Tree Rings and Climate: Sharpening the Focus - Dendroclimatology - page 339

as well as the building of networks of chronologies according to largely consistent
protocols. A number of the basic concepts are discussed by Hughes ( Chapter 2 ,this
volume), and the development of the networks was described by Hughes ( 2002 ) . An
apparent tension has emerged between the empirical-statistical approach, on which
most of the achievements of dendroclimatology have been based, and process-based
modeling. Vaganov et al. ( 2006 , p. 308) suggest that process-based modeling (see
Vaganov et al. Chapter 3 , this volume) should be viewed as a complement to
empirical-statistical techniques (see Cook and Pederson, Chapter 4 , this volume),
not as a substitute. Following Harte ( 2002 ) , they assert that, 'Empirical-statistical
tools are of great value in dealing with particularity and contingency, just as
process-based modeling may be a way of approaching generality and simplicity.'
So it is that the building and analysis of networks of dendroclimatic records have
brought to light emergent phenomena (see Cook and Pederson, Chapter 4 , this vol-
ume) that would likely not have been anticipated based only on the available state of
knowledge of mechanisms; for example, elements of synchrony in drought between
the midlatitudes of North and South America over several centuries (see Villalba
et al. Chapter 7 , this volume), or the possible existence of two continental-scale
modes of drought in North America (Woodhouse et al. 2009 ) . The solutions to some
of the most pressing problems in dendroclimatology, however, must lie in better
understanding of mechanisms; for example, of possible changing of climate control
of ring growth (as in the 'divergence' problem; see Vaganov et al. Chapter 3 ,this
volume; Briffa et al. 1998 ; D'Arrigo et al. 2008 ) , in the effects of stand dynamics,
or in the mechanics and geometry of tree development. It is, of course, also nec-
essary to identify and deal with artifacts produced by combinations of chronology
composition and standardization techniques (see Briffa and Melvin, Chapter 5 ,this
volume).
11.8 The Importance of Networks
Modern dendroclimatology has only been possible because of the development of
networks of chronologies at local, regional, continental, and hemispheric scale (see
Hughes, Chapter 2 , this volume, for a discussion of the centrality of networks to
dendroclimatology). Schulman ( 1956 ) systematically built networks of tree-ring
chronologies collected as records of hydroclimate, demonstrating a strong common
signal across hundreds of kilometers. This development was expanded further by
Fritts and colleagues (Fritts 1965 , 1976 ; Fritts et al. 1971 , 1976 ; LaMarche and Fritts
1971 ; Blasing and Fritts 1976 ) , who used a distributed network of tree-ring sites
to reconstruct spatiotemporal patterns of regional climatic variability prior to the
beginning of instrumental observations. They developed methods for mapping tem-
poral and spatial variations of tree-ring growth for a network of western US conifers
that they linked to changes in large-scale atmospheric circulation patterns (sea level
pressure [SLP], cyclone frequency) in the north Pacific. Fritts ( 1991 ) summarizes
the basic approach and provides examples of the application of these techniques to