The structure of the database has been divided into three sections, described in the data analysis: flora, ecology and production. The database is built with the software -RDBMS File Maker – which has proven efficient and relatively easy to use. The difficulty encountered in structuring the database stems from the fact that vegetation types are sometimes plant combinations whose characteristics result from the statistical analysis of the vegetation and the observation of the environment, summarized in tables, and sometimes mosaics of plant combinations. In the original study (DELMASIG : J. Marie, 2000), we provided only the characteristics of the plant combinations. We therefore had to define what a mosaic represents and what corpus of rules should be applied to compose the floristic, ecological and production characteristics of these mosaics. With the exception of MB, treated separately because of its complexity, the codes used for the mosaics are formed by the juxtaposed acronyms of two elementary plant combinations (e.g. VOR/VH). When the two plant combinations indicated are ecologically close, because they belong to the same level of flooding or to adjacent ones, the code is that of a binary mosaic constituted by the two indicated plant combinations. By convention, the order in which the combinations are indicated places first the one corresponding to the deepest flood. If the combinations have the same flood level, alphabetical order prevails. For PAN, PAR, PAS, which are flooded by local run-off and sometimes by late river flooding, the convention has been adapted as follows. If one of these three associations enters into the composition of a mosaic with a combination that is regularly flooded, the latter association prevails. Thus P (level 3 ]-0.10 - 0.30]) combined with PAN (level 4 ]-0.30 -0.60 ]) is written P/PAN and not PAN/P, according to the depth of flood. On the other hand, if the associations flooded by run-off and late river flood are combined to a dry savanna combination (type TA, TB, ...), the mosaic is written in the order of decreasing flood levels (e.g. PAN/TA). For wide-gradient mosaics, that is to say, when the two associations indicated belong to distant flood levels, the indication is only a shortcut to note a more complex mosaic in which not only the indicated combinations but also others with intermediate flood regimes, participate. For example, B/VSP is a shortcut to indicate a mosaic that includes, in addition to B and VSP, plant combinations such as EOR, O, VOR. The difficulties were solved by determining the characteristics of the mosaic on the basis of those of the component combinations. Calculation rules have been defined so that this can be done systematically and automatically on the basis of the codes given to the combinations and their mosaics. However, these rules may differ, depending on the nature of the information processed. The principle that guided the structuring of the thematic information is to separate the attributes of the 28 basic plant associations (with MB) from the attributes assessed for the mosaics of associations that are derived from that of the component plant associations. The relationship between the source file and the resulting file is performed via a text file called sigle.txt that lists the acronyms of the 120 vegetation types (including TB/TC, treated as a vegetation type) by alphabetical order. When a mosaic composition is launched, the program reads the file sigle.txt containing the acronyms. For each acronym, it draws the necessary resources from the source file and composes the mosaic by computing each field according to the pre-established rules. An acronym with a single plant association is treated as a mosaic of identical plant associations. Special "scripts" allow us to compose tiles by unit, by entering the acronym "by hand", which is convenient for tests. However, these rules may differ, depending on the nature of the information processed : 1 - For the floristic composition of the mosaics, it was decided, regardless of the nature of the mosaic (binary or complex type), that the flora of the mosaic would be a simple "addition" of the floristic compositions of the component combinations. The contribution of species to the vegetation mass is calculated as the average contribution within each combination listed in the mosaic. In the case of a complex mosaic, the presented floristic profile is thus a "shortcut", taking into account only the "starting" and the "ending" plant associations. This simplification of information dispenses with what would have been, in particular for some mosaics, long floristic lists with low contributions of each species. Similarly, the indicator value of the presence/absence of each species in the floristic profile of the association was abandoned in the case of mosaics, as being meaningless in this case. On the other hand, the status of a species in the mosaic is the highest status in each of the two associations (dominant/accompanying/occasional). 2 - For production variables, which are all quantitative, we also decided that the values ​​adopted for the mosaics would be simple arithmetic means of the values ​​of the variables of the two component associations. Systematic tests for complex mosaics were carried out with alternative methods, taking into account the intermediate association values. The differences observed were marginal and did not justify the adoption of such methods. 3 - For ecological variables, however, we decided to take into account the real complexity of the mosaics. When the two characterizations of the mosaic belong to the same flood level or two consecutive ones, that of the mosaic is obtained by a simple "average" - in fact a combination of indices as set out in the table below –of the characterizations of each component associations. On the other hand, when the two associations belong to different and non-consecutive flood levels, the "average" takes into account the intermediate levels. For example, if the two associations belong to flood levels are distant from one level, the "mean" is established for the characters of each of the two associations indicated, and also those of the intermediate level, with equal weight given to each component, which therefore participates as one third to the mosaic. In the previous example, the B/VSP mosaic must be understood as actually the B/LEVEL 5/VSP mosaic in which LEVEL 5 is a mosaic summarizing EOR/O/VOR. The complexity of the mosaics is therefore taken into account in the indexed ecological profiles. Probability profiles are obtained by following empirical rules already used by Pierre Hiernaux for the determination of indexed floristic profiles (eg BP) from the probabilistic indexed relationships linking a plant species (e.g. Echinochloa stagnina ) and a state of the variable (e.g. flood depth class). The combinatory results from the use of the following matrix as indicated in Table 1 Are replaced by a numerical code: +++ code 4; ++ code 3; + code 2; • code 1; 0 code 0
Cod e   Statistical significance   +++   The species is significantly  linked   to the state of the variable at  p<0.001   ++   The species is significantly  linked   to the state of the variable at  p<0.0 1  but not at p <0.001   +   The species is significantly  linked   to the state of the variable at  p<0.05but not at p <0.01      The species is  not  significantly  linked   to the state of the variable at  p<0.05   0   Sampling is not sufficient to conclude
GIS organization (2) CREATION OF THEMATIC DATA
Figure n° 1 :  Operating diagram of thematic databases FLORA part Res ource file    acronym list resulting file ECOLOGY  and PRODUCT part Resource file                  acronym list                                   resulting file      Sigle.txt   AFLORE   MXFLORE   XFLORE                       1   2   3   4   5   AECOLO     APRODUCT     MECOLO     MPRODUCT            Sigle.txt   1   2   3
The resulting file is always in command: the system was imagined and implemented by Isabelle Louise Bisson. It was designed so as to make it possible to update that part of the GIS in several different ways : 1 - the "source" file may be modified by adding files for new plant associations, as long as the acronyms used are alphanumerical codes of no more than 3 characters 2 - without adding new files, it is possible to modify any value related to an existing association file 3 - should VEG4 be changed when updated, new mosaics could be composed simply by modifying the sigle.text list of mosaics. Example of a file processed in MX FLORE :
BP Bourgoutière  basse à  Vossia  cuspidata. Ipomoea  aquatica HA 2 3,75% Ipom_  aqu Nelsonia  canescens HA 2 3,75% Nels_ can Nymphaea lotus HA 4 3,75% Nymp_ lot Polygonum  spp. HA 2 3,75% Poly_  spp Echinochloa  stagnina HD 4 40,00% Echi_  sta Vossia  cuspidata HD 4 40,00% Voss_  cus Centrostachys  aquatica HO 2 1,25% Cent_  aqu Meremia  tridentata HO 2 1,25% Mere_  tri Sesbania  rostrata HO 2 1,25% Sesb_  ros Vetiveria  nigritana HO 2 1,25% Veti_  nig Ziziphus  spina-christi LA 3 15,00% Zizi_  spi Mimosa  pigra LD 2 80,00% Mimo_ pig Acacia  kirkii LO 2 5,00% Acac_  kir
MECOLO file with the names of the items in the database (see page 37 "Ecological variables")
Main Cover   VEG4   Map of plant associations     derived cover   FLORE1   TABLE FLORE         Request about flora     derived cover   CRUE 3   TABLE ECOLOG Y         Request about  ecology   discreet model   of floodable  areas   derived cover   SOL1 & SOL2   derived  cover    VEG 7   Map of the distribution          of vegetal species   Map of Soils   space  operator   Matrix   VEG7   Mathematical  morphology   Curves of equal  flood depth  model     1 -   calculating floodable areas   for all flood heights .  2  -   identifying the main basins 3  –   discretizing of vegetation  map  in main landscapes.   4 –   identifyi  ng the areas of useful  flood for rice crop   5 -   calculating a digital elevation model of the inner Delta   .........      map of floodable  areas for  different   levels  : 6.60 m; 6.50 m;    6,30 m; 6 m; 5.10 m  .
0 1 2 3 4 0 0 0 1 2 3 1 0 1 2 2 3 2 1 2 2 3 3 3 2 2 3 3 4 4 3 3 3 4 4 codes for the probability thresholds (plant association 1) codes for the  probability  thresholds  (plant  association 2)
Figure 2 : GIS organization for the vegetation part