Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Twenty years of drought‐mediated change in snag populations in mixed‐conifer and ponderosa pine forests in Northern Arizona

Twenty years of drought‐mediated change in snag populations in mixed‐conifer and ponderosa pine... Background: Snags (standing dead trees) are important biological legacies in forest systems, providing numerous resources as well as a record of recent tree mortality. From 1997 to 2017, we monitored snag populations in drought-influenced mixed-conifer and ponderosa pine (Pinus ponderosa) forests in northern Arizona. Results: Snag density increased significantly in both forest types. This increase was driven largely by a pulse in snag recruitment that occurred between 2002 and 2007, following an extreme drought year in 2002, with snag recruitment returning to pre-pulse levels in subsequent time periods. Some later years during the study also were warmer and/or drier than average, but these years were not as extreme as 2002 and did not trigger the same level of snag recruitment. Snag recruitment was not equal across tree species and size classes, resulting in significant changes in species composition and size-class distributions of snag populations in both forest types. Because trees were far more abundant than snags in these forests, the effect of this mortality pulse on tree populations was far smaller than its effect on snag populations. Snag loss rates increased over time during the study, even though many snags were newly recruited. This may reflect the increasing prevalence of white fir snags and/or snags in the smaller size classes, which generally decay faster than snags of other species or larger snags. Thus, although total numbers of snags increased, many of the newly recruited snags may not persist long enough to be valuable as nesting substrates for native wildlife. Conclusions: Increases in snag abundance appeared to be due to a short-term tree mortality “event” rather than a longer- term pattern of elevated tree mortality. This mortality event followed a dry and extremely warm year (2002) embedded within a longer-term megadrought. Climate models suggest that years like 2002 may occur with increasing frequency in the southwestern U.S. Such years may result in additional mortalitypulses, whichinturnmay strongly affect trajectories in abundance, structure, and composition of snag populations. Relative effects on tree populations likely will be smaller, but, over time, also could be significant. Keywords: Climate change, Drought, Monitoring, Snag abundance, Snag creation, Snag dynamics, Species composition, Tree mortality Background (Adams et al. 2009; Allen et al. 2010), and climate models Understanding the effects of climate change on structure predict further warming and drying in many areas (e.g., and composition of forest ecosystems presents a major Seager et al. 2007; Stocker et al. 2013;Garfinet al. 2014) challenge for researchers and managers in the twenty-first suggesting that changing climates will continue to affect century (Millar et al. 2007). Warming climates already these systems. Forest systems can be affected directly have profoundly affected forests throughout the world through climate-mediated mortality (Breshears et al. 2005; Mueller et al. 2005; Adams et al. 2009), or indirectly * Correspondence: joe.ganey@nau.edu through altered disturbance regimes that result in in- US Forest Service, Rocky Mountain Research Station, 2500 S. Pine Knoll, AZ creased mortality (e.g., McKenzie et al. 2004; Bentz et al. 86001 Flagstaff, USA © The Author(s). 2021 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Ganey et al. Forest Ecosystems (2021) 8:20 Page 2 of 13 2010; Wan et al. 2019) or reduced regeneration (Davis et al. 2015). Here, we expanded on this earlier work to et al. 2019). Because drought-mediated mortality may dif- summarize drought-mediated changes in snag and tree ferentially affect tree species and sizes, such mortality may abundance, and composition of snag and tree popula- drive significant changes in forest structure and compos- tions within these forest types, over the 20-year period ition (Mueller et al. 2005;Ganey andVojta 2011; Kane from 1997 to 2017, with an emphasis on temporal et al. 2014). changes in the snag population, including rates of snag Snags, or standing dead trees, are important compo- recruitment and loss and the factors driving these pro- nents of forest systems that are directly affected by cesses. Our specific objectives included: (1) Evaluating drought-mediated trends in tree mortality. These snags temporal trends in snag and tree density and snag re- serve as biological legacies in forest systems (Thomas cruitment and loss rates, (2) Summarizing climate et al. 1979; McComb and Lindenmayer 1999; Wolden- during the study period and evaluating potential rela- dorp and Keenan 2005), providing habitat for native tionships between climate patterns and snag recruit- wildlife (e.g., Bull et al. 1997; Rabe et al. 1998), serving ment, and (3) Evaluating temporal trends in composition as an important source of coarse woody debris (Harmon and structure of overall snag populations, as well as in et al. 1986; Laudenslayer et al. 2002; Woldendorp and snags that were recruited or lost during the intervals be- Keenan 2005), and aiding in nutrient cycling and other tween sampling occasions. These data thus provide in- ecosystem functions. Snags also provide an index of re- formation on trends in snag populations during the cent tree mortality (Ganey and Vojta 2011; Wu et al. study period as well as on the factors driving those 2017), with recruitment of new snags reflecting temporal trends. This information should aid forest managers in changes in tree mortality and relative mortality among understanding the potential effects of future climate pat- species and size classes, potentially providing insight into terns on these snag populations, and, to a lesser extent, changes in forest structure and composition. on the tree populations from which they derive. In the southwestern United States, a pronounced in- crease in snag recruitment (i.e., tree mortality) followed Methods an extreme climatic year (2002) embedded within a Study area longer-term mega-drought (Breshears et al. 2005; We established plots for sampling snags and trees within Kopeke et al. 2010; Williams et al. 2020). Several studies a study area of approximately 73,000 ha within the reported on short-term trends in tree mortality (which Coconino and Kaibab National Forests, north-central results in snag recruitment) in important forest types Arizona. Study plots were randomly located in mixed- following this extreme year (Negron et al. 2009 [through conifer (n = 53 plots) and ponderosa pine forests (n =60 2004]; Ganey and Vojta 2011 [through 2007]; Kane et al. plots) within this area (see Ganey 1999) for details on 2014 [through 2008]), but data documenting longer- plot selection). White fir (Abies concolor Lindl. ex Hil- term trends in these forests are lacking, and few studies debr.), Douglas-fir (Pseudotsuga menziesii [Mirb.] focused specifically on snag populations (but see Ganey Franco), and ponderosa pine dominated overstories in and Vojta 2005, 2012, 2014). mixed-conifer forests, accounting for approximately Snag populations are governed by the balance between 90 % of total trees in this forest type (as sampled in gains and losses in snags. Gains occur when new snags 2004; Ganey and Vojta 2011). Other relatively common are created by natural tree senescence processes or dis- species included Gambel oak (Quercus gambelii Nutt.), turbances such as insect outbreaks, diseases, fire, or quaking aspen (Populus tremuloides Michaux), and lim- droughts. Losses occur when snags are lost to timber or ber pine (P. flexilis James), in that order of frequency. fuelwood harvest, fire, or natural decomposition, pro- Ponderosa pine comprised > 90 % of trees in ponderosa cesses which are influenced by factors such as snag spe- pine forest in 2004 (Ganey and Vojta 2011), with Gam- cies and size (Ganey et al. 2015). It is therefore desirable bel oak accounting for approximately 8 % of total trees not only to document populations of existing snags but by frequency. Alligator juniper (Juniperus deppeana also to understand how those populations change over Steud), Douglas-fir, quaking aspen, limber pine, pinyon time and the factors responsible for those changes. pine (P. edulis), and other species of juniper were Since 1997, we have sampled snag and tree popula- present in some stands, typically in relatively small tions periodically in mixed-conifer and ponderosa pine numbers. forests in northern Arizona. Previous papers from this Study plots included a wide range of topographic con- study documented changes in snag populations over 5-, ditions and soil types, ranged in elevation from the tran- 10-, and 15-year increments within this 20-year period sition zone between pinyon − juniper woodland and (Ganey and Vojta 2005, 2012, 2014) as well as patterns ponderosa pine at lower elevations to the ecotone be- in: tree mortality from 2002 to 2007 (Ganey and Vojta tween mixed-conifer and Engelmann spruce (Picea 2011) and snag longevity from 1997 to 2015 (Ganey engelmanni Parry ex Engelm.) − corkbark fir (Abies Ganey et al. Forest Ecosystems (2021) 8:20 Page 3 of 13 lasiocarpa var. arizonica [Merriam] Lemmon) forests at First, our objective was to summarize trends in snag and higher elevations (Brown et al. 1980), and included both tree populations on the overall landscape, including areas commercial forest lands and administratively reserved subject to these disturbances. Second, although the effect lands such as wilderness and other roadless areas. Plots of fire could be large on individual plots, the overall effect also included areas subject to forest thinning and both was relatively small over most 5-year intervals between prescribed burns and wildfires. Consequently, these plots sampling occasions, because relatively few plots were im- sampled a wide range of conditions with respect to for- pacted by moderate-to high severity fire during those in- est structure and composition, and included areas sub- tervals (Table 1). Consequently, estimates for all plots jected to recent disturbance events. were similar to estimates including only plots that did not experience moderate-to high-severity fire. We also in- Sampling snag and tree populations cluded plots subject to forest thinning, but these were We sampled snags in 1-ha plots at five-year intervals be- even fewer in number than burned plots, and thinning ginning in 1997 (i.e., snags were sampled in 1997, 2002, activities therefore had minimal immediate impacts on 2007, 2012, and 2017). Within plot boundaries, we sam- overall snag or tree populations (although they may im- pled all snags ≥ 2 m in height and ≥ 20 cm in diameter at pact future patterns of tree mortality and/or wildfire). breast height (dbh). The focus on snags ≥ 20 cm in dbh reflected the original study objectives related to wildlife We treated years in which we sampled snag (or tree) habitat (Ganey 1999) and the assumption that smaller populations as sampling occasions and the intervals be- snags were less important to wildlife such as cavity- tween those sampling occasions as “sampling periods”. nesting birds (Scott 1978; Cunningham et al. 1980; Con- Thus, for snags we had five sampling occasions (here- way and Martin 1993) and/or roosting bats (Rabe et al. after “years”) and four 5-year sampling periods, whereas 1998; Bernardos et al. 2004; Solvesky and Chambers for trees we had two years and one 10-year sampling 2009). Given this minimum diameter, our inference is period. Years provided estimates of density and compos- limited to populations of snags ≥ 20 cm dbh. ition for snag (or tree) populations at a point in time, We marked all snags with numbered metal tags, allow- and sampling periods provided estimates of change in ing us to distinguish pre-existing snags from new snags those parameters between those points in time. when re-sampling plots (with some exceptions, see As noted earlier, our plots covered a very wide range below). We recorded species and dbh (nearest cm) for of forest structural conditions. Consequently, snag dens- all snags. We sampled plots from May through August, ities varied widely among plots and were so markedly and did not necessarily sample individual plots on the skewed (Ganey and Vojta 2012) that the utility of stand- same date or even in the same month across years. Thus, ard estimators of central tendency such as the mean or the elapsed time between consecutive samples for an in- median was limited. Therefore, we used Huber’sM- dividual plot could range from slightly < 5 years to estimator, a generalized maximum–likelihood estimator slightly > 5 years. We ignored this variability in analysis, that provides robust estimates in distributions contain- and assumed that all intervals between sampling occa- ing outliers (Huber and Ronchetti 2009), to estimate sions represented a 5-year period. central tendency in snag density (by year) and change in We sampled live trees ≥ 20 cm in dbh, which were far snag density (by sampling period). We estimated this more abundant than snags, in a 0.09-ha subplot located parameter and associated 95 % bias-corrected confidence within each snag plot in 2004 and 2014. We adhered to intervals using 1,000 bootstrap iterations (Efron and the minimum 20-cm diameter for consistency with snag Table 1 Number (and percent) of plots in northern Arizona sampling. We recorded tree species and dbh (nearest mixed-conifer and ponderosa pine forest that experienced cm) for all trees. We did not mark individual trees, and moderate-to high-severity fire by 5-year interval between snag consequently were not able to determine fates of individ- sampling occasions. n = 53 and 60 plots sampled in mixed- ual trees. Instead, we focused on overall changes in tree conifer and ponderosa pine forest, respectively. Plots were populations, which were driven by the interactions classified as having experienced moderate- to high-severity fire among ingrowth of small (< 20 cm dbh in 2004) trees if a visual inspection indicated that considerable tree mortality had occurred within the plot due to fire between sampling into our sampled population, growth of trees within that occasions sampled population, and tree mortality. As with snags, our inference is limited to trees ≥ 20 cm dbh. Time interval Mixed-conifer forest Ponderosa pine forest 1997–2002 0 (0.0) 2 (3.3) Analysis of snag and tree populations 2002–2007 3 (5.7) 1 (1.7) We included all sampled plots in our analyses of snag and 2007–2012 2 (3.8) 0 (0.0) tree populations, including plots subject to recent distur- 2012–2017 8 (15.1) 4 (6.7) bances such as wild or prescribed fire, for two reasons. Ganey et al. Forest Ecosystems (2021) 8:20 Page 4 of 13 Tibshirani 1993) in IBM SPSS Statistics v 23 (IBM SPSS We pooled snags or trees across plots within forest type Statistics, IBM Corp., Armonk, NY, 2015). All subse- for these comparisons. quent references to mean values for these and other pa- We included six major species groups in comparisons rameters refer to Huber’sM–estimator. For consistency, of species composition in mixed-conifer forest (white fir, we used the same methods for tree populations, al- Douglas-fir, quaking aspen, ponderosa pine, Gambel though those populations were far less skewed than snag oak, and a sixth group [Other] representing all other populations. species). We included only three species groups (ponder- The skewed distributions for snag densities also lim- osa pine, Gambel oak, and Other) in tests in ponderosa ited the utility of standard hypothesis tests. Conse- pine forest. Other species were present in such small quently, we used the bootstrapped confidence intervals amounts in this forest type that including those species discussed above to assess significance of observed differ- as separate categories resulted in multiple cells with ex- ences in snag density. For comparisons between years or pected values < 5, potentially biasing test results (Con- sampling periods, we assumed that confidence intervals over 1980). In contrast, no cells had expected values < 5 that did not overlap between pairs of years or sampling after collapsing categories. We recognized five diameter periods indicated that those years or sampling periods classes in tests involving diameter-class distributions: differed significantly from each other. For change during 20–29, 30–39, 40–49, 50–59, and ≥ 60 cm dbh. We esti- sampling periods, we assumed that confidence intervals mated percent change in individual snag species or that did not overlap zero indicated that snag (or tree) diameter classes during the 20-year study period as: density changed significantly during that period. Percent change (%) = (2017 value – 1997 value)/(1997 value) × 100. Changes in snag and tree density We estimated snag and tree density within each plot for Climate data each year, then summarized snag and tree density across Because climate can strongly affect tree mortality and plots within forest type for all years. Because snags were thus snag creation, we obtained and summarized data uniquely marked, we also were able to estimate numbers on annual precipitation (AP) and cooling degree days of new snags recruited and existing snags lost during (CDD) for the National Weather Service (NWS) weather each sampling period. Numbers of new snags may be station at Pulliam airport in Flagstaff, AZ (http://w2. slightly overestimated for some plots and sampling pe- weather.gov/climate/xmacis.php?wfo=fgz; downloaded riods, because the metal tags used to mark snags some- 14 Dec 2017). We assumed that this station, which was times melted in plots that experienced moderate- to centrally located within the study area at an elevation of high-severity fire, making it difficult to distinguish new 2136 m, provided a valid index to annual variation in snags recruited post-fire from snags present before the broad-scale climate patterns across the study area. We fire. We suspect that this bias was small, however, for restricted our analysis to the period from 1950 to 2016 two reasons. First, these disturbances affected relatively because data on CDD were not available for most years few plots during most sampling periods (Table 1). Sec- prior to 1950. We assumed that this 67-year period, ond, fires hot enough to melt the metal tags also burned which included the well documented mid-20th century most existing snags, meaning that most apparent “new” drought prominent throughout the southwest (Hereford snags in these areas likely were new. Detection rates for 2007), provided a valid index to broad-scale climate pat- standing snags in unburned plots, estimated using mark- terns across the study area in recent times. resight methodology, were very high, ranging from 0.983 CDD were calculated by NWS using a base temperature to 0.994 across major snag species represented (Ganey of 65°F. Thus, CDD was calculated only for days when the et al. 2015: Table 3). mean temperature was greater than 65°F as: For each sampling period, we estimated numbers of CDD = mean daily temperature (°F) − 65°F. snags gained and lost by plot, then summarized these We used the data from NWS directly rather than con- parameters across plots within forest type. Because num- verting temperatures to °C, because the index is useful bers of snags available at the start of a sampling period primarily in a relative context and the actual units were varied, we also estimated standardized snag loss as: not important. In contrast, we converted AP from inches Percentage of snag loss (%) = (snags lost from time t to cm for consistency with other climate literature. We to t + 5 years)/(snag density at time t) × 100. summarized data for AP and CDD in terms of the differ- ence between annual means for those parameters and Composition of snag and tree populations the 1950–2016 mean for the indicated parameter, which We compared species composition and diameter-class provides an index showing the magnitude and direction distributions of snag and tree populations among years of annual deviation from longer-term normal values. within forest type using chi-square tests (Conover 1980). Ganey et al. Forest Ecosystems (2021) 8:20 Page 5 of 13 These measures provided general indices of relative 2002. Mean loss rate also was considerably greater in wetness (AP) and warmness (CDD; http://www.weather. later sampling periods in ponderosa pine forest, but con- gov/key/climate_heat_cool). To provide a third index fidence intervals around those loss rates were wide and combining relative wetness and warmness, we used data overlapped for all periods. for Palmer’s Drought Severity Index (PDSI; Palmer 1965; Mean tree density declined by 8.3 % and 0.3 % from https://www7.ncdc.noaa.gov/CDO/CDODivisionalSelect. 2004 to 2014 in mixed-conifer and ponderosa pine forest, jsp#; downloaded 18 March 2019) for the region con- respectively, but confidence intervals around mean esti- tainingour studyarea(climatedivision2; https://www. mates were wide and overlapped between years in both esrl.noaa.gov/psd/data/usclimdivs/data/map. forest types (Fig. 1e). Mean tree densities in mixed-conifer html#Arizona). This index uses temperature data and a forest in 2004 and 2014 were approximately ten and five physical water balance model to estimate potential times greater than snag densities in 2002 and 2012, re- evapotranspiration and relative drought severity (Palmer spectively (the closest years in which snags were sampled; 1965, see also https://climatedataguide.ucar.edu/climate- compare Fig. 1a and e). In ponderosa pine forest, mean data/palmer-drought-severity-index-pdsi). Index values tree density was approximately 20 times greater than snag range from − 10 to 10, with lower and higher values in- density in these same among-year comparisons. Conse- dicating drier and wetter conditions, respectively quently, although the observed level of tree mortality was (Palmer 1965). sufficient to significantly alter snag density across time, it did not significantly alter tree density. Note, however, that Results our initial sample of tree populations in 2004 likely oc- Changes in snag and tree density curred after the major mortality pulse during the study, In all years, mean snag densities in mixed-conifer forest which appeared to peak from 2002 to 2003 (Negron et al. were approximately five times greater than mean dens- 2009;Kane et al. 2014). ities in ponderosa pine forest (Fig. 1a). Within forest type, mean snag density was similar in 1997 and 2002, Snag recruitment and climate increased significantly between 2002 and 2007, and de- As discussed above, recruitment of new snags was rela- clined slightly after 2007. In mixed-conifer forest, mean tively low and similar for three of the four sampling pe- snag density remained significantly elevated from 2007 riods, but increased significantly during the period from to 2017 relative to earlier years. In contrast, the confi- 2002 to 2007 in both forest types (Fig. 1b). This spike in dence interval for mean density in 2017 in ponderosa snag recruitment followed an extremely warm and dry pine forest overlapped with all earlier years, indicating year (2002). In the summer of 2002, PDSI reached its convergence toward pre-2007 levels of snag density from lowest value for our study period (Fig. 2a), as well as its 2012 to 2017. Relative to snag density in 1997, peak most extreme value in the past 100 years (Koepke et al. mean density in 2007 was 86 % and 79 % greater in 2010). The year 2002 was not the driest year during our mixed-conifer and ponderosa pine forest, respectively, study (Fig. 2b), but was by far the hottest year during and mean snag density in 2017 was still 83 % and 51 % the study (Fig. 2c), suggesting that the extremely low greater than 1997 density. PDSI in 2002 was driven more by extremely high Mean recruitment of new snags spiked markedly be- temperature than by low moisture. PDSI also exceeded tween 2002 and 2007 in both forest types but was simi- the threshold for extreme drought in the summers of lar among the other sampling periods (Fig. 1b). Mean 2007 and 2015 (Fig. 2a), but did not reach the extreme snag recruitment from 2002 to 2007 was more than level reached in 2002, did not remain low as long as dur- double recruitment in any other sampling period in ing 2002–2003, and 2007 and 2015 overall were not ex- mixed-conifer forest, and differed significantly from tremely dry or warm relative to 2002 (Fig. 2). Snag mean recruitment during all other sampling periods in recruitment did not spike following these later years. this forest type. Mean snag recruitment from 2002 to 2007 was almost double any other sampling period in Composition and structure of snag and tree populations ponderosa pine forest, but due to wide confidence inter- Species composition of snag populations differed signifi- vals differed significantly only from mean recruitment cantly among years in both forest types (Fig. 3a; mixed- from 1997 to 2002. conifer: χ = 905.2, df = 20, P < 0.001, n = 14,857; ponder- Snag loss rates increased with every consecutive sam- osa pine: χ = 24.5, df =8, P < 0.001, n = 3,287). In mixed- pling period in both forest types, and this pattern was conifer forest, the primary temporal trend was a large in- observed both for absolute numbers of snags lost and crease in white fir snags, with smaller increases in all for standardized loss rates (Fig. 1c and d). In mixed- other species. In ponderosa pine forest, proportional conifer forest, mean loss rate was significantly greater representation decreased for ponderosa pine and in- from 2007 to 2012 and 2012 to 2017 than from 1997 to creased for Gambel oak snags. Ganey et al. Forest Ecosystems (2021) 8:20 Page 6 of 13 Fig. 1 Trends in snag and tree populations in northern Arizona mixed-conifer and ponderosa pine forest, 1997–2017. a Snag density by year; snags were sampled every five years. b Density of newly-recruited snags by 5-year interval between sampling occasions. Each year shown indicates the end of the 5-year interval represented. c Density of snags lost during each 5-year interval between sampling occasions. d Percentage of existing snags lost during each 5-year interval between sampling occasions. e Tree density by year; tree populations were sampled only in 2004 and 2014. Shown in all panels are Huber’sM– estimator (solid line), a generalized maximum–likelihood estimator that yields robust estimates of central tendency in distributions containing outliers (Huber and Ronchetti 2009), and lower and upper bounds (dashed lines) of bias-corrected confidence intervals around M Much of the change in species composition was driven 2007. Species composition of snags lost generally was by patterns in recruitment of new snags. In terms of dominated by the same species as newly-recruited snags, magnitude, recruitment of new snags was dominated by but in lower numbers. Increases in snag loss lagged in- white fir in mixed-conifer forest and by ponderosa pine creases in snag recruitment by 5–10 years (Fig. 3c). in ponderosa pine forest (Fig. 3b). Timing of recruitment Diameter-class distributions of snags also differed sig- of new snags varied among species in mixed-conifer for- nificantly among years in both forest types (Fig. 4a, est, peaking from 2002 to 2007 for white fir and quaking mixed-conifer: χ = 67.2, df = 16, P < 0.001, n = 14,857; aspen, from 2007 to 2012 for ponderosa pine, and in- ponderosa pine: χ = 37.3, df =16, P = 0.002, n = 3,476). creasing steadily throughout the study for Douglas-fir. In In both forest types, numbers of snags increased across ponderosa pine forest, recruitment of both ponderosa all size classes, with by far the greatest increase occur- pine and Gambel oak snags peaked between 2002 and ring in the smallest size classes. These changes largely Ganey et al. Forest Ecosystems (2021) 8:20 Page 7 of 13 Fig. 2 Trends in climate in northern Arizona, 1997–2017. a Palmer’s Drought Severity Index (PDSI; Palmer 1965;Alley 1984; Heim 2002)for the region containing our study area by year. This index ranges from − 10 to 10, with higher and lower values indicating wetter and drier conditions, respectively. Horizontal reference lines indicate thresholds for severe (–3.0; dashed line) and extreme (–4.0; solid line) drought, respectively (Palmer 1965). b Difference between annual precipitation (AP) during the study and the 1950 through 2016 mean at the National Weather Service (NWS) station located at Pulliam airport, Flagstaff, AZ, elevation 2136 m. Values above and below the horizontal reference line indicated years that were wetter or drier than average, respectively. c Difference between annual Cooling Degree Days (CDD) during the study and the 1950 through 2016 mean at the National Weather Service (NWS) station located at Pulliam airport. Values above and below the horizontal reference line indicated years that were warmer or cooler than average, respectively. All data sources are described in text were driven by recruitment patterns for new snags, with large numbers of snags recruited in the smaller size clas- ses (Fig. 4b). As with species composition, snags lost were dominated by the same size classes as snags re- cruited, but in lower numbers and with peaks in snag loss lagging peaks in snag recruitment by 5–10 years (Fig. 4c). Species composition of tree populations differed sig- nificantly between 2004 and 2014 only in ponderosa pine forest (χ = 7.105, df =2. P = 0.028; mixed–conifer forest χ = 6.640, df =5, P = 0.359). Ponderosa pine in- creased from 87.4 to 90.4 % of total trees in ponderosa pine forest over this period, whereas proportions of all other species decreased. Size-class distributions for trees did not differ significantly between 2004 and 2014 in ei- ther forest type (mixed-conifer forest χ = 8.773, df =4, P = 0.066; ponderosa pine χ = 5.241, df =4. P = 0.262), and were numerically dominated by trees in the smaller size classes in both years (results not shown here). Discussion Snag numbers increased considerably over our 20-year study period in mixed-conifer forest, with a smaller in- crease in ponderosa pine forest (Fig. 1a). Increased snag density was driven largely by a pulse of snag recruitment (tree mortality) that occurred between 2002 and 2007, with snag recruitment returning to pre-pulse levels in subsequent sampling periods (Fig. 1b). Thus, increases in snag recruitment appeared more indicative of a short- term tree mortality “event” than of a longer-term pattern of elevated tree mortality (with the possible exception of Douglas-fir, Fig. 3a). This mortality pulse appeared to be drought-mediated. Our evaluation of the influence of climate on tree mor- tality was correlative and cannot demonstrate cause and effect relationships. Nevertheless, two aspects of this mortality pulse appear most consistent with drought ef- fects. The first relates to the timing of the mortality Ganey et al. Forest Ecosystems (2021) 8:20 Page 8 of 13 Fig. 3 Trends in species composition of snag populations in northern Arizona mixed-conifer (n = 53 plots) and ponderosa pine (n = 60 plots) forest. a Number of snags by species and year. b Number of newly-recruited snags by species and 5-year period ending in the indicated year. c Number of snags lost within the 5- year period ending in the indicated year, by species. Species acronyms: ABCO = white fir, PIPO = ponderosa pine, POTR = quaking aspen, PSME = Douglas-fir, and QUGA = Gambel oak. Not shown is a composite group comprised of all other species. Snags were pooled across plots within forest type pulse, which immediately followed a single extremely warm and dry year in 2002 (Fig. 2; see also Negron et al. 2009; Kane et al. 2014) embedded within a longer-term megadrought (Williams et al. 2020). The second factor suggesting that mortality was drought-mediated was the spatial synchrony of mortality across plots within both forest types plots. This spatial synchrony in mortality ap- peared more consistent with widespread drought effects than with effects of more spatially limited disturbance events such as fire or insect outbreaks. Those other dis- turbance agents certainly contributed to mortality in some areas, but their effects were not as widespread, and likely were exacerbated by drought. The observed mortality pulse appeared to be triggered by the extreme climate year in 2002. PDSI reached its lowest level in over 100 years in 2002 (Kopeke et al. 2010), and a tree-ring reconstruction indicated that moisture stress during 2002 reached one of the highest levels in the past 1400 years (Salzer and Kipfmueller 2005; see also Williams et al. 2020). The extremely low PDSI in 2002 appeared to be driven more by extremely hot conditions than by extremely dry conditions; 2002 was by far the hottest year during the study but was not the driest year (Fig. 2). Thus, conditions in 2002 ap- peared representative of what Breshears et al. (2005) de- fined as “global climate change type drought”, characterized by dry conditions coupled with extreme heat. Climate change models suggest that such years will occur more frequently in the southwestern U. S. in the future (Seager et al. 2007; Garfin et al. 2014). Many later years during the study were either warmer or drier than average, or both, and annual PDSI values suggested some level of drought stress in 16 of 20 years during the study (Fig. 2a). Those later years and periods did not trigger the same levels of tree mortality seen fol- lowing 2002, however. There are at least four possible explanations for this apparent discrepancy. First, the ex- treme conditions experienced during 2002, especially with regard to temperature (Adams et al. 2009) may have pushed many trees beyond a physiological thresh- old for survival (McDowell et al. 2008, 2016; Koepke et al. 2010) that was not reached again in subsequent years. Second, the lack of increased tree mortality fol- lowing later years that were almost as extreme Ganey et al. Forest Ecosystems (2021) 8:20 Page 9 of 13 Fig. 4 Trends in diameter-class distribution of snag populations in northern Arizona mixed-conifer (n = 53 plots) and ponderosa pine (n = 60 plots) forest. a Number of snags by diameter class and year; b Number of newly-recruited snags by diameter class and 5-year period ending in the indicated year; c Number of snags lost within the 5-year period ending in the indicated year, by diameter class. Snags were pooled across plots within forest type climatically as 2002 could indicate that the most vulner- able trees died in the post-2002 mortality event, leaving tree populations dominated by trees that were more re- silient to drought. Third, remnant trees may not have been more drought-resistant, but drought stress may have been alleviated in subsequent years by previous mortality that reduced tree density and competition for available precipitation (Peet and Christiansen 1987). Fourth, bark beetle populations may have been higher following the extreme climate year in 2002 than in sub- sequent years, swamping the defenses of many trees (e.g., Negron et al. 2009; Kane et al. 2014). At present we cannot distinguish between these hy- potheses, which are not-mutually exclusive. They clearly have very different implications for future trends in for- est structure and composition, however. For example, if hypothesis 1 is correct, we can expect to see similar levels of tree mortality whenever climatic extremes like 2002 occur. In contrast, both hypotheses 2 and 3 suggest that future mortality may be lower even in similarly ex- treme climate years. Hypothesis 4 is particularly intri- guing in the context of understanding future trends, as it may suggest feedback loops that regulate drought- mediated mortality. Bark beetle populations are known to increase in response to drought (e.g., Fettig et al. 2007; Huang et al. 2020), exacerbating mortality in drought-stressed trees (Huang et al. 2020). This increase in bark beetles, along with the associated increase in tree mortality, provides increased food and nest resources for bark insectivores, many of which nest in cavities in snags, resulting in short-term (2–5 years) increases in populations of woodpeckers and other bark insectivores (Koplin 1969; Edworthy et al. 2011; Saab et al. 2019). These populations of bark insectivores then aid in regu- lating bark beetle numbers (Koplin 1972; Fayt et al. 2005). This raises the possibility that severe drought years following 2002 may have seen lower mortality partly because of these feedback loops operating follow- ing earlier spikes in snag recruitment. Regardless of the explanation for the observed mor- tality event, that event was large enough to signifi- cantly alter species composition and structure of snag populations during the study. All species of snags showed increased recruitment during the study, but Ganey et al. Forest Ecosystems (2021) 8:20 Page 10 of 13 the magnitude, timing, and duration of that increase primarily by snag recruitment, they were mediated by varied among species. For example, white fir and patterns in snag longevity. Longevity was related pri- quaking aspen, the species least resistant to drought marily to snag species, diameter, and height (Ganey stress (Ganey and Vojta 2011; after Niinemets and et al. 2015). Standing rates for white fir and ponder- Vallardes 2006), responded strongly and quickly in osa pine snags were lower than for other major spe- mixed-conifer forest, with relatively large increases in cies (Ganey et al. 2015:Table 2),which tended to snag recruitment peaking in the sampling period from work opposite the large increases in recruitment for 2002to2007and returningtopre-2002levelsinsub- those two species (Fig. 3b). And standing rates for sequent periods (Fig. 3b). In contrast, the more smallerdiameter snags were lowerthanthose forlar- drought-resistant Douglas-fir did not respond as ger diameter snags across species (Ganey at al. 2015: strongly from 2002 to 2007, but instead snag recruit- Table 5). Again, this suggests that patterns in snag ment increased over subsequent sampling periods. loss worked opposite to patterns in snag recruitment, Ponderosa pine also responded fairly strongly in which was concentrated in the smaller size classes mixed-conifer forest from 2002 to 2007, but snag re- (Fig. 4b). Snag loss rates also may have been influ- cruitment for this species did not peak until the enced by bark beetle activity, especially later in the period from 2007 to 2012 and remained elevated study. Snags resulting from bark beetle infestations from 2012 to 2017. appear to fall relatively quickly (Chambers and Mast These differences in magnitude and timing of 2014; Rhoades et al. 2020), and this also may have species-specific snag recruitment resulted in signifi- contributed to the increase in snag loss rates observed cant changes in species composition of snag popula- over time in this study (see Fig. 1cand d). tions. In mixed-conifer forest, snag populations in The effect of these changes in composition and struc- 1997 and 2002 were co-dominated by white fir, pon- ture of snag populations on snag use by native wildlife is derosa pine, and Gambel oak in approximately equal unknown. The increase in snag abundance should provide numbers (Fig. 3a). By 2007, those populations were increased foraging opportunities for native birds (Bull dominated by white fir snags, and white fir continued et al. 1997), at least in the short term. Effects on nesting to dominate these populations through 2017, although birds and/or roosting bats are more unclear. Snag popula- it declined in abundance in later years. In contrast, tions in both forest types became increasingly dominated relative abundance of Douglas-fir increased across all by smaller snags during the study, and populations in years, peaking in 2017. This pattern may indicate mixed-conifer forest became increasingly dominated by that, unlike some other species, Douglas-fir was white fir snags. These snags are less likely to be used by responding more to cumulative drought stress than to cavity nesting birds or roosting bats than are larger snags a one-timemortalityevent. or snags of other species (Rabe et al. 1998; Ganey and In the less species diverse ponderosa pine forest type, Vojta 2004; Solvesky and Chambers 2009). They also gen- snag populations were dominated by ponderosa pine in erally decay and fall more rapidly than do larger snags or all years, with Gambel oak the only other species present other species of snags (Ganey et al. 2015). The increased in considerable numbers. Recruitment of ponderosa pine abundance of such snags may have contributed to the ob- snags peaked earlier in the warmer and drier ponderosa served increase in snag loss rates across time in both pine forest (2002–2007; Fig. 3b) than in mixed-conifer forest types during the study (Fig. 1c), despite the large forest. That increase largely was offset by increased loss number of “new” snags in these populations. Thus, al- rates of ponderosa pine snags following 2007, however though snag numbers increased during the study, and that (Fig. 3c). increase likely provided a temporary increase in foraging Diameter distributions of snag populations also chan- substrates and food resources for birds, many of the re- ged significantly during the study in both forest types cruited snags may represent an ephemeral resource that is (Fig. 4a). Numbers of snags recruited increased for all not useful to cavity-nesting birds (Ganey and Vojta 2004). size classes from 2002 to 2007, but the magnitude of in- Note that these rapidly-decaying and falling snags contrib- crease was far greater for the smallest size classes than ute to increases in log populations, however, and this may for the largest size classes. This imbalance was partially benefit native wildlife that feed on, hide in, or den in or but not entirely offset by higher loss rates for snags in under down logs (Bull et al. 1997; Ganey and Vojta 2017). the smallest size classes (Fig. 4c). The net result was in- In contrast to snag populations, tree mortality did creased relative abundance of snags in the smallest size not appear to be adequate to significantly alter tree classes and decreased relative abundance of snags in the populations during the study. Our ability to evaluate largest size classes. changes in tree populations during the study period Note that, although these changes in species com- was hampered by the more limited sampling of those position and diameter distributions were driven populations, however, and especially by the timing of Ganey et al. Forest Ecosystems (2021) 8:20 Page 11 of 13 sampling. We assessed change in snag populations observed following the extreme climatic year in 2002, over a 20-year period, but assessed change in tree coupled with climate models predicting that such years populations only over a 10-year period embedded will become more common in the future (Seager et al. within that longer period. More importantly, we ini- 2007; Garfin et al. 2014; Williams et al. 2020), supports tially sampled tree populations in 2004, after the peak calls for management to increase resilience and drought- of the tree mortality pulse triggered by extreme resistance in these forest types (Millar et al. 2007; Ste- drought conditions in 2002 (2002–2003; Negron et al. phens et al. 2013). 2009; Ganey and Vojta 2011;Kane etal. 2014). Thus, Acknowledgements tree populations likely showed greater change during We thank J. Jenness, G. Martinez, M. Stoddard, B. Strohmeyer, and R. White the 20-year study period than we were able to dem- for assisting in establishing plots, and L. Doll, D. and N. Ganey, and C. Vojta for help with plot sampling. J. Ellenwood, B. Higgins, K. Menasco, C. Nelson, onstrate. Even given these limitations, however, the and G. Sheppard (Kaibab National Forest) and C. Beyerhelm, A. Brown, H. much greater tree densities relative to snag densities Green, T. Randall-Parker, C. Taylor, and M. Whitney (Coconino National Forest) clearly ensure that, for a given mortality level, relative assisted with initial plot selection. effects will be much lower for tree populations than Authors' contributions for snag populations. JG conceived the study, with input from JI, AI, and SV. JG, JI, AI, and SV established the plots and collected the data. JG analyzed the data. All authors contributed to writing the manuscript and read and approved the Conclusions final manuscript. Warming climates have profoundly affected forests throughout the world (Adams et al. 2009; Allen et al. Funding Funding was provided by the USDA Forest Service Rocky Mountain Research 2010), and climate models predict further warming and Station. The funder played no official role (beyond employing the authors) in drying in many areas, including the southwestern United the design of the study and collection, analysis, and interpretation of data States (e.g., Seager et al. 2007; Stocker et al. 2013; Garfin and in writing the manuscript. et al. 2014). Our results demonstrate the dynamic nature Availability of data and materials of snag populations and document the drivers associated The datasets used and/or analysed during the current study are available with rapid changes in those populations over time. That from the corresponding author on reasonable request. is, a drought event in 2002 caused spatially widespread Declarations tree mortality. This mortality significantly increased re- cruitment of new snags, but many of the newly-recruited Ethics approval snags were small in diameter and/or white fir snags. Not applicable. These snags decay and fall faster than larger snags and Consent for publication other species of snags (Ganey et al. 2015), and this con- Not applicable. tributed to an increasing snag loss rate over time. This Competing interests rapid loss of snags did not entirely balance the increased The authors declare that they have no competing interests. recruitment of new snags following the 2002 drought, resulting in higher snag densities after 2007 compared to Received: 29 June 2020 Accepted: 23 February 2021 prior periods, particularly in mixed conifer forests. Sig- nificant questions remain, however, including: (1) how References long do these changes in abundance, structure, and Adams HD, Guardiola-Claramonte M, Barron-Gafford GA, Villegas JC, Breshears composition of snag populations persist? (2) how com- DD, Zou CB, Troch PA, Huxman TE (2009) Temperature sensitivity of drought- induced tree mortality portends increased regional die-off under global- mon will similarly extreme climate years be in the fu- change-type drought. PNAS 106:7063–7066 ture? and (3) will such years produce mortality pulses Allen CD, Macalady AK, Chenchouni H, Bachelet D, McDowell N, Vennetier MM, similar to the one observed in this study, or might future Kitzberger T, Rigling A, Breshears DD, Hogg EH, Gonzalez P, Fensham R, Zhang Z, Castro J, Demidova N, Lim J-H, Allard G, Running SW, Semerci A, patterns of tree mortality be modified by past mortality Cobb N (2010) A global overview of drought and heat-induced tree events that alter forest structure and composition, and mortality reveals emerging climate change risks for forests. Forest Ecol competition for resources? Answering these questions Manag 259:660–684 Alley WA (1984) The Palmer drought severity index: limitation and assumptions. J will require long-term studies, but is critical to under- Clim Appl Meteorol 23:1100–1109 standing how changing climate may affect trends in Bentz BJ, Régnière J, Fettig CJ, Hansen EM, Hayes JL, Hicke JA, Kelsey RG, Negrón structure and composition of future snag populations JF, Seybold SJ (2010) Climate change and bark beetles of the Western United States and Canada: Direct and indirect effects. Bioscience 60:602–613 and the tree populations from which they derive. The Bernardos DA, Chambers CL, Rabe MJ (2004) Selection of Gambel oak roosts by U.S. Forest Service Forest Inventory and Analysis pro- southwestern Myotis in ponderosa pine-dominated forests, northern Arizona. gram (Smith 2002), which repeatedly samples plots J Wildlife Manag 68:595–601 Breshears DD, Cobb NS, Rich PM, Price KP, Allen CD, Balice RG, Romme WH, through time across the landscape, may provide a means Kastens JH, Floyd ML, Belknap J, Anderson JJ, Myers OB, Meyer CW (2005) to accomplish the longer-term monitoring needed to an- Regional vegetation die-off in response to global-change-type drought. swer these questions. In the meantime, the mortality PNAS 102:15144–15148 Ganey et al. Forest Ecosystems (2021) 8:20 Page 12 of 13 Brown DE, Lowe CH, Pase CP (1980) A digitized systematic classification for Hartmann H (2020) Tree defence and bark beetles in a drying world: carbon ecosystems with an illustrated summary of the natural vegetation of North partitioning, functioning and modelling. New Phytol 225:26–36 America. Gen Tech Rep RM-73. U.S Department of Agriculture, Forest Service, Huber PJ, Ronchetti KW (2009) Robust Statistics, Second Ed. John Wiley & Sons, Rocky Mountain Forest and Range Experiment Station, Fort Collins, 93 Inc., Hoboken Bull EL, Parks CG, Torgersen TR (1997) Trees and logs important to wildlife in the Kane JM, Kolb TE, McMillin JD (2014) Stand-scale tree mortality factors differ by interior Columbia River Basin. USDA Forest Service General Technical Report site and species following drought in southwestern mixed–conifer forests. PNW-GTR-391 Forest Ecol Manag 330:171–182 Chambers CL, Mast JN (2014) Snag dynamics and cavity excavation after bark beetle Koepke DF, Kolb TE, Adams HD (2010) Variation in woody plant mortality and outbreaks in southwestern ponderosa pine forests. Forest Sci 60:713–723 dieback from severe drought among soils, plant groups, and species within a Conover WJ (1980) Practical nonparametric statistics, 2nd ed. John Wiley & Sons, northern Arizona ecotone. Oecologia 163:1079–1090 New York Koplin JR (1969) The numerical response of woodpeckers to insect prey in a Conway CJ, Martin TE (1993) Habitat suitability for Williamson’s sapsuckers in subalpine forest in Colorado. Condor 71:436–438 mixed-conifer forests. J Wildlife Manag 57:322–328 Koplin JR (1972) Measuring predator impact of woodpeckers on spruce beetles. J Cunningham JB, Balda RP, Gaud WS (1980) Selection and use of snags by Wildlife Manag 36:308–320 secondary cavity-nesting birds of the ponderosa pine forest. USDA Forest Laudenslayer WF Jr, Shea PJ, Valentine BE, Weatherspoon CP, Lisle TE (tech. cords. Service Research Paper RM-222 ) (2002) Proceedings of the symposium on the ecology and management of Davis KT, Dobrowski SZ, Higuera PE, Holden ZA, Veblen TT, Rother MT, Parks SA, dead wood in western forests. Gen Tech Rep PSW-GTR-181. Albany, CA: U.S. Salag A, Manet MP (2019) Wildfires and climate change push low-elevation Department of Agriculture, Forest Service, Pacific Southwest Research forests across a critical climate threshold for tree regeneration. PNAS 116: Station. 949 6193–6198 McComb W, Lindenmayer D (1999) Dying, dead, and down trees. In: Hunter ML Edworthy AB, Drevera MC, Martin K (2011) Woodpeckers increase in abundance Jr (ed) Maintaining Biodiversity in Forest Ecosystems. Cambridge University but maintain fecundity in response to an outbreak of mountain pine bark Press, Cambridge beetles. Forest Ecol Manag 261:203–210 McDowell NG, Pockman WT, Allen CD, Breshears DD, Cobb N, Kolb T, Plaut J, Efron B, Tibshirani R (1993) An Introduction to the Bootstrap. Chapman & Hall/ Sperry J, West A, Williams DG, Yepez EA (2008) Tansley Review: mechanisms CRC Monographs on Statistics and Applied Probability 57. Boca Raton, FL, of plant survival and mortality during drought: Why do some plants survive USA. 456 while others succumb to drought? New Phytol 178:719–739 Fayta P, Machmerb MM, Steeger C (2005) Regulation of spruce bark beetles by McDowell NG, Williams AP, Xu C, Pockman WT, Dickman LT, Sevanto S, Pangle R, woodpeckers— a literature review. Forest Ecol Manag 206:1–14 Limousin J, Plaut J, Mackay DS, Ogee J, Domec JC, Allen CD, Fisher RA, Jiang Fettig CJ, Klepzig KD, Billings RF, Munson AS, Nebeker TE, Negron JF, Nowak JT X, Muss JD, Breshears DD, Rauscher SA, Koven C (2016) Multi-scale (2007) The effectiveness of vegetation management practices for prevention predictions of massive conifer mortality due to chronic temperature rise. Nat and control of bark beetle infestations in coniferous forests of the western Clim Change 6. doi:https://doi.org/10.1038/NCLIMATE2873 and southern United States. Forest Ecol Manag 238:24 – 53 McKenzie D, Gedalof Z, Peterson DL, Mote P (2004) Climatic change, wildfire, and Ganey JL (1999) Snag density and composition of snag populations on two conservation. Conserv Biol 18:890–902 National Forests in northern Arizona. Forest Ecol Manag 117:169–178 Millar CL, Stephenson NL, Stephens SL (2007) Climate change and forests of the future: managing in the face of uncertainty. Ecol Appl 17:2145–2151 Ganey JL, Vojta SC (2004) Characteristics of snags containing excavated cavities in northern Arizona mixed-conifer and ponderosa pine forests. Forest Ecol Mueller RC, Scudder CM, Porter ME, Trotter RTIII, Gehring CA, Whitham TG (2005) Manag 199:323–332 Differential tree mortality in response to severe drought: evidence for long- Ganey JL, Vojta SC (2005) Changes in snag populations in northern Arizona term vegetation shifts. J Ecol 93:1085–1093 mixed-conifer and ponderosa pine forests, 1997–2002. Forest Sci 51:396–405 Negron JF, McMillin JD, Anhold JA, Coulson D (2009) Bark beetle-caused Ganey JL, Vojta SC (2011) Tree mortality in drought–stressed mixed–conifer and mortality in a drought–affected ponderosa pine landscape in Arizona, USA. ponderosa pine forests, Arizona. Forest Ecol Manag 261:162–168 Forest Ecol Manag 257:1353–1362 Ganey JL, Vojta SC (2012) Trends in snag populations in drought-stressed mixed- Niinemets U, Vallardes F (2006) Tolerance to shade, drought, and waterlogging of conifer and ponderosa pine forests, 1997–2007. Int J Forest Res. doi:https:// temperate Northern Hemisphere trees and shrubs. Ecol Monogr 76:521–547. doi.org/10.1155/2012/529197 http://www.esapubs.org/Archive/mono/M076/020/appendia-A.htm. Accessed 11 April 2008 Ganey JL, Vojta SC (2014) Trends in snag populations in northern Arizona mixed– Palmer WC (1965) Meteorological drought. U. S. Weather Bureau Research Paper conifer and ponderosa pine forests, 1997–2012. USDA Forest Service 45. 58. https://www.ncdc.noaa.gov/temp-and-precip/drought/docs/palmer. Research Paper RMRS-RP-105 pdf. Accessed 19 Dec 2017 Ganey JL, Vojta SC (2017) Comparative trends in log populations in northern Arizona mixed–conifer and ponderosa pine forests following severe drought. Peet RK, Christensen NL (1987) Competition and tree death. Bioscience 37:586–595 West North Am Nat 77:281–292 Rabe MJ, Morrell TE, Green H, deVos JC Jr, Miller CR (1998) Characteristics of Ganey JL, White GC, Jenness JS, Vojta SC (2015) Mark-recapture estimation of ponderosa pine snag roosts used by reproductive bats in northern Arizona. J snag standing rates in northern Arizona mixed-conifer and ponderosa pine Wildlife Manag 62:612–621 forests. J Wildlife Manag 79:1369–1377 Rhoades CC, Hubbard RM, Hood PR, Starr BJ, Tinker DB, Elder K (2020) Snagfall the first decade after severe bark beetle infestation of high-elevation forests Garfin G, Franco G, Blanco H, Comrie A, Gonzalez P, Piechota T, Smyth R, Waskom in Colorado, USA. Ecol Appl e02059 R (2014) Ch. 20: Southwest. In: Melillo JM, Richmond TC, Yohe GW (eds) Climate Change Impacts in the United States: The Third National Climate Saab VA, Latif QS, Dresser MA, Dudley JG (2019) Woodpecker nest survival, Assessment. U.S. Global Change Research Program, 462 density, and a pine beetle outbreak. J Wildlife Manag 83:1387–1400 Harmon ME, Franklin JF, Swanson FJ, Sollins P, Gregory SV, Lattin JD, Anderson Salzer MW, Kipfmueller KF (2005) Reconstructed temperature and precipitation NH, Cline SP, Aumen NG, Sedell JR, Lienkamper GW, Cromack K Jr, Cummins on a millennial timescale from tree-rings in southern Colorado Plateau, USA. KW (1986) Ecology of coarse woody debris in temperate ecosystems. Adv Clim Change 70:465–487 Ecol Res 15:133–302 Scott VE (1978) Characteristics of ponderosa pine snags used by cavity-nesting Heim RH Jr (2002) A review of twentieth-century drought indices used in the birds. J Forest 76:26–28 United States. Bull Am Meteorol Soc 83:1149–1165 Seager R, Ting MF, Held IM, Kushmir Y, Lu J, Vecchi G, Huang H, Harnick N, Hereford R (2007) Climate variation at Flagstaff, Arizona-1950 to 2007. U.S. Leetmaa A, Lau N, Li C, Velez J, Naik N (2007) Model projections of an Geological Survey Open-File Report 2007 – 1410. P 17. https://www.researchga imminent transition to a more arid climate in southwestern United States. te.net/profile/Richard_Hereford/publication/282031501_Climate_Variation_at_ Science 316:1181–1184 Flagstaff_Arizona--1950_to_2007/links/5601c0e008aed9851827cb99.pdf. Smith WB (2002) Forest inventory and analysis: a national inventory and Accessed 19 Dec 2017 monitoring program. Environ Poll 116:S233–S242 Huang J, Kautz M, Trowbridge AM, Hammerbacher A, Raffa KF, Adams HD, Solvesky BG, Chambers CL (2009) Roosts of Allen’s lappet-browed bat in northern Goodsman DW, Xu C, MeddensA JH, Kandasamy D, Gershenzon J, Seidl R, Arizona. J Wildlife Manag 73:677–682 Ganey et al. Forest Ecosystems (2021) 8:20 Page 13 of 13 Stephens SL, Agee JK, Fulé PZ, North MP, Romme WH, Swetnam TW, Turner MG (2013) Managing forests and fire in changing climates. Science 342:41–42 Stocker TF, Qin D, Plattner GK, Tignor MM, Allen SK, Boschung J, Nauels A, Xia Y, Bex V, Midgley PM (2013) IPCC 2013: Climate change 2013: the physical science basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge Thomas JW, Anderson RG, Maser C, Bull EL (1979) Snags. In: Thomas J W (tech. ed.) Wildlife habitats in managed forests: The blue mountains of Oregon and Washington. USDA Agricultural Handbook 553, 60–77. Washington, D.C Wan H-Y, Cushman SA, Ganey JL (2019) Recent and projected wildfire trends across the ranges of three spotted owl subspecies under climate change. Front Ecol Environ 7:37. doi:https://doi.org/10.3389/fevo.2019.00037 Williams AP, Cook ER, Smerdon JE, Cook BI, Abatzoglou JT, Bolles K, Baek SH, Badger AM, Livneh B (2020) Large contribution from anthropogenic warming to an emerging North American megadrought. Science 368:314–318 Woldendorp G, Keenan RJ (2005) Coarse woody debris in Australian forest ecosystems: a review. Aust Ecol 30:834–843 Wu H, Franklin SB, Liu J, Lu Z (2017) Relative importance of density dependence and topography on tree mortality in a subtropical mountain forest. Forest Ecol Manag 384:169–179 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png "Forest Ecosystems" Springer Journals

Twenty years of drought‐mediated change in snag populations in mixed‐conifer and ponderosa pine forests in Northern Arizona

Loading next page...
 
/lp/springer-journals/twenty-years-of-drought-mediated-change-in-snag-populations-in-mixed-dd3gTQbMXp

References (6)

  • HD Adams (2009)

    7063

    PNAS, 106

  • CD Allen (2010)

    660

    Forest Ecol Manag, 259

  • BJ Bentz (2010)

    602

    Bioscience, 60

  • WA Alley (1984)

    1100

    J Clim Appl Meteorol, 23

  • DA Bernardos (2004)

    595

    J Wildlife Manag, 68

  • DD Breshears (2005)

    15144

    PNAS, 102

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2021
eISSN
2197-5620
DOI
10.1186/s40663-021-00298-9
Publisher site
See Article on Publisher Site

Abstract

Background: Snags (standing dead trees) are important biological legacies in forest systems, providing numerous resources as well as a record of recent tree mortality. From 1997 to 2017, we monitored snag populations in drought-influenced mixed-conifer and ponderosa pine (Pinus ponderosa) forests in northern Arizona. Results: Snag density increased significantly in both forest types. This increase was driven largely by a pulse in snag recruitment that occurred between 2002 and 2007, following an extreme drought year in 2002, with snag recruitment returning to pre-pulse levels in subsequent time periods. Some later years during the study also were warmer and/or drier than average, but these years were not as extreme as 2002 and did not trigger the same level of snag recruitment. Snag recruitment was not equal across tree species and size classes, resulting in significant changes in species composition and size-class distributions of snag populations in both forest types. Because trees were far more abundant than snags in these forests, the effect of this mortality pulse on tree populations was far smaller than its effect on snag populations. Snag loss rates increased over time during the study, even though many snags were newly recruited. This may reflect the increasing prevalence of white fir snags and/or snags in the smaller size classes, which generally decay faster than snags of other species or larger snags. Thus, although total numbers of snags increased, many of the newly recruited snags may not persist long enough to be valuable as nesting substrates for native wildlife. Conclusions: Increases in snag abundance appeared to be due to a short-term tree mortality “event” rather than a longer- term pattern of elevated tree mortality. This mortality event followed a dry and extremely warm year (2002) embedded within a longer-term megadrought. Climate models suggest that years like 2002 may occur with increasing frequency in the southwestern U.S. Such years may result in additional mortalitypulses, whichinturnmay strongly affect trajectories in abundance, structure, and composition of snag populations. Relative effects on tree populations likely will be smaller, but, over time, also could be significant. Keywords: Climate change, Drought, Monitoring, Snag abundance, Snag creation, Snag dynamics, Species composition, Tree mortality Background (Adams et al. 2009; Allen et al. 2010), and climate models Understanding the effects of climate change on structure predict further warming and drying in many areas (e.g., and composition of forest ecosystems presents a major Seager et al. 2007; Stocker et al. 2013;Garfinet al. 2014) challenge for researchers and managers in the twenty-first suggesting that changing climates will continue to affect century (Millar et al. 2007). Warming climates already these systems. Forest systems can be affected directly have profoundly affected forests throughout the world through climate-mediated mortality (Breshears et al. 2005; Mueller et al. 2005; Adams et al. 2009), or indirectly * Correspondence: joe.ganey@nau.edu through altered disturbance regimes that result in in- US Forest Service, Rocky Mountain Research Station, 2500 S. Pine Knoll, AZ creased mortality (e.g., McKenzie et al. 2004; Bentz et al. 86001 Flagstaff, USA © The Author(s). 2021 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Ganey et al. Forest Ecosystems (2021) 8:20 Page 2 of 13 2010; Wan et al. 2019) or reduced regeneration (Davis et al. 2015). Here, we expanded on this earlier work to et al. 2019). Because drought-mediated mortality may dif- summarize drought-mediated changes in snag and tree ferentially affect tree species and sizes, such mortality may abundance, and composition of snag and tree popula- drive significant changes in forest structure and compos- tions within these forest types, over the 20-year period ition (Mueller et al. 2005;Ganey andVojta 2011; Kane from 1997 to 2017, with an emphasis on temporal et al. 2014). changes in the snag population, including rates of snag Snags, or standing dead trees, are important compo- recruitment and loss and the factors driving these pro- nents of forest systems that are directly affected by cesses. Our specific objectives included: (1) Evaluating drought-mediated trends in tree mortality. These snags temporal trends in snag and tree density and snag re- serve as biological legacies in forest systems (Thomas cruitment and loss rates, (2) Summarizing climate et al. 1979; McComb and Lindenmayer 1999; Wolden- during the study period and evaluating potential rela- dorp and Keenan 2005), providing habitat for native tionships between climate patterns and snag recruit- wildlife (e.g., Bull et al. 1997; Rabe et al. 1998), serving ment, and (3) Evaluating temporal trends in composition as an important source of coarse woody debris (Harmon and structure of overall snag populations, as well as in et al. 1986; Laudenslayer et al. 2002; Woldendorp and snags that were recruited or lost during the intervals be- Keenan 2005), and aiding in nutrient cycling and other tween sampling occasions. These data thus provide in- ecosystem functions. Snags also provide an index of re- formation on trends in snag populations during the cent tree mortality (Ganey and Vojta 2011; Wu et al. study period as well as on the factors driving those 2017), with recruitment of new snags reflecting temporal trends. This information should aid forest managers in changes in tree mortality and relative mortality among understanding the potential effects of future climate pat- species and size classes, potentially providing insight into terns on these snag populations, and, to a lesser extent, changes in forest structure and composition. on the tree populations from which they derive. In the southwestern United States, a pronounced in- crease in snag recruitment (i.e., tree mortality) followed Methods an extreme climatic year (2002) embedded within a Study area longer-term mega-drought (Breshears et al. 2005; We established plots for sampling snags and trees within Kopeke et al. 2010; Williams et al. 2020). Several studies a study area of approximately 73,000 ha within the reported on short-term trends in tree mortality (which Coconino and Kaibab National Forests, north-central results in snag recruitment) in important forest types Arizona. Study plots were randomly located in mixed- following this extreme year (Negron et al. 2009 [through conifer (n = 53 plots) and ponderosa pine forests (n =60 2004]; Ganey and Vojta 2011 [through 2007]; Kane et al. plots) within this area (see Ganey 1999) for details on 2014 [through 2008]), but data documenting longer- plot selection). White fir (Abies concolor Lindl. ex Hil- term trends in these forests are lacking, and few studies debr.), Douglas-fir (Pseudotsuga menziesii [Mirb.] focused specifically on snag populations (but see Ganey Franco), and ponderosa pine dominated overstories in and Vojta 2005, 2012, 2014). mixed-conifer forests, accounting for approximately Snag populations are governed by the balance between 90 % of total trees in this forest type (as sampled in gains and losses in snags. Gains occur when new snags 2004; Ganey and Vojta 2011). Other relatively common are created by natural tree senescence processes or dis- species included Gambel oak (Quercus gambelii Nutt.), turbances such as insect outbreaks, diseases, fire, or quaking aspen (Populus tremuloides Michaux), and lim- droughts. Losses occur when snags are lost to timber or ber pine (P. flexilis James), in that order of frequency. fuelwood harvest, fire, or natural decomposition, pro- Ponderosa pine comprised > 90 % of trees in ponderosa cesses which are influenced by factors such as snag spe- pine forest in 2004 (Ganey and Vojta 2011), with Gam- cies and size (Ganey et al. 2015). It is therefore desirable bel oak accounting for approximately 8 % of total trees not only to document populations of existing snags but by frequency. Alligator juniper (Juniperus deppeana also to understand how those populations change over Steud), Douglas-fir, quaking aspen, limber pine, pinyon time and the factors responsible for those changes. pine (P. edulis), and other species of juniper were Since 1997, we have sampled snag and tree popula- present in some stands, typically in relatively small tions periodically in mixed-conifer and ponderosa pine numbers. forests in northern Arizona. Previous papers from this Study plots included a wide range of topographic con- study documented changes in snag populations over 5-, ditions and soil types, ranged in elevation from the tran- 10-, and 15-year increments within this 20-year period sition zone between pinyon − juniper woodland and (Ganey and Vojta 2005, 2012, 2014) as well as patterns ponderosa pine at lower elevations to the ecotone be- in: tree mortality from 2002 to 2007 (Ganey and Vojta tween mixed-conifer and Engelmann spruce (Picea 2011) and snag longevity from 1997 to 2015 (Ganey engelmanni Parry ex Engelm.) − corkbark fir (Abies Ganey et al. Forest Ecosystems (2021) 8:20 Page 3 of 13 lasiocarpa var. arizonica [Merriam] Lemmon) forests at First, our objective was to summarize trends in snag and higher elevations (Brown et al. 1980), and included both tree populations on the overall landscape, including areas commercial forest lands and administratively reserved subject to these disturbances. Second, although the effect lands such as wilderness and other roadless areas. Plots of fire could be large on individual plots, the overall effect also included areas subject to forest thinning and both was relatively small over most 5-year intervals between prescribed burns and wildfires. Consequently, these plots sampling occasions, because relatively few plots were im- sampled a wide range of conditions with respect to for- pacted by moderate-to high severity fire during those in- est structure and composition, and included areas sub- tervals (Table 1). Consequently, estimates for all plots jected to recent disturbance events. were similar to estimates including only plots that did not experience moderate-to high-severity fire. We also in- Sampling snag and tree populations cluded plots subject to forest thinning, but these were We sampled snags in 1-ha plots at five-year intervals be- even fewer in number than burned plots, and thinning ginning in 1997 (i.e., snags were sampled in 1997, 2002, activities therefore had minimal immediate impacts on 2007, 2012, and 2017). Within plot boundaries, we sam- overall snag or tree populations (although they may im- pled all snags ≥ 2 m in height and ≥ 20 cm in diameter at pact future patterns of tree mortality and/or wildfire). breast height (dbh). The focus on snags ≥ 20 cm in dbh reflected the original study objectives related to wildlife We treated years in which we sampled snag (or tree) habitat (Ganey 1999) and the assumption that smaller populations as sampling occasions and the intervals be- snags were less important to wildlife such as cavity- tween those sampling occasions as “sampling periods”. nesting birds (Scott 1978; Cunningham et al. 1980; Con- Thus, for snags we had five sampling occasions (here- way and Martin 1993) and/or roosting bats (Rabe et al. after “years”) and four 5-year sampling periods, whereas 1998; Bernardos et al. 2004; Solvesky and Chambers for trees we had two years and one 10-year sampling 2009). Given this minimum diameter, our inference is period. Years provided estimates of density and compos- limited to populations of snags ≥ 20 cm dbh. ition for snag (or tree) populations at a point in time, We marked all snags with numbered metal tags, allow- and sampling periods provided estimates of change in ing us to distinguish pre-existing snags from new snags those parameters between those points in time. when re-sampling plots (with some exceptions, see As noted earlier, our plots covered a very wide range below). We recorded species and dbh (nearest cm) for of forest structural conditions. Consequently, snag dens- all snags. We sampled plots from May through August, ities varied widely among plots and were so markedly and did not necessarily sample individual plots on the skewed (Ganey and Vojta 2012) that the utility of stand- same date or even in the same month across years. Thus, ard estimators of central tendency such as the mean or the elapsed time between consecutive samples for an in- median was limited. Therefore, we used Huber’sM- dividual plot could range from slightly < 5 years to estimator, a generalized maximum–likelihood estimator slightly > 5 years. We ignored this variability in analysis, that provides robust estimates in distributions contain- and assumed that all intervals between sampling occa- ing outliers (Huber and Ronchetti 2009), to estimate sions represented a 5-year period. central tendency in snag density (by year) and change in We sampled live trees ≥ 20 cm in dbh, which were far snag density (by sampling period). We estimated this more abundant than snags, in a 0.09-ha subplot located parameter and associated 95 % bias-corrected confidence within each snag plot in 2004 and 2014. We adhered to intervals using 1,000 bootstrap iterations (Efron and the minimum 20-cm diameter for consistency with snag Table 1 Number (and percent) of plots in northern Arizona sampling. We recorded tree species and dbh (nearest mixed-conifer and ponderosa pine forest that experienced cm) for all trees. We did not mark individual trees, and moderate-to high-severity fire by 5-year interval between snag consequently were not able to determine fates of individ- sampling occasions. n = 53 and 60 plots sampled in mixed- ual trees. Instead, we focused on overall changes in tree conifer and ponderosa pine forest, respectively. Plots were populations, which were driven by the interactions classified as having experienced moderate- to high-severity fire among ingrowth of small (< 20 cm dbh in 2004) trees if a visual inspection indicated that considerable tree mortality had occurred within the plot due to fire between sampling into our sampled population, growth of trees within that occasions sampled population, and tree mortality. As with snags, our inference is limited to trees ≥ 20 cm dbh. Time interval Mixed-conifer forest Ponderosa pine forest 1997–2002 0 (0.0) 2 (3.3) Analysis of snag and tree populations 2002–2007 3 (5.7) 1 (1.7) We included all sampled plots in our analyses of snag and 2007–2012 2 (3.8) 0 (0.0) tree populations, including plots subject to recent distur- 2012–2017 8 (15.1) 4 (6.7) bances such as wild or prescribed fire, for two reasons. Ganey et al. Forest Ecosystems (2021) 8:20 Page 4 of 13 Tibshirani 1993) in IBM SPSS Statistics v 23 (IBM SPSS We pooled snags or trees across plots within forest type Statistics, IBM Corp., Armonk, NY, 2015). All subse- for these comparisons. quent references to mean values for these and other pa- We included six major species groups in comparisons rameters refer to Huber’sM–estimator. For consistency, of species composition in mixed-conifer forest (white fir, we used the same methods for tree populations, al- Douglas-fir, quaking aspen, ponderosa pine, Gambel though those populations were far less skewed than snag oak, and a sixth group [Other] representing all other populations. species). We included only three species groups (ponder- The skewed distributions for snag densities also lim- osa pine, Gambel oak, and Other) in tests in ponderosa ited the utility of standard hypothesis tests. Conse- pine forest. Other species were present in such small quently, we used the bootstrapped confidence intervals amounts in this forest type that including those species discussed above to assess significance of observed differ- as separate categories resulted in multiple cells with ex- ences in snag density. For comparisons between years or pected values < 5, potentially biasing test results (Con- sampling periods, we assumed that confidence intervals over 1980). In contrast, no cells had expected values < 5 that did not overlap between pairs of years or sampling after collapsing categories. We recognized five diameter periods indicated that those years or sampling periods classes in tests involving diameter-class distributions: differed significantly from each other. For change during 20–29, 30–39, 40–49, 50–59, and ≥ 60 cm dbh. We esti- sampling periods, we assumed that confidence intervals mated percent change in individual snag species or that did not overlap zero indicated that snag (or tree) diameter classes during the 20-year study period as: density changed significantly during that period. Percent change (%) = (2017 value – 1997 value)/(1997 value) × 100. Changes in snag and tree density We estimated snag and tree density within each plot for Climate data each year, then summarized snag and tree density across Because climate can strongly affect tree mortality and plots within forest type for all years. Because snags were thus snag creation, we obtained and summarized data uniquely marked, we also were able to estimate numbers on annual precipitation (AP) and cooling degree days of new snags recruited and existing snags lost during (CDD) for the National Weather Service (NWS) weather each sampling period. Numbers of new snags may be station at Pulliam airport in Flagstaff, AZ (http://w2. slightly overestimated for some plots and sampling pe- weather.gov/climate/xmacis.php?wfo=fgz; downloaded riods, because the metal tags used to mark snags some- 14 Dec 2017). We assumed that this station, which was times melted in plots that experienced moderate- to centrally located within the study area at an elevation of high-severity fire, making it difficult to distinguish new 2136 m, provided a valid index to annual variation in snags recruited post-fire from snags present before the broad-scale climate patterns across the study area. We fire. We suspect that this bias was small, however, for restricted our analysis to the period from 1950 to 2016 two reasons. First, these disturbances affected relatively because data on CDD were not available for most years few plots during most sampling periods (Table 1). Sec- prior to 1950. We assumed that this 67-year period, ond, fires hot enough to melt the metal tags also burned which included the well documented mid-20th century most existing snags, meaning that most apparent “new” drought prominent throughout the southwest (Hereford snags in these areas likely were new. Detection rates for 2007), provided a valid index to broad-scale climate pat- standing snags in unburned plots, estimated using mark- terns across the study area in recent times. resight methodology, were very high, ranging from 0.983 CDD were calculated by NWS using a base temperature to 0.994 across major snag species represented (Ganey of 65°F. Thus, CDD was calculated only for days when the et al. 2015: Table 3). mean temperature was greater than 65°F as: For each sampling period, we estimated numbers of CDD = mean daily temperature (°F) − 65°F. snags gained and lost by plot, then summarized these We used the data from NWS directly rather than con- parameters across plots within forest type. Because num- verting temperatures to °C, because the index is useful bers of snags available at the start of a sampling period primarily in a relative context and the actual units were varied, we also estimated standardized snag loss as: not important. In contrast, we converted AP from inches Percentage of snag loss (%) = (snags lost from time t to cm for consistency with other climate literature. We to t + 5 years)/(snag density at time t) × 100. summarized data for AP and CDD in terms of the differ- ence between annual means for those parameters and Composition of snag and tree populations the 1950–2016 mean for the indicated parameter, which We compared species composition and diameter-class provides an index showing the magnitude and direction distributions of snag and tree populations among years of annual deviation from longer-term normal values. within forest type using chi-square tests (Conover 1980). Ganey et al. Forest Ecosystems (2021) 8:20 Page 5 of 13 These measures provided general indices of relative 2002. Mean loss rate also was considerably greater in wetness (AP) and warmness (CDD; http://www.weather. later sampling periods in ponderosa pine forest, but con- gov/key/climate_heat_cool). To provide a third index fidence intervals around those loss rates were wide and combining relative wetness and warmness, we used data overlapped for all periods. for Palmer’s Drought Severity Index (PDSI; Palmer 1965; Mean tree density declined by 8.3 % and 0.3 % from https://www7.ncdc.noaa.gov/CDO/CDODivisionalSelect. 2004 to 2014 in mixed-conifer and ponderosa pine forest, jsp#; downloaded 18 March 2019) for the region con- respectively, but confidence intervals around mean esti- tainingour studyarea(climatedivision2; https://www. mates were wide and overlapped between years in both esrl.noaa.gov/psd/data/usclimdivs/data/map. forest types (Fig. 1e). Mean tree densities in mixed-conifer html#Arizona). This index uses temperature data and a forest in 2004 and 2014 were approximately ten and five physical water balance model to estimate potential times greater than snag densities in 2002 and 2012, re- evapotranspiration and relative drought severity (Palmer spectively (the closest years in which snags were sampled; 1965, see also https://climatedataguide.ucar.edu/climate- compare Fig. 1a and e). In ponderosa pine forest, mean data/palmer-drought-severity-index-pdsi). Index values tree density was approximately 20 times greater than snag range from − 10 to 10, with lower and higher values in- density in these same among-year comparisons. Conse- dicating drier and wetter conditions, respectively quently, although the observed level of tree mortality was (Palmer 1965). sufficient to significantly alter snag density across time, it did not significantly alter tree density. Note, however, that Results our initial sample of tree populations in 2004 likely oc- Changes in snag and tree density curred after the major mortality pulse during the study, In all years, mean snag densities in mixed-conifer forest which appeared to peak from 2002 to 2003 (Negron et al. were approximately five times greater than mean dens- 2009;Kane et al. 2014). ities in ponderosa pine forest (Fig. 1a). Within forest type, mean snag density was similar in 1997 and 2002, Snag recruitment and climate increased significantly between 2002 and 2007, and de- As discussed above, recruitment of new snags was rela- clined slightly after 2007. In mixed-conifer forest, mean tively low and similar for three of the four sampling pe- snag density remained significantly elevated from 2007 riods, but increased significantly during the period from to 2017 relative to earlier years. In contrast, the confi- 2002 to 2007 in both forest types (Fig. 1b). This spike in dence interval for mean density in 2017 in ponderosa snag recruitment followed an extremely warm and dry pine forest overlapped with all earlier years, indicating year (2002). In the summer of 2002, PDSI reached its convergence toward pre-2007 levels of snag density from lowest value for our study period (Fig. 2a), as well as its 2012 to 2017. Relative to snag density in 1997, peak most extreme value in the past 100 years (Koepke et al. mean density in 2007 was 86 % and 79 % greater in 2010). The year 2002 was not the driest year during our mixed-conifer and ponderosa pine forest, respectively, study (Fig. 2b), but was by far the hottest year during and mean snag density in 2017 was still 83 % and 51 % the study (Fig. 2c), suggesting that the extremely low greater than 1997 density. PDSI in 2002 was driven more by extremely high Mean recruitment of new snags spiked markedly be- temperature than by low moisture. PDSI also exceeded tween 2002 and 2007 in both forest types but was simi- the threshold for extreme drought in the summers of lar among the other sampling periods (Fig. 1b). Mean 2007 and 2015 (Fig. 2a), but did not reach the extreme snag recruitment from 2002 to 2007 was more than level reached in 2002, did not remain low as long as dur- double recruitment in any other sampling period in ing 2002–2003, and 2007 and 2015 overall were not ex- mixed-conifer forest, and differed significantly from tremely dry or warm relative to 2002 (Fig. 2). Snag mean recruitment during all other sampling periods in recruitment did not spike following these later years. this forest type. Mean snag recruitment from 2002 to 2007 was almost double any other sampling period in Composition and structure of snag and tree populations ponderosa pine forest, but due to wide confidence inter- Species composition of snag populations differed signifi- vals differed significantly only from mean recruitment cantly among years in both forest types (Fig. 3a; mixed- from 1997 to 2002. conifer: χ = 905.2, df = 20, P < 0.001, n = 14,857; ponder- Snag loss rates increased with every consecutive sam- osa pine: χ = 24.5, df =8, P < 0.001, n = 3,287). In mixed- pling period in both forest types, and this pattern was conifer forest, the primary temporal trend was a large in- observed both for absolute numbers of snags lost and crease in white fir snags, with smaller increases in all for standardized loss rates (Fig. 1c and d). In mixed- other species. In ponderosa pine forest, proportional conifer forest, mean loss rate was significantly greater representation decreased for ponderosa pine and in- from 2007 to 2012 and 2012 to 2017 than from 1997 to creased for Gambel oak snags. Ganey et al. Forest Ecosystems (2021) 8:20 Page 6 of 13 Fig. 1 Trends in snag and tree populations in northern Arizona mixed-conifer and ponderosa pine forest, 1997–2017. a Snag density by year; snags were sampled every five years. b Density of newly-recruited snags by 5-year interval between sampling occasions. Each year shown indicates the end of the 5-year interval represented. c Density of snags lost during each 5-year interval between sampling occasions. d Percentage of existing snags lost during each 5-year interval between sampling occasions. e Tree density by year; tree populations were sampled only in 2004 and 2014. Shown in all panels are Huber’sM– estimator (solid line), a generalized maximum–likelihood estimator that yields robust estimates of central tendency in distributions containing outliers (Huber and Ronchetti 2009), and lower and upper bounds (dashed lines) of bias-corrected confidence intervals around M Much of the change in species composition was driven 2007. Species composition of snags lost generally was by patterns in recruitment of new snags. In terms of dominated by the same species as newly-recruited snags, magnitude, recruitment of new snags was dominated by but in lower numbers. Increases in snag loss lagged in- white fir in mixed-conifer forest and by ponderosa pine creases in snag recruitment by 5–10 years (Fig. 3c). in ponderosa pine forest (Fig. 3b). Timing of recruitment Diameter-class distributions of snags also differed sig- of new snags varied among species in mixed-conifer for- nificantly among years in both forest types (Fig. 4a, est, peaking from 2002 to 2007 for white fir and quaking mixed-conifer: χ = 67.2, df = 16, P < 0.001, n = 14,857; aspen, from 2007 to 2012 for ponderosa pine, and in- ponderosa pine: χ = 37.3, df =16, P = 0.002, n = 3,476). creasing steadily throughout the study for Douglas-fir. In In both forest types, numbers of snags increased across ponderosa pine forest, recruitment of both ponderosa all size classes, with by far the greatest increase occur- pine and Gambel oak snags peaked between 2002 and ring in the smallest size classes. These changes largely Ganey et al. Forest Ecosystems (2021) 8:20 Page 7 of 13 Fig. 2 Trends in climate in northern Arizona, 1997–2017. a Palmer’s Drought Severity Index (PDSI; Palmer 1965;Alley 1984; Heim 2002)for the region containing our study area by year. This index ranges from − 10 to 10, with higher and lower values indicating wetter and drier conditions, respectively. Horizontal reference lines indicate thresholds for severe (–3.0; dashed line) and extreme (–4.0; solid line) drought, respectively (Palmer 1965). b Difference between annual precipitation (AP) during the study and the 1950 through 2016 mean at the National Weather Service (NWS) station located at Pulliam airport, Flagstaff, AZ, elevation 2136 m. Values above and below the horizontal reference line indicated years that were wetter or drier than average, respectively. c Difference between annual Cooling Degree Days (CDD) during the study and the 1950 through 2016 mean at the National Weather Service (NWS) station located at Pulliam airport. Values above and below the horizontal reference line indicated years that were warmer or cooler than average, respectively. All data sources are described in text were driven by recruitment patterns for new snags, with large numbers of snags recruited in the smaller size clas- ses (Fig. 4b). As with species composition, snags lost were dominated by the same size classes as snags re- cruited, but in lower numbers and with peaks in snag loss lagging peaks in snag recruitment by 5–10 years (Fig. 4c). Species composition of tree populations differed sig- nificantly between 2004 and 2014 only in ponderosa pine forest (χ = 7.105, df =2. P = 0.028; mixed–conifer forest χ = 6.640, df =5, P = 0.359). Ponderosa pine in- creased from 87.4 to 90.4 % of total trees in ponderosa pine forest over this period, whereas proportions of all other species decreased. Size-class distributions for trees did not differ significantly between 2004 and 2014 in ei- ther forest type (mixed-conifer forest χ = 8.773, df =4, P = 0.066; ponderosa pine χ = 5.241, df =4. P = 0.262), and were numerically dominated by trees in the smaller size classes in both years (results not shown here). Discussion Snag numbers increased considerably over our 20-year study period in mixed-conifer forest, with a smaller in- crease in ponderosa pine forest (Fig. 1a). Increased snag density was driven largely by a pulse of snag recruitment (tree mortality) that occurred between 2002 and 2007, with snag recruitment returning to pre-pulse levels in subsequent sampling periods (Fig. 1b). Thus, increases in snag recruitment appeared more indicative of a short- term tree mortality “event” than of a longer-term pattern of elevated tree mortality (with the possible exception of Douglas-fir, Fig. 3a). This mortality pulse appeared to be drought-mediated. Our evaluation of the influence of climate on tree mor- tality was correlative and cannot demonstrate cause and effect relationships. Nevertheless, two aspects of this mortality pulse appear most consistent with drought ef- fects. The first relates to the timing of the mortality Ganey et al. Forest Ecosystems (2021) 8:20 Page 8 of 13 Fig. 3 Trends in species composition of snag populations in northern Arizona mixed-conifer (n = 53 plots) and ponderosa pine (n = 60 plots) forest. a Number of snags by species and year. b Number of newly-recruited snags by species and 5-year period ending in the indicated year. c Number of snags lost within the 5- year period ending in the indicated year, by species. Species acronyms: ABCO = white fir, PIPO = ponderosa pine, POTR = quaking aspen, PSME = Douglas-fir, and QUGA = Gambel oak. Not shown is a composite group comprised of all other species. Snags were pooled across plots within forest type pulse, which immediately followed a single extremely warm and dry year in 2002 (Fig. 2; see also Negron et al. 2009; Kane et al. 2014) embedded within a longer-term megadrought (Williams et al. 2020). The second factor suggesting that mortality was drought-mediated was the spatial synchrony of mortality across plots within both forest types plots. This spatial synchrony in mortality ap- peared more consistent with widespread drought effects than with effects of more spatially limited disturbance events such as fire or insect outbreaks. Those other dis- turbance agents certainly contributed to mortality in some areas, but their effects were not as widespread, and likely were exacerbated by drought. The observed mortality pulse appeared to be triggered by the extreme climate year in 2002. PDSI reached its lowest level in over 100 years in 2002 (Kopeke et al. 2010), and a tree-ring reconstruction indicated that moisture stress during 2002 reached one of the highest levels in the past 1400 years (Salzer and Kipfmueller 2005; see also Williams et al. 2020). The extremely low PDSI in 2002 appeared to be driven more by extremely hot conditions than by extremely dry conditions; 2002 was by far the hottest year during the study but was not the driest year (Fig. 2). Thus, conditions in 2002 ap- peared representative of what Breshears et al. (2005) de- fined as “global climate change type drought”, characterized by dry conditions coupled with extreme heat. Climate change models suggest that such years will occur more frequently in the southwestern U. S. in the future (Seager et al. 2007; Garfin et al. 2014). Many later years during the study were either warmer or drier than average, or both, and annual PDSI values suggested some level of drought stress in 16 of 20 years during the study (Fig. 2a). Those later years and periods did not trigger the same levels of tree mortality seen fol- lowing 2002, however. There are at least four possible explanations for this apparent discrepancy. First, the ex- treme conditions experienced during 2002, especially with regard to temperature (Adams et al. 2009) may have pushed many trees beyond a physiological thresh- old for survival (McDowell et al. 2008, 2016; Koepke et al. 2010) that was not reached again in subsequent years. Second, the lack of increased tree mortality fol- lowing later years that were almost as extreme Ganey et al. Forest Ecosystems (2021) 8:20 Page 9 of 13 Fig. 4 Trends in diameter-class distribution of snag populations in northern Arizona mixed-conifer (n = 53 plots) and ponderosa pine (n = 60 plots) forest. a Number of snags by diameter class and year; b Number of newly-recruited snags by diameter class and 5-year period ending in the indicated year; c Number of snags lost within the 5-year period ending in the indicated year, by diameter class. Snags were pooled across plots within forest type climatically as 2002 could indicate that the most vulner- able trees died in the post-2002 mortality event, leaving tree populations dominated by trees that were more re- silient to drought. Third, remnant trees may not have been more drought-resistant, but drought stress may have been alleviated in subsequent years by previous mortality that reduced tree density and competition for available precipitation (Peet and Christiansen 1987). Fourth, bark beetle populations may have been higher following the extreme climate year in 2002 than in sub- sequent years, swamping the defenses of many trees (e.g., Negron et al. 2009; Kane et al. 2014). At present we cannot distinguish between these hy- potheses, which are not-mutually exclusive. They clearly have very different implications for future trends in for- est structure and composition, however. For example, if hypothesis 1 is correct, we can expect to see similar levels of tree mortality whenever climatic extremes like 2002 occur. In contrast, both hypotheses 2 and 3 suggest that future mortality may be lower even in similarly ex- treme climate years. Hypothesis 4 is particularly intri- guing in the context of understanding future trends, as it may suggest feedback loops that regulate drought- mediated mortality. Bark beetle populations are known to increase in response to drought (e.g., Fettig et al. 2007; Huang et al. 2020), exacerbating mortality in drought-stressed trees (Huang et al. 2020). This increase in bark beetles, along with the associated increase in tree mortality, provides increased food and nest resources for bark insectivores, many of which nest in cavities in snags, resulting in short-term (2–5 years) increases in populations of woodpeckers and other bark insectivores (Koplin 1969; Edworthy et al. 2011; Saab et al. 2019). These populations of bark insectivores then aid in regu- lating bark beetle numbers (Koplin 1972; Fayt et al. 2005). This raises the possibility that severe drought years following 2002 may have seen lower mortality partly because of these feedback loops operating follow- ing earlier spikes in snag recruitment. Regardless of the explanation for the observed mor- tality event, that event was large enough to signifi- cantly alter species composition and structure of snag populations during the study. All species of snags showed increased recruitment during the study, but Ganey et al. Forest Ecosystems (2021) 8:20 Page 10 of 13 the magnitude, timing, and duration of that increase primarily by snag recruitment, they were mediated by varied among species. For example, white fir and patterns in snag longevity. Longevity was related pri- quaking aspen, the species least resistant to drought marily to snag species, diameter, and height (Ganey stress (Ganey and Vojta 2011; after Niinemets and et al. 2015). Standing rates for white fir and ponder- Vallardes 2006), responded strongly and quickly in osa pine snags were lower than for other major spe- mixed-conifer forest, with relatively large increases in cies (Ganey et al. 2015:Table 2),which tended to snag recruitment peaking in the sampling period from work opposite the large increases in recruitment for 2002to2007and returningtopre-2002levelsinsub- those two species (Fig. 3b). And standing rates for sequent periods (Fig. 3b). In contrast, the more smallerdiameter snags were lowerthanthose forlar- drought-resistant Douglas-fir did not respond as ger diameter snags across species (Ganey at al. 2015: strongly from 2002 to 2007, but instead snag recruit- Table 5). Again, this suggests that patterns in snag ment increased over subsequent sampling periods. loss worked opposite to patterns in snag recruitment, Ponderosa pine also responded fairly strongly in which was concentrated in the smaller size classes mixed-conifer forest from 2002 to 2007, but snag re- (Fig. 4b). Snag loss rates also may have been influ- cruitment for this species did not peak until the enced by bark beetle activity, especially later in the period from 2007 to 2012 and remained elevated study. Snags resulting from bark beetle infestations from 2012 to 2017. appear to fall relatively quickly (Chambers and Mast These differences in magnitude and timing of 2014; Rhoades et al. 2020), and this also may have species-specific snag recruitment resulted in signifi- contributed to the increase in snag loss rates observed cant changes in species composition of snag popula- over time in this study (see Fig. 1cand d). tions. In mixed-conifer forest, snag populations in The effect of these changes in composition and struc- 1997 and 2002 were co-dominated by white fir, pon- ture of snag populations on snag use by native wildlife is derosa pine, and Gambel oak in approximately equal unknown. The increase in snag abundance should provide numbers (Fig. 3a). By 2007, those populations were increased foraging opportunities for native birds (Bull dominated by white fir snags, and white fir continued et al. 1997), at least in the short term. Effects on nesting to dominate these populations through 2017, although birds and/or roosting bats are more unclear. Snag popula- it declined in abundance in later years. In contrast, tions in both forest types became increasingly dominated relative abundance of Douglas-fir increased across all by smaller snags during the study, and populations in years, peaking in 2017. This pattern may indicate mixed-conifer forest became increasingly dominated by that, unlike some other species, Douglas-fir was white fir snags. These snags are less likely to be used by responding more to cumulative drought stress than to cavity nesting birds or roosting bats than are larger snags a one-timemortalityevent. or snags of other species (Rabe et al. 1998; Ganey and In the less species diverse ponderosa pine forest type, Vojta 2004; Solvesky and Chambers 2009). They also gen- snag populations were dominated by ponderosa pine in erally decay and fall more rapidly than do larger snags or all years, with Gambel oak the only other species present other species of snags (Ganey et al. 2015). The increased in considerable numbers. Recruitment of ponderosa pine abundance of such snags may have contributed to the ob- snags peaked earlier in the warmer and drier ponderosa served increase in snag loss rates across time in both pine forest (2002–2007; Fig. 3b) than in mixed-conifer forest types during the study (Fig. 1c), despite the large forest. That increase largely was offset by increased loss number of “new” snags in these populations. Thus, al- rates of ponderosa pine snags following 2007, however though snag numbers increased during the study, and that (Fig. 3c). increase likely provided a temporary increase in foraging Diameter distributions of snag populations also chan- substrates and food resources for birds, many of the re- ged significantly during the study in both forest types cruited snags may represent an ephemeral resource that is (Fig. 4a). Numbers of snags recruited increased for all not useful to cavity-nesting birds (Ganey and Vojta 2004). size classes from 2002 to 2007, but the magnitude of in- Note that these rapidly-decaying and falling snags contrib- crease was far greater for the smallest size classes than ute to increases in log populations, however, and this may for the largest size classes. This imbalance was partially benefit native wildlife that feed on, hide in, or den in or but not entirely offset by higher loss rates for snags in under down logs (Bull et al. 1997; Ganey and Vojta 2017). the smallest size classes (Fig. 4c). The net result was in- In contrast to snag populations, tree mortality did creased relative abundance of snags in the smallest size not appear to be adequate to significantly alter tree classes and decreased relative abundance of snags in the populations during the study. Our ability to evaluate largest size classes. changes in tree populations during the study period Note that, although these changes in species com- was hampered by the more limited sampling of those position and diameter distributions were driven populations, however, and especially by the timing of Ganey et al. Forest Ecosystems (2021) 8:20 Page 11 of 13 sampling. We assessed change in snag populations observed following the extreme climatic year in 2002, over a 20-year period, but assessed change in tree coupled with climate models predicting that such years populations only over a 10-year period embedded will become more common in the future (Seager et al. within that longer period. More importantly, we ini- 2007; Garfin et al. 2014; Williams et al. 2020), supports tially sampled tree populations in 2004, after the peak calls for management to increase resilience and drought- of the tree mortality pulse triggered by extreme resistance in these forest types (Millar et al. 2007; Ste- drought conditions in 2002 (2002–2003; Negron et al. phens et al. 2013). 2009; Ganey and Vojta 2011;Kane etal. 2014). Thus, Acknowledgements tree populations likely showed greater change during We thank J. Jenness, G. Martinez, M. Stoddard, B. Strohmeyer, and R. White the 20-year study period than we were able to dem- for assisting in establishing plots, and L. Doll, D. and N. Ganey, and C. Vojta for help with plot sampling. J. Ellenwood, B. Higgins, K. Menasco, C. Nelson, onstrate. Even given these limitations, however, the and G. Sheppard (Kaibab National Forest) and C. Beyerhelm, A. Brown, H. much greater tree densities relative to snag densities Green, T. Randall-Parker, C. Taylor, and M. Whitney (Coconino National Forest) clearly ensure that, for a given mortality level, relative assisted with initial plot selection. effects will be much lower for tree populations than Authors' contributions for snag populations. JG conceived the study, with input from JI, AI, and SV. JG, JI, AI, and SV established the plots and collected the data. JG analyzed the data. All authors contributed to writing the manuscript and read and approved the Conclusions final manuscript. Warming climates have profoundly affected forests throughout the world (Adams et al. 2009; Allen et al. Funding Funding was provided by the USDA Forest Service Rocky Mountain Research 2010), and climate models predict further warming and Station. The funder played no official role (beyond employing the authors) in drying in many areas, including the southwestern United the design of the study and collection, analysis, and interpretation of data States (e.g., Seager et al. 2007; Stocker et al. 2013; Garfin and in writing the manuscript. et al. 2014). Our results demonstrate the dynamic nature Availability of data and materials of snag populations and document the drivers associated The datasets used and/or analysed during the current study are available with rapid changes in those populations over time. That from the corresponding author on reasonable request. is, a drought event in 2002 caused spatially widespread Declarations tree mortality. This mortality significantly increased re- cruitment of new snags, but many of the newly-recruited Ethics approval snags were small in diameter and/or white fir snags. Not applicable. These snags decay and fall faster than larger snags and Consent for publication other species of snags (Ganey et al. 2015), and this con- Not applicable. tributed to an increasing snag loss rate over time. This Competing interests rapid loss of snags did not entirely balance the increased The authors declare that they have no competing interests. recruitment of new snags following the 2002 drought, resulting in higher snag densities after 2007 compared to Received: 29 June 2020 Accepted: 23 February 2021 prior periods, particularly in mixed conifer forests. Sig- nificant questions remain, however, including: (1) how References long do these changes in abundance, structure, and Adams HD, Guardiola-Claramonte M, Barron-Gafford GA, Villegas JC, Breshears composition of snag populations persist? (2) how com- DD, Zou CB, Troch PA, Huxman TE (2009) Temperature sensitivity of drought- induced tree mortality portends increased regional die-off under global- mon will similarly extreme climate years be in the fu- change-type drought. PNAS 106:7063–7066 ture? and (3) will such years produce mortality pulses Allen CD, Macalady AK, Chenchouni H, Bachelet D, McDowell N, Vennetier MM, similar to the one observed in this study, or might future Kitzberger T, Rigling A, Breshears DD, Hogg EH, Gonzalez P, Fensham R, Zhang Z, Castro J, Demidova N, Lim J-H, Allard G, Running SW, Semerci A, patterns of tree mortality be modified by past mortality Cobb N (2010) A global overview of drought and heat-induced tree events that alter forest structure and composition, and mortality reveals emerging climate change risks for forests. Forest Ecol competition for resources? Answering these questions Manag 259:660–684 Alley WA (1984) The Palmer drought severity index: limitation and assumptions. J will require long-term studies, but is critical to under- Clim Appl Meteorol 23:1100–1109 standing how changing climate may affect trends in Bentz BJ, Régnière J, Fettig CJ, Hansen EM, Hayes JL, Hicke JA, Kelsey RG, Negrón structure and composition of future snag populations JF, Seybold SJ (2010) Climate change and bark beetles of the Western United States and Canada: Direct and indirect effects. Bioscience 60:602–613 and the tree populations from which they derive. The Bernardos DA, Chambers CL, Rabe MJ (2004) Selection of Gambel oak roosts by U.S. Forest Service Forest Inventory and Analysis pro- southwestern Myotis in ponderosa pine-dominated forests, northern Arizona. gram (Smith 2002), which repeatedly samples plots J Wildlife Manag 68:595–601 Breshears DD, Cobb NS, Rich PM, Price KP, Allen CD, Balice RG, Romme WH, through time across the landscape, may provide a means Kastens JH, Floyd ML, Belknap J, Anderson JJ, Myers OB, Meyer CW (2005) to accomplish the longer-term monitoring needed to an- Regional vegetation die-off in response to global-change-type drought. swer these questions. In the meantime, the mortality PNAS 102:15144–15148 Ganey et al. Forest Ecosystems (2021) 8:20 Page 12 of 13 Brown DE, Lowe CH, Pase CP (1980) A digitized systematic classification for Hartmann H (2020) Tree defence and bark beetles in a drying world: carbon ecosystems with an illustrated summary of the natural vegetation of North partitioning, functioning and modelling. New Phytol 225:26–36 America. Gen Tech Rep RM-73. U.S Department of Agriculture, Forest Service, Huber PJ, Ronchetti KW (2009) Robust Statistics, Second Ed. John Wiley & Sons, Rocky Mountain Forest and Range Experiment Station, Fort Collins, 93 Inc., Hoboken Bull EL, Parks CG, Torgersen TR (1997) Trees and logs important to wildlife in the Kane JM, Kolb TE, McMillin JD (2014) Stand-scale tree mortality factors differ by interior Columbia River Basin. USDA Forest Service General Technical Report site and species following drought in southwestern mixed–conifer forests. PNW-GTR-391 Forest Ecol Manag 330:171–182 Chambers CL, Mast JN (2014) Snag dynamics and cavity excavation after bark beetle Koepke DF, Kolb TE, Adams HD (2010) Variation in woody plant mortality and outbreaks in southwestern ponderosa pine forests. Forest Sci 60:713–723 dieback from severe drought among soils, plant groups, and species within a Conover WJ (1980) Practical nonparametric statistics, 2nd ed. John Wiley & Sons, northern Arizona ecotone. Oecologia 163:1079–1090 New York Koplin JR (1969) The numerical response of woodpeckers to insect prey in a Conway CJ, Martin TE (1993) Habitat suitability for Williamson’s sapsuckers in subalpine forest in Colorado. Condor 71:436–438 mixed-conifer forests. J Wildlife Manag 57:322–328 Koplin JR (1972) Measuring predator impact of woodpeckers on spruce beetles. J Cunningham JB, Balda RP, Gaud WS (1980) Selection and use of snags by Wildlife Manag 36:308–320 secondary cavity-nesting birds of the ponderosa pine forest. USDA Forest Laudenslayer WF Jr, Shea PJ, Valentine BE, Weatherspoon CP, Lisle TE (tech. cords. Service Research Paper RM-222 ) (2002) Proceedings of the symposium on the ecology and management of Davis KT, Dobrowski SZ, Higuera PE, Holden ZA, Veblen TT, Rother MT, Parks SA, dead wood in western forests. Gen Tech Rep PSW-GTR-181. Albany, CA: U.S. Salag A, Manet MP (2019) Wildfires and climate change push low-elevation Department of Agriculture, Forest Service, Pacific Southwest Research forests across a critical climate threshold for tree regeneration. PNAS 116: Station. 949 6193–6198 McComb W, Lindenmayer D (1999) Dying, dead, and down trees. In: Hunter ML Edworthy AB, Drevera MC, Martin K (2011) Woodpeckers increase in abundance Jr (ed) Maintaining Biodiversity in Forest Ecosystems. Cambridge University but maintain fecundity in response to an outbreak of mountain pine bark Press, Cambridge beetles. Forest Ecol Manag 261:203–210 McDowell NG, Pockman WT, Allen CD, Breshears DD, Cobb N, Kolb T, Plaut J, Efron B, Tibshirani R (1993) An Introduction to the Bootstrap. Chapman & Hall/ Sperry J, West A, Williams DG, Yepez EA (2008) Tansley Review: mechanisms CRC Monographs on Statistics and Applied Probability 57. Boca Raton, FL, of plant survival and mortality during drought: Why do some plants survive USA. 456 while others succumb to drought? New Phytol 178:719–739 Fayta P, Machmerb MM, Steeger C (2005) Regulation of spruce bark beetles by McDowell NG, Williams AP, Xu C, Pockman WT, Dickman LT, Sevanto S, Pangle R, woodpeckers— a literature review. Forest Ecol Manag 206:1–14 Limousin J, Plaut J, Mackay DS, Ogee J, Domec JC, Allen CD, Fisher RA, Jiang Fettig CJ, Klepzig KD, Billings RF, Munson AS, Nebeker TE, Negron JF, Nowak JT X, Muss JD, Breshears DD, Rauscher SA, Koven C (2016) Multi-scale (2007) The effectiveness of vegetation management practices for prevention predictions of massive conifer mortality due to chronic temperature rise. Nat and control of bark beetle infestations in coniferous forests of the western Clim Change 6. doi:https://doi.org/10.1038/NCLIMATE2873 and southern United States. Forest Ecol Manag 238:24 – 53 McKenzie D, Gedalof Z, Peterson DL, Mote P (2004) Climatic change, wildfire, and Ganey JL (1999) Snag density and composition of snag populations on two conservation. Conserv Biol 18:890–902 National Forests in northern Arizona. Forest Ecol Manag 117:169–178 Millar CL, Stephenson NL, Stephens SL (2007) Climate change and forests of the future: managing in the face of uncertainty. Ecol Appl 17:2145–2151 Ganey JL, Vojta SC (2004) Characteristics of snags containing excavated cavities in northern Arizona mixed-conifer and ponderosa pine forests. Forest Ecol Mueller RC, Scudder CM, Porter ME, Trotter RTIII, Gehring CA, Whitham TG (2005) Manag 199:323–332 Differential tree mortality in response to severe drought: evidence for long- Ganey JL, Vojta SC (2005) Changes in snag populations in northern Arizona term vegetation shifts. J Ecol 93:1085–1093 mixed-conifer and ponderosa pine forests, 1997–2002. Forest Sci 51:396–405 Negron JF, McMillin JD, Anhold JA, Coulson D (2009) Bark beetle-caused Ganey JL, Vojta SC (2011) Tree mortality in drought–stressed mixed–conifer and mortality in a drought–affected ponderosa pine landscape in Arizona, USA. ponderosa pine forests, Arizona. Forest Ecol Manag 261:162–168 Forest Ecol Manag 257:1353–1362 Ganey JL, Vojta SC (2012) Trends in snag populations in drought-stressed mixed- Niinemets U, Vallardes F (2006) Tolerance to shade, drought, and waterlogging of conifer and ponderosa pine forests, 1997–2007. Int J Forest Res. doi:https:// temperate Northern Hemisphere trees and shrubs. Ecol Monogr 76:521–547. doi.org/10.1155/2012/529197 http://www.esapubs.org/Archive/mono/M076/020/appendia-A.htm. Accessed 11 April 2008 Ganey JL, Vojta SC (2014) Trends in snag populations in northern Arizona mixed– Palmer WC (1965) Meteorological drought. U. S. Weather Bureau Research Paper conifer and ponderosa pine forests, 1997–2012. USDA Forest Service 45. 58. https://www.ncdc.noaa.gov/temp-and-precip/drought/docs/palmer. Research Paper RMRS-RP-105 pdf. Accessed 19 Dec 2017 Ganey JL, Vojta SC (2017) Comparative trends in log populations in northern Arizona mixed–conifer and ponderosa pine forests following severe drought. Peet RK, Christensen NL (1987) Competition and tree death. Bioscience 37:586–595 West North Am Nat 77:281–292 Rabe MJ, Morrell TE, Green H, deVos JC Jr, Miller CR (1998) Characteristics of Ganey JL, White GC, Jenness JS, Vojta SC (2015) Mark-recapture estimation of ponderosa pine snag roosts used by reproductive bats in northern Arizona. J snag standing rates in northern Arizona mixed-conifer and ponderosa pine Wildlife Manag 62:612–621 forests. J Wildlife Manag 79:1369–1377 Rhoades CC, Hubbard RM, Hood PR, Starr BJ, Tinker DB, Elder K (2020) Snagfall the first decade after severe bark beetle infestation of high-elevation forests Garfin G, Franco G, Blanco H, Comrie A, Gonzalez P, Piechota T, Smyth R, Waskom in Colorado, USA. Ecol Appl e02059 R (2014) Ch. 20: Southwest. In: Melillo JM, Richmond TC, Yohe GW (eds) Climate Change Impacts in the United States: The Third National Climate Saab VA, Latif QS, Dresser MA, Dudley JG (2019) Woodpecker nest survival, Assessment. U.S. Global Change Research Program, 462 density, and a pine beetle outbreak. J Wildlife Manag 83:1387–1400 Harmon ME, Franklin JF, Swanson FJ, Sollins P, Gregory SV, Lattin JD, Anderson Salzer MW, Kipfmueller KF (2005) Reconstructed temperature and precipitation NH, Cline SP, Aumen NG, Sedell JR, Lienkamper GW, Cromack K Jr, Cummins on a millennial timescale from tree-rings in southern Colorado Plateau, USA. KW (1986) Ecology of coarse woody debris in temperate ecosystems. Adv Clim Change 70:465–487 Ecol Res 15:133–302 Scott VE (1978) Characteristics of ponderosa pine snags used by cavity-nesting Heim RH Jr (2002) A review of twentieth-century drought indices used in the birds. J Forest 76:26–28 United States. Bull Am Meteorol Soc 83:1149–1165 Seager R, Ting MF, Held IM, Kushmir Y, Lu J, Vecchi G, Huang H, Harnick N, Hereford R (2007) Climate variation at Flagstaff, Arizona-1950 to 2007. U.S. Leetmaa A, Lau N, Li C, Velez J, Naik N (2007) Model projections of an Geological Survey Open-File Report 2007 – 1410. P 17. https://www.researchga imminent transition to a more arid climate in southwestern United States. te.net/profile/Richard_Hereford/publication/282031501_Climate_Variation_at_ Science 316:1181–1184 Flagstaff_Arizona--1950_to_2007/links/5601c0e008aed9851827cb99.pdf. Smith WB (2002) Forest inventory and analysis: a national inventory and Accessed 19 Dec 2017 monitoring program. Environ Poll 116:S233–S242 Huang J, Kautz M, Trowbridge AM, Hammerbacher A, Raffa KF, Adams HD, Solvesky BG, Chambers CL (2009) Roosts of Allen’s lappet-browed bat in northern Goodsman DW, Xu C, MeddensA JH, Kandasamy D, Gershenzon J, Seidl R, Arizona. J Wildlife Manag 73:677–682 Ganey et al. Forest Ecosystems (2021) 8:20 Page 13 of 13 Stephens SL, Agee JK, Fulé PZ, North MP, Romme WH, Swetnam TW, Turner MG (2013) Managing forests and fire in changing climates. Science 342:41–42 Stocker TF, Qin D, Plattner GK, Tignor MM, Allen SK, Boschung J, Nauels A, Xia Y, Bex V, Midgley PM (2013) IPCC 2013: Climate change 2013: the physical science basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge Thomas JW, Anderson RG, Maser C, Bull EL (1979) Snags. In: Thomas J W (tech. ed.) Wildlife habitats in managed forests: The blue mountains of Oregon and Washington. USDA Agricultural Handbook 553, 60–77. Washington, D.C Wan H-Y, Cushman SA, Ganey JL (2019) Recent and projected wildfire trends across the ranges of three spotted owl subspecies under climate change. Front Ecol Environ 7:37. doi:https://doi.org/10.3389/fevo.2019.00037 Williams AP, Cook ER, Smerdon JE, Cook BI, Abatzoglou JT, Bolles K, Baek SH, Badger AM, Livneh B (2020) Large contribution from anthropogenic warming to an emerging North American megadrought. Science 368:314–318 Woldendorp G, Keenan RJ (2005) Coarse woody debris in Australian forest ecosystems: a review. Aust Ecol 30:834–843 Wu H, Franklin SB, Liu J, Lu Z (2017) Relative importance of density dependence and topography on tree mortality in a subtropical mountain forest. Forest Ecol Manag 384:169–179

Journal

"Forest Ecosystems"Springer Journals

Published: Mar 8, 2021

There are no references for this article.