Ecological Archives E093-022-A1

Victor Johansson, Thomas Ranius, and Tord Snäll. 2012. Epiphyte metapopulation dynamics are explained by species traits, connectivity, and patch dynamics. Ecology 93:235–241.

Appendix A. Patch characteristics, tree age model, and the lichen survey.

All oaks with a diameter at breast height (dbh henceforth) exceeding 30 cm were mapped within the study area using GPS. Trees <30 cm are rarely older than 80 years, and therefore not yet suitable for our study species (Table A1). However, we also included a number of smaller, possibly slow growing oaks with deep bark fissures (56 trees). For each tree (in total 2083), we measured tree and environmental variables that were hypothesized to affect the lichen metapopulation dynamics or tree age. We also measured the age of 122 trees by counting the rings of increment cores taken 0.7–1 m above the ground (as in Ranius et al. 2008). The real age of the trees is therefore somewhat underestimated, probably <10 years. The cored trees were selected by stratified random sampling to represent the variation in dbh and ground conditions observed in the study area (Fig. A2).

We estimated the natural logarithm of the age of the uncored trees using a generalized linear model with an identity link function (i.e., multiple linear regression). All variables were normalized to avoid correlations with the intercept parameter. We built the model based on the Deviance Information Criterion (DIC) and on knowledge of the biological system studied (Burnham and Anderson 2002): we first assessed the predictive power of each explanatory variable (Table A2) based on DIC and on the posterior distribution of the parameters. Next, we fitted a multiple model containing the retained explanatory variables. The final model,

ln(agei) = β0 + β1dbhi + β2barki + β3barki2 + β4groundi,

was used to predict the ages of uncored trees (for parameter estimates, see Appendix C: Table C2).

For a few very large trees (0.96%) and trees with very deep bark fissures (0.53%), we did not use the tree age model, because the predicted ages were unreasonable. We tried to core these, but all were hollow making them impossible to date. Instead, we predicted the ages for these trees from uniform distributions with reasonable lower and upper limits, 250 to 500 years based on the ages of the oldest trees in the region. For the cored hollow trees, the oldest age ring was used as the lower limit of the uniform distribution. For seven trees we modeled bark fissure depth as a function of dbh, because the bark at breast height was missing (not shown in Model code, Supplement 1).

We studied five oak-associated crustose lichen species, with varying propagule size and niche breadth (Table A1). The study species were chosen among the crustose lichens associated with old oaks. Hence, common species also occurring on other tree species were excluded. However, there are a few additional species associated with old oaks in the region. All of them except one (Calicium adspersum) are very rare and were therefore not included in the survey. However, we also surveyed the spore dispersed Calicium adspersum. The reason for excluding it was that its pattern of occurrence differed markedly from a pattern that may result from typical metapopulation dynamics. It was completely missing from 25% of the area, which suggests that an unknown environmental factor explains its pattern. However, fitting the model for this species would support the conceptual model (Fig. 1). It would have a high mean colonization rate (1011 occupied trees), it has small propagules (Smith et al. 2009), and it has a wide niche (a patch threshold of approximately 100 years).

For all trees we recorded the occurrence of each of our study species on the lowest 2 m of the trunk. We have found that the risk for missing occurrences resulting from searching on the lowest 2 m is negligible in crustose lichens that are confined to old oaks (Johansson et al. 2010). Other old deciduous trees are rare in the area, and we also searched for the study species on these: on two old Swedish whitebeam (Sorbus intermedia) trees we found C. phaeocephala. Moreover, we mapped the occurrence of the study species on all trees (including other deciduous trees) within a buffer zone of 400 m around the study area. The size of the buffer zone was based on earlier studies of epiphyte colonization ranges and spatial aggregation patterns (e.g. Snäll et al. 2005, Löbel et al. 2006), which have suggested that dispersal sources more than 400 m away from the focal tree have limited explanatory power. In the buffer zone, 103 of the trees (only oaks) were occupied by any of the species. In the analysis, these and the two Swedish whitebeams (Sorbus intermedia) in the area were only used as dispersal sources when estimating connectivity. By including the potential dispersal sources in the buffer zone we avoid to underestimate connectivity of trees close to the border between the study area and the buffer zone. This bias would be greatest in the northeastern part of the study area with the highest tree density in the buffer zone (Fig. A2).


TABLE A1. Characteristics of the five study lichen species.

Species

Soredia size

Spore size

Conidia size

Lower niche breadth limit

Red-list category

Chaenotheca phaeocephala
(Turner) Th. Fr.

6–7

<100 yrs

LC

Cliostomum corrugatum
(Ach.:Fr.) Fr.

10–13 × 2–2.5

2–4 × 1

100–200 yrs

NT

Lecanographa amylacea
(Ehrh. ex Pers.) Egea & Torrente

17–25 × 3–3.5

5–7 × 1–1.2

200–300 yrs

VU

Buellia violaceofusca
G.Thor & Muhr.

15–20

100–200 yrs§

NT

Schismatomma decolorans#
(Turner & Borrer ex Sm.) Clauzade & Vězda

DL

30-37 × 4–5

6–7 × 2

200-300 yrs

NT

Notes: Propagule size in μm from Smith et al. (2009), lower niche breadth limit from Ranius et al. (2008) and red-list category from Gärdenfors (2010); "–" = not applicable; DL = data lacking.

† Based on data from similar oak-rich areas, situated >10 km from the study area, in the same biogeographical region.

‡ LC = least concern; NT = near threatened; VU = vulnerable.

§ V. Johansson et al. (unpublished data).

# Rarely forms spore-producing apothecia and conidia-producing pycnidia (Smith et al. 2009).

 

TABLE A2. Tree and environmental characteristics measured for all 2083 trees at which the study species were surveyed (excluding trees in the buffer zone).

Characteristic

Unit

Description

Mean (min-max)

Age

Years

Estimated from increment cores of 122 trees Predicted ages from the tree age model

168.6 (38.0–372.0)

155.0 (37.4–456.9)†

Bark fissure depth (bark)

mm

Mean depth of four fissures measured at the cardinal points

14.8 (2.75–63.0)

Diameter (dbh)

cm

Trunk diameter 1.3 m above the ground

55.1 (21.3–170)

Openness

 

The vertical projection of foliage around the tree: <25% (0), 25%–75% (1) or >75% (2)

0.73 (0–2)

Stem inclination

°

Inclination of the center of the trunk 0 to 2 m from the ground

3.08 (0–50.9)

Tree condition

%

Proportion dead branches

18.2 (0–100)‡

Ground properties

%

Proportion wooded pasture (compared to agricultural field and lays) within a buffer (radius 15 m) surrounding each tree obtained using GIS. Land classification was based on digital maps, which were corrected by a subjective assessment in the field.

80.0 (0–100)‡

Bare rock

 

Presence of bare rock within 5 m from the tree. Indicates a thin soil-layer.

0.51 (0–1)

† Mean, min and max of the modes from the age distributions of each tree

‡ Variables constituting proportions were arcsine square root transformed in the analysis to improve normality.

 

Figure A1
 
   FIG. A1. Tree age distribution of potentially suitable trees (n = 1579) in the study area as estimated from increment cores and predicted using the tree age sub-model. The remaining surveyed oak trees (n = 504) were younger than 100 years, and thus not suitable for our study species (Fig. 2a).

 

Figure A2
 
   FIG.A2. The study site with the 2083 study oaks (black dots) on a mix of clay-rich agricultural fields or lays (white), or wooded pastures or woods on moraine (gray). Black crosses represent oaks in a 400-m buffer zone around the study area that were occupied by any of our study species.

 

LITERATURE CITED

Burnham, K. P., and D. Anderson. 2002. Model selection and multi-model inference, Second edition. Springer, New York, New York, USA.

Gärdenfors, U. 2010. The 2010 Red List of Swedish species. Artdatabanken, Uppsala, Sweden.

Johansson, V., T. Snäll, P. Johansson, and T. Ranius. 2010. Detection probability and abundance estimation of epiphytic lichens based on height-limited surveys. Journal of Vegetation Science 21:332–341.

Löbel, S., T. Snäll, and H. Rydin. 2006. Metapopulation processes in epiphytes inferred from patterns of regional distribution and local abundance in fragmented forest landscapes. Journal of Ecology 94:856–868.

Ranius, T., P. Johansson, N. Berg, and M. Niklasson. 2008. The influence of tree age and microhabitat quality on the occurrence of crustose lichens associated with old oaks. Journal of Vegetation Science 19:653–662.

Smith, C. W., A. Aptroot, B. J. Coppins, A. Fletcher, O. L. Gilbert, P. W. James, and P. A. Wolseley. 2009. The Lichens of Great Britain and Ireland. British Lichen Society, London, UK.

Snäll, T., J. Ehrlén, and H. Rydin. 2005. Colonization-extinction dynamics of an epiphytemetapopulation in a dynamic landscape. Ecology 86:106–115.


[Back to E093-022]