 ZINBseed3 WPBR severity* DF seedling count + DWD 721.4 3.4 0.12 2.15 ZINBseed4 WPBR severity+ DF seedling count + DWD 723.1 5.1 0.05 2.98 ZINBseed5 DF seedling count + DWD 724.3 6.3 0.03 2.46 ZINBseed-saturated Mgmt * WPBR severity + propJASf + DF seedling count + PP seedling count + shade-tolerant sapling count + DWD + Live SWWP BAg + Live shade-tolerant BA 727 9.0 – 4.69 SWWP sapling count ZINBsap1 canopy closure * Management 499.0 0.0 0.29 1.33 ZINBsap3 canopy closure * Management + litterh 499.9 0.9 0.19 1.45 ZINBsap4 canopy closure * Management + litter + DF sap count 500.3 1.3 0.15 1.39 ZINBsap5 Management*litter + canopy closure 503.4 4.4 0.03 1.74 ZINBsap-saturated Management * WPBR severity + DF sapling count + canopy closure + litter + Live SWWP BA + Live DF BA 507.3 8.3 – 0.52 Dependent variable Model name Parameters AICa ∆AICb AICwc RMSEd SWWP seedling count ZINBseed1 WPBR severity* DF seedling count + Management + DWDe 718 0.0 0.67 2.55 ZINBseed2 WPBR severity* DF seedling count + Management 721.3 3.3 0.13 2.14 ZINBseed3 WPBR severity* DF seedling count + DWD 721.4 3.4 0.12 2.15 ZINBseed4 WPBR severity+ DF seedling count + DWD 723.1 5.1 0.05 2.98 ZINBseed5 DF seedling count + DWD 724.3 6.3 0.03 2.46 ZINBseed-saturated Mgmt * WPBR severity + propJASf + DF seedling count + PP seedling count + shade-tolerant sapling count + DWD + Live SWWP BAg + Live shade-tolerant BA 727 9.0 – 4.69 SWWP sapling count ZINBsap1 canopy closure * Management 499.0 0.0 0.29 1.33 ZINBsap3 canopy closure * Management + litterh 499.9 0.9 0.19 1.45 ZINBsap4 canopy closure * Management + litter + DF sap count 500.3 1.3 0.15 1.39 ZINBsap5 Management*litter + canopy closure 503.4 4.4 0.03 1.74 ZINBsap-saturated Management * WPBR severity + DF sapling count + canopy closure + litter + Live SWWP BA + Live DF BA 507.3 8.3 – 0.52 aAIC = Akaike’s Information Criterion. b∆AIC = change in AIC in comparison to the best model. cAICw = Akaike’s weights. dRMSE = root mean square error. eDWD = percent ground cover of downed-woody debris. fpropJAS = proportion of mean annual precipitation that occurs July–September. gBA = live overstory tree basal area (m2 ha−1). hlitter = percent ground cover litter/duff. Table 3 Zero-inflated negative binomial (ZINB) models predicting southwestern white pine (SWWP) seedling and sapling counts (per subplot). Shaded value indicates the top candidate model based on AIC and all covariates significant at P < 0.05. Dependent variable Model name Parameters AICa ∆AICb AICwc RMSEd SWWP seedling count ZINBseed1 WPBR severity* DF seedling count + Management + DWDe 718 0.0 0.67 2.55 ZINBseed2 WPBR severity* DF seedling count + Management 721.3 3.3 0.13 2.14 ZINBseed3 WPBR severity* DF seedling count + DWD 721.4 3.4 0.12 2.15 ZINBseed4 WPBR severity+ DF seedling count + DWD 723.1 5.1 0.05 2.98 ZINBseed5 DF seedling count + DWD 724.3 6.3 0.03 2.46 ZINBseed-saturated Mgmt * WPBR severity + propJASf + DF seedling count + PP seedling count + shade-tolerant sapling count + DWD + Live SWWP BAg + Live shade-tolerant BA 727 9.0 – 4.69 SWWP sapling count ZINBsap1 canopy closure * Management 499.0 0.0 0.29 1.33 ZINBsap3 canopy closure * Management + litterh 499.9 0.9 0.19 1.45 ZINBsap4 canopy closure * Management + litter + DF sap count 500.3 1.3 0.15 1.39 ZINBsap5 Management*litter + canopy closure 503.4 4.4 0.03 1.74 ZINBsap-saturated Management * WPBR severity + DF sapling count + canopy closure + litter + Live SWWP BA + Live DF BA 507.3 8.3 – 0.52 Dependent variable Model name Parameters AICa ∆AICb AICwc RMSEd SWWP seedling count ZINBseed1 WPBR severity* DF seedling count + Management + DWDe 718 0.0 0.67 2.55 ZINBseed2 WPBR severity* DF seedling count + Management 721.3 3.3 0.13 2.14 ZINBseed3 WPBR severity* DF seedling count + DWD 721.4 3.4 0.12 2.15 ZINBseed4 WPBR severity+ DF seedling count + DWD 723.1 5.1 0.05 2.98 ZINBseed5 DF seedling count + DWD 724.3 6.3 0.03 2.46 ZINBseed-saturated Mgmt * WPBR severity + propJASf + DF seedling count + PP seedling count + shade-tolerant sapling count + DWD + Live SWWP BAg + Live shade-tolerant BA 727 9.0 – 4.69 SWWP sapling count ZINBsap1 canopy closure * Management 499.0 0.0 0.29 1.33 ZINBsap3 canopy closure * Management + litterh 499.9 0.9 0.19 1.45 ZINBsap4 canopy closure * Management + litter + DF sap count 500.3 1.3 0.15 1.39 ZINBsap5 Management*litter + canopy closure 503.4 4.4 0.03 1.74 ZINBsap-saturated Management * WPBR severity + DF sapling count + canopy closure + litter + Live SWWP BA + Live DF BA 507.3 8.3 – 0.52 aAIC = Akaike’s Information Criterion. b∆AIC = change in AIC in comparison to the best model. cAICw = Akaike’s weights. dRMSE = root mean square error. eDWD = percent ground cover of downed-woody debris. fpropJAS = proportion of mean annual precipitation that occurs July–September. gBA = live overstory tree basal area (m2 ha−1). hlitter = percent ground cover litter/duff. Table 4 Final zero-inflated negative binomial (ZINB) model parameter estimates, standard errors (SE), z-values and Pr(z) values for top SWWP seedling and sapling count per subplot models. Estimate SE z Value Pr (|z|) SWWP seedling (ZINB1seed)  Intercept 2.17 0.36 6.04 < 0.0001  WPBR severity −1.48 0.50 −2.97 0.003  DF seedling count 0.02 0.02 1.19 0.2334  rUEAa −1.39 0.60 −2.34 0.0195  rTWOb 0.43 0.49 0.87 0.3818  DWDc −2.89 1.24 −2.34 0.0195  WPBR severity * DF seedling count 0.12 0.04 2.87 0.0041 SWWP sapling (ZINB1sap)  Intercept −2.40 1.58 −1.52 0.129  rUEA −2.69 2.91 −0.92 0.356  rTWO 3.96 1.89 2.10 0.036  canopy closure 0.04 0.02 2.36 0.018  rUEA*canopy closure 0.03 0.04 0.78 0.433  rTWO*canopy closure −0.10 0.04 −2.72 0.007 Estimate SE z Value Pr (|z|) SWWP seedling (ZINB1seed)  Intercept 2.17 0.36 6.04 < 0.0001  WPBR severity −1.48 0.50 −2.97 0.003  DF seedling count 0.02 0.02 1.19 0.2334  rUEAa −1.39 0.60 −2.34 0.0195  rTWOb 0.43 0.49 0.87 0.3818  DWDc −2.89 1.24 −2.34 0.0195  WPBR severity * DF seedling count 0.12 0.04 2.87 0.0041 SWWP sapling (ZINB1sap)  Intercept −2.40 1.58 −1.52 0.129  rUEA −2.69 2.91 −0.92 0.356  rTWO 3.96 1.89 2.10 0.036  canopy closure 0.04 0.02 2.36 0.018  rUEA*canopy closure 0.03 0.04 0.78 0.433  rTWO*canopy closure −0.10 0.04 −2.72 0.007 arUEA = recently-managed uneven-aged stands. brTWO = recently-managed two-aged stands. cDWD = percent ground cover of downed-woody debris. ZINB1seed = Site (random) variance = 0.000000021 (SD = 0.00045), stand within-site (random) variance = 0.5847 (SD = 0.765), negative binomial dispersion parameter = 1.864 (SE = 0.763), zero-inflation = 0.532 (SE = 0.054). ZINB1sap = Site (random) variance = 0.000000021 (SD = 0.00045), stand within-site (random) variance = 0.1677 (SD = 0.410), negative binomial dispersion parameter = 0.698 (SE = 0.471), zero-inflation = 0.534 (SE = 0.121). Table 4 Final zero-inflated negative binomial (ZINB) model parameter estimates, standard errors (SE), z-values and Pr(z) values for top SWWP seedling and sapling count per subplot models. Estimate SE z Value Pr (|z|) SWWP seedling (ZINB1seed)  Intercept 2.17 0.36 6.04 < 0.0001  WPBR severity −1.48 0.50 −2.97 0.003  DF seedling count 0.02 0.02 1.19 0.2334  rUEAa −1.39 0.60 −2.34 0.0195  rTWOb 0.43 0.49 0.87 0.3818  DWDc −2.89 1.24 −2.34 0.0195  WPBR severity * DF seedling count 0.12 0.04 2.87 0.0041 SWWP sapling (ZINB1sap)  Intercept −2.40 1.58 −1.52 0.129  rUEA −2.69 2.91 −0.92 0.356  rTWO 3.96 1.89 2.10 0.036  canopy closure 0.04 0.02 2.36 0.018  rUEA*canopy closure 0.03 0.04 0.78 0.433  rTWO*canopy closure −0.10 0.04 −2.72 0.007 Estimate SE z Value Pr (|z|) SWWP seedling (ZINB1seed)  Intercept 2.17 0.36 6.04 < 0.0001  WPBR severity −1.48 0.50 −2.97 0.003  DF seedling count 0.02 0.02 1.19 0.2334  rUEAa −1.39 0.60 −2.34 0.0195  rTWOb 0.43 0.49 0.87 0.3818  DWDc −2.89 1.24 −2.34 0.0195  WPBR severity * DF seedling count 0.12 0.04 2.87 0.0041 SWWP sapling (ZINB1sap)  Intercept −2.40 1.58 −1.52 0.129  rUEA −2.69 2.91 −0.92 0.356  rTWO 3.96 1.89 2.10 0.036  canopy closure 0.04 0.02 2.36 0.018  rUEA*canopy closure 0.03 0.04 0.78 0.433  rTWO*canopy closure −0.10 0.04 −2.72 0.007 arUEA = recently-managed uneven-aged stands. brTWO = recently-managed two-aged stands. cDWD = percent ground cover of downed-woody debris. ZINB1seed = Site (random) variance = 0.000000021 (SD = 0.00045), stand within-site (random) variance = 0.5847 (SD = 0.765), negative binomial dispersion parameter = 1.864 (SE = 0.763), zero-inflation = 0.532 (SE = 0.054). ZINB1sap = Site (random) variance = 0.000000021 (SD = 0.00045), stand within-site (random) variance = 0.1677 (SD = 0.410), negative binomial dispersion parameter = 0.698 (SE = 0.471), zero-inflation = 0.534 (SE = 0.121). WPBR severity was not associated with a change in the shape of diameter distributions within any silviculture treatment (Figure 5a–c). There was no difference between diameter distributions of SWWP TPH between none/low and high WPBR severity levels in non-managed stands, (Figure 5a; K–S test D = 0.25, pks = 0.98), recently-managed UEA stands (Figure 5b; D = 0.25, pks = 0.98) or recently-managed two-aged stands (Figure 5c; D = 0.375, pks = 0.63). Management did not affect the diameter distributions of WPBR-infected SWWP either between non-managed stands and recently-managed, two-aged stands (D = 0.625, pks = 0.061; Figure 5a,c) or between non-managed stands and recently-managed UEA stands (D = 0.5, pks = 0.18; Figure 5a,b). Figure 5 View largeDownload slide (a–c) Diameter distributions of non-infected and WPBR-infected SWWP per hectare (TPH) in (a) non-managed stands with none/low severity WPBR (n = 27 stands) and severe WPBR (n = 8 stands), (b) recently-managed UEA stands with none/low severity WPBR (n = 8 stands) and severe WPBR (n = 3 stands) and (c) recently-managed two-aged stands with none/low severity WPBR (n = 6 stands) and severe WPBR (n = 3 stands). Figure 5 View largeDownload slide (a–c) Diameter distributions of non-infected and WPBR-infected SWWP per hectare (TPH) in (a) non-managed stands with none/low severity WPBR (n = 27 stands) and severe WPBR (n = 8 stands), (b) recently-managed UEA stands with none/low severity WPBR (n = 8 stands) and severe WPBR (n = 3 stands) and (c) recently-managed two-aged stands with none/low severity WPBR (n = 6 stands) and severe WPBR (n = 3 stands). In non-managed stands, there were large ratios between sapling-sized trees and the 12.7–22.7 cm DBH class, indicating most sapling-sized trees are not likely to reach the next size class in these stands (Figure 5a). Seedling densities necessary to maintain one TPH of SWWP ≥62.7 cm DBH in none/low WPBR severity, non-managed stands were calculated (mean ratio = 2.04). In order to maintain one SWWP TPH >62.7 cm DBH on average, there need to be approximately ≥150 SWWP seedlings per hectare. Trees in non-managed, high severity stands averaged 46% WPBR infection (range = 19.6–63% across diameter classes; Figure 5a), so in order to maintain one TPH of SWWP ≥62.7 cm DBH in non-managed, high severity WPBR stands there should be ≥300 SWWP seedlings ha−1 to account for potential mortality. Sixty percent of stands (n = 33/55) in this study met the >150 TPH criteria; five were recently-managed UEA stands and six were recently-managed two-aged stands. Eighteen stands met the >300 TPH seedling criteria, and four were recently-managed. The mean ratio and minimum seedling density (without accounting for WPBR infection) for non-managed, none/low WPBR severity stands for each site were as follows: San Francisco Peaks ratio = 2.8, minimum seedlings ha−1 = 170; Mogollon Rim ratio = 2.1, minimum seedlings ha−1 = 40; White Mountains ratio = 2.1, minimum seedlings ha−1 = 40; Mt Graham ratio = 3.1, minimum seedlings ha−1 = 290; Signal Peak ratio = 3.1, minimum seedlings ha−1 = 290 and Sacramento Mountains ratio = 2.5, minimum seedlings ha−1 = 100. Discussion SWWP regeneration was more prolific at the northwestern edge of the range in the San Francisco Peaks and along the Mogollon Rim. Density increases in SWWP and other more shade-tolerant species compared with historical reconstructions prior to fire exclusion have been documented in these areas before (Cocke et al., 2005; Huffman et al., 2015; Rodman et al., 2016). The northwestern distribution of SWWP has no WPBR and stands appear fully stocked, but these areas are also potential hybridization zones where SWWP and limber pine (P. flexilis) ranges overlap (Little, 1971; Steinhoff and Andresen, 1971; Menon et al., 2018). In other sites across the range, SWWP was in lower proportions compared with total regeneration, but not as rare as PP regeneration. This supports other surveys that have quantified SWWP as a minor understory component (generally <5% of total regeneration densities) in stands both pre- and post-treatment (Jones, 1971; Ffolliott and Gottfried, 1991; Looney and Waring, 2013). Co-occurrence with other species was a consistent predictor of SWWP regeneration in our study area, along with silviculture treatments, WPBR and other within-site factors related to ground cover and canopy openness. Site conditions appropriate for DF regeneration appear to be similar to those necessary for SWWP, and shade-tolerant regeneration and overstory BA were negatively related to SWWP seedling densities. SWWP pine occupies a range of community types from xeric, PP-dominated stands to more mesic communities, but appeared to regenerate best in open, drier, mixed conifer stands compared with higher elevation stands where more shade-tolerant species are prolific (Goodrich and Waring, 2017). Many factors can interact with and influence regeneration species and densities (Gray et al., 2005; Puhlick et al., 2012; Bataineh et al., 2013), making mixed species regeneration particularly difficult to predict (Fisichelli et al., 2013; Crotteau et al., 2014). Other studies on related five-needle pine species including P. albicaulis, P. monticola and P. strobus have identified source strength (overstory BA, number of cones), climate variables, light, soils (Dovciak et al., 2003; Maloney, 2014), elevation and aspect (Larson and Kipfmueller, 2010) as good predictors of regeneration. Our analyses suggest that understory and overstory environments were more important in predicting SWWP regeneration than stand-level abiotic factors such as climate variables, elevation or aspect. Fisichelli et al. (2013) hypothesized that mean climate may not explain regeneration variation well because more drastic, short term weather events (droughts, frost events, etc.) could be more important to establishing regeneration. Our findings suggest that management may help maintain sustainable densities in areas of the Southwest with severe WPBR, but stands should be monitored for regeneration and WPBR incidence following treatments. We took opportunities to survey stands where recent (<20 years) management had occurred, but comparisons between pre- and post-treatments could not be made. However, differences in current conditions between treated and non-treated stands shed some light on the potential interactions between management, regeneration and disease. Stands where silviculture treatments had been applied in the past 20 years had lower densities of SWWP regeneration compared with stands not recently-managed, but non-managed stands in high severity areas were not sustainable, with 30% of regeneration infected. These infection rates would offset the benefits of higher densities if regeneration dies before reaching the next size class. WPBR negatively affects SWWP seedling counts most likely through rapid mortality of small trees but the disease can also reduce cone crops in heavily infected stands, thus severing functioning regeneration cycles (McKinney and Tomback, 2007; Schoettle and Sniezko, 2007). The fungus that causes WPBR generally requires cool, humid conditions for spore development and germination, and relative humidity can be significantly reduced and temperatures increased following silviculture treatments (Ma et al., 2010). Open conditions are not usually recommended in managing WPBR because it may stimulate Ribes species production (the secondary host required in the life cycle of WPBR and a source of inoculum for the fungus) and can increase fungal spore dispersal throughout the stand (Tomback et al., 1995; Conklin et al., 2009; Zeglen et al., 2010). Site preparation and prescribed burning following silviculture treatments in high hazard sites may be needed to reduce Ribes densities and competing tree species (see Management recommendations Section). Areas where site preparation is needed (see Management recommendations Section) could be prioritized based on our current understanding of WPBR hazard in the Southwest. In the Sacramento Mountains of New Mexico (where WPBR has occurred the longest in southwestern forests), interacting site factors such as higher elevations, canyon bottoms or lower slope positions and presence/abundance of Ribes species are all are associated with higher WPBR hazard (relationships between site factors and WPBR have not been as well-defined in other parts of the Southwest) (Van Arsdel et al., 1998; Geils et al., 1999; Conklin, 2004; Looney et al., 2015). High WPBR infection levels across all diameter classes in non-managed stands indicates that these stands may eventually lose SWWP as a significant component, especially because the ratio between sapling-sized trees to the next size class was high in non-managed stands (Figure 5a). Avoiding drastic BA reduction through silviculture treatments that create open conditions for SWWP because Ribes abundance will increase should be weighed against the possibility of SWWP regeneration not progressing into the overstory because of closed conditions and WPBR infections. If potential WPBR hazard is considered too high on a specific site during pre-treatment surveys (i.e. low numbers of rust-free trees, many stem cankers, dead tops, high Ribes densities in the stand or in surrounding valley bottoms), planting genetically resistant seedlings could be considered instead of encouraging natural regeneration. Research on quantifying genetically resistant families and determining seed transfer models is ongoing (Waring et al., 2016). We lacked pre-treatment SWWP regeneration densities and WPBR information to compare to post-treatment data measured in this study. Because SWWP is a minor species component and often not the highest species preference (Conklin et al., 2009), pre-treatment data were not available in our study area. The mean age of seedlings (measured by whorl count and adjusted with a regression equation modified by growth ring counts, Goodrich and Waring, 2017) illustrated seedlings were stimulated by treatments in most two-aged stands surveyed (Goodrich and Waring, 2017; Supplementary data Table S1). Mean and maximum seedling ages (i.e. advanced regeneration) were older than most treatment dates in recently-managed UEA structures and oldest in non-managed stands (Goodrich and Waring, 2017;,Supplementary data Table S1). We acknowledge that saplings and some seedlings were likely retained during treatments and all SWWP densities were not solely the products of treatments. Seedling ages also limit inferences of WPBR infections in different stand structures, because disease takes time to develop. Seedlings in UEA and non-managed stands have been exposed to WPBR longer than seedlings in two-aged stands as they were older in general (Goodrich and Waring, 2017;Supplementary data Table S1). We were not able to sample similar stand numbers within WPBR severity or silviculture treatment levels within each mountain range due to the geographic distribution of the disease and differences in forest management across land ownership, which led to an unbalanced design. The majority of EA regeneration treatments we sampled were in lower elevation areas where WPBR severity was low, although we did survey recently-managed, two-aged stands in high rust hazard areas, indicated by high rust severity in overstory trees and previous data from permanent plots (Conklin, 2004;Supplementary data Table S1). We attempted to sample across a wide geographic area within sites with disease and included enough stands with high WPBR severity for valid comparisons (at minimum, three stands per disease/management level) to be made. The random effects of site and stand within-site were quantified in ZINB models, and while stand to stand variation was high, estimates and standard deviations of the random effect of site were low (Table 4). Management recommendations Our surveys suggest that young SWWP were stimulated in treatments that created open structures, were more prolific in areas where PP seedlings and DF seedlings and saplings were successful, and were less prevalent where shade-tolerant overstory and saplings dominate. We recommend stimulating or planting SWWP in local areas where DF and PP successfully regenerate. Complementary research indicates that SWWP seedlings also grow fastest in these open areas without shade-tolerant understory species (Goodrich and Waring, 2017). WPBR has not yet spread to many forests throughout the Southwest (Fairweather and Geils, 2011; Looney et al., 2015) and encouraging SWWP natural regeneration or outplanting genetically resistant seedlings follow proactive guidelines in place for other species (Schoettle, 2004; Schoettle and Sniezko, 2007; Conklin et al., 2009). Increasing natural regeneration can increase the genetic, age and size class diversities of a stand and possibly increase resilience following the introduction of WPBR (Waring and O’Hara, 2005; Schoettle and Sniekzo, 2007; Schwandt et al., 2010; Schoettle et al., 2012). Results from predictive models could be helpful in locating or creating the best sites to outplant genetically resistant SWWP seedlings (Schoettle and Sniezko, 2007; Sniekzo et al., 2011), especially in areas where WPBR severity or potential hazard is high (e.g. areas of high elevation, canyon bottoms, prolific Ribes populations) (Van Arsdel et al., 1998; Conklin et al., 2009). We recommend ~300 SWWP seedlings ha−1 are planted (or stimulated in a successful treatment) to safeguard against potential WPBR mortality. Current work is in progress to collect seeds and quantify the frequency of genetic resistance to C. ribicola across the range of SWWP (Conklin et al., 2009; Sniezko et al., 2011; Waring et al., 2016; Goodrich et al., in press), but until these populations and families are identified, stimulating natural, local regeneration may help preserve genetic diversity and increase resilience at the stand level. We recommend managers anticipate that WPBR will eventually reach most mountain ranges in Arizona and New Mexico. The open, two-aged stands in this study were created through establishment phase of shelterwood with reserve trees or patch clearcut with reserve tree regeneration systems. SWWP seedlings also grew fastest in these open stands (Goodrich and Waring, 2017). Eastern white pine, another shade mid-tolerant species, also grows well in stands with more light available if shade-intolerant understory regeneration is controlled, and shelterwood systems are applied to many eastern white pine-dominated stands (Fahey and Lorimer, 2013; Parker et al., 2013). High densities of shade-intolerant seedlings were measured in two-aged stands in this study, but densities of shade-intolerant saplings were low, indicating they were not reaching substantial sizes. A major concern regarding thinning or regeneration systems in areas with WPBR is that the extra light can increase the abundance of Ribes stems (Conklin et al., 2009; Zeglen et al., 2010). Although single-entry broadcast burning may stimulate seedbed Ribes germination and increase proliferation, repeated broadcast burning or another type of site preparation to kill Ribes stems may help limit densities and lower rust hazard. Prescribed burning and reintroducing surface fire also coincide with restoration goals to reintroduce fire back into southwestern mixed conifer forests, which historically had frequent fire regimes (Swetnam and Baisan, 1996; Brown et al., 2001; Huffman et al., 2015). It is possible that an increase in shade-intolerant species densities following treatments outcompeted the Ribes plants in two-aged stands we surveyed (e.g. sprouting species such as New Mexico locust and the abundance of oak species throughout the Southwest), and a lack of Ribes has been noted in intensively managed lower elevation areas of the Mescalero Apache Reservation in previous surveys (Van Arsdel et al., 1998). We recommend prioritizing silviculture treatments that drastically reduce BA in low hazard areas (stands with fewer current Ribes stems, lower elevations or non-canyon bottoms) (Van Arsdel et al., 1998; Geils et al., 1999; Conklin, 2004; Conklin et al., 2009) and evaluation and monitoring of SWWP regeneration in follow-up thinning treatments within open stands. Because the WPBR hazard of a site depends on time, an inoculum source and suitable environmental conditions for the fungus, it would be helpful to quantify whether shelterwood regeneration treatments across a range of elevations increase the WPBR hazard of a stand by monitoring Ribes densities, relative humidity and temperature data for several years pre- and post-treatments. We recommend that open conditions to stimulate SWWP regeneration are created in areas where SWWP exists as an overstory species through either two-aged or UEA regeneration methods that reduce total BA to 9–20 m2 ha−1 and retain high quality overstory SWWP without lethal WPBR infections (i.e. stem cankers). Modified single-tree selection treatments with target BA of 9−10 m2 ha−1 with various size classes left in groups around larger openings may provide similar open conditions to the two-aged structures where disease-free SWWP regeneration occurred in this study and SWWP seedlings tend to grow the fastest (Goodrich and Waring, 2017). This range of residual BA could meet desired conditions for resilient forests as long as sustainable size classes and appropriate spatial structure are maintained (Reynolds et al., 2013). Site preparation and frequent surface fire to reduce Ribes seeds and stems may be necessary, but other pioneer shade-intolerant species, such as oaks, may outcompete Ribes in some southwestern mixed conifer stands. Stimulating natural SWWP regeneration in open conditions in low WPBR severity areas and/or planting genetically resistant SWWP seedlings in high WPBR severity stands with open conditions should lead to faster growth and recruitment of SWWP into the overstory compared with closed canopy stands under both scenarios (Goodrich and Waring, 2017). Precommercial thinning and timber stand improvement should remove SWWP with stem cankers but leave sapling-sized SWWP if they are free of WPBR stem infections. Current recommendations that include favouring seed-bearing SWWP, retaining trees with good form and vigour, and ranking SWWP higher in species preference should be maintained, as they increase the probability that SWWP will continue to occupy the site, provide a seed source and maintain genetic diversity (Conklin et al., 2009). Supplementary data Supplementary data are available at Forestry online. Acknowledgements We thank K. Peterson, C. Lee-Wilkins, T. Largent, C. Harris and T. Steadman for assistance in the field. We are grateful to W. Hornsby, R. Campbell, T. Padilla, E. Enjady, R. Miller, the Mescalero Apache Tribe, the White Mountain Apache Tribe and the USDA Forest Service for assistance in gathering stand exam data, advice on site selections, information on forest treatments, providing housing and site access and supporting this research. Drs T. Kolb, P. Fule and T. Whitham provided reviews that improved earlier versions of this manuscript. Conflict of interest statement None declared. Funding McIntire-Stennis appropriations to the State of Arizona and Northern Arizona University, United States Department of Agriculture Forest Health Protection Evaluation Monitoring Program and the Northern Arizona University Genes to Environment Program through the Department of Biology. References Akaike , H. 1974 A new look at the statistical model identification . IEEE Trans. Autom. Control 19 , 716 – 723 . Google Scholar CrossRef Search ADS Auty , D. , Achim , A. , Bédard , P. and Pothier , D. 2014 StatSAW: modelling lumber product assortment using zero-inflated Poisson regression . Can. J. For. Res. 44 , 638 – 647 . Google Scholar CrossRef Search ADS Bataineh , M. , Kenefic , L. , Weiskittel , A. , Wagner , R. and Brissette , J. 2013 Influence of partial harvesting and site factors on the abundance and composition of natural regeneration in the Acadian Forest of Maine, USA . For. Ecol. Manag. 306 , 96 – 106 . Google Scholar CrossRef Search ADS Brown , P.M. , Kaye , M.W. , Huckaby , L.S. and Baisan , C.H. 2001 Fire history along environmental gradients in the Sacramento Mountains, New Mexico: influences of local patterns and regional processes . Ecoscience 8 , 115 – 126 . Google Scholar CrossRef Search ADS Burnham , K.P. and Anderson , D.R. 2002 Model selection and multimodel inference: a practical information-theoretic approach . Springer Science & Business Media . Conklin , D.A. 2004 Development of the white pine blister rust outbreak in New Mexico . USDA Forest Service , 17 . Southwestern Region, R3-04-01. Conklin , D.A. , Fairweather , M.L. , Ryerson , D.E. , Geils , B.W. and Vogler , D.R. 2009 White pines, blister rust, and management in the Southwest . USDA Forest Service , 16 . Southwestern Region, R3-FH-09-01. Crookston , N.L. and Rehfeldt , G.E. 2008 . Climate estimates and plant-climate relationships. USDA Forest Service, Rocky Mountain Research Station, Moscow, ID. Web Site: http://forest.moscowfsl.wsu.edu/climate/ (accessed March 16 2017). Crotteau , J.S. , Ritchie , M.W. and Varner , J.M. 2014 A mixed-effects heterogeneous negative binomial model for postfire conifer regeneration in northeastern California, USA . For. Sci. 60 , 275 – 287 . DeRose , R.J. and Long , J.N. 2014 Resistance and resilience: a conceptual framework for silviculture . For. Sci. 60 , 1205 – 1212 . Fahey , R.T. and Lorimer , C.G. 2013 Restoring a midtolerant pine species as a component of late-successional forests: results of gap-based planting trials . For. Ecol. Manag. 292 , 139 – 149 . Google Scholar CrossRef Search ADS Fairweather , M.L. and Geils , B.W. 2011 First report of the white pine blister rust pathogen, Cronartium ribicola, in Arizona . Plant Dis. 95 , 494 – 494 . Google Scholar CrossRef Search ADS Ffolliott , P.F. and Gottfried , G.J. 1991 Mixed conifer and aspen regeneration in small clearcuts within a partially harvested Arizona mixed conifer forest. Research Paper Rocky Mountain Forest and Range Experiment Station . USDA Forest Service . (RM-294). Fisichelli , N.A. , Frelich , L.E. and Reich , P.B. 2013 Climate and interrelated tree regeneration drivers in mixed temperate–boreal forests . Landscape Ecol. 28 , 149 – 159 . Google Scholar CrossRef Search ADS Geils , B.W. , Conklin , D.A. and Van Arsdel , E.P. 1999 A preliminary hazard model of white pine blister rust for the Sacramento Ranger District, Lincoln National Forest. Res. Note RMRS-RN-6. Fort Collins, CO: U.S. Department ofAgriculture, Forest Service, Rocky Mountain Research Station. 6 p. [Online]. Available: http://www.treesearch.fs.fed.us/pubs/4702 Goodrich , B.A. and Waring , K.M. 2017 Pinus strobiformis seedling growth in southwestern US mixed conifer forests in managed and non-managed stands . Forestry 90 , 393 – 403 . Google Scholar CrossRef Search ADS Goodrich , B.A. , Waring , K.M. and Flores-Renteria , L. Inpress Pinus strobiformis gene conservation and genecology. In: Schoettle, A.W., R.A. Sniezko, J. Kliejunas, eds. Genetics of Five-Needle Pines, Rusts of Forest Trees, and Strobusphere: Proceedings of the IUFRO Joint Conference. 15–20 June 2014; Fort Collins, CO. Proceedings RMRS-P-xxx (in press). Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station Gray , A.N. , Zald , H.S. , Kern , R.A. and North , M. 2005 Stand conditions associated with tree regeneration in Sierran mixed-conifer forests . For. Sci. 51 , 198 – 210 . Grissino-Mayer , H.D. , Baisan , C.H. and Swetnam , T.W. 1995 Fire history in the Pinaleño Mountains of Southeastern Arizona: effects of human-related disturbances. In Biodiversity and Management of the Madrean Archipelago: The Sky Islands of Southwestern United States and Northwester Mexico . Debano L.F. , Gottfried G.J. , Hamre R.H. , Edminster C.B. , Pfolliott P.F. and Ortega-Rubio A. (eds). USDA Forest Service Rocky Mountain Research Station , pp. 399 – 406 . Gen. Tech. Rep. RM-GTR 264. Huffman , D.W. , Zegler , T.J. and Fulé , P.Z. 2015 Fire history of a mixed conifer forest on the Mogollon Rim, northern Arizona, USA . Int. J. Wildland Fire 24 , 680 – 689 . Google Scholar CrossRef Search ADS Husch , B. , Beers , T.W. and Kershaw , J.A. 2002 Forest Mensuration . John Wiley & Sons Inc , p. 456 . Jones , J.R. 1971 Mixed conifer seedling growth in eastern Arizona. USDA For. Serv. Res. Pap. RM-77, 19 p. Keane , R.E. , Holsinger , L.M. , Mahalovich , M.F. and Tomback , D.F. 2017 Evaluating future success of whitebark pine ecosystem restoration under climate change using simulation modeling . Restoration Ecol. 25 , 220 – 233 . Google Scholar CrossRef Search ADS Keane , R.E. and Schoettle , A.W. 2011 Strategies, tools, and challenges for sustaining and restoring high elevation five-needle white pine forests in western North America. In: Keane, RE.; Tomback, D.F.; Murray, M.P. and Smith, C.M., eds. 2011. The Future of High-elevation, Five-needle White Pines in Western North America: Proceedings of the High Five Symposium. 28–30 June 2010; Missoula, MT. Proceedings RMRS-P-63. Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station. pp. 276 – 294 . Kearns , H.S. and Jacobi , W.R. 2007 The distribution and incidence of white pine blister rust in central and southeastern Wyoming and northern Colorado . Can. J. For. Res. 37 , 462 – 472 . Google Scholar CrossRef Search ADS Korb , J.E. , Fulé , P.Z. and Stoddard , M.T. 2012 Forest restoration in a surface fire-dependent ecosystem: an example from a mixed conifer forest, southwestern Colorado, USA . For. Ecol. Manage. 269 , 10 – 18 . Google Scholar CrossRef Search ADS Larson , E.R. and Kipfmueller , K.F. 2010 Patterns in whitebark pine regeneration and their relationships to biophysical site characteristics in southwest Montana, central Idaho, and Oregon, USA . Can. J. For. Res. 40 , 476 – 487 . Google Scholar CrossRef Search ADS Li , R. , Weiskittel , A.R. and Kershaw , J.A. , Jr. 2011 Modeling annualized occurrence, frequency, and composition of ingrowth using mixed-effects zero-inflated models and permanent plots in the Acadian Forest Region of North America . Can. J. For. Res. 41 , 2077 – 2089 . Google Scholar CrossRef Search ADS Little , E.L. , Jr. 1971 Atlas of United States Trees, Volume 1, Conifers and Important Hardwoods. United States Department of Agriculture Miscellaneous Publication 991. 9 p, 200 maps. Loo , J.A. 2009 Ecological impacts of non-indigenous invasive fungi as forest pathogens . Biol. Invasions 11 , 81 – 96 . Google Scholar CrossRef Search ADS Looney , C.E. and Waring , K.W. 2012 Patterns of forest structure, competition and regeneration in southwestern white pine (Pinus strobiformis) forests . For. Ecol. Manage. 286 , 159 – 170 . Google Scholar CrossRef Search ADS Looney , C.E. and Waring , K.W. 2013 Pinus strobiformis (southwestern white pine) stand dynamics, regeneration, and disturbance ecology: a review . For. Ecol. Manage. 287 , 90 – 102 . Google Scholar CrossRef Search ADS Looney , C.E. , Waring , K.W. and Fairweather , M.L. 2015 Monitoring the health of Pinus strobiformis: early impacts of white pine blister rust invasion. In: Potter, Kevin M.; Conkling, Barbara L. 2015. Forest Health Monitoring: National Status, Trends, and Analysis 2013. Gen. Tech. Rep. SRS-207. Asheville, NC: U.S. Department of Agriculture Forest Service, SouthernResearch Station. pp. 167 − 174 . Ma , S. , Concilio , A. , Oakley , B. , North , M. and Chen , J. 2010 Spatial variability in microclimate in a mixed-conifer forest before and after thinning and burning treatments . For. Ecol. Manage. 259 , 904 – 915 . Google Scholar CrossRef Search ADS Maloney , P.E. 2014 The multivariate underpinnings of recruitment for three Pinus species in montane forests of the Sierra Nevada, USA . Plant Ecol. 215 , 261 – 274 . Google Scholar CrossRef Search ADS Mattson , D.J. and Arundel , T.A. 2013 Consumption of seeds of southwestern white pine (Pinus strobiformis) by black bear (Ursus americanus) . Southwest. Naturalist 58 , 243 – 245 . Google Scholar CrossRef Search ADS McKinney , S.T. and Tomback , D.F. 2007 The influence of white pine blister rust on seed dispersal in whitebark pine . Can. J. For. Res. 37 , 1044 – 1057 . Google Scholar CrossRef Search ADS Menon , M. , Bagley , J.C. , Friedline , C.J. , Whipple , A.V. , Schoettle , A.W. , Leal‐Saenz , A. , et al. 2018 The role of hybridization during ecological divergence of southwestern white pine (Pinus strobiformis) and limber pine (P. flexilis) . Mol. Ecol. doi:10.1111/mec.14505 . Millar , C.I. , Stephenson , N.L. and Stephens , S.L. 2007 Climate change and forests of the future: managing in the face of uncertainty . Ecol. Appl. 17 , 2145 – 2151 . Google Scholar CrossRef Search ADS PubMed Millers , I. , Lachance , D. , Burkman , W.G. and Allen , D.C. 1991 North American sugar maple decline project: organization and field methods. USDA Forest Service Northeastern Forest Experiment Station, General Technical Report NE-154. 26p. Nyland , R.D. 1996 Silviculture: concepts and applications . McGraw-Hill . O’Hara , K.L. and Ramage , B.S. 2013 Silviculture in an uncertain world: utilizing multi-aged management systems to integrate disturbance . Forestry 86 , 401 – 410 . Google Scholar CrossRef Search ADS Paluch , J.G. 2005 The influence of the spatial pattern of trees on forest floor vegetation and silver fir (Abies alba Mill.) regeneration in uneven-aged forests . For. Ecol. Manage. 205 , 283 – 298 . Google Scholar CrossRef Search ADS Parker , A.J. 1982 The topographic relative moisture index: an approach to soil-moisture assessment in mountain terrain . Phys. Geogr. 3 , 160 – 168 . Parker , W.C. , Noland , T.L. and Morneault , A.E. 2013 Comparative mast seed production in unmanaged and shelterwood white pine (Pinus strobus L.) stands in central Ontario . New Forests 44 , 613 – 628 . Google Scholar CrossRef Search ADS Puhlick , J.J. , Laughlin , D.C. and Moore , M.M. 2012 Factors influencing ponderosa pine regeneration in the southwestern USA . For. Ecol. Manage. 264 , 10 – 19 . Google Scholar CrossRef Search ADS R Core Team . 2014 R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria Rehfeldt , G.E. 2006 A spline model of climate for the western United States. General Technical Report 165 . USDA Forest Service, Rocky Mountain Research Station . Reynolds , R.T. , Sánchez Meador , A.J.S. , Youtz , J.A. , Nicolet , T. , Matonis , M.S. , Jackson , P.L. , et al. . 2013 Restoring composition and structure in southwestern firequent-fire forests: A science-based framework for improving ecosystem resiliency . USDA Forest Service-General Technical Report RMRS-GTR, (310 RMRS-GTR), 76pp. Ridout , M. , Hinde , J. and DeméAtrio , C.G. 2001 A score test for testing a zero‐inflated poisson regression model against zero‐inflated negative binomial alternatives . Biometrics 57 , 219 – 223 . Google Scholar CrossRef Search ADS PubMed Roberts , D.W. 1986 Ordination on the basis of fuzzy set theory . Vegetatio 66 , 123 – 131 . Google Scholar CrossRef Search ADS Rodman , K.C. , Sánchez Meador , A.J. , Huffman , D.W. and Waring , K.M. 2016 Reference conditions and historical fine-scale spatial dynamics in a southwestern mixed-conifer forest, Arizona, USA . For. Sci. 62 , 268 – 280 . Samano , S. and Tomback , D.F. 2003 Cone opening phenology, seed dispersal, and seed predation in southwestern white pine (Pinus strobiformis) in southern Colorado . Ecoscience 10 , 319 – 326 . Google Scholar CrossRef Search ADS Schoettle , A.W. 2004 Ecological roles of five-needle pines in Colorado: potential consequences of their loss. Pages 124–135 In: Breeding and Genetic Resources of Five-Needle Pines: Growth, Adaptability, and Pest Resistance. Proc. RMRS-P-32. R. A. Sniezko, S. Samman, S. E. Schlarbaum, and H. B. Kriebel, eds. United States Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, CO. Schoettle , A.W. , Klutsch , J.G. and Sniezko , R.A. 2012 Integrating regeneration, genetic resistance, and timing of intervention for the longterm sustainability of ecosystems challenged by non-native pests—a novel proactive approach. Pages 112–1 23 in: Proc. Fourth Int. Workshop Genet. Host–Parasite Interact. For. Disease and Insect Resistance in Forest Trees, Eugene, OR. Gen. Tech. Rep. PSW-GTR-240. R. A. Sniezko, A. D. Yanchuk, J. T. Kliejunas, K. M. Palmieri, J. M. Alexander, S. J. Frankel, tech. coord. United States Department of Agriculture, Forest Service, Pacific Southwest Research Station, Albany, CA. Schoettle , A.W. and Sniezko , R.A. 2007 Proactive intervention to sustain high elevation pine ecosystems threatened by white pine blister rust . J. For. Res. 12 , 327 – 336 . Google Scholar CrossRef Search ADS Schwandt , J.W. , Lockman , I.B. , Kliejunas , J.T. and Muir , J.A. 2010 Current health issues and management strategies for white pines in the western United States and Canada . For. Pathol. 40 , 226 – 250 . Google Scholar CrossRef Search ADS Sekhon , J.S. 2011 Multivariate and propensity score matching software with automated balance optimization: the matching package for R . J. Stat. Softw. 42 , 1 – 52 . Google Scholar CrossRef Search ADS Shirk , A.J. , Cushman , S.A. , Waring , K.M. , Wehenkel , C.A. , Leal-Sáenz , A. , Toney , C. , et al. 2018 Southwestern white pine (Pinus strobiformis) species distribution models project a large range shift and contraction due to regional climatic changes . For. Ecol. Manage 411 , 176 – 186 . Google Scholar CrossRef Search ADS Skaug , H. , Fournier , D. , Nielsen , A. , Magnusson , A. and Bolker , B. 2011 glmmADMB: Generalized Linear Mixed Models Using AD Model Builder. R Package Version 0.6, 5, r143 Sniezko , R.A. , Mahalovich , M.F. , Schoettle , A.W. and Vogler , D.R. 2011 Past and current investigations of the genetic resistance to Cronartium ribicola in high-elevation five-needle Pines. In: Keane, Robert E.; Tomback, Diana F.; Murray, Michael P.; and Smith, Cyndi M., eds. 2011. The Future Of High-Elevation, Five-Needle White Pines in Western North America: Proceedings of the High Five Symposium. 28–30 June 2010; Missoula, MT. Proceedings RMRS-P-63. Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station. pp. 246 – 264 . Spiess , A.-N. 2014 qpcR: Modelling and Analysis of Real-Time PCR Data. R Package Version 1.4-0. https://CRAN.R-project.org/package=qpcR Steinhoff , R.J. and Andresen , J.W. 1971 Geographic variation in Pinus flexilis and Pinus strobiformis and its bearing on their taxonomic status . Silvae Genet. 20 , 159 – 167 . Stoddard , M.T. , Meador , A.S. , Fulé , P.Z. and Korb , J.E. 2015 Five-year post-restoration conditions and simulated climate-change trajectories in a warm/dry mixed-conifer forest, southwestern Colorado, USA . For. Ecol. Manage. 356 , 253 – 261 . Google Scholar CrossRef Search ADS Swetnam , T. and Baisan , C. 1996 Historical fire regime patterns in the southwestern United States since AD 1700. In: C.D. Allen (ed) Fire Effects in Southwestern Forests: Proceedings of the 2nd La Mesa Fire Symposium, pp. 11 – 32 . USDA Forest Service, Rocky Mountain Research Station, General Technical Report RM-GTR-286. Tomback , D.F. and Achuff , P. 2010 Blister rust and western forest biodiversity: ecology, values and outlook for white pines . For. Pathol. 40 , 186 – 225 . Google Scholar CrossRef Search ADS Tomback , D.F. , Clary , J.K. , Koehler , J. , Hoff , R.J. and Arno , S.F. 1995 The effects of blister rust on post‐fire regeneration of whitebark pine: the Sundance burn of northern Idaho (USA) . Conserv. Biol. 9 , 654 – 664 . Google Scholar CrossRef Search ADS United States Geological Survey . 2006 Digital Representations of Tree Species Range Maps, from Little, E.L., Jr., 1971, Atlas of United States Trees, Conifers and Important Hardwoods, vol. 1. US Department of Agriculture Miscellaneous Publication 1146, 9 p. Van Arsdel , E.P. , Conklin , D.A. , Popp , J.B. and Geils , B.W. 1998 The distribution of white pine blister rust in the Sacramento Mountains of New Mexico. In: Jalkanen R, Crane PE, Walla JA, Aalto T (eds) Proceedings of the 1st IUFRO Rusts of Forest Trees Working Party Conference; 1998 August 2–7; Saariselkä, Finland. Research Paper 712. Finnish Forest Research Institute, Rovaniemi, Finland Waring , K. , Cushman , S. , Eckert , A. , Flores Renteria , L. , Sniezko , R. , Still , S. , et al. 2016 Blending Ecology and Evolution using Emerging Technologies to Determine Species Distributions with a Non-native Pathogen in a Changing Climate. Macrosystems Biology PI Meeting, September 28–30, 2016. Washington, DC. Waring , K.M. and O’Hara , K.L. 2005 Silvicultural strategies in forest ecosystems affected by introduced pests . For. Ecol. Manage. 209 , 27 – 41 . Google Scholar CrossRef Search ADS Zald , H.S. , Gray , A.N. , North , M. and Kern , R.A. 2008 Initial tree regeneration responses to fire and thinning treatments in a Sierra Nevada mixed-conifer forest, USA . For. Ecol. Manage. 256 , 168 – 179 . Google Scholar CrossRef Search ADS Zeglen , S. , Pronos , J. and Merler , H. 2010 Silvicultural management of white pines in western North America . For. Pathol. 40 , 347 – 368 . Google Scholar CrossRef Search ADS © Institute of Chartered Foresters, 2018. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Forestry: An International Journal Of Forest Research Oxford University Press

Interactions of management and white pine blister rust on Pinus strobiformis regeneration abundance in southwestern USA

Loading next page...
 
/lp/ou_press/interactions-of-management-and-white-pine-blister-rust-on-pinus-8f8QwCPQeW
Publisher
Institute of Chartered Foresters
Copyright
© Institute of Chartered Foresters, 2018. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com.
ISSN
0015-752X
eISSN
1464-3626
D.O.I.
10.1093/forestry/cpy009
Publisher site
See Article on Publisher Site

Abstract

Abstract Modelling natural regeneration is complex, and both natural and anthropogenic disturbances can alter forest trajectories. Pinus strobiformis (southwestern white pine, SWWP) is an important component of mixed conifer forests in the Southwest and management recommendations related to natural and planted regenerations are needed to guide conservation of SWWP in the face of an invasive disease (white pine blister rust, WPBR). Regeneration was surveyed across six mountain ranges, three silviculture treatments and two levels of disease severity in the Southwest US. Key findings were: (1) SWWP regeneration in stands with no recent management (<20 years) and high disease severity had unsustainable WPBR infection, (2) SWWP regeneration was less abundant but less likely to be infected in stands with recent management, (3) stands with high disease severity had fewer SWWP seedlings than stands with no or low disease severity and (4) SWWP regeneration densities were best predicted by other understory species abundance. We recommend silviculture treatments that reduce basal area to 9–10 m2 ha−1 and leave large canopy openings to enhance natural SWWP regeneration. Without creating conditions for disease-free regeneration to reach reproductive maturity, some stands may lose SWWP as an overstory component. Results may help refine SWWP management guidelines and expand conservation efforts in forests threatened by WPBR. Introduction Natural regeneration is a complex, stochastic process and understanding the dynamics associated with tree regeneration is critical to forest sustainability (Nyland, 1996; Puhlick et al., 2012; Bataineh et al., 2013). Regeneration abundance can be difficult to predict because it may be directly or indirectly influenced by abiotic factors, biotic factors or interactions between both (Puhlick et al., 2012; Fisichelli et al., 2013). Abiotic factors such as geology, soils, climate and fire influence the potential for germination, establishment and survival of regeneration, and also influence the current biotic conditions of overstory and understory species composition and structure (Paluch, 2005; Puhlick et al., 2012; Fisichelli et al., 2013). Biotic disturbances, such as disease or insect outbreaks, or anthropogenic disturbances including previous silviculture treatments, can alter the trajectory of stand dynamics by creating establishment sites and increasing resource availability to regeneration (Gray et al., 2005; Waring and O’Hara, 2005; Bataineh et al., 2013). However, disease or insect outbreaks may also cause infection and mortality to regeneration, so the affected species may not be able to take advantage of increased resources. Currently, forest structure in many dry mixed conifer forests of the Western US is considered dense and overcrowded with novel species mixtures due to disruption of historical fire regimes and intense logging practices (Gray et al., 2005; Korb et al., 2012; Rodman et al., 2016). Silviculture treatments (reducing basal area (BA) through regeneration treatments or intermediate treatments, such as thinning) may be effective tools for successful regeneration of desired species (e.g. Stoddard et al., 2015) but disease and insect disturbances can alter desired outcomes of such treatments and should be considered in order to create the most resilient future forest structures (Waring and O’Hara, 2005; Zald et al., 2008). Anticipating changes, realignment of disturbed systems and promoting resilience are all important tools in adaptive management in the face of invasive pests and a changing climate (Waring and O’Hara, 2005; Millar et al., 2007; O’Hara and Ramage, 2013; DeRose and Long, 2014). More effective silviculture treatments may be developed when primary abiotic and biotic factors, and regulatory patterns that govern natural regeneration success have been identified. Understanding the baseline ecology of natural regeneration is essential when species sustainability is threatened by an invasive pest (Schoettle and Sniezko, 2007; Loo, 2009; Goodrich and Waring, 2017). Southwestern white pine (Pinus strobiformis Engelm., SWWP) is a five-needle pine species that occurs across mixed conifer forests throughout Mexico and more disjunctly in the Southwest US (Arizona, New Mexico, southwestern Colorado and western Texas) (Little, 1971; Steinhoff and Andresen, 1971; Looney and Waring, 2013). Like all North American five-needle pine species, SWWP is susceptible to infection by Cronartium ribicola (J.C. Fisch.), the invasive fungal pathogen that causes white pine blister rust (WPBR). WPBR is a canker-causing disease that can lead to branch dieback, topkill and mortality of all sizes of SWWP (Conklin et al., 2009; Looney et al., 2015). Although SWWP is generally a minor species component in southwestern US, this species contributes to biodiversity and ecosystem resiliency as a food and habitat resource for wildlife and through the ability of mature trees to survive fire (Grissino-Mayer et al., 1995; Samano and Tomback, 2003; Mattson and Arundel, 2013). High rates of WPBR mortality, delayed forest recovery and loss of ecosystem services in other North America five-needle pine systems threatened by WPBR preview what may occur in some Southwest ecosystems with no proactive conservation and management strategies in place (Schoettle, 2004; Schoettle and Sniezko, 2007; Tomback and Achuff, 2010; Keane and Schoettle, 2011). Conservation strategies for other five-needle pines include mechanical cutting and prescribed burning to reduce competition and enhance natural regeneration and promoting rust resistance through planting genetically resistant seedlings (Schoettle and Sniezko, 2007; Keane and Schoettle, 2011; Schoettle et al., 2012; Keane et al., 2017). Genetic resistance to C. ribicola has been found in SWWP and current work is in progress to quantify the frequency of resistance across the SWWP range (Conklin et al., 2009; Sniezko et al., 2011; Waring et al., 2017; Goodrich et al., in press). The suitable distribution range for SWWP is expected to contract across southwestern US by 2090 in most future climate predictions (Crookston and Rehfeldt, 2008; Shirk et al., 2018), furthering the need to enhance resiliency in SWWP forests. Silviculture treatments best suited to stimulate and encourage SWWP regeneration are not well-defined, in part because we lack knowledge of the species’ ability to grow and compete under different light regimes, amongst different species and across various stand conditions. Retaining overstory SWWP in silviculture prescriptions to increase SWWP regeneration is recommended, as is removing competing vegetation in intermediate treatments (Conklin et al., 2009; Looney and Waring, 2012). Recent work illustrates that SWWP seedlings grow most quickly in open conditions created by regeneration treatments (Goodrich and Waring, 2017). There is a lack of published SWWP regeneration density data across both managed and non-managed southwestern US forests (Jones, 1971; Looney and Waring, 2013, but see Ffolliott and Gottfried, 1991; Looney and Waring, 2012; Goodrich and Waring, 2017). A gradient of WPBR incidence exists across southwestern mixed conifer forests (Conklin, 2004; Fairweather and Geils, 2011; Looney et al., 2015), creating opportunities to study SWWP regeneration dynamics with both the presence and absence of the disease. Our overarching goal was to survey for regeneration within non-managed stands and recently (<20 years) managed stands with a range of previous silviculture treatments and WPBR severities in order to provide better management recommendations to sustain SWWP through enhanced regeneration success. Results may be helpful in guiding conservation and management strategies of SWWP in the face of an invasive pathogen by providing direction on site selection and prescriptions for a variety of objectives (e.g. intermediate treatments, regeneration treatments and seedling outplanting efforts) when restoration is warranted (Schoettle and Sniezko, 2007; Keane and Schoettle, 2011). Our research questions were the following: What are average overstory and regeneration densities and proportions of SWWP compared with co-occurring species across six mountain ranges in southwestern US? What are average regeneration densities of SWWP across past silviculture treatments and two levels of WPBR severity? What between- and within-stand variables are associated with SWWP regeneration abundance? Do differences in SWWP size class distributions occur across silviculture treatments or WPBR severity levels? Methods Study area We sampled mixed conifer forests with SWWP as a substantial overstory component across Arizona and New Mexico National Forest and Native American tribal land in 2012–13 (Figure 1; Goodrich and Waring, 2017). Our surveys included National Forests (NF) land, the Fort Apache Indian Reservation (owned by the White Mountain Apache tribe) in Arizona and the Mescalero Apache Indian Reservation (owned by the Mescalero Apache tribe) in New Mexico. We grouped stands by the mountain ranges (hereafter referred to as sites) in which they occurred: San Francisco Peaks (n = 3 stands), Mogollon Rim (n = 6), White Mountains (n = 11), Mt Graham (n = 4), Signal Mountain (n = 3) and Sacramento Mountains (n = 28) (Figure 1). Our sampling represented a range of topographic relative moisture index (TRMI) values, a scalar index ranging from 0 (xeric) to 60 (mesic) and intended to represent site moisture characteristics by combining aspect, slope and topographic variables (Parker, 1982). Figure 1 View largeDownload slide Fifty-five stands sampled for regeneration across six mountain ranges (sites) in Arizona and New Mexico. White pine distribution layer credit of United States Geological Survey (2006) based on Little (1971). WPBR = white pine blister rust. Figure 1 View largeDownload slide Fifty-five stands sampled for regeneration across six mountain ranges (sites) in Arizona and New Mexico. White pine distribution layer credit of United States Geological Survey (2006) based on Little (1971). WPBR = white pine blister rust. Silviculture treatments and WPBR severity We sampled stands across two current WPBR severity levels and three silviculture treatment categories within ranges of TRMI values. Within categories of low (10–<25), medium (>25–<40) and high (>40−55) TRMI values we attempted to sample two levels of WPBR severity (low and high, defined below), and include at least three stands within each of the following levels of silviculture treatments (representing a gradient of residual BA from high to low): (1) stands with no recent silviculture treatments (no tree removal or prescribed fires) in the past 20 years, (2) recent (<20 years prior) uneven-aged (UEA) regeneration treatments or timber stand improvement treatments and (3) recent (<20 years prior) even-aged (EA) regeneration treatments. UEA structures were generally created through single-tree selection regeneration methods, thinning or a fuels treatment (Supplementary data Table S1). EA structures were created through establishment phase of shelterwood with reserves or clearcut with reserves regeneration methods, leaving two-aged residual structures (Supplementary data Table S1). EA regeneration methods were only found on the Mescalero Apache Indian Reservation in the Sacramento Mountains. We sampled a minimum of three stands per silviculture treatment in every TRMI category with the exception of high TRMI (>40−55), where only two recently-managed UEA stands and one recently-managed two-aged stand were sampled. Stands with high WPBR severity were defined as those with >10% stand incidence of WPBR (percent of overstory trees infected with WPBR, range: 0–88%) and at least 5% mean stand severity (crown dieback due to WPBR cankers, range: 0–25.8%) of infected trees. Stands with no WPBR and low severity WPBR were grouped together because early effects of the disease should not affect overstory or understory composition. Five percent mean severity was selected as the cutoff point where changes in overstory crown conditions could influence understory conditions and has been used as a cutoff between healthy and declining health at the individual tree level in previous research (Kearns and Jacobi, 2007). There were no pre-treatment SWWP regeneration density data or WPBR severity data available for any surveyed stands. Stand selection and plot sampling Stands were defined by stand delineation boundaries on NF lands or treatment boundaries on tribal lands. We followed stand selection methods defined in Looney and Waring (2012) and described in Goodrich and Waring (2017). In brief, we selected stands to scout for suitability by limiting our queries to NF stand exam data that contained ≥6.9 m2 ha−1 BA of SWWP. Permanent plot data were used as scouting information for stands on the Fort Apache Reservation (Looney and Waring, 2012) and local knowledge was used to identify potential reconnaissance stands on the Mescalero Apache Indian Reservation (William Hornsby, pers. comm., Bureau of Indian Affairs Branch of Forestry, 2012–13) where stand exam data were unavailable. Within all potential stands, a random point was generated within the stand boundaries and located in the field using Global Positioning System (GPS). Plots (50 m × 20 m, 0.1 ha) were established if a minimum SWWP overstory density occurred and randomly relocated within the stand if adequate densities did not occur (see Goodrich and Waring, 2017 for details). Fifty-five stands were surveyed using a combination of 0.1 ha rectangular plots and transects with three 0.0016 ha plots per transect using the same methods as described in Goodrich and Waring (2017). Each 0.1 ha rectangular plot was oriented parallel to slope contour and size corrected for slope. For all overstory trees, we recorded species, diameter at breast height (DBH) and condition (live or dead). Tree height, height to base of live crown, crown widths and damage agents were measured on all overstory SWWP stems. All SWWP were visually inspected using binoculars for signs or symptoms (e.g. aecia, branch flagging, branch swelling, squirrel chewing or roughened bark) of WPBR (Looney et al., 2015). Crown dieback due to WPBR was recorded by tracing tree crowns and dieback (branch flagging and recently killed branches) on transparencies and later estimated using Adobe Photoshop CS5.1 (Millers et al., 1991). If a tree had ≤2 branch cankers, an estimate of dieback was visually made, bypassing the transparency sketch (17/83 trees with visual crown dieback estimated). Additional site characteristics measured at each stand included GPS coordinates, percent slope, aspect and topographic and slope positions. Northness (−1 to 1 where −1 represents south and 1 represents north) and eastness (−1 to 1, west to east, respectively) were calculated from aspect (Roberts, 1986). Derived normalized 30-year mean (1961–90) climate variables were estimated from stand coordinates and elevations using thin plate spline surfaces of Rehfeldt (2006). Three 0.01 ha (10 m × 10 m) subplots were nested within the 0.1 ha plot located at distances of 0–10, 20–30 and 40–50 m above or below the centre line of the rectangular plot. All seedlings (<140 cm height) and saplings (≥140 cm height and ≤12.75 cm DBH) were measured for species, height, crown width and condition in each subplot. Within each subplot a line intercept technique (Husch, 2002) was used by placing a ground cover transect across the diagonal of the plot and ground cover type (i.e. grass, forb, shrub, litter, rock, bare, other) was recorded every 0.3 m. Canopy closure was averaged over four cardinal directions at each subplot centre with a spherical densiometer held at 1.2 m above the ground. A random azimuth and distance from the plot corner were used to start an additional 50 m transect with three additional 4 m × 4 m (0.0016 ha) regeneration subplots at 0, 25 and 50 m. A variable radius plot using an angle gauge with a BA factor of 2.3 m2 ha−1 was used to record overstory species BA in each subplot. All overstory and understory measurements were similar to those made in the 0.1-ha rectangular plot in the first transect, but overstory SWWP stems were not tagged and severity was not estimated with transparencies. In each subplot ground cover was estimated using a 1m × 1 m square and crown closure was estimated using a spherical densitometer at 1.2 m above the ground at subplot centre. Statistical analyses All statistical analyses were completed in R statistical programing environment (v. 3.1.1, R Core Team, 2014). We used Kruskal–Wallis (K–W) one-way analysis of variance by ranks tests, which are non-parametric tests to assess whether independent samples originate from the same distribution, to compare regeneration densities and probabilities across sites (n = 6 sites), silviculture treatment and WPBR severity levels (n = 3 levels of silviculture treatment and two levels of WPBR severity). Kendall’s τ correlation analyses were used to compare SWWP seedling and sapling densities to between- and within-stand variables. Subplot-level densities were compared with subplot-level factors (e.g. other species densities, BA and percent canopy closure) and mean stand densities relationship with between-stand variables (e.g. climate variables, percent slope and aspect) were assessed. Because of the wide geographic area that was sampled, common species varied between stands. Regeneration (with seedlings and saplings separated as they may occupy different niche spaces) and overstory species were grouped into (1) common shade-tolerant species: white fir (Abies concolor (Gord. & Glend.) Lindl. Ex Hildebr), corkbark fir (Abies lasiocarpa (Hook.) Nutt. var. arizonica (Merriam) Lemmon), Engelmann spruce (Picea engelmannii Parry ex Engelm.) and blue spruce (Picea pungens Engelm.) and (2) common shade-intolerant species: trembling aspen (Populus tremuloides Michx.), Gambel or other oak species (Quercus gambelii Nutt. or various species of Quercus), alligator juniper (Juniperus deppeana Steud.), Utah juniper (Juniperus osteosperma), Rocky Mountain juniper (Juniperus scopulorum Sarg.) and New Mexico locust (Robinia neomexicana A. Gray). SWWP were neither grouped nor were Douglas-fir (DF; Pseudotsuga menziesii (Mirb.) Franco) and ponderosa pine (PP; Pinus ponderosa Dougl. ex Laws) due to their common occurrence across our study areas. Hierarchical, mixed-effects models were used to compare SWWP regeneration densities across silviculture and WPBR severity levels and build predictive models. Normally, discrete count data are modelled using Poisson regression, so we modelled our seedling and sapling data as stem counts per subplot. Counts within the three smaller subplots were scaled to 0.01 ha counts and the three nested 0.01 ha subplots were averaged (four total subplots per stand). Regeneration count data often exhibit overdispersion (i.e. the variance is greater than the mean) and in such cases a negative binomial (NB) model is more appropriate where the mean and variance can be independently estimated (Ridout et al., 2001; Li et al., 2011; Crotteau et al., 2014). We confirmed SWWP seedling and sapling data were overdispersed by using the ‘dispersiontest’ function in package ‘AER’ (Supplementary data Table S2). Regeneration data also tend to contain more zeros than would be expected under a normal Poisson or NB distribution (zero-inflation) (Crotteau et al., 2014). We confirmed SWWP seedling and sapling data were zero-inflated by comparing the observed and expected proportions of zeros in each data set under a Poisson distribution (Supplementary data Table S2). Zero-inflation is typically accounted for by jointly modelling the zero/non-zero component (using logistic regression) and count component (which can include zeros under a Poisson or NB distribution) (Li et al., 2011). The zero-inflated NB (ZINB) distribution provides flexibility in the variance assumptions and accounts for overdispersion, so we modelled the data with a ZINB probability mass function (Supplementary data equations (S1) and (S2)). Because our data were hierarchical, with subplots nested within stands and stands clustered within sites, we used a mixed-effects modelling approach, which allows for the inclusion of both fixed and random effects (Supplementary data equation (S3)). Models included nested random effects of ‘site’ and ‘stand within-site’ and were fitted using the ‘glmmADMB’ package in R (Skaug et al., 2011). To answer our research questions, we developed two ZINB models for both SWWP seedling and sapling counts that included the following explanatory variables: Silviculture treatment, WPBR severity and an interaction term included as fixed effects (research question 2). Silviculture treatment, WPBR severity, an interaction term and between- and within-stand variables included as fixed effects and reduced to retain best models (research question 3). We chose other potential between- and within-stand variables if they were significant (P ≤ 0.05) in the Kendall’s τ correlation analyses. We then took all potential covariates and reduced them further by viewing all possible ZINB univariate models and only keeping variables that were significantly related to SWWP seedling or sapling counts at P ≤ 0.10. We then built predictive ZINB models by modelling all combinations of the significant variables and compared the models with Akaike’s information criteria (AIC) (Akaike, 1974), the change in AIC value for each model relative to the best model (∆AIC) and Akaike’s weights (AICw) (‘qpcR’ package, akaike.weights function, Spiess, 2014) (Burnham and Anderson, 2002). We chose the best candidate model as the model with the lowest AIC and in which all covariates were significant at P ≤ 0.05. We assessed the predictive ability of the top candidate models by examining plots of observed and predicted seedling and sapling frequencies (Li et al., 2011; Auty et al., 2014) and calculated the root mean square error (RMSE) from the predicted and observed values for each candidate model. Kolmogorov–Smirnov (K–S) bootstrap tests (1000 replicates) were used to compare SWWP diameter distributions (in 10-cm DBH classes) between WPBR severities within each silviculture treatment level (‘Matching’ package, ks.boot function, Sekhon, 2011). We also compared distributions of WPBR-infected SWWP between silviculture treatments in high WPBR severity stands with K–S tests. We used diameter distributions to calculate ratios of each diameter size class to the next largest size class. The average ratio across size classes in none/low WPBR and non-managed stands within each site was used to calculate the number of seedlings needed to regenerate a stand if one tree per hectare (TPH) of SWWP was desired in the largest size class (≥62.7 cm DBH). The number of seedlings needed in high WPBR stands was calculated by taking this value and dividing it by the average incidence of WPBR-infected TPH to account for WPBR-induced mortality. Results SWWP seedlings occurred in 86% and SWWP saplings occurred in 84% of the 55 stands sampled, and densities ranged from 0 to 3 931 seedlings ha−1 and from 0 to 1 304 saplings ha−1 per stand. There were differences in mean ranks for SWWP seedling densities but not SWWP sapling densities across sites (Table 1 and Figure 2a,c). The San Francisco Peaks had the highest SWWP seedling and sapling densities (Figure 2a,c). The proportion of SWWP seedlings to total seedling densities varied across sites (Table 1) and ranged from 2.7% in the Sacramento Mountains to 42.5% on the San Francisco Peaks (Figure 2a). The ratio of SWWP saplings to total sapling densities also varied widely, from 7.1% on Mount Graham to 60.6% on the San Francisco Peaks (Figure 2c), although mean ranks were not significantly different (Table 1). Mean ranks did not differ between sites for DF seedlings or saplings, but there were site-to-site differences in PP seedlings, shade-intolerant seedlings/suckers and shade-tolerant seedlings and saplings (Table 1 and Figure 2a,c). Mean overstory BA ranks differed between sites for live SWWP, dead SWWP, live DF, live PP, live shade-tolerant and live shade-intolerant species (Table 1 and Figure 3a). In general, the San Francisco Peaks had the highest SWWP live BA and Mount Graham had the highest total BA (Table 1 and Figure 3a). The proportion of SWWP BA significantly varied across sites (Table 1) and ranged from 16.3% on Signal Peak to 50.8% on the San Francisco Peaks (Figure 3a). Mean rank differences of some between-stand variables occurred among sites, including elevation, climate variables (mean annual temperature, mean annual precipitation and the proportion of mean annual precipitation that occurs in July, August and September) and percent canopy closure (Supplementary data Table S3). Table 1 Kruskal–Wallis’ (K–W) Chi-squared H statistics, degrees of freedom (df) and P-values for common species regeneration densities (trees per hectare, TPH), proportions and overstory basal areas (BA, m2 ha−1) amongst six sites and six silviculture treatment/WPBR disease severity levels. Kruskal–Wallis’ H statistics significant at P ≤ 0.05 (bolded P-values) indicate differences in the distribution of ranks. Kruskal–Wallis mean ranks by site (n = 6) Kruskal–Wallis mean ranks in silviculture treatment/disease severity levels (n = 6) K–W Chi-squared P-value (df = 5) K–W Chi-squared P-value (df = 5) SWWPa seedlings 18.79 0.002 9.440 0.093 SWWP saplings 5.69 0.337 10.420 0.064 SWWP seedlings/total seedlings 23.95 <0.0001 9.370 0.100 SWWP saplings/total saplings 8.03 0.155 10.970 0.052 DFb seedlings 6.38 0.271 16.540 0.005 DF saplings 2.18 0.824 3.500 0.623 PPc seedlings 34.31 <0.0001 20.770 0.001 PP saplings 7.82 0.166 2.680 0.750 Shade-intolerant seedlings 63.66 <0.001 24.280 0.0002 Shade-intolerant saplings 10.51 0.062 9.180 0.102 Shade-tolerant seedlings 28.02 <0.001 8.440 0.134 Shade-tolerant saplings 23.63 0.0002 2.350 0.799 Live SWWP BA 22.91 0.0003 15.240 0.009 Dead SWWP BA 29.87 <0.0001 11.830 0.037 Live SWWP BA/Live total BA 13.37 0.020 11.84 0.037 Live DF BA 16.77 0.005 57.230 0.0001 Live PP BA 76.36 <0.0001 24.730 0.0002 Live shade-intolerant BA 84.75 <0.0001 25.370 0.0001 Live shade-tolerant BA 41.71 <0.0001 6.530 0.258 Kruskal–Wallis mean ranks by site (n = 6) Kruskal–Wallis mean ranks in silviculture treatment/disease severity levels (n = 6) K–W Chi-squared P-value (df = 5) K–W Chi-squared P-value (df = 5) SWWPa seedlings 18.79 0.002 9.440 0.093 SWWP saplings 5.69 0.337 10.420 0.064 SWWP seedlings/total seedlings 23.95 <0.0001 9.370 0.100 SWWP saplings/total saplings 8.03 0.155 10.970 0.052 DFb seedlings 6.38 0.271 16.540 0.005 DF saplings 2.18 0.824 3.500 0.623 PPc seedlings 34.31 <0.0001 20.770 0.001 PP saplings 7.82 0.166 2.680 0.750 Shade-intolerant seedlings 63.66 <0.001 24.280 0.0002 Shade-intolerant saplings 10.51 0.062 9.180 0.102 Shade-tolerant seedlings 28.02 <0.001 8.440 0.134 Shade-tolerant saplings 23.63 0.0002 2.350 0.799 Live SWWP BA 22.91 0.0003 15.240 0.009 Dead SWWP BA 29.87 <0.0001 11.830 0.037 Live SWWP BA/Live total BA 13.37 0.020 11.84 0.037 Live DF BA 16.77 0.005 57.230 0.0001 Live PP BA 76.36 <0.0001 24.730 0.0002 Live shade-intolerant BA 84.75 <0.0001 25.370 0.0001 Live shade-tolerant BA 41.71 <0.0001 6.530 0.258 aSWWP = southwestern white pine. bDF = Douglas-fir. cPP = ponderosa pine. Table 1 Kruskal–Wallis’ (K–W) Chi-squared H statistics, degrees of freedom (df) and P-values for common species regeneration densities (trees per hectare, TPH), proportions and overstory basal areas (BA, m2 ha−1) amongst six sites and six silviculture treatment/WPBR disease severity levels. Kruskal–Wallis’ H statistics significant at P ≤ 0.05 (bolded P-values) indicate differences in the distribution of ranks. Kruskal–Wallis mean ranks by site (n = 6) Kruskal–Wallis mean ranks in silviculture treatment/disease severity levels (n = 6) K–W Chi-squared P-value (df = 5) K–W Chi-squared P-value (df = 5) SWWPa seedlings 18.79 0.002 9.440 0.093 SWWP saplings 5.69 0.337 10.420 0.064 SWWP seedlings/total seedlings 23.95 <0.0001 9.370 0.100 SWWP saplings/total saplings 8.03 0.155 10.970 0.052 DFb seedlings 6.38 0.271 16.540 0.005 DF saplings 2.18 0.824 3.500 0.623 PPc seedlings 34.31 <0.0001 20.770 0.001 PP saplings 7.82 0.166 2.680 0.750 Shade-intolerant seedlings 63.66 <0.001 24.280 0.0002 Shade-intolerant saplings 10.51 0.062 9.180 0.102 Shade-tolerant seedlings 28.02 <0.001 8.440 0.134 Shade-tolerant saplings 23.63 0.0002 2.350 0.799 Live SWWP BA 22.91 0.0003 15.240 0.009 Dead SWWP BA 29.87 <0.0001 11.830 0.037 Live SWWP BA/Live total BA 13.37 0.020 11.84 0.037 Live DF BA 16.77 0.005 57.230 0.0001 Live PP BA 76.36 <0.0001 24.730 0.0002 Live shade-intolerant BA 84.75 <0.0001 25.370 0.0001 Live shade-tolerant BA 41.71 <0.0001 6.530 0.258 Kruskal–Wallis mean ranks by site (n = 6) Kruskal–Wallis mean ranks in silviculture treatment/disease severity levels (n = 6) K–W Chi-squared P-value (df = 5) K–W Chi-squared P-value (df = 5) SWWPa seedlings 18.79 0.002 9.440 0.093 SWWP saplings 5.69 0.337 10.420 0.064 SWWP seedlings/total seedlings 23.95 <0.0001 9.370 0.100 SWWP saplings/total saplings 8.03 0.155 10.970 0.052 DFb seedlings 6.38 0.271 16.540 0.005 DF saplings 2.18 0.824 3.500 0.623 PPc seedlings 34.31 <0.0001 20.770 0.001 PP saplings 7.82 0.166 2.680 0.750 Shade-intolerant seedlings 63.66 <0.001 24.280 0.0002 Shade-intolerant saplings 10.51 0.062 9.180 0.102 Shade-tolerant seedlings 28.02 <0.001 8.440 0.134 Shade-tolerant saplings 23.63 0.0002 2.350 0.799 Live SWWP BA 22.91 0.0003 15.240 0.009 Dead SWWP BA 29.87 <0.0001 11.830 0.037 Live SWWP BA/Live total BA 13.37 0.020 11.84 0.037 Live DF BA 16.77 0.005 57.230 0.0001 Live PP BA 76.36 <0.0001 24.730 0.0002 Live shade-intolerant BA 84.75 <0.0001 25.370 0.0001 Live shade-tolerant BA 41.71 <0.0001 6.530 0.258 aSWWP = southwestern white pine. bDF = Douglas-fir. cPP = ponderosa pine. Figure 2 View largeDownload slide (a–d) Mean southwestern white pine (SWWP), ponderosa pine (PP), Douglas-fir (DF), shade-tolerant and shade-intolerant regeneration densities across six sites in Arizona and New Mexico (a. seedlings and c. saplings) and across six silviculture treatment/white pine blister rust (WPBR) severity levels (b. seedlings and d. saplings). Percentages indicate proportion of SWWP as total regeneration TPH at the subplot level (Note: subplots without regeneration were calculated as ‘NA’ and not zero, so subplot-level percentages are slightly different than strata-level means in the graph). Silviculture treatments: no recent mgmt (20 years), recent (20 years) uneven-aged (rUEA) treatments and recent (20 years) two-aged (rTWO) treatments. Sites ordered by longitude (west–east). Figure 2 View largeDownload slide (a–d) Mean southwestern white pine (SWWP), ponderosa pine (PP), Douglas-fir (DF), shade-tolerant and shade-intolerant regeneration densities across six sites in Arizona and New Mexico (a. seedlings and c. saplings) and across six silviculture treatment/white pine blister rust (WPBR) severity levels (b. seedlings and d. saplings). Percentages indicate proportion of SWWP as total regeneration TPH at the subplot level (Note: subplots without regeneration were calculated as ‘NA’ and not zero, so subplot-level percentages are slightly different than strata-level means in the graph). Silviculture treatments: no recent mgmt (20 years), recent (20 years) uneven-aged (rUEA) treatments and recent (20 years) two-aged (rTWO) treatments. Sites ordered by longitude (west–east). Figure 3 View largeDownload slide (a and b) Southwestern white pine (SWWP), ponderosa pine (PP), Douglas-fir (DF), shade-tolerant and shade-intolerant overstory basal areas (BA, m2 ha−1) across six sites in Arizona and New Mexico (a) and six silviculture treatment/white pine blister rust (WPBR) severity levels. Percentages indicate proportion of SWWP as total BA at the subplot level (Note: subplots without overstory trees were calculated as ‘NA’ and not zero, so subplot-level percentages are slightly different than strata-level means in the graph). Silviculture treatments: no recent mgmt, recent uneven-aged (rUEA) treatments and recent two-aged (rTWO) treatments. Sites ordered by longitude (west–east). Figure 3 View largeDownload slide (a and b) Southwestern white pine (SWWP), ponderosa pine (PP), Douglas-fir (DF), shade-tolerant and shade-intolerant overstory basal areas (BA, m2 ha−1) across six sites in Arizona and New Mexico (a) and six silviculture treatment/white pine blister rust (WPBR) severity levels. Percentages indicate proportion of SWWP as total BA at the subplot level (Note: subplots without overstory trees were calculated as ‘NA’ and not zero, so subplot-level percentages are slightly different than strata-level means in the graph). Silviculture treatments: no recent mgmt, recent uneven-aged (rUEA) treatments and recent two-aged (rTWO) treatments. Sites ordered by longitude (west–east). There was considerable variation in densities of several species of regeneration across the six levels of silviculture treatments and WPBR severity but there were no differences in mean ranks in SWWP seedling or SWWP sapling densities (P = 0.06 and 0.09, respectively) or the proportions of SWWP seedlings and saplings to total seedling and sapling densities (P = 0.10 and 0.052, respectively) across the six levels in the non-parametric analyses (Table 1 and Figure 2b,d). However, we found significant effects of both silviculture treatment and WPBR severity on SWWP regeneration abundance with the ZINB models that supported trends in non-parametric analyses. A significant effect of silviculture treatment occurred on SWWP seedling counts (P = 0.041) where SWWP seedling counts were greater in non-managed and recently-managed two-aged stands compared with recently-managed UEA stands (Figure 4). There was also a significant effect of WPBR on seedling counts; they were lower in high WPBR severity stands compared with none/low WPBR severity stands across all silviculture treatments (P = 0.032) (Figure 4). SWWP sapling counts were higher in non-managed stands compared with recently-managed two-aged and UEA stands in low/none WPBR severity areas (P = 0.034, Figure 4). There was no effect of WPBR severity alone on sapling counts (P = 0.99) but the interaction between management and WPBR severity was significant (P = 0.027). There were higher SWWP sapling counts in high WPBR severity stands in recently-managed two-aged residual structures compared with none/low severity stands with a two-aged structure (Figure 4, P = 0.03). Figure 4 View largeDownload slide Seedling and sapling zero-inflated negative binomial mixed effect model estimations (seedling and sapling counts per subplot) for the interaction of silviculture treatment (no mgmt, recently treated uneven-aged residual structure (rUEA) and recently treated two-aged residual stand (rTWO) and WPBR severity (none/light and heavy) levels. df = degrees of freedom; Chisq = Chi-square; Pr(Chisq) = P-value. Error bars represent ±1 SE. Figure 4 View largeDownload slide Seedling and sapling zero-inflated negative binomial mixed effect model estimations (seedling and sapling counts per subplot) for the interaction of silviculture treatment (no mgmt, recently treated uneven-aged residual structure (rUEA) and recently treated two-aged residual stand (rTWO) and WPBR severity (none/light and heavy) levels. df = degrees of freedom; Chisq = Chi-square; Pr(Chisq) = P-value. Error bars represent ±1 SE. Overstory species composition and mean BA varied significantly across silviculture treatment and WPBR severity levels (Table 1 and Figure 3b). Live SWWP, dead SWWP, live DF, live PP and live shade-intolerant species BA all varied across levels; non-managed stands had the highest total BA (>25 m2 ha−1), recently-managed two-aged structures had the lowest total BA (<10 m2 ha−1) and recently-managed UEA structures had total BA between these values (~20 m2 ha−1) (Figure 3b). The proportion of SWWP BA to total BA varied across levels and was generally highest in recently-managed two-aged stands (Table 1 and Figure 3b). Mean ranks of between-stand attributes were commonly variable; factors that varied between management and disease levels were eastness, percent slope, percent canopy closure and proportion of mean annual precipitation that occurs July–September (Supplementary data Table S3). Stands with high WPBR severity were more east-facing than the stands with none/low WPBR severity (Supplementary data Table S3). Recently managed stands were less steep than non-managed stands and two-aged stands had lower percent canopy closure than non-managed or recently-managed UEA stands (Supplementary data Table S3). Non-managed stands had the largest proportion of WPBR-infected SWWP regeneration stems in high severity stands (30% of regeneration infected), 4.0% were infected in recently-managed UEA stands and two-aged structures had no regeneration infected in high severity areas (Table 2). Incidence of WPBR in overstory trees and mean crown dieback of infected trees were highest in non-managed, high severity stands but high across all silviculture treatments in stands with high severity WPBR (Table 2). There were no significant differences in mean ranks of Ribes species densities (the most common alternate host in the WPBR disease cycle in the Southwest) across silviculture treatment/WPBR severity levels, although the highest densities occurred in UEA stands and lowest in recently-managed two-aged structures (Table 2). Table 2 Disease related variable means and standard deviation (SD) and Kruskal–Wallis’ one-way analysis of variance by rank test for differences in ranks among disease and silviculture treatment levels. Kruskal–Wallis’ Chi-squared (H statistic) significant at P ≤ 0.05 (bolded values) indicate differences in the distribution of ranks between silviculture treatments and disease levels (n = number of stands in each level). Silviculture treatmenta Disease levelsb n SWWP regeneration WPBR incidence (%) SWWP overstory WPBR incidence (%) SWWP crown dieback from WPBR (%) Ribes density (SPHc) Mean SD Mean SD Mean SD Mean SD No mgmt None/low severity 26 0.02 0.13 0.06 0.38 0.37 0.98 64.8 144.7 rUEA 9 0.00 0.00 0.01 0.46 0.25 0.67 81.0 153.8 rTWO 6 0.00 0.00 0.06 0.16 0.38 0.80 28.6 75.6 No mgmt High severity 8 0.30 0.38 0.50 0.20 13.41 6.64 58.3 101.9 rUEA 3 0.04 0.09 0.49 0.04 9.76 3.02 244.4 234.1 rTWO 3 0.00 0.00 0.15 0.25 11.44 8.94 0.0 0.0 Kruskal–Wallis Chi-squared, P-value (dfd= 5 for all) H = 55.8, P< 0.0001 H = 72.16, P < 0.0001 H = 76.34, P< 0.0001 H = 6.18, P = 0.289 Silviculture treatmenta Disease levelsb n SWWP regeneration WPBR incidence (%) SWWP overstory WPBR incidence (%) SWWP crown dieback from WPBR (%) Ribes density (SPHc) Mean SD Mean SD Mean SD Mean SD No mgmt None/low severity 26 0.02 0.13 0.06 0.38 0.37 0.98 64.8 144.7 rUEA 9 0.00 0.00 0.01 0.46 0.25 0.67 81.0 153.8 rTWO 6 0.00 0.00 0.06 0.16 0.38 0.80 28.6 75.6 No mgmt High severity 8 0.30 0.38 0.50 0.20 13.41 6.64 58.3 101.9 rUEA 3 0.04 0.09 0.49 0.04 9.76 3.02 244.4 234.1 rTWO 3 0.00 0.00 0.15 0.25 11.44 8.94 0.0 0.0 Kruskal–Wallis Chi-squared, P-value (dfd= 5 for all) H = 55.8, P< 0.0001 H = 72.16, P < 0.0001 H = 76.34, P< 0.0001 H = 6.18, P = 0.289 aSilviculture treatments: rUEA = recently-managed, uneven-aged; rTWO = recently-managed, two-aged, disease levels. bWPBR severity level (none/low severity and high severity). cStems per hectare. dDegrees of freedom. Table 2 Disease related variable means and standard deviation (SD) and Kruskal–Wallis’ one-way analysis of variance by rank test for differences in ranks among disease and silviculture treatment levels. Kruskal–Wallis’ Chi-squared (H statistic) significant at P ≤ 0.05 (bolded values) indicate differences in the distribution of ranks between silviculture treatments and disease levels (n = number of stands in each level). Silviculture treatmenta Disease levelsb n SWWP regeneration WPBR incidence (%) SWWP overstory WPBR incidence (%) SWWP crown dieback from WPBR (%) Ribes density (SPHc) Mean SD Mean SD Mean SD Mean SD No mgmt None/low severity 26 0.02 0.13 0.06 0.38 0.37 0.98 64.8 144.7 rUEA 9 0.00 0.00 0.01 0.46 0.25 0.67 81.0 153.8 rTWO 6 0.00 0.00 0.06 0.16 0.38 0.80 28.6 75.6 No mgmt High severity 8 0.30 0.38 0.50 0.20 13.41 6.64 58.3 101.9 rUEA 3 0.04 0.09 0.49 0.04 9.76 3.02 244.4 234.1 rTWO 3 0.00 0.00 0.15 0.25 11.44 8.94 0.0 0.0 Kruskal–Wallis Chi-squared, P-value (dfd= 5 for all) H = 55.8, P< 0.0001 H = 72.16, P < 0.0001 H = 76.34, P< 0.0001 H = 6.18, P = 0.289 Silviculture treatmenta Disease levelsb n SWWP regeneration WPBR incidence (%) SWWP overstory WPBR incidence (%) SWWP crown dieback from WPBR (%) Ribes density (SPHc) Mean SD Mean SD Mean SD Mean SD No mgmt None/low severity 26 0.02 0.13 0.06 0.38 0.37 0.98 64.8 144.7 rUEA 9 0.00 0.00 0.01 0.46 0.25 0.67 81.0 153.8 rTWO 6 0.00 0.00 0.06 0.16 0.38 0.80 28.6 75.6 No mgmt High severity 8 0.30 0.38 0.50 0.20 13.41 6.64 58.3 101.9 rUEA 3 0.04 0.09 0.49 0.04 9.76 3.02 244.4 234.1 rTWO 3 0.00 0.00 0.15 0.25 11.44 8.94 0.0 0.0 Kruskal–Wallis Chi-squared, P-value (dfd= 5 for all) H = 55.8, P< 0.0001 H = 72.16, P < 0.0001 H = 76.34, P< 0.0001 H = 6.18, P = 0.289 aSilviculture treatments: rUEA = recently-managed, uneven-aged; rTWO = recently-managed, two-aged, disease levels. bWPBR severity level (none/low severity and high severity). cStems per hectare. dDegrees of freedom. Most factors significantly associated with SWWP seedling and sapling densities varied within stands, and several of these variables were other species’ regeneration abundance. An increase in both DF and PP seedling densities was significantly correlated with an increase in SWWP seedlings (τ = 0.292 and 0.289, respectively; Supplementary data Table S4). Increases in shade-tolerant sapling densities, shade-tolerant BA and downed-woody debris ground cover were associated with decreases in SWWP seedlings (τ = −0.304, −0.269 and −0.19, respectively; Supplementary data Table S4). The only climate variable that correlated significantly with SWWP seedling densities was the proportion of mean annual precipitation that occurs in July through September (τ = −0.221; Supplementary data Table S4). There were no significant correlations between SWWP sapling densities and any climate or between-stand variables. An increase in DF sapling densities was significantly correlated with increasing SWWP sapling densities (τ = 0.394; Supplementary data Table S4). SWWP sapling densities were also significantly correlated with an increase in canopy closure (τ = 0.35), live SWWP BA (τ = 0.242) and live DF BA (τ = 0.185) (Supplementary data Table S4). The top candidate ZINB model predicted a decrease in SWWP seedling counts in high WPBR severity stands, with higher percent ground cover of downed-woody debris and in recently-managed UEA stands (Tables 3 and 4). A significant interaction between WPBR severity and DF seedling count indicated that the relationship between SWWP seedling counts and DF seedling count varied between disease levels (Tables 3 and 4). This model had a ∆AIC = 9.0 compared with the saturated model, indicating a better fit (Table 3). The RMSE of the top model was 2.55 seedlings (the range of SWWP seedlings was 0–69 seedlings per subplot). The top candidate SWWP sapling count ZINB model predicted higher SWWP sapling counts in recently-managed, two-aged stands and with increasing canopy closure (Tables 3 and 4). A significant interaction between two-aged stands and canopy closure indicates the relationship between SWWP sapling count and canopy closure varied between management levels. The top model had an AIC value 8.4 lower than the saturated model and a RMSE of 1.21 saplings (range of SWWP saplings from 0 to 31 saplings per subplot; Table 3). The best ZINB models predicted zero counts extremely well but were less accurate in predicting regeneration counts above six stems per subplot (Supplementary data Figure S1). Table 3 Zero-inflated negative binomial (ZINB) models predicting southwestern white pine (SWWP) seedling and sapling counts (per subplot). Shaded value indicates the top candidate model based on AIC and all covariates significant at P < 0.05. Dependent variable Model name Parameters AICa ∆AICb AICwc RMSEd SWWP seedling count ZINBseed1 WPBR severity* DF seedling count + Management + DWDe 718 0.0 0.67 2.55 ZINBseed2 WPBR severity* DF seedling count + Management 721.3 3.3 0.13 2.14 ZINBseed3 WPBR severity* DF seedling count + DWD 721.4 3.4 0.12 2.15 ZINBseed4 WPBR severity+ DF seedling count + DWD 723.1 5.1 0.05 2.98 ZINBseed5 DF seedling count + DWD 724.3 6.3 0.03 2.46 ZINBseed-saturated Mgmt * WPBR severity + propJASf + DF seedling count + PP seedling count + shade-tolerant sapling count + DWD + Live SWWP BAg + Live shade-tolerant BA 727 9.0 – 4.69 SWWP sapling count ZINBsap1 canopy closure * Management 499.0 0.0 0.29 1.33 ZINBsap3 canopy closure * Management + litterh 499.9 0.9 0.19 1.45 ZINBsap4 canopy closure * Management + litter + DF sap count 500.3 1.3 0.15 1.39 ZINBsap5 Management*litter + canopy closure 503.4 4.4 0.03 1.74 ZINBsap-saturated Management * WPBR severity + DF sapling count + canopy closure + litter + Live SWWP BA + Live DF BA 507.3 8.3 – 0.52 Dependent variable Model name Parameters AICa ∆AICb AICwc RMSEd SWWP seedling count ZINBseed1 WPBR severity* DF seedling count + Management + DWDe 718 0.0 0.67 2.55 ZINBseed2 WPBR severity* DF seedling count + Management 721.3 3.3 0.13 2.14 ZINBseed3 WPBR severity* DF seedling count + DWD 721.4 3.4 0.12 2.15 ZINBseed4 WPBR severity+ DF seedling count + DWD 723.1 5.1 0.05 2.98 ZINBseed5 DF seedling count + DWD 724.3 6.3 0.03 2.46 ZINBseed-saturated Mgmt * WPBR severity + propJASf + DF seedling count + PP seedling count + shade-tolerant sapling count + DWD + Live SWWP BAg + Live shade-tolerant BA 727 9.0 – 4.69 SWWP sapling count ZINBsap1 canopy closure * Management 499.0 0.0 0.29 1.33 ZINBsap3 canopy closure * Management + litterh 499.9 0.9 0.19 1.45 ZINBsap4 canopy closure * Management + litter + DF sap count 500.3 1.3 0.15 1.39 ZINBsap5 Management*litter + canopy closure 503.4 4.4 0.03 1.74 ZINBsap-saturated Management * WPBR severity + DF sapling count + canopy closure + litter + Live SWWP BA + Live DF BA 507.3 8.3 – 0.52 aAIC = Akaike’s Information Criterion. b∆AIC = change in AIC in comparison to the best model. cAICw = Akaike’s weights. dRMSE = root mean square error. eDWD = percent ground cover of downed-woody debris. fpropJAS = proportion of mean annual precipitation that occurs July–September. gBA = live overstory tree basal area (m2 ha−1). hlitter = percent ground cover litter/duff. Table 3 Zero-inflated negative binomial (ZINB) models predicting southwestern white pine (SWWP) seedling and sapling counts (per subplot). Shaded value indicates the top candidate model based on AIC and all covariates significant at P < 0.05. Dependent variable Model name Parameters AICa ∆AICb AICwc RMSEd SWWP seedling count ZINBseed1 WPBR severity* DF seedling count + Management + DWDe 718 0.0 0.67 2.55 ZINBseed2 WPBR severity* DF seedling count + Management 721.3 3.3 0.13 2.14 ZINBseed3 WPBR severity* DF seedling count + DWD 721.4 3.4 0.12 2.15 ZINBseed4 WPBR severity+ DF seedling count + DWD 723.1 5.1 0.05 2.98 ZINBseed5 DF seedling count + DWD 724.3 6.3 0.03 2.46 ZINBseed-saturated Mgmt * WPBR severity + propJASf + DF seedling count + PP seedling count + shade-tolerant sapling count + DWD + Live SWWP BAg + Live shade-tolerant BA 727 9.0 – 4.69 SWWP sapling count ZINBsap1 canopy closure * Management 499.0 0.0 0.29 1.33 ZINBsap3 canopy closure * Management + litterh 499.9 0.9 0.19 1.45 ZINBsap4 canopy closure * Management + litter + DF sap count 500.3 1.3 0.15 1.39 ZINBsap5 Management*litter + canopy closure 503.4 4.4 0.03 1.74 ZINBsap-saturated Management * WPBR severity + DF sapling count + canopy closure + litter + Live SWWP BA + Live DF BA 507.3 8.3 – 0.52 Dependent variable Model name Parameters AICa ∆AICb AICwc RMSEd SWWP seedling count ZINBseed1 WPBR severity* DF seedling count + Management + DWDe 718 0.0 0.67 2.55 ZINBseed2 WPBR severity* DF seedling count + Management 721.3 3.3 0.13 2.14 ZINBseed3 WPBR severity* DF seedling count + DWD 721.4 3.4 0.12 2.15 ZINBseed4 WPBR severity+ DF seedling count + DWD 723.1 5.1 0.05 2.98 ZINBseed5 DF seedling count + DWD 724.3 6.3 0.03 2.46 ZINBseed-saturated Mgmt * WPBR severity + propJASf + DF seedling count + PP seedling count + shade-tolerant sapling count + DWD + Live SWWP BAg + Live shade-tolerant BA 727 9.0 – 4.69 SWWP sapling count ZINBsap1 canopy closure * Management 499.0 0.0 0.29 1.33 ZINBsap3 canopy closure * Management + litterh 499.9 0.9 0.19 1.45 ZINBsap4 canopy closure * Management + litter + DF sap count 500.3 1.3 0.15 1.39 ZINBsap5 Management*litter + canopy closure 503.4 4.4 0.03 1.74 ZINBsap-saturated Management * WPBR severity + DF sapling count + canopy closure + litter + Live SWWP BA + Live DF BA 507.3 8.3 – 0.52 aAIC = Akaike’s Information Criterion. b∆AIC = change in AIC in comparison to the best model. cAICw = Akaike’s weights. dRMSE = root mean square error. eDWD = percent ground cover of downed-woody debris. fpropJAS = proportion of mean annual precipitation that occurs July–September. gBA = live overstory tree basal area (m2 ha−1). hlitter = percent ground cover litter/duff. Table 4 Final zero-inflated negative binomial (ZINB) model parameter estimates, standard errors (SE), z-values and Pr(z) values for top SWWP seedling and sapling count per subplot models. Estimate SE z Value Pr (|z|) SWWP seedling (ZINB1seed)  Intercept 2.17 0.36 6.04 < 0.0001  WPBR severity −1.48 0.50 −2.97 0.003  DF seedling count 0.02 0.02 1.19 0.2334  rUEAa −1.39 0.60 −2.34 0.0195  rTWOb 0.43 0.49 0.87 0.3818  DWDc −2.89 1.24 −2.34 0.0195  WPBR severity * DF seedling count 0.12 0.04 2.87 0.0041 SWWP sapling (ZINB1sap)  Intercept −2.40 1.58 −1.52 0.129  rUEA −2.69 2.91 −0.92 0.356  rTWO 3.96 1.89 2.10 0.036  canopy closure 0.04 0.02 2.36 0.018  rUEA*canopy closure 0.03 0.04 0.78 0.433  rTWO*canopy closure −0.10 0.04 −2.72 0.007 Estimate SE z Value Pr (|z|) SWWP seedling (ZINB1seed)  Intercept 2.17 0.36 6.04 < 0.0001  WPBR severity −1.48 0.50 −2.97 0.003  DF seedling count 0.02 0.02 1.19 0.2334  rUEAa −1.39 0.60 −2.34 0.0195  rTWOb 0.43 0.49 0.87 0.3818  DWDc −2.89 1.24 −2.34 0.0195  WPBR severity * DF seedling count 0.12 0.04 2.87 0.0041 SWWP sapling (ZINB1sap)  Intercept −2.40 1.58 −1.52 0.129  rUEA −2.69 2.91 −0.92 0.356  rTWO 3.96 1.89 2.10 0.036  canopy closure 0.04 0.02 2.36 0.018  rUEA*canopy closure 0.03 0.04 0.78 0.433  rTWO*canopy closure −0.10 0.04 −2.72 0.007 arUEA = recently-managed uneven-aged stands. brTWO = recently-managed two-aged stands. cDWD = percent ground cover of downed-woody debris. ZINB1seed = Site (random) variance = 0.000000021 (SD = 0.00045), stand within-site (random) variance = 0.5847 (SD = 0.765), negative binomial dispersion parameter = 1.864 (SE = 0.763), zero-inflation = 0.532 (SE = 0.054). ZINB1sap = Site (random) variance = 0.000000021 (SD = 0.00045), stand within-site (random) variance = 0.1677 (SD = 0.410), negative binomial dispersion parameter = 0.698 (SE = 0.471), zero-inflation = 0.534 (SE = 0.121). Table 4 Final zero-inflated negative binomial (ZINB) model parameter estimates, standard errors (SE), z-values and Pr(z) values for top SWWP seedling and sapling count per subplot models. Estimate SE z Value Pr (|z|) SWWP seedling (ZINB1seed)  Intercept 2.17 0.36 6.04 < 0.0001  WPBR severity −1.48 0.50 −2.97 0.003  DF seedling count 0.02 0.02 1.19 0.2334  rUEAa −1.39 0.60 −2.34 0.0195  rTWOb 0.43 0.49 0.87 0.3818  DWDc −2.89 1.24 −2.34 0.0195  WPBR severity * DF seedling count 0.12 0.04 2.87 0.0041 SWWP sapling (ZINB1sap)  Intercept −2.40 1.58 −1.52 0.129  rUEA −2.69 2.91 −0.92 0.356  rTWO 3.96 1.89 2.10 0.036  canopy closure 0.04 0.02 2.36 0.018  rUEA*canopy closure 0.03 0.04 0.78 0.433  rTWO*canopy closure −0.10 0.04 −2.72 0.007 Estimate SE z Value Pr (|z|) SWWP seedling (ZINB1seed)  Intercept 2.17 0.36 6.04 < 0.0001  WPBR severity −1.48 0.50 −2.97 0.003  DF seedling count 0.02 0.02 1.19 0.2334  rUEAa −1.39 0.60 −2.34 0.0195  rTWOb 0.43 0.49 0.87 0.3818  DWDc −2.89 1.24 −2.34 0.0195  WPBR severity * DF seedling count 0.12 0.04 2.87 0.0041 SWWP sapling (ZINB1sap)  Intercept −2.40 1.58 −1.52 0.129  rUEA −2.69 2.91 −0.92 0.356  rTWO 3.96 1.89 2.10 0.036  canopy closure 0.04 0.02 2.36 0.018  rUEA*canopy closure 0.03 0.04 0.78 0.433  rTWO*canopy closure −0.10 0.04 −2.72 0.007 arUEA = recently-managed uneven-aged stands. brTWO = recently-managed two-aged stands. cDWD = percent ground cover of downed-woody debris. ZINB1seed = Site (random) variance = 0.000000021 (SD = 0.00045), stand within-site (random) variance = 0.5847 (SD = 0.765), negative binomial dispersion parameter = 1.864 (SE = 0.763), zero-inflation = 0.532 (SE = 0.054). ZINB1sap = Site (random) variance = 0.000000021 (SD = 0.00045), stand within-site (random) variance = 0.1677 (SD = 0.410), negative binomial dispersion parameter = 0.698 (SE = 0.471), zero-inflation = 0.534 (SE = 0.121). WPBR severity was not associated with a change in the shape of diameter distributions within any silviculture treatment (Figure 5a–c). There was no difference between diameter distributions of SWWP TPH between none/low and high WPBR severity levels in non-managed stands, (Figure 5a; K–S test D = 0.25, pks = 0.98), recently-managed UEA stands (Figure 5b; D = 0.25, pks = 0.98) or recently-managed two-aged stands (Figure 5c; D = 0.375, pks = 0.63). Management did not affect the diameter distributions of WPBR-infected SWWP either between non-managed stands and recently-managed, two-aged stands (D = 0.625, pks = 0.061; Figure 5a,c) or between non-managed stands and recently-managed UEA stands (D = 0.5, pks = 0.18; Figure 5a,b). Figure 5 View largeDownload slide (a–c) Diameter distributions of non-infected and WPBR-infected SWWP per hectare (TPH) in (a) non-managed stands with none/low severity WPBR (n = 27 stands) and severe WPBR (n = 8 stands), (b) recently-managed UEA stands with none/low severity WPBR (n = 8 stands) and severe WPBR (n = 3 stands) and (c) recently-managed two-aged stands with none/low severity WPBR (n = 6 stands) and severe WPBR (n = 3 stands). Figure 5 View largeDownload slide (a–c) Diameter distributions of non-infected and WPBR-infected SWWP per hectare (TPH) in (a) non-managed stands with none/low severity WPBR (n = 27 stands) and severe WPBR (n = 8 stands), (b) recently-managed UEA stands with none/low severity WPBR (n = 8 stands) and severe WPBR (n = 3 stands) and (c) recently-managed two-aged stands with none/low severity WPBR (n = 6 stands) and severe WPBR (n = 3 stands). In non-managed stands, there were large ratios between sapling-sized trees and the 12.7–22.7 cm DBH class, indicating most sapling-sized trees are not likely to reach the next size class in these stands (Figure 5a). Seedling densities necessary to maintain one TPH of SWWP ≥62.7 cm DBH in none/low WPBR severity, non-managed stands were calculated (mean ratio = 2.04). In order to maintain one SWWP TPH >62.7 cm DBH on average, there need to be approximately ≥150 SWWP seedlings per hectare. Trees in non-managed, high severity stands averaged 46% WPBR infection (range = 19.6–63% across diameter classes; Figure 5a), so in order to maintain one TPH of SWWP ≥62.7 cm DBH in non-managed, high severity WPBR stands there should be ≥300 SWWP seedlings ha−1 to account for potential mortality. Sixty percent of stands (n = 33/55) in this study met the >150 TPH criteria; five were recently-managed UEA stands and six were recently-managed two-aged stands. Eighteen stands met the >300 TPH seedling criteria, and four were recently-managed. The mean ratio and minimum seedling density (without accounting for WPBR infection) for non-managed, none/low WPBR severity stands for each site were as follows: San Francisco Peaks ratio = 2.8, minimum seedlings ha−1 = 170; Mogollon Rim ratio = 2.1, minimum seedlings ha−1 = 40; White Mountains ratio = 2.1, minimum seedlings ha−1 = 40; Mt Graham ratio = 3.1, minimum seedlings ha−1 = 290; Signal Peak ratio = 3.1, minimum seedlings ha−1 = 290 and Sacramento Mountains ratio = 2.5, minimum seedlings ha−1 = 100. Discussion SWWP regeneration was more prolific at the northwestern edge of the range in the San Francisco Peaks and along the Mogollon Rim. Density increases in SWWP and other more shade-tolerant species compared with historical reconstructions prior to fire exclusion have been documented in these areas before (Cocke et al., 2005; Huffman et al., 2015; Rodman et al., 2016). The northwestern distribution of SWWP has no WPBR and stands appear fully stocked, but these areas are also potential hybridization zones where SWWP and limber pine (P. flexilis) ranges overlap (Little, 1971; Steinhoff and Andresen, 1971; Menon et al., 2018). In other sites across the range, SWWP was in lower proportions compared with total regeneration, but not as rare as PP regeneration. This supports other surveys that have quantified SWWP as a minor understory component (generally <5% of total regeneration densities) in stands both pre- and post-treatment (Jones, 1971; Ffolliott and Gottfried, 1991; Looney and Waring, 2013). Co-occurrence with other species was a consistent predictor of SWWP regeneration in our study area, along with silviculture treatments, WPBR and other within-site factors related to ground cover and canopy openness. Site conditions appropriate for DF regeneration appear to be similar to those necessary for SWWP, and shade-tolerant regeneration and overstory BA were negatively related to SWWP seedling densities. SWWP pine occupies a range of community types from xeric, PP-dominated stands to more mesic communities, but appeared to regenerate best in open, drier, mixed conifer stands compared with higher elevation stands where more shade-tolerant species are prolific (Goodrich and Waring, 2017). Many factors can interact with and influence regeneration species and densities (Gray et al., 2005; Puhlick et al., 2012; Bataineh et al., 2013), making mixed species regeneration particularly difficult to predict (Fisichelli et al., 2013; Crotteau et al., 2014). Other studies on related five-needle pine species including P. albicaulis, P. monticola and P. strobus have identified source strength (overstory BA, number of cones), climate variables, light, soils (Dovciak et al., 2003; Maloney, 2014), elevation and aspect (Larson and Kipfmueller, 2010) as good predictors of regeneration. Our analyses suggest that understory and overstory environments were more important in predicting SWWP regeneration than stand-level abiotic factors such as climate variables, elevation or aspect. Fisichelli et al. (2013) hypothesized that mean climate may not explain regeneration variation well because more drastic, short term weather events (droughts, frost events, etc.) could be more important to establishing regeneration. Our findings suggest that management may help maintain sustainable densities in areas of the Southwest with severe WPBR, but stands should be monitored for regeneration and WPBR incidence following treatments. We took opportunities to survey stands where recent (<20 years) management had occurred, but comparisons between pre- and post-treatments could not be made. However, differences in current conditions between treated and non-treated stands shed some light on the potential interactions between management, regeneration and disease. Stands where silviculture treatments had been applied in the past 20 years had lower densities of SWWP regeneration compared with stands not recently-managed, but non-managed stands in high severity areas were not sustainable, with 30% of regeneration infected. These infection rates would offset the benefits of higher densities if regeneration dies before reaching the next size class. WPBR negatively affects SWWP seedling counts most likely through rapid mortality of small trees but the disease can also reduce cone crops in heavily infected stands, thus severing functioning regeneration cycles (McKinney and Tomback, 2007; Schoettle and Sniezko, 2007). The fungus that causes WPBR generally requires cool, humid conditions for spore development and germination, and relative humidity can be significantly reduced and temperatures increased following silviculture treatments (Ma et al., 2010). Open conditions are not usually recommended in managing WPBR because it may stimulate Ribes species production (the secondary host required in the life cycle of WPBR and a source of inoculum for the fungus) and can increase fungal spore dispersal throughout the stand (Tomback et al., 1995; Conklin et al., 2009; Zeglen et al., 2010). Site preparation and prescribed burning following silviculture treatments in high hazard sites may be needed to reduce Ribes densities and competing tree species (see Management recommendations Section). Areas where site preparation is needed (see Management recommendations Section) could be prioritized based on our current understanding of WPBR hazard in the Southwest. In the Sacramento Mountains of New Mexico (where WPBR has occurred the longest in southwestern forests), interacting site factors such as higher elevations, canyon bottoms or lower slope positions and presence/abundance of Ribes species are all are associated with higher WPBR hazard (relationships between site factors and WPBR have not been as well-defined in other parts of the Southwest) (Van Arsdel et al., 1998; Geils et al., 1999; Conklin, 2004; Looney et al., 2015). High WPBR infection levels across all diameter classes in non-managed stands indicates that these stands may eventually lose SWWP as a significant component, especially because the ratio between sapling-sized trees to the next size class was high in non-managed stands (Figure 5a). Avoiding drastic BA reduction through silviculture treatments that create open conditions for SWWP because Ribes abundance will increase should be weighed against the possibility of SWWP regeneration not progressing into the overstory because of closed conditions and WPBR infections. If potential WPBR hazard is considered too high on a specific site during pre-treatment surveys (i.e. low numbers of rust-free trees, many stem cankers, dead tops, high Ribes densities in the stand or in surrounding valley bottoms), planting genetically resistant seedlings could be considered instead of encouraging natural regeneration. Research on quantifying genetically resistant families and determining seed transfer models is ongoing (Waring et al., 2016). We lacked pre-treatment SWWP regeneration densities and WPBR information to compare to post-treatment data measured in this study. Because SWWP is a minor species component and often not the highest species preference (Conklin et al., 2009), pre-treatment data were not available in our study area. The mean age of seedlings (measured by whorl count and adjusted with a regression equation modified by growth ring counts, Goodrich and Waring, 2017) illustrated seedlings were stimulated by treatments in most two-aged stands surveyed (Goodrich and Waring, 2017; Supplementary data Table S1). Mean and maximum seedling ages (i.e. advanced regeneration) were older than most treatment dates in recently-managed UEA structures and oldest in non-managed stands (Goodrich and Waring, 2017;,Supplementary data Table S1). We acknowledge that saplings and some seedlings were likely retained during treatments and all SWWP densities were not solely the products of treatments. Seedling ages also limit inferences of WPBR infections in different stand structures, because disease takes time to develop. Seedlings in UEA and non-managed stands have been exposed to WPBR longer than seedlings in two-aged stands as they were older in general (Goodrich and Waring, 2017;Supplementary data Table S1). We were not able to sample similar stand numbers within WPBR severity or silviculture treatment levels within each mountain range due to the geographic distribution of the disease and differences in forest management across land ownership, which led to an unbalanced design. The majority of EA regeneration treatments we sampled were in lower elevation areas where WPBR severity was low, although we did survey recently-managed, two-aged stands in high rust hazard areas, indicated by high rust severity in overstory trees and previous data from permanent plots (Conklin, 2004;Supplementary data Table S1). We attempted to sample across a wide geographic area within sites with disease and included enough stands with high WPBR severity for valid comparisons (at minimum, three stands per disease/management level) to be made. The random effects of site and stand within-site were quantified in ZINB models, and while stand to stand variation was high, estimates and standard deviations of the random effect of site were low (Table 4). Management recommendations Our surveys suggest that young SWWP were stimulated in treatments that created open structures, were more prolific in areas where PP seedlings and DF seedlings and saplings were successful, and were less prevalent where shade-tolerant overstory and saplings dominate. We recommend stimulating or planting SWWP in local areas where DF and PP successfully regenerate. Complementary research indicates that SWWP seedlings also grow fastest in these open areas without shade-tolerant understory species (Goodrich and Waring, 2017). WPBR has not yet spread to many forests throughout the Southwest (Fairweather and Geils, 2011; Looney et al., 2015) and encouraging SWWP natural regeneration or outplanting genetically resistant seedlings follow proactive guidelines in place for other species (Schoettle, 2004; Schoettle and Sniezko, 2007; Conklin et al., 2009). Increasing natural regeneration can increase the genetic, age and size class diversities of a stand and possibly increase resilience following the introduction of WPBR (Waring and O’Hara, 2005; Schoettle and Sniekzo, 2007; Schwandt et al., 2010; Schoettle et al., 2012). Results from predictive models could be helpful in locating or creating the best sites to outplant genetically resistant SWWP seedlings (Schoettle and Sniezko, 2007; Sniekzo et al., 2011), especially in areas where WPBR severity or potential hazard is high (e.g. areas of high elevation, canyon bottoms, prolific Ribes populations) (Van Arsdel et al., 1998; Conklin et al., 2009). We recommend ~300 SWWP seedlings ha−1 are planted (or stimulated in a successful treatment) to safeguard against potential WPBR mortality. Current work is in progress to collect seeds and quantify the frequency of genetic resistance to C. ribicola across the range of SWWP (Conklin et al., 2009; Sniezko et al., 2011; Waring et al., 2016; Goodrich et al., in press), but until these populations and families are identified, stimulating natural, local regeneration may help preserve genetic diversity and increase resilience at the stand level. We recommend managers anticipate that WPBR will eventually reach most mountain ranges in Arizona and New Mexico. The open, two-aged stands in this study were created through establishment phase of shelterwood with reserve trees or patch clearcut with reserve tree regeneration systems. SWWP seedlings also grew fastest in these open stands (Goodrich and Waring, 2017). Eastern white pine, another shade mid-tolerant species, also grows well in stands with more light available if shade-intolerant understory regeneration is controlled, and shelterwood systems are applied to many eastern white pine-dominated stands (Fahey and Lorimer, 2013; Parker et al., 2013). High densities of shade-intolerant seedlings were measured in two-aged stands in this study, but densities of shade-intolerant saplings were low, indicating they were not reaching substantial sizes. A major concern regarding thinning or regeneration systems in areas with WPBR is that the extra light can increase the abundance of Ribes stems (Conklin et al., 2009; Zeglen et al., 2010). Although single-entry broadcast burning may stimulate seedbed Ribes germination and increase proliferation, repeated broadcast burning or another type of site preparation to kill Ribes stems may help limit densities and lower rust hazard. Prescribed burning and reintroducing surface fire also coincide with restoration goals to reintroduce fire back into southwestern mixed conifer forests, which historically had frequent fire regimes (Swetnam and Baisan, 1996; Brown et al., 2001; Huffman et al., 2015). It is possible that an increase in shade-intolerant species densities following treatments outcompeted the Ribes plants in two-aged stands we surveyed (e.g. sprouting species such as New Mexico locust and the abundance of oak species throughout the Southwest), and a lack of Ribes has been noted in intensively managed lower elevation areas of the Mescalero Apache Reservation in previous surveys (Van Arsdel et al., 1998). We recommend prioritizing silviculture treatments that drastically reduce BA in low hazard areas (stands with fewer current Ribes stems, lower elevations or non-canyon bottoms) (Van Arsdel et al., 1998; Geils et al., 1999; Conklin, 2004; Conklin et al., 2009) and evaluation and monitoring of SWWP regeneration in follow-up thinning treatments within open stands. Because the WPBR hazard of a site depends on time, an inoculum source and suitable environmental conditions for the fungus, it would be helpful to quantify whether shelterwood regeneration treatments across a range of elevations increase the WPBR hazard of a stand by monitoring Ribes densities, relative humidity and temperature data for several years pre- and post-treatments. We recommend that open conditions to stimulate SWWP regeneration are created in areas where SWWP exists as an overstory species through either two-aged or UEA regeneration methods that reduce total BA to 9–20 m2 ha−1 and retain high quality overstory SWWP without lethal WPBR infections (i.e. stem cankers). Modified single-tree selection treatments with target BA of 9−10 m2 ha−1 with various size classes left in groups around larger openings may provide similar open conditions to the two-aged structures where disease-free SWWP regeneration occurred in this study and SWWP seedlings tend to grow the fastest (Goodrich and Waring, 2017). This range of residual BA could meet desired conditions for resilient forests as long as sustainable size classes and appropriate spatial structure are maintained (Reynolds et al., 2013). Site preparation and frequent surface fire to reduce Ribes seeds and stems may be necessary, but other pioneer shade-intolerant species, such as oaks, may outcompete Ribes in some southwestern mixed conifer stands. Stimulating natural SWWP regeneration in open conditions in low WPBR severity areas and/or planting genetically resistant SWWP seedlings in high WPBR severity stands with open conditions should lead to faster growth and recruitment of SWWP into the overstory compared with closed canopy stands under both scenarios (Goodrich and Waring, 2017). Precommercial thinning and timber stand improvement should remove SWWP with stem cankers but leave sapling-sized SWWP if they are free of WPBR stem infections. Current recommendations that include favouring seed-bearing SWWP, retaining trees with good form and vigour, and ranking SWWP higher in species preference should be maintained, as they increase the probability that SWWP will continue to occupy the site, provide a seed source and maintain genetic diversity (Conklin et al., 2009). Supplementary data Supplementary data are available at Forestry online. Acknowledgements We thank K. Peterson, C. Lee-Wilkins, T. Largent, C. Harris and T. Steadman for assistance in the field. We are grateful to W. Hornsby, R. Campbell, T. Padilla, E. Enjady, R. Miller, the Mescalero Apache Tribe, the White Mountain Apache Tribe and the USDA Forest Service for assistance in gathering stand exam data, advice on site selections, information on forest treatments, providing housing and site access and supporting this research. Drs T. Kolb, P. Fule and T. Whitham provided reviews that improved earlier versions of this manuscript. Conflict of interest statement None declared. Funding McIntire-Stennis appropriations to the State of Arizona and Northern Arizona University, United States Department of Agriculture Forest Health Protection Evaluation Monitoring Program and the Northern Arizona University Genes to Environment Program through the Department of Biology. References Akaike , H. 1974 A new look at the statistical model identification . IEEE Trans. Autom. Control 19 , 716 – 723 . Google Scholar CrossRef Search ADS Auty , D. , Achim , A. , Bédard , P. and Pothier , D. 2014 StatSAW: modelling lumber product assortment using zero-inflated Poisson regression . Can. J. For. Res. 44 , 638 – 647 . Google Scholar CrossRef Search ADS Bataineh , M. , Kenefic , L. , Weiskittel , A. , Wagner , R. and Brissette , J. 2013 Influence of partial harvesting and site factors on the abundance and composition of natural regeneration in the Acadian Forest of Maine, USA . For. Ecol. Manag. 306 , 96 – 106 . Google Scholar CrossRef Search ADS Brown , P.M. , Kaye , M.W. , Huckaby , L.S. and Baisan , C.H. 2001 Fire history along environmental gradients in the Sacramento Mountains, New Mexico: influences of local patterns and regional processes . Ecoscience 8 , 115 – 126 . Google Scholar CrossRef Search ADS Burnham , K.P. and Anderson , D.R. 2002 Model selection and multimodel inference: a practical information-theoretic approach . Springer Science & Business Media . Conklin , D.A. 2004 Development of the white pine blister rust outbreak in New Mexico . USDA Forest Service , 17 . Southwestern Region, R3-04-01. Conklin , D.A. , Fairweather , M.L. , Ryerson , D.E. , Geils , B.W. and Vogler , D.R. 2009 White pines, blister rust, and management in the Southwest . USDA Forest Service , 16 . Southwestern Region, R3-FH-09-01. Crookston , N.L. and Rehfeldt , G.E. 2008 . Climate estimates and plant-climate relationships. USDA Forest Service, Rocky Mountain Research Station, Moscow, ID. Web Site: http://forest.moscowfsl.wsu.edu/climate/ (accessed March 16 2017). Crotteau , J.S. , Ritchie , M.W. and Varner , J.M. 2014 A mixed-effects heterogeneous negative binomial model for postfire conifer regeneration in northeastern California, USA . For. Sci. 60 , 275 – 287 . DeRose , R.J. and Long , J.N. 2014 Resistance and resilience: a conceptual framework for silviculture . For. Sci. 60 , 1205 – 1212 . Fahey , R.T. and Lorimer , C.G. 2013 Restoring a midtolerant pine species as a component of late-successional forests: results of gap-based planting trials . For. Ecol. Manag. 292 , 139 – 149 . Google Scholar CrossRef Search ADS Fairweather , M.L. and Geils , B.W. 2011 First report of the white pine blister rust pathogen, Cronartium ribicola, in Arizona . Plant Dis. 95 , 494 – 494 . Google Scholar CrossRef Search ADS Ffolliott , P.F. and Gottfried , G.J. 1991 Mixed conifer and aspen regeneration in small clearcuts within a partially harvested Arizona mixed conifer forest. Research Paper Rocky Mountain Forest and Range Experiment Station . USDA Forest Service . (RM-294). Fisichelli , N.A. , Frelich , L.E. and Reich , P.B. 2013 Climate and interrelated tree regeneration drivers in mixed temperate–boreal forests . Landscape Ecol. 28 , 149 – 159 . Google Scholar CrossRef Search ADS Geils , B.W. , Conklin , D.A. and Van Arsdel , E.P. 1999 A preliminary hazard model of white pine blister rust for the Sacramento Ranger District, Lincoln National Forest. Res. Note RMRS-RN-6. Fort Collins, CO: U.S. Department ofAgriculture, Forest Service, Rocky Mountain Research Station. 6 p. [Online]. Available: http://www.treesearch.fs.fed.us/pubs/4702 Goodrich , B.A. and Waring , K.M. 2017 Pinus strobiformis seedling growth in southwestern US mixed conifer forests in managed and non-managed stands . Forestry 90 , 393 – 403 . Google Scholar CrossRef Search ADS Goodrich , B.A. , Waring , K.M. and Flores-Renteria , L. Inpress Pinus strobiformis gene conservation and genecology. In: Schoettle, A.W., R.A. Sniezko, J. Kliejunas, eds. Genetics of Five-Needle Pines, Rusts of Forest Trees, and Strobusphere: Proceedings of the IUFRO Joint Conference. 15–20 June 2014; Fort Collins, CO. Proceedings RMRS-P-xxx (in press). Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station Gray , A.N. , Zald , H.S. , Kern , R.A. and North , M. 2005 Stand conditions associated with tree regeneration in Sierran mixed-conifer forests . For. Sci. 51 , 198 – 210 . Grissino-Mayer , H.D. , Baisan , C.H. and Swetnam , T.W. 1995 Fire history in the Pinaleño Mountains of Southeastern Arizona: effects of human-related disturbances. In Biodiversity and Management of the Madrean Archipelago: The Sky Islands of Southwestern United States and Northwester Mexico . Debano L.F. , Gottfried G.J. , Hamre R.H. , Edminster C.B. , Pfolliott P.F. and Ortega-Rubio A. (eds). USDA Forest Service Rocky Mountain Research Station , pp. 399 – 406 . Gen. Tech. Rep. RM-GTR 264. Huffman , D.W. , Zegler , T.J. and Fulé , P.Z. 2015 Fire history of a mixed conifer forest on the Mogollon Rim, northern Arizona, USA . Int. J. Wildland Fire 24 , 680 – 689 . Google Scholar CrossRef Search ADS Husch , B. , Beers , T.W. and Kershaw , J.A. 2002 Forest Mensuration . John Wiley & Sons Inc , p. 456 . Jones , J.R. 1971 Mixed conifer seedling growth in eastern Arizona. USDA For. Serv. Res. Pap. RM-77, 19 p. Keane , R.E. , Holsinger , L.M. , Mahalovich , M.F. and Tomback , D.F. 2017 Evaluating future success of whitebark pine ecosystem restoration under climate change using simulation modeling . Restoration Ecol. 25 , 220 – 233 . Google Scholar CrossRef Search ADS Keane , R.E. and Schoettle , A.W. 2011 Strategies, tools, and challenges for sustaining and restoring high elevation five-needle white pine forests in western North America. In: Keane, RE.; Tomback, D.F.; Murray, M.P. and Smith, C.M., eds. 2011. The Future of High-elevation, Five-needle White Pines in Western North America: Proceedings of the High Five Symposium. 28–30 June 2010; Missoula, MT. Proceedings RMRS-P-63. Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station. pp. 276 – 294 . Kearns , H.S. and Jacobi , W.R. 2007 The distribution and incidence of white pine blister rust in central and southeastern Wyoming and northern Colorado . Can. J. For. Res. 37 , 462 – 472 . Google Scholar CrossRef Search ADS Korb , J.E. , Fulé , P.Z. and Stoddard , M.T. 2012 Forest restoration in a surface fire-dependent ecosystem: an example from a mixed conifer forest, southwestern Colorado, USA . For. Ecol. Manage. 269 , 10 – 18 . Google Scholar CrossRef Search ADS Larson , E.R. and Kipfmueller , K.F. 2010 Patterns in whitebark pine regeneration and their relationships to biophysical site characteristics in southwest Montana, central Idaho, and Oregon, USA . Can. J. For. Res. 40 , 476 – 487 . Google Scholar CrossRef Search ADS Li , R. , Weiskittel , A.R. and Kershaw , J.A. , Jr. 2011 Modeling annualized occurrence, frequency, and composition of ingrowth using mixed-effects zero-inflated models and permanent plots in the Acadian Forest Region of North America . Can. J. For. Res. 41 , 2077 – 2089 . Google Scholar CrossRef Search ADS Little , E.L. , Jr. 1971 Atlas of United States Trees, Volume 1, Conifers and Important Hardwoods. United States Department of Agriculture Miscellaneous Publication 991. 9 p, 200 maps. Loo , J.A. 2009 Ecological impacts of non-indigenous invasive fungi as forest pathogens . Biol. Invasions 11 , 81 – 96 . Google Scholar CrossRef Search ADS Looney , C.E. and Waring , K.W. 2012 Patterns of forest structure, competition and regeneration in southwestern white pine (Pinus strobiformis) forests . For. Ecol. Manage. 286 , 159 – 170 . Google Scholar CrossRef Search ADS Looney , C.E. and Waring , K.W. 2013 Pinus strobiformis (southwestern white pine) stand dynamics, regeneration, and disturbance ecology: a review . For. Ecol. Manage. 287 , 90 – 102 . Google Scholar CrossRef Search ADS Looney , C.E. , Waring , K.W. and Fairweather , M.L. 2015 Monitoring the health of Pinus strobiformis: early impacts of white pine blister rust invasion. In: Potter, Kevin M.; Conkling, Barbara L. 2015. Forest Health Monitoring: National Status, Trends, and Analysis 2013. Gen. Tech. Rep. SRS-207. Asheville, NC: U.S. Department of Agriculture Forest Service, SouthernResearch Station. pp. 167 − 174 . Ma , S. , Concilio , A. , Oakley , B. , North , M. and Chen , J. 2010 Spatial variability in microclimate in a mixed-conifer forest before and after thinning and burning treatments . For. Ecol. Manage. 259 , 904 – 915 . Google Scholar CrossRef Search ADS Maloney , P.E. 2014 The multivariate underpinnings of recruitment for three Pinus species in montane forests of the Sierra Nevada, USA . Plant Ecol. 215 , 261 – 274 . Google Scholar CrossRef Search ADS Mattson , D.J. and Arundel , T.A. 2013 Consumption of seeds of southwestern white pine (Pinus strobiformis) by black bear (Ursus americanus) . Southwest. Naturalist 58 , 243 – 245 . Google Scholar CrossRef Search ADS McKinney , S.T. and Tomback , D.F. 2007 The influence of white pine blister rust on seed dispersal in whitebark pine . Can. J. For. Res. 37 , 1044 – 1057 . Google Scholar CrossRef Search ADS Menon , M. , Bagley , J.C. , Friedline , C.J. , Whipple , A.V. , Schoettle , A.W. , Leal‐Saenz , A. , et al. 2018 The role of hybridization during ecological divergence of southwestern white pine (Pinus strobiformis) and limber pine (P. flexilis) . Mol. Ecol. doi:10.1111/mec.14505 . Millar , C.I. , Stephenson , N.L. and Stephens , S.L. 2007 Climate change and forests of the future: managing in the face of uncertainty . Ecol. Appl. 17 , 2145 – 2151 . Google Scholar CrossRef Search ADS PubMed Millers , I. , Lachance , D. , Burkman , W.G. and Allen , D.C. 1991 North American sugar maple decline project: organization and field methods. USDA Forest Service Northeastern Forest Experiment Station, General Technical Report NE-154. 26p. Nyland , R.D. 1996 Silviculture: concepts and applications . McGraw-Hill . O’Hara , K.L. and Ramage , B.S. 2013 Silviculture in an uncertain world: utilizing multi-aged management systems to integrate disturbance . Forestry 86 , 401 – 410 . Google Scholar CrossRef Search ADS Paluch , J.G. 2005 The influence of the spatial pattern of trees on forest floor vegetation and silver fir (Abies alba Mill.) regeneration in uneven-aged forests . For. Ecol. Manage. 205 , 283 – 298 . Google Scholar CrossRef Search ADS Parker , A.J. 1982 The topographic relative moisture index: an approach to soil-moisture assessment in mountain terrain . Phys. Geogr. 3 , 160 – 168 . Parker , W.C. , Noland , T.L. and Morneault , A.E. 2013 Comparative mast seed production in unmanaged and shelterwood white pine (Pinus strobus L.) stands in central Ontario . New Forests 44 , 613 – 628 . Google Scholar CrossRef Search ADS Puhlick , J.J. , Laughlin , D.C. and Moore , M.M. 2012 Factors influencing ponderosa pine regeneration in the southwestern USA . For. Ecol. Manage. 264 , 10 – 19 . Google Scholar CrossRef Search ADS R Core Team . 2014 R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria Rehfeldt , G.E. 2006 A spline model of climate for the western United States. General Technical Report 165 . USDA Forest Service, Rocky Mountain Research Station . Reynolds , R.T. , Sánchez Meador , A.J.S. , Youtz , J.A. , Nicolet , T. , Matonis , M.S. , Jackson , P.L. , et al. . 2013 Restoring composition and structure in southwestern firequent-fire forests: A science-based framework for improving ecosystem resiliency . USDA Forest Service-General Technical Report RMRS-GTR, (310 RMRS-GTR), 76pp. Ridout , M. , Hinde , J. and DeméAtrio , C.G. 2001 A score test for testing a zero‐inflated poisson regression model against zero‐inflated negative binomial alternatives . Biometrics 57 , 219 – 223 . Google Scholar CrossRef Search ADS PubMed Roberts , D.W. 1986 Ordination on the basis of fuzzy set theory . Vegetatio 66 , 123 – 131 . Google Scholar CrossRef Search ADS Rodman , K.C. , Sánchez Meador , A.J. , Huffman , D.W. and Waring , K.M. 2016 Reference conditions and historical fine-scale spatial dynamics in a southwestern mixed-conifer forest, Arizona, USA . For. Sci. 62 , 268 – 280 . Samano , S. and Tomback , D.F. 2003 Cone opening phenology, seed dispersal, and seed predation in southwestern white pine (Pinus strobiformis) in southern Colorado . Ecoscience 10 , 319 – 326 . Google Scholar CrossRef Search ADS Schoettle , A.W. 2004 Ecological roles of five-needle pines in Colorado: potential consequences of their loss. Pages 124–135 In: Breeding and Genetic Resources of Five-Needle Pines: Growth, Adaptability, and Pest Resistance. Proc. RMRS-P-32. R. A. Sniezko, S. Samman, S. E. Schlarbaum, and H. B. Kriebel, eds. United States Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, CO. Schoettle , A.W. , Klutsch , J.G. and Sniezko , R.A. 2012 Integrating regeneration, genetic resistance, and timing of intervention for the longterm sustainability of ecosystems challenged by non-native pests—a novel proactive approach. Pages 112–1 23 in: Proc. Fourth Int. Workshop Genet. Host–Parasite Interact. For. Disease and Insect Resistance in Forest Trees, Eugene, OR. Gen. Tech. Rep. PSW-GTR-240. R. A. Sniezko, A. D. Yanchuk, J. T. Kliejunas, K. M. Palmieri, J. M. Alexander, S. J. Frankel, tech. coord. United States Department of Agriculture, Forest Service, Pacific Southwest Research Station, Albany, CA. Schoettle , A.W. and Sniezko , R.A. 2007 Proactive intervention to sustain high elevation pine ecosystems threatened by white pine blister rust . J. For. Res. 12 , 327 – 336 . Google Scholar CrossRef Search ADS Schwandt , J.W. , Lockman , I.B. , Kliejunas , J.T. and Muir , J.A. 2010 Current health issues and management strategies for white pines in the western United States and Canada . For. Pathol. 40 , 226 – 250 . Google Scholar CrossRef Search ADS Sekhon , J.S. 2011 Multivariate and propensity score matching software with automated balance optimization: the matching package for R . J. Stat. Softw. 42 , 1 – 52 . Google Scholar CrossRef Search ADS Shirk , A.J. , Cushman , S.A. , Waring , K.M. , Wehenkel , C.A. , Leal-Sáenz , A. , Toney , C. , et al. 2018 Southwestern white pine (Pinus strobiformis) species distribution models project a large range shift and contraction due to regional climatic changes . For. Ecol. Manage 411 , 176 – 186 . Google Scholar CrossRef Search ADS Skaug , H. , Fournier , D. , Nielsen , A. , Magnusson , A. and Bolker , B. 2011 glmmADMB: Generalized Linear Mixed Models Using AD Model Builder. R Package Version 0.6, 5, r143 Sniezko , R.A. , Mahalovich , M.F. , Schoettle , A.W. and Vogler , D.R. 2011 Past and current investigations of the genetic resistance to Cronartium ribicola in high-elevation five-needle Pines. In: Keane, Robert E.; Tomback, Diana F.; Murray, Michael P.; and Smith, Cyndi M., eds. 2011. The Future Of High-Elevation, Five-Needle White Pines in Western North America: Proceedings of the High Five Symposium. 28–30 June 2010; Missoula, MT. Proceedings RMRS-P-63. Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station. pp. 246 – 264 . Spiess , A.-N. 2014 qpcR: Modelling and Analysis of Real-Time PCR Data. R Package Version 1.4-0. https://CRAN.R-project.org/package=qpcR Steinhoff , R.J. and Andresen , J.W. 1971 Geographic variation in Pinus flexilis and Pinus strobiformis and its bearing on their taxonomic status . Silvae Genet. 20 , 159 – 167 . Stoddard , M.T. , Meador , A.S. , Fulé , P.Z. and Korb , J.E. 2015 Five-year post-restoration conditions and simulated climate-change trajectories in a warm/dry mixed-conifer forest, southwestern Colorado, USA . For. Ecol. Manage. 356 , 253 – 261 . Google Scholar CrossRef Search ADS Swetnam , T. and Baisan , C. 1996 Historical fire regime patterns in the southwestern United States since AD 1700. In: C.D. Allen (ed) Fire Effects in Southwestern Forests: Proceedings of the 2nd La Mesa Fire Symposium, pp. 11 – 32 . USDA Forest Service, Rocky Mountain Research Station, General Technical Report RM-GTR-286. Tomback , D.F. and Achuff , P. 2010 Blister rust and western forest biodiversity: ecology, values and outlook for white pines . For. Pathol. 40 , 186 – 225 . Google Scholar CrossRef Search ADS Tomback , D.F. , Clary , J.K. , Koehler , J. , Hoff , R.J. and Arno , S.F. 1995 The effects of blister rust on post‐fire regeneration of whitebark pine: the Sundance burn of northern Idaho (USA) . Conserv. Biol. 9 , 654 – 664 . Google Scholar CrossRef Search ADS United States Geological Survey . 2006 Digital Representations of Tree Species Range Maps, from Little, E.L., Jr., 1971, Atlas of United States Trees, Conifers and Important Hardwoods, vol. 1. US Department of Agriculture Miscellaneous Publication 1146, 9 p. Van Arsdel , E.P. , Conklin , D.A. , Popp , J.B. and Geils , B.W. 1998 The distribution of white pine blister rust in the Sacramento Mountains of New Mexico. In: Jalkanen R, Crane PE, Walla JA, Aalto T (eds) Proceedings of the 1st IUFRO Rusts of Forest Trees Working Party Conference; 1998 August 2–7; Saariselkä, Finland. Research Paper 712. Finnish Forest Research Institute, Rovaniemi, Finland Waring , K. , Cushman , S. , Eckert , A. , Flores Renteria , L. , Sniezko , R. , Still , S. , et al. 2016 Blending Ecology and Evolution using Emerging Technologies to Determine Species Distributions with a Non-native Pathogen in a Changing Climate. Macrosystems Biology PI Meeting, September 28–30, 2016. Washington, DC. Waring , K.M. and O’Hara , K.L. 2005 Silvicultural strategies in forest ecosystems affected by introduced pests . For. Ecol. Manage. 209 , 27 – 41 . Google Scholar CrossRef Search ADS Zald , H.S. , Gray , A.N. , North , M. and Kern , R.A. 2008 Initial tree regeneration responses to fire and thinning treatments in a Sierra Nevada mixed-conifer forest, USA . For. Ecol. Manage. 256 , 168 – 179 . Google Scholar CrossRef Search ADS Zeglen , S. , Pronos , J. and Merler , H. 2010 Silvicultural management of white pines in western North America . For. Pathol. 40 , 347 – 368 . Google Scholar CrossRef Search ADS © Institute of Chartered Foresters, 2018. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Forestry: An International Journal Of Forest ResearchOxford University Press

Published: Oct 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off