D,evelopnent
of a simulation model for
Onchocerciasis transrnission and control
in
O.C.P.Report 2nd ph.="
Anton Plaisier Gerrit van Oortmarssen
Dik Habbera
Ilept.
of
Public Ee-al-th and Social Hedicine Erasmus University RotterdamP.O. Box 1738 3000 DR Rotterdam, Ttre Netherlands
tel.
010-
4634092January 1987
Ttre model-development
is
carried out onthe basis of
aTechnical Service Agreemnt
(A),
number 08/18f/85, providedby the t{orld Health Organisation, on behalf of
theOnchocerci-asis Control Progranrne (OCp).
1
COTiITBTTS
I
The model VECPARII
Reporton
rapplied modelling anditrs utilisation
in the O.C.P. IfII
Proposalfor
a simple transmission modelIV Exarnple
of
input and outputof a simple
transmission model\tBcPA8l.0
A coqruter prog,rao
for
sinrlatfDs thellfe-cycles of Stmuliuu dr-osro !.1. end
Onchocercavolrnrlus, and
tbeir Lntcractlo.
Antoo
Pleist.r Gerrlt
van OortarrsenDIL Eabbera
Teclrrlcal
relnrt (draft)
preperedfor
IiEO/OGPIlept.
of
Publlc Eealtb and Soclal ]lediclneErasrnus
llnlverslty
RotterdaP.Or Box 1738 3000 DR Botterdqil, The Netherlands
tel.
010-
4634092Septenber 1986
The developnent
of
the VECPAR prograr ras carriedout
onthe basis of
a Techninal Serul.ce Agreeuent(A),
ntuber08/f8fl85,
provided by the ltorldEealth
Organisatlon, onbehalf
of
the Onchocerclasl,s Control Programe (OCp).I
Table
of
contents1.
Introduction.Model-descript ion
2.I Summary..
:2.2
Detailed descriptionof
the model?.2.1
The larvae submodel .2.2.2
The fly-larvae-interaction3.
Discussion items Users guide4.1
Starting a session4.2
Using VECPAR 22
3
submodel
4 6
9
4
.12 .
13Appendix 1: Mathematical summary
(draft)
22appendix
2:
Report on the 4th meeting on applied modellingin
theOCP
25I
IntroductlonThls
report describes VECPAR1.0, thefirst
test-versionof
the model for the analysisof
vector-parasite interactionin
river-blindness transmission.The purpose
of
the documentis to facilitate
testingof
the model,and
toraise
discussion regarding some aspectsof
the modelthat
need further improvement. The VECPAR model describes one partof
the transmlssion cycleof river
blindness (onchocerciasis). Another model, HUMPAR1.0, describes the human-parasite aspectsof
the disease, and was developed by ourgroup
ln 1985.This
model was formerly called ONCHOSIM, we preferto reserve
thisname
for
the whole packageof
the computer programs under development. Theaim of these
modelsis twofold: first
they should be used asan aid
in interpretationof
the data collected by the Onchocerciasis Control Programme(OCp) conducted
in
WesternAfrica;
secondly the two modelsare
detailed Precursorsof
corresponding building blocksof
aful1
transmission model of Onchocerciasisthat will
be developedin
the forthcomingtime. It is
hopedthat
thefull
transmission modelwill
be an importantaid in activities
for the OCP,e.g.
with respectto
planning and long-term predictions (migration, recrudescence, devolution) andin
choosing betweendifferent
policy options.A
numberof
meetings have been held on the subjectof
modellingand its
applicationin
analysis and planninguin OCP. The most relevant meeting forthe
present VECPAR model was the4"",
which took placeat
ourinstitute
in Rotterdam(dec.
16-18, 1985) andin
whichstaff
membersof
OCP/VCU, OCp/EpIand
OCP/STATparticipated. In
the reportof this
meeting (see Appendix 2)the
provisionallist of
contentsof
the model can be foundthat
was agreedupon by the
participants.
Thislist,
and an outlineof the
model, werediscussed
further at
avisit of dr.
Remmeto
Rotterdam(april
18and
2L,1986).
The present versionof
the model has been developed alongthe
linesof these
meetings.The
aspectsthat
arenot yet (or onry partially)
implemented are
listed in
the discussion section.The VECPAR model describes the part
of
the transmission cyclethat
starts with intakeof microfilariae
by Simulium damnosum, and ends when infectivelarvae are
released during a subsequent bloodmeal.fn the
model, twoprocesses interact:
1.
The developmentof
larvaein
thefly,
through stages LL, Lz and L3, to theinfective
stage.2. The life-history of
maturefries
includingbiting
behaviour.The model does not describe the dynanics
of
thefly-population,
which wouldinvolve modelling
of
the breeding behaviour.The output of
the modelis
computed and displayedin
such a waythat it
can be compared
directly
with the resultsof
the OCP/VCU catchingsites.
Inaddition, the
model can display graphs and tablesof
thelife events
forflies
andlarvae.
Thelatter
output can be usefulin
the formulationof
aconcise vector-parasite module
in
thefuII
transmission model.Section 2 starts with a
global overviewof the
modeland of
the assumptions madefor
the 2 processes.Next,
a detailed (more technical) descriptionis
givenof
the two processesseparately. In
the discussion ofsection 4,
aspectsof
the modelthat
could be incorporatedor
improved arepresented. The most,important
point is
probably migrationof flies, that
is notyet
implementedin
the present version. An outlinefor
an implementationof
migrationis
presented. A users guidefor
the VECPAR modelis
presentedin
section 3 and includes technical points regarding the simulation and the performanceof
the model.In
Appendix 1 a mathematical formulationof
the modelis
given.2
2.
l{odel descrlptlon2.1 Sumary
The model
is
confinedto
the vectorial partof
the transmission,i.e.
fromintake of
microfilariae from a hurnan hostuntil
therelease of
infective larvaeto
a human hostat
a subsequent bloodmeal. The goalof
the model isto
simulate asituation similar to
thefield-situation
during a VCU-catchingsession; that is:
afly
population with a certaindistribution
with respectto fly-ager
parousrate
andlarvae-load. It is
assumedthat the fly-
populationis in
an equilibriumsituation, i.e. that
the age-distribution is constant overtime.
Underthis
assumption, the age-distributionof
thefly-
populationis directly
relatedto
thelife-table.
Consequently,for
theequilibrium
situation,
the momentarysituation is
equivalentto
the weightedlife-events of
a specific cohortof flies that is
followedfrom
maturationuntil
death (seefigure below).
This equivalenceis utilized in the
mod,eland all
results are extracted from the simulationof
the courseof life
of such a cohort.fly-
age
4-
+
time catching -
sesston
figuret
'+'denotes deathof
theflies.
For the sakeof simplicity it
assumed
that all flies
die on the same age1S
In the
model, thefate of
thefly-cohort is
determinedby the
followingprocesses !
1.
Intake of
larvae from the humanhost,
possibly followed by deathof
the vector.2
3
Release of infective
larvae'during successive
sugar- and bloodmeals.Development
of the
larvae through intermediate stagesto
theinfective
stage.Ageing and mortality
of
theflles.
Migration
of flies
(notyet
implementedin this
version).Key-variables
in
the modelare:
the lengthof
the bloodmeal-intervals andthe possible durations of
theIarval
stages(Ll, L2, L3(B) and
L3(H)-infective). Other
variablesare:
the naturalfly-mortality, the
larvae- intake by theflies
and the consequent excessfly-rnortality, the
mortalityof the larvae in
thedifferent stages,
and theproportion of
infective larvaethat is
released during sugar- and blood-meaIs.A number
of
simplifications have been madein
the model.First, it
isassumed
that larvae
ingested more than two bloodmealsago can all
be consideredas
Iinfectiver
regardlessof
the ageof these larvae. As
aconsequence,
no provision is
made (thusfar) for
the existenceof
(L3B)larvae that
stayin
the bodyof
thefly
butnever
becomeinfective.
The secondsimplification
concerns therelation
between human microfilariae-loadand intake
of microfilaria
followed by passageof
the gut-walI, whichis
notyet
includedin
the current versionof
the model.Instead it is -
forsimplicity -
assumedthat
an invariable tintakerof
Ll-larvae can actas
a shortcut.Finally,
the processof
migrationof flies,
which undoubtelyis
ofmajor
importance,awaits
inclusioninto
the model. The waywe think
to implementthis
processis
presented as an itemof
discussionin this
report.The output of
the model consistsof
thedistribution of
thebiting fly-
populationover the
numberof
inhabitant larvaeof all types, and
thedistribution of
the numberof
larvae releasedat
a bloodmeal. Partof
theoutput is in a
formatthat
enablesdirect
comparisonswith the
VCU- tabulations.2.2 bXaTled description
of
the nodel2.2.L The larvae-subnnodel
In
the larvae submodel, the input-data on thelife-history of
the larvae are transformedinto
a reference tablethat
gives,for
any day since intake, the statusof
thelarvae.
This tablewill
be consultedduring
calculationswith the fly-larvae
submodel. Four developmental stagesof larvae
aredistinguished: Ll-
(immediatelyafter
intake and passageof
the gut wall), L2-and L3-larvae in the
bodyof the f1y (L3B) and
L3-larvae (morphologicallyidentical to
L3B-larvae)in
the proboscis(L3H). Only
the L3H-larvaeare infective and
can be transmittedduring a
sugar-or
abloodmeal.
The
submodel deals with the developmentof the larvae, starting
from4 5
4
intake until
the momentat
which a larvae becomesinfective
(reaches stageL3H).
The L3H-larvae are not treatedfurther in this
submodel,but
arerecorded
in
thefly-larvae
interaction model. For thefirst
three stages thepossible
durations (program equivalents:durst!!,
dusrtl2iespectively; in aays)
togetherwith [E-p*lr6lFti""
L3-body : o.030 i 1 0.0000
1.0000 z
Results parasite-submodel (reference table) :
L1 LZ L3 L3-head
and durstl3body
( robdurstLl
proportion of larvae, en-
gorged at day 0, ttrat is in each of the stages at a
given age (ttprevaleneerr) proportion of larvae, en-
gorged at day 0, entering the fly-head at given age (Itincidencerr )
probdurstl2 and probdurstl3body respectively) on these durations should be
given. The life-history of
the larvaeis further
influencedby
dailymortality
which should be statedfor
each stage(Ll, L2,
L3B) separately.Starting from these basic data
all
possible life-history-pathsof
the larvaeare calculated and
ultimately summarizedinto the
reference-table. Anindividual
life-history-path
ends when the larva diesor
whenthe
L3H-stage has been reached. Fromthat
moment thefate of
the(infective)
larvawill
bemanaged
by the
f1y-larvae-submodel.An
exampleof input and
output (referencetable) for this
submodel lookslike:
Input-variables parasite-submodel :
Stage Daily
mortality Stagedurations (days)
Probabi-
lities
L1 : o. 100 1
2i
0.00001.0000L2: 0.070 I
2 3
0.0000 0.8000 0.2000
\
Age Stage
I
2 3 4 5 6 7I
0.900 0.810
0.544 0.654 0. 123
Column L1-L3
Colum L3-head 0.753
0.701 0.130
0.501 o. 116
The result
shouldbe
read asfollows: At day 5
13.02of the
larvaeoriginally
takenin at
day 0is in
the L2-stage and 54.42in the
L3B-stage.The remainder 100-(13.0+54.4)=32.62 died during rhe
first
4days. At
day 712.32
is in
the L3B-stage while 50.12just
entered the L3H-stage. This 50.12 L3H-larvaeis further
processedin
the f1y-larvae- submodel. This means that the 11.62 L3[-larvae on day 8 represents theshift
from the L3B-to
the L3H-stage
at that
day only1
z
3
4 5
2.2.2
lhe
fly-larvae-interactlon subuodelIn simulating the life-history of
aspecific fly-cohort the initial
population consistsof
nulliparousflies that
have thefirst
bloodmeal. Theoutput for
the equilibrium-population should be basedon sinulated
eventsfor all fly-ages.
Therefore,the
numberof
days simulated should besufficiently
largeto
guaranteethat at
the end only a negligible proportionof
the cohortis still
alive.It is
assumedthat at
any given moment an individualfly
can be completely characterized by:The number
of
larvae engorgedat
thelast
bloodmeal (NLN) The time (days) since thelast
bloodmeal (TL)The number
of
larvae enBorgedat
thelast
but one bloodmeal (NLP) The time-lag between thelast
andlast
but one bloodmeal (TP) The numberof
L3H-larvae (NLH)In
orderto
accountfor all
possible combinationsof
the5
characteristics(although,
depending on theinput,
notall
combinations arepossible) all calculations are
performedin
a S-dimensionalstate-matrix. In the
state matrix, the fly-frequencyis
recordedfor
each combination (NLN,..rNLH). Thestate-matrix
is
updatedin
simulating the changesin
thefly-population
dayby day. The simulation
starts with
10000 nulliparousflies at
thefirst
f1y-age (A) that, after
emerging from pupae, theflies
can takea
bloodmeal(depends
on the values in
probcvclelenRth,see variables
below anddiscussion-items). The start-population
fits in celt
(NLN=0, TL=A, NLP=0,TP=A, NLH=0)
of
the statematrix.
Thedistribution
over the numberof Ll-,
L2-and
L3(B + H)-larvae can be obtained by combining the state-matrix (NLN throughTP) with the
reference tablethat is
producedby the
larvae-submodel.
Each
day
(=each fly-age) theflies
are assumedto
undergoa
number of processes resultingin
analteration of
the frequenciesin
the state-matrix.These processes
follow
from the input-specificationsof the
fIy-larvea-submodel. These specifications are (names
of
variables underlined):Daily mortality
(fraction) of
theflies
(dailymortSD)Daily mortality
of lnfective
(L3H-)larvae (dailvmortlH)Probabilities for
the possiblelengths of the
bloodmeal-cycle (probcvclelength)Average
fraction
releaseof infective
larvae per sugarmeal(as
arule
preceding a bloodmeal) (propreleaseSugar)Average fraction release of infective larvae per
bloodmeal( propreleaseBlood )
Proportional rintaker during a
bloodmeal,of Ll-larvae
in categoriesthat
must be defined (tlintate)
The
excessmortality (fraction) to
whichflies
are subjected when certain numbersof
larvae (excessmortsD)d
b c d e
f
oD
they engorge
Some
of
the processes (especially when relatedto
larva1 development)also influenced by the reference-table produced by the larvae-submodel.
The processes
that
theflies
are assumedto
undergodaily
are:6
are
I
2
3
4
5
6
Fly-mortality resulting from causes other than intake
of larvae.
Thismortality
reduces the numberof flies in
eachcell of
the state-matrix with a glvenfraction
(dailymortSD).Mortality of infective larvae in
theflies.
Asa result of
thismortality,
characterislic 5 (NLH) changes.In
other words therewill
bea shift along
the5""
dimensionof the state-matrix, following
abinomial
distribution
with the fractionmortality
(dailymortlH) and theresulting
larvae-load (NLH) as thep- and k-
parameters respectively (see also mathematical summary).Based
on
the reference-table fromthe
larvae-submodel,larvae
from previous bloodmealsthat
have reached the L3H-stage are addedto
the poolof infective
larvae, thusaltering
the frequenciesfor
dimension 5(UtH) through a
shift
toward higher numbersof
L3H-larvae.Dependent
on
the time sincelast
bloodmeal(!t)
andthe
distribution (probcyclelength)a
numberof flies will havla
new bloodmeal. Theflies
are placedin
two separate temporary state-matriceswith
equalstructures, one for
the non-biting and onefor
thebiting
population.Biting flies:
Will release a
numberof infective
larvaeduring a
sugarmeal preceding the bloodmeal (on the basisof
propreleaseSugar), thus reducingtheir
L3H-store.Will release
a numberof infective
larvaeduring the
bloodmealitself
(on the basisof
propreleaseBlood)Will at
the same time engorge a numberof L1-larvae
(Llintake)from the
human host(to be
generalizedtrough a
vector/host submodel, see discussion).(The processes
a. b.
andc., that result in
a changeof
the numberof
larvae (NLN,NLP,NLH) are controlled again by the binomialshift
method)
May
die
as aresult of
an (high) intakeof larvae. This is
asimple proportional
reductionof
the content(nr. of flies)
of each elementof
the state-matrix.The
consequencesof
havinga
bloodmealare reflected in all characteristics.
The existing informationin
NI.N and TL regarding thepreceding bloodmeal
is transferred to
NLP-andTP-(last but
one bloodmeal)and
the contentof
NLP(intakelT last 6It one
bloodmeal)is -
dependent on the time since intake and weighted against the larvae reference-table-
addedto
the L3Hstore. Finally,
the time since thelast
bloodmeal(tl,) is
resetto
1 day, the intake follows from 4c.Ageing
of flies that
didnot
have a bloodmeal. The timesince
lastbloodmeal (TL)
is
increased one day.The
temporary state-matricesof biting and non-biting flies
are assembled againin
theoriginal
state-matrix.In the figure below a schematic representation of the simulation processes is given.
To enable a correct comparison between model-results and catching-point data, the model-output is extracted between steps 4a and 4b, i.e. just
a
b c
d
model slructure
agcng ot fles anc larYa.
mortality of flies
monaltty of L3l-l-hrvae
ncldence of L3-+-hrvac
aigarmacl + L3Fl-loeB
bloocfflcal + L3l-t-loss
+
larvae-intake nuillparous pcpulattcrl+
fly-mortalitybefore biting.
This ' information about the exampleof
model-input output lookslike:
at
the sametime,
impliesthat at this
moment onlybiting
partof
the populationwill be displayed.
An (notdisplayed:
the reference tableof
page5)
and -Input-variab les f ly-parasite-suboodel : Categories for number of tarvae :
Proportional Ll-intake in category:
Mortality caused by intake :
Bloodmeal cycle:
Possible cycle-durations (days) Probabilities for durations
0 o.500
I 2.-3 4-6 - 0.300 - 0.100
>6
i0.200 i
0.200 i
Daily mortality of L3-head larrrae : .050 Daily mortality of flies : .I50 Average proportion release of L3-head larvae
duringasugarneal : .100 duringabloodmeal : .400
l1 3
o.600 4i
0.400 i
2
Results f ly-parasite-subnode I
Sirmrlated distributions equilibriuro population:
Parous rate: 0.530932
Stage Larvae-categories:
01
>6
i >0 i Nr. Larv.lflyL1 LZ L3ilI
released LH
100.000
7 5.489 91.538 87 .047 90.32L
0.000 2.795 0.343 0.978 0.011 0.000
o.429 2.408 2.227 4.593
o.000 24. 51r 8.462 12.953 9.679
0.000 2.111 o.497 0.852 77.32t
2-3
.000 .723 .901 .380 .280 o 6 2 5
4
4-6
0.000 14.564 7..809 4.369 0.794
Simrlated VCU-table :
Total number of flies dissected Total number of parous flies Total number of nulliparous flies
:2O040 :10640 :9400
Parous flies wit-h : --) | no i LIandL2 | Llorl2 | L1orLZ I andl3 I orl,3 i
LlorL2 L3only only
L3
I larvae
Nr:rnber of flies 0
No. flies/1000 parous Number of larvae
3913 361 .8
6727 632.2
49L2 46L.7 10368.2
337L 316.8 4548.4
3356 315.4
1815 170.6
Results of catching point pC xx
Iotal nunber of flies dissected
Iotal number of parous flies :63042L2324
Total ntrmber of nulliparous flies :6OiO Parous flies with : --
Nunber of flies
No. flies/1000 parous Number of larvae
>igo llarrrae landl3 !orl3 lHandlzlLlorl2 lLlorl2 i i - m
i -- I Llorlz I Llonlv
I only i
---;;;---
-;;i-
63.6 41.4 s508
87 3.7 zL.3134
796 535
84.9 1937 .0
395 62.7 1551.0 726 3
CompariEon
of
UECPRH results and 0Cp/U[U data date!16- ?-1998time: g:44:eg
UCU catchinssite:
pC xr 1ffi7,807, 607, 407, 207, 07,
PAROU$ [AT[
w7,
Sim U(Ll
WWN
L3 and LllX
LAEUAL STAIIJS
$im
\iCUTM
AI
I
Stases#LAEUf;VPA[[]U:.' FLY
L1
or LI
Li$im l,ttj $im
r/,:l[J60fr
4fr7,
?0v,
07,
0.60 0.50 0.40 0.30 0,20 0,10 0.00
VTN
onls Ll?
lmv LI
onl'Jlegend:
L3-only
=flies
with only L3(S + H)-larvaeonly
L12
=flies
with only L1- and/or L2-IarvaeL3 and LL/LZ =
flies
with L3-rarvae andeither Ll- or
L}-larvaeAll
stages =flies
withall
larvae-staBes (L1- and L2- and L3B and/or L3H-larvae3.
Discusslon itemsIn this section
some aspectsof the current
modelthat might
bequestionable are discussed, and some proposals ar made
for
implementation of factorsthat
are missingin
the current version.1.
Comparisonof
simulated data with VCU-dataIn
additionto
comparisons between summarizingtablesr w€ intend
todevelop the
tools for
a more comprehensive comparison with the detailed catching-point counts we already received onfloppy. Both from
the model-results and the counts a crosstable with larvae-1oad combinations can beextracted.
With such tablesall
possible comparisons (inctuding thedistribution
tablesin
the exampleof
2,3) can be made.2.
Thelife-cvcle of
the lLgThe user
is
offered the opportunityto
lookat
the agespecific life-
events
of
theflies
(see 2.1 and the menuin rstart
simulation'in
4.).Until
now we did not make an attemptto
interpretthis
interim-results, though suchan
interpretation may be important withrespect to
the implementationof a
vector-parasite modulein a
fuI1-transmission model, and therefore suggestions aboutthis
matter are welcome.3.
MigratiolnIn
the current version, migrationof flies is
not implemented. However, wethink that
afairly
simple migration option can bebuilt in
withouttoo
muchtrouble.
This optionis
an additionto
the simulationof
alocal fly-population. A
second,distant
populationof flies
is simulatedfrom
whichflies will
migrateto the local population.
Asimple submodel
for
migration would involve the following steps:The simulation will be started with a local and a distant fly population at day 1 (= fly-age A, see page 6)
The life histories of both populations will be followed simul- taneously (the local population proceeds in the way implemented now).
The simulated part of the distant population consists of those flies that will fly to the 1ocal population
At the first bloodmeal in the distant population, the intake of microfilariae may be different from the loca1 population
A11 distant flies depart after the first bloodmeal
The duration of the flight should be specified. This duration is defined as the time between the last rdistantt bloodmeal, and the next (first tlocalt) bloodmeal.
Upon arrival, the distant flies (and their characteristics) are
mixed with the 1ocal flies. The toriginr of flies in the loca1
population is not recorded.
Points of discussion:
The excess mortality is not modelled explicitly in this approach.
Excess mortality before arrival in the local population could be
taken into
account by incorporatingit in
the sizeof
the distant population, andin
thedistribution of
theflight
duration. Excessmortality (from
exhaustion)after arrival is
neglectedin
thisapproach.
Only flies
departingdirectly after the first
bloodmeal are considered.Only
one distant populationls
considered. Thisone
population' however, could be considered as the sumof
a numberof
populationswith average
distribution for
intake andflight
duration.4.
Intake submodelAs already mentioned
in 2.1, in
the current versionof
the model we usean
imaginary Iintake' of
Ll-larvae asa shortcut for the
chain:Microfilaria-intake -
passing the gut-wallof
thefly
and becoming anLl-larvae.
Technicallyit will
be a rather simple operationto
includethis
chainin
the model, but as long as no data are availableon
theexact nature
of this chain,
alinear
relationwill
be asumed, r*hich isin fact
no more than an extendedshortcut; or in other words:
[Lt-intakel will
be rplaced by lMF-intakex
passaBe-factor], whichis
noreal
amelioration.At this
point we wouldlike
too receive suggestions.Strongly related to this
itemis the
excessfly-mortality
which probablyis
theresult of
a high intakeof microfilariae.
This excessmortality is
nowrelated to the Ll-intake classes, which is
asimplification.
The wayin
which therelation
between MF-intake andthis
mortality may be includedin
the model depends on the solution forthe
t intaket -item mentioned above.5.
PersistinR L3B-larvae, remigrating L3H-larvaeAn item that
might be arestriction in
the current modelis the
factthat
L3B-larvae caneither
become an L3H-larvaeor die.
Once being anLH(infective)-larvae there
is
no wayback.
Somefield
observations,however,
(H.
Remme,pers.
comm.)
point outthat
an L3B-larvae canpersist in
the body and never becomeinfective (tsticking in a leg').
In addition, it
may be possiblefor
aninfective
larvaeto
re-invadethe body.
Dependingon
the significanceof
oneor both options
anadaptation
of
the model may be necessary.6.
Binomial drawThroughout the model, where processes with larvae-numbers (intake, incidence, release) are concerned, it is assumed that these processes
can be described by a binomial draw with the given proportions and the resulting numbers as p- and k-parameters respectively. However, the question is whether a binomial draw is a proper description of these
processes. It is imaginable that another distribution would be
preferable.
7. Constant or age dependent mortality
At the
moment, the(aaify)
mortalityof
thefties is
independent ofage. An option that will
be addedin
the nextversion is that
the mortality can change over age. Two options are possible:l.
The mortality rates follow alinear trend
between user-supplied10
?
mortallty-rates
at
selected days.The
mortality rates follow
somefunction, e.B.
Gompertz or Weibull.8.
SimplificationsIn
the current versionof
the model simplifications are avoided wherepossible. The
consequence becomes apparentin the long
run-times mentionedbefore. In
developing afaster
programme-version,it
is possibleto
include a numberof
simplifications. Twoof
them are:1. Instead of
takingall significant
f1y-ages andtheir events
to assemblethe
rsteadystatet population, it
nay besufficient
toe.g.
take only thoseflies that
have had0, I or
(say) 4 precedingbloodmeals.
2.
Throughout the model a binomial drawis
usedto
realize ashift
tolower
larvae-loads. Applying a binomial draw involvesthe
innerproduct
between a binomial table and thestate-matrix
mentionedbefore. It will
beclear, that
an lnner productwith
such a large spaceis
a time-consuming event andit
may,therefore,
be usefulto
replace a binomial draw by a more simple procedure(see
alsoiten
6).9. Biting
behaviour nulliparous versus parousflies
In the
current approach,for
the nulliparousfly-population, it
isassumed
that
leaving the pupae on day=age=lis -
withrespect to
thetiming of the
bloodmeal-cycle-
equivalent withhaving a
bloodmeal(without intake).
Consequently,as
onlythe
parouspopulation
isinvolved in calculations,
a sJ-mulationstarts at
theearliest
fly-agethat
having a bloodmealis
possible (depending on the given duration(s)of the
bloodmealcycle).
However, the questionis
whetherin
thenulliparous
phase thereexist a
rpre-bloodmealr time-lag analogous tothe
least time lag between subsequent bloodmealsin
the parous phase.In other words: is it justifyable to
use the samedistribution
of durationsof
the bloodmealcyclefor
both parous and nulliparousflies.
11
3
4
4.
Users guide4.1 Starting a session
The following steps must be undertaken:
1. Insert
the STARTUP-(system-)floppyin
drive A: and switch on your P.C.2 l,lhen the APl-package has been loaded the MODEL-floppy must
be
insertedin A:
followed by pressing ENTER.(At the moment) only when your
P.C. is
equipped with an IBM CGA (whichis
assumed by the program) using the graphictools is
possible. At this pointof
the startupit is
asked whether the P.C. has a Hercules board.Be
surethat
you give theright
answer (making graphs witha
Hercules board can even damage yourP.C.).
When a Hercules boardis
saidto
be present, using the graphictools is
prohibited automatically.After
terminationof a session,
a new sessioncan be started
by entering TSTARTT.Some remarks:
a Both floppyts are 360 Kb formatted.
Donrt write-protect the model-floppy; replacing information
in files
is prohibited then.When working on hard-disk (recommended):
-
create a subdirectory VECPAR-
copy the contentsof
both floppyrs(not:
COMMAND.COM, AUTOEXEC.BAT START.BAT and VECPAR.BAT)in this
subdirectory.-
copy VECPAR.BATto
the root-directory- start
a session from the root-directory by entering TVECPARTb
c
l2
4.2 Ustng VBCPAR
The program VECPAR
is written in
APL (SISC APL*PLUS). Each timethe
APL-workspace with the model
is
loaded (realized automatically when booting fromthe STARTUP-floppy), the
last
valuesof
the input variablesof
the prevoiussession
is read from
thefile
FINPLR.ASFthat is
updatedduring
that session. Where single choices are involved, the programis
managed by menus.The main-menu looks
like:
Each
time
an actionis
terminated during asession, the
main-menu re- appers(except, of course,
when the user returnsto
the DOS-environment).Description
of
the actions from the main menu:Start si-uulation: With this
choice the next sequenceof actions
isinitiated:
a)
Thefi1e, to
which both input and outputwill
be written must bespecified (without
extension,which is
always.ASF). If
anexisting filename
is specified,
the user caneither
overwrite the contentsof that file or
choose another name.b)
The following menu appears: ICtroose option
The background of alternatives A (and B) and C (and D) is presented in chapter 2.1 and the figure shown there. A simulation- run always debouches into a calculation of the characteristics of a rsteady-stater biting fly-population. The materials for this calculation are, however, the life-events of a cohort followed from birth to death. With option C the intermediate charact.eristics of biting flies of all simulated ages are
calculated and stored in the output-file for inspection afterwards (main-menu: Tabulate output/Draw graphs). With option D this
1
VECTOR-PARASIIE I,IODH.
Anton Plaisier and Gerrit van Oortmarssen Version 1.0 Sept. 1986
Erasus University Rotterdasr
Make choice by using the cursorkeys Fl=Help!
Special keys: EIITR=Confirn choice; Esc= stop without choice
Start simulation Modify Paraoeters Modify VCl,l-data
Shor current input Tabulate output
Draw graphs
Printer setup Enter HELP leve1 Erase files
Return to DOS
A:Simulation of an equilibritrn situation B:A + Direct output to screen + printer
C:A + Store life events by day
D:C + plsgqt output to screen + printer
E:Show current input
output is
automaticallydlsplayed.
with option Ethe user
is offered the(last)
opportunityto
lookat
the currentinput.
After pressing the ESC-keyor
choosing the (STOP)-optionthe
main-menureturns which may be useful
for
changing the input.c) Initialisation of
therun. In this
part thecalculation of
the reference-tableof
the rarvae-submodel (see 2.2.1) takes place. rnaddition
some other simulation-tools are derived fromthe
input- variables.d) The actual
modelrun.
unfortunatelythis is a
slow-proceedingevent.
To a large exent the speedcan,
however, be influenced bythe
valuesof
the input-variables (see the remarksat
theend
ofthis section).
Of course the choice madeat I b) also
(trighfyl influences the speedof
the run: calculatingall
intermediate age-characteristics and printing the the results during
thecalculations prolongs the run-time.
e)
Input and outputof
the runis
savedin
thefile
chosenat
1 a).2..
l{odify paraneters: The following menu appears:Choose variable(s) to be changed
The meaning
of most
input-parametersis
explainedin chapter
z.z (underlinedterms).
About the others:- Donrt
forgetto
give a relevant run- and subtitle.-
maxflyage
denotesthe
numberolTry-"gillo-u"
invorved incalculations to obtain the equilibrium population. As
anexponential death-rate
of
the populationis
assumed the populationtheoretically never
disappearscompletely. For that
reasonmaxflyage must
-
experimentally stated-
be taken as largeas
to guarantee inclusion of fly-ages that significantly contribute to the total age-spectrum. In test-runs (see example in Z.Z.Z)maxf yaqe=25 appeared to be reasonable, but this is, of course,
highry
dependent on the applied death-rate. rt will be
obviousthat
increasing maxflvaRe prolongs the run-time.Another important parameter
is
nlarvacategories (see again examplein 2.2.2).
Asit
would leadto
extremeS, long run-timesat
highpossible
larvae-loads, afly will
not be characterizedby
exact numbersof of larvae,
but by larvae-load-categories. Depending onnmtitle subtitle
maxflyage nlarvacategories Llintake
excessmortsD daiIynortIJI dailymortSD propreleaseSugar propreleaseBlood probcyclelength durstLl
probdurstLl durstL2 probdurstL2 durstL3body
probdurstL3body dailmortlIL2L3
<STOP=Esc>
TIILE FOR RT'N
L4
the
expected larvae-loads the user can determineboth the
rangeand the precision
of
the categories.In
nlarvacateRories the lower boundariesof
the categories are given. Some remarks:-
The upper boundaryof
thelast
categoryis
derivedfrom
thesize
of
thelast
but one category increased byone;
the sizeof
thelast
categoryis
setto at
least 4.- As
during calculationsat
some points(a.o.
when the outputis calculated)
the exact numberof
larvaerather than
thecategories
is
needed, the contentsof the
categories isspread evenly over the numbers within the
categories. fn
thelast
category the weights decrease 1inearly.Some restrictions:
-
Thefirst
category must comprise only 0 (zero)-
The numberof
categories may not exceed 6-
Thelast
upper boundary may not exceed 20 larvae.An example:
catgories ntrmbers in categories
weights
0 1.0000
1
0
1 I .0000
2 2
3
0.5000 0.5000
4 4
5 6
0. 3333 0.3333 0. 3333
o .437 5 0 . 3125 0. 187s 0.0625
7 7
8 9 10
In
selecting the variableto
be changed, a short explanation appears atthe bottom of the screen. For changing data the
cursor-, (forward)delete-and
backspace-keys can beused.
For entering arrayrsonly
commarscan be
usedto
separatethe elements. After
eachmodification the values of the
input-parametersare
checked,inconsistencies and impossibilities are traced
and,
when anerror
is located, the modification-menu returns and theerror
can berectified.
3.
Ilodify VCU-data: Mediated by the following menu,Ctrange data of selected VCU-catching site
the user is
offered thepossibility to give values to
parametersconcerning the results
of
VCU-catchingpoints.
These values serve as a nrparousfliesnrnullipar nrlarvae
csN (STOP=Esc>