COMPONENT RELATIONSHIPS 



into which the environment has been 

 divided — discrete in thought, though 

 not in reality- And many of these 

 entities have been so sundered as to 

 be the subject of separate disciplines 

 requiring quite different training. 

 The meteorologist and the entomolo- 

 gist, the bryologist and the hydrolo- 

 gist are unlikely to come into contact, 

 and unlikely to understand one an- 

 other if they do. Yet weather and 

 insects, mosses and streams are parts 

 of a common over-all pattern within 

 the landscape, and understanding of 

 each considered in isolation is bound 

 to be imperfect. 



Even within a discipline it has been 

 usual to narrow the focus, so that one 

 is looking at a particular organism, a 

 particular function, a particular organ 

 or tissue — perhaps the role of sto- 

 mata in controlling transpiration, the 

 function of kidney tubules, the en- 

 zyme systems of glycolysis, or the 

 mechanism of adsorption of ions on 

 the surface of clay particles. This 

 analytical approach in science — 

 constantly subdividing one's cate- 

 gories, and getting to know more 

 and more about less and less — has 

 had great success. But there is no 

 doubt that its practitioners have 

 found it difficult to see the woods for 

 the trees. 



Over the past twenty years a reali- 

 zation has been growing that this 

 fragmented attitude is inadequate to 

 the subject matter of scientific study. 

 Science is recognizing the need to 

 try to fit the pieces together again 

 and return to the complex whole that 

 is reality. One form of this newly 

 prominent synthetic effort is what 

 has become known as systems analy- 

 sis, involving the application of math- 

 ematical and computer techniques to 

 the problem. 



Systems Ecology — Systems analy- 

 sis applied to ecology ("systems ecol- 

 ogy") views the ecosystem as a whole 

 and examines processes within it as 

 they depend on all the other com- 

 ponents of the ecosystem — meteoro- 



logical factors, soil, plants, animals, 

 and microorganisms. In the analytic 

 approach, the photosynthetic rate of 

 a leaf was viewed in isolation as de- 

 pendent on the radiation impinging 

 on it, and the temperature and hu- 

 midity of the air around it. Perhaps 

 the analytic approach delved even 

 deeper, and the oxygen exchange of 

 a chloroplast was viewed as a func- 

 tion of the radiation of different 

 wavelengths absorbed by the pig- 

 ments and the ionic balance of the 

 protoplasm in which it was embedded. 

 In systems ecology, in contrast, the 

 focus is broader, and attention is di- 

 rected to the gas exchange of the 

 vegetation as a whole, or perhaps to 

 each of the populations of different 

 species of which it is composed; 

 changes in rate of this process are 

 considered, not in a simpler system 

 actually or conceptually isolated, but 

 in their whole real-world context — 

 affected by the general meteorology 

 of the area, by the soil which deter- 

 mines the supply of water and nutri- 

 ents to the roots, by the animals 

 exerting selective defoliation, polli- 

 nating, or transporting propagules. 



In arriving at this overview, sys- 

 tems ecology may indeed make use 

 of the results of analytic studies 

 covering parts of the system. But the 

 process of synthesis will demonstrate 

 processes and effects in the ecosystem 

 that would never have been recog- 

 nized if the partial processes had been 

 considered only in isolation. 



Systems ecology does not avoid the 

 need for simplification — ecosystems 

 are indeed so complex that to think 

 about them in their full complexity 

 would be beyond human powers, 

 even with any conceivable concentra- 

 tion of mechanical aids. But whereas 

 the scientific approach of earlier dec- 

 ades has been by subdivision and 

 isolation — what one might call a 

 "vertical" simplification — systems 

 analysis requires a "horizontal" sim- 

 plification, in which all major com- 

 ponents are considered but each is 

 whittled down to the bare essentials. 



Models and Submodels 



Generally, the synthesis of partial 

 processes into a representation of the 

 ecosystem as a whole is conceived in 

 terms of a model. The practical proc- 

 ess of building and testing models is 

 closely linked with the use of com- 

 puters, both digital and analogue (or 

 hybrid) — in fact, it is doubtful 

 whether this activity would even have 

 approached its present development 

 without the availability of computers. 



Once a model is built, a computer 

 program representing it may be writ- 

 ten, and repeated operation of the 

 computer program then simulates the 

 behavior of the ecosystem, as simpli- 

 fied in the model, under different sets 

 of conditions. Empirical tests of this 

 sort can then play a valuable part in 

 improving the model, even where the 

 analytical work involved in a direct 

 approach would daunt a mathemati- 

 cian. The process of model devel- 

 opment using computer simulation 

 consequently has a large "boot-strap- 

 ping" component. 



Development of an ecosystem 

 model is sometimes based on obser- 

 vations of the ecosystem as a whole 

 — changes in quantities within it, or 

 rate of processes such as the move- 

 ment of material from one part of it 

 to another. It may take the form of 

 a set of differential equations with 

 coefficients to be estimated, perhaps 

 subject to constraints. Alternatively, 

 the model may be divided into a 

 number of submodels, each of which 

 can be studied separately and its best 

 mathematical representation (again in 

 terms of differential or difference 

 equations) determined. Figure IX-3 

 is an example of one such submodel. 

 The submodels are then combined, 

 and the performance of the model as 

 a whole studied. 



These two approaches may in fact 

 arrive at a model of the same struc- 

 ture, but the estimates of constants 

 will differ. If they are of the same 

 structure, the fit to the set of data 

 used will be better with the first ap- 



281 



