Austral Ecology (2007) 32, ••–••

doi:10.1111/j.1442-9993.2007.01705.x

Crown dieback events as key processes creating cavity habitat for magellanic woodpeckers

RO OF

VALERIA S. OJEDA,* M. LAURA SUAREZ AND THOMAS KITZBERGER Laboratorio Ecotono, Universidad Nacional del Comahue – Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Argentina (Email: [email protected])

RR EC

TE

DP

Abstract Woodpeckers are considered keystone species for webs of cavity nesters and habitat and resource specialists that strongly depend on availability of trees suitable for cavity excavation. Most studies carried out in northern hemisphere temperate coniferous forests emphasize the importance of old growth stages of forests or large dead trees as habitat for cavity builders. We present a study of Nothofagus pumilio tree selection by the magellanic woodpecker (Campephilus magellanicus) that incorporates dendroecological data on long-term growth trends of trees that provides new insights into the processes that create suitable habitat for cavity excavating species. We analysed 351 cavity and neighbouring control trees in terms of age and radial growth patterns, as well as external tree characteristics. In addition, from a subsample of these trees we developed tree-ring chronologies for each group using standard methods in order to analyse potential differences in radial growth patterns between cavity and non-cavity trees. Multivariate models that account for differences between paired cavities versus control trees indicated that growth decline and the degree of crown dieback were the primary variables explaining magellanic woodpecker tree selection for cavity building. In contrast to previous work, neither diameter (above a certain threshold) nor age, were important determinants of selection. Furthermore, trees that became present cavity are those that had synchronously declined in radial growth during the 1943–44 and 1956–57 droughts and the 1985–86 massive caterpillar defoliation. Insect outbreaks and extreme climatic events may episodically reduce vigour, induce partial crown mortality, trigger increased fungal attack and heart rot formation at different tree heights on the bole in a group of trees and thus increase availability of soft substrate and their likelihood of cavity excavation by the magellanic woodpecker. These results underscore the importance of drought/biotically-induced canopy dieback events in creating habitat for woodpeckers and their dependent cavity users. Key words: Campephilus magellanicus, crown dieback, keystone process, Nothofagus forests, tree selection.

INTRODUCTION

UN

CO

Wildlife that use tree cavities as nest, roost or den sites are a major component of the world’s forests. Processes that modify the availability of appropriate cavity-nesting trees will have a key impact not only on primary cavity excavating species, but also on a suite of species that largely rely on the availability of readily excavated cavities (Martin & Eadie 1999; Bonar 2000; Gibbons & Lindenmayer 2002; Bednarz et al. 2004). In this sense, strong emphasis has been given to anthropogenic processes that degrade key habitat characteristics or reduce key substrates by timber harvest and forest fragmentation that have lead several species to extinction or severe threat (Lammertink 1996, 2004; Mikusinski & Angelstam 1997; Styring & Ickes 2001). Less attention has however, been paid to long-term natural processes that improve habitat by increasing the availability/quality of appropriate substrates for primary excavators. *Corresponding author. Accepted for publication September 2006.

© 2007 Ecological Society of Australia

A new emerging issue is the role of fungal decay in creating wood for cavity excavation (Jackson & Jackson 2004). It is increasingly clear that fungal infection is a vital pre-existing condition for transforming wood into cavities for nesters. It is generally assumed that probabilities of fungal attack increase with tree age. However, the processes that lead to fungal infection are multiple, complex and rather system-dependent. Fungal infections can be favoured by insect attacks, mistletoe infections, and injuries, damage by disturbances or climatic fluctuations and even by woodpeckers themselves, and preconditioning causes can be related to tree species, growth condition, and site characteristics, among others (Jackson & Jackson 2004). Thus, rather than analysing static stages of the forest it is necessary to record stand growth/development history better in order to understand the events or chains of events that lead to fungal infection and eventually to selection by cavity excavators. Despite the fact that the Neotropical Region has by far the greatest number of woodpecker species, most studies of woodpecker tree selection have been conducted in conifer forests of the Northern Hemisphere.

V. S . O J E DA ET AL.

growth patterns, to understand factors that create suitable substrates for cavity excavation.

RO OF

METHODS Study area

Tree selection by magellanic woodpeckers was assessed in a resident population in the pristine N. pumilio forests of Challhuaco Valley, Argentina. Cavity excavation patterns and reproductive success of this population have been monitored for several years (see Ojeda 2004, 2006 for details). The study area corresponds to the Andean portion of northern Patagonia, Argentina. Physiographically, the area consists of lakes, glacial valleys, and mountain slopes covered by forests dominated by southern beech (Nothofagus spp.). Elevation ranges from 400 to 3480 m a.s.l. The climate is characterized by a relatively long winter period of precipitation (snow and rainfall ranging 500– 2000 mm year-1) and dry summers (December– March). Mean annual temperature is 8°C (Paruelo et al. 1998). Challhuaco Valley (hereafter, CV) is located about 15 km south-east of Bariloche (41°08′S, 71°12′W). It is a rugged area; between 900 and 1600 m a.s.l, most slopes – from moderate to very steep – are covered by pure old-growth deciduous N. pumilio forest that extends approximately 2370 ha, and is connected to forests from adjacent valleys. Although most of the CV forest is virtually pristine, about 10.4% of its surface was burnt in January 1996, while about 6.3% was partially burnt with patches of living trees (mostly along watercourses) surrounded by burnt forest matrix. Depending on site characteristics, the forest averages 15–24 m tall. The understory is open, dominated by a few shrubs and herbaceous species.

UN

CO

RR EC

TE

A pervasive result of these studies is that woodpeckers in general are habitat and resource specialists that strongly depend on the availability of certain forest habitat elements, such as dead snags or large rotten trees for cavity excavation (Carlson & Aulén 1990; Welsh & Capen 1992). This generalization may be however, strongly biased by the inherent dynamics of northern conifer forests that gradually decline and have lower susceptibility to fungal attack than angiosperm wood. Thus the importance of old growth stages as habitat for cavity nesters has been overemphasized (e.g. Hartwig et al. 2004). Angiospermdominated forests such as many in southern temperate regions have clearly very different dynamics, namely exogeneous (climatic, disturbance) processes that induce extensive dieback irrespective of tree age (Veblen et al. 1996; Suarez et al. 2004), that induce tree decline long before normal ageing would do. We claim that differences in causes and temporal patterns of tree decline may determine important differences in how cavity excavators select trees for nesting. Methodologically, studies of nest-cavity tree selection have generally focused on descriptions of forests and trees based on external features and in a few cases on recent growth and ageing of trees (e.g. Picoides borealis, Jackson et al. 1979; Conner & O’Halloran 1987; Rudolph & Conner 1991). More formal dendroecological approaches to reconstruct growthrelated changes in relation to selection of trees have been lacking. While tree age and recent growth may give an indication of the susceptibility of trees to be selected by cavity nesters, long-term patterns of growth of cavity versus non-cavity trees may provide insight into what determines whether trees become suitable for being selected by excavators and bearing cavities. The magellanic woodpecker (Campephilus magellanicus), a year-round resident of austral temperate forests of South America (Short 1970; Vuilleumier 1985) is the principal cavity excavator in these forests and a potential keystone habitat modifier for at least seven other cavity-nesting bird species (McBride 2000; Ojeda 2006). This picid is of special concern as it has dramatically declined in numbers in areas of Chile where logging and other kinds of forest disturbance have severely reduced native forest habitat (Glade 1993; Willson et al. 1994; Rozzi et al. 1996; Saavedra 1998; Cofré 1999; Estades & Temple 1999; Vázquez & Simonetti 1999; WWF 1999). While there are recent studies on the ecology of this species (Ojeda 2004; Vergara & Schlatter 2004) it is not clear how it selects trees for cavities, nor how forest dynamics affect the suitability of stands for cavity habitat. The aim of this research was to compare attributes of magellanic woodpecker cavity and neighbour trees, with an emphasis on the age and tree-ring reconstructed radial

DP

2

Cavity-tree assessment Between 1998 and 2003, 196 magellanic woodpecker cavity trees were located, tagged and monitored for completion and use throughout the CV. In order to avoid examining trees that may not represent present selection conditions by woodpeckers, we adopted a restrictive sampling design by only considering a subset of cavity trees with current or recent woodpecker use. Discrimination of cavity status (used or inactive) was made on the basis of direct (most cases) or indirect evidence. Direct evidence refers to behavioural observations of individuals engaged in nesting, excavating, or night roosting activities. Indirect evidence consisted of fresh feathers and droppings in the cavity, or signs of cavity maintenance, such as © 2007 Ecological Society of Australia

D I E B AC K A N D C AV I T Y- T R E E S E L E C T I O N

RO OF

beyond a 15 m radius from the cavity tree that met the diameter/mortality criteria was sampled as a control. Therefore, we sampled four control trees for each cavity tree, each control tree being located in a different quadrant. The 15 m minimum distance and quadrant arrangement were imposed to include the nearest neighbours but avoiding sampling trees whose age or other attributes could be influenced by mutual interactions, particularly, the potential effects from the central (cavity) tree, which was a priori expected to have been dominant. For each tree we measured tree height (m) with a clinometer, diameter at breast height (cm, hereafter d.b.h) and degree of crown dieback (six classes: 0 = healthy crown, 1 = 25% crown dead, 2 = 26–50% crown dead, 3 = 51–75% crown dead, 4 = crown nearly dead, 5 = dead crown). Using increment borers, two increment cores at 0.3 m above ground were obtained from all cavity and control trees. Samples were mounted and sanded with progressively finer grades of sand paper. Rings were dated and the last 100 years of growth were measured to the nearest 0.01 mm with a computer-compatible increment tree ring-measuring machine, in order to obtain growth trend patterns of cavity and control trees. We followed Schulman’s (1956) convention for the Southern Hemisphere by assigning tree rings the calendar year when growth began. To estimate age in samples that did not reach the pith we used Duncan’s (1989) geometrical procedure setting a maximum of 10 estimated missing rings above which age was not estimated (e.g. broken cores, trees with very rotten centres). As an indication of recent radial growth trends we integrated the radial width (mm) of the last 19 rings of all core samples. Recent growth of cavity trees can be important in cavity-tree selection because they could affect tree appearance and wood hardness.

Tree sampling protocol

RR EC

TE

DP

substantial enlargement of inner chambers, fresh gouges and missing bark on the entrance rim, and wood chips spread on the ground. Cavity trees without any sign of current use by woodpeckers were excluded from the sample. Sequential observations at the CV indicated that once a cavity was abandoned it developed a plate around the entrance (and its contents broke up) in just a few months, so we are confident that even the indirect status assessment was reliable. Moreover, if the woodpeckers were using cavities they did not maintain, the resulting bias was that we did not sample some cavity trees that were actually active. We classified cavities in discrete categories as follows. Nests were defined as those cavities that had contained magellanic woodpecker eggs or nestlings any time between 1998 and 2002. Starts were defined as incomplete cavities that nevertheless were used as roosts, and/or had undergone an active excavation process (i.e. periodic enlargement) by the time the field work was carried out. Roosts were defined as completed cavities (i.e. about the size of nests, Ojeda 2004) used for roosting. It was impossible to know (but irrelevant) if some of the roosting cavities had originally been nests. Cavity trees located less than 40 m from each other were considered spatially dependent, thus in order to avoid pseudoreplication in those circumstances, nests were prioritized over any other cavity category due to their biological significance. Likewise, starts were prioritized over completed roosting cavities because the former better represented actual tree conditions selected by woodpeckers for cavity excavation. One tree was chosen at random if both trees in a pair were in the same category.

3

UN

CO

Adjacent trees (closest unused neighbours) were used as ‘controls’ for determining tree selection by the woodpeckers. This ‘neighbouring’ sampling scheme limited the possibility that territorial behaviour of conspecifics would prevent the use of these trees by magellanic woodpeckers. Individual control trees were chosen by establishing circular plots with the cavity tree at the centre. In each plot, we defined four quadrants according to the main cardinal directions. A random cardinal orientation of magellanic woodpecker cavities (Ojeda 2006) enabled the convenient use of fixed points to place the sampling quadrants. Control trees were restricted to living and recently dead trees equal to 30 cm in diameter at breast height or larger than 29 cm in diameter at breast height. These conditions represented thresholds for magellanic woodpecker cavity trees in N. pumilio forests (Ojeda 2006). In each quadrant, the nearest tree © 2007 Ecological Society of Australia

Data analyses Because our sampling design may entail similarities among central and control trees due to site topography conditions, the comparisons between cavity and control trees were performed on a paired-samples scheme. Univariate tests involved comparing cavity trees against the mean value of surrounding control trees. Due to non-normality of most data sets, we used the Wilcoxon matched pairs test. Distributions of tree variables between cavity and control groups were performed using the two-sample KolmogorovSmirnov test. We also developed multivariate models that best predicted the occurrence of cavity and control trees based on internal and external tree predictors, using a Conditional Logistic Regression (PROC PHREG Procedure; SAS Institute 1996; hereafter CLR). Since we

4

V. S . O J E DA ET AL.

Table 1. Characteristics of magellanic woodpecker cavity trees compared to those of control trees in Nothofagus pumilio forests from northern Patagonia, Argentina Mean ⫾ SE (n) Cavity trees 0.60 ⫾ 0.02 18.45 ⫾ 0.49 196.17 ⫾ 6.72 17.38 ⫾ 1.00 1.91 ⫾ 0.19

d.b.h. (m) Height (m) Age (years) Recent growth (mm) Crown dieback†

Control trees 0.60 ⫾ 0.01 17.57 ⫾ 0.23 175.58 ⫾ 5.55 23.70 ⫾ 0.89 1.46 ⫾ 0.09

(68) (69) (35) (62) (69)

(282) (281) (115) (214) (282)

RO OF

Characteristics

P 0.7101 0.0134 0.1382 0.0003 0.0257

P-values based on Wilcoxon matched pair test. †Semi-quantitative predictor. Complete name: degree of crown dieback. d.b.h, diameter at breast height.

DP

hand, cavity and control trees were similar in d.b.h. when comparing both mean d.b.h and distributions (Table 1, Fig. 1a). Also, mean age of cavity trees did not differ significantly from the neighbouring control trees (Table 1). However, there was a significant difference in the distribution of age classes where cavity trees showed a significant tendency to be in older (>160 years) age classes than random expectation (P < 0.025, K-S-test, Fig. 1b). Cavity and control trees differed in the degree of crown dieback, height, and recent growth. Cavity trees had 31% more crown dieback than surrounding controls (Table 1) and tended to be concentrated in higher classes of crown dieback (P < 0.05, K-S-test, Fig. 1c). In general, trees with a higher degree of crown dieback also showed suppression in growth (Suarez et al. 2004). In accordance with this, the degree of crown dieback was highly correlated with recent growth (Spearman Rank Correlation –0.36, P < 0.05). Recent growth (cumulative last 19 years) was the variable that differed most between cavity and control trees (Table 1). Cavity trees had grown, on average, 36% less than surrounding control trees. Slow-growing trees (<20 mm or c. <1 mm year-1) represented 70% of cavity trees but only 42% of control trees (Fig. 1d). Many cavity trees were characterized by continuous narrow rings during the last 19 years. We found no correlation between age and recent growth in cavity trees (r = 0.06). In contrast, control trees showed the expected negative relationship between age and growth (r = -0.37, P < 0.05). To avoid confounding effects of age on growth, we performed a univariate test using a subsample (n = 32) with similar mean age. Again, cavity trees were more suppressed in growth than control trees as shown by the highly significant difference in width between the two groups (P = 0.0003 Wilcoxon’s test). The last variable that showed differences between cavity and control trees was height. Cavity trees were significantly taller than surrounding control trees (Table 1). Similar results were found with CLR. Results of Model I of the CLR (Table 2, top) indicated that

UN

RESULTS

CO

RR EC

TE

had four controls for every cavity tree, we used ‘1-M matched’ case of CLR. The model fitted was logit (p) = a |xß where p is the probability a tree in the territory contained a cavity, based on tree characteristics. We followed the guidelines of Hosmer and Lemeshow (1989), and tested for correlations among all independent variables prior to inclusion in the CLR analysis (Spearman Rank Correlation). We excluded age in the models to prevent a loss of information due to a relative small sample size of this variable, compared to that of other variables. In order to analyse more closely past radial growth patterns among cavity and control trees, we developed tree-ring chronologies for each of these two groups using standard dendrochronological procedures (Stokes & Smiley 1968; Fritts 1976). Because we had four times more control-tree samples than cavity-tree samples, we only used a 25% random subsample of the control-tree cores. First, individual series were cross dated using COFECHA (Holmes 1983). Second, individual ring growth series were standardized by subtracting the long-term mean of each individual growth value. This rigid standardization retains most longterm (decadal and multidecadal-scale) trends in radial growth (Veblen et al. 1991). Both chronologies were compared using repeated measures anova, with one between-subject treatment (cavity and control tree) and one within-subject factor (time).

A total of 351 cavity and control trees were sampled between March and July 2003. The final tree sample was composed of nests (n = 23), starts (n = 21), roosts (n = 25) and control trees (n = 282). Most (64%) trees were located in unburnt forest, 32% were located near (10–100 m) burnt forest, and three (4%) trees were located in burnt patches. In a first examination of tree variables, we found significant cavity-habitat resource selection based on individual trees within the stand level. On the one

© 2007 Ecological Society of Australia

D I E B AC K A N D C AV I T Y- T R E E S E L E C T I O N

0.35

0.25

a

0.20

0.25 0.15

0.20 0.15

0.10

0.10

0.05 0.05

0.00

0.00

-80 00 20 40 60 80 00 20 40 60 80 81 60 81-1 01-1 21-1 41-1 61-1 81-2 01-2 21-2 41-2 61-2 >2 1 1 1 1 1 2 2 2 2

. 40 . 50 . 60 .70 .80 .90 0.90 > 0-0 .40-0 .50-0 .60-0 .70-0 .80-0 0. 3 0 0 0 0 0

DBH (m)

Age classes (year)

0.35

c

DP

0.5

d

0.30

0.4

0.25

0.3

0.20

TE

Relative frequency

b

RO OF

Relative frequency

0.30

5

0.15

0.2

0.10

0.1

0.0

RR EC

0.05 0.00

0

1

2

3

4

5

Degree of crown dieback

5-10 10-15 15-20 20-25 25-30 30-35 35-40

>40

Recent growth (mm)

CO

Fig. 1. Distribution of tree variables among cavity (solid bars) and control (empty bars) Nothofagus pumilio trees at Challhuaco Valley (CV) forests. Except when clarifying, n = 68, n = 282 for cavity and control trees, respectively. (a) Diameter distribution of magellanic woodpecker cavity and control trees (P > 0.10, Kolmogorov-Smirnov test); (b) Age distribution of magellanic woodpecker cavity and control trees (P < 0.025, Kolmogorov-Smirnov test); (c) Frequency of magellanic woodpecker cavity (n = 69) and control trees in each crown dieback class (P < 0.05, Kolmogorov-Smirnov test), and (d) Frequency distribution of recent growth (width of the last 19 growth rings) of magellanic woodpecker cavity (n = 62) and control trees (n = 214) (P < 0.001, Kolmogorov-Smirnov test). Table 2. Results of Conditional Logistic Regression analyses (two models, I and II) of magellanic woodpecker cavity trees compared with control trees in Nothofagus pumilio forests from northern Patagonia, Argentina Value

Standard error

P

Height (m) -1.5149 0.1350 -0.0598 Height (m) -1.9691 0.3663

0.1617 1.1673 0.1219 0.0195 0.1556 1.0610 0.1001

0.0685 0.1944 0.2957 0.0022 0.0593 0.0635 0.0003

0.0181

UN

Model I† d.b.h. (m) Crown dieback§ Recent growth (mm) Model‡ d.b.h. (m) Crown dieback§

Variable

0.0088

†Significance of Likelihood Ratio, Score and Wald statistics, respect of Model I ⱕ0.0001, 0.0002, 0.0006; ‡Significance of Likelihood Ratio, Score and Wald statistics, respect of Model II = 0.0004, 0.0003, 0.0007; §Semi-quantitative predictor. Complete name: degree of crown dieback. d.b.h, diameter at breast height.

© 2007 Ecological Society of Australia

V. S . O J E DA ET AL.

Table 3. Repeated measures anova Mean square

1 47 49 49 2303

4.98 1.49 1.52 0.17 0.11

F

P

3.35

0.073

13.7 1.55

0.0001 0.009

DISCUSSION

Between-subject treatment are both cavity and control tree, within-subject factor is time. d.f., degrees of freedom.

Our results strongly suggest that magellanic woodpeckers preferentially selected for excavating their cavities trees that internally had suppressed growth for about the last 50 years and externally were taller and showed signs of moderate-high crown dieback. In contrast, age and particularly diameter were poor predictors of suitability of trees for cavity building.

External attributes for cavity-tree selection Among the external attributes discriminating between cavity and control trees in our study, degree of crown dieback had the greatest potential as a management tool. Contradicting former nest characterizations by Goodall et al. (1946), who described the nest tree as a ‘rotten and limbless bole’, selection of trees with moderate crown dieback was supported by results from this and other recent studies (Saavedra 1998; McBride 2000; Rodríguez 2001). Similarly, the scant data available for the ‘ivory-bills’ (Campephilus imperialis and C. principalis, both probably extinct) indicate a preference for living trees as cavity trees (Tanner 1942; Lammertink 1996). Cavity trees were however, not larger in diameter than control trees. This contradicts the pattern observed for most woodpecker species (e.g. Hartwig et al. 2004) as well as for many secondary cavity nesters (e.g. Gibbons & Lindenmayer 2002), which select the largest boles among those available. The ~30 cm minimum threshold found in this study is similar to that of two well known picids of comparable size, the pileated (Dryocopus pileatus) and black (D. martius) woodpeckers (Rolstad et al. 2000; Hartwig et al. 2004), and to the thresholds recorded by both McBride (2000) and Rodríguez (2001) for magellanic woodpeckers. In concordance with other studies in this species, size distribution and mean size of cavity trees basically reflected availability of diameters in the forest (mean ~60 cm d.b.h., this study; ~53 cm d.b.h. in a N. antartica-dominated forest, McBride 2000; ~82 cm d.b.h. in a N. dombeyidominated forest Rodríguez 2001). In our study, distribution of cavity trees peaked at an optimum 40–50 cm, which means woodpeckers selected not the biggest boles, but those just big enough to accommodate a cavity. Unlike several woodpecker species (e.g. the pileated woodpecker, Hartwig et al. 2004), magellanic woodpeckers do not show a selective use of moribund trees, and do not use snags (old dead standing

UN

CO

RR EC

TE

cavity and control trees were best predicted by recent growth (P < 0.005) and then by tree height (P < 0.05). Model fitting was significant as indicated by Wald, Score and Likelihood Ratio statistics (largest P < 0.0007). Neither d.b.h. nor degree of crown dieback had explanatory power. However, when recent growth was forced out of the model, degree of crown dieback became the variable that best predicted if trees belonged to the cavity or non-cavity (control) category (P < 0.0005). Recent growth and degree of crown dieback seemed to be interchangeable as predictors in the model as they were significantly negatively correlated (-0.36, P < 0.05). Twenty-six (out of 107) cavity tree ring-width series and 27 (out of 67) control tree-ring width series passed the cross-dating procedure (mean sensitivity of composite chronologies 0.294 and 0.275, respectively). Correlation between cavity and control tree ring series was 0.64 (n = 106, P < 0.00001) suggesting good chronological control by independent cross-dating. However, temporal patterns in radial growth showed significant differences between cavity and control trees (Time ¥ Treatment, F = 1.55, degrees of freedom (d.f.) 49, P < 0.01, Table 3, Fig. 2). Before 1950, both cavity and control trees showed similar growth patterns with low growth during the 1907–1914, 1922–1924 and 1942–1943 dry periods. During climatically favourable periods before 1950 cavity trees tended to have relatively higher (compared to each series long-term mean) growth rates than control trees. Starting in the early 1950s but especially after 1956–1957, cavity trees showed a synchronous sudden step-like decline (relative growth ranging 0.4– 0.5) in growth that lasted up to the present. In contrast, control trees maintained growth more or less around the long-term mean up until the early 1980s after which some decline was noticeable. Strikingly, statistical differences in growth between cavity and control trees appeared after the strong 1956 drought (rainfall -1.9 SD below mean, Bariloche Airport) and after the strong caterpillar insect defoliation (100% of trees defoliated, Veblen et al. 1996) of 1986 (Fig. 2). In contrast, control trees appeared to recover growth rates rapidly after these events. Recent increased

RO OF

Treatment Error Time Time ¥ Treatment Error

d.f.

growth may be related to growth releases caused by the 1996 crown fire that equally benefited cavity and control trees.

DP

6

© 2007 Ecological Society of Australia

D I E B AC K A N D C AV I T Y- T R E E S E L E C T I O N

7

1.5

1.0

0.5

0.0 1900

1910 1920

1930 1940

RO OF

Tree Ring Index

2.0

1950 1960 1970

Year

*

1980 1990

2000

DP

Fig. 2. Tree-ring chronologies of cavity (dark line) and control trees (light line). Arrows indicate major regional droughts (1942–1943, 1956–1957 and 1998–1999). Asterisk indicates massive caterpillar defoliation in the years 1985–86.

provide the best predictive power for woodpecker habitat management purposes.

TE

trees) for cavity excavation, despite a surplus of snags in the study area (Ojeda 2006). Almost all dead cavity trees in our sample were not dead or moribund at the time the cavities were excavated. The sudden death of trees with recently excavated (1–5 years) cavities imposes a limitation to habitat selection studies because of the difficulty of reconstructing the actual health condition of the trees before the rapid decline, except for studies with field notes available from longterm observations (as in this case). This may create a confounding situation where it is difficult to determine if the rapid death was a consequence of decline in vigour (precursor of cavity excavation) or if it was accelerated by the excavation itself. Tree height was the second most important external feature that discriminated cavity from control trees (cavity trees about 5% taller than control trees). We propose that woodpeckers possibly do not directly select using height as a feature per se, but taller trees possibly have better characteristics for cavity building. This is evident in Fig. 2, where cavity trees grew faster than control trees during the early 1900’s. Foresters have long recognized that taller trees within even-aged forests are generally those which initially grew fastest and escaped competition from surrounding trees (Oliver & Larson 1996). Trees with faster initial growth may have softer inner wood for excavation or may be more prone to suffer heart-rot infections. Also initially fastest-growing emergent Nothofagus dombeyi trees suffered stronger canopy dieback during the 1998–1999 drought than the average main canopy trees (Suarez et al. 2004). Given however, that dieback and height were not interchangeable (nor correlated), it seems that height (or related initial growth) may in itself provide some extra value to trees for cavity selection and therefore joint use of dieback and height may

Internal attributes for cavity-tree selection

UN

CO

RR EC

Although trees over 150 years old were used more than younger ones, age was a poor predictor of suitability of trees for cavity nesting. Apparently, a threshold age of about 150 years is necessary for trees to attain certain characteristics, beyond which, woodpeckers are not age-selective, but respond more strongly to other tree attributes, such as crown dieback and vigour decline. McBride (2000) found a minimum age for N. antarctica cavity trees of about 100 years. Interestingly, N. antarctica is a short-lived species (around 150 years at optimal sites, and rarely more than 200 years), while N. pumilio can live more than 350 years. Possibly, the processes that make a tree suitable for cavity building occur earlier in the life of a shorter-lived species (around its half-life). Contrary to our findings, however, and probably as a consequence of pooling data from two different species (N. pumilio and N. antarctica), or due to very small sample sizes, McBride (2000) found age as the most important factor after trees had attained a threshold size. In our study radial growth integrated over several decades before cavity construction, was by far the best predictor that differentiated cavity trees used by the magellanic woodpecker from control trees. Thus, possibly after trees attain a threshold size, some conditions that cause or are consequences of slow radial growth make trees suitable for selection by magellanic woodpeckers. Contrary to the idea of a slow suppression process, trees that were later selected for cavity construction by magellanic woodpeckers suffered

© 2007 Ecological Society of Australia

V. S . O J E DA ET AL.

RO OF

discrete dieback events per se as mechanisms that create habitat for animals has been rarely recognized. In general it has been assumed that overmature forests/ trees will ‘die-back’ with time and become suitable for cavity nesters. In many forests of the world, including Australian (Gibbons & Lindenmayer 2002) and North American (Davis et al. 1983) forests, trees in the most advanced states of deterioration generally had the highest likelihood of containing hollows of a range of diameters. Therefore, considerable emphasis in managing forests for cavity-nesting birds has been directed at retaining dead and dying trees as cavity sites. Here we propose that, at least for our system, overmaturity is not a necessary condition for trees to become suitable habitat for cavity excavators because exogenous mechanisms triggering dieback (drought, insect attacks, mistletoe infections) can act well before general stand ageing. This has important habitat management implications because it underscores the importance of strongly suppressed partially dead trees within even-aged stands rather than the dominant paradigm of retaining pockets of old-growth forest, dead snags or old trees.

TE

several discrete events of synchronous, sudden reduction in radial growth, while non-selected trees remained stable or very gradually declined in the long term. Interestingly, the main deflection of growth trajectories between these groups of trees occurred after major documented regional droughts, known to have triggered widespread crown dieback in Nothofagus forests of northern Patagonia (Fig. 2, 1943–44, 1956– 57, Veblen et al. 1996; Villalba et al. 2005). A welldocumented example of massive dieback occurred during the summer 1998–99 following one of the most severe droughts of the 20th century (Suarez et al. 2004) during which about 11 000 hectares of Nothofagus forest of Nahuel Huapi National Park suffered some degree of crown dieback. Another type of event that triggered a major growth suppression pattern was the region-wide Ormiscodes sp. caterpillar outbreak that totally defoliated N. pumilio forests in 1985–86 over at least 250 000 hectares in northern Patagonia (Fig. 2; Veblen et al. 1996; Kitzberger et al. 2000). Both drought and insect attack are well recognized triggers for crown dieback (Mannion 1981). These sources of crown dieback are frequent in N. pumilio, as well as other Nothofagus spp. stands (Cerda & Angulo 2000; Tercero Bucardo & Kitzberger 2004). Although the relationship between the suitability of trees as cavity substrates and these agents is not understood for Nothofagus trees, the loss of functional or photosynthetic tissue (crown dieback) and or mistletoe infections may expose trees to fungal attack at high tree levels (Cwielong & Rajchenberg 1995; Tercero Bucardo & Kitzberger 2004). On the other hand, crown dieback and growth suppression may cause lower limbs to be dropped, and hence, additional heartrot to be formed at different tree heights on the bole. Clearly the need of a relatively soft substrate for cavity excavation would favour the selection of decayed trees by woodpeckers. Heart rots are essential precursors for woodpecker cavity excavation because they make wood tissue physically easier to chisel out (Parks 1998). This would explain the poor representation of fully vigorous individuals in our cavity-tree sample. Unfortunately, the presence of heart rot found at the tree base (i.e. cored portion) is not a reliable indicator of heart rot conditions at higher levels (i.e. cavity heights) because columnar rot in the bole is discontinuous and may derive from infections using different routes (Cwielong & Rajchenberg 1995). We suggest that drought/biotic-induced canopy dieback events may be important processes for creating habitat for woodpeckers and the related web of cavity users. Crown dieback induced by natural triggering factors has long been identified as a pervasive pattern in Nothofagus-dominated landscapes and as an important mechanism influencing the dynamics in many temperate forests (Mueller-Dombois 1988; Ogden 1988; Veblen et al. 1996). However, the role of

DP

8

AKNOWLEDGEMENTS

UN

CO

RR EC

Partial funding was provided by CANON USA Inc. through a scholarship (National Parks Science Scholars Program for the Americas) for Valeria Ojeda and CONICET-Argentina, through Doctoral scholarships. We are grateful to Idea Wild (Fort Collins, CO, USA) and Birders’ Exchange (Colorado Springs, CO, USA) for providing equipment fundamental to this study.

REFERENCES Bednarz J. C., Ripper D. & Radley P. M. (2004) Emerging concepts and research directions in the study of cavitynesting birds: keystone ecological processes. Condor 106, 1–4. Bonar R. L. (2000) Availability of Pileated Woodpecker cavities and use by other species. J. Wildl. Manage. 64, 52–9. Carlson A. & Aulén G. (1990) Conservation and Management of Woodpecker Populations, Report no. 17. Swedish University of Agricultural Sciences, Department of Wildlife Ecology, Uppsala. Cerda L. M. & Angulo A. (2000) Insectos asociados a bosques del centro sur de Chile. In: Agentes de daño en el Bosque Nativo (eds A. Badín & L. Pancel) pp. 201–81. Editorial Universitaria, Santiago de Chile. Cofré H. L. (1999) Patrones de rareza de las aves del bosque templado de Chile: implicancias para su conservación. Boletín Chileno de Ornitología 6, 8–16. Conner R. N. & O’Halloran K. A. (1987) Cavity-tree selection by Red-cockaded Woodpeckers as related to growth dynamics of Southern Pines. Wilson Bull. 99, 38–412.

© 2007 Ecological Society of Australia

D I E B AC K A N D C AV I T Y- T R E E S E L E C T I O N

RO OF

Ojeda V. S. (2006) Magellanic Woodpecker (Campephilus magellanicus) nest site selection and reproductive biology in northern Patagonia, Argentina (PhD Thesis). Universidad Nacional del Comahue, Bariloche. Oliver C. D. & Larson B. C. (1996) Forest Stand Dynamics. John Wiley & Sons, New York. Parks C. G. (1998) Wildlife and wood decay interactions. In: Proceedings of the 46thWestern International Forest DiseaseWork Conference (ed. L. Trummer) pp. 106–7. USDA Forest Service, Reno. Paruelo J. M., Beltrán A., Jobbágy E., Sala O. E. & Golluscio R. A. (1998) The climate of Patagonia: general patterns and controls on biotic processes. Ecol. Austral 8, 85–101. Rodríguez S. P. (2001) Análisis preliminar de cavidades arbóreas de Carpintero Negro Campephilus magellanicus en el Parque Nacional Nahuelbuta, IX Región, Chile (Bsc Thesis). Universidad de Concepción, Chillán. Rolstad J., Rolstad E. & Sæteren Ø. (2000) Black Woodpecker nest sites: characteristics, selection, and reproductive success. J. Wildl. Manage. 64, 1053–66. Rozzi R., Martínez D., Willson M. F. & Sabag C. (1996) Avifauna de los Bosques Templados de Sudamérica. In: Ecología de Los Bosques Nativos de Chile (eds J. J. Armesto, C. Villagrán & M. Kalin Arroyo) pp. 135–52. Editorial Universitaria, Santiago de Chile. Rudolph D. C. & Conner R. N. (1991) Cavity-tree selection by Red-cockaded Woodpeckers in relation to tree age. Wilson Bull. 103, 458–67. Saavedra M. (1998) Avances en el proyecto de conservación del Carpintero Gigante en las Áreas Silvestres Protegidas de la IX Región de la Araucanía. In: La Conservación de la Fauna Nativa de Chile pp. 111–19. CONAF, Santiago de Chile. SAS (1996) SAS/STAT User’s Guide, Release 6.03 edition. SAS Institute, Cary, North Carolina, USA. Short L. L. (1970) The habits and relationships of the Magellanic Woodpecker. Wilson Bull. 82, 115–29. Stokes M. A. & Smiley T. L. (1968) An Introduction to Tree-Ring Dating. University of Chicago Press, Chicago. Styring A. & Ickes K. (2001) Woodpecker abundance in a logged (40 years ago) vs unlogged lowland dipterocarp forest in Penynsular Malaysia. J. Trop Ecol. 17, 261–8. Suarez M. L., Ghermandi L. & Kitzberger T. (2004) Factors predisposing episodic drought-induced tree mortality in Nothofagus – site, climatic sensitivity and growth trends. J. Ecol. 92, 954–66. Tanner J. T. (1942) The Ivory-BilledWoodpecker, Research Report 1. National Audubon Society, New York. Tercero Bucardo N. & Kitzberger T. (2004) Establishment and life history caracteristics of the southern South American mistletoe Misodendrum punctulatum (Misodendraceae). Rev. Chil. Hist. Nat. 77, 509–21. Vázquez R. A. & Simonetti J. A. (1999) Life history and sensitivity to landscape change: the case of birds and mammals of Mediterranean Chile. Rev. Chil. Hist. Nat. 72, 517–25. Veblen T. T., Hadley K. S., Reid M. S. & Rebertus A. J. (1991) Methods of detecting past spruce beetle outbreaks in Rocky Mountain subalpine forest. Can. J. For. Res. 21, 242–54. Veblen T. T., Donoso C., Kitzberger T. & Rebertus A. J. (1996) Ecology of Southern Chilean and Argentinean Nothofagus forests. In: The Ecology and Biogeography of Nothofagus Forests (eds T. T. Veblen, R. S. Hill & J. Read) pp. 293–353. Yale University Press, New Heaven. Vergara P. & Schlatter R. P. (2004) Magellanic Woodpecker (Campephilus magellanicus) abundance and foraging in Tierra del Fuego, Chile. J. Ornithologie 145, 343–51.

UN

CO

RR EC

TE

DP

Cwielong P. P. & Rajchenberg M. (1995) Wood-rotting fungi of Nothofagus pumilio in Patagonia, Argentina. Eur. J. For. Pathol. 25, 47–60. Davis J. W., Goodwin G. A. & Ockenfels R. A., technical coordinators (1983) Snag Habitat Management: Proceedings of the Symposium. Gen.Tech. Rep. RM-99. Department of Agriculture, Forest Service, Rocky Mountain Forest, Fort Collins. Duncan R. P. (1989) An evaluation of errors in tree age estimates based on increments cores in kahikatea (Dacrycarpus dacrydioides). NZ Nat. Sci. 16, 31–7. Estades C. F. & Temple S. A. (1999) Deciduous-forest bird communities in a fragmented landscape dominated by exotic pine plantations. Ecol. Appl. 9, 573–85. Fritts H. C. (1976) Tree Rings and Climate. Academic Press, London. Gibbons P. & Lindenmayer D. (2002) Tree Hollows and Wildlife Conservation in Australia. CSIRO Publishing, Melbourne. Glade A. (1993) Libro Rojo de los Vertebrados Terrestres de Chile, 2nd edn. CONAF, Santiago de Chile. Goodall J. D., Johnson A. W. & Philippi R. A. (1946) Las Aves de Chile, Vol. 1 and Supplement. Platt Establecimientos Gráficos S. A., Buenos Aires. Hartwig C. L., Eastman D. S. & Harestad A. S. (2004) Characteristics of pileated woodpecker (Dryocopus pileatus) cavity trees and their patches on southeastern Vancouver Island, British Columbia, Canada. For. Ecol. Manage. 187, 225–34. Holmes R. L. (1983) Computer-assisted quality control in treering dating and measurement. Tree-Ring Bull. 43, 69–75. Hosmer D. W. Jr & Lemeshow S. (1989) Applied Logistic Regression. John Wiley & Sons, New York. Jackson J. A. & Jackson B. J. S. (2004) Ecological relationships between fungi and woodpecker cavity sites. Condor 106, 37–49. Jackson J. A., Lennartz M. R. & Hooper R. G. (1979) Tree age and cavity initiation by Red-cockaded Woodpeckers. J. For. 77, 102–3. Kitzberger T., Veblen T. T. & Villalba R. (2000) Métodos dendroecológicos y sus aplicaciones en estudios de dinámica de bosques templados de Sudamérica. In: Dendrocronología en América Latina. F. A. Roig (comp.); EDIUNC, Mendoza. Lammertink M. (1996) The lost empire of the Imperial Woodpecker. World Birdwatch 18, 8–11. Lammertink M. (2004) A multiple-site comparison of woodpecker communities in Bornean Lowland and Hill forests. Conserv. Biol. 18, 746–57. McBride P. (2000) Magellanic Woodpecker (Campephilus magellanicus) habitat selection in deciduous Nothofagus forests of Tierra del Fuego (MSc Thesis). Western Washington University, Bellingham. Mannion P. D. (1981) Tree Disease Concepts. Prentice Hall, Englewood Cliffs, New Jersey, USA. Martin K. & Eadie J. M. (1999) Nest webs: a community-wide approach to the management and conservation of cavitynesting forest birds. For. Ecol. Manage. 115, 243–57. Mikusinski G. & Angelstam P. (1997) Economic geography, forest distribution, and woodpecker diversity in Central Europe. Conserv. Biol. 12, 200–8. Mueller-Dombois D. (1988) Towards a unifying theory for stand-level dieback. Geojournal 17, 249–51. Ogden J. (1988) Forest dynamics and stand-level dieback in Zealand’s Nothofagus forest. Geojournal 17, 225–30. Ojeda V. S. (2004) Breeding biology and social behaviour of Magellanic Woodpeckers (Campephilus magellanicus) in Argentine Patagonia. Eur. J. Wildl. Res. 50, 18–24.

© 2007 Ecological Society of Australia

9

V. S . O J E DA ET AL.

Welsh C. J. E. & Capen D. E. (1992) Availability of nesting sites as a limit to woodpecker populations. For. Ecol. Manage. 48, 31–41. Willson M. F., de Santo T. L., Sabag C. & Armesto J. J. (1994) Avian communities of fragmented South-Temperate rainforests in Chile. Conserv. Biol. 8, 508–20. WWF. (1999) Análisis de la Biodiversidad y Conservación de la Eco-RegiónValdiviana. International Workshop Final Report. World Wildlife Fund, Valdivia.

UN

CO

RR EC

TE

DP

Villalba R., Masiokas M., Kitzberger T. & Boninsegna J. A. (2005) Biogeographical consequences of recent climate changes in the southern Andes of Argentina. In: Global Change and Mountain Regions: An Overview of Current Knowledge. Series: Advances in Global Change Research (eds U. M. Huber, H. K. M. Bugmann & M. A. Reasoner) Springer-Verlag, New York. Vuilleumier F. (1985) Forest birds of Patagonia: ecological geography, speciation, endemism, and faunal history. Ornitholog Monogr. 36, 255–305.

RO OF

10

© 2007 Ecological Society of Australia

SNP Best-set Typesetter Ltd. Proofreader: Jason Delivery date: 15 March 2007 Copyeditor:

UN

CO

RR EC

TE

DP

RO OF

Journal Code: AEC Article No: 1705 Page Extent: 10