Simulating disturbance in SORTIE

Presented by Mike Papaik.

Basic approach to studying disturbance

• Wait for a disturbance to your favorite ecosystem/community
• Get there quick, select plots and collect damage/mortality data
• Determine the resistance/susceptibility to it as a species-specific function of some measure of disturbance severity - storm severity index vs. direct measure of intensity
• Determine the disturbance regimes from field studies or literature review that allows you to link storm severity with expected return interval
• Factor in interactions of the disturbance with regeneration, e.g., resource availability, seed substrates

Adding to the approach:

• Determine resistance/susceptibility to disturbance as a spatially explicit function of neighborhood. e.g., using mapped permanent plot stands in UP MI, Kerry Woods and Charlie are analyzing cascading patterns of the effects of a windstorm through a stand.
• Factor in interactions between disturbances. e.g., measuring health status of trees allows us to estimate the resulting change in susceptibility to wind disturbance as a function of tree health.

Co-assessing wind damage and beech bark disease

The storm - New York (July 15th, 1995). 60,000 ha were affected; ~15,000 ha had > 60% mortality

The plots

New York, 1995:

• 43, 1/8th ha plots were assessed for wind damage (2,599 trees)
• 18 of these were also assessed for BBD (1,116 trees)

Michigan, 2002: 83, 1/5th acre USFS permanent plots were assessed (3,894 trees).

The sampling strategy:

• Well mixed stands (size and species distribution)
• Uniform within plot storm damage (no obvious patchiness)
• Range of plot damage greater than none and less than total

The data:

• Species
• DBH
• Damage
  • Tip over > 45 degrees from zenith - dead
  • Stem snap below crown - dead
  • Crown damage
• BBD (Adapted from Burns and Houston 1987)
  • N - None
  • L - Scale / discrete bark lesions < = 3cm
  • M - Necrotic tissue, blocky bark < 50% circ.
  • H - Necrotic tissue > 50% circ.

Simulations

The probability of tree damage was modeled as a species-specific logistic function of dbh and storm severity index. The more shade-tolerant species are more resistant to windthrow - except for American Beech.

Modeled three storm regimes - one from the literature, one with more frequent less intense storms, and one with less frequent more intense storms, all with same average mortality long-term. The effect of windstorm as disturbance increases succession, but also maintains greater diversity - shade-intolerants stick around longer. Dominance of beech is replaced by dominance of yellow birch.

Then we re-modeled windthrow with each level of disease modeled as separate species - which model had best support in the data?

Conclusions

BBD dramatically increases windthrow susceptibility of medium sized beech (20-40cm DBH). As a result, the dominance of beech in these forests is reduced by BBD; however, Beech is expected to remain as an important component of the landscape as a sub-canopy and small canopy tree because the windthrow resistance of smaller adult beech remains high.

High levels of disease has a distinct effect from low or medium levels. There is a strong effect of beech bark disease on the probability of stem snap by wind. The probability of crown damage goes down as a tree is more diseased (because infected trees have thinner crowns?)

Discussion of this talk

Question: Are there things we can never measure effectively? Evidence that species composition is governed by disturbance. In Canada we are trying to document natural disturbance regime - is it possible to quantify natural disturbance? Is natural disturbance and silvicultural disturbance additive in effect? It's worth trying to separate out community response to disturbance from disturbance regime.(?) It's 'simpleminded' to think you can emulate nature and not change forest structure - scientists need to make it clear this isn't possible.

Goodness of fit of mortality model is extremely good. Stem snap fit is weaker - less data. You could have an effect that statistically seems weak - but has a huge effect biologically. When this was modeled, only mortality was used to reduce that.

Question: In Canada - there are fewer large trees and more medium ones in managed forest - so susceptibility to windthrow is lessened - and thus shade-intolerants are favored more? We haven't modeled interactions of thinning and windthrow. We need a storm to hit a thinning study plot to be able to get the data to do this correctly from a management context.

There is a notion that disturbance was necessary for all sorts of positive effects - but the consensus has shifted for wind because it just advances succession and doesn't actually change composition - so its effect is small. It's really the more dramatic disturbance (fire, earthquake) that have the significant, observable effects. Which raises the question of how shade tolerants persist in a system, especially in NE before human presence when fire was rare and wind was the primary disturbance regime.

For mid-tolerant species - some level of disturbance, including lower levels like wind - helps them persist. So when you log big trees and shift the age structure lower, you're impeding the ability of yellow birch to regenerate through gap-phase dynamics. So you could maybe address the question of interaction between silviculture and independent tree fall disturbance with SORTIE. Model is moderately sensitive to substrate changes - we would need to know more about substrate dynamics after windthrow.

There has been sampling in forests with incredibly complicated disturbance regimes - windstorms, ice storms, etc. - we should not underestimate the level and variety of disturbance that forests undergo.

Disturbance and succession in New Zealand forests

Presented by David Coomes.

New Zealand succession

The theories of succession are quite different from the northern hemisphere perspective.

• McKelvey hypothesis: There is still long-term succession going on after volcanic disturbance 1800 years ago - 'twas objected that it was probably environmental gradients instead.
• Ogden's temporal replacement model: If you have a huge disturbance, conifers completely dominate, live ~600 years and begin to decline. When they create gaps, angiosperms colonize and begin to take over, and you need periodic disturbance to restore conifer populations. There's some support from earthquake studies.
• Holloway's beech invasion hypothesis: Beech is a very slow-spreading species - but is the most competitive in the system and is still slowly taking over.

Almost all angiosperms have inverted J-curve for their size distribution - and most are under 100 years old. Conifers have a cohort from 600 years ago in their age distribution (cohort or period of increased recruitment? Can't tell yet). Angiosperms are short-lived and fast growing; conifers are longer-lived and slower-growing.

Competition probably does not affect the survival of large trees. Ferns prevent dense stands, and there is no negative trade-off between angiosperm basal area and conifer basal area (the additive basal area phenomenon).

We don't know the stabilizing effects that allow conifers to persist as strongly as they do - we fear SORTIE will kill them all promptly.

Discussion of this talk

Question: There must be some interaction between adults or we would see more diversity in stand composition, yes? Maybe it's the low-fertility effect - lots of space in the canopy

There seems to be excessive amounts of basal area for some sites. That's due to small plot size - sampling effects. Is it due to non-limiting amounts of water? Could be.

Question: Is McKelvey's theory that angiosperms were less damaged by volcano? No, everything was devastated.

When did beech arrive (beech invasion hypothesis)? Has been present since Gondwana split. But the reason it's re-invading is due to being wiped out by glaciation.

Is the lack of competition between conifers and angiosperms due to angiosperms not taking all resources, and leaving some for the conifers? There IS a strong competitive effect - especially among juveniles - but the effects are smaller for older adults.

Fire in boreal forests

Presented by Yves Claveau.

Fire types:

• Crown fires - most trees die
• Ground fires. Much less damaging to big trees. The organic layer is burned and shrubs and small trees are killed.

Scale of damage differences:

• Microsite - you can see difference in burning intensity in organic layer - seeds and rhizomes can survive and be source of regeneration
• Stand - Standing trees remain - shade new germinates
• Landscape

Discussion of this talk

Someone said that the role of fire in maintaining boreal forest has been overstated - opinions? Fire is a 'sexy' disturbance - but there are areas where return interval is 500 years and don't burn so often.

We've compartmentalized different disturbance types - but it's clear in California that some types of disturbance cause susceptibility changes to other disturbance (disease makes trees more susceptible to fire).

How are people interested in modeling disturbance in the model?

• Messier - very interested in insect disturbance - effects of stand composition on disease outbreaks
• Messier - Ice storms
• Droughts in tropical systems

The challenge is to get the biology right - it is very difficult to write a generic disturbance module without adequate description. One generic behavior possibility - something that is introduced and gradually kills all trees in an area (as with disease or pests).

There are two kinds of disturbance - non-spatially explicit, periodically returning, like wind, and spatially-explicit gradual, like pests.

What about modeling climate change? Nobody's mentioned it. We could model the range of predictions coming out of the global climate models. But it's a fundamentally different type of disturbance and might require a different programming regime.