in both the plant and animal sciences, growth functions have been used for many years, usually

to provide a mathematical summary of time-course data on the growth of an organism or part of an organism.. the term growth function is generally used to denote an analytical function which can be written down in a single equation. thus, a general functions connecting dry weight W to time t is W=f(t) where f denotes some functionalrelationship the development of this area has occurred mostly in the plant sciences, and the early papers of richards(1 ! ,1 " ) are well worth reading. #ore recently, te$tboo%s on the topic have been published& 'unt (1 ()) gives a general and readable account of the sub*ect+ ,ouston and -enus (1 (1) have written a more speciali.ed te$t concentrating on the growth function as the /ichards function (p.(!) the use growth functions is largely empirical& the form of the function f will sometimes be chosen by simply loo%ing at the data and ma%ing an informed guess+ however, it is preferable to try to select or construct a function that has some biological plausibility, and whose parameters may be meaningful 0 that is, they may characteri.e some underlying physiologicalor biochemical mechanism or constraint. 1or e$ample , autocatalysis produces e$ponential growth+ limited substrate or nutrient gives rise to an asymptote+ and senescence or differentiation may cause diminishing growth rates and asymptotic behaviour, as in the 2ompert. equation. 3cientist no longer have the naive e$pectation of finding that chimera+ an analytical function which will describe animal or plant growth under a wide rage of condotions and whose parameters have biological significance. 4his is reflected in the growth of animal and plant modeling, and its emergence as a discipline in its own right. 1or it is recogni.ed that it is most unli%ely that one would ever be able to describe a complicated system such as an animal or plant in a simple manner. 4he use of two or more equtions, many paraeters, numerical methods and simulation is now relatively widespread. 5dmittedly, 4he application of such procedures will only define numericaly the dry weight 0 time relationship of equation (!.1) although now, this will be one of several predicted relationship which are based on the theoretical structure or model, which it self incorporates various physiological or biochemical assumptions about mechanism. 5lthough all models are empirical at some level, growth functions may be of value at the organ, tissue and cell levels, as well as at the whole 0 organism level. 6t is convenient, and also traditional, to approach the topic of growth functions by first discussing the rate of growth, or dw7dt in the case of dry weight W. differentiation of equation (!.1) with respect to time gives 8w7dt=g(t) 9limination of the time variable t between equations (!.1) and (!.)) then gives 8w7dt = h(W) Where h is a function. :ote that equation (!.;) is now in the familiar rate = function of state form (p.1") with W as a state variable. 5n equation in the form of equation (!.;) can often be interpreted biologically , usually in biochemical terms + however , for some growth functions

( certain e$ponential polynomials, for e$ample) , it may not be practical to derive an analytical equation (!.;) from equations( !.1< and (!.)) . many growth functions are obtained by first writing down a relation of the equation (!.;) type, although occasionally, as with the 2omper. equation (!.) ), the form 8w7dt Where u denotes some function of w and t, may be preferred. 3ince growth in dry weight is of most interest, our discussion will be in these terms, although clearly the results may be equally applicable to other variables such as liveweight, leaf area, fresh weight, and so on. 5lso, although units should be chosen to suit the occasion, it is convernient here to ma%e a define choice & day as the unit of time, and %g as the unit of weight. 17W dW7dt Which has dimensions of time-1, is frequently encountered in discussions of growth functions. =nfortunately, various terms are used to describe this growth rate derivative. >lant scientists tal% about the relative growth rate, although this is hardly a satisfactory use the word relative. #ost biologists refer to the specific growth rate + the recommended scientific meaning of specific is divided by mass ( royal society 1 ?!, p.1<) , which then strictly precludes its use for 17@ d@7dt Where @ is a volume, area or some other quantity. 6n other areas (for e$ample economics) the term proportional rate of growth is widely used for all quantities of this type+ this seems to be an e$plicit and pertinent use of proportional , although if is unfamiliar to biologist . we will use any one of the three words , depending upon conte$t, and hope this will not cause confusion

2rowth equations 6n this section, a survey is given of the more commonly encountered growthequations. Where possible, these are derived from a simple model, usually by integration of a differential equation of the type equation (!.;) or equation (!.A). this allows some meaning to be associated with the parameters of the W & t equation. 5lthough the common practice of plotting growth data on a semi - logarithmic scale is sound , it has dangers for the unwary & for e$ample, a sigmoid curve might appear non 0 sigmoid . for this reason a typical member of each type of equation is shown on both linear and semi 0 logarithmic scales. 1igure !.1 5 closed two compartment model for considering growth. 4he system is closed with no inputs or outputs, and the growth process tranfers material from the substrate compartment to the second compartment without loss. -arious assumptions concerning how the rate of the growth process depends on W and 3 enable different growth equations to be derived 6n figure !.1, a simple two 0 compartment models is defined. With the assumption that there is no gain or loss of material from the system, therefore

8w7dt = -ds7dt

3o that w B s = a constant = wo Wo anad so are the initial values of w and s at time t = t , wf and sf are the final values of w and s approached as t == assuming that a steady state is eventually reached+ , is a constant. 4he ne$t step is to write the growth rate as a function v of W and 3, so that 8w7dt 3ince from equation (!.") , 3= , 0 W, and by substituting for 3 in equation (!.?) therefore dW7dt=v where h is a function of the single variable W alone, as in equation (!.;) . the problem has become one of the single state variable , W. the crucial step is now what function v one assumes for equation (!.?). this determines the form of equation !.() ,and after integration , growth equation itself, equation (!.1). we now proceed to derive, by ma%ing different assumptions for v of equation (!.?), several distinct growth equations, all of which may be interpreted in terms of the model of figure (!.1) simple e$ponential growth with an abrupt cut 0 off the assumptions are the quantity of growth machinery is proportional to dry weight W+ the growth machinery wor%s at ma$imal rate so long as there is any substrate available at all+ growth is irreversible and it stops once the substrate is a$hausted. 9quation (!.?) becomes dw7dt where - is a parameter %nown as the specific or relative growth rate, - depends on the proportion of W which constitutes growth machinery, but also on the efficiency or speed with which this machinery can operate. 6ntegration of equation (!. ) gives W=w

1igure !.) simple e$ponential growth with an abrupt cut 0 off. W is the dry weight , t is time , both in arbitrary units. 4he curvesdescribe equations (!.1<) with Wo=1CCCC 5n growth stops abruptly when (putting t=

3imple e$ponential growth limited by the amount of substrate available is illustrated in figure !.) . on the semi 0 logarithmic plot the growth curve is linear, since from equation

4he monomolecular equation 4his equation describes the progressof a simple irreversible first 0 order chemical reaction. 4he assumptions are & the quantity of growth machinery is constant and independent of dry weight W+ this machinery wor%s at a rate proportional to the substrate level 3+ growth is irreversible . instead of equation !. ), we now have dw7dt the growth rate decreases continually and there is no point of infle$ion . this can be seen from second differential