From:    Wesley R. Elsberry                     
To:      All                                    Msg #261, 93-09-10 19:57:24
Subject: Dynamics of Animal Populations 930831


The text that follows is my set of notes for the WFSC 624 course here at
TAMU, "Dynamics of Animal Populations".  The next few messages are further
lecture notes.  Anybody who doesn't like this stuff can holler via netmail
to me here at 1:117/385.

 =========================================================================

WFSC 624 Dynamics of animal populations,
Dr. Kirk O. Winemiller

930829

Labs count for 20% of the grade

Last year, about 1/2 the class thought there was too much quantitative 
material in course, other 1/2 thought too little.

This used to be taught as a modeling course (Folse, maybe?).  Now, it is a
broad survey of the field, concepts, natural history.

Office hours:  for grad course, not really necessary, just drop by.

Text:  last year, Pianka's Evolutionary Ecology, good for concepts.  But
includes geological ecology, etc., that we don't need.  This year, 
Bingham and Mortimer's Population Ecology.

Note the dates in the syllabus.

Have to circumvent the enrollment cap...

Rest of this first period... review of terms and definitions, background
material.

Dynamics -- change over time.  A pattern or history of growth, change, or
development of an entity.  Field of inquiry which deals with motion and
equilibrium of systems under the action of forces usually from outside the
system.

Animal -- heterotrophs.  

Population -- deme: conspecifics which interbreed in a locality.  A group of
conspecifics in a prescribed area.

deme is the more specific term.

evolution -- changes in allele frequencies in a population over time.

Natural selection -- differential reproductive success.  Phenotypic traits 
must have a genetic basis.

directional selection -- selection favors one extreme phenotype.

stabilizing selection -- selection favors the mean phenotype and against the
extreme phenotypes.

disruptive selection -- favors extremes of phenotypes simultaneously.  
Proposed as a mechanism for polymorphisms.

Other hypotheses for polymorphisms:

shifting directional selection 

heterosis -- heterozygous genotype has the advantage, e.g. sickle cell anemia.

frequency depedent selection -- rare phenotype or allele is at a selective
advantage.

selective neutrality --  ???

End of polymorphism hypotheses

Genetic drift -- another mechanism for evolution.  Random changes in allele
frequencies.  Example:  imagine hurricane wipes out half the class, instant
change in allele frequency.

gene flow -- interaction between populations, breeding members.

meiotic drive -- segregation distortion during gametogenesis.  Violates the
independent assortment principle of Mendel.  This can skew the gene
frequencies.

mutation pressure -- force of mutation that is always out there.  1/10^5 or
thereabouts.  

Adaptation -- trait that enhances fitness, usually arrives via natural
selection.

fitness -- relative representation of an individual's offspring in the
population.

Group selection -- naive: "for the good of the species".  deals with a
meta-population view of the world.  Modeling cooperative behavior.  Problem
with most group selection models is the existence of "cheaters".

kin selection -- idea of inclusive fitness : instead of looking at fitness of
individual, look at the fitness of a gene.  Unit of selection then becomes a
group of closely related individuals.  Parent-child relation is 1/2. 
Sibling-sibling relation is 1/4.  Cousins 1/8, aunt-nephew 1/8.

End 930829


--- msgedsq 2.0.5
 * Origin: Central Neural System 409-589-3338 (1:117/385)


From:    Wesley R. Elsberry                     
To:      All                                    Msg #262, 93-09-10 20:00:58
Subject: Dynamics of Animal Populations 930901


WFSC 624 Dynamics of Animal Populations  930901

(got in late... having a car in for repair does that to ya...)

Law of parsimony
Occam's Razor
Assumptions

Verhulst-Pearl logistic growth model
-- assumes stable age distribution

Model testing:
Inductive method: induction -- if a prediction is true 9/10s of the time, it 
isprobably pretty good.

Deductive method:  deduction -- design an experiment or critical test to
falsify the prediction of the model.

We can't toss induction, since not all systems are amenable to the deductive
method.

Resource management field -- Carl Walters at U. Wash.  "adaptive management". 
When you have to make a decision about an action, modify current action to have
a component of the new action, so that you do two things, and see how it works
out.  Combination of both basic and applied research.

Scale in modeling: critical observations, evaluation

Hierarchy in modeling:
Ecosystem
  Population
    Organisms
      Tissues, cells
        Genes
          Molecules

Model selects which level one is looking at.

Other biological hierarchies:
phylogenetic evolutionary hierarchy

Don't want more levels than required, or fewer than necessary.

One of the problems in modeling in ecology is that the systems are complex,
with lots of relationships and interactions, and most components are highly
variable, and individuals that comprise a population are all different.
The downside is that it is difficult to know that you got all the right
factors.

Phenotypic variation
1) ecophenotypic (environmental bias)
2) adaptive variation (genetic basis)
[3) Random variation -- not really expected]


End 930901 
--- msgedsq 2.0.5
 * Origin: Central Neural System 409-589-3338 (1:117/385)


From:    Wesley R. Elsberry                     
To:      All                                    Msg #263, 93-09-10 20:02:22
Subject: Dynamics of Animal Populations 930903

930903 Dynamics of Animal Populations

1840's Leibig's law of the minimum: resource in shortest supply has the
most effect on characteristics of population ecology.  Pianka states that
the resources most frequently indicated as limiting are various nutrients,
water, and temperature.

Bernie Patten - complex networks:
emphasis that indirect interactions between species have considerable effect. 
A predates B predates C, indirect interaction A -> C.

Alan Berryman - perhaps Leibig's law has some validity.  Empirical research
does show that one thing usually does control a population.

Shelford's law of tolerance - too much or too little of any environmental
factor is detrimental to an organism.  Implies a range of optimal levels for 
an organism to survice and perform.

Example: jumping frogs and temperature.  There will be some sort of
distribution (usually normal) of performance that shows limits at both
extremes.

Principle of allocation - To increase performance results in a narrowing of
tolerance limits, in terms of ecological strategies.  Apparently impossible 
to build a superorganism, one that does everything well.

MacArthur and "species packing". When lots of species are around,
specialization is expected.

Distribution of individuals and populations in space.  Hope to finish this
today...  

Environmental heterogeneity (patchiness): Virtually all habitats have this
feature, and we can distinguish different scales (fine-grained vs.
coarse-grained) fine-grained (FG): patchiness is for small patches that are 
fairly regular.  Coarse-grained (CG): big patches relative to organism, so 
organism spends more time in any one patch.

Whether we call a habitat FG or CG depends on patch size, organism size, 
patch distribution, patch density, mobility of patch or organism.

Mobility: a continuum that encompasses 
 sessile, sedentary
 territorial: defended space
 home-ranging: not defended, fuzzy
 free-ranging: organisms that roam broad landscapes

Life history plays a big role in territoriality

Population consequences of level of mobility

Viscous populations v. fluid populations
viscous pop. are comprised of relatively sedentary organisms
fluid pop. are comprised mainly of free-ranging organisms

gene flow tends to be greater in more fluid populations.

tuna on the high seas, wildebeests, etc.
gene flow implies less genetic variability associated with space, since
organisms are so mobile, so you don't find much localized genetic variability.
Weaker meta-population structure (sub-groups on a two or 3 dimensional
landscape) in more fluid populations, stronger in more viscous populations.

In populations that are relatively viscous, distribution often tells about how
the individual organisms view resources.  Even, uniform, hyper-dispersed
distributions show regularity in distribution of individuals.  Examples: Ted
Case UCSD on ants in Australia, competition for limited resources.

Aggregated distributions of individuals is often evidence that Allee effect is
in operation: the fitness of individuals is enhanced when individuals are
tightly packed. Reproductive swarms.

Clonal reproduction: ???

Dan Janzen: recruitment rings in tropical forests.  Parental tree has seedling
recruits in a ring around the parent.  The probability of a seed settling on
the ground will have a negative binomial function of distance from the parent. 
Janzen also noted that the major source of mortality of seeds is predation. 
The prob. of a seed escaping predation is inversely related to distance.
Combine these probabilities, and you find a ring of higher probability of
survival.

End 930903 
--- msgedsq 2.0.5
 * Origin: Central Neural System 409-589-3338 (1:117/385)


From:    Wesley R. Elsberry                     
To:      All                                    Msg #264, 93-09-10 20:03:30
Subject: Dynamics of Animal Populations 930906

Dynamics of Animal Populations 930906

Concepts from last lecture:

Leibig's law of the minimum
Shelford's law of tolerance
Principle of allocation
Distribution in space
Fine-grained v. coarse-grained
Sedentary <-> free-ranging
viscous v. fluid pop.s
uniform (even) v. clumped (aggregated)
Tree recruitment rings

Ants in Australia as hyperdispersed: even dist. of resources can be the
determining factor, but other causes can possibly be in effect.

Start on demography today

Indices of dispersion: how to describe the distribution of organisms.  Lots 
of these approaches rely on sample quadrat: a square area for sampling.  The 
scale of the qradrat is critical.  Count organisms in these qradrats, then 
test the distribution in the quadrats against a random expectation.

(Read _ as an indication of subscript, ^ as exponentiation, strings as single
symbols, string+bar is "string" with a superscripted bar)

P_x = e^(-mu) (mu^x / x!)

P = prob. x individuals in a qradrat
x = integer count of individuals
mu = mean of distribution

Random expectation is the Poisson distribution.  For Poisson, tails off to 
one side (skewed).  The variance is equal to the mean.  sigma^2 = xbar.  I = 
sigma^2 / xbar.  If I = 1.0, then the Poisson distribution is supported.  
If I > 1, then organisms are clumped.  If I < 1, then organisms are evenly 
distributed in space.

Chi square test of goodness of fit is applied.  chi^2 = I(n-1), d.f. = n-1
(where n is number of qradrats).

Standard chi square test:
chi^2 = sum [(observed - expected)^2 / expected]

Another index: nearest neighbor distance.  Hopkin's index of dispersion.  
  H = sum (x_i^2) / sum (r_i^2).  
x_i is distance from random point to nearest organism.  
r_i is distance from random organism to its nearest neighbor.  
I_h = h / (1 + h).  
If I_h = 1.0, then clumped.  
If I_h = 0, uniform (even).  
If I_h ~= 0.5, then random.

Demography: meat & potatoes of this course.  Study of demes: a population 
which is a genetic unit, shares a gene pool.  Starting point: a life 
table: a record of birthdays and dates of death of individuals in a 
population.  Two approaches: longitudinal sample (dynamic sample or cohort 
analysis): follow a cohort (a group of individuals all born within a time 
interval) until all are dead, recording dates of death.  Horizontal sample 
(static sampling or segment): taking a snapshot in time of a population, 
figure out how old each individual in the population is.  Assumes that the 
various rates are not changing within age classes.

Life track graph, age v. time

Cohort sample follows diagonal of life tracks, segment sampling is a vertical
slice through the graph.

Segment sampling requires a good method of aging individuals.  Cohort sampling
takes longer, but is unequivocal.

Plotting deaths(y) against age(x).  Frequency histogram.  Infant mortality, 
old age mortality.  The frequency histogram is not good for comparison to 
other cohorts or populations.

Age specific death rate (q_x) provides a relative measure.  Plot q_x v. age.
  q_x = d_x/ a_(x-1).  
Also termed the "force of mortality".  a_x = number of individuals in an age 
class.  d_x = number of deaths of individuals in an age class.

Age specific survivorship (l_x).  
The proportion of the original cohort which survives through age class x.  
  l_x = a_x / a_0
Monotonically decreasing function.  Fraction of newborns that survive 
through age class x.  All these can be computed with calculus, which leads 
to "survivorship curves".

Type 1 curve: mammals, birds, humans: high survivorship early on.  
Type 2  has relative level force of mortality across age classes.  
Type 3 has extremely high infant mortality, but good survivorship in older 
age classes.  
Trees, inverts, and a lot of others have type 3 curves.

Let's construct a life table.  Five age classes.  Number of individuals a_x.

 x   a_x   l_x   q_x   E_x
 ----------------------------
 0   100   1.0   0     3.3
 1   90    0.9   0.1   2.55
 2   70    0.7   0.22  2.0
 3   50    0.5   0.29  1.4
 4   20    0.2   0.6   1.0
 5   0     0     1.0   0

E_x : expectation of future life.  
 
  E_x = sum (l_y) / l_x 

for all values of y from x to omega (omega is last age interval).  
Know for exam***.  This is the kind of measure life insurance companies 
use to determine premiums.  x is a time interval.  It is also an 
extremely handy concept for evolutionary biology modeling.

Other measures discussed in the text are less often used.

"Killing power" : K_x = log a_x - log a_(x+1) = log (a_x/a_(x+1)).

P_x : P_x = (1 - q_x)  is the probability of survivorship in an age class.

End 930906  Next time: fecundity and life tables.



--- msgedsq 2.0.5
 * Origin: Central Neural System 409-589-3338 (1:117/385)


From:    Wesley R. Elsberry                     
To:      All                                    Msg #265, 93-09-10 20:06:10
Subject: Dynamics of Animal Populations 930908


Dynamics of Animal Populations  930908

Concepts from last lecture:

Indices of Dispersion
Qradrats, Nearest-neighbor
Demography
Life table
Cohort (longitudinal sample)
Static (horizontal sample)
a_x, d_x, q_x, l_x, E_x, K_x, P_x

===========================================

N is number of individuals in a population.

 N_(t+1) = N_t - (N_t (1-P_x))
         = N_t - N_t(q_x)

 a_x = a_x - (a_x (1-P_x))

Look at table 1.2 in text when it comes in.  Red deer in England.  

B_x : age-specific fecundity (also known as m_x), average number of 
offspring produced by a female of age x.

Raw data v. smooth data: in horizontal study, some of the assumptions may 
not be met, or sampling error may be introduced.  Smoothing is integration
over a range of age classes to correct for some of this.

Fecundity curves: m_x (y) v. x (x)

Age of first reproduction is alpha, age of last reproduction is omega.

For semelparous (one-time reproducer) organism

  alpha = TAU

where TAU is mean generation time

For iteroparous organisms,

  TAU ~= (alpha + omega) / 2

 TAU = sum (x l_x m_x), for x from alpha to omega

For fecundity curves, most of the time you only consider females, due to
difficulty in determining paternity.  Females limit the reproductive 
capacity of the population.

Net Reproductive rate: R_0, average number of successful female offspring
produced by average newborn female over its entire lifetime.  R_0 tends 
to be near one, regardless of organism.

  R_0 = sum (l_x m_x), for x from 0 to infinity

R_0 is also called the replacement rate of a population.  It is an index of 
how individuals are replacing themselves in the population.  
 R_0 = 1 indicates a stable population.  
 R_0 > 1 indicates a growing population.  
 R_0 < 1 indicates a declining population.

If R_0 is near one all the time, then an inverse relationship must hold 
between l_x and m_x.  This has implications for life history strategies.

Reproductive value (due to R.A. Fisher) V_x, the age specific expectation 
of future offspring.  The measure of the extent to which members of a 
specific age class contribute to the next generation.

  V_x = sum ((l_t/l_x) m_t), for t from x to omega

for a newborn individual in a stable population, V_0 = R_0 = 1.0.  V_x for a
post-reproductive female equals 0.

V_x starts at one, peaks at alpha, then tails off.  Curve due to mortality
before alpha.

In a changing population, what does that do to the curve?  If a pop. is growing
the value V_x will be relatively lower for younger individuals.  Opposite for a
declining pop.

V_x can be termed present value of future offspring.

V_x can be partitioned into two components:

  V_x = m_x + sum ((l_t/l_(x+1)) m_(x+1)), for t = x+1 to omega

Basically splits the terms into "now" and "future".

If one can record deaths, births for a poulation, you can calculate 
alpha, omega, mean generation time, net reproductive rate, 
reproductive value, E_x.

If l_x and m_x schedules don't change, then the population will achieve a
"stable age distribution".  This means that you have constant percentages 
of individuals in age classes.  This is a "stationary" age distribution.

Demography

birth rate (b) = N births per X individuals per unit time

death rate (d) = N deaths per X individuals per unit time

If the total number of births exceeds the total number of deaths over 
some time interval, the population will increase.  
If total births < total deaths, then population will decline.  
In a closed population, the difference is the intrinsic rate of population 
change or intrinsic rate of population increase, r.

  (Malthusian parameter) = b - d = r

For open population

  r = b + i - (d + e)

where i is immigration rate and e is emigration rate.

 delta N / delta t = bN - dN  (discrete equ.)

 dn/dt = rN  (differential equ.)

This is the basic model of exponential population growth, growth without
limits, "J" growth curve.

One model is E. coli.  N ~= 2^t, t = 20 minutes.  Start with 1 cell, after 36
hours, N = 2^108.

End 930908


--- msgedsq 2.0.5
 * Origin: Central Neural System 409-589-3338 (1:117/385)


From:    Wesley R. Elsberry                     
To:      All                                    Msg #266, 93-09-10 20:07:20
Subject: Dynamics of Animal Populations 930910



Dynamics of Animal Population  930910

Concepts from last lecture:

Age-specific fecundity
Mean generation time

  TAU = sum (x l_x m_x) / R_0

======================================

N_t = N_0 e^(rt)

Euler equation (Lotka's)

  sum ( e^(-rx) l_x m_x) = 1

Estimate of R_0 with basic life table data:

  r ~= (log R_0) / TAU

Note inverse relation between mean generation time and r.  This holds 
pretty well everywhere.

r_max = per capita rate of increase for a population under conditions 
optimal for growth

 taxon      r_max   TAU(days)
 ---------------------------------
 E. coli    60      0.014
 Tribolium  0.12    80
 Rattus     0.015   150
 Homo       0.0003  7000

Any population has inherent capacity for exponential pop. growth.

 delta N / delta t = constant

constant really is r_max

Say we measure two points (N v. t), the slope is an estimate of r

If we track the trajectory of a population, we get the "J" curve

In exponential growth, delta N / delta t = rN

dN/dt = rN (now r is changing instantaneous rate of change per capita)
This changing value of r is r_actual (r_act, r_a are synonyms)

lambda = finite rate of increase

lambda = e^r

N_(t+1) = lambda N_t

1 = sum (lambda^-x l_x m_x)

Intraspecific competition

1. Ultimate effect is to decrease the contribution of individuals to the next
generation (decreasing the fitness of individuals within population)

2. Resource shortages (sometimes not apparent what the resources are)

3. Reciprocity - all individuals suffer a negative effect of increasing
density.

4. density dependence

(see hand notes for graphs)

Other indices of density dependence for populations

Exponential growth most often follows catastrophic population decline

Populations that have "sawtooth" N v. t plots are called 
"opportunistic" (colonizing) species: high capacity for exponential growth

Need a model that lets R_0 change with changing density: dealing with density
dependent dynamics.

Verhulst-Pearl logistic model:

K is carrying capacity of the environment.  K is defined by environment,
expressed in units of population density, it is the density of a population
that the environment can support.  When population is at K, R_0 is 1.0, and 
r_act is 0.

 dN/dt = rN ((K-N)/K)
       = rN (1 - (N/K))

"sigmoidal" (s-shaped, logistic) growth curve results from this


Assumptions:

1. all individuals are equal in pop. (same V_x)

2. r_max and K are constants

3. No time lag between change in r_act and change in pop. density N

Very seldom will one find that these assumptions are supportable.  Model
is still widely used, though, because it gets decent results.


End 930910



--- msgedsq 2.0.5
 * Origin: Central Neural System 409-589-3338 (1:117/385)


From:    Wesley R. Elsberry                     
To:      All                                    Msg #267, 93-09-10 20:10:16
Subject: Fill in the life tables


Here's a couple of life tables for exercise, should anyone out there be
interested in applying the notes...  Answer key later.


 x   a_x   m_x    l_x   q_x   l_x*m_x   x*l_x*m_x   E_x   V_x
 -------------------------------------------------------------
 0   400   0
 1   150   0
 2   100   1.0
 3   25    4.0
 4   10    10.0
 5   5     20.0
 6   0     0

 R_0 =
 TAU =
 

 x   a_x   m_x   l_x   q_x   l_x*m_x   x*l_x*m_x   E_x   V_x
 -------------------------------------------------------------
 0   50    0
 1   40    0.35
 2   30    0.50
 3   20    1.00
 4   0     0

 R_0 =
 TAU =

--- msgedsq 2.0.5
 * Origin: Central Neural System 409-589-3338 (1:117/385)


From:    Wesley R. Elsberry                     
To:      All                                    Msg #268, 93-09-10 20:48:30
Subject: Seminar on ecology notes

930910 Wildlife and Fisheries Sciences Departmental Seminar in Ecology series

Seminar - Ed Rykiel of Biosystems Modeling section, Industrial Engineering 
Dept., TAMU

Rykiel has been a site reviewer for DOE, NSF.  Lots of other indicators
of professional excellence, etc.

Pres., Theoretical Ecology section of Ecological Society

==========

Talk about something that we are familiar with: habitats and species.

There is an interaction between plants and animals, habitats change with 
animal use.

Cluster Phase Dynamics in Vegetation

Current paradigm is "Gap phase dynamics" : fallen tree creates a gap that 
gets filled in.

"Cluster Phase" : succession by clustering around a plant.
 - nurse plants
 - nucleation
 - change to a different vegetation phase

Gap Phase
 - shifting mosaic
 - statistical ensemble
 - successional return to similar vegetation phase

Gap phase
 - dynamics based on competition
 - close seed source
 - affected areas usually shrink
 - animals relatively unimportant
 - gaps do not interact

Cluster phase
 - facilitation eventually gives way to competition
 - distant seed source
 - affected area expands: clusters grow outward
 - animals relatively important
 - clusters can interact

Facilitation (out of fashion in ecology, due to assumption of benignity)
 - active - biotic processes are involved in facilitating differential 
   establishment of vegetation
 - Passive - abiotic and biotic structure attracts seed dispersers;
   differential establishment is simply a matter of seed density

The Role of animals
 - animals speed up the slow plant to plant and cluster to cluster 
   interactions by facilitating the seed dispersal process
 - vegetation change occurs by a combination of rapid saltation and slow
   diffusion
 - woody vegetation can expand into open space

opportunistic tree species - juniper with cedar waxwings (slide)

Where do the CW's disperse the juniper seeds?  (On my car, audience comment).
Not particular about where they put the seeds.  Fence line - we call it a 
fence line attractor: CW's drop seeds near fence lines.

Power pole attractor - another abiotic structural feature for the dispersers
to cluster around.

Windmill attractor

(That's a strange attractor, audience comment)

Facilitation
 - population processes
 - enhanced disperasal
 - enhanced gemination
 - reduced mortality risk - nurse plant reduces mortality for seedlings
 - suppression of herbaceous competitors

- physiological processes
 -- reduced energy load
 -- reduced moisture stress
 -- enhanced nutrient status
 -- increased soil moisture holding capacity

Examples of cluster phase dymamics in Texas
 - post oak savannah
 - rio grande valley

Post Oak Savannah
 - seed deposition under post oak

Note difference of litter within and without the cluster (slide). [Litter 
within is relatively heavy and deep, opposite outside.]

if the nurse plant is relatively small, it doesn't take long for the 
juniper to shade out the nurse plant.

Hypothesized successional sequence
 Post oak with juniper seedlings
 Post oak being shaded out by junipers
 Juniper cluster grows

Now for Rio Grande...

Mesquite seedling - how they get estab. is another question
Once mesquite is established, it becomes a biotic attractor.
Shrub vegetation starts under mesquite.
Cluster then expands around the mesquite nurse plant
Mesquite (Prosopis glanulosa) is very deeply rooted, serves as nitrogen 
fixer, nitrogen pump, was well as water pump.
Shrub patches inhibit the growth of mesquite seedlings.

Nutrient cycling is being changed by cluster as well as by the animals.

Eventually, mesquite dies, leaving cluster, groves can form with 
coalescence of clusters.

Is nucleation/clustering a common phenom.?
- USA: CA oak woodlands, sonoran desert, chihuahuan desert
etc.
??? lots more examples flashed by too quickly

 - clustering has been observed in a wide variety of ecosystems
 - clustering is most likely to be important in ecosystems where at least
   seasonal drought occurs
 - savannah and ecotones appear to be good candidates for clustering

landscape effects
 - interactions at the individual plant level lead to cluster formation and
   expansion
 - cluster expansion leads to interactions between clusters at the landscape
   level
 - the result is a meta-stable two phase vegetation system w/ clusters 
   expanding and contracting in response to climate, fire, grazing, and 
   other influences

Conclusion
 - contrast w/ usual structuring paradigm (Gap Phase)
 - cluster phase succession commonly occurs
 - can be important in changing the system
 - explicit role of animals in veg. structure and composition
 - saltation v. diffusion dispersal process

if you leave out anmals in system, you won't understand how system developed.

Facilitation is a real ecological process: species are not fixed in role, 
roles can change over time.

Cluster Phase succession is more likely to be seen in ecotones, savannas 
because you don't have canopy cover.

Neill: How about advection as well as saltation?  
Neill: saltation implies random, advection implies direction.

Neill: the diff. between gap and cluster seems to be reflected in moisture
gradient: high rainfall -> gap; prairie, savanna -> cluster

Winemiller: Leibig's law of the minimum may play a role rather than simply
moisture gradient.

What you'll find generally is that gap is dominant paradigm, grasslands folk
force gap on observations.

A number of questions remain about junipers: dieocious species, recruitment,
seeding, etc.

Deserts: structures in deserts can be seen at all scales.  Facilitation is 
seen in any relationship that uncouples system from climatic conditions of
environment.

Neill: fractal analysis may point to dominant process.  With landscape
interest, someone should be looking at this.

The question has not yet been framed well yet...

This process is not uncommon: why has it been ignored?

End 



--- msgedsq 2.0.5
 * Origin: Central Neural System 409-589-3338 (1:117/385)


From:    Wesley R. Elsberry                     
To:      All                                    Msg #269, 93-09-10 20:51:00
Subject: Event recorder


Diane had been working on me for a couple of years to program at least
a minimal event recorder.  Well, taking the WFSC 422 Ethology course finally
provided the activation energy, and last night I did a quick and dirty 
program for my trusty TRS-80 Model 100 portable computer in the Mod100 BASIC.

What's an event recorder, some may ask.  Well, I don't have a concrete example
to work from, since they don't come cheap and my only contact has been reading
about them in the literature.  But the basic idea is that an event recorder 
provides a way for an observer to note emitted behaviors in subjects in real
time.  The type of behavior and a timestamp are the minimal pair to be 
recorded.

Today, I tried it out during a student presentation.  The program records 
single characters with a time stamp in the form of a string.  Resolution is
only down to the second.  For my practice run, I came up with a repertoire
of four behaviors, coded as 

 "e": eye contact with audience
 "f": face touch with hands
 "h": any hand gesture excluding "f"
 "u": "uh" or "um" vocalizations

The results are:

 Date: 09/10/93
 u 12:05:23
 u 12:05:28
 e 12:05:31
 h 12:05:39
 e 12:05:51
 e 12:05:56
 e 12:05:58
 u 12:06:01
 e 12:06:04
 h 12:06:05
 e 12:06:08
 e 12:06:11
 e 12:06:15
 h 12:06:16
 h 12:06:18
 h 12:06:21
 e 12:06:23
 h 12:06:24
 f 12:06:26
 e 12:06:30
 e 12:06:34
 u 12:06:40
 u 12:06:42
 e 12:06:44
 u 12:06:50
 e 12:06:53
 h 12:06:55
 e 12:07:05
 h 12:07:07
 u 12:07:08
 e 12:07:10
 h 12:07:14
 e 12:07:15
 u 12:07:17
 h 12:07:20
 e 12:07:30
 e 12:07:33
 e 12:07:53
 e 12:08:06
 e 12:08:11
 h 12:08:21
 e 12:08:26
 e 12:08:40
 h 12:08:48
 h 12:09:00
 f 12:09:16
 h 12:09:27
 f 12:09:32
 e 12:09:35
 f 12:09:45
 e 12:09:46
 h 12:09:51
 u 12:09:52
 f 12:09:53
 f 12:09:56
 e 12:09:57
 h 12:10:04
 u 12:10:05
 u 12:10:09
 f 12:10:11
 e 12:10:12
 u 12:10:12
 e 12:10:14
 u 12:10:19
 f 12:10:21
 e 12:10:24
 f 12:10:29
 u 12:10:31
 e 12:10:35
 h 12:10:42
 f 12:10:43
 e 12:10:44
 u 12:10:44
 e 12:10:54
 u 12:10:56
 e 12:11:03
 u 12:11:14
 h 12:11:17
 u 12:11:21
 u 12:11:24
 e 12:11:25
 u 12:11:26
 u 12:11:31
 h 12:11:33
 u 12:11:38
 e 12:11:50
 h 12:11:57
 f 12:11:59
 u 12:12:00
 h 12:12:03
 f 12:12:05
 e 12:12:10
 h 12:12:13
 h 12:12:14
 u 12:12:16
 h 12:12:18
 f 12:12:21
 f 12:12:23
 e 12:12:23
 u 12:12:29
 h 12:12:30
 e 12:12:36
 h 12:12:38
 
end of talk, end of observations

I'll be working on refining this tool over the course of the semester, and
applying it to interesting behaviors later on.

The program as it stands now, though, is this:

 10 REM event recorder 
 20 ON KEY GOSUB 1000,2000
 30 KEY (1) ON:KEY (2) ON
 40 D=0:G=0
 900 REM
 910 IF (G = 1) AND (D = 0) THEN GOSUB 3000
 920 IF D = 1 THEN GOTO 930 ELSE GOTO 900
 930 KEY (1) OFF : KEY (2) OFF
 940 CALL 23164,0,23366:CALL 27795:REM restore function keys
 990 STOP
 1000 REM F1 int. service
 1010 INPUT "enter file name";F$
 1020 A$ = "ram:" + F$ + ".do"
 1030 OPEN A$ FOR OUTPUT AS 1
 1040 PRINT #1,"Date: ";DATE$
 1050 PRINT "Date: ",DATE$
 1060 G = 1
 1070 RETURN
 2000 REM F2 int. service
 2010 CLOSE 1:D=1
 2020 RETURN
 3000 REM poll for key, then log it and time
 3010 B$ = INKEY$
 3020 IF B$ = "" THEN RETURN
 3030 PRINT #1,B$;" ";TIME$
 3040 PRINT B$,TIME$
 3050 RETURN

That's all, folks.


--- msgedsq 2.0.5
 * Origin: Central Neural System 409-589-3338 (1:117/385)