For a non-mathematical introduction to Gecko Woods, please see the Discussion instead. For how the woods grow, please see the model Details. Exploration gives play suggestions. The Help reader explains how to operate the gizmos on the screen. Further readings are listed in References.
World size (m) : Enter a value in meters. The world is square, with each edge this length. Some Scenarios allow you to enter the size in x and y separately for oblong worlds.
Trunk aversion factor : Trees kill saplings that fall "on" their trunks. The definition
of "on" is tuned via this parameter. For example, a factor of 10.0 means that a larger tree kills off
saplings within 10 x trunk radius: a 10cm Maple would prevent any saplings growing
within a 50 cm (45 cm from its outside) radius. Setting this value lower (even to 0.0)
dramatically increases the number of saplings and run times, with minor change to any
other result.
Site quality : See model details for discussion. Basically, trees can reach their
maximum height/girth/etc. on a site of quality 1.0 or higher. Lower qualities stunt the trees. Qualities greater than 1.0 may make the trees reach maximum size faster. Or not.
Crop crowns : Allows turning on/off the basic model's crown-cropping behavior.
Species random death : If off, makes the model ignore the Species random mortality parameter for all species. More convenient than setting the parameter to zero on each species, and frequently desirable when playing with external death sources (Reapers).
Some Scenarios provide for fragmenting the world. Worlds can be fragmented either randomly, or in a regular array of sites. In conjunction with fragmented worlds, you may be interested in the Fragments Statistic, available on RunControl | AllTrees.
Use random (aot array) fragments : If checked, strew random fragments. If unchecked, use a regular array of same-size fragments.
Options which begin "Random:" or "Arrayed:" only apply to that
fragment style.
Arrayed: sites in x : Using a grid of fragments, how many across.
Arrayed: sites in y : Using a grid of fragments, how many bottom to top.
Arrayed: spacing : If you want all the grids to abut, enter 0. Otherwise, how much space to attempt to leave between two adjacent fragments in both x and y directions. Note that if an impossible request is entered, it will be ignored, and the default of 0 (abutting fragments) used instead.
Site quality for arrayed fragments goes from min to max, in even
increments, from the lower left to the upper right corner of the array.
Random: fraction area covered : Attempt to cover this percentage
of the world with non-overlapping random fragments. Depending on the
percentage and the min/max edges specified, the non-overlapping part might not be satisfied. The fraction should usually be a number between 0.0 and
0.75. Requests for area covered near 1.0 always result in some holes and some overlapping fragments. Fractions greater than 1.0 are allowed. In this case, the fragments always overlap, but there still may be holes in the world. Note that you can use positive or negative quality random fragments. So, for instance, to achieve percent area covered of 75 percent, it may make more sense to use 0.25 fraction area covered and fragment quality of -1.0 on a background of 1.0. I.e., remove 25 percent in holes rather than covering 75 percent in habitat fragments. However, note that the Fragments statistic goes by the holes in that case.
Site quality is randomly selected between min and max for each random fragment.
Random: min fragment edge (m) and Random: max fragment edge (m) : These parameters govern the chunkiness range of rectangles strewn as random fragments. Similar values result in fairly square fragments of the given size. Significantly different values result in a mix of small, large, and oblong (strip-like) fragments. Neither value should exceed half the shortest world dimension, if you want a nice pattern. To get more fragments, use smaller edges. If you really want to analyze an effect, you should consider using arrayed fragments. With seedling disperal off, it's like running N plot experiments in parallel.
The same input values for the parameters specifying the random fragments, always generate the same fragment pattern. You can use slight changes in the input parameters to generate a different pattern. For instance, giving a minimum edge of 10.01 instead of 10.0 meters generates different fragments. The random initialization of Trees is not affected. (Modellers N.B.: an independent random stream is used for the random fragments, with a seed created from the inputs.)
Background quality : The background is that rectangle
described by the world size x/y parameters. The fragments are pieces being strewn on top of that background. It's possible to have a positive background
(quality 1.0), and negative fragments (quality -1) to make hole fragments, rather than habitable fragments. But normally, background quality is 0.0, and site quality is > 0.5, resulting in habitable fragments instead of holes.
Max site quality : The "site quality" assigned to the lower left grid for arrayed fragments. Max of random range for random fragments.
Min site quality : The "site quality" assigned to the upper right grid for arrayed fragments. The grids
between the extremes are evenly stepped in quality. For random fragments, the min of the random range of quality randomly assigned to fragments.
With all fragments, the world is still wrapped (or not) for seed dispersal at the edges. E.g., Trees at the left can introduce saplings at the right. The world isn't wrapped for site quality purposes. E.g., a Tree at the left edge gets no benefit from a site fragment at the right edge.
Depending on the Reaper, it may be difficult to see its effect. Rather than cluttering up the screen with obscure glyphs, many Reapers report their activities to the log. From a web browser, you can see these reports on the "Java console". (In Netscape v 4.5 this is accessible via Communicator | Tools | Java Console. Alas, the option moves even between Netscape versions, so you'll have to consult browser documentation if you can't find it.)
Reapers may either kill, or damage, Trees. Killing obviously affects the whole Tree. Damage generally affects the crown, either in radius or (in the deer case) from the bottom. Some reapers (like fire) may kill some Trees and damage others. Reapers ignore site fragments. Where needed, they take their bearings from the world size boundary.
If desired, the Parameters convenience switch "Species random death" can be disabled to allow all your random deaths to be controlled by external Reapers.
The Reapers and their options are:
Bonus all-tree random death : Independently from other death sources, BonusDeath uses the input percentage as a chance of death on every tree, every timestep (year). Example: say Maple has a Species mortRandom of 0.01 (one percent) and here you input 0.03. Maple's cumulative random death chance is four percent, assuming no other Reapers. In general, each Reaper adds more chances to die.
Old age : OldAge has two different modes. Entering a deathrate, means that this independent deathrate applies to each Tree, each year, after it is "full grown", whatever the Tree itself reports that to be. (Stunted and much-cropped Trees can be "full grown".) The same deathrate applies to all Species. Which can be a problem, since Hemlock may live centuries longer than Maple and Birch at "full grown". The alternative is to turn on the switch labeled "*or* Old age use species mort." In this case, OldAge interprets the Species mortRandom parameter as a maximum age to be enforced. Example: say Hemlock's mortRandom is 0.0025. The reciprocal of that is 400 years. OldAge will force every Hemlock to die by age 400 at the latest, using a random chance which rises based on age of Hemlock until it reaches death probability 100 percent at age 400. It would be nicer if we had a maxAge parameter for the Species, but I don't have that data, and hate to add new parameters just for use by the Reapers.
Hurricane : Hurricane is bigger than whatever you're modeling, so damages the crown of every Tree. Hurricanes only happen some years (frequency 0=none to 1.0=every year). Every random year a Hurricane happens, a random number between 0 and "max severity" is selected for this year's Hurricane severity. That severity is multiplied by another random throw for each Tree to determine the damage to inflict on that Tree. Thus, if "max severity" is 0.5, the year's severity might be only 0.25, and multiplied by another number always less than 1.0, a particular Tree might lose ten percent of its branches. Aside from its relatively global nature, there's nothing especially "hurricane-ish" about hurricane Reapers. It's more like "annual cumulative storm damage, all sources, nobody immune."
Clearcut : Clearcut is a rectangular swath of dead Trees, presumably human in origin. The frequency governs how often Clearcut happens (0.1 means a randomly chosen 10 percent of years). Clearcuts have size in x and y, both chosen randomly, evenly distributed between the min and max edge length. The Clearcut is then placed at random in the world. Clearcut doesn't damage Trees. Any Tree whose trunk is in the Clearcut is killed. Clearcuts are always free to regrow, site quality unaffected. A permanent clearcut would require site fragments (holes).
Roundfire : Roundfire differs from Clearcut in that it is a) round, and b) kills at the center with damage rate falling off to the edges. The Roundfire frequency governs how many years randomly get a (single) Roundfire. The radius of the damage on fire years is randomly selected between 0 and maxRadius meters. Roundfire's center is positioned at random each time. The damage to any particular tree is found by
normalizing its distance to the fire center by the fire radius (call it dist, the percent fire radius this year), and applying damage = 1.2 * dist2. Damage at or above a value of 1.0 kills.
Wind : Non-obviously, Wind is very different from the Hurricane Reaper. The idea in Wind is that, much as Trees compete, they are best suited to life surrounded by other Trees, shielding them from exposure. So Wind is "exposure-related damage" of whatever kind. A Tree is "exposed" if more than 50 percent of its immediate neighborhood is open (not covered by Tree crowns.) The "immediate neighborhood" is defined by the "Wind radius factor", which defaults to 2x this Tree's crown radius. (Wind uses hazy math for the "open" calculation, as the feature is horribly slow otherwise. The hazy math is similar to that used in shadow intersections, as discussed in Details.) Larger neighborhood radii slow down the simulation.
Wind is a constant feature year to year. The random damage ranges from min to max each time an exposed Tree is visited. That random damage amount is multiplied by the Tree's exposure. An additional wrinkle is to optionally multiply the damage by the Tree's crown openness (light coefficient), which is always less than one, and thus scales down the min and max damage entered. The idea there is that a dense crown like that of Hemlock is sort of self-shielding, whereas the airy structure of Maple and Birch (particularly in winter) leave them more susceptible to Wind damage.
Note: Trees are stunted at the edges of sites in Gecko Woods. This is due to straddling the edge of the site, thus experiencing a lower average site quality than their inner neighbors. The only "exposure" involved there is exposure to Tree-negative turf, such as tarmac, granite, or marsh. Reaper Wind does add insult to edges of sites, but only based on exposure. Regrowth in a Clearcut tends to suffer more.
Deer : Deer eat the bottoms of Tree crowns, to whatever height they can reach, specified by "Deer reach (meters)". Deer thus do most damage to low-crowning Trees, such as Hemlock. (I don't know if deer actually like Hemlock, but that's how the Reaper works.) There are good and bad Deer years for Trees, depending primarily on how many deer and how hungry they are. Basically, each year, a Deer damage rate between min and max is selected for the world. There's no variation in this rate between Trees--Deer eat that percent of every crown that they can reach. The crown bottoms are then raised by the amount of damage taken, and the Trees' light-collecting capacity diminished somewhat.
Although deer in nature deeply favor edges of woods, it's unclear to me that stands of the size used online are big enough to impede hungry deer in winter. In a hectare, I'd expect that deer are either there or not there. So, distance from the site edges is ignored. However, Deer is the only Reaper that respects the 5 semi-existence during which seeds turn into saplings. (I.e., it's assumed the saplings are those that survived the deer.)
Please note that there's no particular scientific basis for these Reapers. The idea here
was a varied toolbox of disturbances, each making some sense and inflicting different patterns of damage. These Reapers were the easiest disturbances to
add to Gecko Woods, yet provide a nice variety (old, young, etc.) of damage patterns.
Ideally, the author feels there should be a variety of birth models as well as death models, but that hasn't been done yet.
More plots are available on the RunControl menu, by clicking on the respective Species or CompositeSpecies (like AllTrees) name. Except for Population and Basal Area, the statistics apply to a single Species or Composite. The plots are:
Population : as seen, one trace per Species and CompositeSpecies, headcount vs. time.
Biomass : For Gecko Woods, biomass has been reimplemented as "Basal Area". This is trunk cross-sectional area around human chest height. Its units are square centimeters.
Stage-based statistics : Fertility, Mortality, and Stages are "stage-based", in that each individual supplies a stage (in Woods' case, height small/medium/tall) of their lifecycle, and the statistics are given per-stage. The labels in the Composite case (height less than certain number, etc.) are correct for one of the Species, not all. What's actually composited are the stage answers as chosen for each Species.
Fertility : Per capita fertility per annum per stage, averaged over stat cycle. Since only established saplings are born here, this is recruits that will survive for 5 years.
Mortality : Per capita death per annum per stage, averaged over stat cycle.
Stages : Population per stage, averaged over stat cycle.
Fragments : Fragments are only available on CompositeSpecies (like AllTrees.) This plots number of fragments versus number of Species belonging to the Composite which have inhabited the fragment over the last stat cycle. If site fragments are available, it uses those. (Note that is a poor choice if your fragments are holes rather than habitat.) If there are no fragments, it uses a grid over the world as the "fragments" and is a way to look at mixing.
Biomass (Basal Area) Analyses : The following four statistic types aren't particularly suitable for Gecko Woods analysis, but do something. Biomass, again, is redefined as basal area, since one cannot readily measure trees by biomass.
BiomassTransfers : Absolute basal area growth, theft (none), starve (death), netGrowth (growth - losses), losses (identical to starve).
BiomassRates : Same as BiomassTransfers, except as rates (change over standing biomass).
BiomassGrowth : A "snanglegram", plotting not vs. time, but biomass growth vs. biomass total.
BiomassTheft : Not meaningful in Gecko Woods, since Trees don't eat each other.
If this would be helpful to you, please contact Ginger Booth with a note explaining what you need automated. It's easy to help.
Site Setup Options
Whatever site setup options are available on a Scenario, can be reached at the top of the Parameters menu. As with most Parameters, the simulation must restart for changes to take effect. In this case, you don't get a choice, since OK-ing the popup menu restarts the simulator. The most basic site setup options are:Reapers
For your enjoyment, a collection of external damage inflictors is available on the Parameters menu, button Reapers. Unlike most parameters, the Reapers are effective immediately, without restarting the simulator. However, if the simulator is restarted (using the Restart button, but not by resetting the Scenario), all Reapers continue in effect.Statistics
In most Woods Scenarios, two statistics plots come up automatically, the Population and Basal Area graphs. You can get rid of them, by clicking on the window-close icon (generally in the upper right corner.) They are re-created by selecting a new Scenario. Most plots, once dismissed, can only come back after a new Scenario is selected. All the plots can be zoomed in and out (draw a box up or down respectively) and restored to full size (click on the plot without drawing a box.) The plots can also be grown to full-window size or iconified, relocated and reshaped. They are a lot more tolerant of resizing than the main window, which is under more browser control, thus may not repaint correctly after resizing.Scenario Customization
Clearly, if you're doing a series of experiments with Gecko Woods, using
the popup menus for setup and onscreen plots for data collection can get to be a drag.
But most customizations, including parameterizations, more ornate site setups (including
initiating species by area and letting them spread), statistics, etc., can be accomplished
via the same files that implement the Scenarios. It's a simple text file called a protocol.
And Gecko Woods can be run as an application instead of from a browser. In that mode, all the plot data (and the protocol!) is written to log files suitable for import to stat analysis programs.
Ginger Booth, December 14, 1999