Ecological Archives E086-033-A3

Daniel M. Kashian, Monica G. Turner, William H. Romme, and Craig G. Lorimer. 2005. Variability and convergence in stand structural development on a fire-dominated subalpine landscape. Ecology 86:643–654.

Appendix C. A discussion of potential error sources in age estimation.

Our study includes a series of age distributions that are somewhat broader than have typically been described for post-fire lodgepole pine stand development, particularly for young (<100 years), high-density stands (e.g., stands 0022, 0030, 0119, and 0018 in Appendix D). Stands characterized by these distributions are interpreted as having long periods of tree recruitment, even when much of the available growing space is presumed occupied by a dense overstory that established during the first few decades following the fire. In this Appendix, we evaluate our confidence in the age estimations we present in this study, evidence for similar trends in the literature, and the potential explanations for these trends. We present the following points:

1.  Given the high density and low site index of many of our stands, the potential for missing or indistinguishable rings in increment cores sampled from these stands is moderate to high. If rings are missing from cored trees and ages are underestimated, age distributions that include these series may be broader than they should, suggesting a longer period of recruitment than what has actually occurred. Furthermore, our inability to harvest mature trees within the National Park and our consequent need to extract increment cores 30 cm above ground level may have led to additional errors in age estimation. 

Several authors have recently noted the inaccuracies of estimating tree ages using increment cores taken above the root collar. For example, Gutsell and Johnson (2002) reported up to 43-year underestimations in mid- and late-successional boreal tree species, although early successional species (which, though not sampled in their study, would include lodgepole pine) were far less inaccurate (maximum error = 11 years). Working in white spruce stands younger than 40 years, Peters et al. (2002) noted that trees ages were underestimated by up to 6 years when cored above the root collar. Vasiliauskas and Chen (2002) and Wong and Lertzman (2001) found that the age of early successional species were substantially underestimated when cored at breast height (1.3 m).

2.  Although many tree ages in young dense stands are likely to have been underestimated in our study, we believe that our methodology has been sufficiently cautious in age estimation so that dating errors, where present, have minimal impact on the findings of our study. First, the areas adjacent to our study sites burned by the 1988 YNP fires provided ample opportunity to sample sapling growth rates following a stand-replacing fire. We harvested 400 saplings in the 1988 burned area, at densities ranging from 100 stems/ha to >100,000 stems/ha, and estimated their age at 30 cm; this strategy allowed us to develop a correction factor with which we could correct the age of trees cored at 30 cm. In addition, we harvested a minimum of 5 trees at the base from stands 0022, 0030, 0114, 0119, and 0018, all of which had known stand origin dates and densities >5,000 stems/ha, to verify stand age, and all but one of these 25 trees contained no missing rings. From these data, we estimate our errors in tree dating to be <5 years, which is consistent with dating errors estimated for lodgepole in YNP by Romme (1982). An underestimation of tree age of this magnitude is not nearly large enough to eliminate the long establishment period seen in our young, dense stands, and is hardly enough to affect our age distributions whatsoever.

We reiterate the reasoning of Lieffers and Stadt (2003), who in response to Peters et al. (2002) note two important points. First, accuracy of concept does not necessarily require accuracy of measurement in constructing age distributions. There is a trade-off between reporting absolutely accurate ages, which would require extensive work in harvesting trees that would preclude working in a National Park or other wilderness area, and more rapid, less intensive and expensive techniques that may be less accurate and precise but can be applied to a wider range of stand conditions. Secondly, although fine-scale interpretations of our age distributions are probably not appropriate for our data with a 5-yr error estimate, our strategy of aging trees is certainly most appropriate for a landscape scale study that requires the characterization of nearly 50 sites of varying density (Lieffers and Stadt 2003).

3.  Several previous studies of lodgepole pine have noted multi-aged stand structures. For example, Whipple and Dix (1979) noted that older lodgepole pine stands in Colorado often developed a bimodal age distribution without being replaced by other conifers, and Peet (1981) and Komarkova et al. (1988) were also cognizant of variability in age structures of lodgepole pine in the Rocky Mountains. Brown (1975) and Arno (1980) noted that lodgepole pine stand structures were closely linked to fire severity and frequency, a result echoed by the work of Turner et al. (1997) in Yellowstone. Despain (1983) documented self-replacing lodgepole pine stands that regenerated without disturbance in YNP, and Jakubos and Romme (1993) noted uneven-aged stands of lodgepole pine resulting from conifer invasion of meadows. Muir (1993) also noted continuous recruitment of lodgepole pine over long periods, though most of these stands were found on former clearcuts or other areas characterized by human disturbances. The work of Parker and Parker (1994), who suggested that open, uneven-aged, 120–140-yr stands in Colorado may result from low initial post-fire densities is the most closely related research to our study. Clearly, uneven-aged or multimodal lodgepole pine stands are not uncommon in the Rocky Mountains.

4.  Historically, studies of stand dynamics in lodgepole pine or other coniferous forests characterized by stand-replacing fires have not considered the importance of the spatial distribution of trees within a stand. The traditional dogma assumes that trees at a given density (but especially at high density) are uniformly distributed across a stand, such that little or no growing space is left unoccupied once crown closure occurs. Our study shows that tree spatial distribution in young, dense stands is clustered, such that gaps – with available growing space – exist even in very dense stands. Such "gaps" are not widely distributed in dense stands, but we argue that they are common enough to allow additional tree establishment for up to 40–50 years following a stand replacing fire. We offer three lines of evidence for this argument:

   I.  Sapling spatial patterns observed within 10 years of the 1988 fires in YNP show the existence of gaps in very densely regenerating stands.  In a data set partially published in Plotnick et al. (1996), long transects were established across burned patches of varying size and density, along which post-fire lodgepole pine saplings were counted in continuous 1-m2 quadrats. These transects showed areas of few or no seedlings, ranging from 5 to 100 m wide, even in patches having >50,000 stems/ha; Plotnick et al. (1996) calculated that stem density varied at scales as small as 5 m. Although these gaps have shown little additional recruitment at 15 years following the fire (likely due to a lack of seed source), the potential for additional recruitment is present for the next several decades as saplings mature into cone-bearing trees.

   II.  The establishment of additional trees into gaps is evident in many of the young, dense stands we sampled and is evident from our stand maps.  Figure C1 shows two maps of stands between approximately 70 and 100 years old. In both stands, most trees we have aged as younger (to simplify the figure, we have classified all stems older than the stand origin – 40 years as "older trees", and those younger than this age as "younger trees") are clustered together into areas not occupied by older trees. We interpret these clusters as a "filling-in" of these gaps over time following the initial post-fire recolonization of the stand. Furthermore, we argue that our long plots (10 × 50 m) were more successful in capturing this spatial distribution of tree ages, and will result in relatively broad age distributions perhaps atypical of those often constructed for dense lodgepole pine stands.

   III.  Observations of crown characteristics suggests that tree establishment has occurred over a long period of time even in high density stands.  Fifty random trees were selcted in each stand and their crown characteristics (tree height, crown depth, and canopy position) recorded. As shown in Table C1, the young, dense stands mentioned above each include a significant number of "small trees". This vertical variability suggests the presence of trees of different ages, or at least that the light environment in these stands is favorable to persistence of smaller trees within gaps in dense stands.

In conclusion, our age data exhibit important trends that are not likely artifacts of errors in age estimations. The contagious nature of trees in young, dense stands may allow the establishment of additional trees into the stand for several decades following the disturbance, such that age distributions for high density lodgepole pine stands in YNP may be more broad than traditionally reported for similar stands in the Rocky Mountains.

 

 
   FIG. C1. Stand maps of two stands within 10 × 50 m plots. Open markers indicate trees younger than the stand age – 40 years; closed symbols represent trees older than the stand age – 40 years. Units are in meters, and maps are not to scale. Younger trees tend to be clustered into gaps not occupied by older trees.  

 

TABLE C1. Canopy characteristics for five selected young, dense stands. Note the significance of the smaller-tree component in all stands.

 

Stand
0018

Stand
0022

Stand
0030

Stand
0114

Stand
0119

 

 

 

 

 

 

Stand age, years

125

99

100

45

91

Stand density,
stems/ha

9240

11320

7000

8520

6400

Tree height,
coefficient variation (%)

21

27

21

18

20

Tree height,
range (m)

4.3 – 12.3

2.8 – 8.0

6.0 – 16.5

4.5 – 9.5

6.3 – 15.5

Crown depth,
coefficient variation (%)

36

34

34

29

36

Crown depth,
range (m)

1.0 – 6.8

0.9 – 4.5

1.0 – 6.8

1.3 – 6.0

1.0 – 8.5

Codominant trees (%)

60

48

56

50

78

Intermediate trees (%)

30

24

32

38

20

Overtopped trees (%)

10

28

12

12

2


LITERATURE CITED

Arno, S. F. 1980. Forest fire history in the northern Rockies. Journal of Forestry 78:460–465.

Brown, J. K. 1975. Fire cycles and community dynamics in lodgepole pine forests. Pages 429–455 in  D. Baumgartner, Editor. Symposium proceedings: Management of lodgepole pine ecosystems. Washington State University, Pullman, Washington, USA.

Gutsell, S. L., and E. A. Johnson. 2002. Accurately aging trees and examining their height growth: implications for interpreting forest dynamics. Journal of Ecology 90:153–166.

Lieffers, V. J., and K .J. Stadt. 2003. Comment on “Aging discrepancies of white spruce affect the interpretation of static age structure in boreal mixedwoods.” Canadian Journal of Forest Research 33:2280–2281.

Peters, V. S., S. E. MacDonald, and M. R. T. Dale. 2002. Aging discrepancies of white spruce affect the interpretation of static age structure in boreal mixedwoods. Canadian Journal of Forest Research 32:1496–1501.

Plotnick, R. E, R. H. Gardner, W. W. Hargrove, K. Prestegaard, and M. Perlmutter. 1996. Lacunarity analysis: a general technique for the analysis of spatial patterns. Physical Review E 53:5461–5468.

Vasiliauskas, S., and H. Y. H. Chen. 2002. How long do trees take to reach breast height after fire in northeastern Ontario? Canadian Journal of Forest Research 32:1889–1892.

Wong, C. M., and K. P. Lertzman. 2001. Errors in estimating tree age: implications for studies of stand dynamics. Canadian Journal of Forest Research 31:1262–1271.



[Back to E086-033]