1 Introduction

Forest management activities are designed to alter stand and tree growth, composition, structure, and/or other ecosystem properties (Smith et al. 1997). Management activities such as thinning, prescribed burning, herbicide application, or fertilization change resource availability within forests, altering tree growth (Fox et al. 2007; Johnson and Curtis 2001; Jurgensen et al. 1997; Neary et al. 1999). However, because these activities often change the availability of multiple resources—including moisture, nutrients, and light (e.g. Jurgensen et al. 1997; Neary et al. 1999)—the physiological or ecological mechanisms responsible for changes in tree growth can be unclear. The forestry research community has long been interested in understanding how management activities affect both resource availability and tree physiology. Their wish has been for more mechanistic prescriptions to meet objectives including the optimization of forest productivity (e.g., Campoe et al. 2013; Forrester et al. 2013; Gauthier and Jacobs 2009; Gspaltl et al. 2013), adaptation to climate change (e.g. Chmura et al. 2011; Lindner et al. 2014; Martin-Benito et al. 2011), and the production and maintenance of ecosystem services (Deal et al. 2012; Ford et al. 2011; Luyssaert et al. 2018).

Stable isotopes are an important tool to understand the response to trees, stands and ecosystems to management because they integrate and trace physical and physiological processes over time and space (Dawson et al. 2002; West et al. 2006; Sulzman 2008; Werner et al. 2012). Forestry researchers were early adopters of some stable isotope techniques (e.g., Heilman et al. 1982; Högberg et al. 1993; Nambiar and Bowen 1986). The use of these techniques has expanded rapidly in the last decade to encompass a variety of management practices in ecosystems from the African savanna to giant sequoia forests in California (e.g., Cramer et al. 2010; York et al. 2010). However, although stable isotopes can be powerful analytical tools, the application and interpretation of stable isotope methods can be complex because numerous abiotic and biotic processes alter the relative abundance of individual isotopes (Dawson et al. 2002; Sulzman 2008; Werner et al. 2012; West et al. 2006).

In this chapter, we describe isotopic consequences of management effects and discuss their mechanistic interpretation. We first present, in chronological order, several of the traditional management tools used in even-aged stands from planting to harvest, largely because much of the isotopic literature examining forest management occurred in such stands. This organization is intended to clarify, not to suggest that forest managers are limited to these tools and applications. In fact, these management tools can be used to foster a broad range of ecosystem services, which isotopic data can help inform. We provide both illustrative examples and, in the cases of nutrient additions (fertilization) and thinning, meta-analyses. We present the results of a new meta-analysis here, but we also summarize an earlier meta-analysis (Sohn et al. 2016) of thinning effects on drought responses of tree growth.

We focus here on carbon isotope discrimination (∆13C, Chap. 9) and oxygen isotopic ratios (δ18O, Chap. 10), with brief mention of nitrogen isotope ratios (δ15 N, Chap. 12). We neglect hydrogen isotope ratios (δ2H) because they are often so correlated with δ18O and because hydrogen exchanges so readily with atmospheric water vapor that the samples must undergo complex preparation (Chap. 11). The method is therefore not prevalent in forest management applications. Intense work is going on with δ2H preparation techniques and it seems likely that they will become more feasible in the near future.

2 Choice of Species and Genotype

One of the most powerful tools of forest management is the ability to select the species and genotypes comprising the stand. Foresters can plant favored trees, cull unwanted species or phenotypes, or create conditions that are favorable for particular trees. Tree species differ substantially in ∆13C, which is correlated with intrinsic water-use efficiency (iWUE, Chap. 17) defined as the ratio of net photosynthetic rate to stomatal conductance (A/gs). Numerous studies have evaluated interspecific differences in ∆13C in order to evaluate traits such as drought tolerance and water-use efficiency (e.g., Bonhomme et al. 2008; Jansen et al. 2013; Voltas et al. 2006). Broadly, in interspecific studies, deciduous trees show more positive ∆13C and lower iWUE (intrinsic water-use efficiency) than co-occurring needle-leaved evergreens, which themselves have more positive ∆13C and lower iWUE than scale-leaved evergreens (Marshall and Zhang 1994). Long-lived leaves tend to have greater leaf mass per area (g/m2) (Wright et al. 2004), which is associated with lower mesophyll conductance (Flexas et al. 2014; Niinemets et al. 2009). This relationship is thought (Hultine and Marshall 2000) to be the primary reason why a negative correlation between leaf ∆13C and leaf mass per area has been frequently observed (Araus et al. 1997; Hanba et al. 1999; Hultine and Marshall 2000; Körner et al. 1991; Takahashi and Miyajima 2008; Talhelm et al. 2011). In any case, if a forest is likely to deplete the soil water supply during the growing season, then efficient use of that water generally enhances forest growth (Zhang et al. 1995; Zhang and Marshall 1995) and possibly survival (Timofeeva et al. 2017).

Species differences in water access have also been detected in δ18O of xylem water or wood. These differences in water source may be caused by differences in vertical root distribution, where water deeper in the soil tends to be more depleted in heavy isotopes (18O, 2H) than water found near the soil surface (Hartl-Meier et al. 2015; see Chap. 18). This isotopic difference within the soil profile has been used to understand how species partition water resources (Flanagan et al. 1992) and how water-use patterns change seasonally (Dawson 1998; Meinzer et al. 1999). When species are mixed, the root systems may be deployed at different depths, leading to differences in vertical water-uptake. The resulting niche partitioning may give rise to overcompensation, the tendency of a species mixture to be more productive than monospecific stands of either species alone (Pretzsch et al. 2013). However, species mixtures may also lead to growth reductions and mortality for some of the species (Granda et al. 2014). For example, shallow-rooted spruce trees were more negatively influenced by drought than more deeply rooted beech trees in a mixed stand (Brinkmann et al. 2018). Alternatively, differences in tissue or leaf water δ18O may also reflect differences in transpiration timing or rate (Chap. 10). In either case, δ18O differences provide an index of water sources and water use (Marshall and Monserud 2006). When combined with δ13C, δ18O in cellulose may allow one to distinguish whether water-use efficiency has been modified by changes in photosynthesis or transpiration (Scheidegger et al. 2000; see Chap. 16). However, using δ18O to draw insight into leaf gas exchange requires knowledge or assumptions about the δ18O of xylem water (Gessler et al. 2018; Roden and Siegwolf 2012; Chap. 18).

Seedling planting also offers the opportunity to control genetic composition (e.g., species, genotype, and clone) and initial stand structure. In intensive forestry, “tree improvement” is almost always focused on increased timber yield through rapid growth, wood quality, or increased pest resistance. Careless matching of genotype to environment can be deleterious, causing frost damage when coastal Douglas-fir (Pseudotsuga menziesii) was moved inland (Rehfeldt 1977), or suboptimal growth when cold-tolerant lodgepole pine (Pinus contorta; McLane et al. 2011) or Douglas-fir (Leites et al. 2012) were moved to warmer climates. Planting can also be used to assist the migration of warm climate genotypes toward areas that were once too cold for them, but are no longer (Savolainen et al. 2004). Isotopic data have been used as a screening criterion to describe the populations within a species, often from young trees planted in common gardens (Gornall and Guy 2007; Guerra et al. 2016; Monclus et al. 2006). This approach has also identified provenance differences in rooting depth using δ8O (Voltas et al. 2015). Although these studies have mostly focused on ecological questions related to species adaptations to a particular climate, the observed correlation of ∆13C with height growth or biomass production has also been used to justify genotype selections (Marguerit et al. 2014; Zhang et al. 1995; Zhang and Marshall 1994, 1995). In addition, quantitative trait loci have been applied in native populations to determine whether WUE can be increased without decreasing productivity (Brendel et al. 2002, 2008; Marguerit et al. 2014). In a recent paper, Isaac-Renton et al. (2014) provided a compelling assessment of the climate-change consequences of climate maladaptation in lodgepole pine (Fig. 23.1). The northern populations were defined as “leading edge” because the species is presumably moving northward. These northern populations lacked some of the drought adaptations found in the southern “trailing edge” populations, which may reduce their ability to deal with future droughts. Lodgepole pine has been well studied in this respect, but these same questions can and should be asked about other species as well.

Fig. 23.1
figure 1

Variations in growth and physiology of lodgepole pine (Pinus contorta) in relation to climate and provenance. a serial changes in several growth and tree-ring traits, including stable carbon isotope ratios emphasizing the excursion from the long-term mean in the drought year, 2002. b correlations among traits. c map of the seed sources and the planting sites. From Isaac-Renton et al. (2014)

Full size image

3 Control of Competing Vegetation

Managers often prepare sites for planting or improve growing conditions for young trees by removing competing vegetation with herbicides or mechanical treatments. These treatments increase the availability of light, water, and nutrients and improve tree growth. Because competition control increases the availability of several important resources simultaneously, stable isotopes can be useful for identifying the primary mechanism creating the growth response, particularly when complemented by supporting observations.

Competition control often increases ∆13C by releasing water to the trees, which would increase stomatal conductance. For instance, Ares and colleagues (2007) observed that five years of herbicide treatments reduced canopy cover of competing vegetation from 95 to 7% among newly established Douglas-fir saplings on the western slope of the Coast Range in the northwestern United States. For the Douglas-fir, competition control increased tree growth, foliar N concentrations, and soil moisture availability. In analysis of tree ring ∆13C of the Douglas-fir trees, earlywood ∆13C was unaffected by the herbicide treatments, but the latewood was higher in ∆13C, which led Ares et al. (2008) to conclude that increased soil moisture availability in the late summer was the primary driver of increased tree growth. Likewise, in young ponderosa pine seedlings, reduced competition reduced water stress and increased ∆13C (Pinto et al. 2011, 2012). Similarly, removal of understory shrubs from stands of Pinus densiflora in western Japan increased foliar ∆13C and both stomatal conductance and foliar N. Increased foliar ∆13C suggests that the observed increases in growth were caused primarily by increased water availability (Kume et al. 2003).

In contrast, competition control sometimes improves light or nutrient status, which increases photosynthesis, decreasing ∆13C. For instance, mechanical and herbicide treatments to control grass growing beneath 10-year old white spruce (Picea glauca) trees in boreal Canada also increased tree growth, foliar N, and soil moisture, but reduced foliar ∆13C (Matsushima et al. 2012). This reduction in foliar ∆13C, which was correlated with foliar N concentration and coupled with a concomitant increase in foliar δ15 N, led the authors to conclude that the increase in N availability was the primary mechanism by which competition control increased growth in their study. Similarly, competition control decreased red pine (Pinus resinosa) tree ring Δ13C in the north-central United States (Powers et al. 2010). Clearly, the physiological effects of competition control depend on which resources are released and the ability of trees to utilize these resources to acquire C.

4 Thinning

Managers thin forests primarily to increase growth of the remaining trees or to create stands that are more resilient to climate stress, insects, disease, and wildfire. Like the removal of competing vegetation, decreasing tree density within a stand reduces competition for resources that limit tree productivity. Thinning can increase the availability of water, nutrients, and light; trees respond physiologically to an increase in each of those resources with changes in the isotopic signatures recorded in tree rings.

We assessed the ∆13C response across all years and all studies in a meta-analysis (Fig. 23.2) and found that the direction of the discrimination change depended on the mean annual precipitation at the site. While we note that this analysis contains data from only a relative handful of studies, we presume that this reflects differences in physiological responses to thinning. At low precipitation sites, discrimination increases due to the water released to the remaining trees when the thinned trees are removed. The released water increases conductance more than it does photosynthesis, increasing ∆13C relative to the control. In contrast, at sites with high precipitation, ∆13C decreases after thinning, presumably due to improvements in the light or nitrogen status of the remaining trees, either of which would tend to increase rates of net photosynthesis. The water released by thinning sites must have less influence on ∆13C than the light and nutrient increases on these high-precipitation sites. These responses are unlikely to be species-specific responses as nearly all species respond to water stress by decreasing ∆13C (Granda et al. 2014; Grossiord et al. 2014).

Fig. 23.2
figure 2

Normalized change in discrimination ± SE Versus mean annual precipitation at the study site. ∆13C values are mean responses from each study from the period up to 10 years following the thinning treatment. 1. Brooks and Mitchell (2011) (Thin); 2. Brooks and Mitchell (2011) (Thin x Fertilization); 3. Powers et al. (2010); 4. Sohn et al. (2014); 5. Martin-Benito et al. (2011); 6. Leavitt and Long (1986); 7. Navarro-Cerrillo et al. (2019) (Pinus nigra); 8. Navarro-Cerrillo et al. (2019) (Pinus sylvestris); 9. McDowell et al. (2003) Stand A; 10. McDowell et al. (2003) Stand B; 11. McDowell et al. (2003)) Stand C; 12. York et al. (2010)

Full size image

Under dry conditions, thinning can mitigate drought stress and tree ring stable isotopes have helped illuminate the physiological mechanisms of drought resistance (Sohn et al. 2012, 2014, 2016). In an earlier meta-analysis of thinning effects on drought response, tree ring isotopic data revealed that in conifers, heavy thinning influenced ∆13C in a drought year more strongly than in the other years (Sohn et al. 2016). The heavy thinning treatments also changed δ18O more strongly in the drought year. However, thinning influenced the isotopic data both before and after the drought year as well. An analysis of ring-width data in the same paper showed clearer patterns, in part because it included broadleaf species. For the broadleaves, thinning appeared to increase soil moisture available to the trees during drought, allowing them to resist changes in growth during the drought year. Voelker et al. (2019) also found that stand basal area influenced drought sensitivity; stands at lower basal areas demonstrated decreasing trends in A/gs, and greater resilience to droughts. However, they found stands with high basal area had increasing trends in A/gs over the same period of time, which they attributed to greater water limitations and lower drought resilience. Kruse et al. (2012) also noted that stand density influenced the degree of stomatal control over A/gs.

In more mesic environments, thinning can increase exposure of the remaining canopy to both greater light and higher vapor pressure deficits, as well as increase nutrient availability (Fig. 23.2). For example, Pinus pinaster, Pinus radiata and Pinus resinosa all decreased ∆13C after thinning, presumably from an increase in A relative to gs from greater light exposure and/or greater leaf nitrogen content (Warren et al. 2001; Powers et al. 2010). However, for Pinus nigra, Pinus halepensis, and Pseudotusga menziesii growing in mesic environments, tree ring ∆13C did not change significantly with thinning while growth increased, implying that both A and gs increased in response to increased canopy light exposure and water resources.

Responses of trees to thinning are not always intuitive. Ruzicka et al. (2017) used ∆13C and ring widths to examine stand responses to thinning in three Douglas-fir sites that varied along a moisture gradient. Similar to what was reported above for mesic sites, trees in the thinned stands grew more at the wet and intermediate sites, but had no significant change in iWUE relative to the unthinned control trees. However, at the driest site, trees in the thinned stand did not produce larger growth increments, and increased in iWUE, which implies that water was not the main limiting factor. Instead, they speculated that gs decreased from greater exposure to the surrounding atmosphere and increased evaporative demand.

Forest thinning is usually applied at long intervals, with a decade or more to recover between operations. Therefore, the physiological responses to thinning should exhibit a regular temporal pattern in the subsequent years. As noted above, when water resources are limiting growth, density reduction should result in greater water availability to the remaining trees. An initial increase in water resources should increase stomatal conductance if leaf area, sapwood area and root area do not immediately adjust. As a result, iWUE would be expected to decrease and discrimination would be expected to increase in water-limited site. As shown by the red and pink points in Fig. 23.3, the expected increase did occur, but only on the driest sites and only in the first and second years after thinning. This may be due to increases in water availability leading to increased growth (McDowell et al. 2003, 2006; Giuggiola et al. 2016; Di Matteo et al. 2010). However, these stands varied in the duration of the response (Fig. 23.3). For example in eastern Oregon, McDowell et al. (2003) examined the change in cellulose δ13C in a stand of 250-year old ponderosa pine trees (Pinus ponderosa) where thinning selectively left the largest trees with wide spacing, removing 60–80% of the basal area. ∆13C increased a year after the thinning treatment and remained higher for at least 12 years when they collected cores, whereas growth differences were not noted until 4 years after the treatment. Predawn water potentials of the foliage were significantly less negative in the thinned trees compared to neighboring control stands, indicating that soil water availability was still higher in the thinned stands 12 years after the treatment. In another ponderosa pine stand located in Arizona, McDowell et al. (2006) noted that after a four-year lag, ∆13C values increased for approximately 12 years of a density reduction experiment (shown in Fig. 23.3 as Sohn et al. 2014, who re-analyzed the McDowell et al. 2006 data). In this study, the basal area increment (BAI) produced in a given year was related to this change in ∆13C, with higher levels of thinning producing a greater increase in ∆13C and greater increases in BAI. However, after 12 years, ∆13C returned to pretreatment levels even though stand densities of the treatments were maintained. They found that the whole tree leaf area to sapwood area had adjusted such that trees in the low basal area stands had significantly more leaf area per unit sapwood area, and suspected that increases in tree leaf area returned to a homeostatic balance of A/gs at pretreatment levels. Giuggiola et al. (2016) found similar results in a xeric Scots pine (Pinus sylvestris) thinning experiment they followed for over 40 years. At least in xeric locations, thinning does appear to increase water to the remaining trees initially, increasing stomatal conductance of the existing foliage. However, after a decade or so, the trees at xeric sites adjusted their leaf area and root area and returned to homeostasis in A/gs.

Fig. 23.3
figure 3

Carbon isotope discrimination (± variance) by year after thinning, where year 0 is the year the thinning treatment was applied. Data series displayed in rank order of wet (dark blue) to dry (dark red) based on mean annual precipitation. See Fig. 23.2 for precipitation values and citations

Full size image

Our meta-analysis of δ18O responses to thinning was based on few studies, but we again found different responses depending on the precipitation amount at the site (Fig. 23.4). The change was positive on the wetter sites and negative on the one drier site. Increases in cellulose δ18O indicate that the environment around the trees changed in some way, perhaps with decreased relative humidity, increased leaf temperatures, and/or shallower root water uptake for the thinned trees (Brooks and Mitchell 2011; Martin-Benito et al. 2011; Moreno‐Gutiérrez et al. 2011). These studies with increased δ18O assumed that source water between control and thinned plots did not change, so speculated that the increases in δ18O from thinning were from increased leaf temperatures from increased solar irradiance, and decreased relative humidity around the foliage from increased boundary layer conductance in the open stands (Jarvis and McNaughton 1986). Canopy coupling to the ambient air is generally high for conifers (Jarvis and McNaughton 1986), which would minimize the change in boundary layer with thinning, but thinning in Brooks and Mitchell (2011) removed 2/3 of the basal area, changing the remaining trees from being in dense closed canopies to having completely exposed canopies (Martin et al. 1999). This dramatic shift in canopy density could be enough change in leaf temperature and relative humidity to cause the modest change in δ18O. However, the assumption of constant water sources between treatments may not be accurate as water uptake may shift within the soil profile. An upward shift in root water uptake in thinned stands might favor more enriched source water (Brinkmann et al. 2018). The downward shift after thinning on the drier site is difficult to explain, but the authors speculated a decrease in the isotopic value of source water in the thinned stands (Sohn et al. 2014). Higher transpiration rates in the thinned stands would lead to use of deeper, isotopically more depleted water sources (Chap. 18). This shift in water uptake might be consistent with the hypothesis presented earlier for these same stands, that the root:shoot balance is modified such that the roots can reach deeper depleted water. In future studies, measurements of xylem source water and leaf temperature between treatments would greatly assist in the process of choosing among these potential explanations.

Fig. 23.4
figure 4

Difference in δ18O of tree rings between thinned and control stands as a function of time since thinning. Positive values indicate higher δ18O in the thinned stands. Colors designate mean annual precipitation as in Fig. 23.2. 1. Brooks and Mitchell (2011) (Thin); 2. Brooks and Mitchell (2011) (Thin x Fertilization); 3. Martin-Benito et al. (2011); 4. Sohn et al. (2014); 5. Powers et al. (2010)

Full size image

One additional complication to interpreting tree ring stable isotopes in trees is that stress may increase their reliance on stored carbohydrates (Helle and Schleser 2004). As a result, slow-growing stressed trees appear to have less isotopic variance over time compared to neighboring trees in stands that have been thinned (McDowell et al. 2006; Sohn et al. 2014). Both δ13C and δ18O showed much lower variance between years in control stands and in pretreatment values as compared to after the density reduction (McDowell et al. 2006; Sohn et al. 2014). This difference in isotopic variance for both δ13C and δ18O could be caused by a greater reliance on stored carbohydrates in the slow-growing control trees, which would mute the isotopic response to perturbation in any particular year.

5 Fertilization and Nutrition

Nutrient fertilization is typically applied to forest stands after a pre-commercial or commercial thinning or after stand differentiation has occurred in order to accelerate tree growth and shorten the rotation (Miller 1981). As with thinning, the date of fertilization provides an obvious reference point for analysis of its effects, and tree ring studies can compare pre- and post-treatment responses. As noted above, the response of ∆13C to fertilization depends on the mechanism of response. If fertilization increases the nitrogen concentration within foliage, photosynthesis would be expected to increase (Brix 1971, 1981). Since water resources are not directly altered by fertilization, stomatal conductance could be expected to remain similar to pre-fertilization rates. Thus, increases in foliar N concentration are expected to decrease ∆13C and increase iWUE.

In the meta-analysis of ∆13C, we found a short-term response that generally supported this scenario (Fig. 23.5). Discrimination decreased in years 1 and 2, however the significant effect disappeared thereafter. However, the responses among individual studies were rather different (Fig. 23.6). For example, Brooks and Mitchell (2011) found reduced discrimination, yielding the negative values in Fig. 23.6, for at least four years after fertilization in a nutrient-limited coastal Douglas-fir stand. These decreases in discrimination were correlated with increases in measured leaf nitrogen levels (Mitchell et al. 1996). Similarly, Brooks and Coulombe (2009) found a decrease in ∆13C for four years after fertilization in another nutrient-limited Douglas-fir stand in the Cascade Mountains, and speculated a similar increase in foliage nitrogen. Increasing foliar nitrogen is a common short-term response in these fertilization studies and as with studies of competition control and thinning, the increase in foliar N can help support interpretation of ∆13C and δ18O observations. For this reason, we suggest that measurement of foliar N should be incorporated in future stable isotope studies of leaf gas exchange.

Fig. 23.5
figure 5

Mean ∆13C effect by time since fertilization from meta-analysis. Numbers near means represent the sample size of values (contrasts) included in the analysis for each year

Full size image
Fig. 23.6
figure 6

Differences in carbon isotope discrimination between fertilized and control stands. Negative values indicate lower discrimination and higher iWUE on the fertilized plots. 1. Brooks and Coulombe (2009); 2. Brooks and Mitchell (2011) (Fertilization); 3. Brooks and Mitchell (2011) (Thin xFertilization); 4. Kruse et al. (2012) (Daylesford); 5. Kruse et al. (2012) (Lyons); 6. Elhani et al. (2005); 7. Krause et al. (2012); 8. Walia et al. (2010); 9. Balster et al. (2009) (First Fertilization); 10. Balster et al. (2009) (Second Fertilization)

Full size image

In the longer term, fertilization can also cause increases in foliage leaf area, which may dilute the foliar nitrogen (Balster and Marshall 2000; Binkley and Reid 1985). If leaf area increases without an increase in foliar N levels, then foliar gas-exchange rates and ∆13C may not change. Photosynthetic rates may even decrease due to canopy shading, in which case ∆13C might increase. Note that this would offset the decrease that follows the fertilization immediately after the application. In the Brooks and Mitchell (2011) study, the decrease in ∆13C was observed even in the 10th and 11th years after fertilization. Likewise, the growth increase lasted many more years and both stands were documented to have increased in leaf area. It is plausible that longer-term decreases in ∆13C in response to fertilization could be a consequence of increased leaf area or height growth, either of which might lead to stomatal closure (McDowell et al. 2002, 2011). In contrast to Brooks and Mitchell (2011), Balster et al. (2009) found no significant shifts in ∆13C from fertilization in Douglas-fir at 24 sites located in the interior northwest USA, although growth responses lasted for 8 years after treatment. However, leaf area index was increased (Balster and Marshall, 2000), providing a likely increase in canopy photosynthesis without a change in the gas-exchange setpoint reflected in ∆13C.

Antecedent site fertility influences the longevity of the ∆13C response to fertilization. Cornejo-Oviedo et al. (2017) found an increase in iWUE in both late and earlywood one year after fertilization in a productive coastal stand, in contrast to the 3–4 year increase in more nutrient limited stands mentioned above. Site moisture availability also has an influence. Liles et al. (2019) found no changes in iWUE in their most productive, wettest ponderosa pine site, but found increases in iWUE from fertilization at their drier sites.

Changes in δ18O in response to fertilization might be expected if fertilization influenced transpiration, the canopy microclimate (mostly relative humidity) or the source water accessed by the trees (Chaps. 10, 16, and 18). Some fertilization studies have reported strong changes to δ18O, while others have reported no significant changes. Brooks and Coulombe (2009) found a 2% increase in latewood δ18O with fertilization that lasted nearly a decade with the highest level of fertilization, but no change relative to controls for earlywood. They speculated that the δ18O increase in latewood was caused by the combination of high leaf-area index and late-summer drought in the dry Mediterranean summers of the Pacific Northwestern USA.

Although δ18O data showed significant treatment effects to the fertilization in a range of studies (Fig. 23.7), the effects were in different directions and they occurred at different times. Interpretation is limited by the small number of studies and the many possible sources of variation (Roden and Siegwolf 2012). Especially important are understanding which δ18O differences are caused by changes in canopy conditions and exposure following thinning and fertilization, as well as source-water variation (Sarris et al. 2013). Source water could be tested by measuring the isotopic composition of xylem water (Marshall et al. 2020; Volkmann et al. 2016; White et al. 1985). We strongly recommend these measurements to minimize the uncertainties in these studies.

Fig. 23.7
figure 7

The change in δ18O of tree rings induced by a fertilization event at year zero. Positive values indicate that the δ18O of the fertilized plots were higher than those of the controls. 1. Brooks and Coulombe (2009); 2. Brooks and Mitchell (2011) (Fertilization); 3. Brooks and Mitchell (2011) (Thin x Fertilization); 4. Kruse et al. (2012) (Lyons); 5. Krause et al. (2012)

Full size image

Nitrogen fertilization can also influence the δ15 N of tree rings (Elhani et al. 2005; Peñuelas and Estiarte 1996; Poulson et al. 1995), particularly when the applied nitrogen has a different isotopic composition than native nitrogen in the soil, or when the fertilizer material tends to volatilize, leading to strong kinetic fractionation and a δ15 N increase in the remaining nitrogen. This occurred in a study in which urea was applied to Douglas-fir forests in the northwestern US (Balster et al. 2009). In this study, the urea was converted to NH4+, which is quickly transformed to NH3 and released to the atmosphere. Although the fertilizer could not be detected through increases in wood N concentrations, the δ15 N was significantly enriched in fertilized trees. Tree ring δ15 N analyses are difficult because the nitrogen concentrations in wood are so low compared to leaves or roots (Chap. 12). Further, interpretation of tree ring δ15 N data should recognize that the labeled nitrogen can be translocated across the rings over time (Balster et al. 2009; Hart and Classen 2003; Sheppard and Thompson 2000), unlike the isotopic labels embedded in the C, H, and O in cell walls.

Experimental long-term annual fertilization is sometimes applied in forests. In such cases, changes in δ15 N of plant tissues can be stark (Högberg and Johannisson 1993; Marshall and Linder 2013). Although they do not simulate standard management practice, these studies provide clear insights into mechanism of response. These clear differences are attributed to a combination of decreased N retention by mycorrhizal fungi and increased leaching of depleted nitrate, both of which tend to enrich 15 N in leaves and tree rings (Högberg et al. 2014).

6 Modeling of Management Effects

Although stable isotopes are often used for assessment of treatment responses, the milieu of environmental, ecological, and physiological changes can be complex and difficult to interpret. These complexities can be addressed via models of tree ecophysiology and forest growth (see Chap. 26). This is especially important because forest management decisions are often driven by expectations about future growth. Wei et al. (2014) used ∆13C data to parameterize a physiological growth model in an attempt to explain how large increases in tree growth were obtained from replicated fertilization and herbicide treatments. The model, 3-PG, includes an empirical parameter that describes the influence of nutrition on photosynthate allocation. Wei et al. (2014) set that parameter to its maximum in the treatments that combined fertilizer and herbicide, then empirically fitted the parameter to the other treatments. This was possible because all of the major fluxes of carbon and water were measured, as was the ∆13C of photosynthate. With so many parameters constrained by measurements, the allocation parameter could be fitted with more confidence.

The 3-PG model has also recently been modified to predict tree-ring δ18O (Ulrich et al. 2019). The model was applied to ponderosa pine trees growing over 107 years in a riparian and an upslope area. Key parameters in the δ18O model remained difficult to measure or verify, in particular the Péclet effect, which describes the back-diffusion of enriched water against the xylem flow into the leaf (Chap. 10). However, the model provided a validated estimate of the gas-exchange predictions based on their isotopic consequences. Such constraints would increase the confidence of model predictions as it is used, for example, to predict responses to changed climates, genotypes, or novel management regimes.

These models might be usefully combined with the meta-analyses presented above. In particular, it would be worthwhile to vary precipitation amount or soil depth to create differences in water availability and then thin or fertilize the modeled stand. If such an exercise could recreate the pattern observed in Fig. 23.2, it would increase our confidence in interpreting it as an effect of water availability. Likewise, if the site-specificity of the fertilization response in Fig. 23.6 could be recreated by site-specific parameterization of the models, this would help to explain the differences in direction and timing of the fertilization response.

7 Conclusions

Stable isotope analysis of tree rings is a powerful tool for exploring the long-term and complex physiological responses to forest management. The thinning response differs between wet and dry sites, where the trees reduce their water-use efficiency on dry sites and increase it on wet. These patterns are consistent with the release of water limitations on dry sites and of light and nutrient limitations on wet sites. Carbon isotope discrimination data revealed that fertilization has two dominant effects: a short-term increase in foliar nitrogen, and a longer-term increase in leaf area. We have noted the timing and site-specificity of the isotopic changes after thinning and fertilization, which indicate that physiological responses to these management actions can last for decades, but the responses vary in direction and timing among studies. We encourage the development and parameterization of models to test the mechanisms proposed here and to improve predictions about management responses at a given site. Stable isotope tree-ring studies on management actions could reduce uncertainties and explain mechanisms of response, especially with increased auxiliary measurements of leaf nitrogen, leaf area, and xylem and soil water isotopes.